CLASSICAL

CONDITIONING
AND OPERANT
CONDITIONING
A Response
Pattern Analysis
CLASSICAL
CONDITIONING
AND OPERANT
CONDITIONING
A Response
Pattern Analysis
Wendon W. Henton
Iver H. Iversen

With 106 illustrations

Springer-Verlag New York Heidelberg Berlin

Wendon W. Henton
Senior Research Scientist-Psychology
Dept. of Health, Education, and Welfare
12709 Twinbrook Parkway
HFX-120
Rockville, Maryland 20857

Iver H. Iversen
Research Psychologist
University of Copenhagen
Ryesgade 34A
2200 Copenhagen N
Denmark

Library of Congress Cataloging in Publication Data

Henton, Wendon W., 1941-
Classical conditioning and operant conditioning.
Bibliography: p.
Includes index.
1. Operant conditioning. 2. Conditioned
response. I. Iversen, Iver H., 1948- joint
author. II. Title.
BF319.5.06H46 150 78-16542

All rights reserved.

No part of this book may be translated or reproduced in any form without written permission
from Springer-Verlag.
Copyright 1978 by Springer-Verlag New York Inc.
Softcover reprint of the hardcover 1st edition 1978

987654321

ISBN-13: 978-1-4612-6312-8 e-ISBN-13: 978-1-4612-6310-4

DOl: 10.1007/ 978-1-4612-6310-4
FOREWORD

Since the appearance of the treatise on "Schedules of Reinforcement" by

Ferster and Skinner over two decades ago, the literature in behavior analysis,
both experimental and applied, has been dominated by a range of studies dedi-
cated to providing ever more systematic and refined accounts of these
"mainsprings of behavior control." For the most part, the analysis has been
pursued in the best traditions of "scientific methodology" with careful atten-
tion to the isolation of controlling variables in unitary form. Of late, relatively
simple interaction effects have provided an important additional focus for more
sophisticated analyses. It is clear, however, from even a cursory survey of the
monumental research and conceptual analysis which is represented in this
scholarly volume by Henton and Iversen that the surface ofthis complex "be-
havioral interactions" domain has barely been scratched.
The primary focus of this pioneering effort extends the competing response
analysis across all experimental schedules, both classical and instrumental, as
well as the interactions between the two. Appropriately, the analysis empha-
sizes overt behavioral interactions, beginning with the simplest case of one
operant and one respondent, and inevitably implicating more diverse and subtle
interactions. As the analysis expands to include interactions between multiple
recorded responses, increasingly more precise empirical specifications ofrecip-
rocal interactions in response probabilities are revealed independently of con-
ventional procedural labels (i.e., operants, respondents, collaterals, adjunc-
tives, etc.) and traditional theoretical distinctions. Overt responses are treated
not as measures of "other things" but rather as behavioral interactions to be
explained by an analysis of functional relationships rather than by attribution to
hypothetical constructs or intervening abstractions.
From a methodological vantage point, the analysis described in this treatise
v
vi Foreword

is advantaged by an imaginative and innovative procedural approach. Unfet-

tered by slavish conventions involving "dependent" and "independent" vari-
able assumptions, this dedicated molecular analysis focuses upon specific sets
of events and interactions as mUltiple determinants of behavior. Of particular
interest in this regard is the unique treatment of concurrent classical-operant
conditioning schedules and the ingenious experimental manipulations which
provide the foundation for a novel interpretive departure from the usual' 'base-
line disruption" account of such interactions. Virtually all combinations and per-
mutations of classical-operant' 'superimposition" procedures are explored sys-
tematically in a series of original experiments which emphasize the significance
not of the "baseline disruption," which occurs across procedures, but of the
concurrent and competing response characteristics which are usually unob-
served and unrecorded.
Of at least as great conceptual import would seem to be the frontal experi-
mental attack by Henton and Iversen upon the myth of "response indepen-
dence" which continues to dominate contemporary views of operant schedule
interactions. In this regard, for example, the data presented on increased rates
of orienting behaviors in concurrent response-independent and signaled rein-
forcement schedule experiments constitute a serious challenge to traditional
accounts of reinforcer interaction effects based upon assumptions of negligible
competing response influences. Similarly, this molecular analysis of response
interactions in multiple operant schedules reveals the critical role of interacting
collateral and competing responses which has all too often been obscured by
current labeling practices (e.g., "contrast effects," etc.). Even conventionally
mislabeled "post-reinforcement pause" and "inter-response time" phenomena
observed in simple operant schedules are shown to be influenced directly by
the temporal characteristics of collateral competing responses. Indeed, a telling
point is made with reference to the issue of response versus reinforcement in-
teractions in the analysis of operant schedules, by directing attention away
from a molar analysis based upon data averaged over sessions and focusing
upon the molecular effects of real behaviors in the form of local response rate
changes which occur and interact frequently in the absence of any local change
in reinforcement.
A special place of importance is reserved in the final chapter of this book for
a somewhat unconventional but characteristically innovative treatment of both
simple and complex classical conditioning paradigms within the empirical
framework of the well-developed and demonstrably productive competing re-
sponse analysis. Here again, the authors present a convincing challenge to one
of the more widely held misconceptions regarding the absence of overt re-
sponses and the resultant misguided theoretical interpretations of classical
conditioning effects. Not only is the data on acquisition of delay condition-
ing, extinction, external inhibition, and conditioned inhibition persuasively
marshaled to document a veritable mosaic of response interactions, but the
long-neglected analysis of concurrent classical conditioning effects, suggested
in the early writings of Pavlov, is creatively exposed to experimental scrutiny.
Foreword VII

This singular empirical analysis with its focus upon the broadly based re-
sponse interactions which characterize behavioral transactions at all concep-
tual levels can be seen to represent a "paradigm revolution" in the truest
sense. The novelty and comprehensiveness of its integrative features are im-
pressive, the gaps in knowledge which it reveals are critical, and the new direc-
tions of experimental inquiry which it suggests are the stuff of which progress
in science has traditionally been made.

J. V. Brady
PREFACE

This volume attempts to organize a ten-year experimental analysis of inter-

acting response patterns in behavioral conditioning. The research program ac-
tually began at the Walter Reed laboratories in 1968, as an extension of Brady's
"competing response" analysis of conditioned emotions. The scope of this
work is the development of the "competing response" analysis into a "concur-
rent response" analysis applicable to many of the common experimental tech-
niques used in behavioral psychology today. More specifically, the present
analysis concerns the concurrent and sequential response interactions gen-
erated by Pavlovian as well as Skinnerian conditioning procedures.
This book, written specifically for laboratory psychologists and students, is
divided into three sections corresponding to the developing conceptual and ex-
perimental analysis of interactions in classical-operant schedules ("emo-
tions," Section I), operant schedules (Section II), and classical conditioning
(Section III). Section I is the completed analysis of "emotions" as the inter-
action of Pavlovian and Skinnerian conditioned responses, beginning with a
comprehensive parameter by parameter review of the literature. The concurrent
response analysis is then developed step by step through an experimental
examination, and places Pavlovian-Skinnerian interactions more firmly within
the broader framework of concurrent conditioning procedures. Section II ex-
pands the analysis into an experimental description of Skinnerian or operant
conditioning methods, with chapters devoted to simple operant schedules, con-
current schedules, and mUltiple schedules, and therefore embraces many of the
common Skinnerian procedures. Section III is the current use of the response
interaction analysis in the context of Pavlovian or classical conditioning. This
latter analysis includes many of the standard classical excitatory and inhibitory
procedures, with a further development of Pavlov's work on concurrent
ix
x Preface

classical-classical conditioning. Our continuous concern has been an adherence

to the inductive method of direct and systematic replication in the determination
of lawful generalizations within and across these popular conditioning tech-
niques.
On the other hand, this empirical approach bumps up against the traditional
mentalisms, inferences, and cognitive theories which have gradually evolved
into "laws" in each of the three research areas. In our introduction, we reply to
the opposing arguments and occasionally heated disproofs of response inter-
actions, with the relevant issues carried forward and specifically illustrated in
each of the subsequent sections and chapters.
In all, the volume attempts a systematic position which might have some-
thing to offer investigators using either classical conditioning, operant condi-
tioning, or classical-operant combinations. As a new analysis, the work neces-
sarily contains various new procedures, reanalysis of old procedures, and a
quantitative analysis of response frequencies and patternings. Our thesis is that
behavioral conditioning procedures must control myriad response interactions
which are important to the understanding of psychology, but which are never-
theless ignored by typical methodologies which record and analyze but a single
response. The concurrent response analysis places these single responses
within the ongoing matrix of simultaneous and sequential response interac-
tions.
Lastly, we must express our continuing appreciation to the many individuals
who have contributed to the making of this work. Our obvious debts to the sci-
ence of B. F. Skinner and M. Sidman will be rather apparent. Many portions of
the book have immeasurably profited from our personal and professional asso-
ciations with an ever growing series of colleagues, including A. Brownstein,
J. V. Brady, J. P. Huston, M. Lyon, C. L. Salzberg, andJ. C. Smith. Significant
contributions have also been gratefully received from J. Boren and C. M. Brad-
shaw. We can only hope that these gentlemen might recognize both their con-
tributions and our deep appreciation.
The statements and opinions expressed in this work represent the views of
the authors and may not reflect the views of the Department of Health, Educa-
tion, and Welfare (HEW), nor any agency of the government of the United
States or Denmark.

W.W.H.
I. H. I.
ACKNOWLEDGMENTS

The following journals and publishers have kindly granted permission to re-
print previously published figures and illustrations. The appropriate citations
are also indicated in the captions of each figure: Almquist and Wiksell, three
figures from Scandanavian Journal of Psychology; Kenyon College, four fig-
ures and one table from The Psychological Record; Pergamon Press, one figure
from Physiology and Behavior; Society for the Experimental Analysis of Be-
havior, three figures from the Journal of the Experimental Analysis of Behav-
ior.
Portions of this work and preparation of the manuscript were supported
by grants from the Danish Research Council for the Humanities, Danish Re-
search Council for the Medical Sciences, and a stipend from the University of
Copenhagen.

xi
CONTENTS

Introduction: Different Views of Psychology 1

Wendon W. Henton and Iver H. Iversen
Structural Phrenology 2
Behavioral Phrenology 4
Reinforcer Value: An Example of a Developing
Phrenology 7
Functional Relationships: An Empirical Alternative 10
References 13

SECTION I: CONCURRENT CLASSICAL AND OPERANT

CONDITIONING PROCEDURES 17

1. Review of Classical-Operant Conditioning, Parameter by

Parameter 19
Wendon W. Henton
History: A Confluence of Three Traditions 19
Emotional States 21
Classical Conditioning 22
Classical-Operant Interactions 24
Theoretical Positions Briefly Considered 25
Experimental Analysis: General Methods 29
Subjects 30
Conditioning Procedures 30
Dependent Variables 30
Negative Classical Conditioning Scheduled with Positive
Operant Conditioning 32
xiii
xiv Contents

Classical Conditioning Variables 34

Operant Conditioning Variables 45
Subject Variables 52
Negative Classical Conditioning Scheduled with Negative
Operant Conditioning 59
Classical Conditioning Variables 61
Operant Conditioning Variables 64
Subject Variables 66
Positive Classical Conditioning Scheduled with Positive
Operant Conditioning 67
Classical Conditioning Variables 68
Operant Conditioning Variables 70
Subject Variables 72
Positive Classical Conditioning Scheduled with Negative
Operant Conditioning 73
Concluding Remarks 74
References 76
2. Empirical Analysis of Concurrent Classical-Operant
Schedules 97
Wendon W. Henton
Interactions between Sidman Avoidance and Appetitive
Classically Conditioned Responses (Experiment I) 97
Experiment I 99
Discussion 106
Concurrent Response Rates During Positive Classical-
Positive Operant Conditioning (Experiments II
and III) 110
Experiment II 111
Experiment III 114
Discussion 117
Concurrent Response Rates During Preevent Stimuli
(Experiments IV, V, VI, and VII) 118
Experiment IV 119
Experiment V 122
Experiment VI 126
Experiment VII 129
Discussion 133
Interactions Between Programmed and Superstitious
Operants (Experiment VIII) 134
Experiment VIII 134
Discussion 137
Suppression of Classically Conditioned Responses by
Superimposed Operant Responses (Experiment IX) 138
Experiment IX 138
Discussion 144
Summary 146
References 155
Contents xv

SECTION II: OPERANT CONDITIONING PROCEDURES 161

3. Concurrent Schedules: Response Versus Reinforcement
Interaction 163
Iver H. Iversen
The Issue of Response Versus Reinforcement
Interaction 163
Response Independence and Reinforcement
Interaction 164
Direct or Indirect Reinforcement Interaction 167
Response Pattern Reciprocity in Concurrent FR VI
Schedules (Experiments I and II) 168
Experiment I 169
Experiment II 170
Discussion 173
Changeover Responses Under Discriminative Control with
Discrete-Trial Procedures (Experiments III, IV,
and V) 174
Experiment III 175
Experiment IV 177
Experiment V 181
Discussion 184
Observing Responses with Signaled Reinforcement
(Experiments VI, VII, and VIII) 186
Experiment VI 188
Experiment VII 190
Experiment VIII 195
Discussion 201
Superstitious Responding with Response-Independent
Reinforcement (Experiment IX) 203
Experiment IX 204
Discussion 207
Concluding Comments 213
References 217
4. Concurrent Responses with Multiple Schedules 221
Iver H. Iversen
Explanations of Behavioral Contrast 221
Component Duration and Collateral Responses
(Experiment I) 224
Experiment I 224
Discussion 227
Positive Contrast and Collateral Responses
(Experiment II) 227
Experiment II 229
Discussion 235
Multi-response Interactions and Local Behavioral
Contrast (Experiment III) 236
xvi Contents

Discussion 237
Concluding Comments 242
A Response-Displacement Model of Positive
Contrast 244
Collateral Responses in Multiple Schedules 245
Conclusion 246
References 247

5. Collateral Responses with Simple Schedules 251

Iver H. Iversen
Reinforced Responses Interacting with Collateral
Responses 251
Pattern Reciprocity Between FR Responding and
Collateral Licking (Experiments I and II) 254
Experiment I 255
Experiment II 256
Discussion 256
Pattern Dependency Between FR Responding and
Collateral Licking (Experiment III) 258
Experiment III 258
Discussion 258
Pattern Interdependence Between FR Responding and
Collateral Licking (Experiment IV) 261
Experiment IV A 263
Experiment IVB 268
Discussion 271
Pauses and Collateral Responses with Reinforcement
Omission (Experiment V) 273
Experiment VA 275
Experiment VB 276
Experiment VC 278
Discussion 279
Interresponse Times and Collateral Responses
(Experiment VI) 281
Experiment VI 282
Discussion 287
Concluding Comments 288
References 293

SECTION III: CLASSICAL CONDITIONING

PROCEDURES 297

6. Response Patterning in Classical Conditioning 299

Wendon W. Henton
Review of Conditioning and Competing Responses 299
Simple Classical Conditioning (Experiment I) 305
Contents xvii

Delay Conditioning (Exeriment IA) 307

External Inhibition (Experiment IB) 309
Extinction (Experiment Ie) 311
Delay Conditioning plus Inhibition of Extinction
(Experiment ID) 314
Delay Conditioning: Replication (Experiment IE) 318
Summary 319
Concurrent Classical Conditioning (Experiments II, III,
and IV) 325
Concurrent Delay-Trace Conditioning (Experiments IIA,
B, and C) 326
Concurrent Delay-Delay Conditioning (Experiments lID
and E) 329
Concurrent Delay-Delay Conditioning: Systematic
Replication (Experiment III) 332
Concurrent Trace-Trace Conditioning (Experiment
IV) 334
Summary 340
References 343

Index 347
CLASSICAL
CONDITIONING
AND OPERANT
CONDITIONING
A Response
Pattern Analysis
Introduction

Different Views of Psychology

Phrenology, that age old study of the phrenes, continues to be the accepted pre-
mise of contemporary psychology-unfortunately. The names and numbers of
phrenes have been hotly debated and frequently changed across the decades.
The appropriate measuring device is still at issue. Yet, the fundamental concept
of physical dimensions as definitive measures of hypothetical entities and
causes has not only been unchanged, but assimilated into all current psycholo-
gies. In the beginning, mental states were measured by the shape of the skull.
More recently, physical behavior has supplanted physical structure as the mea-
suring tool. With some irony, Pavlov's rejection of the "fantastic states" and
mentalisms has been ignored, and Pavlovian conditioning is now offered as a
premier technique for measuring drives, fears, helplessness, motivation, etc.
Similarly, the straightforward assertion by Skinner that behavior is only behav-
ior has also been compromised, with once "radical" behaviorists now promot-
ing overt responses as indices of preferences, memories, inhibitions, values,
general emotional states, etc. Within the phrenological systems, behavior as a
dependent variable is transformed into behavior as an index or indicant, an in-
trinsically trivial but convenient epiphenomenon to measure the more interest-
ing workings of the mind (Ebel, 1974).
Phrenology also offers a labor and time-saving alternative to the "thorny
Calvinist path" of manipulating independent variables and recording depen-
dent variables (Skinner, 1975). The bypassing or circumvention of an experi-
mental analysis is accomplished by simply reversing the logic of empiricism.
Phrenology now begins with behavior and develops a logical regression back-
ward to estimate the necessary characteristics of the imputed cause. The actual
efficiency of the system is that the expressed experimental psychologist need
not bother with experimentation. In practice, the nominal author of an experi-
2 Introduction

mental analysis today does not conduct experiments but provides an editorial
evaluation of experiments previously relegated to technicians and students.
The "empirical" analysis is in actuality a theoretical interpretation appended to
the graphs and data reported from the laboratory. The reasoned analysis
usually stops well short of the actual experimental manipulations in favor of a
theoretical inference advocated by the author.
Behavior as indicants of underlying causes has been ably described and de-
fended by many authors, and frequently applied with great skill. Thus, Stevens
(1951) characterized the larger portion of a handbook of experimental psychol-
ogy as concerned with behaviors as indicants of various intervening agents.
Similarly, Bolles (1967) described the inference of intervening causes from be-
havioral effects as the "bread and butter of psychological theorizing." Such
inferred causal agents may be subdivided into "intervening variables" and
"hypothetical constructs" (review by Mac Corquodale and Meehl, 1948). In-
tervening variables are abstract or metaphorical explanatory devices within
particular logic systems and do not have physical attributes (nonspatiotemporal
entities; Kantor, 1976). In contrast, hypothetical constructs refer to potentially
recordable and verifiable events, frequently within the central nervous system.
In both cases, however, the agents are not actually measured or manipulated,
but inferred through presumably isomorphic physical effects. Perhaps one fun-
damental distinction between theoretical and experimental psychology is that
inferred explanations may be the food stuff of theory but have proven to be a
rather dubious experimental method throughout the history of science (Wal-
lace, 1972, 1974; Bernard, 1957). In particular, the continuing genesis of psy-
chology from philosophy and physiology is such a history of fancied causes,
not only promoted as psychical actualities, but also ostensibly measured
through physical attribution (phrenology).

Structural Phrenology
The original phrenology is generally attributed to Gall and Spurzheim in the
early 1800s, although antecedent traditions may be traced through 17th century
philosophers to early Greece (review by Bentley, 1916; Boring, 1957). The term
phrene comes from early Greek philosophers and quite literally means the dia-
phragm or lungs, and figuratively means life spirit, will, or faculties. We might
speCUlate that this divergence of literal and figurative meanings prompted Aris-
totle to warn against the confusion of facts with "reasoned facts" (review by
Wallace, 1972). Centuries later, philosophers Reid and Stewart deduced that
the mind was a collection of 37 mental states or faculties, differing either quan-
titatively or qualitatively. The 37 propensities were then individually localized
within (conveniently) 37 subdivisions of the brain, with Gall and Spurzheim
adding the proposition that the shape of the overlying skull matched and mea-
sured the amount of tissue devoted to the localized mental function. This
matching law eventually ran afoul of the comparative anatomies of the skull
Different Views of Psychology 3

and the brain, and the suggested topographical isomorphism between skull
shape and brain function became untenable by 1911 (Boring, 1929).
The physical identification of mental states, however, did not die out with
the demise of Gall's system. Phrenology merely fractionated into a multitude of
different hypothesis, each advocating a different physical measure or a differ-
ent measured explanation. The simplist device was to substitute a different
overt measure of the inferred faculty. Boring had hardly pronounced phrenol-
ogy dead when Sheldon and various collaborators proposed that the body was
a more precise and appropriate measure of mental propensities. The physical
shape of the body replaced the physical shape of the skull in the phrenological
analysis (Kretschmer, 1925). Sheldon and associates proposed that the observ-
able physique, characterized as a somatotype, may be used to assess the under-
lying biological morphogenotype. The somatotype summarized the pattern of
three components of the physique: endomorphy (soft and fat), mesomorphy
(hard muscle and bone), and ectomorphy (thin and delicate). Next, 650 person-
ality traits were reduced and combined into three categories: visceratonia (glu-
tonous and affectionate), somatonia (athletic), and cerebratonia (inhibited and
socially withdrawn). The association between the three types of bodies and
temperaments thus suggested that fat people are basically jolly, athletes adven-
turous, and frail people withdrawn (Sheldon, 1942). Subsequent studies demon-
strated that the body somatotype accurately predicted the mental disorders of
psychotics and juvenile deliquents as well as the personalities of more normal
subjects (Sheldon, 1949). Unlike Gall's system, the constitutional psychology
of Sheldon does not seem to have attracted many adherents (Hall and Lindzey,
1957).
Ironically, the progress in neuroanatomy critical of Gall's phrenology gave
rise to a different and more severe physiological phrenology. The new phrenol-
ogy simply eliminated skull shape as the physical measure, while continuing to
retain the identity between brain structure and mental function. The later
phrenology proposed that regions of the brain with different histological struc-
ture necessarily have different psychical functions, and, in extreme form, local-
ized different mental processes in different architectural layers of the cortex
(Broadman, 1925). The assertion that isolated, unique functions could be iden-
tified with isolated, unique structures was soon criticized as little more than his-
tological phrenology. Franz (1912) argued that such a severe localization was
antithetical to the interdependent and integrated activity of the central nervous
system. Moreover, Franz questioned the accuracy of operational definitions
equating histological and clinical localization with mental localization: "From
the anatomical and physiological standpoint, we deal solely with associations
of an anatomical and physiological character" (Franz, 1912, p. 327). The stark
claim that physiology is solely physiology seems to have been at least as radical
in 1912 as in 1970. In spite of the early "radical" physiology of Franz, many
predominant theories within psychology continue to identify special attributes
with unique structures (e.g., the hippocampus as the center of motivation,
learning, memory, information storage, or processing; review by Young, 1976).
4 Introduction

Behavioral Phrenology

The early explanatory systems of Gall, Sheldon, and others might be character-
ized as static phrenology, in that fixed and relatively unchanging physical at-
tributes were employed as the measure of the inner cognitive workings. The
attendant problems of measuring a dynamic function with a static characteristic
resulted in the more sophisticated systems of the twentieth century. The prob-
lem has been generally solved by matching a changeable variable rather than a
fixed constant with the presumed dimensions of the conceptual functions.
Within psychology, the physical dimension of choice has been some character-
istic of overt behavior, such as response probability, amplitude, or latency. To
be crude, the differences between static and dynamic phrenology are akin to
the differences between palmistry and theoretical psychology. The palmist pre-
dicts events by analyzing the fixed lines and features of the hand, where as the
psychologist infers events not from the structure but the movement of the
hand, or forepaw, of the subject. Little doubt is left in both cases, however,
that the real matter of interest and importance lies elsewhere, at other levels
and in other dimensions than the corporeal hand.
An early example of such dynamic phrenology is Fechner's law of psycho-
physics. Weber, in 1834, had previously observed that the fraction by which a
physical stimulus must be changed to be minimally discriminable is a constant
and is roughly independent of the absolute magnitude of the initial stimulus
(as/s = C). Fechner (1860) reformulated the Weber fraction to address the
mind-body problem and substituted the term sensation for stimulus. Fechner
proposed that mental sensation or perception was a logarithmic function of
stimulus magnitude, and, in result, initiated the physics of the soul. "Psycho-
physics" remains one of the major areas within contemporary psychology,
with current controversies primarily concerned with the exact relationship be-
tween subjective sensation and stimulus magnitude. The underlying assertion
that behaviors are accurate measures of sensations is either ignored or as-
sumed. Only signal-detection theory currently questions the measurement rela-
tion between responses and sensations (Swets, 1961). The bulk of data from
signal-detection experiments is rather clear: the distribution of responses is not
a direct measure of stimulus characteristics but is also controlled by the be-
havioral contingencies scheduled for each response of "Yes, I see it" and "No,
I do not see it. " However, signal-detection theory maintains the psychophysics
tradition by subdividing responses into separate estimates of sensation and re-
sponse bias, and then proposing that physical behavior, corrected for bias,
measures not only sensation but also the "noise" in the sensing system. The
argument between signal-detection and other theories is not whether behaviors
are measures of underlying sensation or perception, but only whether the psy-
chophysical measure is direct or indirect.
The assumption of independent and exclusive sets of stimulus variables and
bias variables critically simplifies the theoretical correction for sensations but is
not wholly consistent with stimulus and schedule interactions noted in condi-
tioning experiments. Relatively recent studies have reported changes in re-
Different Views of Psychology 5

sponse bias with manipulation of stimulus variables within the signal-detection

paradigm (Nevin, 1970; review by Dusoir, 1975; Nevin et at., 1975; Molinari et
at., 1976). A rather large number of studies, in both the classical and operant
conditioning literature, demonstrate that schedule effects are conditional upon
the presence and absence of associated stimuli, and that stimulus discrimina-
tion is in tum conditional upon differential reinforcement schedules (e.g., Fer-
ster and Skinner, 1957; Sokolov, 1960). As Nevin et al. (1975) suggested, the
signal detection procedure is a particular case of discrimination conditioning,
and the psychophysical interpretations must eventually be consistent with the
related data in animal discrimination experiments. Nevertheless, signal detec-
tion has enabled a large step forward in the recognition that behavior cannot be
directly reduced to a unidimensional sensory construct.
Additional psychophysical systems offer behavior as an indicant of more
central entities and functions rather than afferent sensory processes. For exam-
ple, changes in operant response rates may be used as measures of emotions, as
well as motivation, drive, expectancies, etc. The logical derivation of emo-
tional measurement usually requires a now standard introduction, beginning
with a description of operant rates during the presentation of a classical condi-
tioning procedure. The response rate changes are then equated with condi-
tioned emotional responses, which in tum are effects of underlying emotional
states. The emotional states are next dependent upon associationistic pro-
cesses of excitation and inhibition. By appropriate sequential substitution of
logical identities, the initial semantic definition of emotion as physical behavior
can be transformed into a syntactical relationship between emotions and asso-
ciationistic states. The initial synonyms of behavior and emotions are simply
separated into a causal interpretation, with the conclusion that behavioral
changes are dependent upon and a measure of underlying associative pro-
cesses. Brady (1975) similarly criticized the "use and abuse" of interchange-
able definitions and the "host of semantic, linguistic and taxonomic problems"
that result in emotion' 'persistently reified as a substantive 'thing' ." In a similar
vein, Skinner (1963) suggested that operational definitions have not been espe-
cially successful within psychology, in that the defining operations are widely
applied as logical equations relating behavior to mental events, but in fact set
severe restrictions upon knowledge of subjective events.
Indeed, the current literature abounds with various proofs of specific and
general emotional states as actual and substantive things, with diagrams and
equations depicting the exact relationship between emotional causes and be-
havioral effects. The contemporary operational definitions are frequently ap-
plied as little more than phrenological assertations. The mental entities are not
reduced to the operations of measurement, but the physical measures are of-
fered as indicants and proofs of the mental entities. More generally, matching
an actual dimension of behavior and a hypothetical dimension within a mental
network is an example of the philosophers "category mistake" (Ryle, 1949) or
the "substantialization of abstracta" (Reichenbach, 1958). The advantages and
disadvantages of operational definitions are also discussed at length by Bridge-
man (1927, 1959) and Stevens (1939).
6 Introduction

The behavioral substantialization of cognitive abstracta accepts the double

assumptions of both a close correspondence between overt and covert func-
tions, and a further correspondence between covert and cognitive functions.
Alternative explanations in physiological psychology emphasize only the first
premise and either minimize or disregard the second assumption. The be-
havioral physiological psychology attempts to isolate and quantify the func-
tions of different portions of the actual brain. One of the more common tech-
niques is the double dissociation procedure described by Teuber (1955) and
Rosvold (1959). Double dissociation is an experimental manipulation which
demonstrates that lesion of one area ofthe brain affects behavior A, not behav-
ior B, whereas an injury to a second area only affects behavior B, not behavior
A. Mishkin (1972), for example, reported that graded lesions in the anterior in-
ferotemporal area systematically affected simultaneous visual discrimination,
not pattern learning. Posterior inferotemporallesions, however, affected pat-
tern learning, not simultaneous discrimination. The data suggested that the two
separate behavioral effects are influenced by different subsystems of the brain.
Double dissociation is a rather powerful technique for determining the influ-
ence of specific portions of the brain upon behavior. However, few physiologi-
cal psychologists apply the same logic of dissociation to the supposed measure-
ment relationship between behavior and central nervous system functions. The
conditioning literature again amply demonstrates that behavior can be manipu-
lated across a broad range by altering schedule parameters and contingencies
independent of physiological manipulation. The dissociation then suggests that
behavior is not necessarily an invarient or direct measure of specific functions
of the brain. Similar dissociations have been reported between behavior and
the activity of the motor cortex (Fetz and Finnocchio, 1971) and between neu-
ral responses and eliciting stimuli in sensory systems (Fox and Ruddell, 1970).
These observations would suggest some caution in interpreting temporal correlations
as final evidence for functional relations . . . Such a consistent temporal correlation
under a variety of behavioral conditions would seem to be strong evidence for a func-
tional relation. When cell and muscle activity were simultaneously included in the re-
inforcement contingency, however, we found that the correlated muscle activity
could be readily suppressed. These observations suggest that a possible test of the
stability of an observed temporal correlation would be the operant reinforcement of
its dissociation. (Fetz and Finnocchio, 1971, pp, 434-435.)
The dissociation of physiological and behavioral responses is comparable to
earlier demonstrations that skull shape is not invariantly correlated with brain
shape, and equally questions the phrenological assumption that overt attributes
are invariant measures of covert functions. The bright prospect of conditional
overt-covert response correlations in fact promises a far more comprehensive
understanding of behavior and has long been espoused by diverse psycholo-
gists (James, 1890; Miller, 1969; Pavlov, 1927; Skinner, 1963). The conditioned
modification of covert responses (intereoceptive "feelings," for example,
Harris and Brady, 1974; review by Razran, 1961) is of a piece with the variable
rather than fixed integration of physiological responses within the total be-
havioral pattern. This analysis of conditional rather than invariant correlations
Different Views of Psychology 7

between behavioral and physiological functions begins to fulfill the promise of a

more adequate psychology by considering events within the skin as determinis-
tic responses themselves, rather than mere mediators of behavior.

Reinforcer Value: An Example of a Developing Phrenology

The development of phrenological systems is now difficult to fully appreciate
since most contemporary learning theories have developed within the psycho-
logical traditions established years ago. The original assertions are either ac-
cepted today as traditional and useful truths or rejected as nineteenth century
curios. However, at least one currently developing interpretation is verging
upon becoming another phrenology. For a decade or more, we have known that
rate of reinforcement is one of the independent variables that can control be-
havior (Herrnstein, 1961). The functional relationship is occasionally reversed,
and behavior subsequently offered as an exact measure of reinforcement rate.
Next, reinforcement is redefined as obtained reinforcers, in tum equated with
the value of the reinforcer. Reinforcer or reinforcement value is further defined
to include the effects of parameters other than reinforcement rate. Drugs and
deprivation manipulations, for example, are said to affect response rates by
means of changes in the "reinforcement context" (Cohen, 1973; Herrnstein
and Loveland, 1974). Alternatively, reinforcer matching may involve redefin-
ing responses as "other sources of reinforcement" or "inherent reinforcing
value" (Rachlin, 1973), as well as incorporating various free-floating constants
and exponents. The relative rate of specific behavior can then be argued to
match exclusively the computed reinforcer value, which in tum can be inferred
from the explained response rate. The explanatory parameter of reinforcement
rate has then gradually expanded from a physical variable to a generic value
emerging from the organism and assigned to responses (Baum, 1974). (The de-
velopment of the response-reinforcer value matching law may be found in a
variety of theoretical papers: Baum, 1973; Catania, 1963, 1966, 1969, 1973;
Herrnstein, 1961, 1970, 1974; Herrnstein and Loveland, 1974, 1975; Mazur,
1975; Rachlin, 1971, 1973; Rachlin and Baum, 1969, 1972.)
Additional semantic problems arise when the matching law is said to demon-
strate an independence between responses, with interactions only between
values or reinforcement. Catania previously noted that". . . in order to main-
tain the independence oftwo compatible operants, it is necessary to make them
incompatible . . ." and ". . . [it] is often necessary to program concurrent
schedules in such a way that the operants become even more incompati-
ble . . . " (Catania, 1966, pp. 215-216). However, the term incompatible re-
sponses is not strictly interchangeable with independent responses, since in-
compatible responses are not independent but temporally dependent. The
emission of one response specifically precludes the occurrence of all incompati-
ble responses. Perhaps consistent with the terminological confusion, the
matching law seems to demonstrate that rates of independent responses (de-
fined as incompatible responses) match a single dimension of relative reinforce-
8 Introduction

ment rate (nevertheless defined as multiple public and nonpublic dimensions),

but only by demonstrating that response patterns do not actually match rein-
forcement patterns (Catania, 1969, 1973, Rachlin and Baum, 1%9).
With some irony, the reinforcer matching law is primarily sponsored by the
Journal of the Experimental Analysis of Behavior, which previously editorial-
ized specifically against the use and abuse of interchanging definitions (e.g.,
Editors' comments, 1969, vol. 12, pp. 845-846). However, the hegemony of the
reinforcer value interpretation has been sufficient to almost oust alternative ex-
planations of operant responses. As with previous phrenologies and matching
laws, the interpretive proposition has been that one physical dimension is iso-
morphic with a second physical dimension (response rate matches reinforce-
ment rate, skull shape matches brain shape), followed by the proposition that
one or the other physical dimension is also isomorphic with a more abstract
dimension within a particular logic system. The sequence of logical derivations
has been to identify a causal relationship, reverse it, relabel dependent measure
as an exclusive indicant, and then progressively substitute and interchange def-
initions until a suitable generality is obtained.
Unidimensional explanations also involve the additional constraint that
other previously identified independent variables are actually without effects or
can be redefined and subsumed as special cases of the reference parameter.
The independent variables known to control behavior, however, are exceed-
ingly diverse and numerous, with no single variable both necessary and suffi-
cient to control response rates. Psychological Abstracts, for example, rou-
tinely lists several thousand experimental variables. Similarly, Sidman (1960)
concluded a general discussion of statistical description and interpretation by
noting that:
A sufficient number of experiments have demonstrated that the behavior of the indi-
vidual subject is an orderly function of a large number of so-called independent vari-
ables. Indeed, we may now presume such orderliness to be the rule rather than the
exception . . ." (Sidman, 1960, p. 49).

The more complicated interactions involving three, four, or five variables lit-
erally number in the millions. Considering response rate as an exclusive indi-
cant is thus tantamount to entering a matrix of innumerable behavioral func-
tions, and then knowledgably choosing the one function specifying the
theoretically "correct" relationship. Indeed, the benefit of response rate as a
basic datum lies in it being a generality across many functional relationships
rather than uniquely dependent upon reinforcement rate or any other single pa-
rameter.
The multiplicity of empirical determining events, however, does not neces-
sarily contradict a unidimensional theory of behavioral causality. The empirical
parameters are simply transformed and redefined to converge into an interven-
ing agent, which in tum directly changes the appropriate behavioral indicants.
A set of empirical variables may be said to coalesce into an excitation or value
or sensation that is in a one-to-one relationship with behavior. More elaborate
inferences may be drawn, however, with one set of variables producing a gen-
Different Views of Psychology 9

eral excitation, and a second set of variables producing a general inhibition,

with excitation and inhibition then converging into a single explanatory dimen-
sion of "general action potential," "general motivational state," or the "final
common path." As a most general scheme, the multitude of independent vari-
ables eventually coalesces into a final intervening agent that is isomorphic with
and inferable from observed behavior. No other inferential logic would in fact
make sense. If the intervening agent is not both necessary and sufficient for the
actual behavior, then the proposed explanatory agent could not be unequivo-
cally inferred from the behavioral effects. Most inferential models therefore
critically sidestep the possibility of interdependencies between end terms; that
is, many "effects" also "feedback" to affect "causes." For example, behavior
is said to depend upon reinforcement, but reinforcement also depends upon the
occurrence of the behavior. Similarly, the emission of one response may affect
the patterning of a second response, but the emission of the second response
may as well affect the first response. Such interdependence between end terms
(,'feedback loops") renders inferential causality as quite difficult if not impossi-
ble, even in simple physiological preparations (Horridge, 1969).
More basically, a single event A can only be said to exclusively measure or
reflect event B provided observation reveals an ever-existing correspondence
between A and B. This is not quite the model attained with the inference of
behavioral causality. In no case has a record ever been produced of a covaria-
tion between a measuring behavior and an inferentially measured intervening
agent. More importantly, an exclusive measurement relationship of "A only if
B" is a very strong claim that can only be justified within causal systems that
literally prevent interactions with extraneous events.
At this point, the hapless psychologist is in something of a quandary, with
little more than three options. First, we might attempt to isolate such an experi-
mental preparation, knowing only full well that any results would have abso-
lutely no generality to other closed systems, or to different systems with inter-
acting variables. Second, we might simply deny the possible existence of any
interacting variables, and insist that the A-B function is unconstrained and uni-
versally independent of other variables-thus the requirement that sensation is
independent of reinforcement schedules, that reinforcement interaction is inde-
pendent of response interaction, etc. This independence proposition would
support exclusive causality and inferential measurement and obviate the need
for isolated causal systems; but tends to fly in the face of the interacting rela-
tionships accumulated over the years.
Finally, we might forfeit the ghosts of inference and phrenology to first admit
and then study the matrix of causal interactions within the phenomenal world.
The latter path also suffers disadvantages, by requiring considerable time and
effort-occasionally sweatshop effort. One standard and classic scientific
method approximates an isolated causal system by determining the relationship
between one recorded variable A and one manipulated variable B, while hold-
ing constant all other variables (C, D, . . . ,N). Next, the A-B function is
redetermined again and again when variable C is held constant at different
values, only to be redetermined again when variable D is also changed to a dif-
10 Introduction

ferent value. As an illustration, we might examine the rate of responding as a

function of five different rates of reinforcement, at each of five different depri-
vation levels, at each of five schedule requirements. Even this relatively simple
interaction between only three independent variables would require 125 experi-
mental determinations. At this point, the investigator is quite tempted to retire
gracefully from the laboratory and assume a more comfortable inferential
rather than experimental analysis. Yet, we are quite fortunate that many previ-
ous psychologists have pitched in to at least begin if not finish the empirical
analysis of behavior.

Functional Relationships: An Empirical Alternative

On the other hand, a theoretical generality of "A only ifB" may be offered as
an extreme and starting bargaining position to focus experimental attention
upon a particular variable. Almost any relationship between A and B can be
systematically manipulated, and indeed most laboratory investigators attempt
to do so. In virtually all cases, the initial statement of an unconditional and ex-
clusive relationship is eventually qualified by subsequent experimental effort to
"A if B, conditional upon C, D, . . . ." These empirical functions express the
established dependencies between factual events and are neither hypothetical
nor refutable. Consequently, functional relationships avoid debate about the
possible dimensions, influences, origins, etc. of unrecorded entities, and in-
stead emphasize the observed relationship between actual behavior and actual
events. As such, the described relationships enjoy Wittgenstein's (1922) admo-
nition that what can be said can be said clearly, and interobserver agreement
must soon flourish.
This direct verification of a behavioral function, followed by a systematic
examination of the additional variables previously held constant, will yield a
description ofthe interactions necessary for "positive results" (A ifB, given C,
D, . . . ) and, equally important, the interactions yielding "negative results"
(no A ifB, given E, F, . . . ). The direct and systematic replication techniques
generate families of interrelated data that not only define the accuracy of causal
statements, but also delineate the additional variables in which a particular be-
havioral function mayor may not be obtained. A general principle is then not so
much a universally accurate statement, but a functional relationship that con-
tinues to be found across a broad range of additional and interacting functions.
Recognition of the limits of any generality in turn serves to focus attention upon
the different sets of variables that qualify each general principle. Under the best
conditions, a further analysis of the different sets of qualifying variables may
then lead to a "paradigm shift" or data revolution (Kuhn, 1970)-a third func-
tion that firmly interrelates and merges the formerly disparate generalities. The
systematic evaluation of interacting functions has fortunately been successful
in psychology (Sidman, 1960) as well as medicine (Bernard, 1957). Rather than
some damnable aberration of a science gone wrong, the experimental qualifica-
tion of one function by additional functions may establish first the reliability
and then the generality of auy explanatory principle.
Different Views of Psychology 11

An additional consequence of the array of interacting functions is that the

behavioral analysis tends to become highly specific or "molecular." The anal-
ysis of actual behavior patterns occurring in real time is in fact rather immune
to tjxp}anation by mol"f ~Qn~tructs, Accumulated over many subjects or min-
utes, the averaged rate of one response may be relatively higher than the
averaged rate of a second or a third response, with a plausible inference that the
first response was generated by a higher value or motivation or expectancy; but
the second and third responses may each occur exclusively in any given inter-
val for any given subject. Obviously, this momentary' 'preference" for another
response cannot easily be explained by a higher value assigned to the first re-
sponse. Such molecular behavioral changes are said to be irregular, chaotic, or
indeterminate with respect to the molar analysis of behavior, and are frequently
dismissed as trivial or too microscopic.
Some decades ago, momentary behavior oscillations were said to be due to
an "oscillatory force" acting upon the "effective reaction potential" (Hull,
1943) or attributed to "fleeting emotions" (Woodworth and Schlosberg, 1954).
The local changes in behavior are rather similarly explained today. Thus, Baum
has argued:

From moment to moment, however, the organism engages in one activity or another
and switches from one activity to another. These moment-to-moment relationships
among activities have little to do with value, because at any moment the organism
may be engaging in an activity of any value; it simply engages more often in high-val-
ued activities. The momentary fluctuations in an organisms activities result from mo-
mentary fluctuations of variables that have a constant average effect over extended
periods of time (e.g., deprivation). (Baum, 1973, p. 150)

The argument proposes that specific changes in behavior are unrelated to the
explanatory value, which enters the arena only when the behavioral effects are
summed together and statistically averaged. Given the discrepancy between
explanation and observation, we might reach an opposite conclusion: the molar
explanatory device has little relevance to the control and prediction of the ob-
served changes in behavior. Quite recently, Shimp (1975) also suggested that
moment-to-moment behavioral changes are directly controlled by schedule
contingencies, and molar results are by-products of more fundamental relations
involving local behavioral effects. Shimp more explicitly argued that a molar
analysis becomes superfluous to the extent that a molecular analysis is success-
ful.
Historically, the examination of molecular behavioral effects dates back to
the beginning formulations of an experimental analysis of behavior (Pavlov,
1927; Skinner, 1938). The early Russian investigations of isolated classical con-
ditioned reflexes soon led to the trial-by-trial analysis of interacting reflex sys-
tems (Ukhtomsky, 1954; Anohkin, 1958; Konorski, 1967). Skinner (1938) simi-
larly identified the two principle research tasks within operant psychology as
the identification of the laws of isolated responses, and then the analysis of the
interactions between responses. "Preference," as one case, may be analyzed
as the switching back and forth between responses rather than a construct in-
ferred from statistically averaged response rates (Skinner, 1950). More re-
12 Introduction

cently, interacting behavior patterns have also been reported under the general
rubric of competing responses, and applied to the investigation of functional
interactions within classical and operant schedule combinations ("conditioned
emotional responses," Brady, 1975).
The neglect and occasionally heated denial of local behavioral interactions
by contemporary theorists is then both surprising and disappointing. The mo-
ment-to-moment changes in behavior are simply too apparent and too complex
to be ignored or argued into an explanatory limbo. We suspect that a science of
behavior might eventually require general principles to explain actual be-
havioral changes, perhaps in terms of molecular functional relationships, rather
than averaged behavioral changes in terms of global logical constructs. The ear-
lier reports of behavioral patterns in classical conditioning schedules, operant
conditioning schedules, and classical-operant combinations do in fact seem to
require an explicit analysis of interactions and changeovers between responses.
This analysis of response patterning and behavioral interactions is then both
the purpose and substance of the present work.
In sum, laboratory psychologists are acutely sensitive to assertions that be-
havior is a "measure" of something else, either hypothetical or actual. We
might summarize the objections to both general phrenologies and unidimen-
sional matching laws by paraphrasing Skinner's previous critiques: The objec-
tion to general phrenologies is in the use of responses as a measure which ap-
peals to events taking place somewhere else, at some other level of
observation, described in different (and contradictory) terms, and measured, if
at all, in different dimensions (Skinner's comments on learning theories, 1950,
p. 193). The objection to matching laws is that response rates predicted by
some hypothesis of sensation or rational utility can be generated by manipulat-
ing and balancing a host of variables, but it would be a mistake to regard these
as ultimate conditions, or to stop the search for other functional relationships
(Skinner's comments on concurrent schedules and matching laws, 1966, p. 26).
Reminiscent of the earlier "radical" physiology of Franz and Bernard, Skin-
ner, Sidman, and other "radical" psychologists insist that behavior is solely
behavior, and neither a mental nor dimensional measurement.
To this point, the present volume has been something of a harangue-heart-
felt and long felt, but a harangue nevertheless. This introduction is also a partial
reply to comments over the past decade offered against reporting response pat-
terns in behavioral conditioning ("nonexistant," "trivial," "irrelevant"), and
usually arguing for adherence to some explanatory faculty, sensation, state,
value, or other preferred psychological construct. Fortunately, conformity to
data instead of insistent theory has remained a relatively easy choice year after
year. The remainder of the volume is then a rather more positive description of
the complex behavioral patterns obtained in classical and operant conditioning.
The work is unabashedly indebted to the continuing influences of J. C.
Smith, Aaron Brownstein, and J. V. Brady. Although none of these gentleman
would entirely agree with the present analysis, we would nevertheless hope
that each would accept some responsibility for the effort and scope of the pres-
ent work.
Different Views of Psychology 13

The volume is generally arranged in a chronological organization of 10 years

of research into the extension of Brady's competing resPQnS~ Wlruy~i~ in- "nd
cludes a variety of classical and operant conditioning combinations (Part I),
complex and simple operant schedules (Part II), and simple and concurrent
classical conditioning procedures (Part III).

References
Anohkin, P. K.: The role of the orienting-exploratory reflex in the formation of the conditioned
reflex. In, L. G. Voronin, A. N. Leontiev, A. R. Luria, E. N. Sokolov, and O. S. Vinogradova
(eds.): Orienting Reflex and Exploratory Behavior. Moscow, Academy of Pedagogical Sciences,
1958.
Baum, W. M.: The correlation-based law of effect. J. Exp. Anal. Behav., 20, 137-153, 1973.
Baum, W. M.: Choice in free-ranging wild pigeons. Science, 185,78-79, 1974.
Bentley, M.: On the psychological antecedents of phrenology. Psychol. Monogr., 21, 102-115,
1916.
Bernard, c.: An Introduction to the Study of Experimental Medicine. New York, Dover, 1957.
Bolles, R. C.: Theory of Motivation. New York: Harper & Row, 1967.
Boring, E. G.: A History of Experimental Psychology, 1st ed. New York, Appleton-Century, 1929.
Boring, E. G.: A History of Experimental Psychology, 2nd ed. New York, Appleton-Century-
Crofts, 1957.
Brady, J. V.: Toward a behavioral biology of emotion. In, L. Levi (ed.): Emotions-their Parame-
ters and Measurement. New York, Raven Press, 1975.
Bridgeman, P. W.: The Logic of Modern Physics. New York, MacMillan, 1927.
Bridgeman, P. W.: The Way Things Are. Cambridge, Mass., Harvard University Press, 1959.
Broadman, K.: Vergleichende Lokalisationslehre der Grosshirnrinde. Leipzig, Barth, 1925.
Catania, A. C.: Concurrent performances: reinforcement interaction and response independence.
J. Exp. Anal. Behav., 6,253-263, 1963.
Catania, A. C.: Concurrent operants. In, W. K. Honig (ed.): Operant Behavior: Areas of Research
and Application. New York, Appleton-Century-Crofts, 1966.
Catania, A. C.: Concurrent performances: inhibition of one response by reinforcement of another.
J. Exp. Anal. Behav., 12,731-744, 1969.
Catania, A. C.: Self-inhibiting effects of reinforcement. J. Exp. Anal. Behav., 19,517-526, 1973.
Cohen, I. L.: A note on Herrnstein's equation. J. Exp. Anal. Behav., 19,527-528, 1973.
Dusoir, A. E.: Treatments of bias in detection and recognition models: a review. Percept. Psy-
chophys., 17, 167-178, 1975.
Ebel, R. L.: And still the dryads linger. Am. Psychol., 29,485-492, 1974.
Fechner, G. T.: Elemente der Psychophysik. Leipzig, Breitkopf and Hartel, 1860.
Ferster, C. B., and Skinner, B. F.: Schedules of Reinforcement. New York, Appleton-Century-
Crofts, 1957.
Fetz, E. E., and Finnocchio, D.: Operant conditioning of specific patterns of neural and muscular
activity. Science, 174,431-436, 1971.
Fox, S. S., and Rudell, A. P.: Operant controlled neural events: functional independence in be-
havioral coding by early and late components of visual cortical evoked responses in cats. J.
Neurophysiol., 33, 548-556, 1970.
Franz, S. I.: New Phrenology. Science, 35,321-328, 1912.
HaIl, C. S., and Lindzey, G.: Theories of Personality. New York, Wiley, 1957.
Harris, A. N., and Brady, J. V.: Animal learning: visceral and autonomic conditioning. Annu. Rev.
Psychol., 25, 107-133, 1974.
Herrnstein, R. J.: Relative and absolute strength of response as a function of frequency of rein-
forcement. J. Exp. Anal. Behav., 4,267-272, 1961.
Herrnstein, R. J.: On the law of effect. J. Exp. Anal. Behav., 13,243-266, 1970.
14 Introduction

Herrnstein, R. J.: Formal properties ofthe matching law. J. Exp. Anal. Behav., 21, 159-164, 1974.
Herrnstein, R. J., and Loveland, D. H.: Hunger and contrast in a multiple schedule. J. Exp. Anal.
Behav., 21,511-517, 1974.
Horridge, G. A.: The interpretation of behavior in terms of interneurons. In, M. A. Brazier (ed.);
The Interneuron. Los Angeles, University of California Press, 1969.
Hull, C. L.: Principles ofBehavior: an introduction to Behavior Theory. New York, Appleton-Cen-
tury-Crofts, 1943.
James, W.: Prinicples of Psychology. New York, Holt, 1890.
Kantor, J. R.: Behaviorism, behavior analysis and the career of psychology. Psychol. Record, 26,
305-312, 1976.
Konorski, J.: Integrative Activity of the Brain. Chicago, University of Chicago Press, 1967.
Kretschmer, E.: Physique and Character. New York, Harcourt, 1925.
Kuhn, T. S.: The structure of scientific revolutions. In: International Encyclopedia of Unified Sci-
ence, vol. 2, no. 2. Chicago, University of Chicago Press, 1970.
Mac Corquodale, K., and Meehl, P. E.: On a distinction between hypothetical constructs and inter-
vening variables. Psychol. Rev., 55,95-107, 1958.
Mazur, J. E.: The matching law and quantifications related to Premack' s principle. J. Exp. Psychol.
[Anim. Behav. Processes], 1, 374-386, 1975.
Miller, N. E.: Learning of visceral and glandular responses. Science, 163,434-448,1969.
Mishkin, M.: Cortical visual areas and their interactions. In, A. G. Karcymar and J. C. Eccles
(eds.): Brain and Human Behavior. Heidelberg, Springer-Verlag, 1972.
Molinari, H. H., Rozsa, A. J., and Kenshalo, D. R.: Cool sensitivity measured by signal detection
method. Percept. Psychophys., 19,246-251, 1976.
Nevin, J. A.: On differential stimulation and differential reinforcement. In, W. C. Stebbins (ed.):
Animal Psychophysics: the Design and Conduct of sensory experiments. New York, Appleton-
Century-Crofts, 1970.
Nevin, J. A., Olson, K., Mandell, C., and Yarensky, P.: Differential reinforcement and signal detec-
tion. J. Exp. Anal. Behav., 24,355-367, 1975.
Pavlov, I. P.: Conditioned Ref/exes. London, Oxford University Press, 1927.
Rachlin, H. C.: On the tautology of the matching law. J. Exp. Anal. Behav., 15,249-251, 1971.
Rachlin, H. C.: Contrast and matching. Psychol. Rev., 80,217-234, 1973.
Rachlin, H. C., and Baum, W. M.: Response rate as a function of amount of reinforcement for a
signalled concurrent response. J. Exp. Anal. Behav., 12, 11-16, 1969.
Rachlin, H. C., and Baum, W. M.: Effects of alternative reinforcement: Does the source matter? J.
Exp. Anal. Behav., 18,231-241, 1972.
Razran, G.: The observable unconscious and the inferable conscious in current Soviet psychophy-
siology: interoceptive conditioning, semantic conditioning, and the orienting reflex. Psychol.
Rev., 68,81-147, 1961.
Reichenbach, H.: The Rise of Scientific Philosophy. Berkeley, University of California Press, 1958.
Rosvold, H. N.: Physiological psychology. Annu. Rev. Psychol., 10,415-454, 1959.
Ryle, G.: The Concept of Mind. New York, University Paperbacks, 1949.
Sheldon, W. H.: The Varieties of Tempermanent: A Psychology of Constitutional Differences.
New York, Harper, 1942.
Sheldon, W. H.: Varieties of Delinquent Youth: an Introduction to Constitutional Psychiatry. New
York, Harper, 1949.
Shimp, C. P.: Perspectives on the behavioral unit: choice behavior in animals. In, W. K. Estes
(ed.): Handbook of Learning and Cognitive Processes. New York, Erlbaum Associates, 1975.
Sidman, M.: Tactics of Scientific Research: Evaluating Experimental Data in Psychology. New
York, Basic Books, 1960.
Skinner, B. F.: The Behavior of Organisms. New York, Appleton-Century, 1938.
Skinner, B. F.: Are theories of learning necessary? Psychol. Rev., 57, 193-216, 1950.
Skinner, B. F.: Behaviorism at Fifty. Science, 140,951-958, 1963.
Skinner, B. F.: Operant behavior. In, W. K. Honig (ed.): Operant Behavior: Areas of Research
and Application. New York, Appleton-Century-Crofts, 1966.
Skinner, B. F.: The steep and thorny way to a science of behavior. Am. Psychol., 43,42-49,1975.
Different Views of Psychology 15

Sokolov, E. N.: Neuronal models and the orientin~ influence. In 1 M. A. B. Brazier (ed.): Centrq/
Nervous System and Behavior. New York, Josiah Macy, 1960.
Stevens, S. S.: Psychology and the science of science. Psychol. Bull., 36,221-263, 1939.
Stevens, S. S.: Mathematics, measurement, and psychophysics. In, S. S. Stevens (ed.): Handbook
of Experimental Psychology. New York, Wiley, 1951.
Swets, J. A.: Is there a sensory threshold? Science, 134, 168-177, 1%1.
Teuber, H. L.: Physiological psychology. Annu. Rev. Psychol., 6,267-2%, 1955.
Ukhtomsky, A. A.: Complete Works, vol. 5. Leningrad, State Publishing House, 1954.
Wallace, W. A.: Causality and Scientific Explanation, vol. 1. Ann Arbor, University of Michigan
Press, 1972.
Wallace, W. A.: Causality and Scientific Explanation, vol. 2. Ann Arbor, University of Michigan
Press, 1974.
Wittgenstein, L.: Tractatus Logico-Philosophicus. London, Routledge and Kegan Paul, 1922.
Woodworth, R. S., and Schlosberg, H.: Experimental Psychology. New York, Holt, Reinhart &
Winston, 1954.
Young, G. A.: Electrical activity of the dorsal hippocampus in rats operantly trained to lever press
and to lick. J. Compo Physiol. Psychol., 90, 78-90, 1976.
SECTION I

CONCURRENT CLASSICAL
AND OPERANT
CONDITIONING PROCEDURES
Chapter 1

Review of Classical-Operant Conditioning,

Parameter by Parameter

Wendon W. Henton

History: A Confluence of Three Traditions

The experiments described in this section are concerned with a behavioral anal-
ysis of the various permutations and combinations of classical and operant con-
ditioning schedules. For the present purposes, a schedule describes the interre-
lationship between three primary events: environmental stimuli, responses,
and reinforcer delivery. An operant conditioning schedule describes the stimuli
in which a reinforcing event is delivered following and contingent upon the oc-
currence of the recorded response (Ferster and Skinner, 1957). Dependent vari-
ables in operant conditioning are the rate, duration, force, and latency of the
recorded response (Skinner, 1950; Premack, 1965). A classical conditioning
procedure describes the conditional relationship between an environmental
stimulus (conditioned stimulus, eS) and the subsequent occurrence of an un-
conditionally reinforcing stimulus (unconditioned stimulus, UeS) that reliably
elicits a recorded response (unconditioned response, UeR) (Pavlov, 1927). The
frequency, duration, and latency of responses elicited by the es, not the ues,
are the primary dependent variables in classical conditioning schedules.
The theoretical and methodological issues involved in defining stimuli, re-
sponses, and reinforcers, and the still more subtle distinctions between classi-
cal and operant conditioning schedules, have been systematically reviewed by
a variety of investigators (Miller and Konorski, 1928; Skinner, 1938; Mowrer,
1960; and especially volume two of Psychology: A Study of Science, edited by
S. Koch, 1959*). One interpretative issue, for example, is the definition of a
* The following definitions are used throughout Part I. A response is defined as any environ-
mental change initiated by the organism, either an overt response within the external environment
or a covert response within the internal physiological environment. In operant schedules, rein-
forcement refers to the response-contingent delivery of a reinforcer. A reinforcer is any contingent
(con't.)
20 Section I: Concurrent Classical and Operant Conditioning

classical conditioning ues as a stimulus that reliably elicits a dominant reflex-

ive response (Pavlov, 1927), any reflexive response (Konorski, 1967), or any
measurable response (Stolurow, 1973). The divergent definitions of ues have
somewhat subtle yet pervasive consequences for the distinctions between clas-
sical and operant conditioning procedures (Kimble, 1961; Kling and Riggs,
1971). However, in spite of conceptual and linguistic issues, traditional opera-
tional distinctions between classical and operant conditioning are rather funda-
mental. For classical conditioning, the primary characteristic is that the rein-
forcing event is delivered in a specified temporal relationship to the es,
independent of all responses. In contrast, reinforcer delivery is explicitly de-
pendent upon the occurrence of a response in operant conditioning. Both clas-
sical and operant conditioning schedules may be simple arrangements between
one response, one reinforcer, and one stimulus, or complex programs in which
a variety of responses, reinforcers, and stimuli are combined in simultaneous or
successive components (Pavlov, 1927; Ferster and Skinner, 1957; Findley,
1962). The classical and operant schedule combinations described in this sec-
tion, for example, are concurrent schedules in which the subject is simulta-
neously conditioned on (1) a simple classical conditioning schedule and (2) a
simple operant conditioning schedule (Henton and Brady, 1970; Brady, 1971;
Henton, 1972).

2 III Il~I\\III\1 \ { \IIIUtu

3

4----------------\~---
Figure 1.1. Suppression of behavior during Estes-Skinner procedure. Trace 1: delivery of operant
reinforcer on a VI 2-min schedule. Trace 2: operant key-pecking responses by pigeons. Trace 3:
duration of Pavlovian CS. Trace 4: delivery of shock UCS.

event that alters the subsequent probability of the specified response. In classical conditioning, the
unconditioned stimulus (UCS) is any stimulus that reliably and consistently elicits a specified re-
sponse (UCR). The conditioned stimulus (CS) is any stimulus that does not initially elicit the speci-
fied UCR but may nevertheless elicit other specific responses. The conditioned response (CR) is a
new or acquired behavior during the es as a result of pairing with the ues. Reinforcement in clas-
sical conditioning refers to the presentation ofthe UCS contingent upon the occurrence of the es.
Thus, in classical conditioning, reinforcement is stimulus contingent and response independent,
and, in operant conditioning, reinforcement is specifically response contingent. One common fea-
ture, however, is that positive and negative operant reinforcers may also be arranged to serve as
positive or negative UCS in classical appetitive and defensive conditioning, respectively.
Chapter I: Review of Classical-Operant Conditioning 21

The explicit history of classical-operant combinations is relatively brief,

dating back to an experiment by Estes and Skinner (1941). This report by Estes
and Skinner described the rate of lever pressing reinforced with response-con-
tingent food (operant schedule) during the superimposed presentation of a stim-
ulus terminated by response-independent shock (classical conditioning sched-
ule; Figure 1.1). This recent history, however, is firmly imbedded and
intermixed within several scientific and philosophical traditions. The conflu-
ence of different traditions has now resulted in an occasionally bewildering va-
riety of opinions about classical and operant interactions. However, the writing
and verbal behavior of psychologists, like the behavior of all other organisms,
may be profitably examined within the context of previous and current training
histories (Skinner, 1953). Current formulations of classical and operant sched-
ule combinations might then be viewed as various blends of at least three tradi-
tional considerations within psychology-the analysis of emotions, classical
conditioning, and the elimination or disruption of behavior.

Emotional States
"Anxiety has at least two defining characteristics; (1) it is an emotional state,
somewhat resembling fear . . . " (Estes and Skinner, 1941). Within a fraction
of the introductory first sentence, Estes and Skinner acknowledged the status
of emotional states, anxiety, and fear. The universal and ageless study of emo-
tions continues to be one of the -dominant influences controlling current inter-
pretations of classical-operant schedule effects.
The philosophical interest in feelings and emotions was initially translated
into behavioral measures at least as early as the introduction of Darwinian evo-
lution. The Expression of the Emotions in Man and Animals (Darwin, 1872)
proposed that emotions could be inferred and measured by overt responses
across a broad spectrum of organisms. Baring of teeth and curling of lips in man
as well as animals were inborn, instinctive, and perhaps archetypical measures
of emotions. The analysis of feelings and emotions, as described by Boring
(1957), was also a major element within the psychological system of Wundt.
Wundt's "conscious contents" and "tridimensional theory of feelings" were
both tested and supported by correlations between bodily activity and the three
continuums of pleasantness-unpleasantness, excitement-calm, and strain-re-
laxation.
The interpretation of emotions was profoundly influenced by the simultane-
ous pUblications in the early 1880s by James in America and Lange in Den-
mark. James (1884) attempted a logical simplification by proposing that a spe-
cial and separate center of emotions did not exist within the brain; rather,
emotions corresponded to various combinations of ordinary processes in the
sensory and motor centers, and therefore did not require the assumption of spe-
cial seats and centers. James proposed that the body was a "sort of sounding-
board," with emotions the perception of changes in ". . . the bladder and
bowels, the glands of the mouth, throat, and skin, and the liver . . . and the
continuous co-operation of the voluntary muscles." James further distin-
22 Section I: Concurrent Classical and Operant Conditioning

guished between the primary "emotions" of rage, fear, and anger, and their
pale copies of moral, intellectual, and aesthetic "feelings," in which the bodily
sounding board was mute. Overt responses and bodily changes were thus ex-
pressions of emotions, but not expressions of feelings (James, 1890). In agree-
ment with James, Lange (1885) suggested that emotions included "vasomotor
disturbances, varied dilations of the blood vessels, and consequent excess of
blood, in the separate organs." In contrast to James, however, Lange proposed
that subjective sensations were indirect, secondary disturbances caused by the
cardiovascular changes. Thus, as noted by Wenger (1950), the "James-
Lange" theory is a historical misnomer: for James, emotions were the percep-
tion of bodily changes; for Lange, emotions were the bodily changes.
The neurophysiology and emotional interpretations of James and Lange pro-
moted harsh criticism, and an alternative theory, by Cannon (1914, 1927, 1931;
detailed critique by Fehr and Stem, 1970). In reply to James, Cannon (1927,
1931) proposed that (1) the latency of visceral responses was too slow to ac-
count for emotions, (2) experimental induction of peripheral changes does not
elicit emotions, (3) the viscera are relatively insensitive, (4) the same visceral
changes are found in all emotions, and (5) the elimination of peripheral feed-
back does not alter emotional behavior (also see Sherrington, 1900; Wood-
worth and Sherrington, 1904; Cannon et aI, 1927). Cannon (1927) instead pro-
posed that afferent stimulation from sense organs reaching the thalamus
simultaneously (1) elicit changes in the musculature and viscera, and (2) project
to the cortex, which in tum releases the thalamus from cortical inhibition.
Emotional experience was primarily described as the cortical experience of the
thalamic activation.
Following Cannon, and Woodworth and Sherrington, neurophysiological
speculations have related virtually every aspect of the forebrain and brain stem
to emotional behaviors (review by Brady, 1970a, 1970b). For example, extirpa-
tion and electric stimulation experiments suggest that anterior and dorsomedial
thalamic nuclei may be related to anxiety, tension, emotional "alerting" re-
sponses, and positive reinforcement (Spiegal and Wycis, 1949; Baird et al,
1951; aIds, 1955, 1956). Other brain tissues implicated in emotional behavior
include hypothalamic nuclei (Bard, 1928, 1934; Bard and Rioch, 1937; Rioch,
1938; Bard and Mountcastle, 1948), the limbic system (i.e., the hippocampus,
fornix, mammalary bodies of the hypothalamus, anterior thalamic nuclei, and
amygdala) (Papez, 1937; Kluver and Bucy, 1937, 1938; MacLean, 1949, 1952)
and the brainstem reticular system (Duffy, 1957; Hebb, 1949; Lindsley, 1950,
1957; Malmo, 1957, 1962). Attempts to relate overt behaviors to underlying
neurophysiological events have been described in detail by Arnold (1950,
1970), Lindsley (1951), and Miller et al. (1960), among many others (reviews by
Grossman, 1967; Young, 1943, 1973).

Classical Conditioning
"Anxiety is here defined as an emotional state arising in response to some cur-
rent stimulus which in the past has been followed by a disturbing stimulus"
Chapter 1: Review of Classical-Operant Conditioning 23

(Estes and Skinner, 1941). The second characteristic defining anxiety offered
by Estes and Skinner was that the emotional state is established by classical
conditioning, but with the added proviso that the reaction to the es is not nec-
essarily the same reaction elicited by the ues.
The association of stimulus pairs, of course, was a major issue throughout
the history of philosophy (Boring, 1957). The British philosophers especially
analyzed successive and simultaneous associations of objects, sensations, and
images in terms of the still current laws of contiguity, similarity, frequency, etc.
The experimental analysis of spinal reflexes and classical conditioning by Sher-
rington (1906), Pavlov (1927) and Sechenov (1935) subsequently provided ex-
perimental psychology with a ready behavioral analogy for the study of' 'higher
mental processes."
Watson and Morgan (1917) were among the first theorists to suggest a classi-
cal conditioning basis for affective behaviors. Watson and Morgan proposed
that the gamut of emotions observed in adults are classical conditioned reflexes
based upon three inborn or unconditioned emotional reaction patterns of fear,
rage, and love. Watson and Raynor (1920) subsequently reported the first ex-
perimental analysis of emotional behavior, based upon the classical condition-
ing or associationistic model. A white rat was shown to a 1-year-old boy, and
reaching toward the rat was paired with striking a steel bar immediately behind
the subject's head. After seven trials, presentation of the white rat alone eli-
cited crying and crawling rapidly away. Generalization tests with a rabbit, dog,
fur coat, cotton, wool, etc, elicited similar "fear" reactions. Watson and co-
workers concluded that reflex factors and classical conditioning thus formed
the basis of learned emotional reactions. (Note, however, that the "ues" of
striking the steel bar was an operant punishment procedure and response-con-
tingent, rather than the response-independent DeS of the classical conditioning
model.)
One of the first therapeutic treatments of emotions was a behavioral tech-
nique also based upon the classical conditioning formulation. Jones (1924a,
1924b) described a series of experiments in which stimuli eliciting emotional
responses were carefully paired with eating maintained at a relatively constant
rate by bits of candy. In impressive anticipation of subsequent procedures and
interpretations, Jones reported that (1) the rate of eating could be disrupted by
the responses elicited by the fear stimulus, or, with slightly different proce-
dures, (2) the elicited fear responses could be reduced by eating. Jones (1924a)
suggested that both the deleterious and beneficial effects were due to an inter-
action between two competing response systems controlled by fear objects and
craving objects.
Pavlov (1927) also described a series of classical conditioning procedures for
the study of affective disturbances and pathological states. In each paradigm, a
simple conditioning procedure was made progressively more difficult over suc-
cessive trials until "the animal became quite crazy, unceasingly and violently
moving all parts of its body, howling, barking, and squealing intolerably" (Pav-
lov, 1927, p. 294). The emotional behaviors were produced by (1) gradually in-
creasing the similarity between a circle paired with food and an elipse not
24 Section I: Concurrent Classical and Operant Conditioning

paired with food, (2) systematically increasing the interval between CS and
DCS presentation for each of six different stimuli paired with food, or (3) pro-
gressively altering the locus and intensity of a weak electric current paired with
food. The salivation initially conditioned to the CS was replaced by the violent
emotional reactions in each procedure. The more or less prolonged pathologi-
cal states were thought to result from the gradual alteration of the relative bal-
ance between inhibition and excitation past a critical adjustment value.
Classical conditioned fear and acquired drives continue to be central ele-
ments in current definitions of emotional concepts. The theoretical importance
of conditioned emotions is perhaps best exemplified by the continual analysis
and reanalysis of emotional processes in motivation and reinforcement de-
scribed by Mowrer (1939, 1940, 1950, 1960). More recently, systematic posi-
tions emphasizing the role of classical conditioning in the acquisition and main-
tenance of behavior have been elaborated by many authors (Miller, 1948, 1951,
1959; Brown and Farber, 1951; Brown et aI., 1951; Solomon and Wynne, 1953,
1954; Rescorla and Solomon, 1967; McAllister and McAllister, 1971). More-
over, experiments by Wolpe (1952, 1958) and Masserman (1943) have gen-
erated treatments of emotions based upon antagonistic classical conditioning
responses and processes (Wolpe, 1962; review by Wilson and Davison, 1971).

Classical-Operant Interactions
Estes and Skinner (1941) introduced a third and perhaps predominant influence
affecting interpretations of classically conditioned emotions. The unique con-
tribution of the Estes-Skinner experiment, and the primary subject of the pres-
ent chapter, is the alteration of operant behaviors during the superimposed pre-
sentation of a classical conditioning procedure. After discussing conditioning
and the concept of emotional states, Estes and Skinner proposed that
A stimulus giving rise to "fear," for example, may lead to muscular reactions (includ-
ing facial expressions, startle, and so on) and a widespread autonomic reaction com-
monly emphasized in the study of emotion; but of greater importance in certain re-
spects is the considerable change in the tendencies of the organism to react in various
other ways. Some responses in its current repertoire will be strengthened, others
weakened, in varying degrees. Our concern is most often with anxiety observed in
this way, as an effect upon the normal behavior of the organism. (Estes and Skinner,
1941, pp. 390-391)

The experiments by Estes and Skinner were then designed to examine this
disruptive effect of classically conditioned emotional states. More specifically,
responses by rats were regularly reinforced by the response-contingent deliv-
ery of food every 4 min. The characteristic rate and pattern of the ongoing
operant responses were decreased and disrupted throughout the presentation
ofa 3-min tone (CS) terminated by electric shock (DCS; Figure 1.1). In a sec-
ond experiment, the operant lever-pressing response was no longer reinforced
with food (operant extinction), and the gradually extinguishing lever-pressing
rate was similarly disrupted during the tone-shock pairings. In a final experi-
Chapter 1: Review of Classical-Operant Conditioning 25

mental manipulation, lever pressing was once again reinforced with food, but
the tone was continuously presented alone and unpaired with shock (extinction
of classical conditioning). The ongoing operant response was initially disrupted
at tone onset, but progressively increased to the normal rate during the remain-
ing portion of the session.
This experimental disruption and manipulation of behavior reported by
Estes and Skinner has developed into a major field of experimental research
over the past three decades. Undoubtedly, one of the most comprehensive em-
pirical and conceptual analyses of such classical-operant interactions has
been described by Brady and associates (Brady and Hunt, 1955; Brady, 1971,
1975). Extensive programs in a number of other laboratories have also system-
atically examined the behavioral changes during CS-shock pairings (e.g.
Kamin, 1961, 1965; Hoffman and FleshIer, 1962, 1964; Lyon, 1963, 1967;
Church, 1969). Until quite recently, the analysis of classical-operant combina-
tions has been primarily concentrated on negative classical conditioning proce-
dures superimposed upon positively reinforced operant responses (reviews by
Davis, 1968; Lyon, 1968). The consistency of the experimental analysis is
rather strikingly confirmed by the extraordinary control of operant rates ob-
tained with the careful manipulation of classical and operant schedule parame-
ters (Hendricks, 1966; reviews by Rosenberger, 1970; Smith, 1970).
Following the reviews of Davis and Lyon, a variety of experiments have fur-
ther elaborated the complex interactions between the multitude of variables
controlling operant rates during superimposed CS-shock pairings. The success
of the experimental analysis of "conditioned emotional responses" (Brady and
Hunt, 1955), and influential papers by Rescorla and Solomon (1967), Kamin
(1969), and others, have generated considerable experimental interest in the
analysis of all possible combinations of classical and operant conditioning pro-
cedures. The four-fold analysis of positive and negative classical conditioning
procedures superimposed upon positive and negative operant conditioning pro-
cedures forms the basis of the present review.

Theoretical Positions Briefly Considered

A rich variety of inductive and deductive theoretical formulations have fre-
quently affected experimental research in classical-operant conditioning
schedules. A brief consideration ofthe more prevalent theories would therefore
be appropriate if somewhat tangential to the laboratory analysis of concurrent
classical-operant schedules.
The first systematic formulation was proposed by Brady and Hunt (1955).
After reviewing 54 experimental reports, Brady and Hunt offered an inductive
or "experimental approach to the analysis of emotional behavior. " The experi-
mental approach was, in effect, a formalization of Estes and Skinner's previous
suggestion that some responses in the subject's repertoire will be strengthened
and other responses weakened during classical-operant combinations. A broad
program for the analysis of interacting response systems was based upon the
26 Section I: Concurrent Classical and Operant Conditioning

interrelationships between internal and external respondents, controlled by the

classical conditioning schedule, and internal and external operants, controlled
by the operant conditioning schedule (Brady, 1970a, 1971). The "competing re-
sponse" interpretation has focused upon the variety of interacting yet separa-
ble response systems involved in the conditioned emotional response para-
digm, including endocrine, autonomic, and cardiovascular processes and the
more peripheral muscular systems. For example, changes in baseline operant
response rates may be related to increased frequencies of incompatible behav-
iors, such as crouching and "freezing" elicited by the eS-shock pairings.
Heart rate and cardiovascular responses may also be altered, which in tum
could function as additional discriminative stimuli maintaining the suppressed
operant rate (Brady, 1975). Feelings and emotions were then used as generic
terms summarizing internal and external responses and their appropriate con-
trolling variables. Moreover, the complex interrelationships between response
systems are viewed as schedule dependent, with a variety of response interac-
tions determined by specific experimental parameters, behavioral histories,
and response topographies (Brady, 1971, 1975).
Kamin (1961,1963,1965) proposed a similar incompatible response interpre-
tation, although he emphasized the difficulty of selecting the necessary and/or
sufficient competing responses from the diffuse array of potentially incompati-
ble behaviors. Kamin and co-workers have instead suggested that the alteration
of operant rates may be an indirect quantification of the superimposed classical
conditioning processes. An extensive series of experiments using delayed,
trace, and backward conditioning procedures have become standard references
in the classical-operant conditioning literature (Annau and Kamin, 1961;
Brimer and Kamin, 1963; Kamin and Brimer, 1963; Kamin and Schaub, 1963;
Kamin, 1965). The experiments using different intensities and sequences of
conditioned and unconditioned stimuli do suggest that operant rates may be
sensitive to classical conditioned extinction, inhibition, and disinhibition proce-
dures. More recently, Kamin (1968, 1969) and Kamin and Gaioni (1974) have
proposed a more cognitive interpretation which suggests that the effectiveness
of classical conditioning may vary directly with the "surprising" nature of the
ues. A ues preceded by a well-established es may be more expected, and
therefore less disruptive, than a similar ues preceded by a not as yet condi-
tioned es. Operant rates during a superimposed es would then be dependent
upon the predictability or surprising characteristics of ues onset.
A "conditioned suppression" analysis that eschews emotional inferences
has been proposed by Lyon (1963. 1968). In the conditioned suppression for-
mulation, the operant rate changes are not offered as measures of emotions but
simply accepted as functional or empirical effects. The changes in operant re-
sponding during the es may be partially due to competing responses of crouch-
ing and freezing (interference hypothesis), or to punishment of the operant re-
sponse sequence by the shock ues (punishment hypothesis). [Weiskrantz
(1968) also suggested that the response-independent ues may in fact function
as a response-contingent operant reinforcer, differentially maintaining prepara-
tory or postural responses that reduce the aversive effects of electric shock.]
Chapter 1: Review of Classical-Operant Conditioning 27

Lyon (1968) thus concluded that the conditioned emotional response paradigm
is an aversive control technique that may share nonemotional effects with other
aversive control procedures.
An elegant four-decade series of experiments by Konorski (1967) includes
a comprehensive if provocative formulation of the interactions between classi-
cal and operant conditioning procedures. For Konorski, classical conditioning
involves two distinct sets of unconditioned responses, and therefore simulta-
neously generates two sets of conditioned responses during the es. Preparatory
responses are conditioned to all situational cues, yielding drive-conditioned re-
sponses, and consummatory responses are conditioned to the es, yielding con-
ditioned consummatory responses. (A more detailed description of Konorski's
analysis will be presented in Part III.) In appetitive conditioning, the drive
responses are inhibited by consummatory response systems. In a somewhat
analogous fashion, instrumental responses are suppressed during a superim-
posed stimulus paired with food, with a concomitant increase in consummatory
salivation and approach to the ues food cup. Konorski further proposed that
the disruptive effects did not reflect an interaction between inhibitory and ex-
citatory processes, but an interaction between two excitatory processes con-
ditioned to mutually antagonistic centers. This, the disruption of all motor acts,
such as the suppression of baseline operant responses, occurs by the excitation
of groups of neurons eliciting antagonistic motor acts (Konorski, 1972).
An interacting or mediational state interpretation of classical-operant
schedules has been offered by Rescorla and Solomon (1967). They proposed
that classical conditioning establishes an internal emotional state that may in-
teract with a motivational state maintaining the operant response through a
shared mediator state. Thus, changes in operant response rates could be in-
direct but potentially sensitive measures of classical conditioned emotions. In
contrast, the peripheral classical conditioned responses such as salivation and
cardiac changes are thought to be only imprecise indicators of the emotional
states and do not directly mediate or interfere with operant responses. Rescorla
and Solomon then concluded that the concurrent measurement of both operant
and classical conditioned responses is not merely inefficient but irrelevant to
the analysis of classical-operant interactions. Rescorla and Wagner (1972) pro-
posed a revised model, in which the strength or salience of a es is dependent
upon the total associative ~trength of all es elements. The individual elements
are collectively rather than independently conditioned to an asymptotic value
determined by the strength of the ues. A well-established es paired with
shock will therefore acquire near asymptotic strength and prevent conditioning
of any additional stimuli that are simultaneously paired with the same ues. An
asymptotic conditioned emotional state might then block or prevent the condi-
tioning of a similar associative state (Rescorla and Wagner, 1972) but increment
or facilitate an aversive operant motivational state (Rescorla and Solomon,
1967; also see Wagner, 1969; Rescorla, 1970, 1971a, 1971b, 1973).
Hoffman and associates (Hoffman and FleshIer, 1961, 1964, 1965; Hoffman
et aI, 1963) have also interpretated operant suppression as reflecting emotional
reactions elicited by the aversive classical conditioning procedure. Hoffman
28 Section I: Concurrent Classical and Operant Conditioning

and co-workers have further argued that the disruptive effects of Pavlovian
conditioning cannot be due to an increased frequency of behaviors incompati-
ble with the ongoing operant response; rather, Hoffman and Barrett (1971) and
Stein et al. (1971) suggested that "freezing" during the CS is not a specific con-
ditioned response, but the total inhibition of all overt responses caused by in-
ternal emotional states. Although the details of the inhibitory processes are un-
specified, the emotion-induced behavioral inhibitions are consistent with the
use of operant changes to measure or index internal fear reactions.
A general motivational theory of punishment by Estes (1969) also includes
an interpretation of classical-operant schedule interactions. The Estes inter-
pretation, influenced by stochastic mathematical models (Estes, 1959), sug-
gests that the probability of an instrumental response is proportional to the
facilitative input to amplifier drive elements. Stimuli preceding either response-
dependent or response-independent shock acquire the capacity to inhibit the
amplifier elements maintaining the baseline operant response. This inhibition
in Estes' system is primarily an interaction between drive elements, not direct
peripheral response interactions. The conditioned suppression of overt re-
sponses is therefore said to be mediated by CS-induced reductions in the op-
erant motivational state.
A "general emotional state" has been described by Azrin and Hake (1969) to
account for operant changes during stimuli paired with either positive or nega-
tive reinforcers. The general emotional state theory argues that operant re-
sponding will be suppressed during a stimulus paired with any strong positive
or negative reinforcer. Azrin and Hake proposed that the operant changes may
be related to a state of "heightened preparedness" and are closely associated
with autonomic and cardiac changes, with both covert and overt responses a
product of the same underlying emotional state. The general emotional state
interpretation may also be a reinforcer value theory (Hake and Powell, 1970)
that proposes that a Pavlovian UCS will reduce the value of the operant rein-
forcer during the CS, and operant response rates will in turn decrease to match
the new reinforcer value, as predicted by the response-reinforcer matching in-
terpretation of Catania (1966) and Herrnstein (1970).
The competing response interpretation of Brady and Hunt (1955) was re-
cently extended to a more general concurrent response-concurrent schedules
analysis by Henton and Brady (1970) and Henton (1972). The concurrent re-
sponse analysis suggested that classical conditioning procedures superimposed
upon operant baselines define concurrent schedules in which a classical condi-
tioning schedule and an operant conditioning schedule are programmed simul-
taneously. In consequence, the conditioned emotional response paradigm may
be one of many concurrent schedules that simultaneously control two sets of
responses. The pattern of responses controlled by one conditioning schedule
may then be dependent upon the concurrent pattern of responses controlled by
the second conditioning component. The analysis suggested that the classical
conditioned response rate may be disrupted by the operant schedule, in much
the same fashion that operant response rates are disrupted by the classical con-
Chapter 1: Review of Classical-Operant Conditioning 29

ditioning procedure. Facilitative or suppressive interactions between the con-

current operants and respondents may then be controlled by manipulating the
stimulus, response, or reinforcement contingencies of each schedule compo-
nent (Henton, 1972; Henton and Iversen, 1973; similar interpretations have
been independently reported by Lo Lordo et al., 1974; Schwartz, 1976; and
Schwartz and Gamzu, 1977).
In summary, the heterogeneity of theoretical formulations over the past 30
years has generated widely divergent research strategies and treatments of
classical-operant schedule interactions. The formulations range from inductive
and experimental approaches (Lyon, 1968; Brady, 1970a, 1971; Henton and
Brady, 1970) to inferential and deductive theoretical models (Rescorla and Sol-
omon, 1967; Estes, 1%9; Kamin and Gaioni, 1974). In each system, the altera-
tion of peripheral operant responses may be emotional behavior (Brady, 1975)
or indirect indicants of emotional states (Kamin, 1965; Hoffman and Barrett,
1971) or may be partially determined by nonemotional effects (Lyon, 1968).
Similarly, freezing may be a classical CR (Brady, 1975), the total inhibition of
all responses (Stein et aI., 1971), irrelevant responses (Rescorla and Solomon,
1967), or concurrent interacting responses (Henton, 1972). Furthermore, the
disruption of operant responses may be due to general emotional states (Azrin
and Hake, 1969) or mediational states (Rescorla and Solomon, 1967) shared by
the operant and respondent systems, to excitatory interactions between vari-
ous brain centers eliciting antagonistic responses (Konorski, 1967), or to inhibi-
tory interactions between drive elements (Estes, 1969). Finally, positive and
negative classical conditioning superimposed upon operant baselines may have
identical suppressive effects (Azrin and Hake, 1%9), differential facilitative
and suppressive effects independent of peripheral responses (Rescorla and Sol-
omon, 1%7), or differential effects conditional upon response topographies
(Brady, 1971). The set of theoretical formulations thus contains individual ele-
ments that overlap to greater and lesser degrees, are frequently mutually con-
tradictory, and are occasionally internally inconsistent.

Experimental Analysis: General Methods

The following review is an organization and description of the empirical param-
eters that are used to control response rates during classical-operant sched-
ules. Given the extraordinary number of studies using the Estes-Skinner pro-
cedure, many experiments have surely been overlooked-hopefully without
prejudice. The research in each section is arranged in chronological order, with
at least one and frequently several experiments described in some detail. Refer-
ence to other papers has been simply based on a compromise between compre-
hensive coverage and space limitations. For a more objective presentation, the
theoretical opinions of the authors have generally been deleted, except when
used to augment the historical setting.
30 Section I: Concurrent Classical and Operant Conditioning

Subjects
A strikingly odd variety of species have been successfully used in the Estes-
Skinner procedure, ranging from albino rats (Rosenberger and Ernst, 1971) to
vampire bats (Shumake et aI., 1977). A more complete listing would include
cats (Brady and Conrad, 1960), dogs (Waller and Waller, 1963), fish (Wilson et
aI., 1970), gerbils (Frey et aI., 1972), guinea pigs (Valenstein, 1959), mice (An-
derson and Ressler, 1973), monkeys (Sidman, 1960b), and pigeons (Stein et aI.,
1971). In addition, sensory psychophysical studies have compared suppression
to auditory stimuli in the bushbaby, hedgehog, opposum, potto, slow loris, and
tree shrew (Masterton et al., 1969; Heffner and Masterton, 1970). The previous
experiments unanimously suggest a widespread generality of conditioned sup-
pression across all species, with the singular exception of the subject of pri-
mary concern, homo sapiens (Sachs and May, 1967, 1969; Rand et aI., 1971;
Sachs and Keller, 1972; Skinner, 1974; but also see Di Giusto et aI., 1974).

Conditioning Procedures
Three general types of training procedures have been used; the most prevalent
technique is an "on-the-baseline" procedure. In this procedure, the operant re-
sponse is first trained to a stable rate, with all CS-UCS pairings subsequently
superimposed upon the already established operant baseline (e.g., Lyon and
Felton, 1966b). A second technique is an "off-the-baseline" procedure, in
which the operant and classical conditioning are conducted in separate ses-
sions, and the two schedule components are subsequently combined into one
session (e.g., Geller et al., 1955). The off-the-baseline technique is used to pre-
vent unprogrammed or adventitious reinforcement of the recorded operant re-
sponse by the CS-UCS pairings. A third arrangement is similar to the off-the-
baseline procedure, with the exception that the previously established classical
and operant conditioning components are subsequently combined during
schedule transitions of (1) extinction of the operant conditioning schedule
(Estes and Skinner, 1941), (2) extinction of the classical conditioning schedule
(Leaf and Muller, 1965), or (3) simultaneous extinction of both the operant and
classical conditioning schedules (Rescorla, 1967a). The various extinction pro-
cedures are limited to a relatively brief analysis of a small number of sessions
prior to complete extinction.

Dependent Variables
The "suppression ratio" remains the most popular dependent variable. The
ratio basically compares the operant response rate during the CS with the re-
sponse rate during a control period in the absence of the CS. The control re-
sponse rate is usually, although arbitrarily, recorded during the time period im-
mediately preceding CS onset. The suppression ratio may be calculated with a
variety of different mathematical formulas that differentially weight the CS and
control response rates. Figure 1.2 presents a comparison of the various sup-
Chapter 1: Review of Classical-Operant Conditioning 31

2.0 .1
•
•••
•••
••
•

...o
~

0.0

-1.0

o 60 120
Responses (1 min CS)
Figure 1.2. Comparison of different formulas describing the change in operant responding during
superimposed Pavlovian CS. In this graph baseline responding during control periods is arbitrarily
defined as 60 responses/min (arrow). 1. CS response rate/control response rate (Stein et al., 1958).
2. CS rate - control rate/control rate (Hunt et al., 1952).3. CS rate/CS rate + control rate (Annau
and Kamin, 1961). 4. Control rate/CS rate + control rate (Goldstein, 1966). 5. Control
rate - CS rate/control rate (Hoffman and FleshIer, 1%1).

pression ratios as a function of changes in the operant rate during the CS. For
each ratio, the duration of the CS and control periods is 1 min, .with the re-
sponse rate fixed at 60 responses/min during the control period. The calculated
ratio for a 100% decrease in responding during the CS (relative to control rates)
ranges from - 1.00 to + 1.00; for a 100% increase in relative CS rates, the range
is - 1.00 to + 2.00. Formulas 1, 2, and 5 yield linear suppression ratios, with
equivalent positive and negative changes in CS response rates producing sym-
metrical alterations in the ratio. Formulas 3 and 4 are curvilinear, negatively
accelerated functions, with equivalent positive and negative CS rate changes
producing progressively more asymmetrical alterations in the suppression
ratio. Formulas 2 and 5 seem to be the most logical, with linear functions sym-
metrical about a 0.00 suppression ratio for unchanged CS rates. As previously
32 Section I: Concurrent Classical and Operant Conditioning

described by Lyon (1968), the logic of formulas 3 and 4 is not especially clear
(review of suppression measures by Shimoff, 1972a).
A recovery time measure (not shown in Figure 1.2) developed by Leaf and
Muller (1965) and Leaf and Leaf (1966) is becoming increasingly popular (Carl-
ton and Vogel, 1%7; Hughes, 1969). The recovery time procedure is a one-trial
classical conditioning extinction test in which a previously conditioned CS is
superimposed upon licking responses reinforced with water, sucrose, or con-
densed milk. The dependent recovery time measure is simply the time to com-
plete 10 licking responses during the CS test trial.
The responses controlled by the classical conditioning procedure are gen-
erally unrecorded during classical-operant combinations. Preliminary reports
of freezing and immobility during conditioned suppression were reported by
Brady and Hunt (1955) and Brady and Conrad (1960), and more recently by
Hoffman and Barrett (1971) and Stein et al. (1971). Heart rate, systolic and
diastolic blood pressure, blood flow, and electromyographic activity have also
been reported (Stebbins and Smith, 1964; de Toledo and Black, 1966; Brady et
al., 1969, 1970). Alternatively, observational reports by various authors suggest
a total absence of behavior, or a failure of classical conditioning to elicit any
overt responses when prefood or preshock stimuli are superimposed upon
operant schedules (Azrin and Hake, 1969; Seligman et al., 1971). The tradi-
tional failure to record the responses elicited by the classical conditioning
schedule is especially surprising in view of the heavy explanatory role attrib-
uted to classical conditioning processes in the conditioned emotional response
procedure. At minimum, respondents should be more direct and at least as sen-
sitive as concurrent operant responses as classical conditioned dependent vari-
ables. Moreover, given the well-documented sensitivity of conditioned responses
to seemingly trivial and innocuous procedural changes (Pavlov, 1927; Konorski,
1967), no cogent reasoning would suggest any less sensitivity to the effects of
superimposed operant conditioning schedules. To invert a previous interpreta-
tion by Rescorla and Solomon (1967), any empirical law of operant conditioning
may have profound implications for the control of Pavlovian responses when
the two procedures are interactively combined. (See, for example, Experiment
IX, Chapter 2.)

Negative Classical Conditioning Scheduled

with Positive Operant Conditioning
Negative classical conditioning procedures superimposed upon positive
operant reinforcement schedules is by far the most studied classical-operant
combination. Conditioned suppression of operant responses during stimuli
paired with electric shock is in fact frequently accepted as the prototypical
model of all classical-operant effects. The independent variables fall rather
naturally into three groups: classical conditioning variables, operant condition-
ing variables, and subject variables (Table 1.1). Guided by theory and tradition,
the studies generally emphasize the "emotional" effects of the classical condi-
Chapter 1: Review of Classical-Operant Conditioning 33

Table 1.1 Negative classical-positive operant conditioning

Variables Studies Results

Classical Conditioning
UCS Parameters
intensity 15 suppression as a direct function
duration 1 suppression as a direct function
probability 18 divergent results
random presentation 14 suppression, conditional upon schedules
CS Parameters
duration 15 suppression as an inverse function
intensity 33 conditional upon other variables
generalization 12 graded suppression
preexposure 16 attenuation of suppression
sensory preconditioning 8 suppression during preconditioned CS
second-order stimuli 8 transient suppression
compound stimuli
summation tests 7 increased suppression
inhibition tests 23 usually attenuation of suppression
backward conditioning 9 divergent results
spontaneous recovery and retention 7 increased suppression, maintenance of
suppression over years
Total 186

Operant Conditioning
Reinforcer Parameters
type of reinforcer 8 conditional upon reinforcer
magnitude of reinforcer 4 divergent results
Reinforcement Schedules
response rate measures 12 function of reinforcer proximity
response accuracy 9 generally unchanged during CS
operant vs. consummatory 2 differentially suppressed
responses
Conditioning History 7 suppression conditional upon history
Adventitious Punishment 13 increases suppression
by UCS
Operant Sd compared to CS 5 generally equivalent effects
Concurrent Behaviors 7 inversely related to operant rate
Pre-schedule change stimuli 11 divergent results
Total 78

Subject
Age 11 conditional upon procedures
Central Nervous System
lesions 16 divergent results
stimulation 14 divergent results
electroconvulsive shock 17 transient attenuation of suppression
drugs 45 divergent and conditional effects
Cardiovascular Responses 15 no close correlation with behavior
Total 118
34 Section I: Concurrent Classical and Operant Conditioning

tioning procedure upon the operant responses, rather than the effects of the
operant procedure upon the classical conditioned "emotions." Consequently,
the procedures are most frequently selected to optimize the classical condition-
ing variables relative to the operant conditioning parameters. The procedural
bias has therefore resulted in a far more thorough analysis of the variables
within the classical conditioning component. In addition, the "conditioned
emotional response" (CER) paradigm has generated numerous experiments on
the effects of drugs and other subject variables. As summarized in Table 1.1,
experimental results suggest consistent behavioral changes as a function of the
more powerful independent variables, with the effects of many variables more
markedly conditional upon other parameters, or seemingly inconsistent and di-
vergent within and across experiments.

Classical Conditioning Variables

UCS Parameters

UCS Intensity. With few exceptions, electric shock has been the only aversive
DCS used in conditioned suppression. The intensity of the shock is one of the
most powerful variables determining operant response rates during the CS,
with consistent reports of suppression as a direct function of shock intensity
(Brady and Sulsa, 1955). Annau and Kamin (1961), for example, used a group
design in which a 3-min CS was paired with a 0.5-sec shock of either 0.28, 0.49,
0.85, 1.55, or 2.91 rna. Conditioned suppression was a direct function of DCS
intensity across experimental groups. Response rates were (1) relatively un-
changed by the lowest shock intensity, (2) moderately suppressed, with indi-
vidual differences, by the 0.49-ma shock, and (3) completely suppressed at the
three higher intensities. Extinction of suppression was an inverse function of
DCS intensity during the initial acquisition training (also see Brophy and
Tremblay, 1971; James and Mostoway, 1968; Kamin and Brimer, 1963).
Conditioned suppression is also a monotonic function of DCS intensity
within individual subjects as well as across experimental groups. Henton and
Jordan (1970) superimposed different stimuli paired with 0.0, 0.1, 0.3, 1.0, and
3.0 rna upon a random ratio (RR) operant schedule (Figure 1.3). Initially, sup-
pression during each CS was a direct function of the shock intensity terminat-
ing the immediately preceding trial. Asymptotic suppression over the final ses-
sions was a direct function of the shock intensity terminating that trial (also see
Ayres, 1968; and Experiment IX, Chapter 2).
Related experiments have examined the effects of sequential changes in
shock intensity upon conditioned suppression. Hendry and Van Toller (1965),
for example, reported that lever pressing was substantially less suppressed by a
2.0-ma shock for subjects pretrained with a lower 1.0-ma shock. Conversely,
Rescorla (1974) reported that the unsignalled presentation of relatively higher
shock intensities may facilitate or increase conditioned suppression. Hoffman
et aI. (1963) and Quinsey and Ayres (1969a, 1969b) similarly reported increased
suppression following interpolated un signaled shock during extinction trials
(see also Rohrbaugh et aI., 1972; Rescorla and Heth, 1975).
Chapter 1: Review of Classical-Operant Conditioning 35

5 min.

a a

b
Figure 1.3. Cumulative record oflever pressing by monkey during a session with five different
superimposed CSs paired with shock. Each 30-sec stimulus is presented three times per session and
paired with shock intensity of 0.0 (a), 0.1 (b), 0.3 (c), 1.0 (d), and 3.0 rna (e). CS presentation is
indicated by downward displacement of recording pen. The recording pen moves one unit verti-
cally with each operant response and horizontally with time. A brief displacement of the pen indi-
cates delivery of the operant reinforcer. Data from Henton and Jordan (1970).

ues Duration. Only one study has examined the relationship between sup-
pression and ues duration. Using a O.5-ma shock, Riess and Farrar (l973a)
found that response rates were unchanged during the es when the shock dura-
tion was 0.05 sec, but were completely suppressed when shock duration was
1.0 and 3.0 sec. The rate of extinction of suppression was an inverse function of
the shock duration used in acquisition.

ues Probability. Previous studies of the "partial reinforcement effect" have

resulted in rather divergent and seemingly inconsistent results. The divergent
results have been obtained with acquisition/extinction rates as well as the mag-
nitude of conditioned suppression as the dependent variable. Geller (1964) ini-
tially reported that suppression by goldfish conditioned more rapidly and extin-
quished more slowly in groups trained with 100% es-ues pairings. Geller,
however, described two unpublished studies with rats in which suppression ex-
tinquished more rapidly following 100% es-ues pairings relative to groups
trained with either 50% or 25% pairings. Geller suggested that the opposite par-
tial reinforcement effects with fish and rats may be a species difference. Brimer
and Dockrill (1966) also reported faster acquisition of suppression by rats with
36 Section I: Concurrent Classical and Operant Conditioning

100% pairings relative to 50% and 25% groups. Wagner et al. (1967), however,
found no difference between 50% and 100% es-ues pairings. Equally diver-
gent results have been reported with an extinction measure of suppression in
rats. Scheuer (1969) reported greater resistance to extinction with a 100%
shock schedule, whereas Hilton (1969) reported more resistance to extinction
with a 50% partial reinforcement schedule (also see Brimer and Wickson,
1971).
Suppression is also determined by the shock probability during the intertrial
interval as well as during the es. Rescorla (1968b) reported suppression as an
inverse function of the probability of shock during the intertrial interval. es
response rates are unchanged relative to baseline rates, however, when the
probability of shock is equal during the es and intertrial interval.
Using a within-subject design, Willis and Lundin (1966) reported differential
suppression as a direct function of the percentage of es-ues pairings. How-
ever, using a group design, Willis (1969) reported no difference in suppression
with es-ues pairings of 100%, 90%, 70%, 50%, and 30%, with differential
suppression only in a 10% pairing group. The results of the group design study
thus contrasted with the within-subject function reported by Willis and Lundin
(1966). Willis (1969) suggested that the divergent partial reinforcement effects
in the literature may be partially determined by the experimental design of be-
tween-group or within-subject comparisons (also see Homzie et aI., 1969).
The results of within-subject experiments are even more inconsistent when
the comparison is expanded to include data from related areas. For example,
several investigators have reported accelerated response rates during a es with
a zero shock probability when a second es is paired with shock in the same
session ("inhibition of fear," discussed by Hammond, 1966, 1967; Rescorla,
1969a, 1969c). However, a rather large number of experiments routinely use
the same Pavlovian discrimination procedures in psychophysical investiga-
tions, with unchanged operant rates during the es associated with the zero
shock probability (review by Smith, 1970). Similarly unchanged response rates
during the nonshock stimulus have been reported in a variety of other proce-
dures (de Toledo and Black, 1966; Nathan and Smith, 1968; Henton and Jor-
dan, 1970). The surprising divergence suggests that the effects of ues proba-
bility must be highly conditional upon specific boundary conditions, training
procedures, and other variables.

Random Presentation of es and ues. A special case of ues probability is the

random and independent presentations of the es and the ues. A theoretical
paper by Rescorla (1967b) proposed that random es and ues pairings was the
"proper" control procedure in classical conditioning experiments (also see Re-
scorla, 1966; 1969b). Seligman (1968) subsequently reported a rather complex,
multistage experiment that compared predictable or signaled shock and unpre-
dictable or random shock. Lever pressing by rats was completely suppressed
during both the es and the intertrial interval with the random shock procedure.
Moreover, the random es and ues presentations in one conditioning cycle
retarded the acquisition of suppression to a different stimulus paired with shock
Chapter 1: Review of Classical-Operant Conditioning 37

in the next experimental phase. Seligman suggested that unpredictable shock

procedures may condition a "learned fear" or "learned helplessness," and
therefore would not be appropriate as neutral conditioning techniques (Selig-
man et aI., 1971).
Davis and McIntire (1969) also compared stimuli paired, unpaired, or ran-
domly presented with shock, and similarly reported suppression during both
the intertrial interval and the CS with random CS and UCS delivery. Additional
studies show that suppression during stimuli randomly associated with shock is
a function of the average interval between random CS and UCS presentations
(Kremer and Kamin, 1971) or the relative density ofthe CS and UCS schedules
(Kremer, 1971). Suppression may also be dependent upon sequential effects
within the random CS-UCS pairings. Benedict and Ayres (1972) reported sup-
pression during the CS only if chance pairings of CS and UCS occurred prior to
nonpairings during initial exposure to the random schedule (also see Ayres and
De Costa, 1971; Kremer, 1974; Nageishi and Imada, 1974; Witcher and Ayres,
1975; Baker, 1976).

CS Parameters

CS Duration. The first parameter of conditioned suppression to be exten-

sively studied was the relative duration ofthe CS (Libby, 1951), with functional
relationships subsequently determined for both trace and delay classical condi-
tioning procedures. One standard reference is the study by Stein et aI. (1958).
Using a delay conditioning procedure, Stein et aI. established that suppression
was determined by the ratio of CS on-time to CS off-time (i.e., the intertrial
interval). Response rates were completely suppressed during relatively brief
stimuli when the CS ratio approached 0.00, with higher response rates with in-
creasing CS on/off ratios. However, absolute CS duration per se was not signi-
ficantly related to response rate across a broad range of CS on/off ratios. Stein
et al. proposed that response rates will be reduced during the CS only to the
extent that suppression does not markedly reduce the operant reinforcement
probability.
Carlton and Didamo (1960) varied the CS on/off ratio by using a fixed dura-
tion CS of 3 min and varying intertrial intervals of 4, 6, or 60 min. As in the
study by Stein et al., response rates were completely suppressed during the CS
with the long intertrial interval, and substantially less suppressed with the two
shorter intertrial intervals (also see Davis et aI., 1969; Kling, 1972; Yeo, 1974,
1976).
In a related experiment, Shipley (1974) manipulated CS duration during ex-
tinction rather than acquisition, and reported that suppression was an inverse
function of the total CS on-time during the extinction test. The effect of CS ex-
posure was independent of the number and duration of CS presentations used
to generate the total exposure duration.
In a finely detailed series of experiments, Kamin (1961) reported that condi-
tioned suppression was also dependent upon the temporal characteristics of
trace conditioning procedures. Suppression was an inverse function of the
38 Section I: Concurrent Classical and Operant Conditioning

trace interval between es termination and DeS onset, as well as the interval
between es onset and DeS onset (also see Leaf and Leaf, 1966; Strouthes,
1965).
A seemingly paradoxical facilitation effect has been reported for low re-
sponse rates maintained by differential reinforcement oflow rate (DRL) sched-
ules.
Finnocchio (1963) reported that the low response rate was relatively in-
creased during a 5-min es alternating with a 5-min intertrial interval. The DRL
rate was suppressed, however, when a 2-min es alternated with an 8-min inter-
trial interval (also see Migler and Brady, 1964; Leaf and Muller, 1964; Black-
man, 1968a).

es Intensity. The effects of es intensity have been reported in a series of

papers by Kamin and associates. In the Kamin and Schaub (1963) experiment,
three different groups of rats were exposed to 3-min periods of white noise
paired with shock. The intensity of the white noise was either 49,63, or 81 db,
and acquisition of suppression was a direct function of es intensity. Kamin
(1965) further compared the effects of stimulus intensity per se against the ef-
fects of stimulus discriminability, or change from background noise. For differ-
ent groups of rats, the es was a reduction in the white noise from 80 db to ei-
ther 0, 45,50,60, or 70 db. The acquisition and asymptotic level of conditioned
suppression was a direct function of the magnitude of stimulus change. In a
companion experiment, the direction of stimulus change was reversed, with es
intensity always 80 db, and the background noise level either 0,45,50,60, or 70
db. In contrast to the differential suppression with intensity reduction as the
es, the acquisition and asymptotic level of suppression was equal in all groups
with various increases in noise as the es. The data thus suggest that suppres-
sion is not simply related to stimulus intensity per se, but is also determined by
the direction and magnitude of stimulus change from background (also see Zie-
linski, 1965; Dubin and Levis, 1973).
The effects of es intensity are further complicated by the choice of ues
intensity. Kamin and Brimer (1963) reported a 3 x 3 factorial study of the inter-
actions between three es intensities (47, 60, and 81 db) and three ues intensi-
ties (0.25, 0.50, and 0.80 rna). Response rates were unchanged and high during
all CS intensities paired with a low shock, equally suppressed during all CS in-
tensities paired with a high shock, but a direct function of CS intensity for
groups trained with a medium shock. The series of studies by Kamin provide
clear examples of the diverse effects generated by interacting parameters com-
monly found in all classical-operant schedule combinations. The results would
underscore the marked dependence of functional relationships upon the choice
of constants and other boundary variables (also see Testa, 1975).
A variety of psychophysical experiments have used the Estes-Skinner pro-
cedure and manipulated CS intensity as the major independent variable (Hen-
dricks, 1%6; Sidman et aI., 1966; Shaber et ai. 1967; reviews by Rosenberger,
1970; Smith, 1970). In the Hendricks (1966) experiment, key pecking by pi-
geons was reinforced on a variable interval (VI) 2-min schedule, with a 20-sec
Chapter 1: Review of Classical-Operant Conditioning 39

1.0
o • • • •
-
~
~

z
o
-
1ft
1ft
III
~
IlL
IlL 0.0
::t
1ft

1.0 2.0 3.0

C 5 INTENSITY [ ' concentration]
Figure 1.4. Conditioned suppression by pigeons as a function of the intensity of the odor used as
the CS. Each data point is the mean of three conditioning trials. Data from Henton (1966).

CS of light flickering at 10 cycles/sec. The CS flicker rate was gradually

changed in increments of 1 or 2 cycles/sec, until response rates during the CS
were 50% of the rates during continuous illumination (i.e., threshold). The
flicker fusion threshold was thus determined for eight different light intensities,
yielding critical flicker fusion frequencies as a function of stimulus intensity.
The Hendricks procedure has subsequently been used in ascending, descend-
ing, and constant stimuli psychophysical methods (Figure 1.4), as well as titra-
tion procedures (Rosenberger, 1970). Furthermore, conditioned suppression
has been used to determine absolute and difference thresholds in vision
(Powell, 1967; Powell and Smith, 1968; Shumake et aI., 1968, 1977; Rosen-
berger and Ernst, 1971; La Vail et aI., 1974), audition (Dalton, 1967; Price et aI.
1967; Rivizza et aI, 1969; Heffner and Masterton, 1970; Heffner et aI., 1971),
olfaction (Henton, 1969; Henton et aI, 1966, 1969; Davis, 1973; Pierson, 1974;
Oley et aI., 1975), temperature sensitivity (Kenshalo and Hall, 1974), micro-
wave detection (King et aI., 1971), and x-ray detection (Morris, 1966; Smith
and Tucker, 1969; Chaddock, 1972).

CS Generalization. Conditioned suppression will also generalize to nons hock

test stimuli as a function of the physical similarity of the test stimulus to the CS.
In a study by Ray and Stein (1959), an 1800-Hz tone was paired with shock, and
a 200-Hz tone was simply presented unpaired with shock in the same session.
The generalization test consisted of the occasional presentation of four tones
with intermediate frequencies between the stimuli paired and unpaired with
shock. Conditioned suppression during the test stimuli increased as the tone
frequencies approached the 1800-Hz CS.
A number of generalization experiments have been reported by Hoffman
and FleshIer (FleshIer and Hoffman, 1961; Hoffman and Fleshier, 1961, 1964,
1965; Hoffman et aI., 1963; review by Hoffman, 1969). For example, Hoffman
et al. (1963), presented test stimuli of 300, 450, 670, 2250, and 3400 Hz during
40 Section I: Concurrent Classical and Operant Conditioning

the extinction of suppression previously conditioned to a IOOO-Hz tone. Re-

sponse rates were most suppressed during the lOOO-Hz tone, with diminishing
suppression at test stimuli of progressively higher or lower frequencies. The
inverted U-shaped generalization gradient became steeper and sharper over
successive extinction sessions as the response rates gradually returned to pre-
vious baseline rates during each tone. Bimodal generalization gradients around
two different tones paired with shock have been reported by Hoffman et al.
(1966b) and Desiderato (1964).
In contrast to the previously reported symmetrical gradients around the CS,
Winograd (1965) observed asymmetrical gradients to click frequencies in rats.
The divergent effects were apparently dependent upon unconditioned effects of
the auditory test stimuli, with high click rates unconditionally eliciting more ac-
tivityand subsequently higher response rates. As a result, response rates were
less suppressed during test stimuli of high click rates compared to low click
rates, with a resultant asymmetrical gradient skewed by the higher frequencies.
Hendry et al. (1969) replicated the Ray and Stein discrimination procedures
but added more test stimuli to determine the generalization of suppression at
frequencies beyond the training stimuli paired and unpaired with shock. In gen-
eral, the maximal lever-pressing rate by rats shifted from the nonshock training
stimulus to adjacent stimuli away from the shock CS; conversely, the maximal
suppression shifted from the shock CS in the opposite direction away from the
nonshock CS (i.e., positive and negative peak shift; Hanson, 1959). The au-
thors noted a variety of individual differences in the shifted generalization func-
tions, however, and discussed the previous contradictory data reported by Ray
and Stein (1959), Winograd (1965), and Hoffman, 1967).

CS Preexposure (Latent Inhibition). The alteration of suppression by prior ha-

bituation of the to-be-conditioned stimulus was initially reported by Carlton
and Vogel (1967) and May et al. (1967). In the Carlton and Vogel study, a 10-sec
tone was simply presented without shock during six habituation sessions, and
then paired with shock on a single conditioning trial. In subsequent nonshock
test trials, licking responses were less disrupted in the group of rats previously
habituated to the CS relative to groups trained without tone preexposure or
given comparable preexposure to a different auditory stimulus. May et al.
(1967) further reported that conditioned suppression was an inverse function of
the number of stimulus preexposure trials (also see Leaf et al, 1968; Lubow and
Siebart, 1%9; James, 1971b; Domjan and Siegal, 1971, Reiss and Wagner, 1972;
and Lantz, 1973).
In addition to the above studies using simple delay classical conditioning,
latent inhibition of suppression has also been reported during compound condi-
tioning procedures. Schnur (1971), for example, exposed groups of rats to a
tone or a light, with subsequent classical conditioning with either light, tone, or
light plus tone paired with shock. Suppression was attenuated in the compound
conditioning procedure for subjects preexposed to one of the elements of the
compound CS. The suppression was not attenuated, however, in control
groups preexposed to one CS but tested with the alternative stimulus (also see
Anderson et aI., 1968, 1969a, 1969b; Dexter and Merrill, 1969; Kremer, 1972).
Chapter 1: Review of Classical-Operant Conditioning 41

In a rather complicated experiment, Carr (1974) used a compound condition-

ing procedure to study the latent inhibition of the "overshadowing" effect de-
scribed by Kamin (1969). The overshadowing of one stimulus by another is
found when subjects conditioned to a compound CS, such as light plus tone,
subsequently show less suppression on test trials with one CS element com-
pared to the second CS element (such as overshadowing of noise by light during
the compound conditioning). Carr reported that the overshadowing effect was
eliminated by preexposure to the light CS. Response rates were relatively more
suppressed during the noise test trials for subjects habituated to the potentially
overshadowing light. The inhibition of overshadowing by latent inhibition ex-
emplifies further complex stimulus interactions in the conditioned suppression
procedure (Kumar, 1970; general review of latent inhibition by Lubow, 1973).

Sensory Preconditioning. Sensory preconditioning is a second stimulus preex-

posure technique in which two neutral stimuli are sequentially paired (CS 1-
CS 2), with cs 2 subsequently paired with a UCS of food or shock. The sensory
preconditioning is an increase in the probability of conditioned responses dur-
ing test trials with CS 1 (Brogden, 1939). In 1967, Prewitt reported sensory pre-
conditioning of conditioned suppression in rats, Initially, a 10-sec tone was
paired with a 10-sec light on 1,4, 16, or 64 preconditioning trials. The light was
then paired with shock in phase 2, with suppression during the sensory precon-
ditioned tone recorded during phase 3. Response rates were unchanged during
the tone for subjects receiving random light and tone presentations in the first
preconditioning phase, but were suppressed as a direct function of the number
of preconditioning trials for the paired tone-light groups (also see Parkinson,
1968; Rogers, 1973).
Experiments elaborating the sensory preconditioning of suppression have
been reported by Tait, Suboski, and associates. As one instance, Tait et al.
(1969) reported that suppression during a sensory preconditioned tone was an
inverse function of the number of interposed preconditioning extinction trials.
Subsequent experiments indicated (1) partial reinforcement effects in sensory
preconditioning, with greater suppression on CS 1 test trials as a direct function
of the percentage ofCS l-CS 2 pairings (Tait et aI., 1971), (2) sensory precon-
ditioning in curarized rats (Cousins et al., 1971), (3) discriminative sensory pre-
conditioning, with different tone frequencies paired and unpaired with a light
CS 2 (Tait et al., 1972), and (4) sensory preconditioning as a function of the
intensity of CS 1 and CS 2 (Tait and Suboski, 1972).

Second-order Conditioned Suppression. Second-order procedures and sensory

preconditioning techniques are quite similar, but with a reversed order of clas-
sical conditioning and stimulus pairing phases. In second-order conditioning,
the CS-UCS pairings are conducted first, with subsequent sequential pairings
of a neutral stimulus with the CS. Preliminary reports of second-order condi-
tioned suppression were described by Davenport (1966) and Kamil (1968,
1969).
Kamil (1969), for example, described two experiments on second-order con-
ditioned suppression in rats. In experiment 1, a 10- or tOO-sec tone was termin-
42 Section I: Concurrent Classical and Operant Conditioning

ated by shock, and second-order conditioning consisted of a 30-sec light ter-

minated by the onset of the tone (unpaired with shock). Response rates during
the second-order light CS were transiently suppressed as the conditioned sup-
pression progressively extinquished during the first-order tone formerly paired
with shock. Kamil's second experiment compared the suppression during a
light terminated by tones formerly used in delay and trace classical condition-
ing. Response rates were equally suppressed during the second-order light in all
procedures, in spite of the differential rate of extinction of suppression during
the delay and trace conditioned tones. Rizley and Rescorla (1972) also reported
that second-order conditioned suppression is independent of the suppression
during the first-order CS (also see Byrum and Jackson, 1971). Thus, second-
order suppression does not seem to be a classical conditioned association be-
tween the first and second-order stimuli. As noted by Kamil, the divergent sup-
pression during the first- and second-order CS is not consistent with classical
conditioning formulations of either second-order conditioning or conditioned
suppression (also see Holland and Rescorla, 1975; Rescorla, 1976; Rescorla
and Furrow, 1977).

Compound CS. A number of interesting experiments have used the simultane-

ous presentation of two conditioned stimuli within the conditioned suppression
paradigm. An early investigation (Ayres, 1966) was designed to examine the
information hypothesis of Egger and Miller (1962). In very brief outline, the in-
formation hypothesis would predict that if two stimuli were paired with shock,
then (1) more suppression would be conditioned to the stimulus whose onset is
first (stimulus 2 being redundant), but (2) more suppression would occur to
stimulus 2 if stimulus 1 were also occassionally presented alone and without
shock (stimulus 1 being an unreliable predictor of shock). However, Ayres
found equal suppression on test trials with CS 1 or CS 2, with equal rates of
extinction of suppression across testing sessions. Ayres then concluded that re-
dundant and unreliable conditioned stimuli generate as much suppression as
more informative stimuli (but also see Scheuer and Keeter, 1969; Seger and
Scheuer, 1977).

Summation of Conditioned Suppression. The summation of conditioned sup-

pression during compounding of two CS elements was described by Miller
(1969). Initially, a light or a tone was individually paired with shock intensities
selected to generate 20% to 40% suppression during each CS. Compound trials
with light plus tone were then added to each on-the-baseline session, and
operant responses were more suppressed during the compound than during
trials with each individual CS element. In a second experiment, response rates
were again relatively more suppressed during compound trials, with the com-
pound suppression dependent upon the magnitude of suppression controlled by
each CS element. Furthermore, the compound CS continued to suppress re-
sponding even when the suppression was completely extinquished during the
individual CS elements. More recent experiments have replicated the data of
Miller, with reports of (1) increased suppression during a compound CS relative
Chapter I: Review of Classical-Operant Conditioning 43

to individual CS elements (Weiss and Emurian, 1970; Van Houton et al., 1970;
Booth and Hammond, 1971), (2) continued suppression during a compound CS
after extinction of suppression during the individual CS elements (Reberg,
1972), (3) summation of suppression during compound trials with CS elements
previously paired with different shock intensities (Hendersen, 1975), and (4)
summation of suppression during compounding of stimuli paired with shock
onset ("fear" conditioning) and shock offset ("hope" or "relief' conditioning,
Zelhard, 1972).

Attenuation of Conditioned Suppression. The attenuation of conditioned sup-

pression by external inhibitors and conditioned inhibitors has also been re-
ported in compound conditioned suppression. Hammond (1967) reported an at-
tenuation of suppression during compound trials with a CS formerly paired
with shock and a conditioned inhibitory stimulus previously unpaired with
shock. In 1969, Rescorla further reported that the conditioned inhibition of sup-
pression was a direct function of the shock frequency scheduled in the absence
of the inhibitory CS (Rescorla, 1969a, 1969b). Reberg and Black (1969) repli-
cated the summation of suppression by stimuli paired with shock and the condi-
tioned inhibition of suppression by stimuli previously unpaired with shock, but
reported unchanged suppression during compound trials with an experimen-
tally neutral stimulus (also see Hendry, 1967; Rescorla, 1969c; Cappel et al.,
1970). The external inhibition of suppression by an added neutral stimulus has
been reported, however, by Brimer and Wickson (1971; Brimer, 1971). The dif-
ferent effects of compounding with neutral stimuli may be partially determined
by unconditioned reflexive properties of the added neutral stimulus (Hender-
sen, 1973). In addition, the attenuation of suppression may be related to the
number of suppression trials completed prior to the introduction of the inhibi-
tory stimulus (Brodigan and Trapold, 1974).
A last group of studies using compound CS dealt with the overshadowing
effect (Carr, 1974). The overshadowing procedure was originally reported by
Kamin (1969); a white noise and a light were separately paired with a 1.0-ma
shock, with the two CS subsequently combined during compound conditioning
trials. Later test trials with each CS element revealed that responding was less
suppressed during the noise CS than during the light CS. The results imply that
the compound conditioning phase resulted in suppression primarily condi-
tioned to the light element of the compound CS (overshadowing of noise by
light). However, when the DCS intensity was increased from 1.0 to 4.0 ma,
conditioned suppression was equivalent during test trials with the noise and
light, with apparently equal suppression conditioned to each CS element during
the compound light plus noise conditioning trials (additional studies by Res-
corla, 1971b; Rescorla and Wagner, 1972; Wagner and Rescorla, 1972; Baker,
1974; and Kamin and Gaioni, 1974). Related experiments described above ap-
pear under CS preexposure (latent inhibition).
Gray and Appignanesi (1973) similarly reported that prior conditioning of CS
paired with shock blocked the acquisition of suppression to a second CS when
the two stimuli were subsequently compounded. In contrast, St. Claire-Smith
44 Section I: Concurrent Classical and Operant Conditioning

and Mackintosh (1974) found that the effects of compound stimulus trials may
be rather conditional. Compound trials produced clear increments in suppres-
sion to each CS element if the separate CSs were previously paired with shock
on only a few trials. Alternatively, the compound trials had no consistent effect
upon suppression during individual CS elements that had been frequently
paired with shock prior to compound conditioning. In the St. Claire-Smith and
Mackintosh study, compound trials thus did not produce a decrement in sup-
pression to one or the other stimulus element (also see Mackintosh, 1975, 1976;
Donegan et al., 1977).
In summary, experiments using compound conditioned stimuli consistently
describe summation of suppression during compound trials with CS elements
that individually control moderate suppression. However, the effects of com-
pound trials upon the subsequent suppression elicited by individual CS ele-
ments is not yet clear. Similarly, the effect of a neutral stimulus or external in-
hibitor added to a CS paired with shock has not been consistent across
experiments (general review by Weiss, 1972).

Backward Conditioning. One early report suggested that suppression might

also be conditioned by backward (UCS-CS) pairings (Singh, 1959). Kamin
(1963), in contrast, failed to replicate the backward conditioning of suppression
with either a 0.85- or a 1.5-ma shock terminated by a 3-min CS. Instead, Kamin
found a partial suppression of baseline rates during the intertrial interval, and
slightly accelerated rates during the backward CS (relative to the suppressed
intertrial interval rates). A retardation effect, however, has been reported for
the influence of backward UCS-CS conditioning upon the subsequent forward
conditioning of suppression to the CS (Siegal and Domjan, 1971). Mahoney and
Ayres (1976) also reported that backward conditioning may in fact produce
more suppression than random CS and UCS presentation, although less sup-
pression than forward and simultaneous CS-UCS pairings. (also see James,
1971a; Burdick and James, 1973; Heth and Rescorla, 1973; Siegal and Domjan,
1974; Davis et aI., 1976.)

Spontaneous Recovery and Retention of Suppression. Spontaneous recovery,

or suppression of responses during postextinction test trials, was described in a
paper by Burdick and James (1970). Using a four-phase experiment, the investi-
gators determined the temporal relationship between spontaneous recovery
and the length of the postextinction test interval. Following first the acquisition
and then the extinction of conditioned suppression, different groups of rats
were given a CS test trial at 3.5 min, or 0.5, 1.0,24.0, or 72.0 hr following the
last extinction trial. Suppression during the spontaneous recovery test re-
mained extinquished for the group tested at 3.5 min, with a progressive recov-
ery of suppression across the remaining groups to a maximum at 24 hr. James
(1971a) and Kling and Kling (1973) further reported that spontaneous recovery
diminished across repeated sets of extinction trials and test trials.
The procedures used to measure the retention of conditioned suppression
are basically similar to the spontaneous recovery procedures, with the excep-
Chapter 1: Review of Classical-Operant Conditioning 45

tion that the retention interval follows acquisition training rather than extinc-
tion. The previously cited experiments by Hoffman et al. (1963), and Hoffman
et al. (1966a) indicate that conditioned suppression generalization gradients re-
main relatively intact after 2.5 and 4.0 years in pigeons. Gleitman and Holmes
(1967) also reported that suppression by rats is relatively unchanged after a re-
tention interval of 90 days. Hammond and Maser (1970), however, reported
that the pattern of suppression may be changed in spite of an unchanged overall
suppression during the CS.

Operant Conditioning Variables

Reinforcer Parameters

Types of reinforcers. Conditioned suppression may be quite dependent upon

the specific choice of positive reinforcer selected to maintain the operant base-
line. Brady and Conrad (1960) explicitly compared the suppression of re-
sponses reinforced by response-contingent stimulation of the limbic system to
responses reinforced with water (rats), milk (cats), or sugar pellets (monkeys).
In rats, responses reinforced with water were completely suppressed, whereas
response rates were virtually unaffected during the CS when the reinforcer was
stimulation of the septum or medial forebrain bundle .. Similarly, conditioned
suppression of monkeys was also attenuated when medial forebrain stimulation
was substituted for sugar pellets as the operant reinforcer. Response rates of
cats, however, were equally suppressed when the reinforcer was either milk or
stimulation of the caudate nucleus. Brady and Conrad suggested that the ana-
tomicallocation of the electrode seemed to be a critical determinant of the ef-
fects of reinforcing brain stimulation upon conditioned suppression (also see
Brady, 1957, 1961; McIntire, 1966; Merrill et aI., 1970; Haworth, 1971).
Geller (1960) also studied the effects of qualitatively different reinforcers
upon suppression. He reported that suppression conditioned more rapidly and
extinquished more slowly for subjects reinforced with water relative to subjects
reinforced with sweetened milk.
Kruper and Haude (1964) used a rather novel procedure of CS-shock pair-
ings superimposed upon an operant schedule of response-contingent visual ex-
ploration. Lever pressing by monkeys was reinforced with a 5-sec view of the
external environment, augmented with a panel of30 colored lights. As with the
more conventional operant reinforcers, responses maintained by visual ex-
ploration were suppressed during a tone paired with shock.

Reinforcer Magnitude. An initial study (Vogel and Spear, 1966) suggested that
suppression may be partially determined by the quantity or magnitude of the
operant reinforcer. Vogel and Spear found that responses reinforced with a
32.0% sucrose solution were less suppressed during CS test trials compared to
groups reinforced with 4.0% or 11.5% sucrose. However, Ayres and associates
have reported no differential effects of operant reinforcer magnitude upon sup-
pression, either between groups (Ayres, 1966; Ayres and Quinsey, 1970) or
46 Section I: Concurrent Classical and Operant Conditioning

within individual subjects (Hancock and Ayres, 1974). In the Ayres and Quin-
sey study, for example, licking responses reinforced with either 8.0% or 32.0%
sucrose solutions were equally suppressed. In reviewing the previous studies,
Hancock and Ayres (1974) concluded that incentive and motivational theories
of conditioned suppression may be inconsistent with the experimental results.

Reinforcement Schedules

Response Rate Measures. The effects of CS-shock pairings upon response

rates generated by different operant schedules was initially described by Brady
(1955). The extinction of suppression was differentially dependent upon the
baseline reinforcement schedule, with more rapid extinction using variable and
fixed ratios (VR and FR), a moderate extinction rate with continuous reinforce-
ment (CRF), and slowest recovery with variable interval (VI) schedules (also
see Brady and Hunt, 1955; Goy and Hunt, 1953).
The most detailed analysis of schedule effects is presented in the compre-
hensive series of papers by Lyon and co-workers. Lyon (1963) compared the
magnitude and acquisition/extinction rates of conditioned suppression in VI
schedules. Response rates were relatively more suppressed, with faster acqui-
sition and slower extinction, with a VI 4-min schedule compared to a VI I-min
schedule.
In FR schedules, suppression is conditional upon the location of CS onset
during the response ratio (Lyon, 1964). Response rates were completely sup-
pressed when the CS was presented early in the ratio, but relatively unaffected
by a CS presented later in the ratio immediately prior to reinforcement. Sup-
pression as a function of reinforcement proximity was replicated with FR
schedules by Lyon and Felton (1966a) and with FI schedules by Lyon and Mil-
lar (1969). In contrast, Lyon and Felton (1966b) reported that suppression of
VR responses was independent of the averaged probability of operant rein-
forcement. Response rates were equally suppressed on VR schedules of 50,
100, or 200 responses per reinforcement. During the CS, response rates tended
to have a biphasic pattern and were either completely suppressed or completely
unchanged, with no intermediate or graded suppression.
Blackman (1967a), however, criticized the results of Brady and Lyon as a
confounding of the different effects of the subject's response rate and the
scheduled reinforcement rate. Blackman (1966, 1967b, and 1968b) attempted
to separate response rate from reinforcement rate by using response pacing
schedules. He reported that suppression may be partially determined by the
subject's response rate during the intertrial interval.

Response Accuracy Measures. A derivative question, given the alteration of

overall rates during the CS, concerns the effect of the Estes-Skinner procedure
upon the proportion of reinforced to total, albeit suppressed, responses. The
effects of CS-shock pairings upon the relative distribution of correct versus
incorrect responses was described by Migler and Brady (1964). An A-B re-
sponse chain was reinforced on a DRL schedule, in which the sequence oflever
Chapter 1: Review of Classical-Operant Conditioning 47

A and lever B responses was reinforced only if the A - B interresponse time ex-
ceeded a minimum of 5 sec. The low overall rate of completed A-B response
sequences was suppressed, but the distribution of A to B interresponse times
was virtually unchanged during the CS (also see Leaf and Muller, 1964).
In a related experiment, Kruper (1968) also reported overall suppression
with relatively unchanged response accuracy in an oddity discrimination task.
The rate of choice of the odd stimulus was reduced during a tone paired with a
1.0- to 1.5-ma shock, with further suppression at increased shock intensities of
2.0 to 3.0 rna. However, the proportion of correct responses to the odd stimulus
during the CS was identical to the response accuracy during control periods at
all shock intensities.
Blackman (1970) subsequently replicated the suppression of A-B response
chains using a DRL of 5, 10, or 15 sec. The A to B interresponse times were
unchanged with the 5 sec schedule, as previously reported by Migler and
Brady. However, the interresponse times were proportionally decreased with
the 10- and 15-sec schedules. [Facilitation rather than suppression of DRL re-
sponse rates has also been reported for various combinations of CS duration
and UCS intensity by Finnocchio (1963), and when a limited hold is added to
the DRL by Blackman (1968a).]
Response facilitation has also been reported for the low rate of responses in
the extinction component of a multiple variable interval: extinction (mult VI
EXT) schedule. Hearst (1965) reported that CS-shock pairings during the VI
component paradoxically increased the near zero rate of responding during the
alternate EXT component. However, Weiss (1%8) reported three experiments
using a similar multiple schedule but was unable to confirm Hearst's results.
The VI schedule, shock intensity, and amount of training were manipulated
within and across experiments, but the response rates under EXT were re-
duced rather than increased under all experimental conditions. More recently,
Blackman and Scrutton (1973) also reported no "disinhibition" of extinction
response rates by CS-shock pairings superimposed upon the VI component of
the multiple schedule.
In summary, the current literature consistently demonstrates that the effects
of negative classical conditioning procedures are strongly conditional upon
operant schedule parameters, ranging from increased to unchanged to de-
creased operant rates during the same superimposed classically conditioned
"fear." The suppression of high response rates tends to be an inverse function
of the local probability of response-contingent reinforcement, with the excep-
tion of VR schedules. The data for low response rates are markedly inconsis-
tent and range from conditioned suppression to paradoxical facilitation across
similar experiments.

Suppression of Operant and Consummatory Behaviors. Studies by DeCosta

and Ayres (1971) examined the effects ofthe Estes-Skinner procedure upon the
two different responses involved in an operant contingency-the operant or
reinforced response and the consummatory or reinforcing response (Premack,
1965). Licking a 32% sucrose solution was programmed in one session, and bar
48 Section 1: Concurrent Classical and Operant Conditioning

pressing was reinforced with the sucrose solution on a VI I-min schedule in

alternate sessions. The operant bar-pressing response was suppressed more
than the consummatory sucrose-licking response during a 2-min tone paired
with a O.8-ma shock. In addition, bar pressing was again relatively more sup-
pressed when reinforcement density and adventitious punishment effects were
equated for both responses. Jackson and Delprato (1974) similarly reported that
eS-shock pairings did not suppress the consumption of food pellet reinforcers.
The investigators noted that the suppression of operant responses without a
suppression of consummatory responses is somewhat inconsistent with con-
temporary theories of the Estes-Skinner procedure.

Operant Conditioning History

The conditioned suppression of operant responses is conditional upon and
indeed may be reversed by manipUlating the subject's previous operant training
history. In 1958, facilitation rather than suppression of VI responses was re-
ported in subjects with previous histories of avoidance conditioning (Herrn-
stein and Sidman, 1958). The seemingly aberrant accelerated response rate was
converted to the more typical conditioned suppression by interpolating addi-
tional sessions of avoidance extinction. Herrnstein and Sidman suggested that
the response facilitation may have been dependent upon the operant avoidance
history combining with the unavoidable shock to generate adventitious shock
avoidance contingencies.
An in-depth analysis of operant conditioning history and conditioned sup-
pression has been reported by Sidman (1958). Responses by monkeys on one
lever were reinforced on a VI 4-min schedule, while responses on a second
lever had previously avoided a S.O-ma shock for 20 sec. Responses on both
levers were increased, not suppressed, during superimposed eS-shock pair-
ings. However, various schedule manipulations revealed that the responses on
the two levers were in fact mutually interdependent, with each response par-
tially controlled by both the avoidance and VI conditioning histories. Separa-
ting the concurrent response rates by additional schedule manipulations then
resulted in the typical conditioned suppression of food reinforced responses,
with an acceleration of shock avoidance responses, during the es. Thus, the
atypical acceleration of VI responding was neither abnormal nor pathological,
but rather was determined by an interaction between the normal controlling
variables of the two concurrent schedules (Sidman, 1960a, 1960b; also see Kel-
leher et al., 1963; Waller and Waller, 1963, and Lewis, 1973).

Operant Punishment in Conditioned Suppression

The "response-independent" presentation of a shock DeS is only accurate
as a procedural definition, since shock must be temporally contiguous with
some aspect of the subject's continuous stream of behavior. This functional
nonresponse-independent characteristic of a DeS has prompted an examina-
tion of the uncontrolled operant punishment effects of the shock DeS. Hunt
and Brady (1951b) found that lever-pressing rates were equally and completely
suppressed during stimuli terminated by response-independent shock (condi-
Chapter 1: Review of Classical-Operant Conditioning 49

tioned suppression procedure) or associated with response-contingent shock

(discriminated punishment procedure). However, subjects in the conditioned
suppression group exhibited common behavioral patterns of crouching, freez-
ing, urination, and defecation. In contrast, subjects in the discriminated punish-
ment group simply passively avoided the response lever, while occasionally
emitting abortive lever approach sequences.
The Hunt and Brady experiment was further replicated by Hoffman and
FleshIer (1965), who specifically equated the frequency of shock during each
stimulus. Responding was again equally suppressed by response-independent
and response-contingent shock when the shock intensity was 2.0 rna. However,
Hoffman and FleshIer reported that the operant punishment procedure pro-
duced more suppression when the shock intensity was only 1.0 rna. In contrast,
Orme-Johnson and Yarczower (1974) found greater suppression with a condi-
tioned suppression procedure than with an operant punishment procedure, in-
dependent of shock intensity and schedule manipulations.
Using different procedures, Gottwald (1967) demonstrated that the effects of
conditioned suppression procedures are partially determined by the interval be-
tween the operant response and the classical conditioned UCS on the imme-
diatley preceding trial. The adventitious punishment effect across sequential
trials was subsequently replicated across four different shock procedures by
Gottwald (1969). Church and co-workers have reported an elegant research
program comparing the effects of response-independent and response-contin-
gent shock, and also conclude that the unprogrammed operant response-UCS
shock correlation is not inconsequential in the Estes-Skinner procedure (re-
view by Church, 1969; Church et aI., 1970; also see Herrnstein and Sidman,
1958; Sidman, 1958, 1960a; DeCosta and Ayres, 1971; Desiderato and New-
man, 1971; Matthews et aI., 1974).

Suppression During Stimuli Associated with Shock Avoidance Responses

A discriminative stimulus that controls operant shock avoidance responses
will also function as a conditioned stimulus eliciting suppression of food rein-
forced responses. In one study (Kamin et aI., 1963) a discriminative stimulus
formerly used to control shuttle box avoidance responses was presented (with-
out shock) during VI food reinforcement sessions. The VI responding was sup-
pressed, and the magnitude of suppression was an inverted U-shaped function
of the amount of previous avoidance conditioning. Kamin et al. therefore con-
cluded that suppression was not simply or linearly related to the efficacy of pre-
vious avoidance responses in the shuttle box. Furthermore, conditioned sup-
pression has been reported during a stimulus that controls a concurrent shock
avoidance response. Hoffman and FleshIer (1962) reported that the food rein-
forced response rate was completely suppressed during trials with a successful
avoidance response, but only partially suppressed during trials without a con-
current avoidance response. In contrast to the study by Kamin et aI., suppres-
sion was not attenuated by prolonged shock avoidance training in this concur-
rent procedure. However, Linden (1969) found that criterion acquisition of
shuttle box responses would attenuate conditioned suppression to the CS (also
see Tarpy, 1966; Desiderato and Newman, 1971).
50 Section I: Concurrent Classical and Operant Conditioning

Concurrent Behaviors and Conditioned Suppression

A number of investigators (Hunt and Brady, 1951b; Brady, 1953; Davitz,
1953) have described increased crouching, immobility, defecation, and urina-
tion concomitant to the suppression of operant responding (review by Brady
and Hunt, 1955). More recently, Stein et al. (1971) used a videotape recording
system to analyze the subject's behavior, and reported freezing and wing-flap-
ping responses by pigeons during the suppression of key pecking (also see
Hoffman and Barrett, 1971). Stein et al. argued that freezing was not a classical
conditioned response but a reduction in rate of large classes of overt behaviors
during the CS. Alternatively, CS-shock pairings may elicit increased rather
than decreased activity in pigeons, with freezing during the Estes-Skinner pro-
cedure a paradoxical result of the interaction between CS-elicited activity and
baseline operant responses (see Chapter 2).

Time-out from Positive Reinforcement as the Aversive "ucs"

An operant variation of the Estes-Skinner procedure is the substitution of
extinction or time-out from positive reinforcement for the shock UCS in the
classical conditioning paradigm (Figure 1.5). As a conventional UCS, the al-
teration of the reinforcement schedule is independent of any specific response
but dependent upon the termination of the CS or "warning" signal. Unlike a
Pavlovian UCS, the extinction period requires a conditioning history to control
a specific response analogous to the UCR. The distinction, however, is some-
what academic and misleading. The response-independent presentation of a
food pellet as the UCS also fails to irrevocably elicit approach and con summa-

~1 LEVER 1 = MUlT. V R80:T.O.

15 A --------------------~~r---------------~-
SEC·B --------r-__~--~~----------~~I~I~
C• _ _ _ .'I_."I_'I_.~ •.. ,

50 A- - - - - - - -
SEC·B,------______.-__~--~--------------~---
C_.I_II_I .-. _ _I

.._-_. --
100 A - - - - - - ----,______---i--------------T-
SEC. B -.-------,---.---~-------;_,

VR 10 T.O. : VR 10

Figure 1.5. Event recordings comparing 15-,50-, and 100-sec stimuli terminated by 3-min time out
from operant reinforcement. A. Duration of pre-schedule change stimulus. B. Delivery of operant
reinforcers on variable ratio 80 schedule . C. Lever pressing by monkeys. Data from Henton and
Iversen (1973).
Chapter 1: Review of Classical-Operant Conditioning 51

tory responses, until the subject is thoroughly apparatus habituated, condi-

tioned to approach the food dispenser, and conditioned to retrieve the UCS
from a tray or a bowl (Pavlov, 1927). Similarly, the UCR to electric shock is
first conditional upon eliminating a variety of other prepotent responses from
the subject's behavioral repertoire. (For example, with rat as subject, escape is
prevented by restricting the subject to a box, blatant modification of the shock
is prevented by a polarity scrambler across the shock grids, then by a low ceil-
ing to prevent standing on the hind legs, followed by shock added to the walls,
food cup, and operant manipulandum, and, if necessary, removal of the insulat-
ing fur from the rat's body-only to find that the subject may still minimize the
shock by standing awkwardly on its tail, perhaps crouched and balanced
against the plastic lens of a chamber light or the Plexiglas door.)
All this merely to suggest that the analogy between electric shock and
operant extinction as the UCS may not be as discrepant as cursory examination
might dictate. A more fundamental distinction is that the superimposition of
stimuli terminated by time-out from reinforcement alters the operant schedule
from a simple to a multiple schedule, whereas the addition of a CS-CR-UCS-
UCR sequence changes the conditioning procedures from simple to concurrent
schedules.
Pliskoff (1961, 1%3) first described the operant rate changes during a stimu-
lus terminated by a schedule change to either a lower or a higher density of
operant reinforcement. In both experiments, key pecking by pigeons tended to
increase during the stimulus paired with the lower reinforcement rate, but to
decrease during the stimulus paired with the higher reinforcement rate. The
rate changes in both studies, however, were rather variable both within individ-
ual subjects and across all subjects. Pliskoff is frequently misreferenced as pro-
posing that baseline rates will be increased during stimuli terminated by a lower
reinforcement density but suppressed during stimuli paired with a higher rein-
forcement density. In view of the data, Pliskoff (1963) explicitly rejected this
hypothesis, and instead proposed that response rates would be higher during
one stimulus paired with decreased reinforcement compared to a second stimu-
lus paired with increased reinforcement, but not necessarily increased relative
to control or baseline rates.
Leitenberg (1966) also reported generally increased response rates by pi-
geons during a stimulus terminated by time-out from VI reinforcement. The re-
sponse rates were similarly accelerated when the schedule change to time-out
was response independent or response contingent. Leitenberg et al. (1968),
however, reported suppression rather than acceleration by rats during a pre-
time-out stimulus. Carman (1969) next reported that suppression by rats was
"only slight" during a pre-time-out stimulus, and did not occur under some
conditions (also see Kaufman, 1969). Calef et al. (1971) subsequently reported
substantial response facilitation by rats in early training, but unchanged rates
after prolonged training with stimuli paired with time-out (also see Parker et al.,
1971; Homes, 1972; and Experiment IV, Chapter 2).
In summary, the effects of stimuli paired with time-out from reinforcement
have been inconsistent across different experiments and species, and the con-
trolling variables remain unclear (Coughlin, 1972).
52 Section I: Concurrent Classical and Operant Conditioning

Subject Variables
Age
Relatively recent experiments suggest that conditioned suppression may be
an age-dependent phenomenon. Using rats 40, 90, and 354 days of age, Pare
(1969) reported that the acquisition of differential suppression during a CS
paired with shock and a second CS explicitly unpaired with shock was an in-
verse function of age. However, the differential effect of age was eliminated in
a second experiment by manipulating the auditory stimuli used in the discrimi-
nation and also changing the shock intensity. The relative performance of
young versus older rats may then be a function of CS and UCS parameters. For
example, Frieman et al. (1970) found that conditioned suppression was more
broadly generalized across tone stimuli in adult subjects relative to infants.
Brunner et al. (1970) and Snedden et aI. (1971) have also reported greater sup-
pression with mature rats than infants (also see Campbell and Campbell, 1962;
Solyom and Miller; 1965 Frieman et aI., 1971; Buchanon et al., 1972; Persinger
and Pear, 1972; Wilson and Riccio, 1973; Coulter et aI., 1976).

Central Nervous System Effects

Leisons. Neurophysiological investigations have frequently used the CER to

analyze the anatomical and physiological parameters of affective behaviors.
Brady and Nauta (1953), for example, reported that conditioned suppression
was attenuated by placing lesions in the septum. In comparison, neocortical le-
sions had no detectable effect. Brady and Nauta (1955) replicated the attenua-
tion effects of septal lesions, and further reported that lesions of the thalamic
habenular complex had no discemable effect upon the acquisition or retention
of suppression but may have produced slightly faster extinction. However, the
attenuation of suppression by septal lesions was transient and was paradoxi-
cally opposite to the increase in emotional reactivity recorded on a rating scale
(also see Brady, 1961, 1962, 1970a; Harvey et aI., 1965; Trafton, 1967).
The effect of amygdalectomy on conditioned suppression was described by
Kellicut and Schwartzbaum (1963) and Thompson and Schwartzbaum (1964).
The acquisition of suppression was blocked by large lesions that included the
basolateral and corticomedial nuclei of the amygdala, but was unaffected by
lesions of the lateral amygdala and putamen. Thompson and Schwartzbaum
proposed that the amygdala may participate in somatomotor inhibition, and
therefore lesions would interfere with response inhibition and conditioned sup-
pression. McIntyre and Molino (1972) and Lidsky et al. (1970) also described an
attenuation of suppression by large lesions of the amygdala. However, an in-
creased suppression rather than attenuation has also been reported in amygda-
lectomized rats (Suboski et aI., 1970).
Further neurophysiological studies have implicated hippocampal lesions in
conditioned suppression. Nadel (1968) reported that dorsal hippocampal le-
sions facilitated the acquisition of suppression but had no effect upon the rate
of extinction. In contrast, ventral hippocampal lesions had an opposite result,
Chapter 1: Review of Classical-Operant Conditioning 53

with no change in acquisition but a retarded extinction of suppression. The ac-

quisition attenuation effect of dorsal hippocampal lesions, however, may be de-
pendent upon the operant reinforcer (Freeman et aI., 1974).
In addition, thalamic lesions also seem to influence conditioned suppression.
Lesions of the dorsomedial thalamus may attenuate conditioned suppression,
but not the conditioned cardiac responses, during the Estes-Skinner procedure
(Nathan and Smith, 1971). A similar attenuation effect has been reported by
Dantzer and Delacour (1972), who suggested that the attenuation may be an
indirect result mediated by an increase in overall baseline rates.

Amnesiac Effects of Localized Brain Stimulation. The effects of repetitive

amygdaloid stimulation following conditioned suppression, shuttle box avoid-
ance, or maze learning was reported by Goddard (1964). Suppression was sig-
nificantly attenuated by amygdaloid stimulation delivered immediately posttrial
or throughout the conditioning session, with no effect of stimulation delivered
only during the CS. Goddard proposed that amygdaloid stimulation was func-
tionally equivalent to a lesion and scrambled the orderly flow of neural re-
sponses during the consolidation of emotional conditioning (Goddard, 1969).
Levine et al. (1970), however, were unable to replicate the effect of posttrial
amygdaloid stimulation in monkeys. The retrograde effects were only found
with seizures induced by inferotemporal stimulation which propogated the
amygdala. A second series of experiments (Lidsky et aI., 1970) also failed to
replicate the retrograde disruption by low-level amygdaloid stimulation using
rats (Experiments 2A,B,C). Anterograde and retrograde attenuation effects
were found, however, with high-intensity stimulation that spread diffusely and
elicited grand mal type seizure patterns (Experiments 3A,B). Lidsky et ai. con-
cluded that the anygdaloid complex does not play an essential role in memory
consolidation of aversive processes (also see McIntyre, 1970; McIntyre and
Molino, 1972).
Hippocampal stimulation has also been reported to substantially decrement
conditioned suppression (Barcik, 1970). The attenuation by localized seizures
was in fact comparable to the effects of generalized electroconvulsive seizures.
This blocking of suppression by hippocampal stimulation was also described by
Shinkman and Kaufman (1972a), and it may be a function of stimulation inten-
sity (Shinkman and Kaufman, 1972b; also see Gustafson et al., 1975). The re-
tention of conditioned suppression may also be disrupted by injection of potas-
sium chloride into the hippocampus (Avis and Carlton, 1968; Hughes, 1969;
related experiments by Auerbach and Carlton, 1971; and Avis, 1972).

Generalized Seizures by Electroconvulsive Shock. A long series of experiments

using electroconvulsive shock (ECS) has been reported by Brady and co-work-
ers. In the first study (Hunt and Brady, 1951a) ECS was administered three
times per day for 7 days following the stabilization of suppression, defecation,
and freezing during the CS. The ECS treatments virtually eliminated the sup-
pression of lever pressing, with a concomitant increase in activity and reduc-
tion in defecation during the CS. Subsequent experiments demonstrated that the
54 Section I: Concurrent Classical and Operant Conditioning

attenuation effects were (1) transient, with a reappearance of complete sup-

pression after 30 days (Brady, 1951), (2) dependent upon the number and tem-
poral distribution of ECS treatments (Brady et al., 1954; Hunt and Brady,
1955), (3) inversely related to the delay between CS-UCS pairings and ECS
treatments (Brady, 1952), (4) enhanced by interpolating additional ECS treat-
ments during the retention interval (Brady, Stebbins, and Hunt, 1953), and (5)
independent of the temporal order of classical and operant conditioning (Geller
et aI., 1955) (additional studies by Brady and Hunt, 1951, 1952; Brady, Steb-
bins, and Galambos, 1953; Hunt et aI., 1953).
A similar transient effect of ECS administration has been reported in mice
(Kohlenberg and Trabasso, 1968), and the temporal gradient of recovery from
ECS effects has been described by Nachman et al. (1969; related experiments
by Willi, 1969; Kesner et al., 1970; De Vietti and Holliday, 1972).
The usual ECS methodology was altered in an interesting experiment by
Winocur and Mills (1970), who used the ECS as the Pavlovian UCS within the
conditioned suppression procedure. The suppression during a stimulus termin-
ated by ECS was equivalent to the typical suppressive effects of CS-shock
trials. Suppression during the CS-ECS pairings seemed to be related to elicited
convulsive responses rather than retrograde aftereffects during the recovery
from the seizure.

Drugs. A host of experiments have assessed the impact of pharmacological

agents upon conditioned suppression (reviews by Avis and Pert, 1974; Brady,
1957, 1959; Miczek, 1973). The functional effects of drugs, however, are quite
complex, involve divergent effects across behavioral parameters, and also in-
volve a multitude of interacting biochemical parameters (i.e., differential dose-
response functions, action time courses, pharmacological metabolic rates, pri-
mary and secondary involvement of different neurochemical systems, etc.; re-
views by Dews, 1962; Boren, 1966; Fontaine and Richelle, 1969; Kumar et aI.,
1970; Goodman and Gilman, 1970). For example, Davitz (1953) reported that
tetraethylammonium chloride (TEA) altered the extinction rate of emotional
freezing responses. Brady (1953) subsequently reported that TEA completely
eliminated all activity throughout the conditioning sessions and had no differen-
tial or specific effect upon conditioned suppression per se. Similarly, the same
drug may have different action times across individual subjects and corre-
spondingly variable effects upon baseline and CS response rates. Brady
(1956b), for example, reported that the chronic (daily) administration of reser-
pine consistently increased response rates and attenuated suppression during
the CS but concomitantly decreased response rates during the intertrial inter-
val. The time course of the reserpine effect over sessions was highly variable
across subjects, ranging from rapid and abrupt attenuation to slow and gradual
transitions in suppression ratios over trials and sessions. In comparison,
chronic injection of d-amphetamine, a stimulant, enhanced conditioned sup-
pression but also increased baseline rates during the intertrial interval.

Phenothiazines. Chlorpromazine and other phenothiazines are classified as

major tranquilizers, and they are clinically used as antipsychotic agents. Appel
Chapter 1: Review of Classical-Operant Conditioning 55

(1963) and Torres (1961) reported attenuation of suppression by chronic admin-

istration of phenothiazine compounds (also see Lepore et al., 1974). However,
an acute or single injection of phenothiazines apparently has little if any effect
(Boren, 1961; Kinnard et aI., 1962; Ray, 1964; Cicala and Hartley, 1967; Tenen,
1%7; Palfai and Cornell, 1968; but also see Hunt, 1956; and Lauener, 1963).

Reserpine. Reserpine was once used clinically as a major tranquilizer but has
now been replaced by more effective drugs (such as butyrophenones or thio-
xanthenes). The attenuation of suppression by chronic administration of reser-
pine reported by Brady has been replicated by Weiskrantz and Wilson (1955),
Mason and Brady (1956), Appel (1963), Ray (1964), Frey (1967), and Wilson et
al. (1970). However, at least one study has found that chronic administration
has no effect upon suppression (Yamahiro et aI., 1961). Acute injection of re-
serpine may paradoxically enhance suppression (Valenstein, 1959) or have no
discernable effect (Kinnard et aI., 1962).

Benzodiazepines. Chronic administration of the minor tranquilizers or anti-

anxiety agents of the benzodiazepine family (e.g., librium) has no apparent ef-
fect upon suppression (Stein and Berger, 1969). Other investigators, however,
have reported an attenuation by the acute administration of benzodiazepines
(Lauener, 1%3; Tenen, 1967; Cicala and Hartley, 1967; Holtzman and Villar-
real, 1%9; Scobie and Garske, 1970; Maser and Hammond, 1972; Miczek,
1973; but also Stein and Berger, 1%9).

Meprobamate. Chronic administration of meprobamate seems to attenuate

conditioned suppression (Corson and Corson, 1%7), whereas acute administra-
tion apparently has no clear-cut effect (Hunt, 1957; Lauener, 1%3; Ray, 1964).

Barbituates. Current data indicate that suppression is attenuated by barbi-

tuates in both chronic (Corson and Corson, 1967) and acute administration
(Lauener, 1%3; Tennen, 1967; except Stein and Berger, 1%9).

d-Amphetamine. The stimulant d-amphetamine has been reported to increase

conditioned suppression when used chronically (Brady, 1956a). Glick (1%9),
using acute administration, also reported increased suppression, whereas Cap-
pel et al. (1972) found an attenuation of suppression as a direct function of d-
amphetamine dosage. Miczek (1973), however, found that suppression was un-
affected by various dosages (also see Lauener, 1963).

Scopaiomine. Scopalomine is an anticholinergic drug that does not seem to

affect conditioned suppression in chronic administration (Evans and Patton,
1970). A single injection of scopalomine has been reported to reduce suppres-
sion during extinction tests (Avis and Pert, 1974; Pert and Avis, 1974; addi-
tional studies by Vogel et aI., 1967; Daly, 1968; Berger and Stein, 1969a, 1%9b;
Goldberg et aI., 1971; 1972). Miczek (1973), however, reported no alteration of
suppression by scopalomine injections.
56 Section I: Concurrent Classical and Operant Conditioning

5-hydroxytryptamine (5-HTP). Suppression seems to be increased by acute in-

jection of 5-HTP, which presumably acts indirectly through an increase in brain
serotonin (Wise et aI., 1970). In a similar vein, Hartmann and Geller (1971)
found an attenuation of suppression by p-chlorophenylalanine, which reduces
the serotonin level, and a blunting of the attenuation by additionally injecting
5-HTP to increase serotonin (also see Frey, 1967).

Cannabis. Relatively recent pharmacological studies suggest that suppression

may be attenuated by the marihuana homologue pytahexyl (Abel, 1969). Gon-
zalez et aI. (1972) also reported that the acute administration of marihuana ex-
tract blocked the acquisition and retention of conditioned suppression, and that
chronic injection seemed to disrupt the retention but not the acquisition of sup-
pression.

Drugs as the UCS. A novel varient of the effect of drugs on conditioned sup-
pression was introduced by Goldberg and Schuster (1967), who used drugs
rather than shock as the classical conditioning DCS. Lever pressing by mor-
phine-addicted monkeys was suppressed during a tone paired with nalorphine,
a morphine antagonist. The suppression was accompanied by increased saliva-
tion, bradycardia, and emesis. Suppression was elimianted by substituting a
tone paired with saline in phase 2, and was reconditioned with tone-nalorphine
pairings again in phase 3. The nalorphine conditioned suppression then per-
sisted even when the monkeys were no longer physically dependent upon mor-
phine (Goldberg and Schuster, 1970). Suppression during a tone paired with ly-
sergic acid diethylamide (LSD) has also bee reported, with generalization to
similar tones during extinction test trials (Cameron and Appel, 1972b). The
LSD conditioned suppression was replicated by Cameron and Appel (1972a),
who also reported suppression during a light paired with the tranquilizer chlor-
promazine. Cameron and Appel pointed out that the conditioned suppression
by nonaversive tranquilizers and hallucinogenics does not fit the conditioned
fear or emotional state interpretations of classical-operant interactions. More-
over, chlorpromazine apparently has the ironic effect of inducing conditioned
suppression when used as a DCS but attenuating suppression when used as a
tranquilizer (also see electroconvulsive shock as a DCS; Winocur and Mills,
1970).

Cardiovascular Responses and Conditioned Suppression

Heart rate, blood flow and other cardiovascular responses, like conditioned
suppression, have frequently been used as potential indicants and measures of
fear. If both lever-pressing rates and cardiovascular responses were equally ac-
curate measures of acquired fear, then elementary logic would suggest that
conditioned suppression should be quite closely correlated with heart rate, sys-
tolic and diastolic blood pressure, etc. In 1964, Stebbins and Smith compared
suppression of operant responses with heart rate and blood flow in monkeys.
Lever pressing was suppressed during the CS, with a marked increase in heart
rate, blood flow rate, and general activity. (Heart rate and blood flow also in-
Chapter I: Review of Classical-Operant Conditioning 57

ADAPTATION ACQUISITION

+40

+60

w\~4~1~U~'ll
!
+40

mMl
+20
>-
I-~
-UJ 0
~C>
t;~-20

W1
ct:r
u-4O
C>::l!
~!..
UJ - 60
-80
-100

+60 I

,~lw~~1.
+40

~~ +20
ctUJ
Q: C> 0

~
~1~~ ~~U ~~
UJZ
(/)ct
Z:r-20
~~-4O
ffl~
Q:
-60
-80 I
-100 ~

I ' "'!!'" I IP-IJ-LlP-IfJ---J

2 3 2 3 4 5 6 7 8 9 10 II 12 13 14 15 20 30 40 50
Figure 1.6. Comparison of changes in systolic and diastolic blood pressure, heart rate, EMG activ-
ity, and operant suppression in Monkey S-IOO during Estes-Skinner procedure. Each data point
represents successive I-min intervals of each 3-min CS. EMG recorded with lumbar paraspinal
muscle electrodes, and blood pressure recorded with femoral artery catheter. Data from Brady et
al. (1970).
 58 Section I: Concurrent Classical and Operant Conditioning

creased during the suppression of avoidance extinction rates during CS-shock

pairings.) Stebbins and Smith (1964) therefore proposed an expanded definition
of CER to include the large alterations in the autonomic system.
In opposition, de Toledo and Black (1966) reported relatively independent
conditioning of heart rate and suppression in rats, and suggested that the diver-
gent effects would pose a problem for the use of responses as measures of fear.
Divergent acquisition rates of conditioned suppression and cardiac responses
in rats was also reported by Parrish (1967; also see Smith and Nathan, 1967;
Nathan and Smith, 1968).
Brady et al. (1969) measured heart rate, blood pressure, and lever-pressing
rates during the acquisition, extinction, reconditioning, and reextinction of sup-
pression in monkeys. In general, suppression of lever pressing conditioned and
extinquished more rapidly than the cardiovascular responses, with rather di-
vergent patterns of lever pressing, heart rate, and systolic and diastolic blood
pressure during the CS. Brady et al. also concluded that the causal indepen-
dence of physiological and behavioral measures argued against an interdepen-
dence of behavioral and physiological events or the theoretical primacy of one
or the other response system. The experiment was then replicated with the ad-
ditional analysis of electromyographic activity (EMG) by Brady et al. (1970;
Brady, 1971). Tonic and phasic muscular activity was unrelated to cardiovascu-
lar changes in simple classical conditioning or to conditioned suppression and
cardiovascular responses in the Estes-Skinner procedure (Figures 1.6 and 1.7).
Both EMG acceleration and deceleration were associated with virtually all pos-
sible combinations of increased and decreased heart rate, systolic and diastolic

'i t..- "tIAL 10

.,.,.
~ ?
:? t

Figure 1.7. Amplified polygraph recordings of blood pressure and EMG activity during the 10th
and 20th conditioning trials using a 3-min CS paired with shock superimposed upon operant lever
pressing by monkey. Data from Brady et a\. (1970) .
Chapter 1: Review of Classical-Operant Conditioning 59

blood pressure, and conditioned suppression (also see DeVietti and Porter,
1969; Willi, 1969; Zeiner et al., 1969; de Toledo, 1971; Borgealt et al., 1972;
Dantzer and Baldwin, 1974; and, in humans, Di Guisto et aI., 1974.)

Negative Classical Conditioning Scheduled

with Negative Operant Conditioning
The suppression of food reinforced operants by a concurrent negative classical
conditioning procedure clearly has been the only classical-operant combina-
tion to be studied in some detail. Early investigators pointed out, however, that
the original Estes-Skinner procedure was only one of four general methods in
which a positive or negative Pavlovian procedure could be superimposed upon
positive or negative operant reinforcement schedules (e.g., Estes, 1948; Brady,
1961). The four-fold analysis of classical-operant interactions then progressed
to an arithmetic complement. If CS-shock pairings disrupted operant re-
sponses maintained by positive reinforcers, then the same classical condition-
ing procedure might facilitate responses maintained by negative reinforcers.
Those of us working with conditioned suppression in the 1960s especially en-
joyed the overall simplification and form of such an analysis, with diagrams as-
signing upward pointing arrows for increased response rates when the operant
and classical schedules had the same sign (or the same underlying state, expec-
tancy, motivation, value, etc.), but downward pointing arrows and decreased
rates when the two components had dissimilar signs. The results of early stud-
ies were quickly generalized, and rather overgeneralized, to support the pleas-
ing symmetry apparent in both the figures and logic of conditioned suppres-
sion/acceleration. We might preface the following review section by noting that
the initial theoretical symmetry is no longer especially tenable, or at least re-
quires substantial revision in the light of subsequent experimentation.
The earliest study of avoidance rates during superimposed negative classical
conditioning was reported by Sidman et al. (1957). In that study, monkeys were
trained on a free-operant ("Sidman") avoidance schedule in which each lever
press postponed a 1.0-sec, 5.0-ma shock for 20 sec. Lever-pressing rates in-
creased during both control and CS intervals when a clicker was terminated by
an unavoidable shock of the same intensity and duration of the avoidable
shock. The acceleration gradually declined toward original baseline rates over
sessions, with a somewhat slower return to baseline during the CS. In a second
phase, CS-shock pairings were combined with extinction of the avoidance re-
sponse, and again lever-pressing rates increased during both the intertrial inter-
val and the CS. Throughout the study, lever pressing was similarly increased
by unsignaled or "free' shock. The initial data then suggested that the effects of
negative Pavlovian procedures were not simply restricted to the CS, but also
resulted in overall increased responding throughout the session that was not
substantially different from the effects of unsignaled shock. Sidman et a1. then
concluded that the change in avoidance rates was primarily dependent upon the
response-independent shock rather than a classically conditioned effect of CS-
60 Section I: Concurrent Classical and Operant Conditioning

ues pairings (also see Sidman, 1958, 1960a, I%Ob; Waller and Waller, 1963;
Belleville et aI., 1%3; Grossen and Bolles, 1968; Kamano, 1968; Rescorla,
1%8a).
Table 1.2 summarizes the results of subsequent studies of avoidance behav-
ior during eS-shock pairings. As with the Estes-Skinner procedure, the ex-
periments have predominantly examined the effects of classical conditioning
variables rather than operant or subject variables. The results of anyone vari-
able are again frequently quite conditional upon the selection of other parame-
ters. Indeed, avoidance rates are either suppressed or accelerated by CS-
shock pairings conditional upon specific parametric interactions, in contrast to
the far more consistent conditioned suppression of positively reinforced
operants.

Table 1.2 Negative classical-negative operant conditioning

Variables Studies Results

Classical Conditioning
U CS Parameters
intensity 5 conditional results
duration 4 acceleration as a direct function
probability 2 acceleration/suppression with stimuli paired!
unpaired with shock
type ofUCS 4 conditional upon response
CS Parameters
duration 3 conditional upon other variables
generalization graded acceleration
compound stimuli no effect
backward conditioning suppression a function of UCS-CS interval
Total 21

Operant Conditioning
Reinforcement Schedules
shock intensity 3 conditional upon other variables
response-shock interval 4 perhaps suppression as direct function
partial reinforcement acceleration during CS
Concurrent Behaviors increased during operant suppression
Pre-schedule Change Stimuli acceleration
Total 10

Subject
Central Nervous System
lesions no effect of septal lesions
drugs 3 conditional
Cardiovascular Responses 1 increased during CS
Total 5
Chapter 1: Review of Classical-Operant Conditioning 61

Classical Conditioning Variables

ues Parameters
ues Intensity. Avoidance rates are at least partially controlled by the intensity
of the Pavlovian ues. In a study by Martin and Riess (1%9), a es was paired
with 0.25,0.58, 1.90, or 4.90-ma shock and then superimposed without shock
upon the extinction of shuttle box avoidance responses. In general, the extin-
quishing avoidance response was slightly accelerated as a function of the for-
mer ues intensity. A series of papers by Hurwitz and Roberts (1969, 1971) and
Roberts and Hurwitz (1970) also indicate that operant avoidance rates during a
es are related to the ues intensity.
Interactions between the intensity of the ues and the negative operant rein-
forcer were studied in a 2 x 3 factorial design by Scobie (1972). First, hurdle
jumping avoided either a 0.6- or a l.3-ma shock in two different groups, with a
superimposed es terminated by a ues of approximately 50%, 100%, or 200%
of the avoidable shock. In the 0.6-ma group, avoidance rates increased when
the es was paired with the relatively lower ues of 0.3 or 0.6 rna, but were
suppressed by the higher ues of 1.3 mao In the second group, hurdle jumping
maintained by a 1.3-ma shock was only marginally and transiently increased at
all ues intensities. Scobie thus concluded that the behavioral effects were not
simply determined by the absolute ues intensity, but also by the shock inten-
sity relative to the operant reinforcer.
In hindsight, avoidance rates during es -shock pairings do seem to be af-
fected by ues intensity. However, the direction of the operant rate change is
apparently dependent upon additional variables and has been variously re-
ported to be a direct function (Martin and Riess, 1%9) or an inverse function
(Scobie, 1972) of ues intensity.

UCS Duration. The duration as well as the intensity of the ues has also been
reported to determine avoidance rates during the es; Riess and Farrar (1973b)
have reported avoidance acceleration as a direct function of ues duration. Re-
sponse latency was an inverse function, and response amplitude and frequency
were direct functions of ues durations of 0.05,0.30, 1.00, and 3.0 sec. Riess
and Farrar also discussed a variety of negative and positive correlations be-
tween the different response measures. Related experiments using a "transfer
of training" design would also suggest that avoidance responses previously
conditioned to a discrete visual stimulus might be elicited by tones previously
paired with shock, as a direct function of shock duration Overmeir and Leaf,
1965; Overmeir, 1%6a, 1966b).

Probability of the UCS. Rescorla and Lo Lordo (1965) described a group of

experiments in which avoidance rates were accelerated during one stimulus
paired with shock but suppressed during a second stimulus with a zero shock
probability. The differential effects of shock probability were further replicated
with a discriminative classical conditioning procedure and a conditioned in-
hibition procedure. In one experiment, for example, es 1 was consistently
62 Section I: Concurrent Classical and Operant Conditioning

paired with shock, and es 2 was terminated only byeS 1. The hurdle-jumping
responses were initially accelerated during subsequent es 1 test trials, with a
progressive reduction to control rates with continued testing. In contrast,
avoidance rates were consistently inhibited or suppressed during test trials with
es 2. This acceleration and suppression of avoidance rates during stimuli
paired and unpaired with shock has been replicated by Weisman and Litner
(1969) using a wheel turning avoidance response by rats. The pairing of es 2
with es 1 in the Rescorla and Lo Lordo experiment is analogous to traditional
second-order classical conditioning. The test results for the conditioned in-
hibition group would then suggest a persistent suppression during the second-
order es 2 but transient acceleration during the first-order es 1 formerly paired
with shock. The avoidance rate changes are then similar to the changes in posi-
tively reinforced operants during second-order conditioning, with different
rates and patterns ofresponding during es 2 and es 1 in each case.

Type of UCS. A comparison of stimuli terminated by a loud hom and shock

was reported by Lo Lordo (1967). Avodance rates were increased during the
es paired with a 108-db Klaxon hom, and also during the es paired with
shock. In comparison, avoidance rates were decreased slightly during a stimu-
lus explicitly unpaired with shock and were unchanged by a stimulus explicitly
unpaired with the hom. The increased rates during eS-horn pairings were op-
posite to the unconditioned freezing responses elicited by the hom ues. Lo
Lordo concluded that the fear elicited byeS-horn conditioning summated with
the similar fear motivating the shock avoidance responses.
Studies with rats have also found increased shuttle box avoidance responses
(Riess, 1970a, 1970b) but decreased lever-press avoidance rates (Riess, 1971)
during a light paired with a buzzer.

Conditioned Stimulus Parameters

Relative Conditioned Stimulus Duration. The divergent reported effects of ac-

celeration and suppression of avoidance rates may be dependent upon the dura-
tion of the es relative to the avoidance response-shock interval. Pomerleau
(1970) examined response rate as a function of both es duration and the re-
sponse-produced delay of shock in monkeys. Using an accumulating avoidance
schedule (Field and Boren, 1963), lever pressing was reinforced by time-out
from shocks repeating every 4.7 sec, with each response-produced time-out
added to the previously accumulated shock-free time. The procedures were
systematically replicated over blocks of sessions using avoidance time-out in-
tervals of 1.0,2.5,5.0, and 20.0 sec compared to superimposed es durations of
12,48, and 84 sec. With a 1.0-sec time-out, for example, avoidance rates de-
creased throughout the 12-sec es but followed a biphasic acceleration-decel-
eration pattern throughout the longer 48- and 84-sec ess. Increasing the dura-
tion of the response-produced time-out relative to es duration systematically
decreased avoidance response rates, with responding suppressed at all es du-
Chapter 1: Review of Classical-Operant Conditioning 63

rations when the scheduled time-out was 20 sec. As Pomerleau suggested, rela-
tive CS duration might then be a fundamental parameter, with analogous ef-
fects upon shock avoidance as well as food reinforced response rates.
However, the acceleration of avoidance rates with even relatively long CS
durations is also conditional upon the UCS intensity; that is, avoidance rates
may be suppressed even when the CS duration is relatively longer than the
shock avoidance interval if the UCS intensity is also greater than the intensity
of the avoidable shock (Bryant, 1972).
The effects of relative CS duration in superimposed trace classical condi-
tioning have been described by Shimoff (1972b).

CS Generalization. Avoidance generalization gradients about tones explicitly

paired and unpaired with shock were reported by Desiderato (1969). In off-the-
baseline sessions, an 800-Hz tone or a clicker were explicitly paired or unpaired
with shock in a counterbalanced design across groups. A single generalization
test session consisted of nonshock presentations of 5-sec tones at 250, 450,800,
1500, and 2600 Hz during extinction of hurdle jumping responses. The extin-
quishing avoidance rates were increased for the subjects trained with tone-
shock pairings ("excitatory conditioning"), but decreased for subjects trained
with the tone explicitly unpaired with shock {"inhibitory conditioning"). The
generalization of acceleration and suppression was a U-shaped function across
tone frequencies, with a relatively sharper gradient of excitatory acceleration
than inhibitory suppression.

Compound CS. A compound stimulus test for external inhibition and disinhibi-
tion of avoidance rates was described by Rescorla (1967a). Initially, avoidance
rates decreased at the onset ofa 30-sec CS ("delay of inhibition") followed by a
progressively accelerating rate ("fear conditioning"), which in tum was fol-
lowed by an abruptly decreased response rate just prior to es offset (adventi-
tious punishment by the Pavlovian UeS). A 5-sec flashing light was then pre-
sented during the early portion of the CS with response suppression or the
latter portion with response acceleration. Avoidance rates were relatively un-
changed during both compound procedures, with no indication of "disinhibi-
tion" of the early "inhibition of delay" or "external inhibition" of the latter
"conditioned fear."

Backward conditioning. Suppression or inhibition of Sidman avoidance behav-

ior has recently been reported during a superimposed CS formerly used in
backward UCS-CS conditioning. Maier et al. (1976) found that the suppression
was dependent upon the duration of the UCS-CS interval; that is, the shock
avoidance responses were decreased during the CS when the UCS-CS interval
was 3 sec but were unchanged when the interval was 30 sec. These results ten-
tatively suggest at least some conditions in which backward classical condition-
ing may alter operant avoidance responses.
64 Section I: Concurrent Classical and Operant Conditioning

Operant Conditioning Variables

Avoidance Schedule Parameters

Intensity of the Avoidable Shock. The studies ofUeS intensity by Hurwitz and
Roberts (1971) and Scobie (1972) also directly manipulated the intensity of the
avoidable shock. Hurwitz and Roberts reported that avoidance rates reinforced
with a 0.8-, 1.4-, or 2.0-ma shock were, respectively, suppressed, inconsis-
tently changed, and slightly accelerated during the es. Roberts and Hurwitz
(1970) used a similar procedure but "suspended" the operant avoidance sched-
ule during the es. Response rates were then suppressed at all shock intensities
during the es plus time-out procedure. Scobie (1972) reported that avoidance
rates reinforced by a 1.3-ma shock were relatively independent of ues inten-
sity, whereas avoidance reinforced by a lower 0.6-ma shock was generally in-
creased by low ues intensities but suppressed by a relatively high-intensity
ues.

Response-shock Interval. The previously cited study by Pomerleau (1970) has

shown that avoidance rates during a es are dependent upon the relative dura-
tion of the response-produced delay of shock. The data suggest that avoidance
rates may be suppressed when the response-shock interval is long relative to
es duration, but may be facilitated when the shock delay is shorter than the
es. Brady et al. (1967) also reported that avoidance rates during eS-shock
pairings were dependent upon the operant response-shock interval (also see
Brady and Hunt, 1955). Scobie (1972), however, reported that the influence of
the response-shock interval is conditional upon ues intensity.

Partial Reinforcement of Avoidance Responses. The avoidance rates main-

tained by partial avoidance reinforcement schedules are also disrupted by nega-
tive classical conditioning procedures. Houser (1973) used a procedure in
which lever pressing by monkeys was maintained on a response-shock 20-sec
shock-shock 2-sec schedule, with only 20% of the due shocks actually deliv-
ered. The partially reinforced avoidance rates were substantially increased dur-
ing superimposed eS-shock pairings. The accelerated es avoidance rates
were subsequently replicated across drug and control treatments.

Concurrent Behaviors and Avoidance Rates

A single paper has examined the changes in concurrent responses as well as
operant avoidance responses during classical-operant combinations. Roberts
et al. (1977) recorded 10 different behaviors during initial Sidman avoidance
conditioning and during a superimposed preshock stimulus. Successful avoid-
ance responding consisted of a variety of different topographical lever-pressing
responses, with a reduction in the diverstiy as well as the frequency of avoid-
ance responses during the es. The changed rates and patterns of avoidance be-
Chapter 1: Review of Classical-Operant Conditioning 65

haviors were associated with an increase in other concurrent responses during

the CS. These concurrent response changes seem to be consistent with the sug-
gestion that varients of the Estes-Skinner procedure may control multiple re-
sponses within the subject's repertoire.

Operant Avoidance Schedules as the Aversive "ues"

A voidance rates during stimuli terminated by schedule changes to higher
and lower densities of avoidable shock were examined by Henton (1970).
Avoidance responses, maintained by a response-shock = shock-shock = 30
sec schedule, were irregularly increased during 30-sec stimuli terminated with
(1) a response-shock = shock-shock = 3 sec schedule, and (2) time-out from
avoidance (Figure 1.8). Avoidance rates were equally variable when the dura-
tion of each preschedule change stimulus was changed to 15 sec or 60 sec, with
averaged avoidance rates tending to be a direct function of stimulus duration.

Stimulus A StimulusB
53

~~ {f
2'0~
1.0 :.
0.0 b
• •
• •
- 1.0 .L .L .L

0 52
l-
<{ 3.0
•
•
...•
0:::
Z 2.0 •
0 \.0
i= ••
u 0.0
W
~-1.0 .1. .L

Z
S,1
2.0 •
•
•
1.0 •
O.OH~-I-~

if!I
BLOCKS OF TEN TRIALS
Figure 1.8. Comparison of increase in avoidance response rates by two stimuli terminated by op-
posite schedule changes. Avoidance responding by monkeys was maintained by a response-
shock = shock-shock = 30 sec schedule, stimulus A was terminated by a response-
shock = shock-shock = 3 sec schedule; and stimulus B was terminated by time out from avoid-
ance. Data from Henton (1970).
66 Section I: Concurrent Classical and Operant Conditioning

Subject Variables

Central Nervous System

Lesions. A single study by Dickinson and Morriss (1975) indicated that septal
lesions in rats have no effect upon the acceleration of wheel-turning avoidance
responses during a CS formerly paired with shock.

Drugs. Houser and associates have reported a series of papers analyzing the
effects of various pharmacological agents upon avoidance behaviors.

Chlordiazepoxide. Houser et al. (1975) reported that chlordiazepoxide (a

minor tranquilizer) had no effect upon avoidance rate, heart rate, and general
activity of dogs during the aversive CS. The drug treatments did, however,
suppress baseline avoidance rates during the intertrial interval.

d-Amphetamine and a-methyl-p-tyrosine. Chronic administration of the stim-

ulants d-amphetamine and a-methyl-p-tyrosine had no reliable effect upon con-
ditioned acceleration of avoidance in monkeys (Houser, 1973). Again, how-
ever, each drug fundamentally altered baseline avoidance rates, with recovery
of pre-drug baselines during drug withdrawal sessions.

Scopolamine. A comparison of the central acting scopolamine hydro bromide

and the peripheral acting scopolamine methylnitrate was reported by Houser
and Houser (1973). Initially, avoidance rates were accelerated (four monkeys)
or suppressed (two monkeys) during the CS. The drugging procedure then con-
sisted of daily injection of drug or saline in three-session blocks. The pattern of
avoidance responding was reversed by scopolamine hydrobromide: avoidance
rates were suppressed in subjects with the former acceleration pattern, and ac-
celerated in subjects with the former suppression pattern. Scopolamine methyl-
nitrate had no reliable effect. The authors suggested that the central acting
scopolamine hydrobromide may reduce fear directly by interfering with
cholinergic systems, or indirectly through alteration of memory processes.

Cardiovascular Responses
Heart rate, blood flow, and general activity in monkeys were recorded by
Stebbins and Smith (1964) during CS-shock pairings superimposed upon
avoidance extinction. Heart rate and blood flow consistently increased during
all classical conditioning trials. In contrast, lever pressing accelerated during
the early trials but was suppressed over the latter conditioning trials. The data
thus suggest some independence of cardiovascular and avoidance rates during
CS-shock pairings.
Chapter 1: Review of Classical-Operant Conditioning 67

Positive Classical Conditioning Scheduled

with Positive Operant Conditioning
The operant changes during eS-shock pairings described by Estes and Skinner
were immediatley replicated with eS-food pairings by Estes (1943, 1948).
Estes (1943) reported that the "anticipation" of response-independent food
would also alter the rate of positively reinforced operant responses. The
operant rates were substantially increased during the positive classical condi-
tioning procedure, in striking contrast to the conditioned suppression in the
Estes-Skinner experiment. More specifically, bar pressing by rats was rein-
forced on a fixed interval (FI) 4-min food reinforcement schedule, with off-the-
baseline pairings of a I-min 60-Hz tone with the response-independent delivery
of one food pellet. Extinction lever-pressing rates were increased during 10-min
test trials when the tone was merely presented without food. Estes (1948) re-
versed the initial order of training on the operant and classical conditioning
components, and reported similar acceleration of extinction operant rates dur-
ing es test trials. Estes (1948) therefore suggested that the CS-food pairings
may set the occasion for an anticipatory state that could have discriminatory
effects upon instrumental responses maintained by the same reinforcer.
European experiments by Konorski and Miller antedated the work of Estes
but were largely unknown to Western psychologists until publication of the
popular and comprehensive review by Konorski (1967). Konorski and Miller
found that an instrumental leg-lifting response was immediately suppressed
during a 15-sec CS paired with food, with a concomitant increase in orienting to
the DeS feeder and salivation. Suppression invariably occurred in all subjects,
despite the loss of available response-contingent reinforcers throughout the
CS. In comparison, the instrumental response rates were relatively unchanged
during an "inhibitory" stimulus explicitly unpaired with food. The suppression
during positive classical conditioned stimuli was systematically replicated
across a variety of different experimental procedures. In contrast to Estes,
Konorski concluded than an appetitive type I or classical conditioned stimulus
not only fails to increase the frequency of an appetitive type II instrumental
response, but actively inhibits and decreases the instrumental response rate.
The subsequently reported effects of positive classical conditioning proce-
dures upon positively reinforced operants are summarized in Table 1.3. Al-
though fully as old as the comparable Estes-Skinner procedure, the behavioral
effects of superimposed positive classical conditioning procedures have re-
ceived considerably less experimental examination. The different experimental
frequencies are partially due to the traditional negative connotations of emo-
tions such as "fear," "anxiety," and "anger," and hence a greater theoretical
interest in negative rather than positive classical conditioned "emotions." Sec-
ond, positive classical conditioning is more commonly accepted as eliciting
specific conditioned responses that may be antagonistic to the recorded
operant, and thus complicate and impede the "proper" analysis of underlying
associative or emotional states. As with superimposed negative classical condi-
tioning procedures, studies using positive classical conditioning have been pre-
dominantly biased toward the analysis of the classical conditioning variables.
68 Section I: Concurrent Classical and Operant Conditioning

Table 1.3 Positive classical-positive operant conditioning

Variables Studies Results

Classical Conditioning
UCS Parameters
duration 1 no effect
type of UCS 7 conditional upon other variables
CS Parameters
duration 5 conditional, divergent
Total 13

Operant Conditioning
Reinforcement Schedules
response rate measures 1 conditional upon schedule
response accuracy 2 altered during CS
Concurrent Behaviors 2 inversely related to operant rate
Preschedule Change Stimuli 2 relative suppression

Total 7

Subject
Central Nervous System, drugs attenuation of suppression
Cardiovascular Responses unchanged during CS
Total 2

Finally, the behavioral effects are relatively mixed and conditional, and are
more comparable to the operant changes during negative classical-negative
operant schedules than to the conditioned suppression in negative classical-
positive operant schedule combinations,

Classical Conditioning Variables

ues Parameters

ues Duration. ues duration apparently has little effect upon the food rein-
forced operants during eS-food pairings. Lo Lordo (1971) reported that key
pecking by pigeons on a VI 2-min schedule was slightly increased during a es
paired with either 8-, 4-, or 2-sec access to grain. Response rates were rela-
tively unchanged during a second es explicitly unpaired with the grain ues.
Lo Lordo concluded that the es facilitation may be a summation of operant
responses plus orienting and approach responses elicited by the es presented
on the operant response key.

Type of ues. The conditioned facilitation reported by Estes (1943, 1948) was
replicated by Brady (1961) using intracranial stimulation (IeS) as the response-
independent reinforcer. Operant rates were clearly increased during the es-
Ies interval, with the facilitation effect extinquishing during classical condi-
tioning extinction.
Chapter I: Review of Classical-Operant Conditioning 69

In contrast, the conditioned suppression described by Konorski (1%7) was

immediately replicated by Azrin and Hake (1969). In that study, lever pressing
by rats was reinforced either with food or water, with the subjects then divided
into subgroups in which a 10-sec clicker or red light was terminated by food,
water, or rewarding intracranial stimulation of the median forebrain bundle and
hypothalamus. Operant responses reinforced with food were suppressed during
the 10-sec CS paired with food, water, or intracranial stimulation. Water rein-
forced response rates were also suppressed during the CS paired with food, but
were accelerated during the CS paired with water. Azrin and Hake hypothe-
sized that the conditioned suppression must be due to a general emotional state
of heigthened preparedness rather than competing body movements, head
movements, or freezing during the various CSs. The results were replicated
with slightly different procedures by Hake and Powell (1970) and Van Dyne
(1971).
The following stu9ies by Meltzer and Brahlek (1970), Henton and Brady
(1970), and Miczek and Grossman (1971) also used either qualitatively or quan-
titatively different response-dependent and response-independent reinforcers.

CS Parameters

CS Duration. The reported suppression and facilitation effects of appetitive

Pavlovian procedures seem to be partially determined by the relative CS dura-
tion. The first study ofCS duration was that of Meltzer and Brahlek (1970), who
found that the rate of lever pressing reinforced with food was a direct function
of the duration of a CS paired with either 10% or 25% sucrose solutions. Re-
sponding was accelerated during a 120-sec CS and suppressed during a 12-sec
CS, with intermediate and mixed response rates during a 40-sec CS. Further-
more, the pattern as well as the frequency of responses during the CS was also
dependent upon CS duration, with progressively decreasing response rates
throughout the 12-sec Cs, constant or slightly decreasing rates during the 40-
sec CS, but positively accelerating rates during the 120-sec CS.
The acceleration of DRL response rates is similarly conditional upon the du-
ration of a superimposed CS paired with food (Figure 1.9). Henton and Brady
(1970) reported that the low rate of DRL responding maintained with one food
pellet was relatively unchanged during a 20- or 40-sec CS paired with five food
pellets, but markedly accelerated when the CS was extended to 80 sec. The
response facilitation seemed to be related to the response sequences mediating
the "temporal inhibition" of DRL responding (see, for example, Chapter 5,
Experiment III).
Miczek and Grossman (1971) replicated the Meltzer and Brahlek study and
reported a slightly different function relating operant rates to CS duration.
Lever pressing by monkeys was suppressed during a 15- or 30-sec CS but rela-
tively unchanged rather than accelerated at longer CS durations of 1,2, and 3
min. Suppression of lever pressing by monkeys was also an inverse function of
CS duration in Experiment VII, Chapter 2.
Using pigeons, Smith (1974) reported a third relationship between CS dura-
70 Section I: Concurrent Classical and Operant Conditioning

" acquisition
+200 • ••
•••
•• & reacquisition
••
(I)
u
c
.-
•
••
- •
c
'E •
on
100
•••
"-
•
CD
01
•••
C
•
••
III
~
u
••
'#.
•••
•••••••••• •••
• 6. extinction
0

-50
20 40 80 sec

CSduration
Figure 1.9. Changes in DRL responding as a function of the duration of a CS paired with food.
Each data point is the mean ofthe final 25 conditoning trials in acquisition (filled circles), extinction
(open triangles), and reacquisition (filled triangles). Adapted from Henton and Brady: J. Exp. Anal.
Behav., 13, 205-209, 1970.

tion and response rate. Low baseline rates were accelerated, not suppressed,
during a 5-sec CS, and relatively unchanged during 30-,60-, and 120-sec CS-
food intervals. In comparison, high baseline rates were slightly accelerated dur-
ing the 5-sec CS, suppressed during the 30- and 60-sec stimuli, and unchanged
during the 120-sec CS.
Thus, the effects ofCS duration, like so many other variables, are not totally
consistent across experiments and would seem to be conditional upon addi-
tional parameters.

Operant Conditioning Variables

Schedule Parameters

Response Rate Measures. The previously cited studies have generally used re-
sponse rate measures and reported that the same VI or DRL baseline rate may
be altered by manipulating CS duration. In addition, Kelly (1973b) compared
the effects of CS-food pairings upon the high rate of responses generated by
RR schedules and the low response rates maintained by DRL schedules. The
high RR response rate was suppressed during the CS, independent of changes
in food deprivation or the local intertrial interval response rate. DRL rates were
accelerated, however, independent offood deprivation but partially dependent
upon local variations in the intertrial baseline rates.
Chapter 1: Review of Classical-Operant Conditioning 71

Response Accuracy. Herrnstein and Morse (1957) originally reported the altera-
tion of temporal or spaced responding in DRL schedules during a CS paired
with food. Pecking responses by pigeons were reinforced on a DRL 5-min
schedule, with a 2-min yellow light as the CS projected upon the pecking key.
Response-independent food was delivered 1-min after CS onset. The stimulus-
food pairings produced a generalized increase in responding throughout the ex-
perimental sessions, with a further acceleration during the CS by four subjects,
but a relative suppression for two subjects. Herrnstein and Morse proposed
that the initial change in responding must be related to then unknown variables,
with acceleration or suppression adventitiously maintained by correlations be-
tween the response-independent reinforcer and either key pecking or incompat-
ible behaviors, respectively.
The Herrnstein and Morse study was later replicated by Henton and Brady
(1970), with an additional analysis of the response distribution as a function of
CS duration. The DRL rates were relatively unchanged during a 20- or 40-sec
CS, but accelerated with a redistribution of interresponse times during the 80-
sec CS. As shown in Figure 1.10, the longer CS generated an increased fre-
quency of unreinforced interresponse times of 0-5 and 25-30 sec. The interre-
sponse time distribution shifted back to control patterns during classical condi-
tioning extinction, with a replication of the accelerated response rate during
CS-UCS pairings in reacquisition.

,
15th acquisition

I
I
Ill.
I
• • 'b. •
I
Ill.
I
I
Ill.
I I
I
I
Ill.
I
.tI
•
. I
'i'I
I
I

15th extinction

10 [ • ~t ~
.~I~b.~~.~I~b.~~.~lb.~__.~lb.~~.~I~b.~__I~I~b.~__1~1~1___'~I_I____~~lt_
0
_____

15th reacquisition
, tf
•I
I~
A
I
Ill.
II I
.:
II I

0-5 10-15 20-25 30-35 40-45

Inter-response ti me (sec)
Figure 1.10. Comparison of DRL interresponse times during SO-sec CS paired with food (filled
circles) and equivalent control periods (open triangles). Each data point is the distribution of re-
sponses in 5-sec bins over the last five trials of classical conditioning acquisition, extinction, and
reacquisition. Adapted from Henton and Brady: 1. Exp. Anal. Behav., 13,205-209, 1970.
72 Section I: Concurrent Classical and Operant Conditioning

Concurrent Behaviors
A number of investigators have proposed that the conditioned acceleration
and suppression of operant rates may be related to the temporal pattern of con-
current or competing responses elicited by the appetitive classical conditioning
(Herrnstein and Morse, 1957; Konorski, 1967; Henton and Brady, 1970; Hen-
ton and Iversen, 1973). More recently, Lo Lordo et al. (1974) found that the
acceleration of DRL responding by pigeons was related to an increased rate of
approach and orienting responses to the visual CS presented on the operant
manipulandum. In contrast, a treadle-pressing response reinforced on the
same DRL schedule was suppressed along with the increase in orienting re-
sponses to the pecking key during the visual CS. By comparison, the key-peck-
ing and treadle-pressing operants were inconsistently affected by a similar 10-
sec auditory CS paired with food. The results are consistent with the previous
suggestion that conditioned acceleration may be related to the summation of
topographically similar responses (Lo Lordo, 1971).

Increased Reinforcement Rate as the Positive "UCS"

The previously described studies by Pliskoff (1961, 1963) reported generally
decreased rates during stimuli paired with response-independent schedule
changes to higher operant reinforcement rates relative to stimuli terminated by
lower reinforcement rates (see p. 51).

Subject Variables
Central Nervous System
A single study (Miczek, 1973) has examined the effects of various pharmaco-
logical agents upon operant rates during appetitive classical conditioning in
both rats and monkeys. Response rates were suppressed during a IS-sec tone
paired with sweetened milk (rats) or food pellets (monkeys); the suppression
was attenuated by the stimulant d-amphetamine, but unaffected by benzodiaze-
pines or scopolamine. In contrast, the conditioned suppression during CS-
shock pairings was attenuated by the benzodiazepines, but unchanged by
d-amphetamine or scopolamine. The divergent drug effects were attributed to a
differential susceptibility of the various competing responses elicited by stimuli
paired with food and shock.

Cardiovascular Responses
A two-monkey experiment was described by Kelly (1973a) in which heart
rate and blood pressure were recorded during a 3-min CS paired with food dur-
ing operant lever-pressing sessions. Lever pressing was transiently suppressed
during the initial trials, but relatively unchanged or slightly accelerated
throughout the latter trials of the study; heart rate and blood pressure were un-
affected by the CS-UCS pairings throughout the experiment. Kelly proposed
that the conditioned suppression may be determined by the monkeys adopting
a "why-work-when-you-can-get-it-for-nothing" strategy.
Chapter 1: Review of Classical-Operant Conditioning 73

Positive Classical Conditioning Scheduled

with Negative Operant Conditioning

The analysis of avoidance behaviors during positive classical conditioning pro-

cedures is the most neglected classical-operant schedule. A total of three re-
cent papers have examined shock avoidance rates during superimposed CS-
food pairings. In the first study, Coulson and Walsh (1968) reported that a I-min
CS paired with 10 sucrose pellets had no effect upon lever pressing by rats rein-
forced on a FI or VI shock avoidance schedule. The avoidance rates increased
during the CS, however, when the food deprivations were changed from 80% to
70% of normal body weight. Coulson and Walsh concluded that the response-
independent food had a behavior energizing or motivational effect that would
facilitate ongoing operant behaviors.
In contrast, Davis and Kreuter (1972) found a change in the patterning but
not the overall avoidance rate in rats during superimposed appetitive classical
conditioning. The lever-pressing rates were suppressed at CS onset, followed
by delivery of shock on the programmed avoidance schedule, in turn followed
by a postshock burst of avoidance responses in the latter portion of the CS.
Davis and Kreuter proposed that the cyclical suppression and acceleration pat-
tern may be most adequately analyzed in terms of interacting schedule contin-
gencies.
Using monkeys, Henton (1972; also see Chapter 2) also reported that condi-
tioned acceleration or suppression of avoidance rates was partially controlled
by the CS-food contingencies and the spatial location of the CS and UCS.
Additional studies have used a "transfer of training" design in which a for-
mer CS previously paired with food is combined with the extinction of operant
avoidance responding. Grossen et al. (1969), for example, reported a suppres-
sion of the avoidance extinction rate during tones previously paired with either
the response-independent or response-contingent delivery offood. The authors
argued that the suppression could not be due to competing food retrieval re-
sponses, since the avoidance rates were equally disrupted by tones terminated
by food in a classical conditioning box (requiring no retrieval responses) or in
an alley runway (requiring a specified retrieval response).
A series of papers from Overmeir's laboratory have also used the transfer of
training procedure and similarly describe a suppression of avoidance extinction
rates during the stimuli previously paired with an appetitive UCS (Bull, 1970;
Overmeir and Bull, 1970; Overmeir et al., 1971). Overmeir and associates inter-
pret the suppressed avoidance rates as an interaction between underlying moti-
vational systems.
Two additional studies have reported accelerated avoidance rates during
stimuli terminated by time-out from the operant avoidance schedule in rats
(Baron and Trenholme, 1971), and monkeys (Henton, 1970). The avoidance
rates were highly variable during the pre schedule change stimuli in both stud-
ies.
74 Section I: Concurrent Classical and Operant Conditioning

Concluding Remarks
In general summary, the foregoing review would suggest an almost infinite
complexity of parametric interactions within the various combinations of clas-
sical and operant schedules. Within each major category, consistent behavioral
effects have been reported for only the most powerful classical conditioning
variables, with the effects of many manipulations conditional upon other pa-
rameters, or seemingly inconsistent and divergent within and across experi-
ments. The inconsistent behavioral effects of various parameters would clearly
emphasize the still preliminary nature of current behavioral analysis. The obvi-
ous disparity in research efforts across the four types of schedule combinations
further emphasizes the incomplete and unfinished character of classical-
operant schedule analysis. The absence of empirical facts as well as the pres-
ence of inconsistent results will hopefully generate a more adequate experimen-
tal examination rather than a burgeoning accumulation of "heuristic" and hy-
pothetical opinions. Fortunately, the contemporary literature does indeed
provide a number of variables that may be manipulated in concert to purposely
control operant rates during superimposed classical conditioning procedures-
UCS intensity and CS duration are only two examples. The sheer quantity of
controlling variables, and the resultant intricate interactions between variables,
however, sharply limit the accuracy of global generalizations and molar expla-
nations of classical-operant effects.
One assumption gone wrong, for instance, is the once firm belief that the
Estes-Skinner procedure may be used as a model of classical-operant interac-
tions. Early results could be interpreted as the internalized summation of simi-
lar motivational and emotional states, or subtraction of dissimilar states. Clas-
sical-operant interactions could then be reduced to a few molar generalizations
in the beginning analysis. Conditioned fear would disrupt positive motivations
and enhance fear motivation. In a similar manner, positive conditioned emo-
tions would enhance positive operant motivations but disrupt fear motivation.
The various internal state doctrines thus neatly prescribed the relationship be-
tween emotions and motivations, and, with additional assumptions and opera-
tional definitions, could arguably describe the alteration of behavioral and
physiological indicants of underlying excitatory and inhibitory processes.
The ensuing years found a number of studies attempting to find the causal
functions relating emotional behaviors to underlying states via physiological re-
sponses. The expected close correlation between physiological and behavioral
measures failed to materialize for virtually all classical-operant combinations,
and the conception of behaviors as passive measures of underlying states began
to fail. Internal state theories have not been eliminated, however, but merely
reinterpreted, perhaps citing Canon's early observations that many physiologi-
cal and autonomic responses are after all identical in emotional states. Overt
behavior might therefore be a true measure of covert mental states if not covert
neurophysiological responses. Nevertheless, subsequent investigators have re-
ported similar operant response patterns within different classical-operant
schedules, suggesting that overt behavior is as undifferentiated as covert re-
sponses across the various "emotional" conditioning procedures.
Chapter 1: Review of Classical-Operant Conditioning 75

Additional experiments further question the utility of behaviors as measures

of underlying processes. As only one example, the suppression of operant re-
sponses during the Estes-Skinner procedure is not associated with correspond-
ing changes in food motivational states as measured by eating (Ayres and Quin-
sey, 1970; Jackson and Delprato, 1974). Nevertheless, operant suppression has
been frequently offered as a measure of the disruption of the motivational value
of food by conditioned fear. In reviewing the literature, Hancock and Ayres
(1974) simply concluded that motivational state interpretations have little em-
pirical credibility. For this and many additional reasons, it is not clear that a
phrenology of states and values is any more accurate when based upon the
physical characteristics of behavior than when based upon the physical charac-
teristics of the subject's skull.
Moreover, the rather consistent conditioned suppression during the Estes-
Skinner procedure is in fact atypical and the exception rather than the model of
acceleration and suppression of operant rates within each of the three remain-
ing schedule combinations. The conditioned suppression and/or acceleration
effects pose further problems for underlying state interpretations. The summa-
tion of conditioned fear and operant fear motivation might predict acceleration,
but not suppression, when a negative classical conditioning procedure is su-
perimposed upon shock avoidance responses. Similarly, the disruption of
operant fear motivation by positive conditioned states might predict suppres-
sion, but not acceleration, of avoidance rates during superimposed appetitive
conditioned stimuli. Although unnecessary, most state and value interpreta-
tions assume that the proposed motivational or incentive effects are indifferent
to specific response characteristics and topographies. The simplifying assump-
tion does limit the number of necessary explanatory assumptions, but at the
cost of categorically specifying equivalent disruption or energizing effects for
all behaviors maintained by the same instrumental motivation. In contrast, the
multitude of response rates and patterns in classical-operant schedules sug-
gests that changes in behavior must be at least partially determined by physical
response characteristics (Brady, 1971; Miczek, 1973; Scobie, 1973). Many
years ago, Brady and Hunt discussed much the same point:
Many attempts have been made to order these diversities in behavior to a single broad
principle-i.e., emphasis on the role of conditioned "fear" as a motivational con-
struct mediating the establishment of instrumental responses . . . such monolithic
ordering, prematurely embraced, might serve to obscure important differences as
well as significant similarities and relationships among behaviors. (Brady and Hunt,
1955, p. 322)
The early warnings of Brady and Hunt are still more relevant to the host of
behavioral differences and similarities currently apparent in classical-operant
schedules. The diverse interactions between the many known experimental
variables continue to preclude a deductive or statistical reduction of response
rates to a unidimensional measuring or matching function.
One continuing interpretation, however, has maintained an emphasis upon
behavior qua behavior-perhaps not surprisingly, the analysis is the inductive
rather than deductive experimental analysis of "emotional" behavior intro-
76 Section I: Concurrent Classical and Operant Conditioning

duced by Brady and Hunt. The experimental analysis has proved to be viable
for some 20 years, and in fact is more compelling today than in previous years.
Moreover, the experimental analysis not only accepts the role of operant re-
sponses, but also classical conditioned responses and interacting behavioral
patterns that are commonly denied by deductive models. Admittedly, the anal-
ysis of concurrent or competing response patterns is occasionally harshly un-
popular-"elliptical and irrelevant to operant-respondent processes," "triv-
ial," or "too quantitative and microscopic." The fundamental argument
against a response pattern analysis seems to be the cavalier unconcern with
academic states, motivations, emotions, general emotions, and whatnot, in
favor of observed and recorded behavioral functions. Indeed, the behavioral
interactions may be offered as part and parcel of the behavioral effects and are
perhaps not elliptical and irrelevant to the analysis of classical and operant
schedules-nor trivial.
The experiments presented in Chapter 2 suggest, for example, that avoid-
ance rates may be controlled by the pattern of concurrent UCS approach and
retrieval responses during a superimposed appetitive classical conditioning
procedure (Experiment IA,B,C). Interdependent response patterns also occur
in positive classical-positive operant combinations, with operant rates depend-
ent upon competing response patterns (Experiments II and III). Similarly, dis-
crete trial operant procedures and appetitive classical conditioning procedures
control analogous response patterns when superimposed upon a concurrent
operant schedule, as a function of stimulus duration (Experiments VI and VII).
Importantly, the classical conditioning processes are also disrupted during the
Estes-Skinner procedure and are not passively measured by the concurrent
operant response rates (Experiment IX). The various concurrent schedule ef-
fects do not warrant a collapsing of the diverse independent variables into
global constructs, either within or across classical-operant schedules, and the
present experimental analysis remains unrepentantly behavioral and quantita-
tive, and occasionally microscopic.

References
Abel, E. L.: Effects of marihuana homologue, pyrahexyl, on a conditioned emotional response.
Psychonomic Sci., 16, 44, 1969.
Anderson, D. C., Merril, H. K., Dexter, W., and Alleman, H.: Contextual effects in emotional
learning. Proc. Am. Psychol. Assoc. 3, 147-148, 1968.
Anderson, D. c., O'Farrell, T., Formica, R., and Caponigri, Y.: Preconditioning CS exposure:
variation in place of conditioning and presentation. Psychonomic Sci. 15, 54-55, 1969a.
Anderson, D. C., Wolf, D., and Sullivan, P.: Pre-conditioning exposures to the CS: variations in
place of testing. Psychonomic Sci. 14, 233-234, 1969b.
Anderson, L. T., and Ressler, R. H.: Response to a conditioned aversive event in mice as a func-
tion of frequency of premating maternal shock. Develop. Psychobiol., 6, 113-121, 1973.
Annau, Z., and Kamin, L. J.: The conditioned emotional response as a function of intensity ofthe
US. J. Compo Physiol. Psychol., 54, 428-432, 1961.
Appel, J. B.: Drugs, shock intensity and the CER. Psychopharmacologia, 4, 141-153, 1963.
Arnold, M. B.: An excitatory theory of emotion. In, M. L. Reymert (ed.): Feelings and Emotion.
New York, McGraw-Hill, 1950.
Arnold, M. B. (ed.): Feelings and Emotion. New York, Academic Press, 1970.
Chapter 1: Review of Classical-Operant Conditioning 77

Auerbach, P., and Carlton, P. L.: Retention deficit correlated with a deficit in the corticoid re-
sponse to stress. Science, 173, 1148-1149, 1971.
Avis, H.: Hippocampal injections of KCL and shock-induced bradycardia in the rat. Psychonomic
Sci., 27,283-284, 1972.
Avis, H., and Carlton, P. L.: Retrograde amnesia produced by hippocampal spreading depression.
Science, 161,73-75, 1968.
Avis, H., and Pert, A.: A comparison of the effects of muscarinic and nicotinic anticholingergic
drugs on habituation and fear conditioning in rats. Psychopharmacologia, 34,209-222, 1974.
Ayres, J. J. B.: Conditioned suppression and the information hypothesis. J. Compo Physiol. Psy-
chol., 62,21-25, 1966.
Ayres, J. J. B.: Differentially conditioned suppression as a function of shock intensity and incen-
tive. J. Compo Physiol. Psych., 66,208-210, 1968.
Ayres, J. J. B., and DeCosta, M. J.: The truly random control as an extinction procedure. Psy-
chonomic Sci., 24,31-33, 1971.
Ayres, J. J. B., and Quinsey, V. L.: Between-groups incentive effects on conditioned suppression.
Psychonomic Sci., 21, 294-296, 1970.
Azrin, N. H., and Hake, D. F.: Positive conditioned suppression: conditioned suppression using
positive reinforcers as the unconditioned stimuli. J. Exp. Anal. Behav., 12, 167-173, 1969.
Baird, H. N., Gudetti, B., Reyes, V., Wycis, H. T., and Spiegal, E. G.: Stimulation and elimination
of anterior thalamic nuclei in man and cat. Fed. Proc., 10,8-9. 1951.
Baker, A. G.: Conditioned inhibition is not the symmetrical opposite of conditioned excitation: a
test of the Rescorla-Wagner model. Learning Motivation,S, 369-379, 1974.
Baker, A. G.: Learned irrelevance and learned helplessness: rats learn that stimuli, reinforcers,
and responses are uncorrelated. J. Exp. Psychol. [Anim. Behav. Processes], 2, 130-142, 1976.
Barcik, J. D.: Hippocampal after discharges and conditioned emotional responses. Psychonomic
Sci., 20,297-298, 1970.
Bard, P.: A diencephalic mechanism for the expression of rage with special reference to the sympa-
thetic nervous system. Am. J. Physiol., 84, 490-515, 1928.
Bard, P.: On emotional expression after decortication with some remarks on certain theoretical
views. Parts I and II. Psycholo. Rev., 41, 309-329, 424-449, 1934.
Bard, P., and Mountcastle, V. B.: Some forebrain mechanisms involved in expression of rage with
special reference to suppression of angry behavior. In, J. Fulton (ed.): The Frontal Lobes. Balti-
more, Williams & Wilkins, 1948.
Bard, P., and Rioch, D. M.: A study offour cats deprived of neocortex and additional portions of
the forebrain. Bull. Johns Hopkins Hosp., 60,73-147, 1937.
Baron, A., and Trenholme, I. A.: Response-dependent and response-independent timeout from an
avoidance schedule. J. Exp. Anal. Behav., 16, 123-132, 1971.
Belleville, R. E., Rohles, F. H., Grunzke, M. E., and Clark, F. G.: Development of a complex
multiple schedule in the chimpanzee. J. Exp. Anal. Behav. 6,549-556, 1963.
Benedict, J. 0., and Ayres, J. J. B.: Factors affecting conditioning in the truly random control
procedure. J. Compo Physiol. Psychol., 78,323-330, 1972.
Berger, B. D., and Stein, L.: An analysis of learning deficits produced by scopolamine. Psycho-
pharmacologia, 14,271-283, 1969a.
Berger, B. D., and Stein, L.: Asymmetrical dissociation oflearning produced by scopolamine and
W Y 40 36, a new benzodiazepine. Psychopharmacologia, 14,51-358, 1969b.
Blackman, D. E.: Response rate and conditioned suppression. Psychol. Rep., 19,687-693, 1966.
Blackman, D. E.: Conditioned suppression: comments on Lyon's reply. Psychol. Rep., 20, 909-
910, 1967a.
Blackman, D. E.: Effects of response pacing on conditioned suppression. Q. J. Exp. Psychol., 19,
170-174, 1967b.
Blackman, D. E.: Conditioned suppression or facilitation as a function of the behavioral baseline.
J. Exp. Anal. Behav., 11,53-61, 1968a.
Blackman, D. E.: Response rate, reinforcement frequency and conditioned suppression. J. Exp.
Anal. Behav., 11,503-516, 1968b.
Blackman, D. E.: Effects of a pre-shock stimulus on temporal control of behavior. J. Exp. Anal.
Behav., 14,313-319, 1970.
 78 Section I: Concurrent Classical and Operant Conditioning

Blackman, D. E., and Scrutton, P.: Conditioned suppression and discriminative control ofbehav-
ior. Anim. Learning Behav., 1, 90-92, 1973.
Booth, J. H., and Hammond, L. J.: Configural conditioning: greater fear in rats to the compound
than component through overlapping the compound. J. Exp. Psychol., 87,255-262, 1971.
Boren, J. J.: Some effects of adephenine, benactyzine, and chloropromazine upon several operant
behaviors. Psychopharmacologia, 2,416-424, 1%1.
Boren, J. J.: The study of drugs with operant techniques. In, W. K. Honig (ed.): Operant Behavior:
Areas of Research and Application. New York, Appleton-Century-Crofts, 1%6.
Boring, E. G.: A History of Experimental Psychology, 2nd ed. New York, Appleton-Century-
Crofts, 1957.
Borgealt, A. J., Donahoe, J. W., and Weinstein, A.: Effects of delayed and trace components
on a compound CS on conditioned suppression and heart rate. Psychonomic Sci., 26, 13-15,
1972.
Brady, J. V.: The effect of electro-convulsive shock on a conditioned emotional response: the per-
manence of the effect. J. Compo Physiol. Psychol., 44,507-511, 1951.
Brady, J. V.: The effect of electro-convulsive shock on a conditioned emotional response: the sig-
nificance ofthe interval between the emotional conditioning and the electro-convulsive shock. J.
Compo Physiol. Psychol., 45, 9-13, 1952.
Brady, J. V.: Does tetraethylammonium reduce fear. J. Compo Physiol. Psychol., 46,307-310,
1953.
Brady, J. V.: Extinction of a conditioned "fear" response as a function of reinforcement schedules
for competing behaviors. J. Psychol., 40, 25-34, 1955.
Brady, J. V.: A comparative approach to the evaluation of drug effects upon affective behavior.
Ann. N. Y. Acad. Sci., 64,632-643, 1956a.
Brady, J. V.: The assessment of drug effects on emotional behavior. Science, 123, 1033-1034,
1956b.
Brady, J. V.: A review of comparative behavioral pharmacology. Ann. N. Y. Acad. Sci., 66,719-
732, 1957.
Brady, J. V.: Animal experimental evaluation of drug effects upon behavior. Proc. Assoc. Res.
Nerv. Mental Dis., 37, 104-125, 1959.
Brady, J. V.: Motivation-emotional factors and intracranial self-stimulation. In, D. E. Sheer (ed.):
Electrical Stimulation of the Brain. Austin, University of Texas Press, 1%1.
Brady, J. V.: Psychophysiology of emotional behavior. In, A. J. Bachrach (ed.): Experimental
Foundations of Clinical Psychology. New York, Basic Books, 1%2.
Brady, J. V.: Emotion: some conceptual problems and psychophysical experiments. In, M. Arnold
(ed.): Feelings and Emotion. New York, Academic Press, 1970a.
Brady, J. V.: Endocrine and autonomic correlates of emotional behavior. In, P. Black (ed.): Physi-
ological Correlates of Emotion. New York, Academic Press, 1970b.
Brady, J. V.: Emotion revisited. J. Psychiatr. Res., 8,363-384, 1971.
Brady, J. V.: Toward a behavioral biology of emotion. In, L. Levi (ed.): Emotions-their Parame-
ters and Measurement. New York, Raven Press, 1975.
Brady, J. V., and Conrad, D.: Some effects of limbic system self-stimulation upon conditioned
emotional behavior. J. Compo Physiol. Psychol., 53, 128-137, 1960.
Brady, J. V., and Hunt, H. F.: A further demonstration of the effects of electro-convulsive shocks
on a conditioned emotional response. J. Compo Physiol. Psychol., 44, 204-209, 1951.
Brady, J. V., and Hunt, H. F.: The effect of electro-convulsive shock on a conditioned emotional
response: a control for impaired hearing. J. Compo Physiol. Psychol., 45, 180-182, 1952.
Brady, J. V., and Hunt, H. F.: An experimental approach to the analysis of emotional behavior. J.
Psychol., 40, 313-325, 1955.
Brady, J. V., and Nauta, W. J. H.: Subcortical mechanisms in emotional behavior: affective
changes following septal forebrain lesions in the albino rat. J. Compo Physiol. Psychol., 46,339-
346, 1953.
Brady, J. V., and Nauta, W. J. H.: Subcortical mechanisms in emotional behavior: the duration of
affective changes following septal and habenular lesions in the albino rat. J. Compo Physiol. Psy-
chol., 48, 412-420, 1955.
Chapter 1: Review of Classical-Operant Conditioning 79

Brady, J. Y., and Sulsa, G.: Acquisition and extinction of conditioned "fear" as a function of
shock intensity. Paper read at Eastern Psychological Association, Philadelphia, 1955.
Brady, J. Y., Stebbins, W. C., and Galambos, R.: The effect of audiogenic convulsions on a condi-
tioned emotinal response. J. Compo Physiol. Psychol., 46,363-367, 1953.
Brady, J. Y., Stebbins, W. C., and Hunt, H. F.: The effect of electro-convulsive shock (ECS) on a
conditioned emotional response: the effect of additional ECS convulsions. J. Compo Physiol.
Psychol., 46, 368-372, 1953.
Brady, J. Y., Hunt, H. F., and Geller, I.: The effect of electroconvulsive shock on a conditioned
emotional response as a function of the temporal distribution of the treatments. J. Compo Phy-
siol. Psychol., 47,454-457, 1954.
Brady, J. Y., Libber, S., and Dardano, J.: Some effects of a pre-aversive stimulus on avoidance
behavior. Tech. Rep. 67-14, Space Research Lab, University of Maryland, 1967.
Brady, J. Y., Kelly, D. D., and Plumlee, L.: Autonomic and behavioral responses of the rhesus
monkey to emotional conditioning. Ann. N. Y. Acad. Sci., 150,959-975, 1969.
Brady, J. Y., Henton, W. W., and Ehle, A.: Some effects of emotional conditioning upon auto-
nomic and electromyographic activity. Paper read at Eastern Psychological Association, Atlantic
City, 1970.
Brimer, C. J.: Attention and conditioned suppression. Psychonomic Sci., 22, 131-132, 1971.
Brimer, C. J., and Dockrill, F. J.: Partial reinforcement and the CER. Psychonomic Sci., 5, 185-
186, 1966.
Brimer, C. J., and Kamin, L. J.: Disinhibition, habituation, sensitization, and the conditioned emo-
tional response. J. Compo Physiol. Psychol., 56, 508-516, 1963.
Brimer, C. J., and Wickson, S.: Shock frequency, disinhibition and conditioned suppression.
Learning Motivation, 2, 124-137, 1971.
Brodigan, D. L., and Trapold, M. A.: Recovery from conditioned suppression to a partially over-
lapping compound stimulus. Anim. Learning Behav., 2,89-91, 1974.
Brogden, W. J.: Sensory preconditioning. J. Exp. Psychol., 25,323-332, 1939.
Brophy, J. C., and Tremblay, A. M.: One-trial CER as a function of shock intensity. Psychonomic
Sci., 25, 13-14, 1971.
Brown, J. S., and Farber, I. E.: Emotions conceptualized as intervening variables-with sugges-
tions toward a theory of frustration. Psychol. Bull., 48,465-495, 1951.
Brown, J. S., Kalish, H. I., and Farber, I. E.: Conditioned fear as revealed by the magnitude of
startle response to an auditory stimulus. J. Exp. Psychol., 41, 317-328, 1951.
Brunner, R. L., Roth, T. G., and Rossi, R. R.: Age differences in the development of the condi-
tioned emotional response. Psychonomic Sci., 21, 135-136, 1970.
Bryant, R. C.: Conditioned suppression of free-operant avoidance. J. Exp. Anal. Behav., 17,257-
260,1972.
Buchanan, D. C. Schaefer, G. J., and Caul, W. F.: Effects of infantile handling on heart rate condi-
tioning and response suppression. Psychonomic Sci., 29,279-281, 1972.
Bull, J. A., III: An interaction between appetitive Pavlovian CS's and instrumental avoidance re-
sponding. Learning Motivation, 1, 18-26, 1970.
Burdick, C. K., and James, J. P.: Spontaneous recovery of conditioned suppression of licking by
rats. J. Compo Physiol. Psychol., 72,467-470, 1970.
Burdick, C. K., and James, J. P.: Effects of backward conditioning procedure following acquisition
on extinction of conditioned suppression. Anim. Learning Behav., 1, 137-139, 1973.
Byrum, R. P., and Jackson, D. E.: Response availability and second-order conditioned suppres-
sion. Psychonomic Sci., 23, 106-108, 1971.
Calef, R. S., Kaufman, R. A., Bone, R. N., and Werk, S. A.: Noncontingentnonreinforcementofa
response as the aversive event in a CER paradigm. Psychol. Rep., 29, 1196-1198, 1971.
Cameron, O. G., and Appel, J. B.: Conditioned suppression of bar-pressing behavior by stimuli
associated with drugs. J. Exp. Anal. Behav., 17, 127-137, 1972a.
Cameron, O. G., and Appel, J. B.: Generalization of LSD-induced conditioned suppression. Psy-
chonomic Sci., 27,303-304, 1972b.
Campbell, B. A., and Campbell, E. H.: Retention and extinction ofleamed fear in infant and adult
rats. J. Compo Physiol. Psychol., 55, 1-8, 1962.
80 Section I: Concurrent Classical and Operant Conditioning

Cannon, W. B.: Recent studies of bodily effects of fear, rage, and pain. J. Philosophy, Psychol.
Sci. Methods, 11, 162-165, 1914.
Cannon, W. B.: The James-Lange theory of emotions: a critical examination and alternate theory .
Am. J. Psychol., 39, 106-204, 1927.
Cannon, W. B.: Again the James-Lange theory and the thalamic theories of emotion. Psychol.
Rev., 38,281-295, 1931.
Cannon, W. B., Lewis, J. T., and Britton, S. W.: The dispensibility of the sympathetic division of
the ANS. Boston Med. Surg. J., 197, 514-515, 1927.
Cappel, H. D., Herring, B., and Webster, C. D.: Discriminated conditioned suppression: further
effects of stimulus compounding. Psychonomic Sci., 19, 147-149, 1970.
Cappel, H. D., Ginsberg, R., and Webster, C. D.: Amphetamine and conditioned "anxiety." Br. J.
Pharmacol., 45, 525-431, 1972.
Carlton, P. L., and Didamo, P.: Some notes on the control of conditioned suppression. J. Exp.
Anal. Behav., 3,255-258, 1960.
Carlton, P. L., and Vogel, J. R.: Habituation and conditioning. J. Compo Physiol. Psychol., 63,
348-351, 1967.
Carman, J. B.: Time-out from a short mean-interval reinforcement schedule. Psychonomic Sci., 15,
259-260, 1969.
Carman, J. B.: Deprivation levels and conditioned suppression. Psychol. Rep., 27, 599-602,
1970.
Carr, A. P.: Latent inhibition and overshadowing in conditioned emotional response conditioning
with rats. J. Compo Physiol. Psychol., 86,718-723, 1974.
Catania, A. C.: Concurrent operants. In, W. K. Honig (ed.): Operant behavior: areas of research
and application. New York, Appleton-Century-Crofts, 1966.
Chaddock, T. E.: Visual detection of X ray by the rhesus monkey. J. Compo Physiol. Psychol., 78,
190-201, 1972.
Church, R. M.: Response suppression. In, B. A. Campbell and R. M. Church (eds.): Punishment
and aversive behavior. New York, Appleton-Century-Crofts, 1969.
Church, R. M., Wooten, C. L., and Mathews, T.: Discriminative punishment and the conditioned
emotional response. Learning Motivation, 1, 1-17, 1970.
Cicala, G. A., and Hartley, D. L.: Drugs and the learning and performance offear. J. Compo Phy-
sial. Psychol., 64, 173-178, 1967.
Corson, A., and Corson, E. 0.: Pavlovian conditioning as a method for studying the mechanism of
action of minor tranquilizers. In, H. Brill (ed.): Proceedings of the Fifth International Confer-
ence of Neuropharmacology, Amsterdam, 1967.
Coughlin, R. C.: The aversive properties of withdrawing positive reinforcement: a review of the
recent literature. Psychol. Record, 22,333-354, 1972.
Coulson, G., and Walsh, M.: Facilitation of avoidance responding in white rats during a stimulus
preceding food. Psychol. Rep., 22, 1277-1284, 1968.
Coulter, X., Collier, A. C., and Campbell, B. A.: Long-term retention of early Pavlovian fear con-
ditioning in infant rats. J. Exp. Psychol. [Anim. Behav. Processesl, 2, 48-56, 1976.
Cousins, L. S., Zamble, E., Tait, R. W., and Suboski, M. D.: Sensory preconditioning in curarized
rats. J. Compo Physiol. Psychol., 77, 152-154, 1971.
Dalton, L. W., Jr.: Conditioned suppression as a technique for determination of auditory sensitiv-
ity in pigeons. J. Auditory Res., 7,25-29, 1967.
Daly, H. B.: Disruptive effects of scopolamine on fear conditioning and on instrumental escape
learning. J. Compo Physiol. Psychol., 66,579-583, 1968.
Dantzer, R., and Baldwin, B. A.: Changes in heart rate during suppression of operant responding in
pigs. Physiol. Behav., 12,385-391, 1974.
Dantzer, R., and Delacour, J.: Modification of a phenomenon of conditioned suppression by a tha-
lamic lesion. Physiol. Behav., 8,997-1003, 1972.
Darwin, C.: Expression of the Emotions in Man and Animals. London, Murray, 1872.
Davenport, J. W.: Higher-order conditioning of fear. Psychonomic Sci., 4,27-28,1966.
Davis, H.: Conditioned suppression: a survey of the literature. Psychonomic Monogr. [Suppl.l, 2,
283-291, 1968.
Chapter 1: Review of Classical-Operant Conditioning 81

Davis, H., and Kreuter, C.: Conditioned suppression of an avoidance response by a stimulus
paired with food. J. Exp. Anal. Behav., 17,277-285, 1972.
Davis, H., and McIntire, R. W.: Conditioned suppression under positive, negative, and no contin-
gency between conditioned and unconditioned stimuli. J. Exp. Anal. Behav., 12,633-640, 1969.
Davis, H., McIntire, R. W., and Cohen, S. I.: Fixed and variable duration warning stimuli and
conditioned suppression. J. Psychol., 73, 19-25, 1969.
Davis, H., Memmott, J., and Hurwitz, H. M. B.: Effects of signals preceding and following shock
on baseline responding during a conditioned suppression procedure. J. Exp. Anal. Behav., 25,
263-277, 1976.
Davis, R. G.: Olfactory psychophysical parameters in man, rat, dog, and pigeon. J. Compo Physiol.
Psychol., 85,221-232, 1973.
Davitz, J. P.: Decreased autonomic functioning and extinction of a conditioned emotional re-
sponse. J. Compo Physiol. Psychol., 46,311-313, 1953.
DeCosta, M. J., and Ayres, J. J. B.: Suppression of operant vs consummatory behavior. J. Exp.
Anal. Behav., 16, 133-142, 1971.
Desiderato, 0.: Generalization of conditioned suppression. J. Compo Physiol. Psychol., 57,434-
437, 1964.
Desiderato, 0.: Generalization of extinction and inhibition in control of avoidance responding by
Pavlovian CSs in dogs. J. Compo Physiol. Psychol., 68,611-616, 1969.
Desiderato, 0., and Newman, A.: Conditioned suppression produced in rats by tones paired with
escapable and inescapable shock. J. Compo Physiol. Psychol., 77,427-431, 1971.
de Toledo, L.: Changes in heart rate during conditioned suppression in rats as a function of US
intensity and type of CS. J. Compo Physiol. Psychol., 77,528-538, 1971.
de Toledo, L., and Black, A. H.: Heart rate: changes during conditioned suppression in rats. Sci-
ence, 152, 1404-1406, 1966.
DeVietti, T. L., and Holliday, J. H.: Retrograde amnesia produced by electroconvulsive shock
after reactivation of a consolidated memory trace: a replication. Psychonomic Sci., 29, 137 -138,
1972.
DeVietti, T. L., and Porter, P. B.: Modification of the autonomic component of the conditioned
emotional response. Psychol. Rep., 24,951-958, 1969.
Dews, P. B.: Psychopharmacology. In, A. J. Bachrach (ed.): Experimental Foundations o/Clinical
Psychology. New York, Basic Books, 1962.
Dexter, W. R., and Merrill, H. K.: Role of contextual discrimination in fear conditioning. J. Compo
Physiol. Psychol., 69,677-681, 1969.
Dickinson, A., and Morriss, R. G. M.: Conditioned acceleration and free-operant wheel-turn
avoidance following septal lesions in rats. Physiol. Psychol., 3, 107-112, 1975.
Di Giusto, J. A., Di Guisto, E. L., and King, M. G.: Heart rate and muscle tension correlates of
conditioned suppression in humans. J. Exp. Psychol., 103,515-521, 1974.
Domjan, M., and Siegal, S.: Conditioned suppression following CS preexposure. Psychonomic
Sci., 25, 11-12, 1971.
Donegan, N. H., Whitlow, J. W., and Wagner, A. R.: Posttrial reinstatement of the CS in Pavlo-
vian conditioning: facilitation or impairment of acquisition as a function of individual differences
in responsiveness to the CS. J. Exp. Psychol. [Anim. Behav. Processes], 3,357-376, 1977.
Dubin, W. J., and Levis, D. J.: Influence of similarity of components of a serial conditioned stimu-
lus on conditioned fear in rats. J. Compo Physiol. Psychol., 85,304-312, 1973.
Duffy, E.: The psychological significance of the concept of "arousal" or "activation." Psychol.
Rev., 64,265-275, 1957.
Egger, M. D., and Miller, N. E.: Secondary reinforcement in rats as a function of information value
and reliability of the stimulus. J. Exp. Psychol., 64,97-104,1962.
Estes, W. K.: Discriminative conditioning. I. A discriminative property of conditioned anticipa-
tion. J. Exp. Psychol., 32, 150-155, 1943.
Estes, W. K.: Discriminative conditioning. II. Effects of a Pavlovian conditioned stimulus upon a
subsequently established operant response. J. Exp. Psychol., 38, 173-177, 1948.
Estes, W. K.: The statistical approach to learning theory. In S. Koch (ed.): Psychology: A Study 0/
Science, vol. 2. New York, McGraw-Hill, 1959.
82 Section I: Concurrent Classical and Operant Conditioning

Estes, W. K.: Outline of a theory of punishment. In, B. A. Campbell and R. M. Church (eds.):
Punishment and Aversive Behavior. New York, Appleton-Century-Crofts, 1969.
Estes, W. K., and Skinner, B. F.: Some quantitative properties of anxiety. J. Exp. Psychol., 29,
390-400, 1941.
Evans, H. L., and Patton, R. A.: Scopolamine effects on conditioned suppression: influence of
diurnal cycle and transitions between normal and drugged states. Psychopharmacologia. 17, 1-
13, 1970.
Fehr, F. S., and Stern, J. A.: Peripheral physiological variables and emotion: the James-Lange
theory revisited. Psychol. Bull., 74,411-424, 1970.
Ferster, C. B., and Skinner, B. F.: Schedules of Reinforcement. New York, Appleton-Century-
Crofts, 1957.
Field, G., and Boren, J. J.: An adjusting avoidance procedure with multiple auditory and visual
warning stimuli. J. Exp. Anal. Behav., 6,537-543, 1963.
Findley, J. D.: An experimental outline for building and exploring multi-operant behavior reper-
toires. J. Exp. Anal. Behav. [Suppl.], 5, 113-166, 1962.
Finnocchio, D.: Changes in temporally space responding as a measure of conditioned emotional
behavior. Paper read at Eastern Psychological Association, New York, 1963.
Fleshier, M., and Hoffman, H. S.: Stimulus generalization of conditioned suppression. Science,
133,753-755, 1961.
Fontaine, 0., and Richelle, M.: Comparative study of the effects of chlorpromazine and chlordia-
zepoxide on a series of programs of positive and negative reinforcement of behavior in the rat.
Psychol. Belg., 9, 17-29, 1969.
Freeman, F. G., Mikulka, P. J., and d' Auteuil, P.: Conditioned suppression of a licking response in
rats with hippocampal lesions. Behav. Bioi., 12,257-263, 1974.
Frey, A. H.: Modification of the conditioned emotional response by treatment with small negative
air ions. J. Compo Physiol. Psychol., 63, 121-125, 1967.
Frey, P., Eng, S., and Gavin, W.: Conditioned suppression in the gerbil. Behav. Res. Methods
Instr. 4, 245-249, 1972.
Frieman, J. P., Warner, L., and Riccio, D. C.: Age differences in conditioning and generalization of
fear in young and adult rats. Develop. Psychol., 3, 119-123, 1970.
Frieman, J. P., Frieman, J., Wright, W., and Hegberg, W.: Developmental trends in the acquisition
and extinction of conditioned suppression in rats. Develop. Psychol., 4,425-428, 1971.
Geller, 1.: The acquisition and extinction of conditioned suppression as a function of the baseline.
J. Exp. Anal. Behav., 3,235-240, 1960.
Geller, 1.: Conditioned suppression in goldfish as a function of shock-reinforcement schedule. J.
Exp. Anal. Behav., 7,345-349, 1964.
Geller, 1., Sidman, M., and Brady, J. V.: The effect of electroconvulsive shock on a conditioned
emotional response: a control for acquisition recency. J. Camp. Physiol. Psychol., 48, 130-131,
1955.
Gleitman, H., and Holmes, P. A.: Retention of incompletely learned CER in rats. Psychonomic
Sci., 7, 19-20, 1967.
Glick, S. D.: Effects of d-amphetamine and frontal ablations on response suppression in rats. J.
Compo Physiol. Psychol., 69, 49-54, 1969.
Goddard, G. V.: Amygdaloid stimulation and learning in the rat. J. Camp. Physiol. Psychol., 58,
23-30, 1964.
Goddard, G. V.: Analysis of avoidance conditioning following cholinergic stimulation of amygdala
in rats. J. Camp. Physiol. Psychol. Monogr., 68, (2, pt 2), 1969.
Goldberg, M. E., Sledge, K., Hefner, M., and Robichaud, R. C.: Learning impairment after three
classes of agents which modify cholinergic function. Arch. Int. Pharmacodyn. Ther., 193,226-
235, 1971.
Goldberg, M. E., Sledge, K., Robichaud, R. C., and Dubinsky, B.: A comparative study of the
behavioral effects of scopolamine and 4-(1-naphthylvinyl) pyridine hydrochloride (NVP), an in-
hibitor of choline acetyltransferase. Psychopharmacologia, 23, 34-47, 1972.
Goldberg, S. R, and Schuster, C. R: Conditioned suppression by a stimulus associated with nalor-
phine in morphine-dependent monkeys. J. Exp. Anal. Behav., 10,235-242, 1967.
Chapter 1: Review of Classical-Operant Conditioning 83

Goldberg, S. R., and Schuster, C. R.: Conditioned nalorphine-induced abstinence changes: persis-
tance in morphine dependent monkeys. J. Exp. Anal. Behav., 14,33-36, 1970.
Goldstein, R.: Effects of non-contingent septal stimulation on the CER in the rat. J. Compo Physiol.
Psychol., 61, 132-135, 1966.
Gonzalez, S. C., Karinol, I. G., and Carlini, E. A.: Effects of Cannabis sativa extract on condi-
tioned fear. Behav. Bioi., 7,83-94, 1972.
Goodman, L. S., and Gilman, A. (eds.): The Pharmacological Basis of Therapeutics, 4th ed. New
York, MacMillan, 1970.
Gottwald, P.: The role of punishment in the development of conditioned suppression. Physio.
Behav., 2,283-286, 1967.
Gottwald, P.: Adventitious punishment in two conditioned suppression procedures. Psychol. Rep.,
25,699-703, 1969.
Goy, R. W., and Hunt, H. F.: The resistance of an instrumental response to suppression by condi-
tioned fear. Am. Psychol., 8,509, 1953.
Gray, T., and Appignanesi, A. A.: Elimination of the blocking effect. Learning Motivation, 4, 374-
380, 1973.
Grossen, N. E., and Bolles, R. C.: Effects of classical conditioned "fear signal" and "safety sig-
nal" on nondiscriminated avoidance behavior. Psychonomic Sci., 11,321-322, 1968.
Grossen, N. E., Kostansek, D. J., and Bolles, R. C.: Effects of appetitive discriminative stimuli
upon avoidance behavior. J. Exp. Psychol., 81,340-343, 1969.
Grossman, S. P.: A Textbook of Physiological Psychology. New York, Wiley, 1967.
Gustafson, J. W., Lidsky, T. I., and Schwartzbaum, J. S.: Effects of hippocampal stimulation on
acquisition, extinction and generalization of conditioned suppression in the rat. J. Compo Phy-
siol. Psychol., 10,1136-1147, 1975.
Hake, D. F., and Powell, J.: Positive reinforcement and suppression from the same occurrence of
the unconditioned stimulus in a positive conditioned suppression paradigm. J. Exp. Anal.
Behav., 14,247-257,1970.
Hammond, L. J.: Increased responding to CS - in differential CER. Psychonomic Sci., 5,337-339,
1966.
Hammond, L. J.: A traditional demonstration of the active properties of Pavlovian inhibition using
differential CER. Psychonomic Sci., 9,65-66, 1967.
Hammond, L. J., and Maser, J.: Forgetting and conditioned suppression: role oftemporal discrimi-
nation. J. Exp. Anal. Behav., 13,333-338, 1970.
Hancock, R. A., and Ayres, J. J. B.: Within-subject effects of sucrose concentration on condi-
tioned suppression of licking. Psychol. Record, 24,325-331, 1974.
Hanson H. M.: Effects of discrimination training on stimulus generalization. J. Exp. Psychol., 58,
321-323, 1959.
Hartmann, G. J., and Geller, I.: P-Chlorophenylalanine effects on a conditioned emotional re-
sponse in rats. Life Sci., 10,927-933, 1971.
Harvey, J. A., Lints, C. E., Jacobson, L. E., and Hunt, H. F.: Effects oflesions in the septal area
on conditioned fear and discriminated instrumental punishment in the albino rat. J. Compo Phy-
siol. Psychol., 59,37-48, 1965.
Haworth, J. T.: Conditioned emotional response phenomena and brain stimulation. Br. J. Psychol.,
62,97-103, 1971.
Hearst, E.: Stress-induced breakdown of an appetitive discrimination. J. Exp. Anal. Behav., 8,
135-146, 1965.
Hebb, D.O.: The organization of behavior. New York, Wiley, 1949.
Heffner, H., and Masterton, B.: Hearing in primitive primates: slow loris (Nycticebus coucang)
and potto (Perodicticus potto). J. Compo Physiol. Psychol., 71, 175-182, 1970.
Heffner, R., Heffner, H., and Masterton, B.: Behavioral measurement of absolute and frequency-
difference thresholds in the quinea pig. J. Acoust. Soc. Am., 49, 1888-1895, 1971.
Hendersen, R. W.: Conditioned and unconditioned fear inhibition in rats. J. Compo Physiol. Psy-
chol., 84,554-561, 1973.
Hendersen, R. W.: Compounds of conditioned fear stimuli. Learning Motivation, 6, 28-42,
1975.
84 Section I: Concurrent Classical and Operant Conditioning

Hendricks, J.: Flicker thresholds as determinted by a modified conditioned suppression procedure.

J. Exp. Anal. Behav., 9,501-506,1%6.
Hendry, D. P.: Conditioned inhibition of conditioned suppression. Psychonomic Sci., 9,261-262,
1967.
Hendry, D. P., and Van Toller, C.: Alleviation of conditioned suppression. J. Compo Physiol. Psy-
chol., 59,458-460, 1965.
Hendry, D. P., Switalski, R., and Yarczower, M.: Generalization of conditioned suppression after
differential training. J. Exp. Anal. Behav., 12,799-806, 1%9.
Henton, W. W.: Suppression behavior to odorous stimuli in the pigeon. Unpublished Doctoral Dis-
sertation, Florida State University, 1%6.
Henton, W. W.: Conditioned suppression to odorous stimuli in pigeons. J. Exp. Anal. Behav., 12,
175-185, 1969.
Henton, W. W.: The effects of pre-schedule change stimuli upon avoidance responding. Rep.
Neurophysiol. Inst. (Kbh), 34-35, 1970.
Henton, W. W. Avoidance response rates during a pre-food stimulus in monkeys. J. Exp. Anal.
Behav., 17,269-275, 1972.
Henton, W. W., and Brady, J. V.: Operant acceleration during a pre-reward stimulus. J. Exp. Anal.
Behav., 13, 205-209, 1970.
Henton, W. W., and Iversen, 1. H.: Concurrent response rates during preevent stimuli. Paper read
at the Easter Conference, Cambridge, March 1973.
Henton, W. W., and Jordan, J. J.: Differential conditioned suppression during pre-shock stimuli as
a function of shock intensity. Psychol. Record, 20, 9-16, 1970.
Henton, W. W., Smith, J. c., and Tucker, D.: Odor discrimination in pigeons. Science, 153, 1138-
1139, 1966.
Henton, W. W., Smith, J. C., and Tucker, D.: Odor discrimination in pigeons following section of
the olfactory nerves. J. Compo Physiol. Psychol., 69,317-323, 1%9.
Herrnstein, R. J.: On the law of effect. J. Exp. Anal. Behav., 13,243-266, 1970.
Herrnstein, R. J., and Morse, W. H.: Some effects of response-independent positive reinforcement
on maintained operant behavior. J. Compo Physiol. Psychol., 50,461-467, 1957.
Herrnstein, R. J., and Sidman, M.: Avoidance conditioning as a factor in the effects of unavoidable
shocks on food-reinforced behavior. J. Compo Physiol. Psychol., 51,380-385, 1958.
Heth, C. D., and Rescor1a, R. A.: Simultaneous and backward fear conditioning in the rat. J.
Compo Physiol. Psychol., 82,434-443, 1973.
Hilton, A.: Partial reinforcement of a conditioned emotional response in rats. J. Compo Physiol.
Psychol., 69,253-260, 1%9.
Hoffman, H. S.: Discrimination processes are relative to aversive controls. In, R. Gilbert (ed.):
Discrimination Learning. Aberdeen, University of Aberdeen, 1%7.
Hoffman, H. S.: Stimulus factors in conditioned suppression. In, B. A. Campbell and R. M.
Church (eds.): Punishment and Aversive Behavior. New York, Appleton-Century-Crofts, 1969.
Hoffman, H. S., and Barrett, J.: Overt activity during conditioned suppression: a search for pun-
ishment artifacts. J. Exp. Anal. Behav., 16,343-348, 1971.
Hoffman, H. S., and Fleshier, M.: Stimulus factors in aversive controls: the generalization of con-
ditioned suppression. J. Exp. Anal. Behav., 4,371-378, 1%1.
Hoffman, H. S., and Fleshier, M. The course of emotionality in the development of avoidance. J.
Exp. Psychol., 64,288-294, 1962.
Hoffman, H. S., and Fleshier, M.: Stimulus aspects of aversive controls: stimulus generalization of
conditioned suppression following discrimination training. J. Exp. Anal. Behav., 7, 233-239,
1964.
Hoffman, H. S., and Fleshier, M.: Stimulus aspects of aversive controls: the effects of response
contingent shock. J. Exp. Anal. Behav., 8,89-96, 1965.
Hoffman, H. S., Fleshier, M., and Jensen, P.: Stimulus aspects of aversive controls: the retention
of conditioned suppression. J. Exp. Anal. Behav., 6, 575-583, 1%3.
Hoffman, H. S., Selekman, W., and Fleshier, M.: Stimulus aspects of aversive controls: long term
effects of suppression procedures. J. Exp. Anal. Behav., 9,659-662, 1966a.
Chapter 1: Review of Classical-Operant Conditioning 85

Hoffman, H. S., Selekman, W. L., and Fleshier, M.: Stimulus factors in aversive controls: condi-
tioned suppression after equal training to two stimuli. J. Exp. Anal. Behav., 9,649-653, 1966b.
Holland, P. C., and Rescoria, R. A.: The effect of two ways of devaluing the unconditioned stimu-
lus after first- and second-order appetitive conditioning. J. Exp. Psychol. [Anim. Behav. Proc-
esses], 1, 355-363, 1975.
Holmes, P.: Conditioned suppression with extinction as the signalled stimulus. J. Exp. Anal.
Behav., 18, 129-132, 1972.
Holtzman, S. G., and Villarreal, J. E.: The effects of morphine on conditioned suppression in rhe-
sus monkeys. Psvchonomic Sci., 17, 161-162, 1%9.
Homzie, M. J., Shucard, D. W., and Trost, R. c.: Effects of percentage of reinforcement in fear
conditioning upon the acquisition of an instrumental response. Psychonomic Sci., 17,143-145,
1%9.
Houser, V. P.: Modulation of avoidance behavior in squirrel monkeys after chronic administration
and withdrawl of d-amphetamine or a-methyl-p-tyrosine. Psychopharmacologia, 28, 213-234,
1973.
Houser, V. P. and Houser, F. L.: The effects of agents that modify muscarinic tone upon behavior
controlled by an avoidance schedule that employs signalled shock. Psychopharmacologia, 32,
133-150, 1973.
Houser, V. P., Rothfeld, B., and Varady, A., Jr.: Effects of chlordiazepoxide upon fear-motivated
behavior in dogs. Psychol. Rep., 36,987-998, 1975.
Hughes, R. A.: Retrograde amnesia in rats produced by hippocampal injections of potassium chlo-
ride: gradient of effect and recovery. J. Compo Physioi. Psychol., 68,637-644, 1%9.
Hunt, H. F.: Some effects of drugs on classical (type S) conditioning. Ann. N. Y. Acad. Sci., 65,
258-267, 1956.
Hunt, H. F.: Some effects of meprobanate on conditioned fear and emotional behavior. Ann. N. Y.
Acad. Sci., 67,712-723, 1957.
Hunt, H. F., and Brady, J. V.: Some effects of electro-convulsive shock on a conditioned emo-
tional response ("anxiety"). J. Compo Physiol. Psychol., 44, 88-98, 1951a.
Hunt, H. F., and Brady, J. V.: Some quantitative and qualitative differences between "anxiety"
and "punishment" conditioning. Am. Psychol., 6,276-277, 1951b.
Hunt, H. F., and Brady, J. V.: Some effects of punishment and intercurrent "anxiety" on a simple
operant. J. Compo Physioi. Psychol., 48,305-310, 1955.
Hunt, H. F., Jernberg, P., and Brady, J. V.,: The effect of electroconvulsive shock (ECS) on a
conditioned emotional response: the effect of post-ECS extinction on the reappearance of the
response. J. Compo Physiol. Psychol., 45, 589-599, 1952.
Hunt, H. F., Jernberg, P., and Lawlor, W. G.: The effect of electro-convulsive shock on a condi-
tioned emotional response: the effect of electro-convulsive shock under anesthesia. J. Compo
Physiol. Psychol., 46,64-68, 1953.
Hurwitz, H. M. B., and Roberts, A. E.: Suppressing an avoidance response by a pre-aversive stim-
ulus. Psychonomic Sci., 17,305-306, 1%9.
Hurwitz, H. M. B., and Roberts, A. E.: Conditioned suppression of an avoidance response. J. Exp.
Anal. Behav., 16,275-281, 1971.
Jackson, D. E., and Delprato, D. J.: Aversive CSs suppress lever pressing for food, but not the
eating of free food. Learning Motivation, 5, 448-458, 1974.
James, J. P.: Acquisition, extinction and spontaneous recovery of conditioned suppression. Psy-
chonomic Sci., 22, 156-158, 1971a.
James, J. P.: Latent inhibition and the preconditioning-conditioning interval. Psychonomic Sci.,
24,97-98, 1971b.
James, J. P., and Mostoway, W. W.: Conditioned suppression of licking as a function of shock
intensity. Psychonomic Sci., 13, 161-162, 1%8.
James, W.: What is emotion? Mind, 9, 188-204, 1884.
James, W.: Principles of Psychology. New York, Holt, 1890.
Jones, M. C.: A laboratory study of fear: the case of Peter. J. Genet. Psychol., 31,308-315, 1924a.
Jones, M. C.: The elimination of children's fears. J. Exp. Psychol., 7,382-390, 1924b.
86 Section I: Concurrent Classical and Operant Conditioning

Kamano, D. K.: Effects of an extinquished fear stimulus on avoidance behavior. Psychonomic

Sci., 13,271-272, 1968.
Kamil, A. C.: The second-order conditioning of fear in rats. Psychonomic Sci., 10,99-100, 1968.
Kamil, A. C.: Some parameters on the second-order conditioning of fear in rats. J. Compo Physiol.
Psychol., 67,364-369, 1969.
Kamin, L. J.: Trace conditioning of the conditioned emotional response. J. Compo Physiol. Psy-
chol., 54, 149-153, 1961.
Kamin, L. J.: Backward conditioning and the conditioned emotional response. J. Compo Physiol.
Psychol., 56,517-519, 1963.
Kamin, L. J.: Temporal and intensity characteristics of the conditoned stimulus. In, W. F. Prokasy
(ed.): Classical Conditioning. New York, Appleton-Century-Crofts, 1965.
Kamin, L. J.: "Attention-like" processes in classical conditioning. In, M. Jones (ed.): Miami Sym-
posium on the Prediction of Behavior: Aversive Stimulation. Miami, University of Miami Press,
1968.
Kamin, L. J.: Predictability, surprise, attention and conditoning. In, B. Campbell and R. Church
(eds.): Punishment: a Symposium. New York, Appleton-Century-Crofts, 1969.
Kamin, L. J., and Brimer, C. J.: The effects of intensity of conditioned and unconditioned stimuli
on a conditioned emotional response. Can. J. Psychol., 17, 194-200, 1963.
Kamin, L. J., and Gaioni, S. J.: Compound conditioned emotional response conditioning with dif-
ferentially salient elements in rats. J. Compo Physiol. Psychol., 87,591-597, 1974.
Kamin, L. J., and Schaub, R. E.: Effects of conditioned stimulus intensity on the conditioned emo-
tional response. J. Compo Physiol. Psychol., 56,502-507, 1963.
Kamin, L. J., Brimer, C. J., and Black, A. H.: Conditioned suppression as a monitor of fear ofthe
CS in the course of avoidance training. J. Compo Physiol. Psychol., 56,497-501, 1963.
Kaufman, A.: Response suppression in the CER paradigm with extinction as the aversive event.
Psychonomic Sci., 15, 15-16, 1969.
Kelleher, R., Riddle, W., and Cook, L.: Persistent behavior maintained by unavoidable shocks. J.
Exp. Anal. Behav., 6,507-517, 1963.
Kellicut, M. H., and Schwartzbaum, J. S.: Formation of a conditioned emotional response (CER)
following lesions of the amygdaloid complex in rats. Psychol. Rep., 12,351-358,1963.
Kelly, D. D.: Long-term prereward suppression in monkeys unaccompanied by cardiovascular
conditionining. J. Exp. Anal. Behav., 20,93-104, 1973a.
Kelly, D. D.: Suppression of random-ratio and acceleration of temporally spaced responding by the
same prereward stimulus in monkeys. J. Exp. Anal. Behav., 20,363-373, 1973b.
Kenshalo, D. R., and Hall, E. C.: Thermal thresholds of the rhesus monkey (Macaca mulatta). J.
Compo Physiol. Psychol., 86, 902-910, 1974.
Kesner, R. P., Gibson, W. E., and Leclair, M. J.: ECS as a punishing stimulus: dependency on
route of administration. Physiol. Behav., 5,683-686, 1970.
Kimble, G. A.: Hilgard and Marquis' Conditioning and Learning. New York, Appleton-Century-
Crofts, 1961.
Kinnard, W. J., Aceto, M. D. G., and Buckley, J. P.: The effects of certain psychotropic agents on
the conditioned emotional response behavior pattern of the albino rat. Psychopharmacologia, 3,
227-230, 1962.
King, N. W., Justesen, D. R., and Clarke, R. L.: Behavioral sensitivity to microwave irradiation.
Science, 172,398-401, 1971.
Kling. J. 0.: Interstimulus intervals in conditioned suppression. J. Gen. Psychol., 87,297-299,
1972.
Kling. J. 0., and Kling, K.: Spontaneous recovery of conditioned suppression. J. Gen. Psychol.,
88, 151-152, 1973.
Kling, J. W., and Riggs, L. A.: Woodworth and Schlosberg's Experimental Psychology. New
York, Holt, Rinehart & Winston, 1971.
Kluver, H., and Bucy, P. C.: "Psychic blindness" and other symptoms following bilateral tem-
porallobectomy in rhesus monkeys. Am. J. Physiol., 119,352-353, 1937.
Kluver, H., and Bucy, P. C.: An analysis of certain effects of bilateral temporal lobectomy in the
rhesus monkey with special reference to "psychic blindness." J. Psychol., 5,33-54, 1938.
Chapter 1: Review of Classical-Operant Conditioning 87

Koch, S. (ed.): Psychology; a Study of Science, vol. 2. New York, McGraw-Hill, 1959.
Kohlenberg, R., and Trabasso, T.: Recovery of a conditioned emotional response after one or two
electroconvulsive shocks. J. Comp. Physiol. Psychol., 65,270-273, 1968.
Konorski, J.: Integrative Activity of the Brain. Chicago, University of Chicago Press, 1967.
Konorski, J.: Some ideas concerning physiological mechanisms of so-called internal inhibition. In,
R. S. Boakes and M. S. Halliday (eds.): Inhibition and Learning. New York, Academic Press,
1972.
Kremer, E. F.: Truly random and traditional control procedures in CER conditioning in the rat. J.
Comp. Physiol. Psychol., 76,441-448, 1971.
Kremer, E. F.: Properties of a preexposed stimulus. Psychonomic Sci., 27,45--47, 1972.
Kremer, E. F.: The truly random control procedure: conditioning to the static cues. J. Comp.
Physiol. Psychol., 86,700-707, 1974.
Kremer, E. F., and Kamin, L. J.: The truly random control procedure: associative or nonassocia-
tive effects in rats. J. Comp. Physiol. Psychol., 74,203-210, 1971.
Kruper, D. C.: Effects of a pre-aversive stimulus upon oddity performance in monkeys. J. Exp.
Anal. Behav., 11,71-75, 1968.
Kruper, D. C., and Haude, R.: Visual exploration in a conditioned emotional response paradigm. J.
Exp. Anal. Behav., 7,381-382, 1964.
Kumar, R.: Incubation of fear: experiments on the "Kamin effect" in rats. J. Comp. Physiol.
Psychol., 70,258-263, 1970.
Kumar, R., Stolerman, I. P., and Steinberg, H.: Psychopharmacology. Annu. Rev. Psychol., 21,
595-628, 1970.
Lange, C. G.: Om Sindsbevoegesler, (1885.) Leipzig, Theodor Thomas, 1887.
Lantz, A. E.: Effect of number oftrials, interstimulus interval, and dishabituation during CS habit-
uation on subsequent conditioning in a CER paradigm. Anim. Learning Behav., 1, 273-277,
1973.
Lauener, H.: Conditioned suppression in rats and the effect of pharmacological agents thereon.
Psychopharmacologia, 4, 311-325, 1963.
La Vail, M. M., Sidman, M., Rausin, R., and Sidman, R.: Discrimination oflight intensity by rats
with inherited retinal degeneration: a behavioral and cytological study. Vision Res., 14,693-702,
1974.
Leaf, R. C., and Leaf, S. R. P.: Recovery time as a measure of conditioned suppression. Psychol.
Rep., 18,265-266, 1966.
Leaf, R. C., and Muller, S. A.: Effect ofCER on DRL responding. J. Exp. Anal. Behav., 7,405-
407, 1964.
Leaf, R. C., and Muller, S. A.: Simple method for CER conditioning and measurement. Psycho/,
Rep., 17,211-215, 1965.
Leaf, R. C., Kayser, R. J., Andrews, J. S., Jr., Adkins, J. W., and Leaf, S. R. P.: Block of fear
conditioning induced by habituation or extinction. Psychonomic Sci., 10, 198-199, 1968.
Leitenberg, H.: Conditioned acceleration and conditioned suppression in pigeons. J. Exp. Anal.
Behav., 9,205-212, 1966.
Leitenberg, H., Bertsch, G. J., and Coughlin, R. C., Jr.: "Time-out from positive reinforcement"
as the UCS in a CER paradigm with rats. Psychonomic Sci., 13,3--4, 1968.
Lepore, F., Ptito, M., Frieberg, V., and Gullemot, J. P.: Effects oflow dose of chlorapromazine on
a conditioned emotional response in the rat. Psychol. Rep., 34,231-237, 1974.
Levine, M. S., Goldrich, S. G., Pond, F. J., Livesky, P., and Schwartzbaum, J. S.: Retrograde
amnestic effects of inferotemporal and amygdaloid seizures upon conditioned suppression of
lever-pressing in monkeys. Neuropsychologia, 8,431--442, 1970.
Lewis, J.: Conditioned suppression of a VI baseline using a two-bar multiple VI shock-avoidance
schedule. Anim. Learning Behav., 1,247-250, 1973.
Libby, A.: Two variables in the acquisition of depressant properties by a stimulus. J. Exp.
Psychol., 42, 100-107, 1951.
Lidsky, T. I., Levine, M. S., Kreinich, C. J., and Schwartzbaum J. S.: Retrograde effects ofamyg-
daloid stimulation on conditioned suppression (CER) in rats. J. Comp. Physiol. Psychol., 73,
135-149, 1970.
88 Section I: Concurrent Classical and Operant Conditioning

Linden, D. R.:Attenuation and reestablishment of the CER by discriminated avoidance condition-

ing in rats. J. Compo Physiol. Psychol., 69,573-578, 1969.
Lindsley, D. B.: Emotions and the electroencephalogram. In, M. L. Reymert (ed.): Feelings and
Emotions. New York, McGraw-Hill, 1950.
Lindsley, D. B.: Emotion. In, S. S. Stevens (ed.): Handbook oj Experimental Psychology. New
York, Wiley, 1951.
Lindsley, D. B.: Psychophysiology and motivation. In, M. R. Jones (ed.): Nebraska Symposium on
Motivation. Lincoln, University of Nebraska Press, 1957.
Lo Lordo, V. M.: Similarity of conditioned fear responses based upon different aversive events. J.
Compo Physiol. Psychol., 64, 154-158, 1967.
Lo Lordo, V. M.: Facilitation offood-reinforced responding by a signal for response-independent
food. J. Exp. Anal. Behav., 15,49-55, 1971.
Lo Lordo, V. M., McMillan, J. C., and Riley, A. L.: The effects of food-reinforced pecking and
treadle-pressing of auditory and visual signals for response-independent food. Learning Motiva-
tion, 5,24-41, 1974.
Lubow, R. E.: Latent inhibition. Psychol. Bull., 79,398-407, 1973.
Lubow, R. E., and Siebart, L.: Latent inhibition within the CER paradigm. J. Compo Physiol.
Psychol., 68, 136-138, 1969.
Lyon, D.O.: Frequency of reinforcement as a parameter of conditioned suppression. J. Exp. Anal.
Behav., 6,95-98, 1963.
Lyon, D.O.: Some notes on conditioned suppression and reinforcement schedules. J. Exp. Anal.
Behav., 7,289-291, 1964.
Lyon, D.O.: CER methodology: reply to Blackman. Psychol. Rep., 20,206, 1967.
Lyon, D.O.: Conditioned suppression: operant variables and aversive control. Psychol. Record,
18,317-338, 1968.
Lyon, D. 0., and Felton, M.: Conditioned suppression and fixed ratio schedules of reinforcement.
Psychol. Record, 16,433-440, 1966a.
Lyon, D.O., and Felton, M.: Conditioned suppression and variable-ratio reinforcement. J. Exp.
Anal. Behav., 9,245-250, 1966b.
Lyon, D.O., and Millar, R. D.: Conditioned suppression on a fixed interval schedule ofreinforce-
ment. Psychonomic Sci., 17,31-32, 1969.
Mackintosh, N. J.: Blocking of conditioned suppression: role of the first compound trial. J. Exp.
Psychol. [Anim. Behav. Processes], 1,335-345, 1975.
Mackintosh, N. J.: Overshadowing and stimulus intensity. Anim. Learning Behav., 4, 186-192,
1976.
MacLean, P. D.: Psychosomatic disease and the "visceral brain:" recent developments bearing on
the Papez theory of emotion. Psychosom. Med., 11,388-353, 1949.
MacLean, P. D.: Some psychiatric implications of physiological studies on frontotemporal portion
of limbic system (visceral brain). Electroencephal. Clin. Neuropsychiatry, 4,407-418,1952.
Mahoney, W. J., and Ayres, J. J. B.: One-trial simultaneous and backward fear conditioning as
reflected in conditioned suppression of licking in rats. Anim. Learning Behav., 4, 357-362,
1976.
Maier, S. F., Rapoport, P., and Wheatley, K.: Conditioned inhibition and the UCS-CS interval.
Anim. Learning Behav., 4,217-220, 1976.
Malmo, R. B.: Anxiety and behavioral arousal. Psychol. Rev., 64,276-287, 1957.
Malmo, R. B.: Activation. In, A. J. Bachrach (ed.): Experimental Foundations oj Clinical Psychol-
ogy. New York, Basic Books, 1962.
Martin, L. K., and Riess, D.: Effects of US intensity during previous discrete delay conditioning on
conditioned acceleration during avoidance extinction. J. Compo Physiol. Psychol., 69, 196-200,
1969.
Maser, J. D., and Hammond, L. J.: Disruption of a temporal discrimination by the minor tranquil-
izer, oxazepam. Psychopharmacologia, 25,69-76, 1972.
Mason, J. W., and Brady, J. V.: Plasma 17-hydroxycorticosteroid changes related to reserpine ef-
fects on emotional behavior. Science, 124, 983-984, 1956.
Masserman, J. H.: Behavior and Neurosis. Chicago, University of Chicago Press, 1943.
Chapter 1: Review of Classical-Operant Conditioning 89

Masterton, B., Heffner, H., and Ravizza, R.: The evolution of human hearing. J. Acoust. Soc.
Am., 45,966-985, 1969.
Matthews, T. J., McHugh, T. G., and Carr, L. D.: Pavlovian and instrumental determinants of
response suppression in the pigeon. J. Compo Physiol. Psychol., 87,500-506, 1974.
May, R. B., Tolman, C. W., and Schoenfeldt, M. G.: Effects of pre-training exposure to the CS on
conditioned suppression. Psychonomic Sci., 9,61-62, 1967.
McAllister, W. R., and McAllister, D. E.: Behavioral measurement of conditioned fear. In, F. R.
Brush (ed.): Aversive Conditioning and Learning. New York, Academic Press, 1971.
McIntire, R. W.: Conditioned suppression and self-stimulation. Psychonomic Sci., 5, 273-274,
1966.
McIntyre, D. C.: Differential amnestic effect of cortical vs amygdaloid elicited convulsions in rats.
Physiol. Behav., 5,747-753, 1970.
McIntyre, D. C., and Molino, A.: Amygdala lesions and CER learning: long term effects of kin-
dling. Physiol. Behav., 8, 1055-1058, 1972.
Meltzer, D., and Brahlek, J. A.: Conditioned suppression and conditioned enhancement with the
same positive UCS: an effect of CS duration. J. Exp. Anal. Behav., 13,67-73, 1970.
Merrill, H., Lott, W. J., and Bergen, B. J.: Attenuation of a conditioned emotional response via
reinforcing intracranial stimulation in rats. J. Compo Physiol. Psychol., 71,426-434, 1970.
Miczek, K.: Effects of scopolamine, amphetamine, and benzodiazepines on conditioned suppres-
sion. Pharmacol. Biochem. Behav., 1,401-411, 1973.
Miczek, K. A., and Grossman, S. P.: Positive conditioned suppression: effects of CS duration. J.
Exp. Anal. Behav., 15,243-247, 1971.
Migler, B., and Brady, J. Y.: Timing behavior and conditioned fear. J. Exp. Anal. Behav., 7,247-
251, 1964.
Miller, G. A., Galanter, E., and Pribram, K.: Plans and the Structure of Behavior. New York,
Holt, Rinehart & Winston, 1960.
Miller, L.: Compounding of pre-aversive stimuli. J. Exp. Anal. Behav., 12,293-299, 1969.
Miller, N. E.: Studies of fear as an acquirable drive: I. Fear as motivation and fear-reduction as
reinforcement in the learning of new responses. J. Exp. Psychol., 38,89-101, 1948.
Miller, N. E.: Learnable drives and rewards. In, S. S. Stevens (ed.): Handbook of Experimental
Psychology. New York, Wiley, 1951.
Miller, N. E.: Liberalization of basic S-R concepts: extenstion to conflict behavior, motivation and
social learning. In, S. Koch (ed.): Psychology: a Study of Science, vol. 2. New York, McGraw-
Hill, 1959.
Miller, S., and Konorski, J.: Sur une forme particuliere des reflexes conditionnels. C. R. Soc. Pol.
Bioi. 99, 1155-1157, 1928.
Morris, D. D.: Threshold for conditioned suppression using x rays as the pre-aversive stimulus. J.
Exp. Anal. Behav., 9,29-34, 1966.
Mowrer, O. H.: A stimUlus-response analysis of anxiety and its role as a reinforcing agent.
Psychol. Rev., 46, 553-565, 1939.
Mowrer, O. H.: Anxiety reduction and learning. J. Exp. Psychol., 27,497-516, 1940.
Mowrer, O. H.: Learning Theory and Personality Dynamics. New York, Ronald Press, 1950.
Mowrer, O. H.: Learning Theory and Behavior. New York, Wiley, 1960.
Nachman, M., Meinecke, R. 0., and Baumbach, H. D.: Temporal gradient of recovery ofa condi-
tioned emotional response following a single electroconvulsive shock. Psychonomic Sci., 17, 137
-138, 1969.
Nadel, L.: Dorsal and ventral hippocampal lesions and behavior. Physiol. Behav., 3, 891-900,
1968.
N ageishi, Y., and Imada, H.: Suppression of licking behavior in rats as a function of predictability
of shock and probability of conditioned stimulus-shock pairings. J. Compo Physiol. Psychol., 87,
1165-1173, 1974.
Nathan, M. A., and Smith, O. A., Jr.: Differential conditional emotional and caridovascular re-
sponses-a training technique for monkeys. J. Exp. Anal. Behav., 11,77-82, 1968.
Nathan, M. A., and Smith, O. A., Jr.: Conditional cardiac and suppression responses after lesions
in the dorsomedial thalamus of monkeys. J. Compo Physiol. Psychol., 76,66-73, 1971.
90 Section I: Concurrent Classical and Operant Conditioning

Olds, J.: Physiological mechanisms of reward. In, M. R. Jones (ed.): Nebraska Syposium on Moti-
vation. Lincoln, University of Nebraska Press, 1955.
Olds, J.: A preliminary mapping of electrical reinforcing effects in the rat brain. J. Compo Physiol.
Psychol., 49,281-285, 1956.
Oley, N., DeRan, R. S., Tucker, D., Smith, J. c., and Graziadei, P. P. c.: Recovery of structure
and function following transection ofthe primary olfactory nerves in pigeons. J. Compo Physiol.
Psychol., 88,477-495, 1975.
Orme-Johnson, D. W., and Yarczower, M.: Conditioned suppression, punishment and aversion. J.
Exp. Anal. Behav., 21,57-74, 1974.
Overmeir. J. B.: Differential transfer of avoidance responses as a function of UCS duration. Psy-
chonomic Sci., 5, 25-26, 1%6a.
Overmeir, J. B.: Instrumental and cardiac indices of Pavlovian fear conditioning as a function of
US duration. J. Compo Physioi. Psychol., 62, 15-20, 1%6b.
Overmeir, J. B., and Bull, J. A., III: An interaction between appetitive Pavlovian CSs and instru-
mental avoidance responding. Learning Motivation, 1, 18-26, 1970.
Overmeir, J. B., and Leaf, R. C.: Effects of discriminative Pavlovian fear conditioning upon pre-
viously or subsequently acquired avoidance responding. J. Compo Physiol. Psychol., 60, 213-
217, 1%5.
Overmeir, J. B., Bull, J. A., III, and Pack, K.: On instrumental response interaction as explaining
the influences of Pavlovian CS+s upon avoidance behavior. Learning Motivation, 2, 103-112,
1971.
Palfai, T., and Cornell, J. M.: Effect of drugs on consolidation of classically conditioned fear. J.
Compo Physiol. Psychol., 66, 584-589. 1968.
Papez, J. W.: A proposed mechanism of emotion. Arch. Neurol. Psychiatry, 38, 725-743,
1937.
Pare, W. P.: Interaction of age and shock intensity on acquisition of a discriminated conditioned
emotional response. J. Compo Physiol. Psychol., 68, 364-369, 1969.
Parker, B., Barker, D. L., and Topping, J.: The differential effects of timeout and shock on DRL
responding in the CER paradigm. Psychonomic Sci., 22, 133-135, 1971.
Parkinson, S. R.: Sensory preconditioning of a conditioned emotional response. Psychonomic Sci.
11, 119, 1968.
Parrish, J.: Classical discrimination conditioning of heart rate and bar press suppression in the rat.
Psychonomic Sci., 9,267-268, 1%7.
Pavlov, I. P.: Conditioned Reflexes. New York, Dover, 1927.
Persinger, M. A., and Pear, J. J.: Prenatal exposure to an ELF-rotating magnetic field and subse-
quent increase in conditioned suppression. Develop. Psychol., 5, 269-274, 1972.
Pert, A., and Avis, R. R.: Dissociation between scopolamine and mecamylamine during fear con-
ditioning in rats. Physiol. Psychol., 2, 111-116, 1974.
Pierson, S. C.: Conditioned suppression to odorous stimuli in the rat. J. Compo Physiol. Psychol.,
86,708-717, 1974.
Pliskoff, S.: Rate-change effects during a pre-schedule change stimulus. J. Exp. Anal. Behav., 4,
383-386, 1%1.
Pliskoff, S.: Rate-change effects with equal potential reinforcements during the "warning" stimu-
lus. J. Exp. Anal. Behav., 6,557-562, 1963.
Pomerleau, 0.: The effects of stimuli followed by response-independent shock on shock-avoidance
behavior. J. Exp. Anal. Behav., 14, 11-21, 1970.
Powell, R. W.: The pulse-to-cycle fraction as a determinant of critical flicker fusion in the pigeon.
Psychol. Rec., 17, 151-160, 1%7.
Powell, R. W., and Smith, J. C.: Critical-flicker-fusion thresholds as a function of very small pulse-
to-cycle fractions. Psychol. Record, 18, 35-40, 1%8.
Premack, D.: Reinforcement theory. In, D. Levine (ed.): Nebraska Syposium on Motivation. Lin-
coln, University of Nebraska Press, 1965.
Prewett, E. P.: Number of preconditioning trials in sensory preconditioning using CER training. J.
Compo Physiol. Psychol., 64,360-362, 1%7.
Price, L. L., Dalton, L. W., Jr., and Smith, J. C.: Frequency DL in the pigeon as determined by
conditioned suppression. J. Auditory Res., 7,229-239, 1%7.
Chapter 1: Review of Classical-Operant Conditioning 91

Quinsey, V. L., and Ayres, J. J. B.: Shock induced facilitation of a partially extinquished CER.
Psychonomic Sci., 14,213-214, 1969a.
Quinsey, V. L., and Ayres, J. J. B.: The effect ofCS-induced fear on a partially extinguished CER.
Psychonomic Sci., 14, 242-244, 1969b.
Rand, G. V., Sloane, H. N., and Dobson, W. R.: Some variables affecting conditioned suppression
in humans. Behav. Ther., 2, 554-559, 1971.
Ravizza, R. J., Heffner, H. E., and Masterton, B.: Hearing in primitive mammals: I. Opossum
(Didelphis virginianus). J. Auditory Res., 9, 1-7, 1969.
Ray, 0.: Tranquilizer effects on conditioned suppression. Psychopharmacologia, 5, 136-146,
1964.
Ray, 0., and Stein, L.: Generalization of conditioned suppression. J. Exp. Anal. Behav., 2,357-
361, 1959.
Reberg, D.: Compound tests for excitation in early acquisition and after prolonged extinction of
conditioned suppression. Learning Motivation, 3, 246-258, 1972.
Reberg, D., and Black, A. H.: Compound testing of individually conditioned stimuli as an index of
excitatory and inhibitory properties. Psychonomic Sci., 17,30-31, 1969.
Reiss, S., and Wagner, A. R.: CS habituation produces a "latent inhibition effect" but no active
"conditioned inhibition." Learning Motivation, 3,237-245, 1972.
Rescorla, R. A.: Predictability and number of pairings in Pavlovian fear conditioning. Psychonomic
Sci., 4,383-384, 1966.
Rescorla, R. A.: Inhibition of delay in Pavlovian fear conditioning. J. Compo Physiol. Psychol., 64,
114-120, 1967a.
Rescorla, R. A.: Pavlovian conditioning and its proper control procedure. Psychol. Rev., 74,71-
80, 1967b.
Rescorla, R. A.: Pavlovian conditioned fear in Sidman avoidance learning. J. Compo Physiol.
Psychol., 65, 55-60, 1968a.
Rescorla, R. A.: Probability of shock in the presence and absence of CS in fear conditioning. J.
Compo Physiol. Psychol., 66, 1-5, 1968b.
Rescorla, R. A.: Conditioned inhibition offear. In, W. K. Honig and N. J. Mackintosh (eds.): Fun-
damental Issues in Associative Learning. Halifax, Dalhousie University Press, 1969a.
Rescorla, R. A.: Conditioned inhibition of fear resulting from negative CS-UCS contingencies. J.
Compo Physiol. Psychol., 67, 504-509, 1969b.
Rescorla, R. A.: Pavlovian conditioned inhibition. Psychol. Bull., 72,77-94, 1969c.
Rescorla, R. A.: Reduction in the effectiveness of reinforcement following prior excitatory train-
ing. Learning Motivation, 1,372-381, 1970.
Rescorla, R. A.: Summation and retardation tests of latent inhibition. J. Compo Physiol. Psychol.,
75,77-81, 1971a.
Rescorla, R. A.: Variations in the effectiveness of reinforcement and nonreinforcement following
prior inhibitory conditioning. Learning Motivation, 2, 113-123, 1971b.
Rescorla, R. A.: Effects of US habituation following conditioning. J. Compo Physiol. Psychol., 82,
137-143, 1973.
Rescorla, R. A.: Effect of inflation of the value of the unconditioned stimulus following condition-
ing. J. Compo Physiol. Psychol., 86, 101-106, 1974.
Rescorla, R. A.: Second-order conditioning of Pavlovian conditioned inhibition. Learning Motiva-
tion, 7, 161-172, 1976.
Rescorla, R. A., and Furrow, D. R.: Stimulus similarity as a determinant of Pavlovian condition-
ing. J. Exp. Psychol. [Anim. Behav. Processes], 3,203-215, 1977.
Rescorla, R. A., and Heth, C. D.: Reinstatement of fear to an extinguished conditioned stimulus. J.
Exp. Psychol. [Anim. Behav. Processes], 104, 88-96, 1975.
Rescorla, R. A., and Lo Lordo, V. M.: Inhibition of avoidance behavior. J. Compo Physiol.
Psychol., 59,406-412, 1965.
Rescorla, R. A., and Solomon, R.: Two-process learning theory: relationships between Pavlovian
conditioning and instrumental learning. Psychol. Rev., 74, 151-182, 1967.
Rescorla, R. A., and Wagner, A. R.: A theory of Pavlovian conditioning: variations in the effective-
ness of reinforcement and nonreinforcement. In, A. H. Black and W. F. Prokasy (eds.): Classi-
cal Conditioning, vol. 2. New York, Appleton-Century-Crofts, 1972.
92 Section I: Concurrent Classical and Operant Conditioning

Riess, D.: The buzzer as a primary aversive stimulus: I. Unconditioned acceleration and summa-
tion of conditioned and unconditioned acceleration. Psychonomic Sci., 21, 167-169, 1970a.
Riess, D.: The buzzer as a primary aversive stimulus: II. Unavoidable buzzer presentation and
conditioned acceleration. Psychonomic Sci., 21,302-304, 1970b.
Riess, D.: The buzzer as a primary aversive stimulus: III. Unconditioned and conditioned suppres-
sion of barpress avoidance. Psychonomic Sci., 24, 212-214, 1971.
Riess, D., and Farrar, C. H.: UCS duration and conditioned suppression: acquisition and extinc-
tion between-groups and terminal performance within-subjects. Learning Motivation, 4, 366-
373, 1973a.
Riess, D., and Farrar, C. H.: Us duration, conditioned acceleration, mUltiple CR measurement and
Pavlovian R-R laws in rats. J. Compo Physiol. Psychol., 82, 144-151, 1973b.
Rioch, D. M.: Certain aspects of the behavior of decorticate cats. Psychiatry, 1,339-345, 1938.
Rizley, R. C., and Rescorla, R. A.: Associations in second-order conditioning and sensory precon-
ditioning. J. Compo Physiol. Psychol., 81, 1-11, 1972.
Roberts, A. E., and Hurwitz, H. M. B.: The effect of a pre-shock signal on a free operant avoid-
ance response. J. Exp. Anal. behav., 14, 331-340, 1970.
Roberts, A. E., Copper, K. G., and Richey, T. L.: Rat behaviors duringunsignalled avoidance and
conditioned suppression training. Bull. Psychonomic Soc., 9,373-376, 1977.
Rogers, J. D.: Stimulus intensity and trace intervals in sensory preconditioning using the CER.
Bull. Psychonomic Soc., 1, 107-109, 1973.
Rohrbaugh, M., Riccio, D. S., and Arthur, A.: Paradoxical enhancement of conditioned suppres-
sion. Behav. Res. Ther., 10, 125-130, 1972.
Rosenberger, P. B.: Response-adjusting stimulus intensity. In, W. C. Stebbins (ed.): Animal Psy-
chophysics: the Design and Conduct oj Sensory Experiments. New York, Appleton-Century-
Crofts, 1970.
Rosenberger, P. B., and Ernst, J. T.: Behavioral assessment of absolute visual thresholds in the
albino rat. Vision Res., 11, 199-207, 1971.
Sachs, D. A., and Keller, T.: Intensity and temporal characteristics of the CER paradigm with
humans. J. Gen. Psychol., 86, 181-188, 1972.
Sachs, D. A., and May, J. G., Jr.: Conditioned emotional response with humans: the effect of a
variable interstimulus interval using a trace conditioning paradigm. Psychonomic Sci., 9,343-
344,1967.
Sachs, D. A., and May, J. G. Jr.: The presence of a temporal discrimination in the conditioned
emotional response with humans. J. Exp. Anal. Behav., 12, 1003-1007, 1969.
Scheuer, C.: Resistance to extinction of the CER as a function of shock-reinforcement training
schedules. Psychonomic Sci., 17, 181-182, 1969.
Scheuer, C., and Keeter, W. H.: Temporal vs discriminative factors in the maintenance of condi-
tioned suppression: a test of the information hypothesis. Psychonomic Sci., 15,21-23, 1969.
Schnur, P.: Selective attention: effect of element preexposure on compound conditioning in rats. J.
Compo Physiol. Psychol., 76, 123-130, 1971.
Schwartz, B.: Positive and negative conditioned suppression in the pigeon: effects of the locus and
modality of the CS. Learning Motivation, 7,86-100, 1976.
Schwartz, B., and Garnzu, E.: Pavlovian control of operant behavior: an analysis of auto shaping
and of interactions between multiple schedules of reinforcement. In, W. K. Honig and
J. E. R. Staddon (eds.): Handbook oj Operant Behavior. Englewood Cliffs, N. J., Prentice-Hall,
1977.
Scobie, S. R.: Interaction of an aversive Pavlovian conditioned stimulus with aversively and appe-
titively motivated operants in rats. J. Compo Physiol. Psychol., 79, 171-188, 1972.
Scobie, S. R.: The response-shock interval and conditioned suppression of avoidance in rats.
Anim. Learning Behav., 1, 17-20, 1973.
Scobie, S. R., and Garske, G.: Chlordiazepoxide and conditioned suppression. Psychopharmaco-
logia, 16, 272-280, 1970.
Sechenov, I. Selected Works. Moscow, State Publishing, 1935.
Seger, K. A., and Scheuer, C.: The informational properties of SI, S2, and SI-S2 sequence on
conditioned suppression. Anim. Learning Behav., 5,39-41, 1977.
Chapter I: Review of Classical-Operant Conditioning 93

Seligman, M. E.: Chronic fear produced by unpredictable electric shock. J. Compo PhysioL. Psy-
choL., 66,402-411, 1968.
Seligman, M. E., Maier, S. F., and Solomon, R. L.: Unsignalled and uncontrollable aversive
events. In R. R. Brush (ed.): Aversive Conditioning and Learning. New York, Academic Press,
1971.
Shaber, G. S., Ramsey, J. A., Ill., Dorn, B. C., and Brent, R. L.: Saccharin behavior taste thresh-
old in the rat. Fed. Proc., 26,543, 1967.
Sherrington, C. S.: Experiments on the value of vascular and visceral factors for the genisis of
emotion. Proc. R. Soc., 366,390-433, 1900.
Sherrington, C. S.: The Integrative Action of the Nervous System. New Haven, Yale University
Press, 1906.
Shimoff, E.: Measurement of behavioral effects of the CER procedure. PsychoL. Rep., 30, 67-71,
1972a.
Shimoff, E.: Avoidance responding as a function of stimulus duration and relation to free shock. 1.
Exp. Anal. Behav., 17,451-461, 1972b.
Shinkman, P. G., and Kaufman, K. P.: Posttrial hippocampal stimulation and CER acquisition in
the rat. 1. Compo PhysioL. Psychol., 80,283-292, 1972a.
Shinkman, P. G., and Kaufman, K. P.: Time course of retroactive effects of hippocampal stimula-
tion on learning. Exp. Neur., 34,476-483, 1972b.
Shipley, R.: Extinction of conditioned fear in rats as a function of several parameters of CS expo-
sure. 1. Compo Physiol. Psychol., 87,699-707, 1974.
Shumake, S. A., Smith, J. C., and Taylor, H. L.: Critical fusion frequency in rhesus monkeys using
the conditioned suppression technique. 6571 st Aeromedical Research Laboratory, Alaomgordo,
N. Mex., Tech. Rep. ARL-TR-67-22, 1967.
Shumake, S. A., Smith, J. C., and Taylor, H. L.: Critical fusion frequency in rhesus monkeys.
Psychol. Record, 8,537 -542, 1%8.
Shumake, S. A., Thompson, R. D., and Caudill, C. J.: A technique for visual threshold measure-
ment in vampire bats. Physiol. Behav., 18,325-327, 1977.
Sidman, M.: By-products of aversive control. 1. Exp. Anal. Behav., 1,265-280, 1958.
Sidman, M.: Normal sources of pathological behavior. Science, 132,61-68. I 960a.
Sidman, M.: Tactics of scientific research: evaluating experimental data in psychology. New York,
Basic Books, 1960b.
Sidman, M., Hermstein, R. J., and Conrad, D. G.: Maintenance of avoidance behavior by unavoid-
able shock. 1. Compo Physiol. Psychol., 50,553-557, 1957.
Sidman, M., Ray, B. A., Sidman, R. L., and Klinger, J. M.: Hearing and vision in neurological
mutant mice: a method for their evaluation. Exp. Neurol., 16,377-402, 1%6.
Siegal, S., and Domjan, M.: Backward conditioning as an inhibitory procedure. Learning Motiva-
tion, 2, 1-11, 1971.
Siegal, S., and Domjan, M.: The inhibitory effect of backward conditioning as a function of the
number of backward pairings. Bull. Psychonomic Soc., 4, 122-124, 1974.
Singh, S. D.: Conditioned emotional response in the rat. I. Constitutional and situational determi-
nants.l. Compo Physiol. Psychol., 52, 574-578, 1959.
Skinner, B. F.: The Behavior of Organisms. New York, Appleton-Century-Crofts, 1938.
Skinner, B. F.: Are theories of learning necessary? Psychol. Rev., 57, 193-216, 1950.
Skinner, B. F.: Science and Human Behavior. New York, MacMillan, 1953.
Skinner, N. P.: Comment on Sachs and May's paper on the conditioned emotional response with
humans. Psychol. Rep., 34, 1069-1070, 1974.
Smith, J. B.: Effects of response rate, reinforcement frequency, and the duration of a stimulus pre-
ceding response-independent food. 1. Exp. Anal. Behav., 21,215-221, 1974.
Smith, J. c.: Conditioned suppression as an animal psychophysical technique. In W. C. Stebbins
(ed.): Animal Psychophysics: the Design and Conduct of Sensory Experiments. New York,
Appleton-Century-Crofts, 1970.
Smith, J. C., and Tucker, D.: Olfactory mediation of immediate x-ray detection. In, C. Pfaffmann
(ed.): Olfaction and Taste III. New York, Rockfeller University Press, 1969.
Smith, O. A., Jr., and Nathan, M. A.: The development of cardiac and blood flow conditional re-
94 Section I: Concurrent Classical and Operant Conditioning

sponses during the acquisition of a differentiated "conditioned emotional response" in monkeys.

Conditional Reflex, 2, 155-156, 1%7.
Snedden, D. S., Spevack, A. A., and Thompson, W. R.: Conditioned and unconditioned suppres-
sion as a function of age in rats. Can. J. Psychol., 25,313-322, 1971.
Solomon, R. L., and Wynne, L. C.: Traumatic avoidance learning: acquisition in normal dogs.
Psychol. Monogr., 67, 1-19, 1953.
Solomon, R. L., and Wynne, L. C.: Traumatic avoidance learning: the principle of anxiety conser-
vation and partial irreversibility. Psychol. Rev., 61, 353-385, 1954.
Solyom, L., and Miller, S.: The etTect of age differences on the acquisition of operant and classical
conditioned responses in rats. J. Gerontol., 20,311-314, 1965.
Spiegal, E. A., and Wycis, H. T.: Physiological and psychological results of thaiamontomy. Proc.
R. Soc. Med. [Suppl.], 42, 1-9, 1949.
St. Claire-Smith, R., and Mackintosh, N. J.: Complete suppression to a compound CS does not
block further conditioning to each element. Can. J. Psychol., 28,92-101, 1974.
Stebbins, W. C., and Smith, O. A., Jr.: Cardiovascular concomitants of the conditioned emotional
response in the monkey. Science, 144,881-883, 1964.
Stein, L., and Berger, B. D.: Paradoxical fear-increasing etTects of tranquilizers: evidence of re-
pression of memory in the rat. Science, 166,253-256, 1%9.
Stein, L., Sidman, M., and Brady, J. V.: Some etTect of two temporal variables on conditioned
suppression. J. Exp. Anal. Behav., 1, 153-162, 1958.
Stein, N., HotTman, H. S., and Stitt, C.: Collateral behavior of the pigeon during conditioned
suppressIon of key pecking. J. Exp. Anal. Behav., 15,83-93, 1971.
Stolurow, L. M.: Conditioning. In, B. B. Wolman (ed.): Handbook of General Psychology. Eng-
lewood Cliffs, N.J., Prentice-Hall, 1973.
Strouthes, A.: EtTect of CS-onset, UCS-termination delay, UCS duration, CS-onset interval and
number of CS-UCS pairings on conditioned fear response. J. Exp. Psychol., 69, 287-291,
1%5.
Suboski, M. D., Marquis, H. A., Black, M., and Platenius, P.: Adrenal and amygdala function in
the incubation of aversively conditioned responses. Physiol. Behav., 5,283-289, 1970.
Tait, R. W., and Suboski, M. D.: Stimulus intensity in sensory preconditioning of rats. Can. J.
Psychol., 26,374-381, 1972.
Tait, R. W., Marquis, H. A., Williams, R., Weinstein, L., and Suboski, M. D.: Extinction of sen-
sory preconditioning using CER training. J. Compo Physiol. Psychol., 69, 170-172, 1%9.
Tait, R. W., Simon, E., and Suboski, M. D.: "Partial reinforcement" in sensory preconditioning
with rats. Can. J. Psychol., 25,427-435, 1971.
Tait, R. W., Black, M., Katz, M., and Suboski, M. D.: Discriminative sensory preconditioning.
Can. J. Psychol., 26,201-205, 1972.
Tarpy, R. M.: Incubation of anxiety as measured by response suppression. Psychonomic Sci., 4,
189-190, 1%6.
Tenen, S.: Recovery time as a measure of CER strength: etTect of benzodiazepines, amobarbital,
chloropromazine, and amphetamine. Psychopharmacologia, 12, 1-17, 1%7.
Testa, T. J.: EtTects of similarity of location and temporal intensity pattern of conditioned and un-
conditioned stimuli on the acquisition of conditioned suppression in rats. J. Exp. Psychol. [Anim.
Behav. Processes], 104, 114-121, 1975.
Thompson, J. B., and Schwartzbaum, J. S.: Discrimination behavior and conditioned suppression
(CER) following localized lesions in the amygdala and putamen. Psychol. Rep., 15, 587-606,
1964.
Torres, A. A.: Anxiety vs escape conditioning and tranquilizing agents. J. Compo Physiol.
Psychol., 54,349-353, 1961.
Trafton, C. L.: EtTects oflesions in the septal area and cingulate cortical areas on conditioned sup-
pression of activity and avoidance behavior in rats. J. Compo Physiol. Psychol., 63, 191-197,
1967.
Valenstein, E. S.: The etTects ofreserpine on the conditioned emotional response of the quinea pig.
J. Exp. Anal. Behav., 2,219-225, 1959.
Van Dyne, G.: Conditioned suppression with a positive US in the rat. J. Compo Physiol. Psychol.,
77, 131-135, 1971.
Chapter 1: Review of Classical-Operant Conditioning 95

Van Houten, R., O'Leary, K. D., and Weiss, S. J.: Summation of conditioned suppression.J. Exp.
Anal. Behav., 13,75-81, 1970.
Vogel, J. R., and Spear, N. E.: Interaction of reward magnitude and conditioned fear on the con-
sumatory response. Psychonomic Sci., 5,263-264, 1966.
Vogel, J. R., Hughes, R. A., and Carlton, P. L.: Scopolamine, atropine, and conditioned fear. Psy-
chopharmacologia, 10,409-416, 1967.
Wagner, A. R.: Stimulus validity and stimulus selection. In, W. K. Honig and N. J. Mackintosh
(eds.): Fundamental Issues in Associative Learning. Halifax, Dalhousie University Press, 1969.
Wagner, A. R., and Rescorla, R. A.: Inhibition in Pavlovian conditioning: application of a theory.
In, R. S. Boakes and M. S. Halliday (eds.): Inhibition and Learning. New York, Academic
Press, 1972.
Wagner, A. R., Siegal, L. S., and Fein, G. G.: Extinction of conditioned fear as a function of per-
centage of reinforcement. J. Compo Physiol. Psychol., 63, 160-164, 1967.
Waller, M. B., and Waller, D. F.: The effects of unavoidable shocks on a multiple schedule having
an avoidance component. J. Exp. Anal. Behav., 6,29-37, 1963.
Watson, J. B., and Morgan, J. J. B.: Emotional reactions and psychological experimentation. Am.
J. Psychol., 28, 163-174, 1917.
Watson, J. B., and Raynor, R.: Conditioned emotional reactions. J. Exp. Psychol., 3, 1-14, 1920.
Weiss, K. M.: Some effects of the conditioned suppression paradigm on operant discrimination
performance. J. Exp. Anal. Behav., 11,767-775, 1968.
Weiskrantz, L.: Emotion. In, L. Weiskrantz (ed.): Analysis of Behavioral Change. New York,
Harper & Row, 1968.
Weiskrantz. L., and Wilson, W. A., Jr.: The effects of reserpine on emotional behavior of normal
and brain-operated monkeys. Ann. N. Y. Acad. Sci., 61,36-55, 1955.
Weisman, R. G., and Litner, J. S.: The course of Pavlovian extinction and inhibition of fear in rats.
J. Compo Physiol. Psychol., 69,667-672, 1969.
Weiss, S. J.: Stimulus compounding in free-operant and classical conditioning: a review and analy-
sis. Psychol. Bull., 78, 189-208, 1972.
Weiss, S. J., and Emurian, H. H.: Stimulus control during the summation of conditioned suppres-
sion. J. Exp. Psychol., 85, 204-209, 1970.
Wenger, M. A.: Emotion as visceral action: an extension of Lange's theory. In, M. L. Reymert
(ed.): Feelings and Emotions. New York, McGraw-Hill, 1950.
Whitney, G. D., and Trost, J. D.: Response disruption following amphetamine self- and
programmed-administration. In, R. T. Harris, W. M. McIsaac, and C. R. Shuster (eds.): Drug De-
pendence. Austin, University of Texas Press, 1970.
Willi, F. J.: The effect of electroconvulsive shock on a conditioned emotional response in the rhe-
sus monkey. Br. J. Psychol., 60, 509-521, 1969.
Willis, R. D.: The partial reinforcement of conditioned suppression. J. Compo Physiol. Psychol.,
68, 289-295, 1969.
Willis, R. D., and Lundin, R. W.: Conditioned suppression in the rat as a function of shock rein-
forcement schedule. Psychonomic Sci., 6, 107-108, 1966.
Wilson, G. T., and Davison, G. C.: Processes of fear reduction in systematic desensitization: ani-
mal studies. Psychol. Bull., 76, 1-14, 1971.
Wilson, L. M., and Riccio, D. C.: CS facilitation and conditioned suppression in weanling and adult
albino rats. Bull. Psychonomic Soc., 1, 184-186, 1973.
Wilson, W. L., Darcy, J. M., and Haralson, J. V.: Reserpine and conditioned suppression in the
fish Tilapia H Marcrocephala. Psychonomic Sci., 20,47-48, 1970.
Winocur, G., and Mills, J. A.: Aversive consequences of electroconvulsive shock. Physiol.
Behav., 5,631-634, 1970.
Winograd, E.: Maintained generalization testing of conditioned suppression. J. Exp. Anal. Behav.,
8,47-51, 1965.
Wise, C. D., Berger, B. D., and Stein, L.: Serotonin: a possible mediator of behavioral suppression
induced by anxiety. Dis. Nerv. Syst., 31,34-37, 1970.
Witcher, E. S., and Ayres, J. J. B.: Effect of removing background white noise during CS presenta-
tion on conditioning in the truly random control procedure. Bull. Psychonomic Soc., 6,25-27,
1975.
96 Section I: Concurrent Classical and Operant Conditioning

Wolpe, J.: The formation of negative habits: a neurological view. Psychol. Rev., 59,290-299, 1952.
Wolpe, J.: Psychotherapy by reciprocal inhibition. Stanford, Calif., Stanford University Press,
1958.
Wolpe, J.: The experimental foundations of some new psychotherapeutic methods. In, A. J.
Bachrach (ed.): Experimental Foundations of Clinical Psychology. New York, Basic Books,
1%2.
Woodworth, R. S., and Sherrington, C. S.: A pseudoaffective reflex and its spinal path. J. Physiol.,
31, 234-243, 1904.
Yamahiro, H. S., Bell, E. C., and Hill, H. E.: The effects of reserpine on a strongly conditioned
emotional response. Psychopharmacologia, 2, 197-202, 1%1.
Yeo, A. G.: The acquisition of conditioned suppression as a function of interstimulus interval dura-
tion. Q. J. Exp. Psychol., 26,405-416, 1974.
Yeo, A.: The acquisition of conditioned emotional response as a function of intertrial interval. Q. J.
Exp. Psychol., 28,449-458, 1976.
Young, P. T.: Emotion in Man and Animal. New York, Wiley, 1943.
Young, P. T.: Motivation and emotion. In, B. B. Wolman (ed.): Handbook of General Psychology.
Englewood Cliffs, N.J.: Prentice-Hall, 1973.
Zeiner, A. R., Nathan, M. A., and Smith, O. A., Jr.: Conditioned emotional responding (CER)
mediated by interference. Physiol. Behav., 4,645-648, 1%9.
Zelhard, P. F.: Conditioned reinforcement based on shock termination. J. Gen. Psychol., 86, 131-
139, 1972.
Zielinski, K.: The direction of change versus the absolute level of noise intensity as a cue in the
CER situation. Acta Bioi. Exp., 25, 337-357, 1965.
Chapter 2

Empirical Analysis of Concurrent

Classical-Operant Schedules
Wendon W. Henton

The following series of experiments is loosely based upon the competing re-
sponse interpretation of conditioned suppression by Brady and Hunt (1955).
This type of analysis also seems to account for the effects of conditioned stim-
uli paired with food as well as shock, and the interpretation has been slightly
extended to a more general analysis of concurrent schedules controlling con-
current responses (Henton and Brady, 1970). The present studies were con-
ducted between 1970 and 1973 as a developing experimental analysis of the re-
sponse patterns within all classical-operant schedule combinations. The
experiments were consequently designed to purposely manipulate and directly
record the response interactions generated by the various possible classical-
operant combinations.

Interactions Between Sidman Avoidance and Appetitive

Classically Conditioned Responses (Experiment 1)*
Our behavioral analysis was actually initiated at the Walter Reed laboratories
in 1968 with a study on the effects of eS-food pairings upon shock avoidance
behavior (described by Brady, 1971). The results simply suggested that avoid-
ance rates of monkeys were highly dependent upon response contingencies
buried within the superimposed appetitive conditioning procedure. The lever-
pressing response was accelerated by a visual es, in spite of an explicit contin-
gency preventing DeS delivery within 5 sec of a previous operant avoidance

* Experiments conducted at the University of Copenhagen, from September, 1970 to February

1971, with portions previously reported by W. Henton (1972).
98 Section I: Concurrent Classical and Operant Conditioning

response. Suspiciously, lever pressing also increased when the classical condi-
tioning was presented alone, without any shock avoidance contingency. Addi-
tional manipulations demonstrated that the acceleration was primarily main-
tained by spurious reinforcement of lever pressing during rather than preceding
the 3-sec ues pellet delivery cycle. Another contributing factor was a substan-
tial decrease in shock avoidance rates during the intertrial interval, with bursts
of responses alternating with 3- to IS-sec periods of pawing at the unilluminated
es or orienting toward and licking the ues dispenser.
The observed behavioral interactions were then rather similar to the adventi-
tious chaining of two responses described by Sidman (1958) when defensive
classical conditioning was superimposed upon a similar shock avoidance base-
line. Following Sidman's lead, schedule manipulations and a physical separa-
tion of the responses eliminated the response alternation during the intertrial
interval, and also produced a suppression of avoidance rates and acceleration
of orienting and approach responses during the es. Figure 2.1 shows the typi-
cal performance of two monkeys with the latter procedures.
At that time, however, the effects of positive classical conditioning proce-
dures upon negatively reinforced operants had received only limited experi-
mental attention and were not entirely understood. The latency of a shuttle box
avoidance response was reported to increase during stimuli formerly paired
with either response-independent or response-contingent food (Grossen et al.,
1969). Using a transfer of training design, Bull and Overmeir (1969) also found a
decreased operant rate when a stimulus associated with shock avoidance was
combined with a second stimulus formerly paired with food. Yet, our prelimi-
nary data seemed to indicate that maintained shock avoidance rates would be

55 57

.\1)
a:::
o
o
~ L -_ _ _. - . I

10 Min
Figure 2.1. Cumulative records demonstrating the suppression of avoidance responses by su-
perimposed stimulus-food pairings for two monkeys. Each presentation of the 30-sec CS is indi-
cated by downward displacement of recording pen. UCS was delivery of five response-independent
food pellets. Avoidance responding maintained by response-shock = shock-shock = 50 sec
schedule. .
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 99

either accelerated or suppressed by a prefood stimulus, dependent upon spe-

cific experimental arrangements.
The next experiment was simply a replication of our previous preliminary
study and sought to further examine the control of avoidance rates by manipu-
lating the characteristics of the superimposed eS-DeS pairing procedure.
Within each procedural phase, a stimulus terminated by two food pellets was
superimposed upon a free operant shock avoidance baseline. In Phase A, the
pellets remained indefinitely available in a food cup until retrieved by a simple
movement of the subject's right hand. In Phase B, the DeS retrieval sequence
was extended to a slightly longer response chain by locating the es and DeS
immediately behind the subject's head. Furthermore, the pellets were only
available during a limited interval, comparable to the duration of DeSs such as
electric shock, grain reinforcement, intracranial stimulation, etc. In Phase e,
the DeS pellets were again available during a limited interval, and the chain of
pellet retrieval responses was further extended by delivering the pellets at a dis-
tance from the shock avoidance lever.

Experiment I
Figure 2.2 presents the general features of the apparatus used in each phase of
the experiment. In initial acquisition, the subjects were placed in a standard
primate restraint chair, and DeS food pellets were delivered into a cup
mounted on the right-hand side of the restraint chair. A 3-W red light, which
would later be used as the es, was mounted on a response panel in front of the
subject's head, and a Plexiglas response lever was centered in front of the sub-
ject's waist. A 7.0-ma, 1500-V, 0.5-sec shock was delivered through wire elec-
trodes attached around the waist and to a brass footplate.
All subjects had a previous shock avoidance history, and were retrained on a
response-shock = shock-shock = 30 sec schedule for 35 additional sessions;
that is, shocks occurred every 30 sec in the absence of responses, and a lever

Figure 2.2 . Apparatus used to study the effects of CS and DCS location upon lever-press avoid-
ance responding by monkeys. H, DCS food hopper; S, blinking red light CS; L, lever for operant
avoidance responses; SH, shield to prevent lever pressing with left hand (reacquisition I only).
Left, acquisition; middle, reacquisition 1; right, reacquisition 2.
 100 Section I: Concurrent Classical and Operant Conditioning

S3
1.0

0.0

-1.0 L..._ _--lL...._ _....._ - _....._ - -..

S2
1.0

0
.;:;
~
0.0

"0
'B
'"
:;::
.£
III"
::iE
-1.0

Sl
2.0

1.0

0.0

-1.0 ....- - -.....- - -......

5 10
----.&...---""
15 20

Sessions

Figure 2.3. Mean inflection ratio for avoidance responses during stimulus-food pairings for each
acquisition session and subject. Each data point is the mean of five trials. The frequency of deliv-
ered shock is shown by the X's above each session. Adapted from Henton: J. Exp. Anal. Behav.,
17, 269-275, 1972.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 101

press postponed shock for 30 sec. After the avoidance rates had stabilized,
each subject was habituated to a 15-sec blinking red light that was presented 5
times per session for 10 sessions (stimulus adaptation). Finally, the red light
was terminated by delivery of two food pellets, with the food hopper activated
13 sec after CS onset and actual pellet delivery occuring 1.5 and 2.3 sec later.
(A delay procedure prevented activation of the food hopper until 0.5 sec had
elapsed since the previous avoidance response.)
Figure 2.3 presents the mean inflection ratio for avoidance rates in each of
the 20 sessions with superimposed classical conditioning. Positive values indi-
cate accelerated avoidance rates during the CS, and negative values indicate
suppressed avoidance rates. The inflection ratio for each subject increased to a
maximum during the first two to four sessions, followed by an approximate sta-
bilization at a somewhat lower value over the final acquisition sessions. During
the CS, each subject continued to emit shock avoidance responses and only
initiated the pellet-retrieval sequence following actual pellet delivery into the
food cup. One crucial consequence was that one or more avoidance responses
were frequently emitted during the 2.3-sec pellet delivery cycle and were there-
fore immediately contiguous with the actual pellet delivery. The lower baseline
avoidance rate of subject 1 was relatively more accelerated, and the higher
baseline rate of subject 3 was less accelerated by these stimulus-food pairings.
Of some interest, the variability in the computed inflection ratio was equally
determined by the variability in responding during the intertrial interval as well
as during the CS (Table 2.1). In addition, the local rate and pattern during the
intertrial interval was relatively disrupted when compared to previous avoid-
ance sessions, especially for subject 1 with the lowest baseline avoidance rate.
Monitored on a closed-circuit television system, each subject alternated bursts
of avoidance responses with various idiosyncratic behaviors, such as head and
hand movements toward the unilluminated CS during the intertrial interval. Al-
though such collateral responses are difficult to anticipate and record electroni-
cally, unprogrammed lever-holding responses were fortuitously recorded
throughout the acquisition sessions for subject 2.
Figure 2.4 presents a sample strain gauge recording of the form and ampli-
tude of lever pressing by subject 2 compared to previous avoidance baselines.
The normal avoidance pattern was a regular and cyclical stream of responses
with relatively fixed amplitude. The superimposed conditioning procedure
disrupted this cyclical responding and generated holding of the response lever
at maximum depression, which required 450 g of pressure and a 3-inch dis-
placement of the lever. This lever-holding behavior was occasionally correlated
with and perhaps maintained by activation ofthe food hopper, as in Figure 2.4.
Second, the amplitude rather than the frequency of responding was most
disrupted, with irregular response amplitude during both the intertrial interval
and the CS. In Figure 2.4, only 11 of the 20 lever presses in the pre stimulus
control period and 15 of the 22 lever presses during the CS were sufficient to
cross both the upper and lower force requirements of the response micro-
switch. The irregular disruption of lever-pressing amplitUde was associated
with collateral responses, such as observing and pawing responses toward the
102 Section I: Concurrent Classical and Operant Conditioning

Table 2.1 Avoidance response frequencies during control and stimulus periods
Session and trial Subject 1 Subject 2 Subject 3

1 10/12 18117 27/30

2 819 18/22 21121
3 17119 16/19 21124
4 10/18 13117 17/19
5 9/20 13/19 16/18

5
1 7111 15117 20/21
2 7111 18/21 23/18
3 519 15/19 24/22
4 5/13 13119 25/26
5 12/12 15/19 16/21

10
1 7111 17/22 27/21
2 7110 19/26 22/33
3 8/15 19/24 26/29
4 12/14 20/26 25/38
5 12/10 14/20 26/37

20
1 6121 11114 27/28
2 7120 9/19 22/29
3 12/16 16/16 19/30
4 8120 16/23 24130
5 7/13 13117 19/24

_ us

B - - - - - - _ w_ _ _ _ __ cs
A
Figure 2.4. Comparison of the form and amplitude oflever pressing by subject 2. A. During avoid-
ance baselines. B. During control and stimulus intervals of acquisition. R, upper and lower micro-
switch response requirements; US, delivery of UCS food pellets; CS, I5-sec blinking red light.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 103

red light. Similar head and hand movements were clearly apparent in subject 1,
but occurred only infrequently in subject 3, who had the least disrupted avoid-
ance rate.
The second phase of the study began with classical conditioning extinction,
in which the red light was merely presented without pellet delivery for 20 ses-
sions. The inflection ratio for avoidance responses decreased and stabilized at
approximately 0.00 across the first three to five extinction sessions. The appa-
ratus was then modified after the 10th session, with the food hopper and red
light mounted immediately behind the subject's head (Figure 2.2). This modifi-
cation would then require the subject to rotate its head approximately 180 de-
grees to orient toward the red light and food hopper. A sheet metal shield was
placed 2 inches to the left of the avoidance lever and prevented lever pressing
with the left hand. [The shield was added in an unsuccessful attempt to examine
physical incompatibility between lever-pressing and pellet-retrieval resonses.
If the subject retrieved the food pellets by clockwise rotation of the head and
shoulders, then pellet retrieval would be incompatible with lever pressing with
the right hand, and the shield would prevent lever pressing with the left hand.
Conversely, if the subject retrieved the food pellets by counterclockwise rota-
tion of the head and shoulders, then pellet retrieval would be compatible with
lever pressing with the right hand. All subjects, however, retrieved the pellets
by clockwise rotation of the body, and pellet retrieval was therefore incompati-
ble with avoidance responses for each subject].
Given stable shock avoidance baselines, the red light was again terminated
by response-independent pellet delivery in each of the next 20 sessions (classi-
cal conditioning reacquisition). As before, the food hopper was activated 13 sec
after CS onset, but the food pellets now did not remain constantly available.
Instead, the pellets fell from the delivery tube 0.8 and 1.5 sec later, and those
pellets not retrieved during the pellet delivery cycle fell onto the floor and were
"lost. "
Figure 2.5 presents the inflection ratio for each subject during the 20 reacqui-
sition sessions of Phase B. The inflection ratio initially increased and then stabi-
lized at a positive value (acceleration) for subjects 1 and 2, but at a negative
value (suppression) for subject 3. Changing the location and temporal charac-
teristics of the CS and UCS thus resulted in relatively lower inflection ratios
compared to the previous Phase A for all subjects. The inflection ratio was
again higher for the subject with the lowest baseline avoidance rate (subject 1)
and relatively lower, and suppressed, for the subject with the highest baseline
rate (subject 3). However, the overall suppression for subject 3 resulted from a
progressive diminution in the amplitude rather than the frequency of avoidance
responses throughout the CS-UCS interval.
Similar to the initial acquisition phase, each subject again alternated bursts
of avoidance responses with various unprogrammed behaviors during the inter-
trial interval. Figure 2.6 presents strain gauge recordings of lever holding by
subject 2, which again occurred during Phase B. Although the frequency of
lever pressing was approximately equal during the control and CS intervals, the
response amplitUde was nevertheless relatively more disrupted during the con-
104 Section I: Concurrent Classical and Operant Conditioning

1.0

0.0

-1.0

S2
1.0

0
.;:;
~
c:
0
'fl
Q)
;;:: 0.0
.=c:
8l
:2:

-1.0

S1
X X

1.0

0.0

-1.0
5 10 15 20

Sessions
Figure 2.5. Mean inflection ratio for avoidance responses during stimulus-food pairings of reac-
quisition 1. Each data point is the mean of five trials. Shock frequency is shown by the X's above
each session. Adapted from Henton: J. Exp. Anal. Behav., 17,269-275, 1972.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 105

us

CS
Figure 2.6. Strain gauge recording oflever holding by subject 2 during control and stimulus inter-
vals of reacquisition 1. R, upper and lower microswitch response requirement; US, delivery of
UCS food pellets; CS, 15-sec blinking red light.

trol period. This differential response amplitude thus resulted in more re-
sponses being counted during the es. For all subjects, the local rate of avoid-
ance responses during the intertrial interval appeared to be associated with
head and hand movements toward the unilluminated es. As in Phase A, the
variability in the inflection ratio was therefore equally determined by the avoid-
ance patterning during the control and es-ues intervals. One variable which
determined this local avoidance rate was the recent history of shock. Shock
delivered on the operant schedule immediately increased avoidance rates and
decreased the frequency of unprogrammed behaviors during both the control
and es intervals throughout the remainder of the session.
In the final Phase e, avoidance rates were first reestablished by extinguish-
ing the classical conditioning component. The red light was merely presented
but not terminated by ues food pellets in each of 20 extinction sessions. The
extinction procedure was interrupted after the 10 session, and each subject was
retrained in a 79 x 60 x 51 cm metal and Plexiglas cage (Figure 2.2). The
avoidance lever was positioned in the front left comer, 26 cm above the floor,
and the red light and food hopper were mounted on top of the cage in the back
right comer. The subjects were trained only on the Sidman avoidance schedule
for 5 sessions, and then the classical conditioning extinction procedure was re-
sumed for 10 additional sessions.
In this apparatus, stimulus-orienting and pellet-retrieval responses would
physically direct the subject away from the shock avoidance lever. Therefore,
one of two possibilities existed. Either orienting and retrieval responses would
occur during the stimulus, with a concomitant suppression of the spatially dis-
tant operant responses, or, conversely, avoidance responses would be emitted
with a resulting suppression of orienting and retrieval responses.
106 Section I: Concurrent Classical and Operant Conditioning

The results of stimulus-food pairings in this apparatus are shown in Figure

2.7. Again, the food pellets were presented for a short duration (approximately
1.3 sec), and pellets not retrieved during the delivery cycle were lost to the sub-
ject. In contrast to Phases A and B, the superimposed classical conditioning
produced a suppression of avoidance rates for all subjects. The inflection ratio
progressively changed across sessions and followed a slightly different pattern
for each subject. Avoidance rates were moderately decreased (subjects 2 and 3)
or relatively unchanged (subject 1) during sessions 1 to 9. This initial phase was
followed by a more marked suppression over the final reacquisition sessions.
The sequential changes in the inflection ratio were associated with the temporal
pattern of pellet retrieval responses adopted by each subject. Initially, the
changeover from avoidance to retrieval responses occurred only following ac-
tual pellet delivery. However, since the duration of the changeover response
sequence was longer than the availability of the food pellets, this response pat-
tern resulted in retrieval of approximately 30% of the delivered pellets. Over
the latter sessions, the changeover to pellet retrieval responses was initiated at
CS onset rather than UCS onset, and avoidance responses were consequently
suppressed through the prefood stimulus.
During the intertrial interval, bursts of avoidance responses again alternated
with idiosyncratic responses, including jumping or climbing responses (subject
1), rocking back and forth with the lever in the right foot (subject 2), and brief
orienting and postural responses (subject 3). The range in local avoidance rates
was approximately equal to the variability observed in Phases A and B (Table
2.1) and was similarly dependent upon the local history of delivered shock, as
noted in Phase B.

Discussion
The results then demonstrate that avoidance rates during the prefood stimuli
may be controlled by selecting the appropriate apparatus and schedule charac-
teristics. Throughout the study, avoidance responses were decreased when
concurrent pellet-retrieval responses were increased, either after pellet deliv-
ery (Phase A), during pellet delivery (Phase B), or during the stimulus preced-
ing pellet delivery (Phase C). In the initial acquisition, pellet-retrieval re-
sponses were at low rates during the prefood stimulus and only increased after
actual pellet delivery into the food cup. Concomitantly, avoidance responses
emitted throughout the stimulus were temporally contiguous with the subse-
quent delivery of the UCS food pellets. This unscheduled relationship would
favor adventitious operant reinforcement and maintenance of avoidance re-
sponses by the response-independent food (Skinner, 1948). (The brief 0.5-sec
delay contingency between avoidance responses and pellet delivery used in
Phase A would preserve the response-independent delivery of the UCS, but it
was less than the avoidance interresponse times of each subject and would not
therefore be expected to prevent the adventitious effect.) This ever-present re-
lationship between behavior and response-independent events has been pre-
viously described in other classical-operant combinations (Hermstein and
Morse, 1957; Sidman et aI., 1957; Gottwald, 1967).
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 107

S3
1.0

0.0

-1.0

S2
1.0

0
.;;
e
c
.£0
"
;;:::
.5
0.0
c
III
::!

-1.0

S1
1.0

0.0

-1.0

5 10 15 20

Sessions
Figure 2.7. Mean inflection ratio for avoidance responses during stimulus -food pairings in reac-
quisition 2. Each data point is the mean of five trials, and shock frequency is given by the X's above
each session. Adapted from Henton: J. Exp. Anal. Behav., 17,269-275, 1972.
108 Section I: Concurrent Classical and Operant Conditioning

The response patterning was altered, however, by changing the characteris-

tics of pellet delivery over subsequent experimental phases. In the last phase,
for example, avoidance responses contiguous with the more distant pellet de-
livery would result in a loss of pellets. Avoidance rates were therefore sup-
pressed by the changeover to pellet retrieval responses during the CS rather
than during the UCS. The location of pellet retrieval responses within the clas-
sical conditioning component might then be determined by- the temporal char-
acteristics of the UCS relative to the duration of the UCS retrieval responses.
As a consequence, the pattern of operant responses and retrieval responses
should be dependent upon the temporal probabilities of response-contingent
and response-independent events, as reported for operant rates in other classi-
cal-operant combinations (Stein et aI., 1958; Meltzer and Brahlek, 1970; Pom-
erleau, 1970). More importantly, additional studies have now reported qualita-
tive observations of concurrent response patterns in these classical-operant
schedules.
Konorski (1967), for example, reported that operant suppression during pre-
food stimuli was associated with approach responses to the UCS food tray (also
see Stein et al., 1971). In contrast, Azrin and Hake (1969) replicated the Kon-
orski experiment but added that the suppression could not be explained by "ap-
pealing" to competing responses. Hake and Powell (1970), however, seemed to
reverse the claim of Azrin and Hake, and also seemed to support the response
interaction interpretation (see also Van Dyne, 1971). [Azrin and Hake (1969)
had superimposed stimuli paired with intracranial stimulation, food, or water
upon positively reinforced operant responses and concluded that the operant
suppression could not be explained by competing responses. The Azrin and
Hake interpretation is somewhat complicated by the inclusion of freely avail-
able water and food within the conditioning chamber for different subgroups.
The authors believed that food-deprived rats would respond very little for food
reinforcers unless water was freely available, or that water-deprived rats would
respond very poorly forwater reinforcers unless food was also available (Azrin
and Hake, 1969, p. 168; see also Hake and Powell, 1970). The remarkable ab-
sence of approach and orienting responses to the CS, food trays, and freely
available water is not wholly consistent with previous data on schedule-
induced drinking (Falk, 1961) or stimulation-induced orienting and consumma-
tory responses (e.g., Pliskoff et aI., 1964; Mogenson and Stevenson, 1966;
Glickman and Schiff, 1967; Valenstein and Cox, 1970; Huston and Brozek,
1972).]
The current data thus suggest that overt responses may be directly if adven-
titiously affected by the UCS if all other responses are at a zero rate during the
superimposed CS. Conversely, conditioned suppression may be dependent
upon the schedule parameters generating an increased rate of concurrent re-
sponses during the CS-UCS interval. Operant rates during superimposed clas-
sical conditioning may thus be determined by the many parameters that control
the frequency and temporal patterning of each concurrent response as well as
the changeover between responses.
As a related issue, the analysis of classical-operant combinations has been
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 109

primarily influenced by theoretical interpretations that emphasize operant rates

only within the CS-UCS interval. Recent data would suggest, however, that
superimposed conditioning procedures may have more extensive effects, and
under some conditions may also influence response rates during the intertrial
interval. Smith (1970) noted that intertrial response rates are often disrupted by
CS-shock pairings, and that the frequency of such pausing may be controlled
by the CS and UCS parameters. The pausing and bursts of avoidance responses
in the present study would further suggest that stimulus-food pairings similarly
disrupt baseline avoidance rates. The additional observations of competing re-
sponses then at least tentatively indicate that response disruption during both
the intertrial interval and the CS are equally associated with increased frequen-
cies of other behaviors.
These behavioral patterns are particularly congruent with the incompatible
respondent analysis (Brady and Hunt, 1955) and the incompatible operant anal-
ysis (Weiskrantz, 1968) of negative classical conditioning superimposed upon
positive operant baselines. Both concurrent response analyses may be ex-
tended to account for the effects of positive or negative classical conditioning
combined with positive or negative operant reinforcement schedules. How-
ever, the distinctions between classical and operant processes are theoretical
rather than empirical (Anohkin, 1958; Kimmel, 1965; Rescorla and Solomon,
1967). A more atheoretical position would suggest that classical-operant com-
binations may be viewed as concurrent schedules in which two components are
operative during the CS (Henton and Brady, 1970).
Formally, classical conditioning procedures superimposed upon operant
schedules are temporally concurrent schedules of response-independent and
response-dependent events (Ferster and Skinner, 1957). Moreover, similar mu-
tual interactions have been described in concurrent classical-operant sched-
ules and concurrent operant-operant schedules. Disruption of response rates
in both forms of concurrent schedules are (1) observed when the responses are
conditioned in each component prior to schedule combination, (2) similarly af-
fected by unprogrammed or adventitious reinforcement, and (3) dependent
upon similar parameters (e.g., Catania, 1966; Davis, 1968; Lyon, 1968). Classi-
cal and operant combinations may therefore have many of the functional as
well as formal characteristics of concurrent schedules controlling concurrent
responses.
Such concurrent classical-operant schedules are most commonly inter-
preted in terms of interactions between underlying processes and states. Re-
cent data, however, demonstrate that the conditioned suppression of operant
responses is not clearly correlated with presumed covert or autonomic indica-
tors of underlying associative states, such as heart rate, blood pressure, or
EMG activity (e.g., Brady et aI., 1969; Brady et aI., 1970). A similar indepen-
dence between operant responding and conditioned salivation was previously
reported by Knitisch and White (1962). The experimental evidence thus sup-
ports Skinner's suggestion that overt responses are not an index or measure of
physiological responses, not to mention presumed emotional constructs or
states (Skinner, 1950). We might then conclude that classical and operant con-
110 Section I: Concurrent Classical and Operant Conditioning

ditioning procedures may be more amenable to an empirical analysis of concur-

rently scheduled events, at least until the assumed mental states are directly
measurable. In contrast to previous accounts that emphasize inferred and inter-
posed constructs, a concurrent schedules analysis might advantageously em-
phasize observed events.

Concurrent Response Rates During Positive

Classical-Positive Operant Conditioning
(Experiments nand m)*
The original behavioral analysis of Brady and Hunt (1955) suggested that re-
sponse interactions must occur when defensive classical conditioning is su-
perimposed upon appetitive operant baselines. We have reason to believe that
stimuli terminated by food, intracranial stimulation, and other positive rein-
forcers also control overt responses (Pavlov, 1927; Pliskoff et al., 1964; Kon-
orski, 1967; Brown and Jenkins, 1968; Huston and Brozek, 1972), which in tum
inductively suggests that respondents and operants also interact when appeti-
tive classical conditioning procedures are combined with operant baselines.
Davis and Kreuter (1972), for example, reported changes in shock avoidance
rates during prefood stimuli may be associated with food cup approach re-
sponses-a finding confirmed by our Experiment I. A concurrent response-
concurrent schedules analysis has also been applied to the changes in positively
reinforced operants during stimuli paired with food (Henton and Brady, 1970)
and intracranial stimulation (Van Dyne, 1971). The present experiments were
therefore designed to examine various aspects of a more general concurrent
responses interpretation.
The primary purpose was to record the response patterns maintained by a
positive classical-positive operant conditioning procedure. To investigate the
relationship between concurrent responses, the parameters in Experiments II
and III were selected to generate high and low operant rates, respectively, dur-
ing the prefood stimulus, with the suspicion that concurrent conditioned
responses would be at reciprocally low and high rates.
Second, the previous competing response analyses have emphasized that
physically incompatible responses are competing responses, by definition. The
reverse assertion-that competing responses are necessarily physically incom-
patible-would be misleading. Interactions between compatible responses
have been previously reported during classical conditioning (Pavlov, 1927) and
operant conditioning (Sidman, 1958). Unless classical-operant schedules are
assumed to have unique properties, physical incompatibility may also be suffi-
cient but not necessary for response interactions during positive classical-posi-
tive operant conditioning. The apparatus in the following studies was purposely
arranged such that retrieval of the ues food pellet was physically compatible

* Experiments conducted at the University of Copenhagen, from June to September 1971, by

W. Henton and I. Iversen.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 111

with lever pressing. Monkeys of the same size and species have been trained
with response chaining contingencies in our laboratory to simultaneously con-
tact and hold one manipulandum while repeatedly lever pressing with the other
hand. Based on this information, the distance between the operant response
lever and the ues food cup was selected to be well within the known limits of
simultaneous responding. Thus, the concurrent responses were behaviorally
compatible as well as physically compatible.
Third, operant suppression during classical-operant procedures has been
proposed to be dependent upon qualitative or quantitative differences between
the response-independent and response-dependent reinforcers (Azrin and
Hake, 1969). To examine the reinforcer value interpretation, the operant and
Pavlovian reinforcers were quantitatively and qualitatively equivalent in the
present experiments.
Three experimentally naive Mrican green monkeys were trained in the metal
and Plexiglas cage described in Experiment I and Figure 2.2. The operant re-
sponse lever was mounted in the right comer (10 cm from the right wall and
26 cm from the floor), and response-contingent reinforcers were delivered into
a food cup 11.1 cm to the left of the response lever. The ues food pellets were
delivered into a second food cup mounted on the right wall. With the subject
seated directly in front of the response lever, the ues pellets were delivered
approximately 19 cm to the right of the subject's head and within easy reach of
the right arm (approximately 27 cm). Both the response-independent and re-
sponse-contingent reinforcers were I-g banana-flavored food pellets. Contact
with the ues food cup was recorded with a voltage comparator and measured
in 200 millisec units. A 5-W flashing light (on/off phases of 100 millisec) was
used as the es and was placed 12.7 cm directly above the ues delivery tube.

Experiment II
The specific purpose of Experiment II was to analyze the concurrent response
patterns when the parameters were purposely selected to yield relatively high
and unchanged operant rates during the prefood stimulus. Specifically, lever
pressing was reinforced after an average of 100 responses by pellet delivery
into food cup 1 (VR 100 schedule). Initially, the subjects were trained on this
operant schedule for 30 sessions. A 30-sec flashing red light was then presented
five times per session and terminated by one food pellet delivered into food cup
2. The ues food pellet remained indefinitely available until retrieved by the
subject. Retrieval of the ues was recorded in all sessions and was physically
compatible with lever pressing.
Figure 2.8 presents the results of this classical-operant schedule for each
subject and acquisition session compared to the effects of stimulus only presen-
tation during two adaptations sessions. Lever-pressing rates were relatively
high and unchanged during the prefood stimulus for all subjects. The duration
of retrieval responses was simultaneously zero (sessions 1 to 6) or low (less
than 5 millisec/sec, sessions 7 to 15) for each subject. One informative excep-
tion, however, was that the lever-pressing rate of subject 2 in session 13 de-
112 Section I: Concurrent Classical and Operant Conditioning

53
2.0
1.0

0.0 .... o. • •• • •
LJ
.~ . . . . . . . . .0
I
0.0

u 52
! 1.0 2.0
~ s::
j., !!.
iii'
.~
/ !::. "ii'
...
I:
0 !::. ~
m
f!
"
.... ....../vo
iii'
'C
I:
1Il
.!!!
.,
'C
~
0.0
L-J
0 •• ••••• I
0.0

51
1.0 2.0

0.0 .... 0 ••••••••••••••• 0 0.0

A L.J B I I I
ad pt 5 10 15

Sessions
Figure 2.8. Operant response rate (filled triangles) and duration of concurrent UCS retrieval re-
sponses (filled circles) A. During stimulus adaptation. B. During sumperimposed stimulus-food
pairings. For comparison, the open triangles and open circles show the operant rate and retrieval
response duration, respectively, during control periods of sessions 1 and 15. Each data point is the
mean of five trials.

creased approximately 34% when concurrent pellet retrieval responses in-

creased to approximately 200 millisec/sec during the prefood stimuli.
Five control trials were also presented each session in which lever pressing
and pellet retrieval responses were merely recorded during two consecutive 30-
sec periods. Lever pressing was at a high rate and retrieval responses occurred
less than 5 millisec/sec for each subject during these control trials.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 113

Strip chart records for individual conditioning trials are presented in Figure
2.9. Operant responses were emitted at high rates when the retrieval response
duration was zero during both the intertrial interval and the prefood stimulus.
However, lever pressing was consistently suppressed by pellet retrieval re-
sponses during the 1 to 3 sec following ues delivery on each trial. The local
lever-pressing rate was also transiently suppressed when retrieval responses
briefly occurred during the intertrial interval or during the stimulus (Figure 2.9,
subject 1, trial 75; subject 2, trials 25 and 50). For each subject, a high rate of
one recorded response was therefore associated with a low rate of a second
recorded response. The data further indicate that reciprocal interactions occur
when the responses maintained by the classical and operant schedules are phy-
sically compatible. Moreover, the reciprocal interactions were not limited to
the two specific responses of lever pressing and retrieval of the response-inde-
pendent pellets, but also included the effects of other physically compatible re-
sponses, such as retrieval of the operant reinforcers. Lever pressing and re-
trieval responses were both at zero rates, for example, following delivery and
retrieval of the operant reinforcers on the VR 100 schedule.
The temporal patterning of positively reinforced operant responses might
therefore be related to alterations in the rate of other concurrently available re-
sponses. The results in fact imply that conditioned stimuli that do not control
competing responses have little if any disruptive effect upon baseline operant
responses. Similarly, other stimuli, such as ues pellet delivery, were disrup-
tive only to the degree and duration that concurrent responses were increased.
The results are then consistent with the suggestion that the superimposed ef-

TRIALS:
2 10 25 50 75
~1 A---___-- ~ -- ~ --
B - - -- ---~~ r ---~-

C:II I ' r dill' II 1111111 \11111 1\1 1 .... 1111 11 11 III RIll 11' 111 I
0 - -- - -

~2A -_-­

B ---~ ----r--~ ,r- ~ I---~- - -- ---.-,--

C _n il . (l1 I_ • • \ ,1, \ \ • __ II ft.lr l it! ' :
D- - -- -

B ------.... \.

C 111111111111 II I 111I111Il1_ \lll I 1111._1111 III_II II In I _._1111 I. III

0 - -----,---- -

Figure 2.9. Strip chart records showing the response patterns during individual trials of c1assical-
operant conditioning for each subject. A, es duration; B, ues retrieval responses; e, operant
lever pressing; D, delivery of operant reinforcer.
114 Section I: Concurrent Classical and Operant Conditioning

fects of Pavlovian conditioning are limited to those temporal intervals in which

concurrent responses increase and interact with the reference operant re-
sponse. This possibility was further investigated in the next experiment.

Experiment III
The basic purpose of Experiment III was to determine the response patterns
generated by schedules selected to suppress operant rates during the prefood
stimulus. The schedule manipUlations were based on previous reports that
operant suppression is an inverse function of relative es duration (Stein et al. ,
1958; Meltzer and Brahlek, 1970). However, acceleration of operant respond-
ing during even relatively brief stimuli (Azrin and Hake, 1969; Henton, 1972)
clearly indicated that manipulation of es duration alone would not precisely
control operant rates during the es. Alternatively, the experimental literature
suggests that manipUlation of both es and ues parameters might be sufficient
to specify and maintain a low operant response rate during the superimposed
stimulus. A concurrent response analysis suggests that any such parametric
manipulations will also increase the rate of competing responses during the pre-
food stimulus. To eliminate lever pressing during the es, the duration of the
red light was reduced to 15 sec and the response-independent pellets were
available only during the 200-millisec pellet delivery cycle.
Second, if the concurrent response analysis is correct, then the sporadic oc-
currence of retrieval responses during the intertrial interval should also be con-
trollable by manipulating the competing rate of lever pressing. We could, for
example, increase retrieval responses during the intertrial interval by decreas-
ing the rate of concurrent lever pressing. However, a low lever-pressing rate
during the intertrial interval would counter the primary purpose of demonstra-
ting a conditioned suppression of lever pressing during the es. Therefore, the
parameters were also manipulated to completely eliminate retrieval responses
from the intertrial interval by increasing concurrent lever pressing to a slightly
higher and more uniform rate. To this end, the operant schedule was changed
from the previous VR 100 to a VR 80 reinforcement schedule.
Each subject from Experiment II was consequently trained in two 50-min
sessions on each ofthe next 5 days. Session A and session B were separated by
3 hr. During session A, lever pressing was reinforced on a VR 80 schedule, and
the prefood stimulus was not presented. During session B, the operant re-
sponse lever and associated food cup were removed, and the subjects were
trained to retrieve the ues pellets directly from a delivery tube. The flashing
red light was presented for 15 sec and terminated by the response-independent
delivery of one food pellet from feeder 2. The pellets fell directly from the deliv-
ery tube approximately 200 millisec after feeder operation, and those pellets
not retrieved were lost to the subject. Five stimulus trials and five control trials
were given each session in an irregular sequence.
The operant conditioning from session A and the classical conditioning from
session B were then combined in each of the next 15 sessions. Figure 2.10 pre-
sents the overall response patterns with this schedule combination for each
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 115

S3
1.0 2.0

0.0 0.0
,0

u
"
~ S2 2.0
~ 1.0
~ s:
"
.;:
~ .rc.'"
::J

"0
''-
m-
~ ~
::J ~

"C l;l
" ~
~'" "
:!: 0.0
0.0

Sl
1.0 2.0

0.0 ...... At
,
... ...... 0 0.0

5 10 15

A B
Sessions
Figure 2.10. Operant response rate (filled triangles) and retrieval response duration (filled circles)
A. During simple classical conditioning. B. During concurrent classical-operant conditioning.
Open triangles and open circles give the operant response rate and retrieval response duration,
respectively, during control periods of session 1 and session 15. Each data point is the mean of live
trials.

subject and session. Strip chart records of lever pressing and retrieval re-
sponses for individual trials are presented in Figure 2.11. During the prelimi-
nary training with stimulus-food pairings in separate off-the-baseline sessions,
retrieval response durations progressively increased to 700 or 800 millisec/sec
for all subjects (Figure 2.10A). Each subject oriented the head and body toward
116 Section I: Concurrent Classical and Operant Conditioning

C.C.-25 Acq. 1 2 10 25 50 75
81 A~ ~ ~ ~ ~ ~ ~

B -...su- --:,r- --.r...r- ~ ~- -W~

C 111111 IlUi 1111 1.ln III .lln Imll! lliii I I

0

C.C.-25 Acq.2 10 30 43 45 75

.._-
52 A~ ~ ~ ~ ~ -~ ~

••• _n
B~ -ww- -.a.r -'----- ~

C a- lB 11l1li ID I_II III

0 ---
C.C.-25 Acq.2 5 17 20 50

-
75
53

.. - • -
A~ ~ ~ ~ ~ -~ ~~,-

B --.........r- ----,.- -----....r- ---u...r- ~

C
0
•• la _lllIla IW
• I

Figure 2.11. Strip chart records of individual trials of simple classical conditioning (C.C.-2S) and
concurrent classical-operant conditioning (Acq. 1,2, etc.). A, CS duration; B, UCS retrieval re-
sponses; C, operant lever pressing; D, delivery of operant reinforcer.

the stimulus during the first 1 to 3 sec of each trial, followed by a changeover to
retrieval responses throughout the remaining 10 to 12 sec of the stimulus (Fig-
ure 2.11). Following ues delivery, the pellets were transferred to the mouth,
either by a simple movement of the right hand or by covering the end of the
delivery tube with the mouth. Subsequent combination of the operant and clas-
sical conditioning procedures produced a transient suppression of both lever
pressing and recorded retrieval responses during the first one to three trials. All
subjects, monitored with the television system, oriented the head and eyes to-
ward the red light and feeder 2 throughout the stimulus but did not physically
contact the pellet delivery tube during these initial trials. Over subsequent
trials, the pattern oflever pressing and retrieval responses was slightly different
for each subject.
Subject 1 made a retrieval response during the last 3 sec of the second acqui-
sition trial (Figure 2.11), and retrieval responses increased to approximately 700
millisec/sec by the 10th trial (Figure 2.10). Lever pressing was therefore sup-
pressed throughout the remaining acquisition trials.
For subject 2, retrieval responses similarly increased across the first 10 to 15
trials, then decreased to zero until trial 43 (Figure 2.11), followed by an immedi-
ate increase to a high and stable rate after trial 45. Reciprocally, the lever-
pressing rate of subject 2 was initially low, progressively increased to baseline
rates until trial 43, and then decreased to near zero over subsequent trials.
For subject 3, retrieval responses were suppressed until trial 17, then sys-
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 117

tematically increased to approximately 800 millisec/sec across trials 17 to 20,

and remained high over subsequent trials. Lever pressing concurrently in-
creased to baseline rates until trial 17, then decreased to near zero during the
remaining trials of the experiment.
During the intertrial interval and control trials, retrieval response duration
was zero and lever-pressing rates remained high and unchanged for all subjects.
However, Figure 2.11 shows that lever pressing and ues retrieval responses
were both locally suppressed during and immediately following delivery of the
operant reinforcers. Again, observation indicated that each subject oriented to-
ward food cup 1 and physically retrieved the food pellets following the VR 80
reinforcement.
The recorded behavioral patterns then indicate that suppressed operant
rates may be associated with a high rate of concurrent responses during a pre-
food stimulus. The response transitions of subjects 2 and 3 especially demon-
strate that even phasic changes in lever-pressing rates are associated with re-
ciprocal alterations in concurrent retrieval responses. The alterations in
operant responding in the present study were not dependent upon any qualita-
tive or quantitative difference between the response-independent and re-
sponse-dependent reinforcers; rather, the reciprocal rate patterns may be de-
pendent upon quantitative and qualitative differences between concurrent
responses, not concurrent reinforcers. The operant suppression may thus be de-
pendent upon an increase in the rate of topographically different responses
rather than a strict behavioral or physical incompatibility.
In contrast to Experiment II, the selected parameters in the present study
suppressed the intermittent pellet retrieval rate to zero during the intertrial in-
terval. In consequence, the results suggest that response patterns during the
intertrial interval also may be controlled by competing response manipulations.

Discussion
Experiments II and III therefore suggest that the rate of positively reinforced
operants may be controlled by specifying the temporal location of competing re-
sponses during superimposed appetitive classical conditioning. Increased pellet
retrieval responses produced a decrease in concurrent lever pressing following
ues pellet delivery in Experiment II, but during the prefood stimulus in Ex-
periment III. In both experiments, the changeover from exclusive lever press-
ing to exclusive retrieval responses could not be attributed to reinforcer differ-
ences or response incompatibility, nor were the reciprocal interactions
restricted to a simple correlation between the two specifically manipulated and
recorded responses. The findings thus suggest that any parameter that alters
the local rate of one response may very well have more elaborate effects, and
may also modify the concurrent rates of a variety of topographically different
responses.
Sidman (1%0) has previously suggested that typically unrecorded behaviors
may effectively alter the frequency and temporal patterning of the experi-
menter-selected response. The present data would also argue that recording
118 Section I: Concurrent Classical and Operant Conditioning

only one response may not provide a sufficient description of classical-operant

interactions. One typically unrecorded interaction, for example, is the effect of
the operant schedule upon the classical conditioning procedure. Most contem-
porary theories assume the operant responses passively measure but do not ac-
tively determine the superimposed Pavlovian processes. Classical-operant
analyses therefore usually emphasize only the modulation of operant rates by
classical conditioning procedures. The present results instead suggest that the
modulation effects are reciprocal: responses maintained by the classical condi-
tioning procedure may be altered and disrupted by the operant conditioning
schedule. The initial disruption ofUCS retrieval responses of all subjects in Ex-
periment III, and the subsequent response transitions of subjects 2 and 3, dem-
onstrate that responses controlled by the classical conditioning component may
be neither constant nor unaffected by classical-operant combination. Similar
modulation effects occur in concurrent classical-classical conditioning (re-
viewed in Part III) and in concurrent operant-operant conditioning (see Part
II). The combined literature suggests that mutual modulation effects occur in
each of the four general types of concurrent schedules involving an operant or
Pavlovian procedure combined with a second operant or Pavlovian procedure.
Interdependent response patterns therefore seem to be ordered and empirical
characteristics of concurrent schedules in general, rather than a haphazard col-
lection of passive or unobstrusive devices of inferrential measurement.

Concurrent Response Rates During Preevent Stimuli

(Experiments IV, V, VI, and VII)*
Classical-operant schedules have attracted numerous theoretical perspectives
and treatments since their inception some 30 years ago (Estes and Skinner,
1941). The observed effects can and have been attributed to general emotional
states (Azrin and Hake, 1969), emotional-motivational interactions (Rescorla
and Solomon, 1967), reinforcer value (Hake and Powell, 1970), motivational
states (Estes, 1969), incompatible responses (Brady and Hunt, 1955), multiple
schedule effects (Hake and Powell, 1970), and concurrent schedule effects
(Henton and Brady, 1970), among others. The mUltiple schedule and the con-
current schedules analyses similarly identify observable events as the variables
that directly control behaviors during classical-operant combinations, and
therefore both interpretations are amenable to direct examination.
The mUltiple schedule interpretation, a derivative of the general emotional
state account, suggests that a superimposed UCS may be equivalent to a
change in the ongoing operant reinforcement rate. Superimposed CS-UCS
pairings might then have some of the same effects as a stimulus signaling a
change in the rate of operant reinforcement (Hake and Powell, 1970). The con-
current schedules analysis, a derivative of the incompatible response interpre-

* Experiments conducted at the University of Copenhagen, from February to December 1971,

by W. Henton.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 119

tation, proposes that a superimposed CS-UCS procedure is a concurrent

schedule that controls different responses and would therefore be more analo-
gous to a discriminative stimulus controlling a competing operant response. If
the multiple schedule or concurrent schedules analogies are correct, then clas-
sical-operant effects should be duplicated either by stimuli paired with sequen-
tial transitions in the baseline operant schedule, or stimuli paired with changes
in a second, concurrent operant schedule. Given the context of previous re-
ports on relative CS duration in the classical-operant literature (Stein et al.,
1958; Meltzer and Brahlek, 1970; Pomerleau, 1970), we might then expect that
manipulating stimulus duration would yield similar behavioral functions with
stimuli paired with mUltiple or concurrent schedule changes. The purpose of
the following experiments was simply to compare operant response rates dur-
ing different duration stimuli paried with various response-independent events.
The parameters of the response-independent changes were systematically al-
tered across experiments to progressively approximate the characteristics of
superimposed stimuli paired with food or shock.

Experiment IV
Experiment IV began with the multiple schedule interpretation, and the sugges-
tion that the value of the operant reinforcer will increase or decrease, respec-
tively, during stimuli signaling a subsequent decrease or increase in the proba-
bility of operant reinforcement. This account further suggests that operant
response rates will correspondingly increase or decrease during the stimulus to
match the calculated value of the reinforcer (Azrin and Hake, 1959; Hake and
Powell, 1970). Increased operant rates during stimuli terminated by time out
(Leitenberg, 1966) and decreased rates during stimuli terminated by a more
dense reinforcement schedule (Pliskoff, 1961, 1963) support the multiple sched-
ule interpretation. The purpose of Experiment IV was to determine the func-
tion relating response rate to the duration of stimuli terminated by time out
from positive reinforcement.
Initially, three experimentally naive monkeys (Cercopithicus aethiops) were
trained in the previously described metal and Plexiglas cage. For this study,
two transparent Plexiglas levers were mounted on the front wall 26 cm apart,
and a panel containing a food cup and red, green, and white lights was centered
between lever 1 and lever 2. Each lever could be trans illuminated by an inter-
nally mounted white light, and responses on the illuminated lever 1 (right lever)
were reinforced with food pellets on a VR 80 schedule (range 1 to 371). Lever 2
was used in the succeeding experiments but had no programmed consequences
in this experiment. When lever 1 response rates had stabilized (30 sessions), the
light within the lever was turned off for 3-min periods at irregular intervals
throughout each session. The VR 80 schedule was then in effect for a total of 36
min, and the 3-min time out (TO) periods were scheduled six times per session
(i.e., mult VR 80 TO). Training with this mUltiple schedule continued until re-
sponses occurred at a high and uniform rate during the VR 80 component, and
at a stable virtually zero rate during the TO component (sessions 31 to 76).
 120 Section I: Concurrent Classical and Operant Conditioning

After five stimulus habituation sessions (sessions 77 to 81), the TO compo-

nent was preceded by either a 15-sec white light, a 50-sec red light, or a l00-sec
green light in each of the next 25 sessions (sessions 82 to 106). The stimuli were
each presented twice per session in an irregular order, with response rates dur-
ing each stimulus compared to a control interval of comparable duration pre-
ceding stimulus onset. This procedure would thus yield a function relating

••
1.5

•• •• •••
•• • ••
. ~-------.!----------!
•
•• • •

0.0 ....- - - -...........- - - " " - - - - -.....- - -......

&l
.,.
-l!!
VI
1.5

..
•
•• •
••
•
••
•••
--.•••
•
c
.
0
Q.

~
c
.,
co
:::r!1

0.0

1.5

••
•
••• ••• ••• •••••
• •
•
0.0 .....- -......- - - - " " - - - - -......- - -......
c 15 50 100

Stimulus duration (sec)

Figure 2.12. Relationship between operant response rate and the duration of pre-TO stimuli. For
comparison, the data plotted at C are the response rates during the control periods preceding the
IS-sec stimulus. Each data point is the mean of two trials for each of the final five sessions.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 121

operant response rates to the duration of the pre-TO stimuli within each ses-
sion.
Figure 2.12 presents the operant rates during each stimulus for each subject.
The VR 80 response rates were unchanged during the pre-TO stimuli relative to
baseline rates throughout the acquisition sessions. Strip chart records of indi-
vidual trials at each stimulus duration are presented in Figure 2.13; they simi-
larly provide little indication of altered response patterns during the pre-TO
stimuli. The negligible effects of stimulus duration are then in sharp contrast to
the behavioral effects of CS duration in classical-operant schedules. This find-
ing would suggest that a sequential alteration in the proportional value of rein-
forcers may not be sufficient to change response rates during pre-schedule
change stimuli. Previous results have also been unclear, with indications that
operant rates may be unchanged, increased, or decreased during stimuli ter-
minated by time out from positive reinforcement (Pliskoff, 1963; Leitenberg,
1966; Leitenberg et aI., 1968; Kaufman, 1969). Similarly, the identical altera-
tion of avoidance rates during stimuli terminated by opposing schedule changes
suggests that the effects of pre-schedule change stimuli are dependent upon
variables other than reinforcer value (Henton, 1970).
Moreover, Brownstein and Hughes (1970) and Brownstein and Newsom
(1970) noted that response rate changes in mUltiple schedules ("contrast" ef-
fects) may be partially dependent upon changes in response rate rather than
reinforcer rates during the successive schedule components. In a somewhat re-
lated analysis, Dunham (1971) and Iversen (1974) proposed that multiple sched-
ule effects may be dependent upon the concurrent rate of unrecorded responses
within a schedule component, and therefore only indirectly related to the re-
corded operant rate during the subsequent schedule component (see Chapter 4

.._,
!5SEC. 50 SEC. !OOSEC.
51 A r- ----------'-________ ~r-

.-
~-

B ----,
C ••• 10• • • W'iI • _bill'

52 A r- ----------~--------~,-

....
~

B I i

C .. , . . . . . . . .1 . . . . . . . . . . . . . . 1•

A r- - - - - - - - - L -______~,-
S3 --"L..r-

B -,
C IInl.! •• lIlIlIo.,ullllllll \III .1,"IIIIWIIIIII1 11111111111111. Imll
MULT. VR 80:1.0.
Figure 2.13. Strip chart records of individual trials using the 15-,50-, and lOO-sec pre-TO stimuli.
A, stimulus duration; B, delivery of the operant reinforcer; C, operant responses on lever 1.
122 Section I: Concurrent Classical and Operant Conditioning

for a more detailed review and analysis). The variety of rate alterations asso-
ciated with multiple schedule changes could thus result from the absence of
contingencies specifically controlling the rates of concurrent responses.
An extended competing or concurrent response analysis would in fact sug-
gest that the changeover from baseline responses to concurrently reinforced re-
sponses may be a fundamental parameter in classical-operant schedules. Such
changeover responses are a distinguishing characteristic of concurrent per-
formances, but not mUltiple schedules (Catania, 1969). The concurrent re-
sponse interpretation thus suggests that classical-operant combinations may
be more analogous to pre-schedule change stimuli superimposed upon concur-
rent schedules than to mUltiple schedules. [The distinction between mUltiple and
concurrent schedules, however, is sometimes confused in current terminology.
For example, a two-component procedure in which reinforcement for one re-
sponse sequentially alternates between positive reinforcement and time-out,
while the schedule for a second response simultaneously alternates between
time-out and positive reinforcement could be described as a multiple schedule;
e.g., mult VI VI. This description, however, ignores the TO component and the
concurrently available response. Since TO specifies the relationship between
stimuli, responses, and reinforcers, and therefore defines a reinforcement
schedule, the above procedure would be more accurately described as a con-
current schedule simultaneously controlling two responses; e.g., cone (muit VI
TO) (mult TO VI).]

Experiment V
The procedures of Experiment IV were therefore modified to more closely ap-
proximate concurrent classical-operant schedules by specifying a changeover
from the baseline response to a second response at the offset of each pre-sched-
ule change stimulus. The purpose of Experiment V was to record the rate of
each concurrent response as a function of the duration of the pre-schedule
change stimuli. Specifically, Experiment IV was repeated, with the addition
that responses on a second lever were reinforced during the 3-min TO for lever
1 responses.
As preliminary training, lever 2 was illuminated, and responses on lever 2
were reinforced on the VR 80 schedule in each of the next five sessions (ses-
sions 107 to 111). In these sessions, lever 2 rates increased and equalled the
previous lever 1 rates by the third session.
In sessions 112 to 136, lever 1 responses were again reinforced on the mult
VR 80 TO schedule, and the TO component was again preceded by the 15-sec
white light, the 50-sec red light, or the 100-sec green light. However, lever 2
was now illuminated during the 3-min TO for lever 1, and responses on lever 2
were reinforced on the VR 80 schedule. Each stimulus was thus followed by
changes in concurrent schedules, from VR 80 to TO for lever 1, and from TO to
VR 80 for lever 2 [cone (mult VR 80 TO) (mult TO VR 80)]. Responses on the
two levers were physically compatible, and also were compatible with retrieval
of the VR 80 reinforcers delivered into the food cup between the levers.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 123

baseline r's .'1---.'

..
lever 1
imposed r's x ••••• x lever 2

1.5
••
•• • •
~ • • •
• ••

0.0 xxxxx xxxxx ••••• xxxxx ••••• xxxxx

1.5 •
•••
al
~
•
!G
In
c:

•
o
Q.

...
In
Ql

c:
It!
Ql
:2

0.0 xxxxx xxxxx ••••• xxxxx • •• ··xxxxx

•
1.5

••
•••
• ••
•• •
'. '.

- -
'" xxxxx·····xxxxx

- -
0.0
xxxxx

c 15 50 100

Stimulus duration (sec)

Figure 2.14. Baseline response rates on lever 1 and concurrent rates on lever 2 as a function of the
duration of stimuli terminated by changes in two concurrent schedules. For comparison, the data
plotted at C give the response rates during the control period preceding the IS-sec stimulus. Each
data point is the mean of two trials in each of the final five acquisition sessions.
124 Section I: Concurrent Classical and Operant Conditioning

With this schedule, lever 1 response rates were decreased during the 15-sec
stimulus and remained unchanged during the 50- and l00-sec stimuli throughout
the 25 acquisition sessions. The concurrent response rates during stimulus
trials are given in Figure 2.14 for each ofthe final five sessions. Responses on
lever 2 increased slightly during the 15-sec stimulus for subject 3, but not for
subjects 1 and 2. Lever 2 responses during the 50- and 100-sec stimuli remained
unchanged at virtually zero rates for subjects 1 and 2 and at low, irregular rates
for subject 3.
Strip chart records of individual trials are given in Figure 2.15. The temporal
pattern oflever 1 responses changed during the 15-sec stimulus, but not during
the 50- and l00-sec stimuli, for all subjects. Observed on the television monitor,
each subject either oriented toward the visual stimulus or made postural adjust-
ments toward lever 2 during the 15-sec trials. Subjects 1 and 2, however, did
not physically contact and depress lever 2 during any pre-schedule change
stimulus. For subject 3, lever 2 responses increased during the 15-sec stimulus,
and also occurred at lower intermittent rates during the intertrial interval and
the 50- and 100-sec stimuli. For all subjects, responses on either lever 1 or lever
2 were associated with a zero local rate on the other available lever before, dur-
ing, and after the pre-schedule change stimuli. Similarly, responses on both

15SEC. 50 SEC. 100 SEC.

_It.....!. ....___
~1A~
8---
.I1...___..rn__. -
I I I

C 1_11\1 ••

o I_ !II

~2A~ --~_-----1r- -----------c__________ ~.__

8---
C_II 111- _ _ • • • • I •• ,_IMIIIIII
o n • 1 •••
~3 A --c....s- --------.--------~.--

B
C I'll Iii 1111 II II_ 1 ..111 .11 1_.11111,111111111 111111111
o II 11111 I I I I I I .. II I I It !\ II II Itll1l
LEVER1=MULT. VR80:T.O. LEVER 2 = MUL T. T.O.: V R 80
Figure 2.15. Strip chart records of concurrent response patterns during stimuli terminated by si-
multaneous changes in concurrent operant schedules. A, stimulus duration; B, delivery of operant
reinforcer; C, lever 1 responses reinforced on mult VR 80 TO; D, lever 2 responses reinforced on
mult TO VR 80.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 125

levers were suppressed when concurrent pellet retrieval responses were emit-
ted during the delivery of VR reinforcers (Figure 2.15).
In summary, the results of Experiment V demonstrate that operant rates
may be decreased during relatively brief stimuli terminated by simultaneous
changes in concurrent schedules. Further, the suppressive effects of the 15-sec
stimulus appeared to be associated with an increase in other recorded and unre-
corded behaviors. The baseline responses on lever 1 were then decreased inde-
pendent of whether the emitted concurrent response was pressing lever 2, ori-
enting to the pre-schedule change stimuli, or retrieving the operant reinforcer.
The stimulus-bound changes in lever 1 rates would not seem to be wholly
dependent upon reinforcer value per se, since reinforcers were delivered on the
same VR 80 schedule before, during, and after stimulus presentation. Indeed,
lever 1 rates were locally suppressed during the 15-sec stimulus, independent of
whether the burst of concurrent responses was associated with explicit operant
reinforcement, but were unchanged during the 50- and 100-sec stimuli in spite
of the subsequent shift in reinforcer value to a different response. These find-
ings support the contention that operant rates do not necessarily match the se-
quential value of reinforcers (Brownstein and Hughes, 1970; Brownstein and
Newsom, 1970) and that the same distribution of reinforcers may control a di-
versity of response patterns even in simple operant schedules (Ferster and
Skinner, 1957). The local response interactions are, however, at least qualita-
tively similar to the concurrent response patterns recorded in positive classical-
positive operant conditioning (Experiments II and III). The possibility then
remains that an increase in the rate of concurrently available responses may be
sufficient to alter baseline operant rates during pre-schedule change stimuli, as
well as during superimposed prefood or preshock stimuli.
The response-independent presentation of an operant schedule component,
however, is fundamentally different from the delivery of food or shock in a typ-
ical classical conditioning procedure. One difference is that the baseline
operant schedule is usually unchanged during a superimposed classical condi-
tioning procedure but is frequently altered following the offset of pre-schedule
change stimuli. More importantly, in the classical conditioning procedure the
stimulus is terminated by the immediate delivery of reinforcers, but in the pre-
schedule change procedure it is terminated by the intermittent delivery of rein-
forcers throughout several minutes. The pre-schedule change stimuli in Experi-
ment V, for example, were typically terminated by three pellets scattered
throughout a 3-min VR 80 component, rather than the undelayed delivery of
Pavlovian reinforcers within milliseconds. These temporal characteristics of
the response-independent event are rather powerful controlling variables in
classical-operant schedules (e.g., Kamin, 1965), as evidenced by the differen-
tial effects of superimposed delay and trace classical conditioning. If pre-sched-
ule change stimuli superimposed upon concurrent operant schedules are in fact
analogous to classical-operant combinations, then the response patterns may
also be expected to depend upon the parameters of the response-independent
event.
126 Section I: Concurrent Classical and Operant Conditioning

Experiment VI

The procedures of the previous experiment were therefore modified to dupli-

cate the temporal characteristics of prefood stimuli superimposed upon an on-
going operant reinforcement schedule. In Experiment VI, the pre-schedule
change stimuli were terminated only by a brief change in the contingencies con-
trolling a second operant response. More specifically, responses on lever 2
were maintained on the VR 80 schedule throughout each session. Each of the
three pre-schedule change stimuli was terminated by a 3-sec interval in which a
concurrent response on lever 1 was immediately reinforced with three food pel-
lets. (The reader may note that the baseline VR 80 was assigned to lever 2 in the
present study, but lever 1 in Experiments IV and V. The change was necessi-
tated by the intention of exchanging one of the levers for an additional pellet
feeder in the following experiment, and the physical characteristics of the sup-
porting system required that the added feeder replace the right-hand lever 1,
leaving lever 2 as the comparable baseline response for both Experiments VI
and VII.)
Beginning with the first session (session 137), a fading procedure (Terrace,
1963) was used to establish the changeover from lever 2 to lever 1 during a lim-
ited period following stimulus offset. In the first four sessions, each stimulus
was terminated by a to-sec period in which (1) the light in lever 2 was turned off
and lever 2 responses had no programmed effect, and (2) lever 1 was illumi-
nated and the first response was followed by three food pellets. The schedules
and associated stimuli changed back to VR 80 (lever 2) and TO (lever 1) follow-
ing the first lever 1 response, or after 10 sec. In all subsequent sessions the VR
80 for lever 2 remained in effect throughout the session and was not changed
following stimulus offset. The limited hold for lever 1 reinforcement was re-
duced to 7.5 sec (three sessions) and 5.0 sec (three sessions). Each subject
consistently changed over from lever 2 to lever 1 during this limited hold, with
the latency of the changeover approximately 2.0 to 2.5 sec.
In the next 25 sessions (sessions 147 to 171), the 15-sec white light, the 50-
sec red light, and the 100-sec green light were paired with a 3.0-sec interval in
which the first lever 1 response was reinforced with three food pellets. The be-
havioral procedure thus consisted of a simple VR 80 for lever 2 and a mult TO
CRF (limited hold 3 sec) for lever 1. With this concurrent schedules procedure,
lever 2 and lever 1 rates were both dependent upon the duration of the pre-
schedule change stimuli. The mean response rates on each lever during stimu-
lus periods are given in Figure 2.16 for each subject. For subjects 1 and 3, the
baseline rates on lever 2 were suppressed during the 15- and 50-sec stimuli, but
remained relatively high and unchanged during the 100-sec stimulus. At the
same time, lever 1 rates increased during the 15- and 50-sec stimuli, but were
relatively low and unchanged during the 100-sec stimulus. For subject 2, lever 2
rates were decreased, and lever 1 rates irregularly increased during the 15-sec
stimulus, and both responses were at unchanged rates during the 50- and 100-
sec stimuli.
Interactions in the frequency and patterning of responses during individual
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 127

baseline r's ...... --"""'4.. lever 2

imposed r's x ••••••• x lever 1

.:.
1.5

•
x
0.0 xxxx

1.5
•• • ••
c.i
••
•
Q)
~
en
Q)
en
c:
0
a.
en
...
••
Q)

c:
'"
Q)
::2
xx
0.0 xxxxx
- -
xxx ••• ·xxxxx·····xxxxx

1.5

•
•••
•
".xx·..xxx
••
•••• x

-
x xx
0.0 xxx x xx
c 15 50 100

Stimulus duration (sec)

Figure 2.16. Baseline rates on lever 2 and concurrent rates on lever 1 as a function of the duration
of stimuli terminated by reinforcement of lever 1 responses. For comparison, the data plotted at C
give the concurrent response rates during the control period preceding the IS-sec stimulus. Each
data point is the mean of two trials in each of the final five sessions.
128 Section I: Concurrent Classical and Operant Conditioning

trials are given in Figure 2.17. Subject 1 consistently emitted lever 1 responses
throughout the IS-sec stimulus, alternated bursts of responses on lever 2 and
lever 1 throughout the 50-sec stimulus, and emitted only baseline lever 2 re-
sponses during the 100-sec stimulus. Subject 2 alternated responses on lever 2
with orienting responses and occasional lever 1 responses during the IS-sec
stimulus, but changed over to lever 1 only following the offset of the 50- and
100-sec stimuli. Subject 3 alternated responses on lever 2 and lever 1 through-
out all pre-schedule change stimuli, with the frequency of each response per
changeover dependent upon stimulus duration. Bursts of responses on either
lever 2 or lever 1 were associated with a zero local rate on the other available
lever before, during, and after the pre-schedule change stimuli. Similarly, re-
sponse rates on both levers were decreased when pellet retrieval responses oc-
curred during delivery of the three food pellets following stimulus offset or dur-
ing delivery of the baseline VR 80 reinforcers (Figure 2.17). All subjects also
emitted other unrecorded behaviors, such as approach and contact with the vis-
ual stimuli.
The rate of a positively reinforced operant was therefore altered by increas-
ing the rate of other responses during the pre-schedule change stimuli. The data
suggest that the local rates and patterns of the competing responses were de-
pendent upon the duration of the signal for concurrent reinforcement. For each
subject, baseline response rates on lever 2 decreased and approximated a direct
function of stimulus duration. The concurrent lever 1 rate was increased and
approximated an inverse function of stimulus duration. This functional rela-

15SEC. SO SEC. 100 SEC.

~1 A ~- .-
B ----,
C Hlml w-
III!I!! 1I~!Ii :1! Ut I 11 1 In 1I1!!ili11 111'1 I&nE( !1121111(IIH!lII'Iii!1Lt lil~1t III
0 !~ I !I! 1\ IHI Po !lUll II

~2 A - --..r-
B ---.--
C HI ( W ......"•• _1. ..11 ItI1Il(I(IIWI1tlII !1
o ! II

S3A-~
B ---..--
C 11111 I nU ll Idhl III 1111111111 11111 II1I1I1WIIIIIII 1111 ilm mill 1l III
o I 1111111 IIllttMIr.\t1U I ( I ( I ( I (mell
I i.! I I ! II ( I
Figure 2.17. Strip chart records of concurrent response patterns during signaled concurrent rein-
forcement. A, stimulus duration; B, delivery of operant reinforcers; C, baseline lever 2 responses
maintained on VR 80; D, lever 1 responses maintained by signaled reinforcement.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 129

tionship, moreover, is quantitatively comparable to the effects of CS duration

in classical-operant schedules (e.g., Stein et aI., 1958). Stimulus duration may
then be a common parameter controlling concurrent response rates during
stimuli paired with concurrent response-contingent or response-independent
reinforcers.
One troubling aspect of these observations, however, is that previous au-
thors have reported that baseline rate changes during signaled concurrent rein-
forcement cannot be attributed to response interactions, but instead match the
proportional value of reinforcers (Catania, 1963; Rachlin and Baum, 1969). In
contrast, the baseline response rates in the present experiment did not match
the proportional rate of obtained reinforcers for any subject. Furthermore, the
rate of competing responses maintained by the pre-schedule change stimuli did
not match the relative value of the signaled reinforcers. In comparison, both
Catania (1963) and Rachlin and Baum (1969) reported low, near zero rates of
competing preparatory and observing responses, which therefore also did not
match the relatively high rate of signaled reinforcers. The absence of concur-
rent or competing responses in the previous signaled reinforcement studies is
rather puzzling and seemingly inconsistent with response-reinforcer value
matching as well as the concurrent response interpretation. The present data
are consistent, however, with the suggestion that response rates do not match
reinforcer rates except within specific values of changeover delays and other
experimental constraints (Catania, 1966), or when all controlling variables are
carefully balanced across all responses (Rachlin, 1971). Balancing procedures
were not used in the present study and are not used in classical-operant combi-
nations. The response rates would therefore not be expected to conform to the
matching theory. Indeed, the constrained matching of concurrent response
rates to reinforcer rates may be a specific example of response interactions
(Part 11).

Experiment VII
We then suspect that the response interdependencies in Experiment VI are
closely similar to the behavioral patterns generated by analogous classical-
operant schedules. If operant rates are a direct function of the duration of the
superimposed CS, then we would expect that the rates of other, competing re-
sponses must be inversely related to stimulus duration. Such an inductive spec-
ulation is consistent with previous data demonstrating overt orienting responses
toward the conditioned stimulus, as well as conditioned consummatory re-
sponses, during Pavlovian conditioning (Pavlov, 1927; Razran, 1961; Konorski,
1967; Patton and Rudy, 1967; Brown and Jenkins, 1968). The results then im-
ply that a variety of response interactions must occur when conditioned orient-
ing and consummatory responses are scheduled with operant responses in
classical-operant conditioning. Although the sequence of orienting and con-
summatory responses may involve numerous postural changes and head and
hand movements, the terminal response of physical contact with the CS and
the DCS may be directly recorded. In Experiment VII, contact with the con-
130 Section I: Concurrent Classical and Operant Conditioning

ditioned stimuli was recorded with a voltage comparator connected to unin-

sulated wire embedded in the stimulus display panel; physical contact with the
ues pellet delivery tube was recorded with a second voltage comparator.
The purpose of the experiment was then to determine the interactions be-
tween stimulus-orienting responses, preparatory ues retrieval responses, and
operant lever pressing when prefood stimuli are superimposed upon a positive
operant reinforcement schedule. Lever 1 was replaced with a pellet delivery
tube and the procedures of Experiment VI were repeated, with the exception
that the three food pellets following stimulus offset were delivered independent
of the subject's behavior.
To convert the pre-schedule change stimuli to classical conditioning stimuli,
an extinction procedure was first used in which each stimulus was presented
but not terminated by a schedule change in each of the next 15 sessions (ses-
sions 172 to 186). Baseline responses on lever 2 were reinforced on the VR 80
schedule throughout each session. Lever 2 rates during the stimuli increased to
normal baseline rates, and lever 1 responses decreased to a zero rate by the
10th extinction session. The apparatus was then modified and lever 2 was re-
moved and replaced with a ues pellet delivery tube.
Each subject was trained to retrieve ues pellets during off-the-baseline ses-
sions following each of the last five extinction sessions. In these additional ses-
sions, lever 2 and the stimulus display panel were removed from the training
cage. To match the temporal characteristics of the operant limited-hold proce-
dure of Experiment VI, a white light mounted above the pellet delivery tube
was turned on for 3.0 sec, and the ues pellet feeder was operated 2.0 sec after
light onset. The three pellets were delivered approximately 300 millisec after
feeder operation, which corresponds to the changeover latency to lever 1 in the
previous procedure. Pellets not retrieved fell through the grid floor and were
not available to the subject.
The two types of sessions were subsequently combined, with lever 2 re-
sponses reinforced on the VR 80, and the 15-, 50-, and tOO-sec stimuli were
paired with the ues pellet delivery cycle (sessions 187 to 211). The compatibil-
ity of lever 2 responses and ues retrieval responses was matched to the previ-
ous compatibility of lever 2 and lever 1 responses by positioning the pellet de-
livery tube 26 cm from lever 2. The results of this classical-operant schedule
are presented in Figure 2.18. Baseline response rates on lever 2 were dependent
upon the duration of the prefood stimulus for all subjects. For subjects 1 and 3,
lever 2 rates were suppressed during the 15- and 50-sec stimuli and slightly de-
creased during the l00-sec stimulus. Concomitantly, the rate of recorded stimu-
lus-orienting responses was most increased during the 15- and 50-sec stimuli
and least increased during the tOO-sec stimulus. For subject 2, lever 2 rates
were suppressed during the 15-sec stimulus, marginally decreased during the
50-sec stimulus, and remained unchanged during the l00-sec stimulus. Stimu-
lus-orienting responses concurrently increased during the 15-sec stimulus, mar-
ginally increased during the 50-sec stimulus, and remained relatively un-
changed during the 100-sec stimulus. For all subjects, recorded ues retrieval
responses were at a zero rate during the stimulus and only increased during the
ues pellet delivery cycle.
 baseline r's ••. - - - - . . . . lever 2

imposed r's x••••••••• x orienting r's

•
x

I
x

XX • • • • • • • • xx
•
• x
x
x

x
x xx
x .tM.---~~ x
xxx ••
- x

•• •
.., 1.0
••
•
x

x
Q)
~
x
en
Q)
en
C
0 •••••
'. •
.
0- x
en

•
~ xx '
C x • x x... x
co • • • • eX X X
Q) X
~ xxx x x

x
xx x

2.0 ••• eX.

X ••••• e.
I

••
I
I
I

•
I
x I
I

••
'. X

S'x
". X

•
x x
x

•
x
x
0.0 xx

C 15 50 - 100

Stimulus duration (sec)

Figure 2.18. Baseline rate on lever 2 and concurrent rate of conditioned orienting responses as a
function of the duration of an appetitive Pavlovian CS. For comparison, the data plotted at C give
the response rates during the control period preceding the IS-sec stimulus. The rate of orienting
responses is the frequency of contact with the CS measured in 200-millisec units divided by CS
duration. Each data point is the mean of two trials in each of the final five sessions.
131
132 Section I: Concurrent Classical and Operant Conditioning

The interactions between lever 2 responses, stimulus-orienting responses,

and ues retrieval responses are presented in Figure 2.19 for individual trials at
each stimulus duration. A changeover from lever-pressing to stimulus-orienting
responses occurred at the onset of the 15-sec stimulus, followed by a second
changeover from orienting responses to ues retrieval responses after stimulus
offset, which in turn was followed by a third changeover from ues retrieval
responses back to lever pressing. Subjects 1 and 3 emitted stimulus-orienting
responses throughout the 50-sec stimulus, while subject 2 irregularly alternated
between lever pressing and brief orienting responses. During the 100-sec stimu-
lus, each subject only intermittently changed over to stimulus-orienting re-
sponses. A high local rate of either lever-pressing, stimulus-orienting, or ues
retrieval responses was associated with a low local rate of the other two avail-
able responses. Lever-pressing and ues retrieval responses were simulta-
neously decreased during retrieval of the operant reinforcers from food cup 1
mounted on the stimulus display panel (Figure 2.19, traces D and E).
These findings are consistent with the notion that prefood stimuli superim-
posed upon operant procedures simultaneously control the rates of concurrent
responses. The alteration in the rate of each response seems to be intimately
related to the duration of the prefood stimulus. For all subjects, the operant
rate was decreased and approximated a direct function of stimulus duration,
while stimulus-orienting responses concomitantly increased and approximated

155EC. 505EC. 1005EC.

51 A ~ -----------.__________--Lr--
B ~
• ------------------------.Ir-
C III II1 I 1111 • • Mil 1111111
D
E -ur~ ....---..;.-......:.r.LL-:I

'.
52 A~
B - , .-
C
D
.!~ I
-:---,
at._ IEIA_._hl_1. t=!t:! . !:

E --;-;-, r:-- :-:--nr---r-u--

53A-~
B ----,r,- -------------------' - ..
C 111111 ,111111111 Idalllll ll :11 lid ~U1I1U i.l lUI! I.l I::!. , I t
D
E --ur- , t I I , III i"

Figure 2.19. Concurrent response patterns when classically conditioned stimuli are superimposed
upon operant baselines. A, CS duration; B, contact with ues food cup; e, baseline lever 2 re-
sponses reinforced on VR 80; D, delivery of operant reinforcers; E, contact with es display panel.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 133

an inverse function of stimulus duration. Similar to the results of Experiment

VI, the local rate of one recorded response was inversely related to the local
rate of concurrently available responses. Furthermore, the results again indi-
cate that physical incompatibility is not necessary for reciprocal interactions
between concurrent responses. The present results, in agreement with previous
data (Davis and Kreuter, 1972; Henton, 1972), then amplify the suggestion that
prefood stimuli combined with operant schedules may have more extensive ef-
fects than the simple alteration of a single response.
The multiple response interactions and changeovers during the prefood stim-
uli are quite similar to the response interactions during the comparable pre-
schedule change stimuli in Experiment VI: in both experiments, baseline
operant rates were decreased when the rate of a second response was in-
creased; the rates of both recorded responses were simultaneously decreased
when the rate of a third response was increased; and the interactions between
responses did not match reinforcer value and were not dependent upon re-
sponse incompatibility. In combination, the data suggest that the competing re-
sponse analysis might be successfully applied to a variety of concurrent rein-
forcement procedures.

Discussion
A common thread binding this series of experiments together is the fact that
alterations in the patterning of one response seem to be sufficient to change the
rate characteristics of other concurrently available responses. In hindsight, the
rate interactions are not limited to a few isolated occurrences, but are found
before, during, and after presentation of a stimulus terminated by a response-
independent event. The evidence suggests that the pattern of concurrent re-
sponses during such preevent stimuli is related to both stimulus duration and
the parameters of the response-independent event. The mutual response inter-
actions recorded in Experiments V, VI, and VII are particularly consistent with
the competing response analyses of Brady and Hunt, Weiskrantz, and others.
The competing response interpretation is also quite similar to analyses that sug-
gest that conditioned reinforcers, such as stimuli paired with food or shock,
may function as discriminative stimuli that control specific responses (review
by Kelleher and Gollub, 1962). More recent results, for example, demonstrate
that stimuli paired with intracranial stimulation, which do not "require" re-
sponses, nevertheless control the rate of preparatory and orienting behaviors
(e.g., Pliskoff et al., 1964; Huston and Brozek, 1972).
A general form of the competing response interpretation has suggested that
concurrent classical-operant schedules may be analyzed as procedures that su-
perimpose one set of responses upon a second, divergent set of responses
(Henton and Brady, 1970; Henton, 1972). The local rate of individual elements
in each response set may be intimately dependent upon the local rate of other
behaviors in the subject's repertoire. Additive response interactions ("summa-
tion of excitation") as well as competing response interactions (,'inhibition")
may then be at least partially dependent upon the qualitative and quantitative
topographical characteristics of each response element.
134 Section I: Concurrent Classical and Operant Conditioning

Unfortunately, this type of analysis clashes with contemporary emotional,

motivational, and inhibition theories, which argue that classical-operant inter-
actions are independent of peripheral response patterns. Nevertheless, the data
are consistent with suggestions that subjects are in fact responding during be-
havioral "pauses" and inhibition procedures, and that. the patterns of unre-
corded responses are determined by observable events (Sidman, 1960). The in-
teractions between stimulus-orienting responses and operant responses
observed in Experiments V and VI and recorded in Experiment VII also indi-
cate that relevant behaviors may be omitted by recording only a single re-
sponse. The results of Experiment VII, for example, demonstrate that appeti-
tive classical-appetitive operant conditioning may have multiple behavioral
effects and may not be a simple interaction between Pavlovian and operant
processes. This rich variety and complexity of local response interactions is
simply not effectively described by one or two explanatory states. Perhaps the
diversity of observed rates and patterns of responding could be assumed to
match a corresponding molecular diversity of emotions, motivations, or other
underlying processes. In the latter assumption, however, the hypothesized
state variable would be redundant and little more than the specific response
pattern from which it is inferred; rather, the interactions between lever-press-
ing responses, orienting responses, retrieval responses, changeover responses,
etc., may be more directly related to empirical parameters. This interpretation,
of course, is simply the reiteration of previous functional analyses in which re-
sponse rates are directly related to real variables and experimental procedures
(Sidman, 1960).

Interactions Between Programmed and

Superstitious Operants (Experiment VllI)*

Experiment VIII
Our next concern was the application of the response pattern analysis to the
third schedule combination of negative classical-positive operant conditioning.
Hoffman and Barrett (1971) and Stein et al. (1971) had just reported that sup-
pression of operant responses may be associated with freezing during preshock
stimuli. A concurrent response interpretation would then anticipate that the in-
teractions between freezing and operant responses must be keyed to schedule
parameters such as ues intensity and es duration. The initial purpose of this
study was to record the concurrent rates of operant and classical conditioned
responses as a function of ues shock intensity. However, a flaw in the shock
delivery system resulted in an unprogrammed or adventitious operant avoid-
ance contingency-all behaviors with the common {'roperty of moving the
shock electrodes would disrupt electrical continuity in the shock circuit and

* Experiment conducted at the University of Oxford, from October to December 1972, by W.

Henton.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 135

modify the 30-millisec shock. The equipment artifact effectively altered the
negative classical conditioning into a discrete trial operant avoidance schedule.
The resultant behaviors, although clearly artifactual, are directly relevant to
several conflicting interpretations. Weiskrantz (1968) suggested that suppres-
sion during a preshock stimulus may be due to unprogrammed competing re-
sponses that operantly modify the unconditioned stimulus. Second, negative
classical conditioning and discrete trial operant avoidance conditioning might
have common properties when superimposed upon reference baselines (e.g.,
Henton and Iversen, 1973). Alternatively, Maier et al. (1969) argued that ad-
ventitious reinforcement has received limited experimental examination and
may not be sufficient to generate superstitious behaviors in procedures using
response-independent shock. The present experiment was therefore continued
in order to examine the theoretical and empirical implications of superstitious
behaviors generated by preshock stimuli. The serendipitous purpose was to re-
cord the mutual interactions between unprogrammed shock avoidance re-
sponses and positively reinforced operant responses as a function of shock in-
tensity.
The subjects were two white Careneaux pigeons trained in a standard Cam-
den Instrument pigeon box. Pecking responses on the center key were rein-
forced with 3.0-sec access to grain on a random interval 64-sec schedule. Click
rates of 5, 15, and SO/sec were used as the preshock stimuli and were diffusely
presented throughout the experimental room containing the conditioning cham-
ber (to prevent specific orienting responses elicited by a localized CS). Each
auditory stimulus was presented for three 30-sec trials in each of five adaptation
sessions and then terminated by a 30-millisec shock of 0.1, 1.3, and 2.6 ma,
respectively. The subject's activity was observed via a closed-circuit television
system and was recorded with an ultrasonic activity recorder. Key pecking and
overt activity were recorded during each stimulus and during the immediately
preceding 30-sec interval.
Key-pecking rates during the three preshock stimuli are given in Figure 2.20
for each subject and session. The acquisition of differential key-pecking rates
during the stimuli followed a similar biphasic pattern for both subjects. Initially,
key-pecking rates were approximately equal during all stimuli and independent
of the shock intensity terminating each trial. Over subsequent sessions, key
pecking stabilized at baseline rates during the O.I-ma stimulus but was progres-
sively and differentially suppressed during the 1.3- and 2.6-ma stimuli.
Observed daily on the closed-circuit television system, each subject progres-
sively developed idiosyncratic, stereotyped behaviors during the stimuli. Onset
of either the 1.3- or 2.6-ma stimulus produced a changeover from key pecking
to pecking the shock electrodes and vigorous jumping (subject 1), or to wing
flapping and clockwise turning movements (subject 2). Both subjects continued
to key peck during the O.I-ma stimulus. The stereotyped behaviors prompted a
reexamination of the shock system following the 12th session, and we found
that the shock circuit was closed only when the shock commutator was station-
ary. Virtually any rotation or vibration of the commutator resulted in a disrup-
tion of the electrical continuity within the circuit. Secondly, shock was deliv-
136 Section I: Concurrent Classical and Operant Conditioning

0.1ma _

1.3ma ___

2.6 rna .--.

p1 p2

40

,,
•
,
I
I

I
I
20
I
I

•I,
I

.
,,,
I

. .
1 ..
I , , I
1 "

....
I

'-' "•
1/

"
I
,
I

. I '. '..-,I.·.·.
"
\' '
o
10 15 10 15

Sessions

Figure 2.20. Differential response rates during 30·sec stimuli terminated by 0.1·,1.3·, and 2.6·ma
shocks. Each data point is the mean of three trials per session.

ered through beaded-chain wing-band electrodes, which also have poor

electrical continuity when in motion. The flaws in the shock circuit thus com-
bined to define an unprogrammed class of behaviors that would attenuate the
response-independent shock. The experiment was nevertheless continued, and
the interactions between key pecking and stereotyped avoidance behaviors
were recorded during the preshock stimuli.
Figure 2.21 presents the relationship between key pecking and activity as a
function of shock intensity. Key pecking decreased and molar activity concur-
rently increased as monotonic functions of the shock intensity paired with each
stimulus. Relative to control periods, the wing flapping and electrode pecking
during the preshock stimuli generated more gross activity than key pecking at
the moderately high rate of approximately 1.3 responses/sec. As a supplement
to the gross activity data, the avoidance behaviors observed throughout the ex-
periment were manually recorded during sessions 17 and 18 (one observer).
The mean frequency of pecking the shock electrodes by subject 1 was 0.0,12.0,
and 19.5, and the mean frequency of wing flapping by subject 2 was 0.0, 8.5,
and 14.0 during the stimuli terminated by 0.1, 1.3, and 2.6 rna, respectively. At
all shock intensities for both subjects, key pecking was at a zero rate when the
unprogrammed behaviors increased, and conversely, key pecking was at lo-
cally high rates when wing flapping or electrode pecking was at a zero rate.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 137

120

pl p2

•
80

~
0
Co
~
C
co
"
::i: •
40 0
0

0.0 0.1 1.3 2.6 0.0 0.1 1.3 2.6

Shock rna

Figure 2.21. Key pecking and activity during 30-sec preshock stimuli as a function of shock inten-
sity. For comparison, the data plotted at 0.0 are the response rates during the 30-sec control period
preceding the O.I-ma stimulus. Each data point is the mean of the final 10 trials. Black circles, activ-
ity; white circles, key peck.

Discussion
The results suggest that key pecking by pigeons may be suppressed during dis-
crete stimuli that maintain a high rate of unprogrammed shock avoidance re-
sponse. The results also demonstrate that adventitious reinforcement by the
response-independent delivery of shock is not only sufficient to generate super-
stitious behaviors, but that the effects are differential and dependent upon
shock intensity. The graded suppression is quite similar to the differential ef-
fects of shock intensity in negative classical-positive operant conditioning
(Annau and Kamin, 1961; Henton and Jordan, 1970). This behavioral compari-
son suggests that operant responses may be suppressed by unprogrammed
shock avoidance responses as well as classical conditioned freezing responses
during discrete trial procedures.
One distinction between the competing respondent and competing operant
analyses of conditioned suppression (Brady and Hunt, 1955; Weiskrantz, 1%8)
is this possibility of adventitious operant reinforcement by the response-inde-
138 Section I: Concurrent Classical and Operant Conditioning

pendent ues. The competing operant and competing respondent accounts,

however, are frequently placed in theoretical opposition as mutually contradic-
tory systems. Alternatively, the effects of classical-operant and operant-
operant combinations may be quite similar and equally dependent upon mutual
interactions between the local rates of concurrent resonses (Henton, 1972;
Henton and Iversen, 1973; Iversen, 1974, 1975a, 1975c). The competing
operant and competing respondent interpretations may then be both accurate
and similar concurrent schedules analyses rather than mutually conflicting in-
terpretations.

Suppression of Classically Conditioned Responses by

Superimposed Operant Responses (Experiment IX)*
The influence of emotional theories upon conditioned suppression research has
resulted in a singular emphasis upon the alteration of operant rates during the
superimposed classical conditioning procedures. The strength and perhaps
weakness of such emotional interpretations is the proposition that underlying
classical conditioning processes are measured not by classically conditioned re-
sponses, but by the concurrent operant response (Rescorla and Soloman,
1967). Nevertheless, we have reason to believe that the behavioral effects may
be more extensive than the stimulus-bound disruption of operant responses. As
one instance, the responses controlled by the classical conditioning procedures
may be reciprocally disrupted by the superimposed operant responses (Experi-
ment III). Behaviors maintained by the classical conditioning contingencies
may then be suppressed by concurrent operant responses, in the same fashion
that baseline operant responses are altered by concurrent classical condition-
ing. In this perspective, operant rates would substantially alter rather than mea-
sure classical conditioning processes.
The current literature, unfortunately, does not provide a direct comparison
of conditioned responses in the presence and absence of concurrent operant
procedures. However, previous data indicate that activity by pigeons is in-
creased during simple negative classical conditioning (Longo et al., 1962), but
decreased when the negative classical conditioning is combined with operant
conditioning (Hoffman and Barrett, 1971; Stein et aI., 1971). Together, the two
sets of data imply that the operant procedure must suppress the behaviors elic-
ited by superimposed classical conditioning.

Experiment IX
The purpose of the present study was to further examine the suppressive ef-
fects of operantly reinforced key pecking upon classically conditioned freezing
responses. In Phase A, three different auditory stimuli were randomly pre-
sented and terminated by different shock intensities. In Phase B, key pecking

* Experiment conducted at the University of Oxford, from January to Apri11973, by W. Hen-

ton.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 139

•.. , ......•
0.10 rna 0.75ma 1.50ma

---I A············A

6
150
54

.....
X-I~!.~.
: ".~ •. .i~·.~;.l· ~.
_x-'\ ~" 1;tI"

. ···X·."
•• : '.:. 1 .A··

I'''· ..x/'.. .1 .
A.. A....

. .. ,"
: A"; • • • •: IS' •••• •.... A· •t.
A: • ~ A"'" ... A··A
'0:
l\..
'. .
•..: .. ,
•• • ~/"" x.......... , ~ I,. :.~. .•• 1',
.

I. ..:
'0,' .. --Z '., • •
~:
x •• ••
~

;;

o ~--------~--------~--------~---------'
150

OL-_ _ _...._ _ _ _...L_ _ _ _"'-_ _ _ ~

150 S6

I'_~
A"A
.'. j\X ..
x-x \x-x/ /1 X

:.. ;...,t.·1··.,":~a. .A/l:n. A.'~. :. · ne.,.

• ·A .11.. _ _
• • .... I . . A
• ••.• eo ./l ..4
• x •• • A
..~....
I
I. I,
J
.,- I,
';0..

.......
. . A.:,'
....

_ •••
I: •
• I

o
o Sessions
20

Figure 2.22. Freezing responses by each subject during 30-sec stimuli paired with response-inde-
pendent shock of 0.10, 0.75, and 1.50 rna. Each data point is the mean of three trials per session.
140 Section I: Concurrent Classical and Operant Conditioning

reinforced on an operant schedule was superimposed upon the classical condi-

tioning baseline of Phase A. The previous reports (Longo et aI., 1962; Hoffman
and Barrett, 1971) suggest that freezing responses would be an inverse function
of shock intensity in Phase A, but a direct function of shock intensity in Phase
B. Further, a response interaction analysis suggests that the local rate of con-
current key pecking would be inversely related to the freezing response rate
across all shock intensities.
The apparatus described in Experiment VIn was repaired and used in the
present study. Three experimentally naive pigeons were first habituated to the
experimental chamber (3 sessions) and then given 20 sessions of classical con-
ditioning acquisition. The 5, 15, and 50 clicks/sec stimuli were presented in a
randomized order within each session and terminated by a 30-millisec shock of
0.10,0.75, and 1.50 rna, respectively. Stimulus duration was 30 sec on all trials,
and each stimulus was presented three times per session. A freezing response
was defined as a zero frequency of activity recorded by the ultrasonic monitor
during a 200-millisec interval.
The results of this simple classical conditioning procedure are given in Fig-
ure 2.22. Freezing during each stimulus was rather variable across the first 5 to
10 sessions for each subject. A differential pattern during the conditioned stim-
uli was apparent over the final 10 sessions, with freezing at low rates during the
1.50-ma stimulus, and at progressively higher rates during the 0.75- and 0.10-
rna stimuli. Freezing was an inverse function of shock intensity for all subjects
over the final sessions. Freezing was relatively unchanged during additional
blank control trials, and was approximately comparable to the freezing re-
corded during the O.lO-ma stimulus.
The seemingly inconsistent pattern of freezing across subjects during early
acquisition (Figure 2.22) was partially determined by a sequential dependence
offreezing upon the preceding ues, and therefore at least partially determined
by the randomized sequence of es-ues trials. Figure 2.23 presents the fre-
quency of freezing during each es as a function of the shock intensity on the
preceding trial (N - 1) during the initial and final sessions. In the first five ses-
sions, freezing during each es was related to the ues intensity on the immedi-
ately preceding trial, with the amount of freezing during any stimulus a direct
function of the preceding ues intensity. The single exception is that freezing
by subject 6 during the 0.75-ma stimulus was independent of the preceding trial.
In comparison, over the final five sessions, freezing during one trial was rela-
tively independent of the ues on the preceding trial.
For a more detailed analysis, freezing was also recorded in successive 6-sec
intervals within the pre stimulus , stimulus, and poststimulus periods of each
trial. Figure 2.24 presents the within-trial analysis for trials 1, 15, and 60 with
each es. Each column compares the pattern of freezing across successive trials
with each es, and each row compares freezing across the three different stim-
uli within the same trial. The trial-by-trial data for subjects 5 and 6 were qualita-
tively comparable to the data for subject 4. In the pre stimulus period, freezing
had no systematic pattern either within or across successive trials. Within each
es, the freezing pattern was initially dependent upon the previous ues (trial
 Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 141

S4
150

.........

150 S5

&
VI
u
t:
C>
t:
'N

~"
t:

•
ill
::i!

0
- -
S6
150

~ -----
~

o trial

0.10 0.75 1.50 0.10 0.75 1.50 0.10 0.75 1.50 n-l

0.10 0.75 1.50 n

Shock rna

Figure 2.23. Freezing during each CS on trial N as a function of the DeS intensity on trial N - 1.
Data for the first five sessions (squares) and the last five sessions (circles) are presented for each
subject.
142 Section I: Concurrent Classical and Operant Conditioning

Trial
30

0
-- ----
.J-y..\.J
30

Cl
c:
'N
Q)
Q)
....
u.

0
-- ---
30

V..rf
o
pre cs post pre cs post pre cs post
0.10 0.75 1.50
Shock ma
Figure 2.24. Distribution offreezing in the 30-sec pre-CS, CS, and post-CS intervals of individual
trials by subject 4. Each data point represents the absolute frequency of freezing in successive 6-
sec intervals within trials I, 15, and 60 of each CS.

1), but only dependent upon the paired shock intensity during trial 60. In these
final trials, freezing decreased at es onset, followed by an increase to higher
rates across succeeding 6-sec periods within the es-ues interval. Both the
initial decrease and final rate of freezing was an inverse function of ues inten-
sity. During the postshock period of each trial, freezing was generally an in-
verse function ofVeS intensity in the early acquisition sessions. However, the
freezing elicited by the 1.50-ma ves gradually increased over sessions and
eventually equalled the pattern offreezing elicited by the lower 0.75-ma ves in
the final acquisition sessions. Postshock freezing had no systematic pattern fol-
lowing the O.lO-ma ves either within trials or across subjects.
In Phase B, stimulus and control trials were not presented during the next 20
sessions, and each subject was operantly trained to peck the center response
key. The reinforcement schedule was gradually changed from continuous rein-
forcement to random interval (RI) 64 sec across the first 12 sessions, with key-
pecking rates stabilizing over the next 8 sessions with the RI 64 sec schedule.
The operant reinforcement procedure was then superimposed upon the pre-
viously established defensive classical conditioning baseline for 20 additional
sessions.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 143

As shown in Figure 2.25, the superimposed operant schedule disrupted the

pattern of freezing during each CS for all subjects. Freezing was most
disrupted, and suppressed, during the O.lO-ma CS, and least disrupted during
the 1.50-ma CS throughout the 20 classical-operant sessions. For subjects 4
and 6, freezing decreased to zero during the O.lO-ma CS, when the rate of key
pecking was relatively high. Conversely, key pecking was reduced to zero by
concurrent freezing during the 0.75- and 1.50-ma stimuli. For subject 5, key
pecking suppressed freezing during the 0.10-rna CS, freezing suppressed key
pecking during the I.50-ma CS, and there was intermediate rates of both re-
sponses during the 0.75-ma CS.
Patterns of key pecking and freezing during successive 6-sec intervals within
individual trials are described in Figure 2.26 for subject 4. Again, the behavior
of subjects 5 and 6 was qualitatively comparable to the response patterns of
subject 4. During triall, a changeover from key pecking to freezing occurred at
the onset of each of the three different CSs. A differential response pattern
gradually developed over sessions, with key pecking at relatively high and un-
changed rates during the O.lO-ma CS but suppressed during the CSs paired with
the 0.75- and 1.50-ma shock. On the other hand, freezing was at a zero rate
throughout the O.lO-ma CS, but progressively increased to higher rates during

0.10 rna 0.75 rna 1.50rna

j
4 \ ..
~.. .. S4 2

~
:: :;.
.':
~
:
2 :::
~""J:
. \/ ..:. . . . \1...... 1

() XI
o ............ .

'\ ". :j:

w
U) ---~- 0
"rn-<
"'"
<!l
z
N
W
W
II:
u.
. \ . . ·/·. . V··I· .
IX
/I. 0

.•.
. .;. . .,. . ./:j:
. .....
~ '

o
-1..
' ..........._ .........' _" , _ I..
' ..........._ .....~I
1 20 1 20 1 20
SESSIONS

Figure 2.25. Mean rate of freezing (unconnected dots) and key pecking (connected dots) during
superimposed 30-sec stimuli paired with 0.10-, 0.75-, and 1.50-ma shock (columns) for each ofthree
subjects (rows). For comparison , freezing during the preceding simple classical conditioning is
given in the left portion of each panel. All data points are the means of three trials per session.
144 Section I: Concurrent Classical and Operant Conditioning

,.'.
T rial
18 !
,

f~ - - ~~ - - ~
!*,.x 'j., 1,1.
~.I.X

1~ i\ , ~

j\
,! ;!\
I

,.t ! ~:'

-
I i'x,i
~
t..I, ·'
0 ~

18

~
~~ 'V ~y' M
,
~C
, x ',x
25

0 !,I .. ,
!k
9;
a:
0 J·I· I·I·I
~
X·.-l-1·1
i
1·. . . . .
~/
,.,.,.,., i _

"~~ '--I V h: --
x
,

~!
I' i 60

o 1-1 .1.1.1 1-1·1 ·1·1 I·.-X·. -I

t
:'
..I

/
1-1·I·X,. i_
;
t-
1·';·1-1'.

p'. e, post p'. es POSI p,. e, post

0 .1 0 0.75 1.50

Shock rna

Figure 2.26. Distribution of freezing (x) and key pecking (circles) within the 30-sec pre-CS, CS,
and post-CS intervals of individual trials for subject 4. Each data point is the absolute frequency of
each response in successive 6-sec intervals within trials 1,25, and 60 for each CS. Squares repre-
sent the delivery of grain reinforcement contingent upon key pecking.

the 0.75- and 1.50-ma ess. Within the post-Ues interval, freezing occurred at
high rates during the initial trials and then systematically decreased with in-
creasing key-pecking rates over trials with each es. The local rate of key peck-
ing was also suppressed during the 3-sec presentation of the response-contin-
gent operant reinforcer (for example, during the pre stimulus periods of trial 60
with each eS). Thus, a high rate of either key pecking or freezing would predict
a low rate of the alternative response. However, a low rate of one recorded
response would not invariably predict a high rate of the second response rather
than the third response of consuming the operant reinforcer.

Discussion
The present data clearly demonstrate that the pattern of classically conditioned
freezing was markedly altered by the superimposed operant reinforcement
schedule; that is, freezing was an inverse function of shock intensity with the
simple classical conditioning procedure (Phase A), but a direct function of
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 145

shock intensity with the concurrent classical-operant procedure (Phase B).

The disruptive effects of a concurrent operant schedule were therefore quite
substantial, and in fact reversed the functional relationship between condi-
tioned freezing and ues intensity. The es effects then replicate the seemingly
paradoxical reports of decreased freezing during defensive classical condition-
ing (Longo et al., 1962) but increased freezing when the classical conditioning is
superimposed upon operant baselines (Stein et al., 1971).
The paradoxical effects of classical and concurrent classical-operant condi-
tioning might be resolved, however, by an analysis of the reciprocal relation-
ship between key pecking and freezing. Because of this relationship, freezing
was not only differentially affected during each es, but was also disrupted dur-
ing the intertrial interval. As a result, freezing during the es is compared to
different control levels in the two procedures, with opposite response patterns
of high freezing-low key-pecking rates during the intertrial interval of the clas-
sical conditioning (Phase A), but low freezing-high key-pecking rates during
the same intertrial interval with the classical-operant procedure (Phase B). As
one example, the es paired with 1.50-ma shock elicited approximately equal
freezing during the classical conditioning and the concurrent classical-operant
conditioning. However, the same rate of freezing was decreased relative to the
high control rates in Phase A, but increased relative to the low control rates in
Phase B. Similar effects are apparent in a comparison of the freezing pattern
across the different ess, with the inverse relationship between freezing and
ues intensity changed to a direct function by the reciprocal interactions with
key pecking. The seemingly contradictory effects of classical conditioning in
the presence and absence of a concurrent operant schedule may thus be due to
systematic response interactions, which in tum provides some strong evidence
for reciprocal response patterning in classical-operant schedules.
The interdependent nature of classical and operant responses would also ad-
dress the frequent attribution of invariant and unconditional properties to the
Pavlovian DeS. Previous investigators such as Kimmel (1965) have observed
that the responses elicited by a ues may be neither constant nor uncon-
strained. Kimble (1961), for example, reported changes in the magnitude of un-
conditioned responses elicited by the ues in classical eyelid conditioning, and
concluded that the unconditioned responses may be inhibited by the sequential
pairing of es and Des. Similarly, the progressive changes in postshock freez-
ing in Phase A also argue that the effects of a Des are not constant over suc-
cessive acquisition trials. The point that we wish to make here, however, is that
freezing elicited by each ues was also markedly altered by superimposing a
concurrent operant reinforcement schedule in Phase B. The unconditioned be-
havioral effects of a ues, like the conditioned effects of a es, might therefore
be equally conditional upon concurrent as well as sequential experimental
events.
Such conditional rather than unconditional ues effects would then be inti-
mately involved in concurrent classical-operant schedules. Although post-
ues effects are not frequently discussed, occasional papers have reported
transient changes in baseline operant rates following ues delivery (Smith,
146 Section I: Concurrent Classical and Operant Conditioning

1970). Key pecking rates in the present study were similarly disrupted follow-
ing ues delivery in the early sessions of Phase B (i.e., trial 1, Figure 2.26). The
transiently decreased operant rates were generated by an equally transient high
rate of freezing responses following the ues and extending into the subsequent
intertrial interval. As noted above, however, this pattern quickly reversed after
a few trials or sessions, with operant key pecking returning to baseline rates
immediately postshock with a concomitant suppression of the unconditioned
freezing elicited by the ues. We therefore have some reason to believe that
extensive response interactions occur throughout classical-operant schedules,
including interactions with unconditioned as well as conditioned Pavlovian re-
sponses.
As a final note, reciprocal interactions of this type do not seem to be re-
stricted to the concurrent responses separately maintained by each condition-
ing component. The within-trial analysis also shows similar behavioral interac-
tions occurring within the operant schedule. For all subjects, key pecking was
locally inhibited by the retrieval and consumption of the grain reinforcers deliv-
ered on the operant schedule. Identical local interactions were observed be-
tween operant lever pressing and consummatory responses in Experiments II
to VII, and also between operant and schedule-induced collateral responses
(Iversen, 1975b, 1976). Apparently, reciprocal interactions occur between the
responses generated within each conditioning component and are not limited to
concurrent responses generated by different conditioning components.
The primary and secondary response interactions in the present study might
therefore extend the generality of a competing or concurrent response analysis.
The alteration and indeed reversal of classical conditioning functions, however,
is not wholly consistent with the traditional conception of baseline response
rates as measures of potential processes underlying a superimposed schedule.
Thus, operant rates may not measure superimposed classical conditioning
processes, or, in the present study, classical conditioned responses may not
measure superimposed operant processes. Rather, the "measuring" response
may fundamentally alter the "measured" process. In a similar fashion, the pat-
tern of one response may be a determinant, not a measure, of other responses
and processes in simple and complex operant schedules (Part 11). Behavioral
conditioning procedures therefore seem to be more accurately characterized by
response interdependence rather than response independence.

Summary
The present set of experiments consistently shows that the local rate of one
recorded response will be effected by changes in the local rate of other concur-
rently available responses. The interdependent relationship seems to be true
across a variety of parameters and classical-operant schedules, ranging from
es duration to ues intensity across positive or negative classical conditioning
procedures scheduled with positive or negative operant conditioning proce-
dures. Our current working generality is summarized in Figure 2.27. Four gen-
eral types of concurrent schedules may be generated by superimposing a classi-
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 147

A
CC OP OP CC

.................................................................................................................
cs
R J1J u
ucs

oprnt
R~

reinf R
Lf

CSon CSon

B c
Figure 2.27. A. Graph of the possible combinations of positive and negative classical or operant
conditioning superimposed upon positive or negative classical or operant baselines. B. Mutual in-
teractions between physically incompatible responses when classical conditioning is superimposed
upon operant conditioning. Event pens deflect downward for the duration of each response. C.
Interactions between classical and operant conditioned responses when two responses are physi-
cally identical. es R, orienting responses to the es; DeS R, responses elicited by DeS; oprnt R,
operant or reinforced response; reinf R, responses generated by contingent reinforcer.

cal or operant conditioning schedule upon another classical or operant

schedule. Within each of the general combinations, a positive or negative con-
ditioning procedure may be superimposed upon another positive or negative
conditioning procedure, yielding four specific procedures within each general
combination. The first general concurrent schedule is the more common classi-
cal conditioning superimposed upon operant baselines (Experiments I, II, and
VII). Second, a discrete-trial operant procedure could be combined with a
more continuous operant conditioning baseline (Experiments V and VI; see
also Part II). Third, operant procedures may be superimposed upon preestab-
lished classical conditioning baselines (Experiments III and IXB). Finally, pos-
itive or negative classical conditioning could be superimposed upon another
positive or negative classical conditioning procedure (Part III).
Each of the four general combinations, or 16 specific procedures, may con-
148 Section I: Concurrent Classical and Operant Conditioning

trol the rate of at least four responses in the subject's repertoire. Classical-
operant schedules, for example, may control (1) orienting or "autoshaping" re-
sponses toward or away from a CS paired with food or shock, (2) retrieval of
the UCS food pellet or responses elicited by shock, (3) baseline operant re-
sponses, and (4) the reinforcing or contingent response. If the responses are
topographically dissimilar (Figure 2.27B), then an increase in the local rate of
one response may be sufficient to suppress the local rate of the other three re-
sponses. For instance, any procedure that increases stimulus-orienting re-
sponses during the CS would simultaneously decrease the rate of preparatory
UCS responses. Conversely, procedures that increase UCS responses may de-
crease CS-orienting responses. This is, if you will, a competing response analy-
sis within the classical conditioning component itself. Similar relationships
have also been described within the operant conditioning schedule (e.g., Dun-
ham, 1972; review in Part II, this volume). This interpretation has some obvi-
ous similarities with Premack's suggestion that response-response relation-
ships may be critical factors within conditioning schedules. However, the
Premackian analysis is primarily concerned with the reinforcing and punishing
effects of sequential responses (Schaeffer and Premack, 1961; Premack, 1965,
1971; Schaeffer, 1965), whereas the present formulation is an analysis of the
incremental and decremental effects of concurrent responses.
If two of the responses controlled by the classical and operant schedules are
identical, as the operant and CS-orienting responses in Figure 2.27C, then an
increase in the rate of one response would obviously increase the "alternate"
identical response. Lo Lordo (1971) has previously suggested that key pecking
by pigeons will be increased by the addition of autoshaped responses to the
same key during a prefood stimulus. If this type of analysis is accurate, then we
should be able to increase as well as decrease operant rates during superim-
posed classical conditioning procedures by (1) separately controlling low or
moderate rates of the same physical response with each schedule component,
and (2) maintaining a low or zero rate of other concurrent responses.
The incremental effect has been confirmed in unpublished student projects
in the Oxford laboratories, in which CS location was manipulated relative to
the operant manipulandum. Key pecking by pigeons was relatively higher dur-
ing a prefood stimulus when the auditory CS was placed immediately below the
operant manipulandum, but suppressed when the CS was presented on the
other side of the chamber. Observation indicated that orienting responses to-
ward the location of the CS directed the subject to the operant pecking key in
the former case, but led the subject away from the pecking key in the latter
case. Of some interest, a reverse interaction occurs during preshock stimuli,
with the subject moving away from rather than approaching the CS. In this
case, we have found that operant key pecking is more suppressed when the
negative CS is located immediately beneath the pecking key rather than at
some distance. The relative location of the CS also seems to playa role in the
relative location of rats during superimposed stimuli formerly paired with
shock or food. Although tentative, these results are consistent with Lo Lordo's
suggestion and would be supportive of a concurrent responses-concurrent
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 149

schedules analysis. We suspect, however, that space, like time, may be a rather
empty variable, with the more important aspect being the pattern of responses
elicited by the different spatial and temporal characteristics of the CS. The CS
location per se may be as ineffectual and irrelevant as any experimentally neu-
tral stimulus, and may only effectively alter classical-operant rates when the
different CS proximities control different overt response patterns (see also Lo
Lordo et aI., 1974).
Changes in operant responding during a superimposed CS, however, are
only relative descriptions comparing response rates in two different time sam-
ples, the intertrial interval versus the CS-UCS interval. Previous investigators
have importantly noted that operant acceleration may be only relative to a
disrupted intertrial interval response rate, not to previous operant baselines
(e.g., Meltzer and Brahlek, 1970). Conversely, operant responding during a
negative classical conditioning procedure may be suppressed relative to previ-
ous operant baselines, but unchanged relative to the equally suppressed inter-
trial interval response rate (Brady, 1953). The relative characterization of CS
response rates in classical-operant schedules can then be substantially depend-
ent upon changes in responding during the intertrial interval. In the present Ex-
periment I, for example, baseline avoidance rates gradually decreased across
classical-operant sessions as well as gradually increasing during the CS. As a
result, the recorded CS avoidance rate was slightly accelerated if compared to
previous avoidance baselines, but markedly accelerated when compared to the
disrupted avoidance pattern during the intertrial interval. An increased CS rate
is then not necessarily an excitation of operant responding by superimposed
classical conditioning, but may also involve a partial disruption of response pat-
terns during the intertrial interval control period.
A similar alternation of classically conditioned freezing was found before
and after superimposing a positive operant procedure in Experiments IXA and
B. The freezing responses were maximally decreased during the intertrial inter-
val and differentially suppressed during the CS as a function of UCS intensity.
The description of increased freezing during the CS is then only relative to the
low rates during the intertrial interval, and not an absolute facilitation of freez-
ing by superimposed operant responses. The present results as well as previous
data suggest that relatively increased CS response rates should not be categori-
cally attributed to a generalized summation or facilitation of classical and
operant processes; rather, relative and absolute acceleration could involve di-
vergent operations of disrupting intertrial interval rates or accelerating CS re-
sponse rates, respectively.
The above analysis is admittedly an oversimplification of the behaviors con-
trolled by concurrent schedules. Given the four basic responses within classi-
cal-operant schedules, there must be sequences of changeover responses-a
changeover from lever pressing to CS orienting responses, for example. Figure
2.28 presents a more detailed representation of the sequence of primary and
changeover responses controlled by classical-operant conditioning. We be-
lieve that the changeover responses are both important and basically similar to
the four primary responses. The changeover from operant responses to concur-
150 Section I: Concurrent Classical and Operant Conditioning

changeover

-- -- .. ....
csR
A ~ a-b
..
:--

-
--
: a-c

......
. a-d
ucs R
B : b-a ...
- --
: b-c

..
: b-d
oprnt R. : ...
C ~ c-a ... ~

-- ..
: c-b

rainf R .
D :d-a
~ d-b
: d-c

: CS • us:
Figure 2.28. Interactions between basic responses and changeover responses during classical-
operant conditioning. Time scale from left to right.

rent ues responses, for example, may be partially controlled by the spatial and
temporal characteristics of the response-independent event during eS-food
pairings (Experiments I, II, and III) or pre schedule change stimuli (Experi-
ments V and VI). The changeover characteristics may be especially involved in
determining the relative rates and patterning of physically incompatible re-
sponses-incompatible responses generally having longer sequences and dura-
tions of changeover responses than compatible responses. As shown in Experi-
ment I, the changeover from lever pressing to ues retrieval responses may be
dependent upon the duration of the changeover sequence relative to the latency
and duration of the ues; that is, the changeover to ues pellet retrieval oc-
curred during the ues when the changeover response was brief relative to the
indefinitely available food pellets, but occurred during the es when the change-
over sequence was relatively longer than the limited availability of the ues.
Quite simply, any effective response must precede the delivery of a brief ues.
The schedule control of unrecorded changeover responses may therefore spec-
ify the temporal location of operant suppression either during the es or during
the ues. Indeed, the continuum of changeover response duration may be at
least one of the defining dimensions of physical compatibility -incompatibility.
All this again emphasizes that the unrecorded changeover responses sum to an
important fraction of the subject's behavior and may be an interesting set of
variables in the analysis of classical-operant schedules. Unfortunately, we
have at best been recording only one-fourth of the data during classical-operant
combinations; at worst, one-sixteenth. This is not a very good record for any
empirical science.
As shown in Figure 2.27, the present analysis holds that the effects of su-
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 151

perimposed classical conditioning procedures may be duplicated by superim-

posed discrete-trial operant procedures. Previous data would suggest, for ex-
ample, that discrete-trial avoidance procedures and CS shock pairings may
have analogous effects upon baseline operant response rates. This interpreta-
tion is derived from the previous data of Desiderato and Newman (1971), the
adventitious discrete-trial avoidance procedure of Experiment VIII, and previ-
ous comparisons of superimposed discriminative punishment and negative
classical conditioning procedures (Brady and Hunt, 1955; Church, 1969; Orme-
Johnson and Yarczower, 1974). In general, the previous studies indicate an ap-
proximately similar suppression of baseline responding, with perhaps slightly
different parametric relationships, within discriminative punishment and nega-
tive classical conditioning procedures. More basically, the striking difference
between signaled response-contingent and response-independent shock seems
to be the changes in concurrent responses during the two procedures. Brady
and Hunt (1955) reported that the response-contingent shock procedure engen-
dered only a passive avoidance of the baseline response, with bursts of various
other behaviors emitted throughout the discriminative stimulus. The response-
independent shock procedure, however, had an opposite effect, with a high
rate of one specific concurrent response (freezing) and a low rate of all other
responses throughout the conditioned stimulus. The divergent response pat-
terns suggest fundamentally different mechanisms underlying the admittedly
similar operant suppression during superimposed discriminative punishment
and negative classical conditioning procedures. In the former case, the punish-
ment contingency only eliminates one response from the subject's repertoire,
with no explicit control over other concurrent responses. In the latter case, the
negative classical conditioning schedule actively increases the rate of one con-
current response, with a concomitant local suppression of all other responses
(including the recorded baseline response). To speak loosely, the baseline sup-
pression is a "cause" in the discriminative punishment procedure, but an "ef-
fect" in the negative classical conditioning procedure, of increased concurrent
response rates.
The previous results could then suggest that conditioned suppression may be
more analogous to a superimposed discrete-trial avoidance procedure, specifi-
cally controlling an increased rate of one response, than to a passive avoidance
or discriminative punishment procedure, which specifically decreases the rate
of one response. [Dinsmoor (1954) and Sidman (1960) have previously dis-
cussed the opposing functional contingencies in punishment and avoidance
schedules.] The more recent work of Desiderato and Newman (1971) does sug-
gest that a superimposed discrete-trial active avoidance procedure will gen-
erate concurrent response patterns similar to the conditioned suppression para-
digm. Related experiments by Hoffman and FleshIer (1962) also demonstrate a
suppression of operant baselines during a superimposed avoidance discrimina-
tive stimulus. Furthermore, the results of Experiment VIII suggest that dis-
crete-trial shock avoidance responses and conditioned suppression of baseline
responses are symmetrically related to UCS intensity. Thus, the preliminary
data consistently demonstrate analogous response patterning when active
152 Section 1: Concurrent Classical and Operant Conditioning

avoidance or negative classical conditioning is superimposed upon operant

baselines.
The discriminative punishment analogy, however, would seem to be espe-
cially appropriate during the acquisition of conditioned suppression in the on-
the-baseline procedure. Initially, baseline response rates are suppressed during
the CS without accompanying freezing responses; rather, the subjects move
away from the operant manipulandum with abortive approach and tentative
operant responses throughout the CS. The initial behavioral pattern is typical
of the passive avoidance responses noted in discriminative punishment proce-
dures. A consistent pattern of increased freezing throughout the CS only de-
velops after elimination of the serial responses within the operant response
chain. Our current interpretation is that conditioned suppression is a two-stage
acquisition process in the on-the-baseline procedure, with passive avoidance
effects in early trials followed by the classically conditioned acquisition of spe-
cific response patterns over subsequent trials.
Additional data suggest a similar equivalence of concurrent response pat-
terns during superimposed stimuli paired with response-independent and re-
sponse-contingent food. A comparison across the matched parameters and pro-
cedures of Experiments VI and VII, for example, would demonstrate a
qualitative and quantitative similarity of interdependent behaviors when dis-
criminative or conditioned stimuli are superimposed upon operant baselines. A
concurrent response analysis would therefore be commensurate with the be-
havioral effects of discrete operant and classical conditioning procedures
scheduled with continuous, ongoing operant baselines.
The concurrent response analysis is, of course, neither new nor currently
unique. The examination of behavioral patterns in negative classical-positive
operant schedules has been a signal feature of the competing response analysis
of Brady and co-workers (reviews by Brady, 1971, 1975). The present analysis
is at the same time somewhat more extensive yet more restricted than the origi-
nal competing response interpretation. Given the expanded data base, the cur-
rent interpretation is an extension of the formulation to include additional con-
current schedules (Henton and Brady, 1970; Henton, 1972). The approach is
more restricted, however, in the analysis of only overt response interactions
and does not include overt-covert response patterns. In addition, the interde-
pendent relationship between responses during superimposed positive classical
conditioning was previously discussed by Konorski (1967), and further concur-
rent response interpretations have more recently been reported by other inves-
tigators (Lo Lordo, 1971; Van Dyne, 1971; Miczek, 1973; Lo Lordo et aI.,
1974). Moreover, response interactions within operant conditioning were pre-
viously emphasized by Sidman (1960). Finally, some of the recording tech-
niques and procedures described in the present experiments were rapidly repli-
cated by Divas and Wikmark, in our laboratory, and subsequently by their
associate E. Hearst in another laboratory.
And yet, in spite of substantial amounts of data, many theorists routinely
"disprove" the existence of concurrent responses in classical-operant sched-
ules. Indeed, the logical derivation of virtually all mediating hypothetical states
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 153

within the conditioned suppression paradigm rests on the denial of behavioral

interactions. One logic proposes that conditioned suppression must be due to
fear, not competing behaviors, when the classical and operant procedures are
separately conditioned prior to test combinations of the two components. The
particular argument usually emphasizes the absence of any response require-
ment and the stimulus-stimulus contiguous nature of classical conditioning.
Since classical conditioning does not "require" any particular response, no re-
sponses are available to interact with baseline operants during the classical-
operant test sessions. The argument, however, would require a rather sweep-
ing denial of some 50 years of data describing orienting, approach, and consum-
matory responses during classical conditioning procedures (Pavlov, 1927;
Zener, 1937; Razran, 1961; Konorski, 1967). Most laboratory psychologists are
more than reluctant to accept such an extreme and negative theoretical posi-
tion. A related argument proposes that classical conditioning under curare
somehow eliminates behavioral interactions in subsequent classical-operant
sessions. However, curare administration hours or days before testing does not
in any way eliminate behavior in the subsequent noncurarized state. Experi-
ments using response prevention techniques during conditioning have clearly
reported the presence, not the absence, of responses following the removal of
the physical restraints (Pavlov, 1927; Zener, 1937; also see Solomon and
Turner, 1962). The fundamental issue of conditioning under curare is not
whether behavior is or is not required for classical conditioning, but whether
classical conditioning is suffiCient to generate behavior. The implication that
procedural necessity is equivalent to behavioral sufficiency would seem to be
inaccurate but is obviously amenable to empirical rather than logical demon-
stration.
A more common argument disproves response interactions by describing the
responses that are not elicited during classical-operant schedules. The conspic-
uous element is a reluctance to describe the responses that are occurring. One
paper "proved" that conditioned suppression could not be due to concurrent
responses since CS-orienting behaviors occurred at a very low rate during the
CS-UCS interval. Only later, in criticizing a different formulation, did the au-
thors mention that the subjects spent most of the CS with the head in the food
hopper. The absence of CS-orienting responses in this example is at best a
rather elliptical description of the presence of competing food cup responses.
The multitude of responses that are not occurring simply provides little or no
information about the actual behaviors that are occurring. A zero rate of one
response displayed on electronic counters does not exhaust the subject's re-
sponse repertoire, not demonstrate a "behavioral inhibition" or a "behavioral
pause," only a "recording pause". The empirical difficulty is that behavior is
continuous and always occurring. Total inhibition may well be more descrip-
tive of the observing behavior of the experimenter than the conditioned behav-
ior of the subject.
A related issue is the absence of perfect negative correlations between the
frequencies of any two responses in classical-operant conditioning. Imperfect
correlational pairs are occasionally offered as a statistical demonstration of the
154 Section I: Concurrent Classical and Operant Conditioning

independence of concurrent responses. However, as shown in Figures 2.9,

2.11,2.13,2.15,2.17, and 2.19, operant responses may be locally suppressed
by a number of different concurrent behaviors, ranging from eS-orienting re-
sponses, preparatory DeS approach responses, adventitious Des avoidance
responses, retrieval of the operant reinforcer, changeover responses, etc. A
perfect correlation between the absence of one response and the presence of a
second response would require a subject with but two responses in the be-
havioral repertoire. The expectation of a perfect "match" between one arbi-
trary pair of responses would be as improbable as expecting a response to
"match" only one variable selected from the host of known controlling vari-
ables. An analogous error would be the negative proof that shock avoidance
occurs in the absence of food reinforcement contingencies, with the proposi-
tion that positive reinforcement therefore cannot control operant responses.
The fact that behavior can be controlled by a variety of different contingencies
does not refute the fact that a specific contingency is sufficient, although not
necessary, to control response rates. In a similar manner, a negative logic that
the absence of response A does not specify concurrent behaviors is fundamen-
tally irrelevant. The suppression of one response A neither requires nor specifies
the specific presence of response B rather than response e, or D, or N. The
argument is misleading and surely does not deny that the occurrence of re-
sponse A is sufficient to decrease response B, as well as e, D, and N-suffi-
cient but not necessary, since many other competing responses will also de-
crease the rate of response B. A partial independence of "not responding"
should not be accepted as an independence "of responding." Such partial cor-
relations summed over responses and time may accurately describe the com-
plex interdependence of behaviors, but not response independence.
Our previous papers prepared over the last few years have uniformly ac-
cepted the theory based device of relabeling positive and negative classical
conditioning as prefood or preshock stimuli. We have become progressively
concerned with both the neutrality and accuracy of the relabeled terms. Tradi-
tionally, prefood and preshock stimuli are used as noncommital terms to pre-
vent even the appearance of implying two-process dichotomies between
operant and classical conditioning. The traditional theory testing of operant-
respondant distinctions, such as the skeletal-autonomic or overt-covert di-
chotomies (Kimble, 1961), has resulted in the dismissal of "suspect" responses
that do not neatly fit into the theoretical requirements. Overt motor responses
are frequently ignored or denied in classical conditioning as irrelevant to the
proper analysis of stimulus-stimulus association processes. The recent redis-
covery of overt responses in classical conditioning (pavlov, 1927, Zenor, 1937;
Patton and Rudy, 1967) has now promoted the relabeling of positive classical
conditioning as autoshaping. If overt motor responses are recorded during
stimuli paired with food, then the procedure is autoshaping. However, if saliva-
tion is recorded, then the procedure is true classical conditioning. This segrega-
tion of overt responses from classical conditioning maintains the view that Pav-
lovian conditioning is the autonomic, covert, contiguity conditioning of
associative states, uncontaminated by offending overt auto shaping responses.
The role of stimulus-reinforcer and response-reinforcer effects in autoshap-
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 155

ing (Brown and Jenkins, 1968; Boakes et aI., 1975; Schwartz and Gamzu, 1977)
currently parallels previous discussions of associative states versus overt re-
sponses in "conditioned anxiety." Rather oddly, the autoshaping argument
uniformly concedes whereas the "anxiety" argument most frequently denies
that classical conditioning controls overt behaviors. The growing concern is that
the description of overt responses during prefood and preshock stimuli are all too
easily dismissed as mere examples of autoshaping rather than proper classical
conditioning and may actually circumvent the accurate analysis of classical
conditioning effects. If so, the cost of theory maintenance would be a subser-
vient and segregated data analysis in place of the accepted view of inductive
and deductive theory subservient to data. Hopefully, the time has at least re-
cently past when theoretical explanations could profitably gainsay behavioral
patterns in favor of more convenient "behavioral pauses" during superim-
posed Pavlovian conditioning.
In conclusion, the present data seem to be fundamentally inconsistent with
the current use of a few molar state variables to explain classical and operant
schedule effects. If response rates were in fact a measure of explanatory con-
structs, then the broad expanse of response patterns and interactions would
seem to require an equal number of explanatory emotions, general emotional
states, incentives, expectancies, inhibitions, etc. The varied effects of manipu-
lating CS location, for example, could be attributed to a corresponding number
of underlying motivations. Alternatively, the changes in operant baselines
might be more directly attributed to the altered environmental space, or per-
haps still more directly to the elicited behaviors maintained by the altered en-
vironment. The purpose of the present experiments, however, was not to create
problems for other interpretations but to analyze the concurrent response
patterns within a variety of different procedures and parameters. Ultimately,
our argument is not so much that emotional-motivational theory cannot specify
the patterns and distributions of concurrent responses, but that they should.
Eventually, all interpretations, including the concurrent response analysis, will
have to describe and explain the total pattern of responses within classical-
operant conditioning.

Note:
This review of our work between 1970 and 1973 is taken from a prepared
speech first presented at the Easter Conference, Cambridge, England, in March
1973. To emphasize that the analysis is neither new nor unique, related experi-
ments and similar interpretations have now been reported by Hearst and Jen-
kins (1974), Karpicke et aI. (1977), Roberts et aI. (1977), and Schwartz and
Gamzu (1977).

References
Annau, Z., and Kamin, L. J.: The conditioned emotional response as a function of the intensity of
the US. J. Compo Physiol. Psychol., 54,428-432, 1961.
Anohkin, P. K.: The role of the orienting-exploratory reflex in the formation of the conditioned
156 Section I: Concurrent Classical and Operant Conditioning

reflex. In, L. G. Voronin, A. N. Leontiev, A. R. Luria, E. N. Sokolov, and O. S. Vinogradova

(eds.): Orienting Reflex and Exploratory Behavior. Moscow, Academy of Pedagogical Science,
1958.
Azrin, N. H., and Hake, D. F.: Positive conditioned suppression: conditioned suppression using
positive reinforcers as the unconditioned stimuli. J. Exp. Anal. Behav., 12, 167-173, 1%9.
Boakes, R. A., Halliday, M. S., and Poli, M.: Response additivity: effects of superimposed free
reinforcement on a variable-interval baseline. J. Exp. Anal. Behav., 23, 177-191, 1975.
Brady, J. V.: Does tetraethylammonium reduce fear. J. Compo Physiol. Psychol., 46, 307-310,
1953.
Brady, J. V.: Emotion revisited. J. Psychiatr. Res., 8, 363-384, 1971.
Brady, J. V.: Toward a behavioral biology of emotion. In, L. Levi (ed.): Emotions-Their Parame-
ters and Measurement. New York, Raven Press, 1975.
Brady, J. V., and Hunt, H. F.: An experimental approach to the analysis of emotional behavior. J.
Psychol., 40, 313-325, 1955.
Brady, J. V., Kelly, D. D., and Plumlee, L.: Autonomic and behavioral responses of the rhesus
monkey to emotional conditioning. Annu. N. Y. Acad. Sci., 159,959-971, 1969.
Brady, J. V., Henton, W. W., and Ehle, A.: Some effects of emotional conditioning upon auto-
nomic and electromyographic activity. Paper read at the Eastern Psychological Association,
1970. (Also described in Brady, 1971.)
Brown, P. L., and Jenkins, H. M.: Auto-shaping of the pigeon's key-peck. J. Exp. Anal. Behav.,
11, 1-8, 1968.
Brownstein, A. J., and Hughes, R. G.: The role of response suppression in behavioral contrast:
signalled reinforcement. Psychonomic Sci., 18,50-52, 1970.
Brownstein, A. J., and Newsom, C.: Behavioral contrast in multiple schedules with equal rein-
forcement rates. Psychonomic Sci., 18,25-26, 1970.
Bull, J. A., and Overmeir, J. B.: Incompatibility of appetitive and aversive conditioned motivation.
Proceedings of the 77th Annual Convention of the American Psychological Association, vol. 4
(part 1), 1969, pp. 97-98.
Catania, A. C.: Concurrent performances: reinforcement interaction and response independence.
J. Exp. Anal. Behav., 6, 253-263, 1963.
Catania, A. C.: Concurrent operants. In, W. K. Honig (ed.): Operant Behavior: Areas of Research
and Application. New York, Appleton-Century-Crofts, 1966.
Catania, A. C.: Concurrent performances: inhibition of one response by reinforcement of another.
J. Exp. Anal. Behav., 12,731-734, 1969.
Church, R. M.: Response suppression. In, B. A. Campbell and R. M. Church (eds.): Punishment
and Aversive Behavior. New York, App1eton-Century-Crofts, 1%9.
Davis, H.: Conditioned suppression. a survey of the literature. Psychonomic Monogr. [Suppl.], 2,
1968.
Davis, H., and Kreuter, C.: Conditioned suppression of an avoidance response by a stimulus
paired with food. J. Exp. Anal. Behav., 17,277-285, 1972.
Desiderato, 0., and Newman, A.: Conditioned suppression produced in rats by tones paired with
escapable or inescapable shock. J. Compo Physiol. Psychol., 77,427-431, 1971.
Dinsmoor, J. A.: Punishment: I. The avoidance hypothesis. Psychol. Rev., 61, 34-46, 1954.
Dunham, P. J.: Punishment: method and theory. Psychol. Rev., 78,58-70, 1971.
Dunham, P. J.: Some effects of punishment upon unpunished responding. J. Exp. Anal. Behav., 17,
443-450, 1972.
Estes, W. K.: Outline of a theory of punishment. In, B. A. Campbell and R. M. Church (eds.):
Punishment and Aversive Behavior. New York, Appleton-Century-Crofts, 1%9.
Estes, W. K., and Skinner, B. F.: Some quantatative properties of anxiety. J. Exp. Psychol., 29,
390-400, 1941.
Falk, J. L.: Production of polydipsia in normal rats by an intermittent food schedule. Science, 133,
195-196, 1961.
Ferster, C. B., and Skinner, B. F.: Schedules of Reinforcement, New York, Appleton-Century-
Crofts, 1957.
Glickman, S. E., and Schiff, B. B.: A biological theory of reinforcement. Psychol. Rev., 74,81-
109, 1967.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 157

Gottwald, P.: The role of punishment in the development of conditioned suppression. Physiol.
Behav., 2,283-286, 1967.
Grossen, N. E., Kostansek, D. J., and Bolles, R. C.: Effects of appetitive discriminative stimuli on
avoidance behavior. J. Exp. Psychol., 81,340-343, 1969.
Hake, D. F., and Powell, J.: Positive reinforcement and suppression from the same occurrence of
the unconditioned stimulus in a positive conditioned suppression procedure. J. Exp. Anal.
Behav., 14,247-257, 1970.
Hearst, E., and Jenkins, H. M.: Sign Tracking: The Stimulus-Reinforcer Relation and Directed
Action. Monograph of the Psychonomic Society, Austin, Texas, 1974.
Henton, W. W.: The effects of pre-schedule change stimuli upon avoidance behavior. Rep. Neuro-
physiol.Inst. (Kbh.) 34-35, 1970.
Henton, W. W.: Avoidance response rates during a pre-food stimulus in monkeys. J. Exp. Anal.
Behav., 17,269-275, 1972.
Henton, W. W., and Brady, J. V.: Operant acceleration during a pre-reward stimulus.J. Exp. Anal.
Behav., 13,205-211, 1970.
Henton, W. W., and Iversen, I. H.: Concurrent response rates during pre-event stimuli. Paper pre-
sented at the Easter Conference, Cambridge, March 1973, Abstract pp. 2-3.
Henton, W. W., and Jordan, J. J.: Differential conditioned suppression during pre-shock stimuli as
a function of shock intensity. Psychol. Record, 20,9-16, 1970.
Herrnstein, R. J., and Morse, W. H.: Some effects ofresponse independent positive reinforcement
on maintained operant behavior. J. Compo Physiol. Psychol., 50,461-467,1957.
Hoffman, H. S., and Barrett, J.: Overt activity during conditioned suppression: a search for pun-
ishment artifacts. J. Exp. Anal. Behav., 16,343-348, 1971.
Hoffman, H. S., and Fleshier, M.: The course of emotionality in the development of avoidance. J.
Exp. Psychol., 64,288-294, 1962.
Huston, J. P., and Brozek, G.: Attempt to classically condition eating and drinking elicited by hy-
pothalamic stimulation in rats. Physiol. Behav., 8,973-977, 1972.
Iversen, I. H.: Behavioral interactions in concurrent reinforcement schedules. Paper presented at
the English Experimental Analysis of Behavior Group, Bangor, 1974.
Iversen, I. H.: Concurrent responses during multiple schedules in rats. Scand. J. Psychol., 16,49-
54, 1975a.
Iversen, I. H.: Interactions between lever pressing and collateral drinking during VI with limited
hold. Psychol. Record, 25,47-50, 1975b.
Iversen, I. H.: Reciprocal response interactions in concurrent variable-interval and discrete-trial
fixed-ratio schedules. Scand. J. Psychol., 16, 280-284, 1975c.
Iversen, I. H.: Interactions between reinforced responses and collateral responses. Psychol. Rec-
ord, 26,399-413, 1976.
Kamin, L. J.: Temporal and intensity characteristics of the conditioned stimulus. In, W. F. Pro-
kasy (ed.): Classical Conditioning. New York, Appleton-Century-Crofts, 1965.
Karpicke, J., Christoph, G., Petersen, G., and Hearst, E.: Signal location and positive vs negative
conditioned suppression in the rat. J. Exp. Psychol. [Anim. Behav. Processes], 3,105-118, 1977.
Kaufman, A.: Response suppression in the CER paradigm with extinction as the aversive event.
Psychonomic Sci., 15, 15-17, 1969.
Kelleher, R. T., and Gollub, L. R.: A review of positive conditioned reinforcers. J. Exp. Anal.
Behav., 5, 543-595, 1962.
Kimble, G. A.: Hilgard and Marquis' Conditioning and Learning. New York, Appleton-Century-
Crofts, 1961.
Kimmel, H. D.: Instrumental inhibitory factors in classical conditioning. In, W. F. Prokasy (ed.):
Classical Conditioning: A Symposium. New York, Appleton-Century-Crofts, 1965.
Knitisch, W., and White, R. S.: Concurrent conditioning of bar press and salivation responses. J.
Compo Physiol. Psychol., 55, 863-968, 1962.
Konorski, J.: Integrative Activity of the Brain. Chicago, University of Chicago Press, 1967.
Leitenberg, H.: Conditioned acceleration and conditioned suppression in pigeons. J. Exp. Anal.
Behav., 9,205-212, 1966.
Leitenberg, H., Bertsch, G. J., and Coughlin, R. C.: "Timeout from positive reinforcement" as the
UCS in a CER paradigm with rats. Psychonomic Sci., 13,3-4, 1968.
158 Section I: Concurrent Classical and Operant Conditioning

Lo Lordo, V. M.: Facilitation offood-reinforced responding by a signal for response-independent

food. 1. Exp. Anal. Behav., 15,49-55, 1971.
Lo Lordo, V. M., McMillan, J. C., and Riley, A. L.: The effects offood-reinforced pecking and
treadle-pressing of auditory and visual signals for response-independent food. Learning Motiva-
tion, 5,24-41, 1974.
Longo, N., Milstein, S., and Bittennan, M.: Classical conditioning in the pigeon: effects of partial
reinforcement. 1. Compo Physiol. Psychol., 55, 983-986, 1962.
Lyon, D.O.: Conditioned suppression: operant variables and aversive control. Psychol. Record,
18,317-338, 1968.
Maier, S. F., Seligman, M. E. P., and Solomon, R. I.: Pavlovian fear conditioning and learned
helplessness: effects on escape and avoidance behavior of (a) the CS-US contingency and (b) the
independence of the US and voluntary responding. In, B. A. Campbell and R. M. Church (eds.):
Punishment and Aversive Behavior. New York, Appleton-Century-Crofts, 1969.
Meltzer, D., and Brahlek, J. A.: Conditioned suppression and conditioned enhancement with the
same positive UCS: an effect of CS duration. 1. Exp. Anal. Behav., 13,67-75, 1970.
Miczek, K.: Effects of scopolamine, amphetamine, and benzodiazepines on conditioned suppres-
sion. Pharmac., Biochem., Behav., 1,401-411, 1973.
Mogenson, G. J., and Stevenson, J. A. F.: Drinking and self-stimulation with electrical stimulation
of the lateral hypothalamus. Physiol. Behav., 1,251-254,1966.
Onne-Johnson, D. W., and Yarczower, M.: Conditioned suppression, punishment, and aversion.
1. Exp. Anal. Behav., 21,57-74, 1974.
Patton, R. L., and Rudy, J. W.: Orienting during classical conditioning: acquired vs. unconditioned
responding. Psychonomic Sci., 7,27-28, 1967.
Pavlov, I. P.: Conditioned Reflexes. London, Oxford Press, 1927.
Pliskoff, S.: Rate change effects during a pre-schedule change stimulus. 1. Exp. Anal. Behav., 4,
383-386, 1961.
Pliskoff, S.: Rate-change effects with equal potential reinforcements during the "warning" stimu-
lus. 1. Exp. Anal. Behav., 6,557-562, 1963.
Pliskoff, S. S., Hawkins, T. D., and Wright, J. E.: Some observations on the discriminative stimu-
lus hypothesis and rewarding electrical stimulation of the brain. Psychol. Record, 14, 179-184,
1964.
Pomerleau, O. F.: The effects of stimuli followed by response-independent shock on shock-avoid-
ance behavior. 1. Exp. Anal. Behav., 14, 11-21, 1970.
Premack, D.: Reinforcement theory. In, D. Levine (ed.): Nebraska Symposium on Motivation.
Lincoln, University of Nebraska Press, 1965.
Premack, D.: Catching up on common sense, or two sides of a generalization: reinforcement and
punishment. In, R. Glazer (ed.): On the Nature of Reinforcement. New York, Academic Press,
1971.
Rachlin, H.: On the tautology of the matching law. 1. Exp. Anal. Behav., 15,249-251, 1971.
Rachlin, H., and Baum, W. M.: Response rate as a function of the amount of reinforcement for a
signalled concurrent response. 1. Exp. Anal. Behav., 12, 11-16, 1969.
Razran, G.: The observable unconscious and the inferable conscious in current Soviet psychophys-
iology: Intereoceptive conditioning, semantic conditioning, and the orienting reflex. Psychol.
Rev., 68,81-147, 1961.
Rescorla, R. A., and Solomon, R. L.: Two-process learning theory: relationships between Pavlo-
vian conditioning and and instrumental learning. Psychol. Rev., 74, 151-184, 1967.
Roberts, A. E., Cooper, K. G., and Richey, T. L.: Rat behaviors during unsignalled avoidance and
conditioned suppression training. Bull. Psychonomic Soc., 9,373-376, 1977.
Schaeffer, R. W.: The reinforcement relation as a function of instrumental response base rate. 1.
Exp. Psychol., 69,419-425, 1965.
Schaeffer, R. W., and Premack, D.: Licking rates in infant albino rats. Science, 134, 1980-1981,
1961.
Schwartz, B., and Gamzu, E.: Pavlovian control of operant behavior. In, W. K. Honig and J. E. R.
Staddon (eds.): Handbook of Operant Behavior. Englewood Cliffs, N. J., Prentice-Hall, 1977.
Sidman, M.: By-products of aversive control. 1. Exp. Anal. Behav., 1,265-280, 1958.
Chapter 2: Empirical Analysis of Concurrent Classical-Operant Conditioning 159

Sidman, M,: Tactics of Scientific Research: Evaluating Experimental Data in Psychology. New
York, Basic Books, 1960.
Sidman, M., Herrnstein, R. J., and Conrad, D. G.: Maintenance of avoidance behavior by unavoid-
able shock. J. Compo Physiol. Psychol., 50,553-557,1957.
Skinner, B. F.: "Superstition" in the pigeon. J. Exp. Psychol., 38, 168-172, 1948.
Skinner, B. F.: Are theories of learning necessary. Psychol. Rev., 57, 193-216, 1950.
Smith, J. C.: Conditioned suppression as an animal psychophysical technique. In W. Stebbins
(ed.): Animal Psychophysics. New York, Appleton-Century-Crofts, 1970.
Solomon, R. L., and Turner, L. H.: Discriminative classical conditioning in dogs paralyzed by
curare can later control discriminative avoidance responses in the normal state. Psychol. Rev.,
69, 202-219, 1962.
Stein, L., Sidman, M., and Brady, J. V.: Some effects of two temporal variables on conditioned
suppression. J. Exp. Anal. Behav., 1, 153-162, 1958.
Stein, N., Hoffman, H. S., and Stitt, c.: Collateral behavior of the pigeon during conditioned sup-
pression of key pecking. J. Exp. Anal. Behav., 15,83-93, 1971.
Terrace, H.: Discrimination learning with and without "errors." J. Exp. Anal. Behav., 6, 1-27,
1963.
Valenstein, E. S., and Cox, V. C.: Influence of hunger, thirst, and previous experience in the test
chamber on stimulus bound eating and drinking. J. Compo Physiol. Psychol., 70, 189-199, 1970.
Van Dyne, G.: Conditioned suppression with a positive US in the rat. J. Compo Physiol. Psychol.,
77, 131-135, 1971.
Weiskrantz, L.: Emotion. In, L. Weiskrantz (ed.): Analysis of Behavioral Change. New York,
Harper & Row, 1968.
Zener, K.: The significance of behavior accompanying conditioned salivary secretion for theories
of the conditioned response. Am. J. Psychol., 50,384-403, 1937.
SECTION II

OPERANT CONDITIONING
PROCEDURES
Chapter 3

Concurrent Schedules: Response

versus Reinforcement Interaction
Iver H. Iversen

The Issue of Response Versus Reinforcement Interaction

The fact is now well established that presentation of reinforcers contingent
upon a specific response will usually increase the rate of that response, and
subsequent removal of reinforcers will decrease the response rate (Skinner,
1938). Skinner once briefly introduced and quickly withdrew a principle stating
that the rate (strength) of one response (R) might be directly proportional to the
absolute reinforcement rate (r) for that response; expressed mathematically,
R =Kr. (1)
This simplistic relationship between an independent variable (reinforcement
rate, r) and a dependent variable (response rate, R) was abandoned because
experimental findings did not accord well with the assumption of a direct pro-
portionality between response and reinforcement rates. Rather, Skinner (1938,
1950, 1953, 1969) clearly argued that the rate of one response may realistically
be dependent upon more than just one variable.
More recently, a principle reminiscent of equation 1 has been put forward as
applicable to a wide range of behavioral settings. Thus, Herrnstein (1970) sug-
gested that the rate of one response might be directly proportional to the rela-
tive rate of reinforcement for that response; expressed mathematically,
R =K_r_. (2)
r + ro
Herrnstein argued that the reinforcement rate assigned to one response should
be considered within the context of (or relative to) the total operating reinforce-
ment rate for all responses. The symbol ro thus refers to the aggregate rein-
forcement rate from all responses concurrent with the reference response.
164 Section II: Operant Conditioning Procedures

The assumptions encompassed by equation 2 explicitly express, first, an in-

dependence between the absolute rate of one response and the absolute rates of
concurrently occurring responses. The right-hand side of equation 2 simply
does not contain a parameter representing the rates of concurrent responses
(Catania, 1963, 1969; Pliskoff et al., 1968; Rachlin and Baum, 1969; Nevin,
1973). Second, the absolute rate of one response is considered to be inversely
related to the absolute reinforcement rates of concurrent responses (increases
in ro enlarge the denominator and thereby decrease the size of R). The first as-
sumption has been referred to as response independence or constancy (Ca-
tania, 1966) and the second as reinforcement interaction or contrast, or as an
inhibitory effect of concurrent reinforcement (Catania, 1969, 1973; Herrnstein,
1970). Significantly, equation 2 proposes that the rate of one response is fully
described by the reinforcement rate for that response and the reinforcement
rate for concurrent responses.
These assumptions of response independence and reinforcement interaction
-referred to as the reinforcement model-have mainly been supported by ex-
periments with combinations of positive operant conditioning procedures, bet-
ter known as concurrent reinforcement schedules.
In a previous review, Catania summarized the theoretical position of rein-
forcement interaction:
Concurrent performances, therefore may be described as the product of reinforce-
ment interaction, in that the rate of each operant is determined not only by its own
rate of reinforcement but also by the concurrent rates of reinforcement of other
operants, and response independence, in that the rate of each operant is not deter-
mined by the concurrent rates of other operants. (Catania, 1966, p. 248)
Response independence and reinforcement interaction would appear, how.
ever, to be somewhat different from, if not directly opposed to, the principle of
response interaction that applies for other conditioning procedures. Thus, re-
sponse interactions are said to describe well the findings from combinations of
operant and classical conditioning procedures (Chapter 2), classical condition-
ing procedures (Chapter 6), positive operant and avoidance conditioning proce-
dures (Sidman, 1962), and some avoidance conditioning procedures (Verhave,
1961). Nevertheless, response interaction has been vigorously dismissed as a
possible characterization of concurrent performances in combinations of posi-
tive operant reinforcement procedures. Response interactions have therefore
not been generalized to an extent warranting any serious limitation of the re-
sponse independence principle in concurrent operant schedules. Instead, nega-
tively, we must conclude that one and the same behavioral principle apparently
cannot encompass the different conditioning procedures.
The purpose of this chapter is to examine the opposing models of concurrent
performances as they apply to combinations of positive operant conditioning
procedures.

Response Independence and Reinforcement Interaction

In the early literature on concurrent reinforcement schedules the emphasis was
on establishing generality of schedule control obtained with single-response
Chapter 3: Concurrent Schedules 165

schedules. A significant question was whether the rate and pattern of a re-
sponse continuously maintained on one schedule (an ongoing response) was the
same whether or not a second concurrently available response was simulta-
neously maintained on a separate schedule. * Ferster and Skinner (1957), for
example, reported that each of two responses separately maintained on VI
schedules occurred at a uniform rate characteristic of VI. Ferster (1957) also
reported that concurrent responses maintained by different schedules such as
FR and VI showed patterns characteristic of the same schedules in isolation: a
"break and run" pattern of FR responding and a more uniform rate of VI re-
sponding. These early results led to the more general definition that concurrent
operants are "capable of being executed with little mutual interference" (Fer-
ster and Skinner, 1957, p. 724). Ferster suggested that concurrent schedules
could then be used to study "bilateral independence" and that "the perform-
ance on one key could be used as a baseline for the emotional side effects of a
change in the schedule of reinforcement on a second key" (Ferster, 1957, p.
1091).
The study of changes of an ongoing response after schedule changes for a
second response has in fact been a major concern of subsequent research on
concurrent schedules. A more general finding, if not the prime feature of con-
current performances, has been that increased reinforcement for a concurrent
response not only increases the rate of that response, but also simultaneously
decreases the rate of an ongoing response (Catania, 1963, 1966, 1969, 1973;
Herrnstein, 1961, 1970, 1974; Baum, 1973, 1974; Rachlin, 1973). In this sense,
the rate of an ongoing response may certainly be affected by the schedule per se
for a second concurrent response.
The terminology employed in concurrent performances is now slightly more
descriptive, and the term emotional effect has been abandoned. However, the
fundamental assumption of response independence proved to be far more in-
fluential. Since both the response and the reinforcement rate vary in the same
direction under the manipulated schedule, a significant question has been
whether the rate change of the ongoing response should be attributed to the
response or the reinforcement rate change on the manipulated schedule.
With response interaction, the increased time allocated to concurrent re-
sponding after an increase in concurrent reinforcement rate must decrease the
time available to, and thereby the rate of the ongoing response. With reinforce-
ment interaction, on the other hand, the decreased rate of the ongoing response
is said to be independent of the increased time allocated to concurrent respond-
ing.
While an abundant number of experiments have been done with concurrent
schedules, relatively few have investigated the issue of response versus rein-
forcement interaction. The experimental approach has been to separate or pre-

* In the following, an ongoing response refers to a response continuously maintained by a

schedule. Concurrent responses refers to other responses available within the same session as the
ongoing response. Concurrent performances, on the other hand, is used in describing the overall
situation of two or more responses maintained on separate but concurrently operating schedules.
Reinforcements refer to reinforcers delivered contingent on an ongoing response, delivered contin-
gent upon a concurrent response, or delivered independent of any response.
166 Section II: Operant Conditioning Procedures

vent concomitant changes in concurrent response and reinforcement rates. Ca-

tania (1%3) reported that the rate of an ongoing response was inversely related
to the rate of concurrent reinforcement on a signaled VI schedule (i.e., a signal
was presented when a concurrent reinforcer was eligible, and the first concur-
rent response during the signal produced signaled reinforcement). This finding
was replicated in subsequent experiments with roughly similar procedures (Ca-
tania, 1969; Rachlin and Baum, 1%9; Catania and Dobson, 1972; Catania et al.,
1974). Rachlin and Baum (1972) further reported that the rate of an ongoing re-
sponse was inversely related to the rate of concurrent reinforcement on a
response-independent schedule (i.e., no responses were required for concur-
rent reinforcements). Since both procedures maintained a low or zero recorded
concurrent response rate, the change of the ongoing response was said to be
dependent upon concurrent reinforcement rates and to be independent of con-
current response rates.
Simply, the demonstrations of apparent response independence within these
procedures were considered to provide substantial confirmation of the assump-
tion of reinforcement interaction (Herrnstein, 1970; Catania, 1973; Rachlin,
1973). Hermstein, for example, in a review of signaled reinforcement, con-
cludes that "contrast depends upon the reinforcement for the other alternative,
not on the responding there" (1970, p. 257). Importantly, the implication is that
the rate of an ongoing response decreases not because of the simultaneously
increased rate of the concurrent response, but only because of the increased
reinforcement rate for the concurrent response.
The data obtained with the signaled and response-independent procedures
appear to constitute the empirical support for the claim of reinforcement inter-
action in the absence of response interaction (Catania, 1%9; Hermstein, 1970,
1974; Rachlin and Baum, 1972; Nevin, 1973; Rachlin, 1973). Note that the par-
ticular logic of reinforcement interaction requires not only that the procedures
yield an unchanged (low or zero) rate ofthe second response, but also that they
do not otherwise generate concurrent responses that compete for time with the
ongoing response. Some emphasis was therefore placed in arguing against the
possibility that concurrent responses other than those recorded might have af-
fected the rate of the ongoing response. The generation and eventual influence
of observing responses toward the signal in the signaled reinforcement proce-
dure, and of superstitious responses in the response-independent procedure,
were clearly dismissed by various arguments (Catania, 1969; Rachlin and
Baum, 1%9, 1972; Catania and Dobson, 1972). (The details of the arguments
are given in full in Experiments VI, VII, VIII, and IX in this chapter).
The characterization of concurrent performances in terms of response inde-
pendence and reinforcement interaction appears to have received tremendous
support during the past decade. A relatively large number of experimental and
theoretical papers have been devoted to an application of the reinforcement
theory to concurrent schedules in general, and the obtained reciprocal response
rate relations were treated as an example of reinforcement interaction. This he-
gemony of the reinforcement model would suggest an apparent settlement of
the issue of response versus reinforcement interaction.
Chapter 3: Concurrent Schedules 167

In general, the results from concurrent schedules are treated as a relation-

ship between relative response and reinforcement rates. When the response
ratio equals the reinforcement ratio, the response distribution is said to match
the reinforcement distribution (Herrnstein, 1970, 1974). Expressed mathemati-
cally, this matching law is
R
------
r (3)
R + Ro r + ro
We will not go into detail as to whether or not relative response rates invari-
ably match relative reinforcement rates. Several experiments have in fact dem-
onstrated deviations from matching. Suffice to mention here that systematic de-
viations from matching tend to be obtained when parameters other than
reinforcement are scheduled asymmetrically on concurrent schedules with
equal reinforcement rates, or are symmetrically arranged on schedules with
unequal reinforcement rates (Chung, 1965; Shull and Pliskoff, 1967; Thomas,
1968; Todorov, 1971; Deluty, 1976a). Shull and Pliskoff (1967), for example,
reported that the difference between the relative response and reinforcement
rates on VI schedules with unequal reinforcement rates was a decreasing func-
tion of the duration of a changeover delay (COD) scheduled in both compo-
nents (see also Brownstein and Pliskoff, 1968; Schroeder, 1975). Similarly, rel-
ative response and reinforcement rates did not match when response-contingent
punishment was presented with different rates in equal RI schedules (Deluty,
1976a; see also Carlson and Aroksaar, 1970; Katz, 1973).

Direct or Indirect Reinforcement Interaction

Before turning to the experimental analysis of response versus reinforcement
interaction, we may briefly consider the theoretical consequences of an even-
tual change in emphasis from response independence to response interaction in
the characterization of concurrent performances. Certainly, the rate of an on-
going response (as the dependent variable) has been shown to be inversely re-
lated to the rate of concurrent reinforcement (as the independent variable) in
several experiments. If this well-established functional relationship is not me-
diated by changes in concurrent response rates, a relevant question is how the
relationship then comes about. This issue has long been explained by saying
that an ongoing response is affected directly by concurrent reinforcements (Ca-
tania, 1966, 1973). However, the interpretative emphasis on a direct connection
between concurrent reinforcement and ongoing response may not necessarily
imply a final settlement of the empirical analysis of concurrent performances.
The relationship between concurrent reinforcement and an ongoing response
might be further investigated, and perhaps the theoretical issues have not yet
been sufficiently debated to warrant a complete rejection of alternative inter-
pretations.
In the literature of concurrent schedules, the alternative interpretation of re-
sponse interactions has occasionally questioned response independence and
reinforcement interaction. While response interaction has been dismissed sev-
168 Section II: Operant Conditioning Procedures

eral times, the possible characterization of concurrent performances in terms of

observed response interactions was in fact recognized at least to have the ad-
vantage of not appealing to unobserved inhibitory processes that are implicit in
the reinforcement model (Catania, 1969).
Let us accept for the sake of argument the alternative interpretation of re-
sponse interaction. The inverse relationship between the rate of an ongoing re-
sponse and concurrent reinforcement rates would still hold as a functional rela-
tionship, but now the relation would be indirect rather than direct. If the rate of
concurrent reinforcement is increased, for example, and the rate of concurrent
responses thereby increases simultaneously, then a decreased rate of the ongo-
ing response would be attributed directly to the increased concurrent response
rate, and only indirectly to the increased concurrent reinforcement rate. With
response interaction, concurrent reinforcements simply act upon concurrent
responses, which in tum directly influence the ongoing response.
The following experiments will argue in favor of a change in the causal vo-
cabulary from a direct to indirect dependent relationship between an ongoing
response and concurrent reinforcement rates. This vocabulary change may ap-
pear small and insignificant in itself. The assumptions of response indepen-
dence and exclusive reinforcement interaction would, however, be abandoned
if concurrent performances could be sufficiently described as interactions be-
tween responses.

Response Pattern Reciprocity in Concurrent

FR VI Schedules (Experiments I and 11)*
While overall changes in an ongoing response are generally considered to be
independent of a second concurrent response, more local changes were not
thought to be entirely independent of concurrent responses in the early litera-
ture. Ferster (1957, 1959) noted, for example, that VI responding was tem-
porally interrupted during high local rates of a concurrent FR response. More
gross local inversions were also described between the rates of responses main-
tained on schedules such as concurrent (cone) FR FI (Ferster and Skinner,
1957), cone VI FR (Catania, 1966), and cone VI FI (Nevin, 1971). In the Nevin
study, for example, the FI response occurred in the typical scallop, with a grad-
ually increasing rate between reinforcements. The concurrent VI response was
disrupted, however, with a gradually decreasing rate between the FI reinforce-
ments. Thus the pattern of an ongoing response may, at least under some con-
ditions, show a partial dependence on the simultaneous pattern of a concurrent
response.
Since response interactions already have been admitted at the level of local
response rates, a good starting point for a concurrent response analysis would
be a further investigation of the relationships between local response patterns
in concurrent schedules. In the following two experiments, the pattern of FR
responding was simply compared to the pattern of concurrent VI responding.
* Experiments were conducted at the University of Copenhagen, from January to October
1971, and reported by I. Iversen (1974).
Chapter 3: Concurrent Schedules 169

Experiment I
The beginning experiment examined changes in FR and VI response patterns
established by successively presented schedules when the schedules were
made simultaneously available. More specifically, two male pigeons were
maintained at 80% of their free-feeding body weights. The pigeons had served
in previous experiments with single and cone VI FR schedules. When the cen-
ter key of a three-key pigeon chamber (Iversen, 1975a) was red, pecks on this
key were reinforced on VI 2 min. The interreinforcement intervals were distrib-
uted according to the formula of FleshIer and Hoffman (1962). A reinforcer set
up on this schedule was only accessible for a 3-sec period. The first peck in this
period produced the reinforcer, but reinforcer delivery was canceled if no
pecks occurred [limited hold (LH) procedure]. When the left key was lit green,
pecks on this key were reinforced on FR 50. The FR counter was only reset
after reinforcer delivery for FR responding. The right key was never lit and
pecks on this key had no effect. In the first experimental phase (30 sessions),
the schedules were presented successively. A session of 3200 sec was divided
into eight 400-sec periods. During each period, only one key was lit, and rein-
forcers were only scheduled for pecks on that key. Pecks on the other, nonillu-
minated key had no effect. The schedules were always presented in strict alter-
nation.
In the second phase (18 sessions), both keys were lit simultaneously and the
associated schedules operated concurrently throughout a session. The sched-
ules were independently programmed so that pecks on one key did not affect
the response-reinforcement contingency on the second key. Reinforcer deliv-
ery was 3 sec of access to mixed grain.
With the cone VI FR in the second phase, the terminal performance was in-
variably a substantially reduced rate on one key and an approximately un-
changed rate on the second key (Table 3.1). The pauses in FR responding in-
creased for both pigeons and VI responding occurred in bursts after FR
reinforcement. For pigeon A5, the FR pauses were short with few interspersed
VI responses, whereas for pigeon A6, the pauses were long with a high number
of VI responses (Table 3.1).

Table 3.1 Response and reinforcement rates (responses/min, reinforcements/min) for

FR and VI responding, pause durations and run times (sec) in FR responding, and the
number of VI responses emitted during FR pauses and runs
Response Reinforce- Number of VI
rates ment rates responses
FR FR run
Pigeon no. Schedules FR VI FR VI pausea time FR pausea FR run

A5 in isolation 89.8 61.2 1.79 0.44 6.8 26.6 0.0 0.0

simultaneously 83.4 8.6 1.66 0.15 9.4 27.0 4.1 1.1

A6 in isolation 119.6 63.6 2.38 0.43 6.3 18.8 0.0 0.0

simultaneously 10.2 66.0 0.20 0.44 89.5 205.6 103.5 221.5
a FR pauses were measured to the first FR response after FR reinforcement. Values are means of the last six
sessions of each phase.
 --
170 Section II: Operant Conditioning Procedures

~L 6 min

Figure 3.1. Sample cumulative record showing reciprocity between FR responding and VI re-
sponding. VI responding is shown on the event pen with segments displaced to the appropriate
pause for easy comparison. Reinforcements for both responses are indicated as hatchmarks on the
stepping pen.

The cumulative record of FR responding and the event traces of VI respond-

ing in Figure 3.1 clearly show the response pattern reciprocity. The burst of VI
responding and the concomitant pause in FR responding were closely co-
variant, and the FR pause was apparently independent of whether or not the
burst of VI responding was reinforced.
We should note that the recorded responses were physically incompatible,
which by necessity assures some degree of response interdependence. Thus,
pauses in one response clearly cannot be shorter than bursts of a concurrent
incompatible response. However, the pause in one response can be indefinitely
longer than (and therefore patrially independent of) the burst of a second re-
sponse. Since the FR pause approached the burst duration of VI responding in
this study, FR pausing was then not totally independent of the concurrent VI
response. The results would then seem to suggest some form of interdepen-
dence between concurrently reinforced responses.

Experiment II
To further analyze the interrelationships between concurrent FR and VI re-
sponding, the next experiment directly manipulated the duration of access to
VI responding. The experiment also sought to determine whether physical in-
compatibility is necessary for inverse response rate relations. Therefore, mon-
keys were trained in a cage with two response levers that could easily be de-
pressed simultaneously. The experimentally naive monkeys were
Cercopithecus aethiops (African green monkeys) and were deprived offood for
approximately 21 hr prior to each daily session. The levers were 9.0 cm long
and 2.0 cm in diameter: they operated a switch when depressed 2.0 cm with a
Chapter 3: Concurrent Schedules 171

force of 150 g. The distance between the levers was 26.0 cm. The 1-g food pel-
lets were delivered from a food cup centered between the two levers. During
sessions, presses on the left lever were followed by response feedback of a 0.1-
sec darkening of a white light in the lever and a O.I-sec 4000-Hz tone. When a
red stimulus light was lit above the right lever, presses on the right lever were
followed by a O.I-sec illumination of a white light in the lever and a 0.1-sec 400-
Hz tone.
Mter pretraining, the terminal schedules in effect throughout a session were
either an FR 100 (monkey 1) or FR 80 (monkey 2) on the left lever. The right
lever was programmed with a discrete-trial VI 40 sec with a limited hold of 5
sec. In the first phase, the discrete VI trials and intertrials (EXT) occurred in
strictly alternating 150-sec periods. In the second phase, trial durations of 10,
20,40,60, 120,240, or 300 sec occurred in mixed order separated by variable
intertrial intervals of 40 to 360 sec. In both phases, shifts between trial and in-
tertrial intervals were independent of responses on any lever. A session was
terminated after delivery of 50 pellets.
This paradigm of presenting and retracting the discriminative stimulus for
right lever responding could formally be described either as a discrete-trial pro-
cedure or a multiple schedule. In the following, trials therefore refers to the VI
component of the mult VI (LH) EXT schedule for the right lever, and intertrial
intervals refers to the EXT component.
Presenting the VI discriminative stimulus resulted not only in large rate in-
creases in VI responding, but also in decreases in FR responding (Table 3.2).
The manipulated trial durations, furthermore, produced different burst dura-
tions of VI responding, which in tum affected the ongoing pattern of FR re-
sponding.
Figure 3.2 presents sample cumulative records for the fixed and variable trial
durations. The interruption in FR responding was complete and closely fol-
lowed the trial duration for monkey 2 since trial onset controlled an immediate
changeover to VI responding (a) and trial offset controlled an immediate
changeover back to FR responding (d and m). For monkey 1, FR responding

Table 3.2 Mean response and reinforcement rates calculated for the last five
sessions of each phase
Response rates Reinforcement rates
(responses/min) (reinforcements/min)

Left lever Right lever Left lever Right lever

Monkey Trial
no. duration VIa EXT" VIa EXT" VIa EXT" VIa EXT"

fixed 38.1 153.1 7004 204 0.38 1.53 1.36 0.0

variable 38.3 190.0 61.5 1.8 0.38 1.90 lAO 0.0

2 fixed 0.3 66.5 75.2 0.3 0.0 0.83 1.43 0.0

variable 0.3 61.4 70.2 0.2 0.0 0.77 1.39 0.0
a Schedule for right lever.
172 Section II: Operant Conditioning Procedures

MONKEY MONKEY 2

!L 6 MIN.

300 60 260
Figure 3.2. Sample cumulative records illustrating the effects of discrete VI trials upon ongoing
FR responding. VI responding is shown on the event pen as downward deflections during trials and
upward deflections during intertrial intervals. Pellet deliveries for both responses are shown as
hatchmarks on the stepping pen. Upper records are from the first phase, with fixed trial duration;
lower records are from the second phase, with variable trial duration. Numbers refer to the trial
duration, and letters refer to details discussed in the text.

continued occasionally after trial onset, with the changeover to VI responding

then occurring immediately after FR pellet delivery (b and k). Moreover, change-
overs from VI responding to brief bursts (c) or complete runs 0) ofFR respond-
ing occurred during some trials for monkey 1. Therefore, FR responding was
suppressed during trials to the extent that trial presentation controlled emission
of VI responses. For both monkeys, trials presented immediately after FR rein-
forcement produced a postreinforcement pause approximating the duration of
the concurrent VI trial. In a similar fashion, trials presented during the FR run
increased the run time by an amount approximating the trial duration (h and j).
During intertrial intervals, VI responding was at zero rate with the major ex-
ception of brief bursts after FR pellet deliveries for monkey 1 (e and n), and
occasionally for Monkey 2.
Overall, the interruptions in FR responding more closely approximated the
actual duration of VI responding than the programmed trial duration. Further-
more, the responses were physically compatible rather than incompatible,
hence one response did not prevent the occurrence of the second response. Re-
Chapter 3: Concurrent Schedules 173

ciprocal interactions in concurrent schedules may therefore not necessarily re-

sult from or be reduced to response incompatibility.

Discussion
Demonstrations of response pattern interdependence in concurrent schedules
may involve two levels of comparison. At the level of averaged response and
reinforcement rates, the reinforcement model specifically states that inhibition
of an ongoing response does not result from an increased rate of a concurrent
response but only from an increased concurrent reinforcement rate. The pres-
ent data would not disagree qualitatively with this model for averaged, overall
response and reinforcement rates (Tables 3.1 and 3.2); that is, the decreased
FR response rate was associated with increases in both concurrent response
and reinforcement rates. At the molecular level of response changes during in-
dividual trials, however, the reinforcement model only uneasily explains the re-
sponse rate changes. The burst duration of concurrent VI responding ranged
widely in Experiment I and was purposely manipulated between 10 and 300 sec
in Experiment II. Therefore, the local VI reinforcement rate was highly vari-
able from one VI trial to the next, with an absence of VI reinforcement during
many of the trials. If the reinforcement model can be applied to the consistent
response rate changes during trials, the inhibition from VI reinforcement upon
FR responding must exactly match the duration of VI responding, not the vari-
able rate of obtained VI reinforcement.
In the present experiments, alterations of the ongoing FR response pattern
were far more closely associated with the concurrent pattern of VI responding
than with the actually delivered VI reinforcers. Similarly, FR pauses may be
directly affected by bursts of collateral responding that does not require deliv-
ery of reinforcers (Iversen, 1976). Alterations of FR responding are thus more
closely accounted for by interacting response patterns than by interacting rein-
forcement patterns.
The reinforcement model would seem to require essentially that inhibition
by concurrent reinforcement must somehow be tightly associated with the
emission of a concurrent response. This analysis does not appear to be peculiar
to the specific procedures of the present experiments. As mentioned before,
Nevin (1971) employed continuously operating concurrent VI and FI schedules
and found a gradually decreasing VI response rate between FI reinforcements.
Again, the effect of the FI schedule on the VI response pattern must somehow
be connected with the generation of a scallop of FI responding.
Response changes at a more local level go beyond the premises of the rein-
forcement model, and are therefore usually excluded from analysis. Thus
Baum, for example, asserted that "orderly relations between behavior and en-
vironment should emerge at the level of aggregate flow in time, rather than mo-
mentary events" (1973, p. 137). A critical question then is the degree of data
averaging. Whether reinforcement interaction can account for local response
rate changes remains unclear. In contrast, response interaction may easily ac-
count for both overall and local changes in responding.
174 Section II: Operant Conditioning Procedures

Changeover Responses Under Discriminative Control with

Discrete-Trial Procedures (Experiments ill, IV, and V)*
A changeover from an ongoing response to a concurrent response may involve
not only concurrent changes but also sequential alternation between responses.
The sequential changes were not explicitly controlled in Experiment I, with the
changeover from FR to VI responding consistently only after FR reinforce-
ment. The more explicit experimental control of burst durations in Experiment
II established changeovers controlled by the onset of the VI discriminative
stimulus. The results would suggest that changeover responses might profitably
be investigated by various paradigms of discriminative stimulus control. The
explicit investigation of the discriminative control of changeover responses,
however, has been undertaken in only a few experiments.
Sidman (1956) studied the changes in DRL responding during a discrimina-
tive stimulus controlling a concurrent response. With this design, the probabil-
ity of the second response was a decreasing function of the time of stimulus
onset within the DRL requirement. Boren (1961) similarly studied the effects
upon an FR response of an added stimulus during which a concurrent response
was reinforced on a CRF schedule for only one reinforcement. The superim-
posed discriminative stimulus was presented at different positions within the
FR run. The probability of a changeover to the concurrent response was a de-
creasing function of the number of FR responses emitted prior to presentation
of the discriminative stimulus. (In the experiments by Sidman and Boren, the
response contingencies were not exactly independent. The DRL in Sidman's
experiment and the FR in Boren's experiment were both "reset" by the second
response.) The procedure of signaling reinforcement for a second response has
subsequently been used in a series of experiments with VI schedules (Catania,
1963, 1969; Rachlin and Baum, 1969; Pliskoff and Green, 1972).
The changeover response has also been investigated when the second re-
sponse changed over to merely alters the schedule for the first response
changed over from (Findley, 1958; Sherman and Thomas, 1968). More re-
cently, discriminative control of changeovers has been studied with concurrent
operant and classical conditioning procedures (Henton and Iversen, 1973;
Chapter 2, this volume). In these schedules, a changeover from the operant to
the classically conditioned response immediately occurred at onset of condi-
tioned stimuli controlling a high rate of the classically conditioned response.
The following three experiments were an attempt to investigate further the
variables that control changeover responses within concurrent schedules. A
central question is: To what extent is the interruption of an ongoing response
related to concurrent responses controlled by a discriminative stimulus?" All
three experiments maintained one response on a continuously accessible
schedule and a second response on a discrete-trial procedure. The advantage of
the trial procedure, in addition to the elaborate discriminative control of

* Experiments conducted at the University of Copenhagen, from January 1971 to January 1974,
and reported by I. Iversen (1974, 1975a).
Chapter 3: Concurrent Schedules 175

changeovers, is the explicit within-session manipulation and dissociation of

concurrent response and reinforcement parameters.

Experiment III
This experiment was designed to study the rate and pattern of VI responding
during the acquisition, extinction, and reacquisition of discriminative control of
concurrent FR responding (Iversen, 1975a). Pigeons were maintained on 80%
of their free-feeding body weights and pecking one red key was reinforced on a
VI2 min, with an LH of3.6 sec. The concurrent discrete trial was illumination
of a second key by green light. In the first phase (acquisition), responding on
the green key was reinforced on FR 40. After 40 pecks, a reinforcer was pre-
sented and the green light was turned off. To assess the effects of FR extinc-
tion, pecks on the green key were no longer reinforced and the trial duration
was 40 sec in the second phase (EXT). Finally, pecking the green key was again
reinforced on FR 40 in the third phase (reacquisition). In all phases, 12 trials
occurred with variable intertrial intervals and were independent of pecks on
any key. Each phase consisted of fifteen 70-min sessions and the reinforcer was
access to mixed grain for 4 sec.
The intermittent trial presentation during acquisition controlled an immedi-
ate changeover from VI to FR responding. VI responding was completely sup-
pressed during all trials (Figure 3.3). Removal of the concurrent FR reinforcer

~
ACQUISITION

. '1-"--""'-I f
EXTINCTION

-fR-KEY
VI - KEY
C>-<)

~L
r
REACQUISITION

15 30 45
SESSIONS
Figure 3.3. Relationships between FR and VI response rates during the acquisition, extinction
and reacquisition of discrete-trial FR responding. Data are means of 12 trials presented during
each session. From Iversen: Scand. J. Physioi., 16, 280-284, 1975.
176 Section II : Operant Conditioning Procedures

(EXT) resulted in a gradual increase in the VI response rate along with the de-
crease in the FR response rate. Reacquisition of FR responding reversed the
response pattern changes, with a gradually decreased rate of VI responding
along with the increased rate of FR responding.
According to the reinforcement model, removal of reinforcers for a concur-
rent response increases an ongoing response rate because of an increased rela-
tive reinforcement rate . The model would thus seem to predict an abrupt in-
crease in VI response rates to match the immediate increase in relative VI
reinforcement rate during the FR EXT. Therefore, the reinforcement model
may apparently not account for the gradually increasing VI response rate dur-
ing extinction of FR responding, unless an additional inhibitory effect of previ-
ous FR reinforcers is assumed to decay gradually. Such an inferred inhibitory
after-effect of previous concurrent reinforcement must at least closely follow if
not exactly mimic the extinction curve for the concurrent response. However,
the VI response rate was precisely decreased only during those EXT trials in
which FR responding still occurred, not during trials in which FR responses
were absent. Similarly, during reacquisition, the VI response rate decreased
only when FR responses occurred during trials. Examples of the minute inter-
relationships between the responses are shown in sample event records in Fig-
ure 3.4.
The influence of FR reinforcement must then closely follow changes in the
FR response if an inhibitory effect is to cogently explain changes in the VI re-
sponse rate. More importantly, perhaps, changes in the VI response rate would
then seem to be as well described by reference to the FR response rate changes
in and of themselves, as by reinforcement inhibition inferred from changes in
VI responding.
Although the reinforcement model does not address local or molecular re-
sponse changes (Catania, 1966; Baum, 1973), the very close negative correla-
tions between local rates of an ongoing response and a concurrent response
nevertheless remain to be explained. Molecular response changes have in fact
been left in limbo for quite sometime. Perhaps the argument against local or

ACQ. EXT. REACQ.

2hi. _.IIAIIi._ .101 ••• 11 ..mi••ii1lll_II1,............rr,- -____
• ..-alii ._ ,.iiURi"", Ii' i1....... 1111 ..
3---....--......

.Nil8iii• • _ ,_

-
• • • 18'1111 _ M 1 . _ •••• i
. . .w
1181_

-
Ii iilAiNiii,lilAlllllliililiiiill iiliiill,"

••__....
.-
' _ _••rnl•. - -- ._._111._ ••• liJ,i1.IMUiiIIM,Iibldii •• AN_1i1 i.e/ii.i HI' iI_il i,AIIi

- '• • 11111'111 I.,

30see.

Figure 3.4. Event records of VI and FR responding for each phase for pigeon A3. On pen 3, ex-
tended deflections refer to trial presentations and brief deflections refer to reinforcer delivery. I,
VI-Key; 2, FR-Key; 3, FR-triaIs or reinforcement. Adapted from Iversen: Scand. J. Physiol., 16,
280-284 , 1975.
Chapter 3: Concurrent Schedules 177

transitional interactions could be reversed, so that any general description of

concurrent performances might attempt to account for all performance aspects
rather than only overall response rates.

Experiment IV
In Experiment III both the response rate and the reinforcement rate were rela-
tively higher for the concurrent response than for the ongoing response. This
finding raises the question of whether the relatively higher rate of the concur-
rent response is necessary for the complete suppression of the ongoing re-
sponse. The first purpose of Experiment IV was to examine this question; it's
second purpose was to determine the immediacy of the discriminative control
of the changeover as a function of the concurrent reinforcement rate.
Three male pigeons were maintained at 80% of their free-feeding body
weights. The pigeons had served in a previous experiment with simple VI
schedules, and one pigeon (AI) had served in a preliminary experiment. Peck-
ing of a red key (key 1), centered on the chamber wall with the food hopper,
was reinforced on a VI 120 sec with an LH of 3.6 sec throughout sessions.
Pecking of a second key (key 2), centered on the adjoining left-hand wall, was
reinforced on a discrete-trial procedure.
One-minute trials of white or green illumination of key 2 each occurred six
times per session, intermixed with intertrial intervals (key 2 dark) of 90 to 360
sec. During white trials, key 2 pecking was reinforced on VI 60 sec (Phase A),
VI 40 sec (Phase B), VI 120 sec (Phase C), and VI 40 sec again (Phase D). Dur-
ing green trials, key 2 pecking was always reinforced on VI 180 sec. To prevent
variability in the key 2 reinforcement rate over sessions (because of the rela-
tively brief exposure time to the discrete trials), the programmer was arranged
so that each schedule would assign a fixed number of reinforcer deliveries dur-
ing a session. The number of sessions in Phases A, B, C, and D was 18, 18,30,
and 10, respectively, for pigeon AI; for pigeon A2 it was 41,30,43, and 40,
respectively; and for pigeon A4 it was 30, 30, 50, and 32, respectively.
Figure 3.5 presents the results of this cone VI (mult VI VI EXT) schedule.
During trials, the concurrent key 2 response rate was an increasing function,
and the ongoing key 1 response rate was a decreasing function of the key 2 rein-
forcement rate. Note that the key 1 response rate decreased to zero with the
highest key 2 reinforcement rates. For pigeons A2 and A4, the decrease in key
1 response rate exceeded the increase in key 2 response rate from intertrial to
trial intervals. The complete suppression of key 1 responding thus did not re-
quire a changeover to a relatively higher key 2 response rate; that is, the key 1
baseline rate was higher than the key 2 trial rate. In other words, the key 1 de-
crease did not necessarily balance the key 2 increase in responses/min. Hence,
the interruption or total suppression of an ongoing response would not seem to
require changeover to a relatively higher concurrent response rate.
At the level of averaged data, the change in the ongoing response was related
to both an increased rate of concurrent responding as well as an increased rate
of concurrent reinforcement. However, analyzed during individual trials, the
178 Section II: Operant Conditioning Procedures

A1 ~
80 , 0- --

-z
~
60
40
a:en 20

-
I.&J
a:::
I.&J A4
140
~
a::: 120 120
I.&J
en 100 100
Z
0 80 80
0....
en J:r - --()
I.&J 60 60
~
40 40
20 20
,

0.0 03305 to 1.5 0.0 0.3305 1.0 1.5

REINF/MIN ON KEY 2
Figure 3.5. Mean key 2 (white circles) and key 1 (black circles) response rates as a function of the
key 2 reinforcement rate. The response rates were averaged for all trials with equal key 2 sched-
ules, and for intertrial intervals (zero key 2 reinforcement rate). Data are presented as means for the
last 10 sessions of each phase.

ongoing key 1 rate appeared to be related only to the concurrent response rate
on key 2. Figure 3.6 shows the suppression of key I responding during trials
and the key 2 reinforcement rate (Figure 3.6A,B) compared with the key 2 re-
sponse rate (Figure 3 .6C,D) for overall and local response measures. For com-
parison of means (Figure 3.6A,C), the percent suppression of key I responding
is an increasing function of the key 2 reinforcement rate as well as the key 2
response rate. For individual trials (Figure 3 .6B,D), the percent suppression of
key I responding is unrelated to the key 2 reinforcement rate but remains an
increasing function of the key 2 response rate. In spite of the scatter in data
points (each dot represents the rates for only one I-min period), the key I rate
is minimally suppressed only when the key 2 response rate is low, but is maxi-
mally suppressed only when the key 2 response rate is high. The formally con-
founded effects of key 2 response rate and reinforcement rate upon the key I
suppression is clearly apparent only at the level of mean comparisons. An in-
verse relationship between concurrent responses was obtained, however, for
overall as well as local data analysis. These data therefore support a model
based on response interaction, rather than response independence, of concur-
rent performances.
Chapter 3: Concurrent Schedules 179

100 A C/ll00 ':r.

T
... '!' B

fI
..J
W « !
~ 80 ~ 80

I
0
a.
C/l
W 60
Z
-SO ~ !
0::
~ ~
40 ~ 40 ~
> .i.
W C/l
~ 20 ~ 20
lL. ':
o O+--,,,-----.----r > 0
z 0.0 Q.33 0.5 lD t5 W 0 2 3
o MEAN KEY 2 REINFORCEMENT
~
KEY 2 REINFORCEMENT RATE
iii (REINF/MIN) IN TRIALS
C/l
W
RATE (REINFIM IN)
t5 D
!tl00
a. c ZI00
o .: : /:..:-.:'.:};'; ~ ':" '.
:::I
C/l 80 gj 80 . ':.: .~. .. .'
~
:. :':~}:f :~;~. '.:':"
• • •: " .. " 0 •• 0" "0
I-
Z
w 60 & 60
~ i7l . ::: ..., ••...1..:.... .1:
.. ...
0
• .-

~ 40 I- 40 0" ".' i::..!.. . " ..• '

Z
::.' .,,':
..... ;::... ',"! : •
z
~ 20
W
~ 20
w
. '::0 I.:,' ..
~
O~~~~~~~-r~,-~T a. Oi-~r-~~~-r-r'-,-~~ro-'~-r-r
o 10 20 30 40 50 60 o 10 20 30 40 50 60 70 80 90
MEAN KEY 2 RESPONSE RATE KEY 2 RESPONSE RATE (RESPIMIN) IN TRIALS
(RESP/MIN)
Figure 3.6. Percent suppression of ongoing key 1 responding. A and B: As a function of the con-
current key 2 reinforcement rate. C and D: As a function of key 2 response rate. Average data
(panels A and C) and raw data from individual trials (panels B and D) are presented. Data are from
the last 10 sessions of Phases A, B, and C for pigeon A2. To avoid crowding of data, the horizontal
line in panel D represents the absolute range of key 2 response rates during trials with 100% sup-
pression of key 1 responding.

W
I--
:::l 28 •
Z 24 •
..
~
Al 20 A2 20 A4
0: 16 16
W
a... 12 12 •• 12
• •

- . .
C/'l 8 8 8
0: 4
w 4 4 •
>
0
w 0.5 0.6 0.7 0.8 U9 1.0
<!>
z (U41 (U31 (all (UII (UO)
<
J: RELATIVE RESPONSE
U
RATE K:~i+~EY 2
Figure 3.7. Relationship between the mean rate of changeovers and the mean relative key 2 re-
sponse rate. Means are based upon the last 10 sessions of each phase. Eight data points are pre-
sented for each pigeon, corresponding to VI 60, 40, 120, and 40 sec scheduled during white trials
plus four replications of VI 180 sec during green trials. Relative response rates are plotted as sym-
metrical values around 0.5. A relative response rate ofO.2, for example, is plotted at 0.8. Triangles,
<0.5; circles, ",,0.5.
180 Section II: Operant Conditioning Procedures

As a further analysis, the mean changeover rates are shown in Figure 3.7 as
a function of the relative key 2 response rate. The changeover rate was highest
when the relative response rate was approximately 0.5 and then decreased as
the relative response rate approached the extremes of 1.0 or 0.0. An example of
the symmetry between changeover rate and relative key 2 response rate for in-
dividual trials is presented in Figure 3.8.
Previous reports suggest similar relationships between the changeover rate
and relative response rate for continuously operating concurrent reinforcement
schedules (LaBounty and Reynolds, 1973; Schneider, 1973).
The discriminative control of changeover responses and the interresponse
relationships are perhaps best illustrated by sample event recorder segments
(Figure 3.9). During intertrial intervals, key 1 responding occurred at a uniform
rate, only interrupted by feeder operation (a). With key 2 reinforcement rate of
1.5/min, an immediate changeover from key 1 to key 2 occurred at trial onset
(b). Key 2 responding then occurred at a high uniform rate, again only inter-
rupted by feeder operation (c), and key 1 responding was simultaneously sup-
pressed throughout trials (d). With a different key 2 reinforcement rate of
1.0/min. a changeover to key 2 responding also occurred immediately at trial
presentation, and key 1 responding was suppressed during trials for pigeon Al
(e). For pigeons A2 and A4, the changeover to key 2 also typically occurred at
trial presentation (t), but with intermittent changeovers back to brief bursts of

• •
48
w
....
~40
Z
~
a:: 32
~
(/)24
a::
lLJ
e5lLJ 16 GOO
o 0
o
•• • ••
•
~8
~
U
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
RELATIVE RESPONSE
RATE K1~2K2
Figure 3.8. Relationship between changeover rate and relative key 2 response rate during individ-
ual trials with O.5/min (black circles) and 0.33/min (white circles) key 2 reinforcement rates for pi-
geon A4.
Chapter 3: Concurrent Schedules 181

PIGEON Al PIGEON A2 PIGEON A4

KEY 2 REINF/MIN : lS

",.' . "..,.. " ........

3---------, ..mm""
~~ir··~dw-C·Nu.im
~"''"' :t·c'··- ,:,"" .".':'"' "-"'. '
lD

,"";;"".........
------~-----~
nn.[,.
--------~----~

'".,i'd "f'''FIi
.. ..
1 ,,,' •••• ' ..' .... I t I i • ......._ ••••• i i ; ' ; ' ; = .. I "
" 'lil
,i ,""'og" , ",,,, I

•
_hllllN' ''M, hi ' " II

Q5
1- - - - - - - ,
2"""""',, .... , ,OJ,,' in',.,.=,
..
''''jlII II f i ,. . . . . . , ....... I _ ... iiI ........ ' .. ' • "_, ..... , " ' " . 'hl
3- - - - - - - - - -•••r-~ ...~,~.
,~".~.~
,," 'I' ,,,.. , i .no h' i ""'''' , , ,
h I

0.33

ftl"Ir,.,•• fa ;'••
~~--'-------~
=+, ............... "'" 11.4""" 44. i it ;Wi • h Nil ji II. if ••"iPi'i "'i ii' hHi Ii "'i1i" ' ........ '•• i
~ "' ''!'''' k '(" " ""iii'"'
m
,II iii '

--lmin--

Figure 3.9. Sample event records for pigeons AI , A2, and A4 (left to right) of key I and key 2
responding during trials and intertrial intervals for each key 2 reinforcement rate and pigeon. Pen I
was deflected during trials and reinforcer delivery. I, trials or reinforcements; 2, key I; 3,
key 2.

key 1 responding (g). The third key 2 reinforcement rate of 0.5/min produced
far more frequent changeovers back and forth between the keys (h). With the
lowest key 2 reinforcement rate of 0.33/min, changeovers to key 2 were some-
what delayed, and usually occurred within the first 0 to 10 sec after trial onset
(k). The key 1 rate was only moderately decreased during these trials, with sev-
eral key 1 response bursts typically interspersed between a single or a few key 2
responses (1).
In a recent review of concurrent schedules, deVilliers (1977) suggested that
relative response rates can be used to quantify the underlying reward value of
different reinforcement conditions. Opposed to this view is a previous sugges-
tion by Skinner (1950) that preference or choice merely lies in changing over
between responses. With the relationship between changeover rates and rela-
tive response rates obtained in the present experiment, an intermediate relative
response rate is associated with frequent changeovers or choices, whereas a
high or low relative response rate occurs with few choices. Therefore, the rela-
tive response rate may be simply related to frequency of choice (e.g., change-
overs) rather than unequivocally measuring "value" of choice outcome.

Experiment V
If a response model of concurrent performances is accurate, then one response
should be affected by alterations in concurrent response-reinforcement rela-
tions only to the extent that such alterations change concurrent responding.
The purpose of Experiment V was simply to investigate the changes in an ongo-
182 Section II: Operant Conditioning Procedures

ing response after alterations in concurrent response-reinforcement relation-

ships. In principle, the previous results (e.g., Experiment III) suggest that an
ongoing response may be interrupted during a stimulus only if that stimulus
controls the emission of a concurrent response. This relationship was further
investigated in the present experiment by occasionally presenting either one of
two discrete stimuli during sessions of FR reinforcement. During one stimulus,
concurrent reinforcement was contingent on a concurrent VI response. During
the second stimulus, however, concurrent reinforcement was contingent on not
emitting the concurrent response. The rates of delivered concurrent reinforcers
were exactly equal during the two stimuli, which nonetheless differentially con-
trolled high and low concurrent response rates. According to the reinforcement
model, the ongoing response should be equally interrupted during the two stim-
uli because the concurrent reinforcement rates are equal. According to the re-
sponse model, the ongoing response should in contrast be interrupted only dur-
ing a stimulus controlling changeovers to a concurrent response.
The experiment used two male and one female Mrican green monkeys main-
tained at 21 hr of food deprivation prior to each daily session. The monkeys
were trained to respond on two levers that could be depressed simultaneously
using the same equipment as in Experiment II. Mter pretraining, left lever re-
sponding was maintained on FR 80 throughout sessions. Each lever press was
followed by response feedback of a 0.1 sec darkening of the lever and a high-
pitch tone. Phase A analyzed the discriminative control of changeovers from
left to right lever responding. Therefore, right lever responding was maintained
on a two-ply discrete-trial procedure (or mUltiple schedule).
For most of the session, the red and green stimulus lights above the right
lever were off and right lever responding was not reinforced (EXT). Occasion-
ally, the red light was turned on for 100 sec and right lever responding was rein-
forced on VII min. Occasionally the green light was turned on for 100 sec and
right lever responding was then reinforced on a variable-time schedule with a
differential reinforcement of other behavior contingency (VTDRO ) for 1 min.
During green trials, food pellets were merely delivered on the VT schedule pro-
vided right lever responding was at low or zero rate. The DRO contingency was
fixed at 5 sec and was introduced to force right lever responding to a low or
zero rate during green while maintaining the rate of pellet delivery similar to
that during red. Two red and green trials occurred in each session and were
always separated by a variable intertrial interval (no stimuli) associated with
extinction of right lever responding.
A complementary procedure was used in Phase B, with discriminative con-
trol of the reverse changeover, from right to left lever responding. The schedule
for the right lever was consequently altered so that VI (red stimulus) was in
effect for most of the session with occasionall00-sec trials of either EXT (the
red light turned oft) or VTDRo (the red light turned off and the green light turned
on). Essentially, VI and EXT (and their associated stimUli) were reversed from
the previous phase. Each phase lasted 15 sessions, and each session terminated
after delivery of fifty I-g food pellets. The schedules were procedurally inde-
pendent so that responding on one lever did not affect the response-reinforce-
ment contingency on the second lever.
 MONKEY VI _ •• _11_. __.._. ____ .. YTd,o _._ ~
PHASE ~. - - - _ _ I _ _::. •• _ _ _ _ '.!":- - 4 • •• ,' I,

A ,
5,
d

EXT V\t,.
_ • U _ ' _ .lI_ . _ _ _ _ _ I, _ _ i_ lIi _ _ _ .n_"'•• Jllliil I J ,l i D_
'i _"Z [_ ~." .... __ . _ .""" W_riit1 Ii ...... ',ik" _ 'MdIII
B

MONKEY 3 Vi VTdro
1,_ " ' _ , • • • ...------T 1 I rJ----.--..--l - . ~D_ 1 J _ _ . ._ '_ ••• _i'.....'TIJW"..-.--u ••••••• n . .

EX T V TdlO
"8 , j , Ji • • _ JIM _ , . I "I I .. _ Il I . . . _ · rrTT 1 J r I
iii II .. hi ... ... " . . . . . . . . . " i .. . . . . . .i t. . . . . . . . . . . ~ 4.. . ••• h i l i . . . . . . . . . , • •
II.i. -_-,..,......,...
,. .... "_ ••••

MONKEY 4 VI V'dro
11161 ' i. - ' JIi J ..., _ iMAM . ,- - ,' .. _ 11 1011 __ 11 ,"11 - . . 1 " 1,_.1) ••,__,11 ,. , 1 _ _ 1i r~

Ext V'd,O
• • 11 . . . . . . . 11.111. 1)1' _ _ I,' iiiJD, _ iI Jll, _ .,, tI,8l iT ---.-,.. ...rl rr
I
, i' i.____ ill .. _ • _ • • i ,i ---w--n

- -- - 100 SiC - - - -

Figure 3.10. Event records showing the interactions between FR responding on the left lever and VI responding
on the right lever. For each monkey, the upper record is from Phase A with VI and VT DRO trials, and the lower record
is from Phase B with EXT and VT DRO trials . The schedule for right lever responding is indicated for each trial. 1, Left
lever; 2, right lever; 3, pellet for left lever; 4, pellet for right lever; 5, trial.
00
v.>
184 Section II: Operant Conditioning Procedures

Figure 3.10 presents sample event records under the two procedural condi-
tions. In Phase A, onset of VI trials controlled a changeover from the left to the
right lever immediately (a) or shortly after completion of the ongoing ratio run
(b). The response rate on the right lever was typically high during the discrete
VI trials, with only occasional changeovers back to bursts ofleft lever respond-
ing (c). Presentation of the red light then led to a radical suppression of left
lever responding along with the almost exclusive emission of right lever re-
sponding. During VTDRO trials, on the other hand, right lever responses were
infrequent and the pattern of left lever responding was unchanged relative to
intertrial intervals, except for brief interruptions after concurrent VTDRO rein-
forcement (d). However, for monkey 4, left lever responding was intermittently
interrupted during VTDRo trials (e).
With the reversed discriminative control of changeovers in Phase B, right
lever responding decreased at the onset of EXT or VTDRo trials, with a simulta-
neously marked increase ofleft lever responding for monkeys 1 and 3 (f amd h).
For monkey 4, right lever responding also decreased during trials, but left lever
responding did not simultaneously increase. Instead, changeovers occurred to
idiosyncratic responses such as exaggeratedly turning the head upward or left
lever responding at insufficient force to activate the recording system (g).
These behaviors were rarely seen during intertrial intervals. Consequently, re-
corded left lever responding did not typically increase during trials relative to
intertrials for this monkey (i).
The overall schedule effects of Phases A and B are summarized in Figure
3.11. A stimulus signaling increased concurrent reinforcement rates may inter-
rupt an ongoing response only when that stimulus also controls a changeover to
a concurrent response. Furthermore, a stimulus effecting a changeover away
from a concurrent response may control a simultaneous changeover to and
hence increased rate of a specific ongoing response, in spite of maintained con-
current reinforcement rates. The latter relationship is deliberately expressed
with some caution, as a changeover from one response would not necessarily
result in a changeover to only one particular response. For monkey 4, for ex-
ample, the stimuli controlling a decreased rate of right lever responding in
Phase B simultaneously controlled changeovers to observed but unrecorded
concurrent responses in addition to the ongoing left lever response. The overall
data seem to establish that an ongoing response rate will be decreased by con-
current reinforcements that generate increased rates of concurrent responses.

Discussion
The present series of experiments systematically examined the concurrent re-
sponse analysis previously developed for the study of behavioral interactions
generated within combinations of operant and classical conditioning proce-
dures (Chapter 2). The application of the discrete-trial analysis to concurrent
operant schedules significantly established that the interruption of an ongoing
response was related to the changeover to a concurrent response, rather than to
a changed concurrent reinforcement rate.
Chapter 3: Concurrent Schedules 185

PHASE: A B
INTE~TRIAL

,i VI EXT VI VT.

/I
3

We

:~ >I~ II ( I~ I ~ >I:; II ~ I~ I
~E
o::~
~tI
z=
-tl
WQ,
0:: ......

4
200

100 \ '- ........ -.

/
'" '"
Figure 3.11. Relationship between the mean response and reinforcement rates for each phase. For
each monkey, the two left-hand columns are from Phase A, and the two right-hand columns are
from Phase B. Response rates during intertrial intervals and during trials are given as the left and
right data points within each column. The first column for each monkey, for example, compares the
rates during intertrial intervals (EXT) to the rates during VI trials; the next column compares the
rates during intertrial intervals to the rates during VTDRO trials. Black circles, left lever; white cir-
cles, right lever.

The results seem to be consistent with previous research using continuously

operating concurrent schedules. With a variety of parametric manipUlations,
the rate of an ongoing response may thus be inversely related to the rate of a
concurrent reinforced response without a simultaneous "match" to reinforce-
ment rates (Chung, 1%5; Chung and Herrnstein, 1967; ShuUand Pliskoff, 1967;
Brownstein and Pliskoff, 1968; Carlson and Aroksaar, 1970; Herbert, 1970;
Moffit and Shimp, 1971; Todorov, 1971; Katz, 1973; Schroeder, 1975; Deluty,
1976a).
The response model would in fact emphasize that suppression of an ongoing
response is importantly dependent upon a changeover to concurrent responses.
186 Section II: Operant Conditioning Procedures

Likewise, enhancement of an ongoing response may be dependent upon a

changeover away from a concurrent response to the ongoing response. How-
ever, the response model certainly does not deny a relationship between the
rate of one response and concurrent reinforcement rates; rather, this relation-
ship is considered to be indirect rather than direct. The ongoing response rate
may then more broadly be affected by changes in concurrent reinforcement
rates by virtue of simultaneous changes in concurrent response rates. At the
level oflocal performances, we clearly saw that the inhibitory influence of con-
current reinforcement closely coincides with emission of concurrent responses.
The pattern of one response was then typically in close inversion to the pattern
of a concurrent response-from the minute interruption associated with a brief
burst of a concurrent response to the more radical suppression associated with
the steady emission of a concurrent response. Indeed, a response may be "in-
hibited" by concurrent reinforcements only during periods of emission of con-
current responses.

Observing Responses with Signaled Reinforcement

(Experiments VI, VII, and VIII)*
Given the response model, the interest centers on the special procedures hith-
erto employed to demonstrate reinforcement interaction in the absence of re-
sponse interaction. In particular, the procedure of signaling concurrent rein-
forcement has been used to examine response competition in concurrent
schedules. The data from the signaled reinforcement procedure have been cited
as critical evidence in support for a reinforcement interaction as opposed to a
response interaction interpretation of concurrent performances (Catania, 1966;
Herrnstein, 1970; Rachlin, 1973).
In the typical concurrent VI VI experiment, the effects of manipulating con-
current reinforcement parameters are usually confounded with the simultane-
ous changes in the rate of the concurrent response. Thus, as the reinforcement
rate is increased for a concurrent response, the rate of that response also in-
creases. Any change in the ongoing response could be attributed to either the
increased rate of concurrent responses or concurrent reinforcers. To prevent
the concomitant variation of response and reinforcement rates, Catania (1963)
designed an experiment in which concurrent reinforcement could be manipu-
lated while maintaining a very low concurrent response rate. Employing a
changeover-key procedure (Findley, 1958), Catania changed one of two con-
current VI schedules to a signaled VI; that is, one VI was continuously in ef-
fect, and occasionally a second, signal key was lit. A response on this key dur-
ing the signal switched the main key from VI to continuous reinforcement for
only one reinforcer delivery. The experiment showed that the response rate on
the main key was inversely related to the rate of signaled reinforcement. As the
* Experiments VI and VII were conducted at the University of Copenhagen, from January to
November 1974, and reported by I. Iversen (1975b); Experiment VII was conducted at the Univer-
sity of Copenhagen, from September to December 1977.
Chapter 3: Concurrent Schedules 187

procedure required only a single response for each signaled reinforcement, the
results were said to show that the response rate decrease on the main key was
determined directly by the signaled reinforcers and not by response interfer-
ence.
Subsequently, Rachlin and Baum (1969) argued that the suppressive effect of
signaled reinforcement might have come about by changes in the rate of unmea-
sured responses, such as observing and orienting toward the stimulus signaling
availability of the concurrent reinforcer. Rachlin and Baum then hypothesized
that competing observing responses toward the signal source could be mea-
sured by latencies between signal onset and the response to the signal. Rachlin
and Baum repeated the experiment by Catania (1963), with a somewhat simpli-
fied procedure and with a manipulation of duration rather than rate of signaled
reinforcement. A VI was continuously assigned to one key. A second key was
lit on a VT schedule, and the first response on the second key during the signal
produced a signaled reinforcer. The results showed that the VI response rate
was inversely related to the reinforcer duration on the signal key. Rachlin and
Baum also suggested that the decreased VI response rate was not a result of
increased observing responses toward the stimulus source since the response
latencies to the signal key were uncorrelated with the VI response rate. How-
ever, whether observing responses actually occurred toward the signal key was
not reported. The data thus confirmed the previous position that the ongoing
response is directly influenced by concurrent reinforcement rather than by con-
current responses.
Catania (1969) then replicated the procedure by Rachlin and Baum, with the
exception that the VI key was darkened whenever the concurrent signal key
was lit. This procedural change would make observing responses redundant,
and thus reduce the likelihood of observing responses toward the signal key as
a competing response. (In Catania's experiment, however, the VI key was also
darkened during reinforcer deliveries. The dark VI key therefore did not reli-
ably indicate signal key illumination.) The results again confirmed the findings
of the previous experiments. Catania did not report, however, whether the ex-
perimental procedures actually prevented the development of observing re-
sponses toward the signal key. The relationship between an ongoing response
and concurrent reinforcement was then generalized to propose that the rate of
an ongoing response is directly inhibited by reinforcement of a concurrent re-
sponse.
Using similar procedures, Catania and Dobson (1972) suggested that re-
sponding might have been interrupted frequently by observing responses but
reported that visual inspection did not reveal any consistent head movements
specific to the schedule component with signaled reinforcement. Catania et al.
similarly argued that "the effects of signaled reinforcement might be attributed
to observing responses . . . , but it is not plausible that such looking would
consume as much time as the movement between the two keys when concur-
rent reinforcement is unsignalled" (1974, p. 106).
In contrast, a recent experiment found that changeovers between concur-
rently reinforced responses appeared to be associated with high rates of ob-
188 Section II: Operant Conditioning Procedures

serving responses toward the discriminative stimulus (Iversen, 1975a). The

sheer observation of any signal-observing response with a signaled reinforce-
ment procedure suggests that the previous experiments might profitably be rep-
licated with the additional recording and manipulation of observing responses.
The issue of response versus reinforcement interaction might be considerably
clarified by a direct and quantitative analysis of response interactions within
the signaled reinforcement procedure.

Experiment VI
This experiment simply investigated the extent to which an ongoing VI re-
sponse and concurrent signaled reinforcement might be associated with
changes in observing responses toward the signal source. More specifically,
two male homing pigeons were trained with VI on one key and a signaled rein-
forcement procedure on a second key. The pigeons had a previous history of
key pecking reinforced on cone VI FR. The experimental chamber had a floor
area of 30 x 30 cm and a height of 36 cm. Three 2.5-cm diameter keys with a
center to center distance of 5.0 cm were positioned 20.0 cm above the floor.
Pecks on the left key were reinforced on a VI 3 min with arithmetically distrib-
uted interreinforcement intervals. The left key was transilluminated by orange
light during sessions except during grain delivery for pecks on this key.
The experiment was conducted in three phases. In Phase A, only the VI for
left key responding was in effect. Signaled reinforcement was introduced in
Phases Band C. Consequently, the right key (10 cm from the left key) was oc-
casionally lit green. A single peck on this lit key produced grain delivery and
turned off the green light. The trials on the right key were distributed arithmeti-
cally on a VT schedule with a mean intertrial interval of 2 min in Phase B and 1
min in Phase C. The rates of signaled reinforcement on the right key were thus
0.0, 0.5, and 1.0/min during Phases A, B, and C, respectively. A session was
terminated after 50 min for Phases A and B but after 40 min for Phase C due to
the higher rates of grain delivery. Reinforcer delivery was 3 sec of access to
mixed grain. Phases A, B, and C lasted 18, 10, and 15 sessions, respectively.
To record observing responses toward the right key, a O.4-cm diameter pho-
tocell was positioned 3.0 cm below the center of the key. The chamber was illu-
minated by one white light in the ceiling, 8.5 cm from the left cage wall and
4.0 cm from the back wall of the chamber. The houselight was thus behind and
above the head of the pigeon when pecking the left VI key. Movements of the
head and upper neck to the right of the chamber midline toward the area around
the right key occluded the photocell and defined observing responses.
The times allocated to responses in concurrent schedules are typically mea-
sured as the cumulative time elapsing from a changeover t9 one response and
the subsequent changeover to a concurrent response. The indirect measure of
time allocation therefore includes long interresponse times (IRT) presumeably
devoted to unrecorded, not explicitly reinforced responses. Inclusion of such
unmeasured response durations as time allocated to a recorded response would
be avoided by the direct recording of response durations (Dunham, 1972). A
Chapter 3: Concurrent Schedules 189

second advantage of actual response durations is the applicability of units of

measure across behaviors of different topography, such as pecking one key and
observing toward a second key (Premack, 1965). The time spent pecking the
left key was defined as the number ofO.5-sec periods in which pecking occurred
on the left key. Only one time unit was counted if more than one key peck oc-
curred within the O.5-sec period. The time spent observing was defined as the
number of O.5-sec periods with the photocell occluded. For each session, the
probabilities (or relative durations) of key pecking and observing were calcu-
lated by dividing the number of O.5-sec units entered for each response by the
total session duration minus the duration of reinforcer delivery (Dunham,
1972).
The experimental procedures established that observing responses may not
only occur toward the signal source, but may also increase in probability with
increases in the signaled reinforcement rate (Figure 3.12). The increased proba-

Q30 A9

Q20

0.10
_0

>
~
:::; ---------0-
CD 0.0
~ 0.0 0.5 lD
m
0
0:::
a.. A10
0.40 ....0
w
I/)
z /CY"
0
8; 0.30 /
/

I&J /
0:::
/
/
0.20 /
/
/
I
/

OlO

0.0-+------.-------,
0.0 0.5 lD
Signaled reinforcements
per min

Figure 3.12. Relationship between the probability of pecking the VI key (black circles) and the
probability of observing the signal key (white circles) as a function of the signaled reinforcement
rate. Data points are means calculated from the last five sessions of each phase.
190 Section II: Operant Conditioning Procedures

bility of observing responses as a function of the signaled reinforcement rate

would suggest that observing responses may be systematically affected by
manipUlations of signaled reinforcement parameters. The simultaneously de-
creased probability of the ongoing response was at least correlated with an in-
creased probability of observing responses. Note that the VI response and the
observing probabilities remained unchanged for pigeon A9 when the signaled
reinforcement rate was increased from 0.0 to 0.5 reinforcements/min. Overall,
the probability of VI responding was then in fact more closely correlated with
the probability of observing than with the rate of signaled reinforcement.
The present data tentatively suggest an alternative interpretation of the sig-
naled reinforcement procedure. The decreased probability of the VI response
may simply have come about by the increased probability of observing re-
sponses toward the signal source.

Experiment VII

The purpose of this experiment was to investigate directly the influence of ob-
serving responses upon VI responding. The probability and pattern of observ-
ing the signal key was manipulated by inserting an opaque shield between the
signal key and the VI key, without simultaneously changing the reinforcement
parameters.
A second purpose was to measure the time allocation between schedule
components. However, the traditional measure as elapsed time between
changeover responses cannot be applied to the signaled reinforcement pro-
cedure. Because of the low number of changeovers to the signal key response
(only one for each signal presentation) and the brief duration of the signaled com-
ponent (from signal onset to the first response on the signal key), virtually all
of the session time is spent in the VI component. Hence, the time allocated to
VI remains relatively invariant across the range of reinforcement parameters on
the signaled VI. A similar argument applies to the relative response rate meas-
ure. Thus, neither time allocation nor relative response rate in VI can equal the
relative reinforcement rate with the procedure of signaled reinforcement.
Time allocation between schedule components has also been measured as
the time elapsed in either half of experimental chambers associated with each
response and schedule component (Rachlin and Baum, 1969; Bacotti, 1977).
Such a measure was used in the present experiment. Therefore, the experimen-
tal chamber, which was the same as in Experiment VI, had the following modi-
fications: a 2.5-mm diameter metal bar was fixed under the midline of the grid
floor, and the floor would tilt lightly whenever the pig~on moved from one side
of the chamber to the other.
To manipulate the probability of observing responses an opaque shield was
inserted between the response keys in Phases C and D. The O.OI-cm metal
shield protruded 4.0 cm into the cage at a right angle to the wall with the keys.
The shield was 12.5 cm long and was positioned with the lower edge 15.0 cm
from the floor. The shield thus extended 7.5 cm above and 5.0 cm below the
keys. Observations made during a preliminary experiment with one pigeon in-
Section II: Operant Conditioning Procedures 191

dicated that the insertion of a shield between the keys resulted in observing re-
sponses more directly in front of the signal key. A second houselight was there-
fore added to record observing responses with the altered topography.
Houselight 2 was positioned in the ceiling, 4.0 cm from the right cage wall and
4.0 cm from the back wall of the chamber. With the addition of the second house-
light, the photocell under the signal key would also be occluded by the pigeon
standing closely in front ofthe signal key. The two types of observing responses
are shown in Figure 3.13. In Figure 3.13A, pigeon A12 is standing in front of the
signal key; in Figure 3.13B, pigeon A13 is shown with an excursion of the head
away from the VI key and toward the signal key.
Experimentally naive homing pigeons were initially trained to peck on the
left key with reinforcement on VI 150 sec (arithmetically distributed interrein-
forcement intervals). Only the VI for left key pecking was in effect in Phase A,
and this established the baseline of photocell activation in the absence of sig-
naled reinforcement. The nonzero observing probability with no signaled rein-
forcement in Phase A resulted from the relative positions of the pigeon during
VI key pecking and collateral behaviors such as turning and bending motions.

Figure 3.13. Two topographies of observing responses recorded by the photocell under the right
key (signal key) following insertion of the opaque shield. Top, pigeon A12; Bottom, pigeon AI3.
192 Section II: Operant Conditioning Procedures

For example, during bursts of VI key pecking pigeon All stood along the floor
diagonal. The pulling back of the head and upper neck between individual
pecks on the VI key would then briefly occlude the photocell under the signal
key. A high rate of very brief photocell occlusions was therefore the baseline of
observing responses for this pigeon in Phase A.
In direct replication of Experiment VI, a signaled VI 150 sec was introduced
on the right key in Phase B. The probability of observing responses was then
experimentally increased by inserting the shield between the keys in Phase C.
Finally, the change in ongoing VI responding following extinction of observing
responses was studied in Phase D by preventing reinforcement after the signal.
Pecks on the right key then merely terminated the signal. Sessions were 40 min,

>
I-
...J
m
~
m
o
a::
CL.

w
(/')
z
oQ..
(/')
W
a::

KEY 1: VI VI VI VI
KEY 2: SVI SVI+ EXT+
SHIELD SHIELD

Figure 3.14. Probability of pecking the VI key (black bars) and probability of observing the signal
key (white bars) for each experimental phase. SVI: signaled VI on key 2. Data are means from the
last five sessions of each phase.
Chapter 3: Concurrent Schedules 193

and grain delivery was 3 sec throughout the experiment. Phases A, B, C, and D
lasted 24, 30, 12, and 12 sessions, respectively.
The probability measures generally provided both direct and systematic rep-
lication of the data from previous experiments (Figure 3.14). First, the VI re-
sponse decreased and the observing response increased after the introduction
of signaled reinforcement. Second, the VI response further decreased when the
observing response was increased by inserting the sheild between the keys. Fi-
nally, the VI response increased again when the observing response decreased
with extinction of the signal key response. The data then clearly show an in-
verse relationship between VI responding and concurrent observing responses.
The time allocation measures showed that the proportion of time allocated to
the signal side did not equal the proportion of reinforcer delivery for any pigeon
(Table 3.3). Furthermore, the marked change in the time allocated to the signal
side from Phase B to Phase C (when the shield was inserted) was not associated
with any alterations of the rates of reinforcer delivery. The rate of shifting
chamber side roughly covaried with the proportion of the time spent on the sig-
nal side. The proportion of time allocation then appears not to be an unequivo-
cal "measure" of the reinforcement value of the associated schedule compo-
nents.
The minute interactions between VI responding, observing, and standing on
the signal side are probably best described with reference to sample event rec-
ords (Figure 3.15). For all pigeons, the VI response pattern changed concomi-
tantly with the development of observing responses toward the signal key in
Phase B. The number oflong IRT on the VI key increased for each pigeon, with
virtually all of the long IRT associated with an instance of observing. Standing
on the signal side frequently occurred in synchrony with observing. The pattern

Table 3.3 Proportion of time on signal side, rate of shifting side (shifts/min), and
proportion of signaled reinforcement for each phase
Proportion Rate of Proportion
Pigeon of time on shifting of signaled
no. Phase signal side side reinforcement

All A 0.09 8.1 0.0

B 0.39 28.7 0.5
C 0.54 32.7 0.5
D 0.35 22.5 0.0

AI2 A 0.60 36.3 0.0

B 0.74 40.2 0.5
C 0.64 36.6 0.5
D 0.28 28.9 0.0

Al3 A 0.18 10.5 0.0

B 0.25 14.3 0.5
C 0040 23.9 0.5
D 0.17 14.0 0.0
Data are means of the last five sessions of each phase.
'0
.j>. VI A13
All A12
---w u---
_=,••11\11 ellNt '.11" It 8. *'1 It . " I. I' ILl" I I -.r II I • • • , I Niill Ii Ii iii i til II I • Ii iii" m itt Ii ti iii, ( I I n
2ihHiiM'''ii':i,,~M~I~.,,"'_111
__n.n.r...~~nn.~
'I. ilhIJ"I.JiI,if'11TII""U" hi ~ .... i.,1 .'fi¥"""'"...av-rTT
- u
3 I I j Ii .. ......,.",,-,.,....-:rT"W"OJ----r1..JI\..J"".-...w.,r-..r..II.IJT..."........J...J"'""'oI'T........r. ~--."-'-"--"~"""--
4~ VI SVI

;.J -----.,..J ~ ----------------------,~

21 [~ii!'lIriL[t fill hitH", (linUIIt lii\l~i IiUt TTTT""T""T-r-.---w--rr..- -- -;..- It l ' , 1 1 II I • rr-rnr-rr-rIr,\ " n In II U II m III III n : 111111 n .11 II

3~,J'-'V'..r~.II-I"TT1I'"",~ 1'T1IJJ'''''' J ".,.,. Ji'A""J1'M " •• '1' hunt,.- ~~

41- -,. -.-:. --. -.- :'':..1---:''-':'-.-:.-_T."".L:.--r.....n..---....n.-..-..-.",,-.-...,- ~'\tILo-.J.+:£'\I.J'.....r""'..1.lI-~, ..-..I..~--...u.aa.~~...u+Ll\.U' "'""'\..----....--,..r--..~...J'""'"""...____,._......"._....-...______---".-..----.. ... '"-r-_-.
5-----------------------
VI SVI.SHIELD
,. oJ l.J :r----'...r
21 it K, t. It:: , ii U " It t I II' III ILl • m I I." If' I I I I '" I 11 I T i l ' I ' (111 "!In --------;J
Iii till :l l l i:tt l t l l ' l l .
_.'. \i i Ii i I I i~
31r~~.M"lITII' .r\Q""1f1W..,.,......~-..rr--.6TVlV"'.r-JTT""'1'~~ :,r.r..r_'T""n.O",....-,.a.\.V.cJ1,.y"..,...,r.r;.r.a.TT.........-;r.n.TTTl-,..-......'\.vJ'..'\IirlIl}
~-...-.....rr_"'7
....-+".. ___.....__"T'"".....&-l'..t...-..--,......,._·..r. . . . .'l"4-...~.: r.-.-.. ~.:
. -:--_-..... -.__ .._.......... -.-.~.- ....._..~.~ ~r
. ____..-.. ....- .. ·."'l-.::.~ ••. ' •••......
-
4t-.. ."...l-[-'"-1O-...- -......,...- - ....- ""I......."'-'--:-~- .....\.."..L............":-.....---:--.:
5 .-- ------------------,------u---------- ~r------------------------------

VI EXT+SHIELD
w- ~~r----------------------
2. iii • IN 1111[1 I i . lin it i II It [i , U Hi II lim { • lit hut I 1 .- _I I ' I.. Wt •• • I • • • U "1ft 'n ...... 1 1URIW" H tllflI II till ne • ,.n _IT I ' t
3;.1151 lUI ii II Iii. mil ••1' ...... »N i,.1i i 'hi r - t - -...." ,,1 II j I ',I 11 i ji • » a j, ) 1111 Ii j » ,1 ,,, ~

4~~~~YV~rr~~vmnn~>n~~v- ~~,r-r

r 30 SEC
Figure 3.15. Segments of event records showing interactions between ongoing VI responding, observing, and standing on the signal side
for each experimental phase. I, reinforcement; 2, key pecking; 3, observing; 4, signal side; 5, signal.
Chapter 3: Concurrent Schedules 195

of VI responding and observing further changed when the shield was inserted
between the keys (Phase C). The duration of each observing response instance
then increased, and the duration of the concomitant IRT in VI responding in-
creased accordingly for pigeons All and A12. For pigeon A13, observing re-
sponse episodes decreased somewhat in duration but markedly increased in
number. At the same time the number of intermediate IRTs and VI responding
increased for pigeon A13. When pecking the signal key was no longer rein-
forced in Phase D, the observing response disappeared for all pigeons, resulting
in a return of photocell activation to baseline. The patterns of VI responding
simultaneously increased to approach those obtained in Phase A.
To provide an analysis of the reliability of the automatic recording of observ-
ing responses, the observing behavior was recorded by two human observers in
Phase C. An observing response was recorded by pressing a button as long as
"the pigeon looked or moved directly toward the right key behind the shield."
As an accuracy check, the pigeons were observed from the left, and the ob-
server could not see the exact moment ofphotocell occlusion (Figure 3 .13). The
observers also recorded turning behavior whenever' 'the pigeon made a tum in
whichever direction."
Figure 3 .16 presents sample event records of the automatically recorded be-
haviors and the behaviors recorded by the human observers. Recordings of ob-
serving responses from both observers precisely covaried with the photocell
occlusions for pigeons All and A13, whereas observing responses were occa-
sionally recorded by both observers without a simultaneous occlusion of the
photocell for pigeon A12. For all pigeons, photocell occlusions rarely occurred
without a simultaneous recording of observing responses from the observers.
Photocell occlusions and instances of observing responses recorded by the ob-
servers were thus closely covariant. The duration of individual photocell occlu-
sions was also closely covariant with the simultaneous observing durations re-
corded by the observers. Observing responses were frequently observed
simultaneously with the behavior of turning in circles. For each pigeon, one or
more instances of observing the signal key typically occurred during the turning
behavior.

Experiment VIII
To further establish the species generality of interacting responses with concur-
rent schedules, Experiment VIII examined the development of signal-observ-
ing responses in rats. Our previous emphasis was on affecting the ongoing VI
response by manipulating concurrent observing responses. This experiment ex-
amined the interdependency between the responses by instead manipulating
the VI response to change the concurrent observing response. First, observing
responses were developed with a signaled VI. Then a VI for lever pressing was
introduced to suppress the probability of observing. Third, to further suppress
the observing probability without changing the rate of VI reinforcement, the
probability of VI responding was increased experimentally by adding a differ-
ential reinforcement of high rate (DRH) contingency.
 196 Section II: Operant Conditioning Procedures

1~~------------------------------------
2.
II I Ii tt i it II , i Ii II i , Ii it li ({

3~~~rn~~~~~~~~~~~~~~
4 .....~,...J\..J"\.J'~~
5- '-I\.f'T...J")JJ"""\J"r~...n-rnnn-u'~JV""'V
6-----------~------~ .---~.---.---.r-------~----·---·--·--OBS EK'
I~~----------------w~r------------
2 ii' 8" • ti it i i t i , iii iii Ii i ti I i Ii i i i Ii iii itt Ii I i
~~_n.n.mr_.rllJTr.J'~~.J1.1"W".srnr
4 ~~~,...&J\IlJT\.J
5~.r-.rn-rr.......lr..tr,lT".rrrr.rv-.rr.r.rnrr.u-.r.-.rr.rrrrr.-..rr--rrrJ-rr"J'r.rJ"V"JV-rJ
6r. .__---~ , .. _-...r-..-_--v-~_-_-_-_-.--~----.,--------
OBS ll.
.
I~A~1~2~_________________________ ~~~-------~
2. ,. i ai • i it Ii Ii it 'Ii ii , Ii t Itl ii iti I Iii ttl Ii •

4~~-'-'-r~-r-v~~~-n~~~--~~~~r-~--~.--
5 1f"V'.nry--r---\..f'.r-~.r-; - 'r-t..r"""""'1r-' <I""'V".r_J""""TJ".rv--..-
6 __-.-----r-r--· - _ _- - - - -___--·--..-----_ ------------- . . - . - - - OBS EN.

1~~ ~~-----------------------------8_J
2t iii Ii • 11 Ii i i i Nil lit i i Ii { ti
3·",....TT--T.r~I1_.n~..J'Tl-I"l-~-'----nrr---)JY_vn....r__>_r.J>.~·l_".-:-V·T.....nn_r__;..r_
ii if( Ii hi . ii' t
i'
4~~~ __rw~~......r;~. .JV~,-~~_.~,_
5r~..J •
:..r---r...r-:~J""V"...I"Lr-~~
6 • _----.---__ ---_----_~-.--.-----_~ ~
OBS U.

II~A~1~3~______________~~r------------------------.~~----
2, (i t i i i Ii i ! i i i i i i t it i I II (I i i: i i i i 1 ( i i i ( It i 11'11 i
3~~HlJ""V")J"JV".r.rH'.r___v_l_f',JlJV"1r,.r...ru_............~.

~~.I1J"1J"V""J"'.r...---.r-.J"V"V""~.I"V.....'.t..,r"-v'n~"""'
08SEI(

--------------------------------............,~~,------------,~~---
2. iii I lit t t ,I I I I I t I1I1 I t (I' li t 1\ t I i i ii ih !
3-,.r...nnnrur~~""\.(""~..r..r..ntV".n.nn.~__

4 ~
5~.JV"""""""'.r_.rJ".J""\.r.r~.I"V1..nl",r~
ii--------·----~ ·-----------------------------------------=OB~S~lI~.

IS SEC

Figure 3.16. Event recorder segments showing VI responding relative to observing responses re-
corded automatically (A) and observing responses recorded by observers (B). Observers also re-
corded turning motions. Note that the recording speed is higher than for the records in Figure 3.15.
I, signal or reinforcement; 2, key pecking; 3, observing A; 4, signal side; 5, observing B; 6, turning.

Observing responses directed toward the source of signaled reinforcement

were analyzed in a modified Campden Instruments rodent test chamber. An
aluminum plate (0.2 x 8.0 x 8.0 cm) was centered on a Plexiglas wall opposite
to the lever and the food tray. This left a floor area of 21.0 x 21.0 cm. The sig-
nal was a blinking white light, with on-off phases of 0.15 sec, which appeared
Chapter 3: Concurrent Schedules 197

behind a 1.6-cm diameter hole centered on the aluminum plate, 9.0 cm above
the floor. In previous experiments, signaled reinforcement was contingent upon
emission of one peck on the key with the signal, the signal response. In this
experiment, nose poking to the signal source served as the signal response and
was recorded as a 3 0. -mm forward push of the 4.0-mm arm of a microswitch
recessed 4.0 mm behind the hole .
The observing responses were automatically recorded by a body-capaci-
tance sensitivity system adjusted to detect any activity within 1.0 cm of the alu-
minum plate. Direct contact with the plate also activated the sensing system.
This response served as an equivalent to the observing response recorded as
photocell occlusion in the previous experiments.
The additional response of entry into the food tray was recorded by an elec-
trical contact attached to the covering flap. Lever pressing, observing, signal
responses, and food tray entry were recorded in 0.3-sec bins.
The three male Wistar albino rats were maintained at 80% of their free-feed-
ing body weights and had a previous history of response-independent reinforce-
ment on a VT 1 min for 60 sessions. The reinforcer was a 45-mg food pellet and
sessions lasted 30 min.

PHASE: A B C D
VT SVI VI SVI V~rhSVI

0.30
RAT 1
0.20
0.10
0.00
>
t::
:d
m 0.50 RAT2
<{
m 0.40
0
0:: 0.30
a..
0.20
w
C/') 0·10
z 0.00
0
a..
C/')
l1J
0:: RAT 3
0.30
0.20
0.10
0.00

Figure 3.17. Probabilities of lever pressing, observing, and food tray entry for each experimental
phase. Data are means from the last three sessions of each phase. White bars, lever; black bars,
observing ; shaded bars, tray.
198 Section II: Operant Conditioning Procedures

Baseline probabilities of lever pressing, observing, signal responses, and

food-tray entry were first established with food pellets merely delivered re-
sponse independently on a VT 1 min for 10 sessions (phase A). Observing re-
sponses were then introduced by changing the schedule to a signaled VI 1 min.
The signal was turned on by the VT schedule and the first signal response pro-
duced a food pellet and turned off the signal (Phase B, 10 sessions).
This change to signaled VI dramatically increased the probability of observ-
ing for all rats (Figure 3.17). The topography of the observing responses in-
cluded sniffing the signal source without emission of the signal response, fixat-
ing the nose about 0.5 cm from the signal, or touching the plate with a forepaw
during sniffing at the signal. The signal response (activation of the nose contact
in the hole) rarely occurred in the absence of the signal and was clearly disso-
ciated from the observing response in the absence of the signal (Figure 3.18 and
Table 3.4). The observing response necessarily occurred in conjunction with
the signal response during the signal. These results lend considerable support
to the conclusion that signaled reinforcement gives rise to observing responses
directed toward the source of the signal.
The probability and pattern of the developed observing response was then
affected by increasing the concurrent probability of lever pressing in Phase C
(10 sessions). Lever pressing was first shaped with 10 extra food pellets in the
first 5 min of the first session, and thereafter reinforced on a VI 1 min. The
probability of lever pressing of course increased, but the observing probability
simultaneously decreased for all rats (Figure 3.17). The pattern of observing
also appreciably changed with the increased lever-pressing probability. Now
bursts of observing were shorter and more frequent, and they commonly oc-
curred in rapid alternation with a single or a few lever presses (Figure 3.18).
The probability and pattern of observing responses developed on signalled VI
are then clearly modifiable by concurrent lever pressing on a VI. However, the
increase in lever pressing with VI is formally confounded with the increase in
reinforcement rate of lever pressing (from zero to 1.0/min).

Table 3.4 Average signal response latency (seconds) and rate ofthe signal
response (response/min) for each phase
Phase
Rat
no. A B C D E F
signal latency 3.0 3.4 4.1 2.9 1.6
signal response 1.2 5.4 2.8 2.3 4.4 21.2

2 signal latency 2.1 2.8 3.1 2.4 1.9

signal response 1.0 2.1 1.1 1.5 1.6 6.0

3 signal latency 5.4 5.4 5.2 4.4 1.4

signal response 1.0 15.8 10.3 10.6 7.6 17.8
Data are calculated from the last five sessions of each phase, except for Phase F, for which the data are calcu-
lated from the last 10 min of the session.
Chapter 3: Concurrent Schedules 199

PHASE
A. VT
1-
2 - - -
"\.T"T~ --V--- r -
,....
3
4
5
B SVI
r r

C VI SVI

3 •Ii 'i I ",,, .... ,, .. i , i ...... II .... , ...... , , Ii """ J'TT"'""'.oi" II .. ii, "Ii i " ... , , ' Iii II • Ii ~li" " I I .,ii, i l l . ii, f' •• Ii I I i "ii' , ... .
j i If

-4 ~ j,,; I,;"".' ' •• "'i"T"I'~.rT'''''''-'-'"'''~'''''''''''''''''''''''-T'''....,T'I,',.,.,"'T.rr"....,.,..., ' ' ' "'T' TT'' ' ' ' ' ' ' '''''......,..,.~
5----r-------------~--~----~----------------~--------~_r_

o V rh SVI

......... I i Ii . . . . . . . aWi Wi I i ,ali ..... IHi'.~.".'

..... ..... Ie • • • , ...~ . . . . .

............
, ...........

~
I •

Figure 3.18. Event recorder segments for rat 2 showing the interactions between the recorded
responses for each experimental phase. 1, signal; 2, food tray entry; 3, lever/pellet; 4, observing; 5,
signal response.

The probability of lever pressing was therefore further increased without

any changes in reinforcement parameters by adding a DRH contingency to
lever pressing (Phase D, 12 sessions). The DRH was introduced gradually, with
a VI reinforcer contingent upon emission of two lever presses within 0.75 sec
(first session), three lever presses within 2.0 sec (second and third sessions),
and then four lever presses within 2.0 sec (remaining sessions).
The added DRH contingency increased the probability of lever pressing and
concomitantly decreased the probability of observing (Figure 3.17). Lever
pressing now occurred in brief bursts, which is typical of VI schedules incor-
porating a DRH contingency (Ferster and Skinner, 1957). The pattern of ob-
serving also changed from frequent short bursts to less frequent bursts of a
slightly longer duration relative to Phase C (Figure 3.18).
The results clearly extend the previous experiments in showing a direct
modification of both the probability and pattern of observing responses by man-
ipulating the concurrent probability and pattern of lever pressing. In a compli-
mentary fashion, increasing observing responses decreased the probability of
VI responding (Experiment VII), and an increase in VI responding decreased
the probability of observing (this experiment) without any corresponding
changes in reinforcement parameters.
200 Section II: Operant Conditioning Procedures

In previous signaled reinforcement experiments, interest focused on the sig-

nal response latency as a possible measure of the extent of observing responses
(Rachlin and Baum, 1%9). In the present experiment, the signal response la-
tency and the observing probability were not clearly related (Table 3.4 and Fig-
ure 3.17). Therefore, an increase in observing is apparently not unequivocally
associated with a decrease in signal response latency and vice versa.
However, the signal response latency may still be determined by the re-
sponse context. A more molecular analysis reveals that one can indeed predict
the signal latency from the particular response occurring at signal onset. The
event records presented in Figure 3.18 show long signal latencies when the rat
was in the food tray at signal onset, whereas the latency was quite brief when
the signal occurred during observing responses.
Table 3.5 gives the matrix of signal response latencies associated with spe-
cific responses at signal onset. The initial response (either lever pressing, food
tray entry, or observing) was noted at each signal presentation along with the
corresponding latency of the signal response. Short, intermediate, and long la-
tencies occurred with observing, lever pressing, and food tray entry as the
emitted response at signal onset, respectively. These results indicate that a sub-
sequent signal response latency may be predicted simply on the basis of the
response occurring at signal onset. For this reason, the extent of observing or
any other particular response cannot be predicted by the average latency of the
signal response. This analysis is particularly consistent with the predictions of
VI IRT from the initiating collateral response (Experiment VI in Chapter 5).
In sum, the results leave little doubt that observing responses do occur under
schedules that signal reinforcement. Perhaps more importantly for a concurrent
response analysis, the observing responses modify concurrent responses and
are themselves modified by alterations in concurrent responses.

Table 3.5 Signal response latency (seconds) with respect to the response occurring
at signal onset
Phase
Rat Occurring
no. response A B C

lever pressing 3.7 4.1 4.1

food tray entry 14.5 8.7 8.4
observing 1.4 2.2 1.1

2 lever pressing 3.4 4.4 3.1

food tray entry 7.7 8.0 5.9
observing 1.8 1.1 1.6

3 lever pressing 5.9 5.9 3.6

food tray entry 11.2 13.0 12.0
observing 1.7 1.8 1.3
Data are calculated from event records from the last three sessions of each phase.
Chapter 3: Concurrent Schedules 201

Discussion

The results extend the response model to the theoretically important signaled
reinforcement procedure by showing that the probability of an ongoing re-
sponse may be inversely related to the probability of observing responses con-
trolled by a concurrent schedule of signaled reinforcement.
The signaled reinforcement procedure is important historically because of
the apparent demonstration that an ongoing response is directly dependent
upon concurrent reinforcement rates. The argument is predicated on the fact
that the signaling procedure controls a very low response probability on the sig-
nal key, and the inhibitory effect upon the ongoing VI response therefore does
not result from' 'some kind of competition between the two responses for avail-
able time" (Catania, 1973, p. 518). However, can the absence ofresponses on
the signal key support the broader generalization that response competition is
absent in the signaled reinforcement procedure? More clearly, is responding on
the signal key really representative of all responses concurrent to the VI re-
sponse? This problem was in fact recognized in previous theoretical discus-
sions of the signaled reinforcement experiments, which clearly dismissed ob-
serving as a possible interfering response.
One argument raised against response competition was that observing re-
sponses would presumeably not consume as much time as the required move-
ment between the response keys in regular cone VI VI. Therefore, the similar
effect of concurrent signaled and un signaled reinforcement upon an ongoing VI
response could not depend upon response competition (Catania et al., 1974).
Whatever the relationship may be between the movement between response
keys and observing responses with different procedures, the present experi-
ments do demonstrate competition between VI responding on one key and con-
current observing toward a second signal key. In fact, the rate of moving from
side to side in the experimental chamber (Experiment VII) did increase with
introduction of signaled reinforcement, indicating considerable' 'movement be-
tween the keys."
A second argument against response competition is based upon the assump-
tion that the key peck latency "measures" the extent of observing responses
toward the signal key (Rachlin and Baum, 1969). Rachlin and Baum found no
consistent relationship between the signal latency and the VI response rate,
and concluded that changes in the VI response did not result from interference
from observing responses toward the signal key. A similar inconsistent rela-
tionship was obtained in the present experiments, with the response latency to
the signal and the observing probability negatively correlated in Experiment
VI, positively correlated in Experiment VII (Figure 3.19), but not clearly corre-
lated in Experiment VIII. Nevertheless, the present results cannot support the
additional assumption that observing responses therefore do not occur or can-
not interfere with VI responding. The results instead clearly demonstrate that
observing responses are in fact generated by signaled concurrent reinforce-
ment. The observing responses were furthermore systematically related to the
rate of signaled reinforcement and apparatus manipulations, and they were
202 Section II: Operant Conditioning Procedures

- A11
\5 0 A12
oA9
X A13
-A10

0<Il A ao B
.!! QO 0.1 0.2 OJ 00 0.1 02 OJ
>-
u PROBABILITY OF PECKING KEY 1
z
w
I-

/-;.
:5

c 0
OD 0.1 0.2 OJ 0.4 02 0:4 0.&
PROBABILITY OF OBSERVING KEY 2
Figure 3.19. Relationships between the latency of pecking the signal key 2 and: A,B. Probability
of pecking the VI key or C,D. Probability of observing key 2. Data from Experiments VI (panels
A and C) and VII (panels B and D) are shown.

directly recorded and observed rather than indirectly derived through other
"measures" such as the signal response latency.
Observing responses toward the source of a discriminative stimulus may
perhaps be a necessary outcome of the functional control of discriminative
stimuli. A changeover to the reinforced response controlled by the discrimina-
tive stimulus would seem to require a prior orientation of the relevant extero-
ceptor toward the source of the stimulus (Iversen, 1975a). The emphasis on the
necessity of observing responses is by no means unique or new. Spence, for
example, argued as follows:
The animal learns many other responses in addition to the final, selective approaching
reaction. Prominent and important among these are what have been termed, for want
of a better name, "preparatory" responses. These latter consist of the responses
which lead to the reception of the appropriate aspects of the total environmental com-
plex on the animals' sensorium, eg., the orientation and fixation of the head and eyes
toward the critical stimuli. That is, an animal learns to "look at" one aspect of the
situation rather than another because of the fact that this response has always been
followed within a short temporal interval by the final goal response. (Spence, 1937,
p.432)
More recently, Browne and Dinsmoor similarly argued that' 'in order to re-
spond differentially to stimuli in their environment, animals must learn to ob-
serve those stimuli, i.e., to establish contact between the stimulus energy and
the sensory receptors" (1972, p. 745).
Chapter 3: Concurrent Schedules 203

Clearly, the present data do not support the assumptions of response inde-
pendence and reinforcement interaction within the signaled reinforcement pro-
cedure. On the contrary, the probability of an ongoing response appears to be
inversely related to the concurrent signaled reinforcement rate precisely be-
cause the procedures generate specific observing responses that interfere with
the VI response. The present results therefore suggest a change in the emphasis
from a direct to an indirect relationship between the probability of an ongoing
response and concurrent reinforcement rates. Importantly, the prior claims of
reinforcement interaction and simultaneous response independence are thus in-
consistent with the present data. On the other hand, the single claim of re-
sponse interaction would appear to be consistent not only with the present data
but also with the previous effect described by the data giving rise to the rein-
forcement model.

Superstitious Responding with Response- Independent

Reinforcement (Experiment IX)*
The preceding experiments showed that a decrement in an ongoing response
after introduction of signaled concurrent reinforcement cannot be considered
an unequivocal demonstration of reinforcement interaction. A second proce-
dure of presenting concurrent reinforcers independently of responding has also
been used as an apparent demonstration of reinforcement interaction. Thus,
Rachlin and Baum (1972) reported that the rate of ongoing VI responding was
inversely related to the rate of concurrent reinforcers presented independently
of any response on a VT schedule. FR, Fl, or RI responding may also be de-
creased by response-independent concurrent reinforcement on fixed-time (FT)
or VT schedules (Edwards et aI., 1970; Deluty, 1976b; Lattal and Bryan, 1976;
Zeiler, 1976).
Within the controversy ofresponse versus reinforcement interaction, the ra-
tionale behind response-independent reinforcement is that responses are not
required for concurrent reinforcer delivery; therefore changes in the rate of the
ongoing response are said not to depend upon concurrent response rate
changes (Rachlin and Baum, 1972; Catania, 1973). The procedure of response-
independent reinforcement is thus logically equivalent to that of signaled rein-
forcement; both manipulate concurrent reinforcement rates with an assumed
absence of any changes in concurrent response rates. As with signaled rein-
forcement, this assumption can only be upheld provided response-independent
reinforcement does not alter or generate concurrent competing responses.
Clearly, response independence does not seem to hold for the signaled rein-
forcement procedures of Experiments VI, VII, and VIII. What about the re-
sponse-independent procedure? One suggestive feature of response-independ-
ent reinforcement is that responses of various topographies may increase in
rate in spite of the absence of any requirement between responses and rein-
forcer delivery; this is, of course, the well-known finding of superstitious re-
* Experiment was conducted at the University of Copenhagen, from February to April 1976.
204 Section II: Operant Conditioning Procedures

sponding (Skinner, 1948). Almost a century of research has similarly showed

alterations in responses of various topographies with classical conditioning pro-
cedures, genuinely characterized by an absence of response requirements (Pav-
lov, 1927; Zener, 1937; Kimble, 1967).
Applied to the procedure employed by Rachlin and Baum (1972), response-
independent concurrent reinforcement might then decrease the rate of an ongo-
ing response because of an increased rate of concurrent superstitious re-
sponses. Rachlin and Baum specifically argued, however, that superstitious re-
sponses could not explain the decrement of the ongoing response. Similar to
observing responses in the signaled reinforcement procedure, "other" re-
sponses were in fact admitted a possible existence but certainly no importance
in the response-independent reinforcement procedure. Rachlin and Baum thus
argued that "other" responses might be changed but would not affect the rate
of the ongoing response. Instead, response-independent reinforcement was
claimed to make the subjects "distractable," "that is, the procedures somehow
enhance the value . . . of sources of reinforcement (grooming, exercise, ex-
ploration, etc.) other than reinforcement -A and -B" (Rachlin and Baum, 1972,
p. 240). (The terms reinforcement -A and -B were used for the response-de-
pendent and response-independent reinforcements, respectively.) In essence,
"other" responses were defined as "other" sources of reinforcement and were
not affected by temporal contiguity with the response-independent reinforcers.
An opposite conclusion supporting the development of superstitious re-
sponding was reached in related research on response-independent reinforce-
ment effects in simple schedules. Zeiler, for example, concluded that changing
a simple schedule from response-dependent to response-independent reinforce-
ment "acts to make this particular competing behavior predominate, thereby
resulting in the rapid decrease in the frequency of emission of the original re-
sponse" (1971, p. 404). Similar conclusions were reached by Schoenfeld and
Farmer (1970), Lachter (1971), Lattal (1972), Davis et al. (1973), Alleman and
Zeiler (1974), and Buel (1975).
A considerable complication for the eventual resolution of the theoretical
opposition is that "other" responses have not been recorded with either proce-
dure. Since the relationship between "other" responses and response-inde-
pendent reinforcement is so essential, observation and recording of "other" re-
sponses would seem to be a natural prerequisite for the eventual solution of the
issue of response versus reinforcement interaction.

Experiment IX
One purpose of this experiment was simply to repeat the procedure of Rachlin
and Baum with the additional recording of such "other" responses concurrent
to an ongoing response. Second, a variety of experiments suggest that a critical
determinant of response rates is the frequency with which response-independ-
ent reinforcement is actually in close temporal contiguity with a particular re-
sponse (Davis and Bitterman, 1971; Lachter, 1971; Henton, 1972; Lattal, 1973,
1974). Applied to concurrent schedules, the effect of concurrent reinforcers
Chapter 3: Concurrent Schedules 205

might then depend upon how many of such reinforcers occur in close temporal
contiguity with "other" responses. A second purpose of the experiment was
then to assess more precisely multiresponse changes following manipUlations
of the temporal contiguity between" other" responses and response-independ-
ent reinforcement.
Two albino rats maintained at 80% of their free-feeding body weights were
used as subjects. They had a previous history of reinforcement of lever press-
ing on cone VI VI. The experiment was done in a two-lever Campden Instru-
ments rodent test chamber with the right lever removed and the hole covered
with a metal plate.
The rats were routinely observed through a one-way window in the sound-
attenuating chamber, and behaviors other than lever pressing and food tray
entry (movement of the flap covering the food tray) were recorded by the ex-
perimenter during the last two sessions of each phase. Exploration was re-
corded when a rat walked around and sniffed corners, walls, or the floors.
Standing (rearing) was recorded when the rat was in an upright position sniffing
the ceiling of the chamber. Finally, grooming was recorded during scratching
and licking of the body. A fourth response, being immobile, was only observed
and recorded in Phase G. For each response, the investigator pressed a button
for as long as the response occurred. Each response was recorded in 0.3-sec
units.
To establish an ongoing response, lever pressing was maintained on VI 1 min
in Phases A to E. Reinforcer delivery was always one 45-mg food pellet. The
effects of concurrent response-independent reinforcement were analyzed in
Phase B, in which a VT I-min was simply presented concurrently with the on-
going VI response. For ease of description and consistency with the Rachlin
and Baum terminology, pellets delivered on VI were termed A pellets, whereas
pellets delivered on VT or the subsequent modifications of the VT schedule
were termedB pellets. The programmers for A andB pellets ran independently
in all phases. Each phase lasted eight sessions and each session was 40 min
long.
The introduction of concurrent response-independent pellets in Phase B de-
creased the probability of lever pressing for both rats and simultaneously in-
creased the probability of some of the "other" responses (Figure 3.20). This
finding is in agreement with the previously reported effects of response-inde-
pendent reinforcement in concurrent schedules (Rachlin and Baum, 1972).
However, the previous argument (that the changes in VI responding were ex-
clusively determined by concurrent reinforcement per se and not by changes in
"other" responses) was based upon a logical interpretation of procedures other
than truly response-independent reinforcement. Rachlin and Baum added a
second procedure, in which the response contingencies were manipulated to
increase the possibility of temporal contiguity between" other" responses and
concurrent reinforcement. Their manipUlation was to never deliver concurrent
reinforcers within 2 sec of emission of the ongoing VI response. The VI re-
sponse rate was reported to be equally changed whether the concurrent rein-
forcers were response independent or delayed, in spite of the possibility of dif-
206 Section II: Operant Conditioning Procedures

RAT 1 RAT 2
PHASE: A B C 0 E F G ABC 0 E F G
70 0 10o 68 80 17 0 100 29 61

o.5 r-
0:0.1.
~o. 3
-
n n n
~o.2
> o.1- -
~
o.0
....J .
n nn n
iii ~o2
~
m
o~
0:0.1
~
o.0 n n nn n n -
~o2 -
a..
l£J
en
~o. 1·
°0.0
n n ,.., ,.., n n n ,.., ..... n n M n
z 0 03
oa.. ~O2
:;;0.1 nn
en
l£J
~
wD.4
a.0 M
- n

rl'
~o1.3-
~o2 ~BI -
><
Wo.1
o'".- n
PROB. 5 UM: 0.82 0.8£ 0.82 0.8£ 083 089 096
I
I nn n n
085 081 Q81 086 083 OB6 0.94
"II
I
I

Figure 3.20. Probabilities oflever pressing, food tray entry, grooming, standing, and exploration
for each rat and experimental phase. Percentages of the delivered B pellets preceded by a I.O-sec
pause in lever pressing are shown at the top of the columns for Phases B to F. Summed response
probabilities are shown at the bottom of each column. A new response, imobility, is added in Phase
G and is shown in the row for exploration. Data are means of the last two sessions of each phase.

ferential rates of "other" competing responses. Rachlin and Baum then argued
that the similar effects on the ongoing VI response must be directly related to
the similar rates of concurrent reinforcement rather than differential rates of
concurrent responses. The Rachlin and Baum manipulations were repeated in
the present experiment, but with somewhat different results.
Although B pellets were delivered response independently in Phase B, they
were in fact in close temporal contiguity with emitted responses. Thus some of
the B pellets happened to be temporally close to a lever press, with the rest of
the B pellets close to the "other" responses. The number of such contiguities
between "other" responses and a B pellet was defined as the number of in-
stances in which a B pellet was delivered in the absence of a lever press for 1
sec. An operate-reset timer remained on as long as lever pressing occurred
with IRT less than 1 sec. AB pellet delivered while this "response timer" was
off thus counted as a B pellet contiguous with "other" responses.
Chapter 3: Concurrent Schedules 207

The first step in the manipulation of temporal contiguity between B pellets

and "other" responses involved the simple prevention of delivery of B pellets
in contiguity with "other" responses (phase C). The procedure was modified
so that B pellets were now conjointly contingent upon the VT schedule and the
occurrence of a lever press within 1 sec preceding B pellet delivery. Simply, B
pellets were now only delivered when the response timer was on; that is, be-
tween zero and 1 sec of the emission of a lever press.
As a result, the fraction of B pellets closely contiguous with "other" re-
sponses was experimentally decreased to zero. The effect of this manipulation
was that the lever-pressing probability became the same as in the absence of B
pellets in Phase A (Figure 3.20). Introduction of concurrent reinforcement thus
reduces an ongoing response only when at least some of the concurrent rein-
forcers occur in close temporal contiguity with "other" responses.
To further pursue this relationship the possibility of close contiguity be-
tween "other" responses and B pellet delivery was increased to maximal in
Phase D. The B pellets were conjointly contingent upon the VT schedule and
the absence of a lever press within 1 sec preceding B pellet delivery. Thus, B
pellets could only be delivered when the response timer was off; that is, B pel-
lets were never delivered between zero and 1 sec of the emission of a lever
press.
With this optimal condition for the development of superstitious responding,
the probability of some of the "other" responses increased, and simulta-
neously the probability of lever pressing decreased relative to the truly re-
sponse-independent procedure in Phase B (Figure 3.20). The reliability of the
findings were then assessed by again changing the procedure back to truly re-
sponse-independent B pellet delivery (Phase E). The probability of lever press-
ing then increased to approximately the same level as in Phase B.
In the two final procedures, the influence of B pellets upon lever pressing
was further analyzed by first changing the operant schedule to extinction (no A
pellets were delivered, Phase F) and then totally preventing pellet delivery (nei-
ther A nor B pellets were delivered, Phase G). Lever pressing decreased, but
not to zero, by removing the explicitly contingent A pellets in Phase F. At the
same time, some of the "other" responses showed marked increases. The en-
tire organization of recorded responses changed with the total elimination of all
pellet deliveries in Phase G, with increases in immobility and grooming, and de-
creases in lever pressing, food tray entry, and exploration.

Discussion
Overall, the probability of lever pressing was rather precisely related to the
sum of the probabilities of the "other" responses. Note that the probability of
all recorded responses did not sum to unity. The sums ranged between 0.81 and
0.96, and were highest in Phase G. Although an attempt was made to record all
responses exhaustively, some fraction of the session time was not represented
in the measured response durations. Related analysis shows that differences in
the sum of recorded response durations may be directly related to changes in
208 Section II: Operant Conditioning Procedures

the additional set of changeover responses controlled by the different schedules

(Chapters 4 and 6).
One essential finding was that equal B pellet rates did not generate equal VI
response probabilities across experimental phases (B, C, D, and E). B pellets
simply had no decremental effect upon the ongoing VI response when none of
them were contiguous with "other" responses. Given the premise of reinforce-
ment interaction, equal rates of concurrent reinforcement should produce equal
decrements in an ongoing response regardless of how the concurrent rein-
forcers are delivered. A previous very crucial argument was that concurrent
reinforcement can be explicitly contingent upon concurrent responses, inde-
pendent of any response, either signaled or delayed with respect to a specific
ongoing response; the" source" of concurrent reinforcement should not matter
for the decremental effect upon an ongoing response (Rachlin and Baum, 1972;
Catania, 1973; Rachlin, 1973). The results of the present experiment clearly in-
dicate, however, that the probability of an ongoing response may in fact be dif-
ferently affected by equal rates of delivered concurrent reinforcers.
Recent reports provide substantial support for the present findings. Zeiler
(1976, 1977) found that the rate of ongoing FI, VI, or VR responding was rela-
tively higher when concurrent reinforcements were truly response independent
than when explicitly contiguous with pauses in the ongoing response (i.e., de-
livered on a DRO schedule). Similarly, for a previously reinforced response un-
dergoing extinction, the response rate may be higher when concurrent rein-
forcers are response independent then when contingent upon pausing in the
extinguishing response (Davis and Bitterman, 1971; Lowry and Lachter, 1977).
Therefore, the effects of concurrent reinforcements upon either an ongoing
response or a response undergoing extinction are not sufficiently described by
the mere overall rates of delivered concurrent reinforcers. The probabilities of
lever pressing and the summed probabilities of all "other" recorded responses
were instead clear functions of the percentage of all pellet deliveries contiguous
with the "other" responses (Figure 3.21). (Note that the abscissa does not give
the relative rates of obtained reinforcement, which were 0.5 in Phases B, C, D,
and E.) For both rats, the summed probabilities of the "other" responses and
the lever-pressing probabilities were simply increasing and decreasing func-
tions, respectively, of the percentages of all pellets contiguous with "other"
responses. Thus, across all phases, the higher the percent of pellets contiguous
with the "other" responses, the higher the probability of the "other" re-
sponses, and the lower the probability of lever pressing. The present data
therefore rather specifically indicate that the decremental or "inhibitory" ef-
fect of a given rate of concurrent reinforcement upon an ongoing response may
depend upon the fraction of the concurrent reinforcements being contiguous
with "other" responses.
The previous argument that an ongoing response is inhibited by concurrent
reinforcements independently of source (Rachlin and Baum, 1972; Catania,
1973; Rachlin, 1973) might, however, be formally preserved by arguing that
only those concurrent reinforcements that are contingent on or contiguous with
"other" responses can be classified as coming from another source. If that is
Chapter 3: Concurrent Schedules 209

RAT 1 RAT 2
0.8

0.7
> D
t-
:::i 0.6 A
iii B /
/

<{
m 0.5 / C
E
0
a: /
/

0.. 0.4 A.C

, (f
W
(/') A, d
z 0.3 E o
0 B C
0.. F
(/') 0.2
w
a:
0.1

0.0
0 20 40 60 80 o 20 40 60

PERCENT OF ALL DELIVERED PELLETS

CONTIGUOUS WITH OTHER RESPONSES
Figure 3.21. Lever-pressing probability (black circles) and the summed probabilities of "other"
responses (white circles) for each rat as a function of the percent of all pellet deliveries contiguous
with "other" responses. Letter for each data point indicates experimental phase.

the case, then concurrent reinforcements that are contiguous with the ongoing
response might be added to the reinforcements that are directly contingent on
that response. In essence, the inhibitory effect of concurrent reinforcement
would be smaller if some fraction of the concurrent reinforcements were con-
tiguous with the ongoing response. Applied to the present experiment, an ongo-
ing response would not be inhibited by the concurrent response-independent
reinforcements that "happened" to be contiguous with the ongoing response.
[The absence of an incremental effect by the addition of reinforcements in close
temporal contiguity with the VI response may not be empirically exceptional.
The function relating rate of VI responding to rate of reinforcement may be
roughly asymptotic at the reinforcement rates used in the present study (Ca-
tania and Reynolds, 1968). Therefore, the addition of reinforcements may not
necessarily increase the overall response probability].
Therefore, the functional effect attributed to concurrent reinforcement de-
pends upon the precise definition given to concurrent or "other" reinforce-
ment. If only reinforcements contiguous with "other" responses qualify as
concurrent reinforcement, then concurrent reinforcement is always inhibitory.
Conversely, if response-independent reinforcements contiguous with an ongo-
ing response qualify as concurrent reinforcement, then concurrent reinforce-
ment is not always inhibitory. However, with either definition, the present ex-
periment suggests that concurrent reinforcements decrease an ongoing
response to the extent that they are contiguous to and increase the probability
of concurrent responses.
210 Section II: Operant Conditioning Procedures

W1C2 L Wl L Wl L T L P

T L Wl L Wl L T Wl L Wl~P

3 ~L P

T L~L Wl L(!)L W~L W1C1L C~lL P

T L Wl L Wl L T L Wl L Cl L Wl L T L Wl L T P

T L P
T Wl L T Wl L Wl L Wl L T L F T L Wl L Cl L Wl C4 L Wl L Wl L Wl L Wl L Wl 5 T Wl L Wl -

L Wl L Wl L Wl C4 L Wl L G P

TGL T Wl L G L e T LeT L Wl L T WleLG)LQLG)T LG)L P

T LTWP

10 T L T ~~ Cl L Wl L Wl L Wl L Wl T L Wl L Cl C4 L Wl L Wl L Cl L C3 G P
11 T L T L Cl L Cl W2 L W2 C3 F L F W4 P

12 T L T L T Wl L W4 L T Wl L C4 5 F L Cl L T L Wl L T ~ P

13 T L C4 L T L C3 ~ 5 C3 ~ C4 5 F ~ L P

14 T L T L C4 L F L C4~C4 L~S~C4~C4 C2 W2 L~G F L~C3~F W2 L T~ C3-

5 ~ F L Cl C2 C3 ~ F G L P

15 T L Cl T L Cl L Cl L Cl L Cl ~ P

16 T L W T L Wl T L EL3 ~ Wl 5 C4 L ~ C4 L T ~ Cl L Cl L Cl T WW Cc
L T L T L Cl L Wl L Cl L Cl L Wl L Wl L T L Cl L Wl L Cl T Wl L Cl L Cl-

L Wl L Wl L Wl L P

17 T L T L P
18 T L T L Cl L T L Wl L C4 L Wl L Wl L Wl L Cl L Wl L Wl L Wl L Cl L Wl 5 F Cl L P

19 G!)p
20 ®P

21 T F L®~P

22 T F(l')~Wl L~ L~ ~Wl L~LG!)L P

23 T L P
24 TL~~P
25 T ~ Wl L ~ L ~ L Wl ® Wl L Wl L Wl ®~ Wl L Cl Wl L Wl L Wl Cl P
26 T (session stop)

Figure 3.22. Successive response sequences for rat I from the last session of Phase E. Each letter
represents one burst of a given response class (see text). Each sequence starts from the left with T
(food tray entry after pellet delivery) and ends in P (pellet delivery). A bar at the end of a row
indicates that the sequence is continued in the next row.

The above analysis was concerned with the overall or molar relationship be-
tween ongoing response and contingency of concurrent reinforcement. What
would be convincing evidence that concurrent reinforcements provide a source
for the development of "other" superstitious responses, and thereby alter the
probability of the ongoing response? Perhaps a mere increase in the overall
probability of some "other" response is not sufficient to label these responses
superstitious. Therefore, the present experiment provided a more molecular
analysis of sequential alterations in bursts of individual responses.
Chapter 3: Concurrent Schedules 211

T c 1 L c 1 c 2 c 3 G F L c 1 L c 1 L c 2 w3 c 3 w4 F c 4 L c 1 w2 L c 1 L w2 S L w 2 L w2 c 2 L-

w2 S L S c 1 L c 1 w2 S L 5 L C1 C2 5 G P

G W4 F
e
T F L T W1 L C1 T W1 L C1 P

T L C1 T W1 L C1 L C1 L <:1 T C1 L W1 L C1 L W1 L W1 C4 S W1 C4 p

T G T G S G S H2 F L C1 L C4 W4 C4 SI!!iIP

T ~ C4 E!I C4 S E§] C4 EQ T E§] S C4 S E§] S ~ L C1 C4 S C4 W3 S W2 C1 L C1 C4 E!j F E!j-

C 3 W3 C 3 ~ C 3 Cl L T C 4 S C4 S E!I C 3 C 2 L C1 W2 L W2 L W2 C 2 C 3 F T C4 C 3 -

C2 L Cl E!j C4 C2 P

T L Cl T Cl Wl C4 W1 C1 L W2 L C1 W2 C 2 C 3 (!)P

T(!)T Wl Cl L Cl L C l C2 C 3 (!)T(!)C 4 P

T C4 T ~ P

T ~ C4 1Sj T Cl W4 C 4 F L T W4 C 3 C 2 F C 2 G F L Cl F C4 C 3 C2 F C1 F W4 C4 C 2 W2 C 2 W2 -

e ee
LG)P

®P
e
10 TeL T C4 L W2
11 T T Cl L C l T L C1 L C l L C1 Wl C l W1 C 4 C 3 C 2 W1 F C1 F C l W2 C2 C 3 G F C1 W2 Cl W2 -

C1 L C l W2 C 2 W3 S C 3 C4 Cl W2 C 2 W2 S Ell P
12 T 5] C3 W3 F Cl W21S1 W3 C 3 W4 C 4 W2 C l W21S1 W3 W4 S F W3 F LIS] C 3 G F L F W4 C 3 1S] F-
S ~ W2 F C4 S L C l W21S1 W3 W2 ~ S W2 C l W2 S L C1 W2 Cl W2 Cl W2 C l W1 C 4-

Wl C l C4 S C4 S ~ P

13 T~ C3 ~ F T Cl C4 F Wl C 4 ~ F C 3 C 2 F W2 F Cl ~ C3 ~ S C 3 S c40p

tS) P
e e eee e
14 T Cl c 4 0 T C 4 Wl

15 T T C4 Wl T Wl L L W2 C2 C 3 F T C4 W2 C 2 ~ P
16 T C4 ~ C3 S ~ c20p

17 T(!)C 4 W4 C 3 (!)T CI C!)C 2 W3 C2 (!)C l (!)C 3 W3 S C2 W2 Cl W2 SC!)C 4 W4 C 3 C2 W2 S W2 Wl W4 -

C4 S C 3 C2 F Cl W2 C2 W3 C2 W2 C2 C 3 W4 F S Wl L Wl S W4 C4 W4 S C4 F C1 W2 S-

Cl Wl W2 C 2 C 3 W3 S W3 cl[!JP

18 T F[!JL Cl[!JW1[!]W 1 c412lW1 C 4 w1[!JW l S C4 [!JC 1 Wl S L C4 ~ P

19 T Cl L T ~ C4 ~ T ~ C4 ~ Cl ~ T C4 ~ C l T C4 C 3 F T C 4 ~ T C4 ~ S ~ C4 T 6l S L-
T~C1~T~ Cl Wl S C4 C 3 C2 C1 W1 C1 T C4 W4 C3 W3 C2 W2 Cl F W4 C 3 W3 -

C2 w2 C 2 W2 CI WI CI L WI C 4 C2 W3 C 3 W3 S W3 C 2 W2 C 2 W2 C 2 C 3 F C I W3 C 3 W3--
F T Wl C 4 W4 C4 S C 3 G0p
20 T Cl G)[E]P

21 TmG)m~m Cl ~m~m~[!]~ C4 ~ C l wllJ W1 C 4 Cl ~ (session stop)

Figure 3.23. Successive response sequences for rat 1 from the last session of Phase F. See Figure
3.22 legend and text for definitions.

Figures 3.22 and 3.23 present representative samples of successive interpel-

let response sequences from the last sessions of Phases E and F for rat 1. Simi-
lar data were obtained for rat 2. Each response sequence is identified by num-
ber (left-hand column) and shows all responses occurring between pellet
deliveries. The specific sequence of all emitted responses was recorded
throughout the session, and each possible response was identified by location
within the chamber. The recorded responses included: exploring or sniffing
each corner of the chamber (C l , C2 , C3 , and C4 ), each wall (W l , W2 , W3 , and
W4 ) or the grid floor (F), as well as lever pressing (L), food tray entry (T),
212 Section II: Operant Conditioning Procedures

grooming (G), and standing (S). [Comer one, Cb was between the wall with the
lever and food tray (wall one, WI) and the left-hand wall (wall two, W2 ); comer
two, C2 , was between the left-hand wall and the back wall, etc.] This method
recorded the number of bursts but not the relative duration of individual re-
sponses.
The last 25 sequences in the last session of Phase E are shown in Figure 3.22.
The first sequence (seq 1) consisted mainly of alterations between L and WI, L
and W2, and Land CI . T preceded pellet delivery in seq 2, and in the brief seq 3,
and then became more frequent in seq 4. (Examples of significant changes in
structural composition from one sequence to the next are indicated by circles
or squares around the last response in one sequence and the same response in
the next sequence.) The long seq 7 terminated in C4 preceding pellet delivery,
and C4 was immediately apparent as a dominant response throughout seq 8. (C4
was typically a very rare response.) In seq 9, WI L preceded pellet delivery as a
response pair and then dominated seq 10. Seqs 11 and 12 both terminated in
W4 , and W4 was a most frequent response in the relatively long seq 14. Then in
seq 15, the response pair WIL again preceded pellet delivery and appeared fre-
quently in seq 16. T terminated the very brief seqs 19,20, and 21 and was then a
very frequent response in seq 22 compared to many of the preceding se-
quences. T again terminated seq 24 and appeared frequently in seq 25. Note
that the mini sequence WILT that preceded pellet delivery in seq 21 was re-
peated in seqs 22, 24, and 25.
Similar response sequence changes were clearly apparent in Phase F. The
last 21 sequences of the last session are shown in Figure 3.23. C4 terminated seq
1 and then appeared in seq 2. In seq 2, the response pair LCI preceded pellet
delivery and dominated seq 3. In virtually all of the following sequences, the
last response preceding pellet delivery in one sequence not only occurred fre-
quently in the next sequence, but also was immediately emitted after pellet re-
trieval (after T); see seqs 4, 5, 7 through 13, 17, and 21. In the lastthree sequen-
cies, F terminated seq 19 and the response pair TF terminated the brief seq 20.
Both F and T entirely dominated the last seq 21. L was typically not as domi-
nant as in Phase E and was either entirely absent or emitted at low frequency
(see for example, seqs 13, 17, and 19 in Figure 3.23).
These examples clearly show that the particular response preceding pellet
delivery was often emitted again soon after pellet retrieval. Typically, the effect
was relatively transient but became more persistent, perhaps dominating one or
more successive sequences, if that response was followed relatively quickly by
a second pellet delivery. This clear incremental effect upon exactly the re-
sponse that preceded reinforcer delivery beyond doubt extends the principle of
superstitious responding to include response-independent reinforcement in
concurrent schedules. Skinner (1938), Fenner (1969), and Neuringer (1970) pre-
viously found that a very few pairings, even only one, between a given re-
sponse and reinforcement was similarly sufficient for a marked increase in the
frequency of that specific response.
In the present experiment, the relatively increased frequency of a given re-
sponse after temporal contiguity with reinforcer delivery appeared to supplant
Chapter 3: Concurrent Schedules 213

the previously dominating responses. In many response sequences, the proba-

bility of a previously infrequent response was locally increased after contiguity
with reinforcer delivery. In fact, a particular composition of response frequen-
cies could be radically altered by only a few contiguities between reinforcer de-
livery and a "new" response. The recorded "other" responses then did not
merely constitute a class of homogeneous responses passively filling the pauses
of the operant response. On the contrary, they participated in the formation of
the pattern of the operant response.
In summary, the present data do not provide much support for recent argu-
ments that question the development of superstitious responding with re-
sponse-independent reinforcement (Rachlin and Baum, 1972, Staddon, 1975).
More critically, the present results extend the response interaction model to the
hitherto theoretically important procedure of response-independent concurrent
reinforcement. The results are specifically congruent with the results from Ex-
periment V and with Henton (1972). These results also combine with the results
from the signaled reinforcement procedures of Experiments VI, VII, and VIII
and suggest that contiguities between concurrent responses and concurrent re-
inforcements are basic to concurrent schedule effects. In general, the inhibitory
effect of concurrent reinforcements may literally depend upon the extent to
which they control concurrent responses that compete with the ongoing re-
sponse.

Concluding Comments
The behavioral effects of the various concurrent schedules have proven to be
well described in terms of interactions between response. Briefly, an ongoing
response was characteristically interrupted during emission of a concurrent re-
sponse. The features of response interaction within combinations of operant
and classical conditioning procedures were easily replicated in the present ex-
periments. An ongoing response was interrupted only when stimulus presenta-
tion controlled a changeover to a concurrent response (i.e., when the stimulus
was a discriminative stimulus for a different response). Furthermore, the pro-
cedure of withdrawing rather than presenting a discriminative stimulus sug-
gested at least some conditions in which the elimination of a concurrent re-
sponse may be followed by a changeover back to and enhancement of the
ongoing response. Finally, the ongoing response was not suppressed when the
discriminative stimulus did not control a changeover to concurrent responses.
In previous investigations specific concurrent responses were physically
prevented in an attempt to assess whether response interactions are important
for concurrent schedules, either classical-operant or concurrent operant-
operant schedules. As in recent analyses of classical-operant combinations,
the present analysis also encountered the" discovery" of concurrent responses
and response interactions within the operant procedures hitherto throught to
disprove response interactions. The extensive response interdependencies sug-
gest that the central assumption of response independence cannot be upheld as
214 Section II: Operant Conditioning Procedures

a universal description of concurrent performances. Consequently, the empiri-

cal basis for a strict reinforcement model is rather weakened, and the model
may not provide a satisfactory description of concurrent performances.
The present results therefore confront the assumptions and predictions
based upon previous criticisms of the role of response competition for concur-
rent performances. Some of the more critical arguments are elaborated below.
Traditionally, changes in an ongoing response associated with the emission
of a concurrent response are, first of all, confounded with the reinforcements
for the concurrent response. This confounding is usually separated by a logical
resolution that only concurrent reinforcements and not concurrent responses
are responsible for the altered probability of an ongoing response. In the pres-
ent experiments, however, the probability changes of the ongoing response
were more clearly related to the probability of concurrent responses than to the
rates of delivered concurrent reinforcers. For example, in Experiment II, the
pauses in FR responding covaried with the manipulated burst durations of con-
current VI responding. This effect was perhaps even more clear in Experiment
III, in which an ongoing VI response was interrupted precisely during those
trials in which the extinguishing concurrent response still occurred, not during
trials without the concurrent response. Furthermore, in Experiment IV, an on-
going response was maximally suppressed during trials only when the concur-
rent response rate was high, and minimally suppressed only when the concur-
rent response rate was low, irrespective of the associated rate of concurrent
reinforcers delivered during trials.
The point raised by these examples is, of course, that response interactions
and reinforcement interactions are experimentally separated at the level of
local response changes. Moreover, in the case of interactions. involving a con-
current response undergoing extinction, any confounding of concurrent re-
sponses and concurrent reinforcements is obviously eliminated. Only an in-
ferred "inhibitory aftereffect" of previously delivered concurrent reinforcers
might provide a reinforcement explanation in the absence of actually delivered
reinforcers-and even then, only when the assumed decaying aftereffects of
previous reinforcers precisely follow the local pattern of the extinguishing re-
sponse. In all cases, the local rate and pattern of an ongoing response were
more closely related to the concurrent response rate than to the concurrent re-
inforcement rate.
The claim of reinforcement interaction was previously supported with dem-
onstrations of inhibition in the apparent absence of concurrent responses
within the procedures of signaled and response-independent concurrent rein-
forcement. These procedures were repeated in four ofthe present experiments.
Clearly, the concurrent reinforcements did in fact control concurrent re-
sponses, and the effects may come about by virtue of changes in concurrent
responses. Furthermore, the proportional rate of delivered reinforcers was not
sufficient to describe the probability of the ongoing response. In Experiment
VII, for example, the probability of the ongoing VI response decreased when
the probability of concurrent observing responses was increased, without any
changes in reinforcement rates. Similarly, in Experiments V and XI, equal
Chapter 3: Concurrent Schedules 215

rates of delivered concurrent reinforcers yielded different probabilities of an

ongoing response, and different rates of concurrent reinforcers generated equal
response probabilities. Therefore, reinforcement factors are not sufficiently
precise to provide a satisfactory description of response changes.
The primary question is whether concurrent reinforcements can have an in-
hibitory effect without reference to concurrent response changes. The very
close response-response correlations in the present experiments, even within
quite minute response patterns, simply suggest that the inhibitory influence of
concurrent reinforcements must somehow be related to the simultaneous oc-
currence of concurrent responses. Ultimately, the response model states that
an ongoing response is "inhibited" by concurrent reinforcements exactly and
only during the emission of concurrent responses.
Catania (1%9) has argued in a historical perspective that the concept of in-
hibition has been ambiguously applied because of failures to specify clearly the
inhibited and the inhibiting events. Catania suggested that the property of in-
hibition should encompass not only a reduction of an ongoing response but also
the agent responsible for the reduction. In the case of concurrent operant
schedules, the inhibitory agent was said to be concurrent reinforcements. This
now standard and often cited definition, that "concurrent reinforcements are
inhibitory" (Catania, 1%9), would seem to be imprecise, since the introduction
of concurrent reinforcements does not always lead to a reduction of an ongoing
response (Experiment V). When concurrent reinforcements do lead to a reduc-
tion, that effect stems from an increase in concurrent responses (e.g., Experi-
ments VI, VII, VIII, and IX). Response reductions may also be produced by
increases in concurrent responses experimentally dissociated from increases in
concurrent reinforcement rates (Experiment VII). In summary, the term in-
hibition is misleading if the inhibiting event refers to concurrent reinforcement,
and superflous if the inhibiting event refers to a concurrent response. Assigning
inhibitory properties to concurrent responses simply does not add to the ob-
served fact that an ongoing response decreases after increases in concurrent
responses.
The reinforcement model essentially states that interactions in concurrent
schedules are sufficiently accounted for by one functional relationship between
concurrent reinforcement rate and ongoing response probability. Hence, an on-
going response is said to be acted upon directly by concurrent reinforcements.
The present experiments suggest that the action of concurrent reinforcement
on the ongoing response may be further analyzed as two rather than just one
functional relationship. Concurrent reinforcements seem to control concurrent
responses directly (one functional relationship), with concurrent responses
then directly affecting the ongoing response (a second functional relationship).
Apparently, concurrent reinforcements affect an ongoing response only in-
directly, via alterations in concurrent responses.
The identification of concurrent responses as independent variables there-
fore argues that response probabilities are dependent upon more than just the
rates of contingent and concurrent reinforcement. This conclusion does not ap-
pear to be peculiar to the present analysis as previous experiments have shown
216 Section II: Operant Conditioning Procedures

that response probabilities do not "match" reinforcement rates in a range of

procedures involving additional parameters. At best, each additional parameter
influencing a given response could be represented as a quantifier in a general
mathematical formulation predicting the probability of responding. Of course,
the task of quantifying all behavioral variables into one large formula may be a
thankless attempt to outrun infinity. The problem here is that one particular for-
mula might fit the data from one procedure but might not equally apply to a
different procedure in which previous conditions "held constant" were now
changed.
Attempts have been made to extend the generality of reinforcement inter-
action to other than reinforcement parameters by introducing the terms sponta-
neous or hidden reinforcement (Herrnstein, 1970). The spontaneous reinforce-
ment is not an empirical parameter but rather is a quantity extracted from
obtained data. Of late, spontaneous reinforcement has been considered to be
intimately related to or imbedded in unspecified concurrent responses (Herrn-
stein, 1974; deVilliers, 1977). This assignment of reinforcement value to un-
known or at best potential concurrent responses apparently preserves the theo-
retical claim that response probability is only determined by reinforcement
context (Herrnstein, 1970). We believe, however, that as long as spontaneous
reinforcement value is only indirectly derived, such that it exactly covaries
with the probability of a response, then a change in reinforcement value is a
circular explanation of behavioral changes (Powers and Osborne, 1976).
Important to note, nevertheless, is that both the reinforcement and the re-
sponse models apparently concur that the probability of an ongoing response
and of concurrent responses may change in opposite directions. Reinforcement
interaction might positively propose that a change of one response is indirectly
associated with a change of concurrent responses. The response model would
push the critical step further and propose that a change of one response is
directly dependent upon changes of concurrent responses in and of themselves.
As Herrnstein has noted, "choice is nothing but behavior set into the context of
other behavior" (1970, p. 255). Within the reinforcement model, however, the
"other" behaviors are represented not in terms of their rate or probability, but
in terms of their contingent and inferred (inherent) reinforcement value. This is
precisely where the analyses separate. The reinforcement model specifies that
an ongoing response is at least partially dependent upon derived and inferred
entities. In contrast, the concurrent response analysis highlights that an ongo-
ing response is only dependent upon empirical parameters.
Response interaction, of course, emphasizes foremost concurrent responses
as determinants of an ongoing response. However, concurrent reinforcements
are also potent parameters. The present experiments suggest that concurrent
reinforcements and concurrent response parameters combine, so that the prob-
ability of an ongoing response may be decreased by concurrent reinforcements
to the extent that they increase the probability of concurrent responses. Con-
current performances may then be described as the precise relationship be-
tween concurrent reinforcements and concurrent responses, on the one hand,
and the equally precise relationship between the concurrent responses and the
ongoing response on the other hand.
Chapter 3: Concurrent Schedules 217

This analysis of concurrent schedules most satisfactorally meets the analysis

of operant-classical conditioning combinations. The effect of various arrange-
ments of concurrent operant and classical conditioning procedures could thus
be accounted for by the direct relationship between concurrent reinforcements
and concurrent responses. An ongoing baseline response is suppressed when
concurrent responses precede concurrent reinforcements but is unchanged or
facilitated when the baseline response precedes concurrent reinforcement
(Henton, 1972). A further consistency is that the probability of an ongoing re-
sponse appears to be inversely related to the probability of concurrent re-
sponses across a wealth of parametric manipulations in concurrent schedules.
In conclusion, the considerable mutual reinforcement provided by these re-
lated analyses suggests that the functional characteristics of concurrent
operant-classical-conditioning schedules and concurrent operant schedules
may be encompassed by the same basic behavioral principles. Additionally,
the response- response emphasis has been extended to complex sequential
conditioning procedures (multiple schedules; Chapter 4), to simple operant
conditioning procedures (Chapter 5), and to combinations of classical condi-
tioning procedures (Chapter 6).
This analysis of concurrent performances furthermore ties in well with both
previous (Skinner, 1950; Sidman, 1960) and more recent arguments (Dunham,
1971, 1972; Staddon, 1977) emphasizing the importance of interacting re-
sponses in conditioning procedures. A particular tribute is paid to the long held
insistence by Kantor that a behavior occurs in a field of other behaviors, never
in isolation (Kantor and Smith, 1975).
The growing generality of interresponse patterns strengthens the argument
that response factors are essential elements in concurrent performances. Such
repeated observations of interbehavioral relationships across many different
procedures suggest that response interactions are real and relevant and merit
systematic investigation.

References
Alleman, H. D., and Zeiler, M. D.: Patterning with fixed-time schedules of response-independent
reinforcement. J. Exp. Anal. Behav., 22, 135-141, 1974.
Bacotti, A. Y.: Matching under concurrent fixed-ratio variable-interval schedules of food presenta-
tion. J. Exp. Anal. Behav., 27, 171-182, 1977.
Baum, W. M.: The correlation-based law of effect. J. Exp. Anal. Behav., 20, 137-153, 1973.
Baum, W. M.: On two types of deviation from the matching law: bias and undermatching. J. Exp.
Anal. Behav., 22,231-242, 1974.
Boren, J. J.: Stimulus probes of the fixed ratio run. Paper delivered at Eastern Psychological Asso-
ciation, Philadelphia, 1961.
Browne, M. P., and Dinsmoor, J. A.: Selective observing of discriminative stimuli. Proceedings of
the 80th Annual Convention of the American Psychological Association, 1972, pp. 745-746.
Brownstein, A. J., and Pliskoff, S. S.: Some effects of relative reinforcement rate and changeover
delay in response-independent concurrent schedules of reinforcement. J. Exp. Anal. Behav., 11,
683-688, 1968.
Buel, C. L.: Investigation of the temporal parameters in omission training with humans in a two-
key situation. Psychol. Record, 25,99-109, 1975.
218 Section II: Operant Conditioning Procedures

Carlson, J. G., and Aroksaar, R. E.: Effects of time-out upon concurrent operant responding. Psy-
chol. Record, 20,365-371, 1970.
Catania, A. C.: Concurrent performances: reinforcement interaction and response independence.
J. Exp. Anal. Behav., 6,253-263, 1963.
Catania, A. C.: Concurrent operants. In, W. K. Honig (ed.): Operant Behavior: Areas of Research
and Application. New York, Appleton-Century-Crofts, 1966.
Catania, A. C.: Concurrent performances: inhibition of one response by reinforcement of another.
J. Exp. Anal. Behav., 12,731-744, 1969.
Catania, A. C.: Self-inhibiting effects of reinforcement. J. Exp. Anal. Behav., 19,517-526,1973.
Catania, A. C., and Dobson, R.: Concurrent performances: rate and accuracy of free-operant odd-
ity responding. J. Exp. Anal. Behav., 17,25-35, 1972.
Catania, A. C., and Reynolds, G. S.: A quantitative analysis of the responding maintained by inter-
val schedules of reinforcement. J. Exp. Anal. Behav., 11,327-383, 1968.
Catania, A. C., Silverman, P. J., and Stubbs, D. A.: Concurrent performances: stimulus-control
gradients during schedules of signaled and un signaled concurrent reinforcement. J. Exp. Anal.
Behav., 21,99-107, 1974.
Chung, S. H.: Effects of delayed reinforcement in a concurrent situation. J. Exp. Anal. Behav., 8,
439-444, 1965.
Chung, S. H., and Herrnstein, R. J.: Choice and delay of reinforcement. J. Exp. Anal. Behav., 10,
67-74, 1967.
Davis, H., Iriye, C., and Hubbard, J.: Response independent food as an extinction procedure for
responding on DRL schedules. Psychol. Record, 23,33-38, 1973.
Davis, J., and Bitterman, M. E.: Differential reinforcement of other behavior (DRO): a yoked-con-
trol comparison. J. Exp. Anal. Behav., 15,237-241, 1971.
Deluty, M. Z.: Choice and the rate of punishment in concurrent schedules. J. Exp. Anal. Behav.,
25, 75-80, 1976a.
Deluty, M. Z.: Excitatory and inhibitory effects of free reinforcers. Anim. Learning Behav., 4,436-
440, 1976b.
deVilliers, P.: Choice in concurrent schedules and a quantitative formulation of the law of effect.
In, W. K. Honig and J. E. R. Staddon (eds.): Handbook of Operant Behavior. Englewood Cliffs,
N.J., Prentice-Hall, 1977.
Dunham, P. J.: Punishment: method and theory. Psychol. Rev., 78,58-70, 1971.
Dunham, P. J.: Some effects of punishment upon unpunished responding. J. Exp. Anal. Behav., 17,
443-450, 1972.
Edwards, D. D., Peek, V., and Wolfe, F.: Independently delivered food decelerates fixed ratio
rates. J. Exp. Anal. Behav., 14,301-307, 1970.
Fenner, D. H.: Key pecking in pigeons maintained by short-interval adventitious schedules of rein-
forcement. Proceedings of the 77th Annual Convention of the American Psychological Associa-
tion, 1969, pp. 831-832.
Ferster, C. B.: Concurrent schedules of reinforcement in the chimpanzee. Science, 125, 1090-
1091, 1957.
Ferster, C. B.: A complex concurrent schedule of reinforcement. J. Exp. Anal. Behav., 2,65-80,
1959.
Ferster, C. B., and Skinner, B. F.: Schedules of Reinforcement. New York, Appleton-Century-
Crofts, 1957.
Findley, J. D.: Preference and switching under concurrent scheduling. J. Exp. Anal. Behav., 1,
123-144, 1958.
FleshIer, M., and Hoffman, H. S.: A progression for generating variable-interval schedules. J. Exp.
Anal. Behav., 5,529-530, 1962.
Henton, W. W.: Avoidance response rates during a pre-food stimulus in monkeys. J. Exp. Anal.
Behav., 17,269-275, 1972.
Henton, W. W., and Iversen, I. H.: Concurrent response rates during pre-event stimuli. Paper pre-
sented at the Easter Conference: English Experimental Analysis of Behavior Group, Cambridge,
March 1973.
Herbert, E. W.: Two-key concurrent responding: response-reinforcement dependencies and black-
out. J. Exp. Anal. Behav., 14,61-70, 1970.
Chapter 3: Concurrent Schedules 219

Rerrnstein, R. J.: Relative and absolute strength of response as a function of frequency of rein-
forcement. J. Exp. Anal. Behav., 4,267-272, 1961.
Rerrnstein, R. J.: On the law of effect. J. Exp. Anal. Behav., 13,243-266, 1970.
Rerrnstein, R. J.: Formal properties ofthe matching law. J. Exp. Anal. Behav., 21, 159-164, 1974.
Iversen, I. R.: Behavioral interactions in concurrent reinforcement schedules. Paper presented at
Easter Conference of the English Experimental Analysis of Behavior Group, Bangor, April
1974.
Iversen, I. R.: Reciprocal response interactions in concurrent variable-interval and discrete-trial
fixed-ratio schedules. Scand. J. Psychol., 16,280-284, 1975a.
Iversen, I. R.: Response versus reinforcement interaction in concurrent reinforcement schedules.
Paper presented at Easter Conference of the English Experimental Analysis of Behavior Group.
Exeter, March 1975b.
Iversen, I. R.: Interactions between reinforced responses and collateral responses. Psychol. Rec-
ord, 26,399-413, 1976.
Kantor, J. R., and Smith, N. W.: The Science of Psychology: An Interbehavioral Survey. Chicago,
Ill., Principia Press, 1975.
Katz, R. C.: Effects of punishment in an alternative response context as a function of relative rein-
forcement rate. Psychol. Record. 23,65-74, 1973.
Kimble, G. A.: Foundations of Conditioning and Learning. New York, Appleton-Century-Crofts,
1967.
LaBounty, C. E., and Reynolds, G. S.: An analysis of response and time matching to reinforce-
ment in concurrent ratio-interval schedules. J. Exp. Anal. Behav., 19, 155-166, 1973.
Lachter, G. D.: Some temporal parameters of non-contingent reinforcement. J. Exp. Anal. Behav.,
16,207-217,1971.
Lattal, K. A.: Response-reinforcer independence and conventional extinction after fixed-interval
and variable-interval schedules. J. Exp. Anal. Behav., 18, 133-140, 1972.
Lattal, K. A.: Response-reinforcer dependence and independence in multiple and mixed sched-
ules. J. Exp. Anal. Behav., 20,265-271, 1973.
Lattal, K. A.: Combinations ofresponse-reinforcer dependence and independence. J. Exp. Anal.
Behav., 22,357-362, 1974.
Lattal, K. A., and Bryan, A. J.: Effects of concurrent response-independent reinforcement on
fixed-interval schedule performance. J. Exp. Anal. Behav., 26,495-504, 1976.
Lowry, M. A., and Lachter, G. D.: Response elimination: a comparison of four procedures. Learn-
ing Motivation, 8,69-76, 1977.
Moffit, M., and Shimp, C. P.: Two-key concurrent paced variable-interval paced variable-interval
schedules of reinforcement. J. Exp. Anal. Behav., 16,39-49, 1971.
Neuringer, A. J.: Superstitious key pecking after three peck-produced reinforcements. J. Exp.
Anal. Behav., 13, 127-134, 1970.
Nevin, J. A.: Rates and patterns of responding with concurrent fixed-interval and variable-interval
reinforcement. J. Exp. Anal. Behav., 16,241-247, 1971.
Nevin, J. A.: The maintenance of behavior. In, J. A. Nevin and G. S. Reynolds (eds.): The Study of
Behavior. Glenview, Ill., Scott, Foresman, 1973.
Pavlov, I. P.: Conditioned Reflexes (Translated by G. V. Anrep). London, Oxford, 1927 (reprinted,
New York, Dover, 1960).
Pliskoff, S. S., and Green, D. :Effects on concurrent performances of stimulus correlated with rein-
forcer availability. J. Exp. Anal. Behav., 17,221-227, 1972.
Pliskoff, S. S., Shull, R. L., and Gollub, L. R.: The relation between response rates and reinforce-
ment rates in a multiple schedule. J. Exp. Anal. Behav., 11,271-284, 1968.
Powers, R. B., and Osborne, J. G.: Fundamentals of Behavior. San Francisco, West, 1976.
Premack, D.: Reinforcement theory. In, D. Levine (ed.): Nebraska Symposium on Motivation.
Lincoln, University of Nebraska Press, 1965.
Rachlin, R.: Contrast and matching. Psychol. Rev., 80,217-234, 1973.
Rachlin, R., and Baum, W. M.: Response rate as a function of amount of reinforcement for a sig-
nalled concurrent response. J. Exp. Anal. Behav., 12, 11-16, 1969.
Rachlin, R., and Baum, W. M.: Effects of alternate reinforcement: Does the source matter?J. Exp.
Anal. Behav., 18,231-241, 1972.
220 Section II: Operant Conditioning Procedures

Schneider, J. W.: Reinforcer effectiveness as a function of reinforcer rate and magnitude: a com-
parison of concurrent performancesJ. Exp. Anal. Behav., 20,461-471, 1973.
Schoenfeld, W. N., and Farmer, J.: Reinforcement schedules and the "behavior stream." In,
W. N. Schoenfeld (ed.): The Theory ofReinforcement Schedules. New York, Appleton-Century-
Crofts, 1970.
Schroeder, S. R.: Perseveration in concurrent performances by the developmentally retarded. Psy-
chol. Record, 25,51-64, 1975.
Sherman, J. A., and Thomas, J. R.: Some factors controlling preference between fixed-ratio and
variable-ratio schedules of reinforcement. J. Exp. Anal. Behav., 11,689-702,1968.
Shull, R. L., and Pliskoff, S. S.: Changeover delay and concurrent schedules: some effects on rela-
tive performance measures. J. Exp. Anal. Behav., 10,517-527, 1967.
Sidman, M.: Time discrimination and behavioral interaction in a free operant situation. J. Compo
Physiol. Psychol., 49,469-473, 1956.
Sidman, M.: Tactics of Scientific Research: Evaluating Experimental Data in Psychology. New
York, Basic Books, 1960.
Sidman, M.: Time out from avoidance as a reinforcer: a study of response interactions. J. Exp.
Anal. Behav., 5,423-434, 1962.
Skinner, B. F.: The Behavior of Organisms. New York, Appleton-Century-Crofts, 1938.
Skinner, B. F.: Superstition in the pigeon. J. Exp. Psychol., 38, 168-172, 1948.
Skinner, B. F.: Are theories of learning necessary? Psychol. Rev., 57, 193-216, 1950.
Skinner, B. F.: Science and Human Behavior. New York, Macmillan, 1953.
Skinner, B. F.: Contingencies of Reinforcement. New York, Appleton-Century-Crofts, 1969.
Spence, K. W.: The differential response in animals to stimuli varying within a single dimension.
Psychol. Rev., 44,430-444, 1937.
Staddon, J. E. R.: Learning as adaptation. In, W. K. Estes (ed.): Handbook of Learning and Cog-
nitive Processes, vol. 2. Hillsdale, N. J., Lawrence Erlbaum, 1975.
Staddon, J. E. R.: Schedule-induced behavior. In, W. K. Honig and J. E. R. Staddon (eds.): Hand-
book of Operant Behavior. Englewood Cliffs, N. J., Prentice-Hall, 1977.
Thomas, J. R.: Fixed-ratio punishment by timeout of concurrent variable-interval behavior. J. Exp.
Anal. Behav., 11,609-616, 1968.
Todorov, J. C.: Concurrent performances: effect of punishment contingent on the switching re-
sponse. J. Exp. Anal. Behav., 16,51-62, 1971.
Verhave, T.: Some observations concerning prepotency and probability of postponing shock with a
two-lever avoidance procedure. J. Exp. Anal. Behav., 4, 187-192, 1961.
Zeiler, M. D.: Eliminating behavior with reinforcement. J. Exp. Anal. Behav. 16,401-405, 1971
Zeiler, M. D.: Positive reinforcement and the elimination of reinforced responses. J. Exp. Anal.
Behav. 26,37-44, 1976.
Zeiler, M. D.: Elimination of reinforced behavior: intermittent schedules of not-responding. J. Exp.
Anal. Behav. 27,23-32, 1977.
Zener, K.: The significance of behavior accompanying conditioned salivary secretion for theories
of the conditioned response. Am. J. Psychol., SO, 384-403, 1937.
Chapter 4

Concurrent Responses with

Multiple Schedules

Iver H. Iversen

Explanations of Behavioral Contrast

The response rate in the presence of one stimulus and its associated schedule
depends also upon the schedules associated with temporally adjacent stimuli.
Alternations between schedule components associated with different stimuli
are traditionally referred to as multiple schedules (Ferster and Skinner, 1957).
Reynolds (1961b) suggested that interactions in mUltiple schedules might be de-
scribed as contrast or induction, depending upon whether the changes in re-
sponse rate diverge or converge across schedule components. Furthermore,
contrast and induction may be either positive or negative, depending upon di-
rection of change in response rate. Positive contrast, for example, refers to an
increase in response rate in one schedule component simultaneously with a de-
creased rate in an adjacent component.
Several interpretations have been offered to account for or explain interac-
tions in multiple schedules. However, since most interpretations primarily deal
with positive contrast the following brief overview only deals with this phe-
nomenon.
Perhaps the most well-known explanation is an extension of reinforcement
interaction outlined in Chapter 3. Based on experiments that assessed response
rate in one schedule component when the reinforcement rate was manipulated
in an adjacent component (Reynolds, 1961a, 1961b, 1961c; Bloomfield, 1967;
Nevin, 1968), the reinforcement model specifically suggests that the response
rate in one component depends upon the relative reinforcement rate (Herrn-
stein, 1970; Catania, 1973). However, the reinforcement model of multiple
schedules does not easily account for response changes occurring in one com-
ponent when the response rate but not the reinforcement rate changes in the
adjacent component (Brownstein and Hughes, 1970; Brownstein and Newsom,
222 Section II: Operant Conditioning Procedures

1970; Hughes, 1971; Lander, 1971; Wilkie, 1973). Conversely, positive contrast
may not occur after a decrease in reinforcement in an adjacent component with
particular errorless discrimination procedures (Terrace, 1963; compare to Rill-
ing, 1977) or in conjunction with certain brain lesions (Henke et aI., 1972) or
drugs (Bloomfield, 1972). Apparently, a change in relative reinforcement rate
may be neither necessary nor sufficient for positive contrast.
Frustration and emotions have also been evoked in the explanation of posi-
tive contrast. One theory holds that positive contrast is a manifestation of emo-
tional responses, generated by the aversiveness of not receiving reinforcement
for previously reinforced responses under an adjacent component (Terrace,
1966). However, as mentioned above, reinforcement rate changes are neither
necessary nor sufficient for positive contrast, therefore positive contrast may
be produced in situations that do not involve emission of nonreinforced re-
sponses. Moreover, several experiments have shown positive contrast without
nonreinforced responses in an adjacent component (Reynolds, 1961b; Freid-
man and Guttman, 1965; Vieth and Rilling, 1972; Sadowsky, 1973: Halliday and
Boakes, 1974) and, conversely, an absence of positive contrast with nonrein-
forced responses in an adjacent component (Halliday and Boakes, 1971; Mack-
intosh et al., 1972). The occurrence of nonreinforced responses is then neither
necessary nor sufficient for positive contrast.
Changes in preference among schedule components have also been offered
as an explanation of positive contrast. Specifically, positive contrast has been
suggested to result when the adjacent component is relatively less preferred
(Bloomfield, 1969; Premack, 1969). However, relative preference for one
schedule component, as assessed with concurrent schedules, may not be a reli-
able predictor of positive contrast with the same schedule components ar-
ranged as a mUltiple schedule. Wilkie (1973), for example, reported that posi-
tive contrast was obtained with signaled reinforcement in an adjacent
component, whether or not preference for signaled reinforcement had been ob-
tained in the same subjects when the schedules were arranged concurrently.
Also, Halliday and Boakes (1972) found preference for response-dependent re-
inforcement over response-independent reinforcement but not positive con-
trast when the same schedules were arranged as a multiple schedule. There-
fore, relative preference for schedule components also appears to be neither
necessary nor sufficient for positive contrast.
According to a more recent formulation, interactions between classical and
operant conditioning may produce positive contrast (the response additivity
theory; Gamzu and Schwartz, 1973; Hemmes, 1973; Redford and Perkins,
1974; Schwartz and Gamzu, 1977). The response additivity theory suggests that
responses directed toward a stimulus associated with a relatively higher rein-
forcement rate (stimulus-reinforcer relation) may add to the operant response
producing the reinforcer (response-reinforcer relation). Positive contrast then
results when the component stimuli are located on the operant manipulandum,
with stimulus-directed responses and operant responses thereby activating the
same manipulandum. Experiments accordingly demonstrated that positive con-
trast may not occur when the component stimuli are not on the operant manipu-
Chapter 4: Concurrent Responses with Multiple Schedules 223

landum (Keller, 1974; Schwartz, 1975; Schwartz et aI., 1975; Spealman, 1976).
However, other experiments have shown that positive contrast may easily be
obtained when the component stimuli are not located on the manipulandum for
the operant response (Beecroft, 1969; Bloomfield, 1972; Boakes, 1972; Chit-
wood and Griffin, 1972; Coates, 1972; Henke et aI., 1972; Gaffan, 1973;
Hemmes, 1973; Bradshaw, 1975; Farthing, 1975; Gutman et aI., 1975; Gutman
and Minor, 1976; Henke, 1976; Bradshaw et al., 1978; Gutman, 1977). On the
other hand, contrast is not invariably obtained when the component stimuli are
in fact located on the operant manipulandum (Terrace, 1963). Placement of the
component stimuli on the operant manipulandum therefore seem to be neither
necessary nor sufficient for positive contrast.
A relatively novel account suggests that positive contrast results from the
stimulus of the adjacent component signaling a decrease in local reinforcement
rate (Marcucella, 1976; B. A. Williams, 1976; Wilkie, 1977). For example, with
a maintained overall reinforcement rate, positive contrast appears to depend
upon the temporal characteristics of reinforcement in the adjacent component
(B. A. Williams, 1976). Positive contrast thus occurred when reinforcement
was available late but not early in the adjacent component. This model has not
yet been extended further and any determination of whether the critical vari-
able is either necessary or sufficient for contrast is perhaps premature. How-
ever, the available evidence that positive contrast may not occur with some
forms of errorless training (Terrace, 1%3) or in subjects with certain central
nervous system lesions (Henke et al., 1972) suggests that a change in local rein-
forcement rate in an adjacent component is perhaps not sufficient for positive
contrast.
Each of the above theoretical positions suggests that positive contrast re-
sults from one common effect. The theories differ according to what sort of
"change" best describes the common effect. Changes in reinforcement rate,
response rate, stimulus-directed responses, frustration, aversiveness, prefer-
ence, or local reinforcement rate have each been proposed to account for posi-
tive contrast. None of the explanations, however, seems to be able to account
for all the various experimental findings with mUltiple schedules. Some of the
theoretical positions make reference to intervening variables, such as frustra-
tion or aversiveness, and generally suggest that a number of experimental man-
ipulations affect the intervening variable, which in tum affects the response
rates. Measurement of intervening emotions by responses such as wing-flap-
ping and attack has been attempted, but a comparison of reports indicates that
the "measures" are inconsistently related to both the independent and depend-
ent variables of interest (Terrace, 1971; Coughlin, 1973; Rilling and Caplan,
1973, 1975; Rilling et aI., 1973).
A significant feature of multiple schedules is that interactions occur across
components and are therefore separated temporally. Interactions in multiple
schedules are thereby critically different from interactions obtained with con-
current reinforcement schedules. Response rate changes, however, are often
very similar in concurrent and mUltiple schedules. For example, changing muit
VI VI to mult VI EXT, or changing cone VI VI to cone VI EXT, will equally
224 Section II: Operant Conditioning Procedures

decrease the response rate under EXT and increase the response rate under VI
(Herrnstein, 1970). In concurrent schedules, the time allocated to one response
detracts from the time available to a second concurrent response. In multiple
schedules, however, the time allocated to responding in one component cannot
(directly) affect the time available for responding in. temporally adjacent com-
ponents. Response interactions across components in mUltiple schedules
would therefore seem to be formally different from response interactions be-
tween components of concurrent schedules.
Recently, investigations have attempted to assess how or by what means re-
sponse interactions across mUltiple schedule components may be related to
concurrent response rate changes in each component.
The stimulus-directed responses generated by differential reinforcement
rates across multiple schedule components suggest one possible route to an un-
derstanding of positive contrast under at least some procedural arrangements.
Note that the stimulus-directed responses are generated concurrently with the
operant rate changes described as positive contrast effects. Furthermore, pro-
cedures that prevent positive contrast do so by selectively adding more concur-
rent responses to the component with the relatively higher reinforcement rate
(Keller, 1974; Redford and Perkins, 1974; Schwartz, 1975). Concurrent re-
sponses dissociated across multiple schedule components may therefore criti-
cally participate in the formation of positive contrast.
The purpose of the following three experiments was to offer an analysis of
multiple schedules in terms of interactions among concurrently occurring re-
sponses. These experiments represent an attempt to interrelate changes in the
rate and pattern of operant responses and collateral responses under the com-
ponents of mUltiple schedules. In two experiments, the operant response and
the collateral responses were explicitly manipulated to cast some light upon
possible mechanisms whereby responses in one component may be acted upon
by events in a second, temporally adjacent component.

Component Duration and Collateral

Responses (Experiment 1)*
In the first study, interactions between reinforced and collateral responses
were merely illustrated, with an experimental manipulation of component dura-
tions of a mUltiple schedule.

Experiment I
Three male naive Wi star albino rats were trained on a mult FR 20 FR 40 sched-
ule. The rats responded for a 20% sucrose solution in a two-lever chamber (de-
scribed by Iversen, 1975b) and were maintained at 80% of their free-feeding
body weights. Lever pressing was reinforced according to a two-ply multiple
schedule. In one component (A), responses on the left lever (A) were rein-

* Experiment was conducted at the University of Copenhagen, from January to August 1972.
Chapter 4: Concurrent Responses with Multiple Schedules 225

forced on FR 20 and responses on the right lever (B) had no scheduled conse-
quences. In the other component (B), responses on lever B were reinforced on
FR 40 and responses on lever A had no effect. Responses on the appropriate
lever were followed by a relay click in each schedule component. A 9OO-Hz
tone sounded from the ceiling of the chamber when component B (FR 40) was
scheduled. Licking on the sucrose delivery tube was recorded by a drinkometer
circuit.
Initially the duration of each component was 240 sec (for 45 sessions) and
was then changed to 400 sec (15 sessions), 120 sec (15 sessions), and 600 sec (15
sessions). A session was terminated after 40 min. Component changes were in-
dependent of responses on any lever, and the FR counters were not reset at
component changes. The components occurred in strict alternation with the
first component in a session selected at random.
The results of this parametric manipulation are presented in Table 4.1. For

Table 4.1 Means of the last five sessions of each component duration

Response
rate for Response
reinforced Licking rate for
Rat Component lever duration other lever
no. Schedule duration (resp/min) (sec/min) (resp/min)

SI FR20 120 104.4 7.8 0.7

240 93.7 7.9 0.6
400 97.7 9.6 0.2
600 82.1 21.9 0.2

FR40 120 28.8 13.4 7.7

240 37.2 13.1 8.0
400 35.4 17.3 6.1
600 28.8 19.9 6.0

S2 FR20 120 88.6 13.2 0.6

240 103.0 9.4 0.8
400 85.7 13.6 0.1
600 53.5 30.1 0.1

FR40 120 19.8 14.4 7.0

240 23.4 12.5 7.8
400 19.8 12.4 6.4
600 16.8 30.8 5.7

S3 FR20 120 64.6 6.0 0.7

240 73.8 3.0 0.5
400 66.6 6.3 0.2
600 61.1 16.9 0.1

FR40 120 26.4 5.3 5.0

240 21.6 4.8 5.9
400 25.8 6.0 5.0
600 22.8 11.6 4.7
226 Section II: Operant Conditioning Procedures

all rats, the overall rate of reinforced lever pressing was higher under FR 20
than under FR 40. The only systematic effect of the component duration was a
tendency for a decreased rate under FR 20 for the longest component duration.
Although the overall response measures remained relatively invariant, the
postreinforcement pause and the run rate measures depended upon the manipu-
lated component duration (Figure 4.1) . Under FR 40, both the postreinforce-
ment pause and the run rate were increasing functions of the component dura-
tion. These effects approximately balanced, resulting in an almost constant
overall response rate. Under FR 20, the postreinforcement pause increased for
the longest component duration, whereas the run rate remained approximately
the same for all component durations. The overall rate under FR 20 was there-
fore decreased during the longest component duration because of an increased
postreinforcement pause.
Hence, invariant overall response rates do not necessarily signify invariant
response patterns. In particular, opposing changes may balance one another
and leave the overall response rate unchanged. However, more molecular
changes may be systematically related to experimental manipulations.

I
fR20 fR 40 FR 20 FR 40
p

,l' so
~
z 12 -51 60 i
~rn ..... 52
. - . 53 I
I
~~
""''''
!U58
~100 ~
I
40 Z 40
au

~
0.. .... I "
...... _.
~!a,
g:l)!
~
20 ~ w-·· .. · -... . ... ', • •• • .Jt 20
<fl:O Z 60
::J <
A ~~ a:

8 8
,

8
O~
__ ...... '
.x '
X"" ~-
' . ... -.0

lnffi .,. ....

~~L
~!
w
a: a:
2
.w
B a~
z-'

z
o
12
.
I
40
I
I

~:0 8
!s
Z 20~
~~k
0;;; , 20
C)~4 ~
~
~ ~, 10
_0 ~ O o __ • -o.- ... .A
><u 2 ...g.. _

C ~~ ' - " - v'

120 240 1,1)() SOO 120 240 400 600 120 240 400 600 120 240 400 600

COMPONENT DURATION (SECONDS)

Figure 4.1. Response interactions as a function of the component duration of a mull FR 10 FR 40
schedule. A. FR postreinforcement pauses and run times. B. Nonreinforced responses on alternate
lever. C. Collateral licking. Data are means from the last five sessions.
Chapter 4: Concurrent Responses with Multiple Schedules 227

Discussion
The differentiation in the operant response pattern across components was
clearly associated with a differentiation in the pattern of the recorded collateral
responses. The longer postreinforcement pauses under FR 40 were associated
with longer durations of licking the empty sucrose delivery tube plus a higher
frequency ofresponses on the nonfunctional lever (Figure 4.1). This finding is
in agreement with several previous reports showing an increased postreinforce-
ment pause (Ferster and Skinner, 1957; Felton and Lyon, 1966) and/or an in-
creased duration of collateral responding (Schaeffer and Diehl, 1966; Hutchin-
son et aI., 1968; Flory, 1969; Carlisle, 1971) with increasing FR size. In the
present experiment, the duration of collateral licking systematically increased
in each component along with the increases in postreinforcement pauses as a
function of the manipulated component duration. Thus, postreinforcement
pauses and the associated collateral responses may expand not only when the
FR size is increased, but also when stimuli enclosing FR schedules are in-
creased in duration.
Apparently, operant as well as collateral responses may change rate and pat-
tern as a function of changes in a single independent variable (component dura-
tion in this experiment). More broadly, the experiment suggests that mUltiple
schedules generate local interactions among concurrently available responses
in each schedule component.

Positive Contrast and

Collateral Responses (Experiment 11)*
Typically, response interactions in multiple schedules refer only to sequential
changes in one response across the successive components. The traditional ex-
ample is that an operant response rate may decrease under the EXT component
and increase under the VI component when a simple VI or a mult VI VI is
changed to mult VI EXT (Reynolds, 1961b). The results from Experiment I
suggest, however, that concurrent interactions may also occur between the
operant response and collateral responses within each component of the multi-
ple schedule.
Collateral responses have occasionally been recorded, but not typically re-
lated to operant responses, in the analysis of mUltiple schedule effects. Some
available evidence indicates that collateral responses, such as attack, schedule
escape, wheel running, and wing flapping, may occur almost exclusively under
the EXT component of various mUltiple schedules, including mult CRF EXT,
mult FI EXT, and mult VI EXT (Skinner and Morse, 1957; Azrin et aI., 1966;
Catania, 1966; Terrace, 1966,1971; Rilling et aI., 1969, 1973; Rilling and Cap-
lan, 1973). Likewise, collateral attack, mirror pecking, and unreinforced lever

* Experiment was conducted at the University of Copenhagen. from January to December

1974, and reported by I. Iversen (1975a). The author thanks Mrs. E. Nielsen for her assistance
during the experiment.
228 Section II: Operant Conditioning Procedures

pressing may occur most frequently in the component with the lower operant
response rate in mult FR FR (Keehn and Bratbak, 1967; Flory, 1969; Cohen
and Looney, 1973; Iversen, 1975b). In all the above schedules, a high rate of
collateral responding was then associated with a low rate of operant responding
and a low rate of collateral responding was associated with a high rate of
operant responding. The pattern of still other collateral responses, however,
may be highly dependent upon, and indeed reversed by schedule manipula-
tions. Thus, collateral drinking primarily occurs under the EXT component of
mult CRF EXT (Keehn and Colotla, 1971) but under the VI component of mult
VI EXT (Jacquet, 1972). Similarly, collateral responses directed toward dis-
criminative stimuli signaling schedule components occur primarily under the
VI component of mult VI EXT (Keller, 1974).
Quite recently a few experiments have purposely manipulated collateral re-
sponses in mUltiple schedules. In a study by Allen and Porter (1975), access to
collateral drinking was prevented in one component of mult FI FI. The drinking
rate increased in the unchanged component, but unfortunately no data were
provided on possible simultaneous changes in the operant response rate. Simi-
larly, Rilling and Caplan (1973) reported that collateral attack responses were
more frequent under EXT and lower under VI of mult VI EXT relative to sim-
ple VI. Again the concomitant changes in operant response rates were not pre-
sented. Rilling and Caplan did note, however, that the changes in the rate of
collateral attack precluded an analysis of operant responding because the at-
tack response "competed with and lowered" the rate of VI responding. In a
subsequent study, Rilling and Caplan (1975) accordingly excluded attack re-
sponses from the VI component to prevent competition with the operant re-
sponse. When a simple VI was changed to mult VI EXT, the operant rate in-
creased under VI for four of seven subjects, but decreased or remained
unchanged for the remaining three subjects. According to Rilling and Caplan,
the latter three subjects showed an increase of competing preattack responses
under VI which may have prevented positive operant contrast.
Keller (1974), Schwartz (1975), and Spealman (1976) recorded operant re-
sponses on one key as well as collateral responses directed toward the compo-
nent stimuli of the multiple schedule. The overall rate of the operant response
either did not increase (Schwartz, 1975), or actually decreased (Keller, 1974;
Spealman, 1976) with the simultaneous emergence of competing stimulus-
directed responses during the VI component of a mult VI EXT schedule. Speal-
man (1976) further reported that the rates of the operant and the stimulus-
directed responses were increasing and decreasing functions, respectively, of
the component durations of the mult VI EXT schedule. The interactions in
Spealman's experiment were also more local, with an initial low rate of the
operant response (in the first 3 sec of each VI component) associated with an
initial high frequency of stimulus-directed responses.
Across experiments, the data provide some indication that specific collateral
responses change rate during mUltiple schedules. The inverse relationships be-
tween the operant and collateral response rates within components particularly
invite examination of possible causal relationships between response rate
Chapter 4: Concurrent Responses with Multiple Schedules 229

changes in multiple schedules. Such response interdependencies suggest that a

functional analysis of sequential interactions might be incomplete by neglecting
collateral response changes systematically associated with the schedule mani-
pulations. By and large, relevant interactions between operant and collateral
responses in multiple schedules have often been sidestepped by selectively re-
cording only the collateral response changes or only the operant response
changes. The extensive interactions between concurrently occurring responses
reported in this volume would indicate that changes in collateral responses may
perhaps participate in the formation of operant response changes. The rate of
collateral responding in multiple schedules may then effectively prevent or pro-
mote changes in the operant response which are described as positive and nega-
tive behavioral contrast.

Experiment II

The purpose of this experiment was to investigate systematically the relation-

ship between operant responses and collateral responses during a typical multi-
ple schedule manipulation, a change from simple VI to mult VI EXT.
Initially, three naive male Wistar albino rats were reduced to 80% of their
free-feeding body weights and trained in a general apparatus consisting of a re-
sponse lever and a food cup mounted on one side of a running wheel. The wheel
was 33.0 cm in diameter, and the running surface consisted of 15.5-cm long
metal bars spaced every 8.5 mm. Four magnets mounted equidistant on the cir-
cumference of the wheel could activate two reed relays spaced 9.0 cm apart on
the supporting base of the running wheel. The wheel could turn in both direc-
tions. A modified relay served as a brake to the wheel. A 3.0 x 2.0 x 1.0 cm
metal food cup was mounted on the right side, 1.0 cm above the running sur-
face. Small holes in the bottom of the food cup prevented accumulation of pel-
let dust. A 3.0 x 3.0 cm flat metal lever was mounted with the right edge 4.5
em from the left edge of the food cup. The lever was also 1.0 em above the
surface of the running wheel. A lever press was recorded as a 3.0-mm de-
pression with a force greater than 10 g. Since lever pressing and wheel running
were topographically different and incompatible responses, each response was
recorded in duration rather than frequency (Premack, 1965).
The duration of each response was recorded as the number of O.5-sec
periods with one or more instances of that response (Dunham, 1972; and Ex-
periment VI, Chapter 3, this volume). Each quarter revolution of the wheel
constituted a response provided both reed relays had been activated in a given
O.5-sec period. Thus, running was only recorded when the wheel was rotated at
a minimal speed of 9.0 cmlO.5 sec along the circumference (l0.8m1min).
Throughout the experiment, lever pressing was reinforced on a VI I-min
schedule in the presence of a houselight. The reinforcer was one 45-mg food
pellet and a session was terminated after 42 min. The running wheel was locked
before and after the session.
The first question posed was simply as follows: What will happen to collat-
eral wheel running during the establishment of behavioral contrast for operant
230 Section II: Operant Conditioning Procedures

responding? After pretraining, lever pressing was maintained on VI 1 min for 15

sessions with free access to wheel running (Phase A). The VI for lever pressing
was then changed to mult VI EXT with strictly alternating 3-min components
(Phase B). The houselight was on during VI and off during EXT.
This change from VI to mult VI EXT produced the data presented in Figure
4.2. For all rats, the probability oflever pressing decreased under EXT and in-
creased under VI compared to the simple VI. Wheel running simultaneously
increased under EXT and decreased under VI. The wheel-running probability
thus changed in a direction opposite to the lever-pressing probability for each
component of the mult VI EXT. The data were then directly replicated with the
schedule changed back to VI (Phase C), and then to mult VI EXT again (Phase
D).
The increased lever-pressing probability under VI, along with the decreased
probability under EXT, is an example of positive contrast for an operant re-
sponse (Reynolds, 1961b). The simultaneously decreased probability of collat-
eral wheel running under VI and increased probability under EXT would be an
example of negative behavioral contrast for a collateral response. These oppos-
ing contrast effects for operant and collateral responses appear to have at least
some generality, since Rand (1977) recently reported that certain collateral re-
sponses in pigeons increased under EXT and decreased under VI along with
the typical reverse changes for the operant response. Positive contrast for an
operant response may thus be obtained along with negative contrast for a spe-
cific collateral response.
The response changes associated with the schedule changes from simple VI
to mult VI EXT invite the following questions: (1) Did the changing probability
of collateral responding influence the probability of operant responding, or (2)
did the change in operant responding influence the collateral responding? The
same questions may be posed for each component of the mUltiple schedule.
The possible relations between collateral and operant responding were there-
fore examined by manipulating access to collateral wheel running across ex-
perimental phases.
When the wheel was locked under EXT in Phase E, the wheel-running prob-
ability of course decreased immediately to zero under EXT; but wheel running
simultaneously increased under VI, with a concurrent decrease in lever press-
ing (Figure 4.2). The lever-pressing and wheel-running probabilities under VI
were now comparable to the data obtained with simple VI in Phase C. The in-
crease in lever pressing that previously occurred with the change from VI to
mult VI EXT was thus counteracted by only allowing wheel running under VI.
Hence, positive contrast for the operant response was effectively prevented by
manipUlating access to the collateral response. This effect was clearly under
experimental control, since lever-pressing and wheel-running probabilities re-
turned to the contrast patterns when wheel running was again possible through-
out a session (Phase F compared to Phase D).
Again the data are consistent with recent reports. Allen and Porter (1975)
showed that preventing access to collateral water drinking under one compo-
nent of a multiple schedule for food reinforcement increased the duration of
 .
Chapter 4: Concurrent Responses with Multiple Schedules 231

LEVER WHEEL
~
• : VI,F VI,F VI,F VI,F VI,F VI,F VI.L VI.L VI,L
0: EXT,F EXT,F EXT,L EXT,F EXlFEXT,L -
PHASE : A 8 C 0 E F G H I
0.1. - R1

e:i 0.3_ ..... / ~

.r-- w-- .-v- ..... r
---
>
w 0.2~-
...J
0.1 - "\y., 1\
~ ~ 00<>",. I ...........

-
~ <>000/'0 ~
...... r~ ~
...J 0.3
W
~ 0.2-
IV
>-
~ 0.1 -
I~ ........
~

...... . ~ ~ ~ .......... ..... ..... ......

....J R2
m a:: 0.4 ,- ~
« WO.3 -
m> -
o ~0.2
a:: 0.1 -
~ ~ I~ ~ ~ ~
Cl.

r
:c. ......... ----
...J 0.3
Ww
(/) ~0.2~- ~ ~
Z ~0.11- ........ ~
o ~
Cl.
(/)

I .....,- ........ .r .r ...... ....-

w R3
a:: 0.4 - ......... ~
O.3-
Q:
w
>
wO.2,...
...J ~
0.1 -
~ ~ ~ \00 aDoo
...J
~0.2 -
:I:
~0.1 -~
oJ:l.cpoO

'.... .............
I
~
1 ......... -
....... rrf>%oo
I.......
~

'T
10 20 30 1.0 SO 60
BLOCKS OF THREE SESSIONS
Figure 4.2. Relationships between lever-pressing and wheel-running probabilities in three-session
blocks for each rat and experimental phase. Experimental conditions are indicated at the top of
each column. The schedule for lever pressing was either VI or EXT and the wheel was free (F) or
locked (L) in each schedule component.
232 Section II : Operant Conditioning Procedures

drinking under the second component. The available data seem to suggest that
reducing the probability of a collateral response in one component may result in
an increase in the probability of the same collateral response in a second com-
ponent.
More minute interactions between lever pressing and wheel running are
shown in representative event recorder segments in Figure 4.3 and clearly con-

RAT 1 VI
---,ailll.",,"""
~:,,~~.:;:,;,....-..."T•.,."T•..,"M...."..."...., r.T,"....n.ft"n....!rrln-" :,,-•••-:,--:-.~.~Iii.' .... " i" ii'
j I
PHASE

3, •••• , ••• _ . . . . ......... I i "ii • • • •

i
•••• i
it Ii iii lili i .. M,
A'" . . . . . _ c
Cl b

.. ...,
EXT VI
1------------------------------;.--.-----,--.-
2----~--------------------------~.~.~
I.~.n"~"~n.~"n"'~.~i1.ft~.rl~~~"n.~.~.T. ~.n.~I~.m"~I.~.n.~1 •• •••• __ o
3" .." '_"_'_'1•• Ii • • • , " d

EXT WHEEL LOCKED VI

I------------------------------------------------~----~
2r-------------r-----·----------~.n.~.""n'.m'.".~.n.r.r.".~.r.~I.n'~ ••~
I~"nl~.m~'mm'_.T.r.~ ••rn"n.M"•• •• E
31--------------------------~nn~~.nmn--nn~__--_n~
4,------____________________~·_·'_'-
__ --_~~·~
'·-
'·-·_.,._._
•• __'_"_.__ ••~.
••_. ____._._

RAT 2
Irl--------------r-------,---------------·-------------------------r
2'_ •• iI . , , " ."'_ii •• I '
ii'_
I _ ' i l i l i " i i i i , . M i h _ i i 8 i i • • • i , d . l i l t• • • • •II . . . ." 11_'ii"
3.." • • • _". he. 'i _ _ • • • • , • • Ii i . i i" ...... ,_._ a _ a
Q b
•

.
1---------------------------------
2", • i i Ii iii Ii '_'i .aMtiiWmiMe'••
& i i , i iiii'W"II"hW". _
11 I
it' i 8 ••
ill i .... • ..... I ..........._ .. " • • I 'd ,.,. .M." • "
~r-----------------------------~--~------------------------~

2- -.ano
..
m'T.n"~----~----------------~'.~.n.~lim.n'.n'~.~.~.~.~.~'w.~I.n'n.~.~'r1.~.~.r.~.~.'m'."".~I'
3 , • i • • NW aM Wi i • •, i , • iii. ,.It
•

RAT 3

.n I.
1 J , I i i I j I
2_*_.." __ 00'1111"' ••• , •• 00 I.' "" .•• ,"",.,., _ •• i •••• to., ,. . ........... '., •. M. -, ••
3 b .. •• cf"" •• ...... . . " . WI.. •• . ·•·•
2 •• _, liill. i • Ii
IIi ii i Ii
1.'*111111_1 _ _ " _ _ _ _ ._.11_ .WiI&, lilli., iii

".,1 _"8 liJili. "8"W"1i I i• • i ii, i

d
II

2." .,
3
Ii. " . , " I i " "" ii. Ii I , Ii
Ii i

ih' • M.
•
i... .
1118• • ;"111 • • • • • • • • • • _ _ _ _ \,,,._,11 ••• _.

i , • i . .. ",
- - - - - - - - - - - 3 MIN - - - - - - - - - - -

Figure 4.3. Event recorder segments of operant lever pressing and collateral wheel running from
Phases C, D, and E for each of three rats. I, pellet; 2, lever; 3, wheel running; 4, stimulus.
Chapter 4: Concurrent Responses with Multiple Schedules 233

firm the relationships described in Figure 4.2. With VI (Phase C), lever pressing
bursts regularly alternated with wheel running (a). Periods with a high lever-
pressing density were associated with an absence or infrequent wheel running
bursts (b). Conversely, a relatively low lever-pressing density was associated
with frequent wheel-running bursts (c). With mult VI EXT in Phase D, the fre-
quent alternation between lever pressing and wheel running was now replaced
by a more infrequent alternation, largely under control of the component stim-
uli. At the onset of the VI component, lever pressing emerged and wheel run-
ning ceased, whereas onset of the EXT component produced a reverse pattern,
with a cessation of lever pressing and emergence of wheel running. The pattern
of lever pressing under the VI component now resembled the pattern during VI
alone (Phase C), with wheel running either absent or infrequent (d, compare to
b). Finally, with wheel running prevented under EXT (Phase E), alternation be-
tween lever-pressing and wheel-running bursts returned in the VI component
(e) as with the VI in Phase C. Wheel-running bursts, however, were largely
confined to the first half of the VI component (f). IRT in lever pressing conco-
mitantly tended to be longer during the first half of the VI component.
Summing up at this point, the critical feature of this experiment was that pre-
vention of wheel running under EXT (Phase E) resulted in an increase in wheel-
running probability under VI, which in tum resulted in a concurrent decrease in
lever-pressing probability under VI and prevention of contrast. However, this
change in lever pressing under VI might be attributed to the simultaneous
changes in wheel running under VI, or EXT. This possibility was further ana-
lyzed by preventing wheel running under VI (Phase G) and then under EXT
(Phase H). The results were clear: lever pressing under VI was unaffected by
the gross changes in wheel running under EXT if wheel running was also pre-
vented under VI (Figure 4.2).
In combination, the lever-pressing probability under VI appeared to be
directly dependent upon the concurrent probability of wheel running within the
same VI component. Yet, an increase in wheel running under VI was produced
by preventing wheel running under EXT. Therefore, the changes in VI lever
pressing were also indirectly dependent upon the wheel-running probability
under EXT. The experimental manipUlations thus indicate that positive con-
trast may be prevented by allowing a specific collateral response to occur only
concurrently with the operant response under VI.
The present experiment did not directly assess the possible influence of col-
lateral wheel running on the course of extinction of the operant response under
EXT. The results tentatively suggest, however, that a low probability of the
operant response under EXT may not be increased when the probability of the
concurrently occurring collateral wheel-running response is decreased to zero
(Phase E). Therefore, collateral wheel running under EXT did not seem to be
responsible for the maintained low probability of the operant response under
EXT; rather, the decreased probability of the extinguished operant response
allowed for an increase in the probability of wheel running.
The changing interrelations in the VI component from one phase to the next
are shown in Figure 4.4. Phase-to-phase changes in lever-pressing probability
234 Section II: Operant Conditioning Procedures

RAT 1 RAT 2
~ <D
0.08 0.08

o.oL
• 0.04

-0.12
0.08

o.OL -O.oL
Y=-o.L8x -0.001 Y= - 0.73x - 0.003
0.08 -0.08

RAT 3
LEVER PRESSING
PROBABILITY
(Pj - Pi)
• 0.08

o.OL

0.08 WHEEL RUNNING

PROBABILITY
(Pj-Pi)
y =- 0.97. - o.oos
-0.08

Figure 4.4. Changes in the probabilities of lever pressing and wheel running under the VI compo-
nent across successive phases of the experiment. Positive and negative values indicate increased
and decreased probabilities, respectively. Data represent the differences in response probabilities
for the last five sessions of each successive phase. Regression lines were calculated by the method
of least squares. (Data are not presented for probability changes from Phase G to Phase I with the
wheel locked.)

are plotted on the ordinate and changes in wheel running are plotted on the ab-
scissa. An increase in lever pressing was clearly associated with a decrease in
wheel running (data in the second quadrant), and a decrease in lever-pressing
was associated with an increase in wheel-running (data in the fourth quadrant).
For each rat, the intercept of the regression lines was close to zero, and the
lines had a negative slope. Overall, the numerical changes (independent of sign)
in the probability of wheel running exceeded or equaled the changes in the
probability of lever pressing.
Inverse relations between lever-pressing and wheel-running probabilities
were also obtained under EXT in Phases B and D. For all rats, the numerical
change in the probability of wheel running was much smaller than the simulta-
neous change in the probability of lever pressing (Figure 4.2). Under EXT, the
Chapter 4: Concurrent Responses with Multiple Schedules 235

change in wheel-running thus fell short of the simultaneous change in lever

pressing. In subsequent phases, the lever-pressing probability remained low
under EXT across low or high probabilities of wheel running resulting from
manipulations of access to run. The results therefore indicate that the de-
creased probability of the operant response under EXT allowed for the concur-
rent increase in the probability of wheel running, rather than vice versa.

Discussion
These results are immediately relevant to recent descriptions of collateral re-
sponses during multiple schedules. The probability changes in wheel running
are analogous to probability changes in collateral attack (Rilling and Caplan,
1973) and collateral turning motions (Rand, 1977). Rilling and Caplan (1975) ob-
served that specific components of collateral attack interfered with and pre-
vented an increased probability of operant responding under VI of mult VI
EXT. In agreement with the present data, Keller (1974), Schwartz (1975), and
Spealman (1976) reported that an increase in operant response rate under VI
may not be obtained when concurrent stimulus-directed responses are simulta-
neously generated predominantly under VI in mult VI EXT.
The available data are in agreement that an increased probability of operant
responding under VI may be (1) prevented when a collateral response can only
occur under VI, or (2) produced by a decrease in the probability of a collateral
response under VI. In combination, positive contrast may not occur when com-
peting responses occur exclusively or predominantly under VI.
We previously posed the question of how a change in the operant response
probability in one component may affect the probability of the same operant in
succeeding schedule components. Although tentative, the results suggest that
this effect is mediated by collateral responses, or more specifically that positive
contrast may result from a decreased probability of a collateral response under
VI with the schedule change to mult VI EXT. At least for wheel running inves-
tigated in the present experiment, the relatively unrestricted opportunity to run
in the wheel under EXT may decrease the wheel-running probability under VI
when running is relatively more constrained by the operant response. Such a
displacement of a collateral response away from VI and into EXT would in-
crease the available time for the operant response under VI, resulting in posi-
tive contrast. Further, positive contrast was effectively counteracted by lock-
ing the wheel under EXT and thereby preventing displacement of collateral
wheel running. Wheel running was thus restricted to the VI component and
competed with and decreased the probability of the reinforced operant.
Operant response interactions across multiple schedule components may then
be related to interactions between concurrently available responses within
each schedule component. Displacement and/or differential generation of col-
lateral responses across schedule components would seem to provide at least
some of the possible means by which a manipUlation in one schedule compo-
nent may affect responding in alternate schedule components.
236 Section II: Operant Conditioning Procedures

Multi-response Interactions and Local

Behavioral Contrast (Experiment 111)*
A much debated issue is whether positive contrast is related to a decreased
probability of operant responses or operant reinforcement in an adjacent com-
ponent (Dunham, 1968; Freeman, 1972). An extensively used experimental
strategy has been either to decrease the operant response probability in one
component while maintaining a constant reinforcement rate, or to decrease the
reinforcement rate while maintaining a near zero probability of the operant re- .
sponse in that component. In common, both manipulations entail a low operant
probability in the manipulated component. The purpose of Experiment III was
to attempt a complementary design: (1) to decrease the reinforcement rate in
one component while maintaining a relatively high operant response probability
in that component, and (2) to decrease the operant response probability to zero
while maintaining a zero reinforcement rate.
A second purpose was to analyze local changes for both operant and collat-
eral responses. The experiment purposely used an operant response with an
above zero baseline probability-wheel running in rats. Wheel running was
simply reinforced on VI, and food cup activity, grooming, and exploration were
recorded simultaneously as collateral responses.
Three naive Spraque-Dawley rats, maintained at 80% of their free-feeding
body weights, were given daily sessions with reinforcement of wheel running
by 45-mg food pellets. The apparatus described in Experiment II was also used
in this experiment. Physical contact and activity within 0.1 cm of the food cup
was recorded with a modified body-capacitance sensitivity system. Grooming
and exploration were recorded manually for the two last sessions of each ex-
perimental phase by a button being pressed for each response. The rats were
viewed from above through a one-way window in the side of the sound-atten-
uating chamber by means of mirror arrangements. Grooming was recorded
when a rat scratched or licked his body. Exploration was recorded whenever a
rat was standing up or sniffing the surface or walls of the running wheel. A slow
"wheel-walking" response frequently occurred with sniffing the bars of the
wheel. However, wheel running was only recorded when the circumferential
speed of the turning wheel was above 15.0 em/sec. Wheel running, food cup
activity, grooming, and exploration were recorded in duration. Pulses from a
continuously operating recycling timer (time base 0.6 sec) were counted as long
as the recording device was activated for each each response. The frequency of
changeover responses or switching between the recorded responses was also
recorded.
After pretraining, wheel running was reinforced on VI 1 min for 16 sessions
(Phase A). The houselight was on during the 42-min sessions, and the wheel
was locked before and after each session. In Phase B, the VI was changed to a
mult VI EXT with strictly alternating 3-min components (20 sessions). The

* Experiment was conducted at the University of Zurich, from February to April 1975, sup-
ported by the European Brain and Behavior Society. The author thanks Dr. J. P. Huston for the use
of his laboratory facilities.
Chapter 4: Concurrent Responses with Multiple Schedules 237

houselight was on during VI and flickered with on-off periods of OJ sec under
EXT. The flickering houselight in EXT was accompanied by a clicking sound
from the relay controlling the houselight. In Phase C, the schedule remained
mull VI EXT as in Phase B, but the wheel was locked under EXT (10 sessions).
In Phase D, response-independent pellets were delivered under VT 1 min in the
component with the locked wheel (the former EXT component; 10 sessions).
The schedule was thus mull VI VT, with the wheel locked under VT. Finally, in
Phase E, wheel running was again maintained on VI throughout a session (6
sessions).

Discussion
OveraU Contrast
The results suggest that an increase in the average probability of operant re-
sponding in one component (positive contrast) may be partially related to the
operant response probability in the adjacent component. Operant wheel run-
ning first increased under VI in mull VI EXT when the response probability
was experimentally decreased to zero under EXT (Figure 4.5). Furthermore,
this positive contrast remained apparent when the reinforcement rate was again
increased in the adjacent component with VT but the response probability was
maintained at zero. A marked decrease in operant responding in one compo-
nent thus may override a decrease in the reinforcement rate in the production
of positive contrast. Indeed, within the procedures used in this experiment, a
decreased rate of reinforcement in one component was not necessary for posi-
tive contrast.
The present data closely parallel the findings of a previous report using
wheel running as the operant response with mult VR VR (D. R. Williams,
1965). Previous reports also agree in showing a diminished positive contrast in
mult VI EXT when the probability of the operant response is only moderately
decreased under EXT (Mackintosh et al., 1972; Halliday and Boakes, 1974).
The results of this experiment seem to fit with a model of contrast emphasiz-
ing interactions between concurrently occurring responses in each schedule
component. The results of Experiment II suggested that positive contrast is as-
sociated with displacement of a collateral response that previously occurred in
the VI component. Such displacement of a collateral response from VI to EXT
would also appear to involve a low operant response probability under EXT.
This possibility was tested by purposely manipulating the operant probability
under EXT.
Exploration did not shift out of the VI component and positive contrast was
not obtained when the operant response still occurred at a moderate probability
in the EXT component (Phase B) (Figure 4.5). Exploration first decreased
under VI when the operant response probability under EXT was manipulated
to zero, allowing for a further increase in exploration under EXT (Phase C).
This displacement of exploration, and the accompanying positive contrast for
operant wheel running, were clearly maintained in spite of an increased rein-
forcement rate in the adjacent component conditional upon a maintained zero
operant probability (Phase D). The low operant probability under the adjacent
238 Section II: Operant Conditioning Procedures

RAT 1 wheellock!d RAT 2

VI VI EXT VI ~T VI VT VI

0.8
0.6
WHEEL
RUNNING O.L
0.2
M IT I, 11 11 If ~ It II Ii
FOOD-CUP

- ~
0.2
ACTIVITY
n n. ro.. ...... ..., n.. ...... n

GROOMING 0.2

0.8
• I • -- • -
O.S
>
t- EXPLORING
0.4
::J
m 0.2
n r r r- Q n r r r n
~ PROS. SUM' 0.95 096· 094' 095: 0.90 0.94 0.95 : 0.97 : 0.97; Q94
0
a:: 0.96 0.98 0.97 0.97 0.99 0.95
CL.

w RAT 3
U'l
Z
~ 0.8
U'l 0.6
W WHEEL
a:: RUNNING O.L
0.2 10
It I, t It Ir
FOOD-CUP 0.2
ACTIVITY
r-t n. ...- ...I ......

--
GROOMING 0.2

o.e
0.6
EXPLORING
O.L
0.2
n ..I.- ~

PROS. SUM : 0.97 0.98 : 0..98 : 0.99: 0.ge

0.97 099 093

Figure 4.5. Probabilities of four responses for each rat and experimental phase. Sums of the proba-
bilities are shown below each column. Changeover rates between responses are inserted in the col-
umns for wheel running. Data are means obtained during successive 3-min periods for Phases A
and E, and during the 3-min components for Phases B, C, and D. White bars-VI; stippled bars-
EXT or VT.

component would therefore seem to be one of the variables controlling the

component shift of collateral responses, which in tum affects the operant re-
sponse under the VI component.
In addition to the operant contrast effects and displacement of exploration,
changes in grooming and food cup activity under the adjacent component were
also related to both the operant response and the reinforcement rate manipula-
Chapter 4: Concurrent Responses with Multiple Schedules 239

tions. Thus, grooming increased under EXT when the operant response was
decreased in spite of a maintained zero rate of reinforcement (Phases B to C).
However, grooming decreased in the adjacent component when the associated
schedule was changed from EXT to VT with a maintained zero probability of
the operant response (Phase D). Food cup activity was primarily "tied" to the
schedule components with reinforcement. At the same time, however, the
probability of food cup activity was also dependent upon the simultaneous
operant response probability; that is, for rats 2 and 3, food cup activity oc-
curred at a relatively higher probability in the component with a zero operant
probability when reinforcement rate was equal in the two components (Phase
D).
The shifting response changes across components and phases entailed
changes in an additional set of responses-changeovers between the recorded
responses. The rate of changeover responses under VI components was
roughly inversely related to the probability of operant responding (Figure 4.5).
Furthermore, the summed probability of wheel running, food cup activity,
grooming, and exploration appeared to approach 1.0 as the rate of changeover
responses decreased toward zero (Figure 4.5). Changes in the summed proba-
bility across phases may also be plotted against changes in the changeover rate
(Figure 4.6). The summed probability increased for decreasing changeover
rates and decreased for increasing changeover rates. Therefore, although an at-
tempt was made to record and account for all responses, the summed probabili-
ties did not quite total 1.0. However, such "pausing" in recorded responses
was systematically related to the additional response of changing over among
the recorded responses.
This finding is immediately relevant to the analysis of classical conditioning
presented in Chapter 6. The over- and underestimation of the probability of one
response from the probability of concurrent responses was gradually dimin-
ished as more responses were included in the analysis. Clearly, accuracy in

SUM DIFFERENCE
(increase)
.06

.OL
o
.02

6 CO RATE
- ..........--+--+--+---+---+>-Itf4--+-+-+--..-.....,..........--+ DI FFERE NCE
-8 -6 -L
(increase)
(decrease)
- .02 •
o
-.OL o

-.06
(decrease)
Figure 4.6. Changes in the summed probability of all responses plotted against alterations in
changeover rate for each phase, schedule component, and rat. The regression line was determined
by the method of least squares. Black circles, rat I; white circles, rat 2; x, rat 3.
240 Section II: Operant Conditioning Procedures

prediction of behavioral changes is greatly augmented by the additional record-

ing of all concurrently occurring responses.

Local Contrast
In recent years, the analysis of positive behavioral contrast has become con-
cerned with the more minute response changes within components of multiple
schedules. Thus, operant responding may gradually decrease within the tem-
porallimits of one schedule component that is followed by a second component
associated with a relatively lower reinforcement rate (local behavioral contrast;
Boneau and Axelrod, 1962; Catania and Gill, 1964; Nevin and Shettleworth,
1966; Terrace, 1966; Bernheim and Williams, 1967; Staddon, 1969; Gonzalez
and Champlin, 1974).
To this point the concurrent response analysis has revealed that overall con-
trast appears to be related to a simultaneous and opposing behavioral contrast
for collateral responding. In an extension of this analysis, a significant question
arises: How does local contrast for operant responding relate to collateral re-
sponding?
Figure 4.7 presents an analysis of the recorded responses in successive 15-
sec intervals within each mUltiple schedule component for Phases B, C, and D.
The local probability of operant wheel running initially increased and then grad-
ually decreased within the VI component of mult VI EXT for rats 1 and 2
(Phases B and C). This finding is clearly in accord with the results from the
previous experiments. However, in some experiments with apparently similar
procedures, the pattern of local contrast may be reversed-there may be an
initial decrease in the operant probability followed by a gradual increase within
VI (Ferster, 1958; Buck et al., 1975). Local positive contrast may also occur in
the absence of overall positive contrast, as in Phase B; this finding is confirmed
by recent data (Gutman and Minor, 1976; B. A. Williams, 1976). In addition,
overall positive contrast may occur without local contrast, as the local operant
response probability was relatively constant within the VI component of the
mult VI VT schedule (Phase D). Consequently, local positive contrast may be
obtained with or without overall positive contrast, and vice versa.
We suspect that within-component or local operant contrast is intimately as-
sociated with opposing local contrast effects for collateral responses. The local
distribution of exploration, and to some extent food cup activity, in the VI com-
ponent was thus inversely related to the distribution of the operant response for
the two subjects with local positive contrast for the operant response (rats 1
and 2). Hence, local positive contrast for operant responding is clearly asso-
ciated with negative local contrast for one or more collateral responses. Of par-
ticular interest is the effect of the introduction of response-independent rein-
forcement into the component with the wheel locked (Phase D). The local
probabilities of food cup activity and exploration then became fairly constant
within the VI component, as did the corresponding local operant probability.
Therefore, the reverse relationship between contrast for operant and collateral
responding in one component appears to be under the control of reinforcement
factors in the adjacent component. In agreement with this conclusion, B. A.
 A B C o E
RAT 1

--....,.
VI VI (XT VI EXT VI VT VI
l ock"d tockld

~
8 whul wh •• t
WHEn 0
., t
0.ll.
~ •
I~
RUNNI NG

Ql~

2·
.... A -\.

A
0.6
GROOlol ING 0
•
o1
1
.Av\. ~

£~P\.ORING
0..8
o.6
o.t.
o2-,
r-Y'
\.y-
~
Iv ..... J...
~

•
RAT 2

0.
8 t \tvv ~ ~

.
/'
WHEEL 0~
~ RUNNING 0

::::; Q,~
in
'" moo-cup Vv-'v ,
III ACT IVI TY 0,2
y...-.-- ~ I\.,., .....
~
a... 0. 6
GROOIoIING 0
•
UJ
en
o1
A i~ .-

V
z
~ o,8
~
II)
~
EXPLOR ING 0
0,t.-
.6
\{v-
Q.2-
~ ---.-v --~
t

RAT 3

(
08
...-....- ~
•
WHEEl 0.6
RUNNING
O.
Ol

rooo-cup v..;v....
ACTIVITY
02
~I' ~ A- I\.-...
02
I~
GROO,",ING

..-r-
~ h

oe ..yyv-
06
(xPLORINV
D.
02

60 120 eo
~ 60 120 180 60 120 180 60 120 leo 60 120 180 60 120 180
COMPONENT TIME luconds'

Figure 4.7. Local interations between wheel running, food cup activity , grooming, and exploration
in successive IS-sec intervals within each schedule component in Phases B, C, and D. Means and
absolute ranges of the local probabilities for Phases A and E (VI alone) are shown in the left-hand
and right-hand columns, respectively.

241
242 Section II: Operant Conditioning Procedures

Williams (1976) has found that the distribution of operant responding in one
component may depend upon the temporal location of reinforcement in the ad-
jacent component.
Perhaps a relatively unheeded finding is a local negative contrast for the
operant response under EXT. In this experiment, the local operant probability
initially decreased and then gradually increased under EXT (Phase B). D. R.
Williams (1965) found a similarly increasing probability profile for operant
wheel running within the component with the lower reinforcement rate (also
see Malone, 1976). The collateral response changes were more complex under
the components with extinction or response-independent reinforcement. First,
local positive contrast appeared for grooming and exploration (and to some ex-
tent for food cup activity), along with the local negative contrast for the operant
response under EXT (Phase B). Second, the pattern of exploration reversed,
from local positive to local negative contrast, when the operant response was
decreased to zero under EXT (Phase C). Finally, all local probabilities of collat-
eral responses were relatively constant with equal rates of reinforcement under
the two schedule components.
Little doubt, the present findings only begin to explore how responding in
one schedule component is affected by events in a temporally adjacent compo-
nent. However, the available data do demonstrate continuous associations be-
tween local changes in operant and collateral responses. Moreover, the specific
response patterns were modifiable by environmental manipulations. The close
response interrelationships suggest that a single-response analysis is an insuffi-
cient description of mUltiple schedule effects; rather, the apparent simultane-
ous changes in local patterns of operant and collateral responses within and
across multiple schedule components seem to argue for a multiresponse analy-
sis.

Concluding Comments
This series of experiments seems to provide considerable evidence for an ex-
tension of the concurrent response analysis to encompass sequential operant
schedules. More specifically, alterations in the probabilities of collateral re-
sponses systematically affect the operant response probabilities in mUltiple
schedules. Foremost, perhaps, a particular interaction-positive contrast-
may be directly dependent upon differentiation of specific collateral responses
across the multiple schedule components.
The findings generally agree with several previous experiments but appear to
be incongruent with some previous theoretical accounts of positive contrast.
The principles and predictions of the theories might then be analyzed in terms
of the available data.
A direct relationship between positive contrast for operant responding and
collateral wheel running (Experiment II) seems to suggest that alterations in
specific collateral responses may at least override the effect of differential rein-
forcement rates. Further, in Experiment III, positive contrast was not obtained
Chapter 4: Concurrent Responses with Multiple Schedules 243

with EXT in the adjacent component until the probability of the operant re-
sponse was also reduced to zero. In addition, positive operant contrast was
maintained by equal rates of reinforcement across components, conditional
upon the absence of the operant response in the adjacent component. The pres-
ent results thus firmly support previous suggestions that changes in reinforce-
ment rate in an adjacent component are neither necessary nor sufficient for pos-
itive operant contrast. However, local contrast effects are more supportive of a
reinforcement interpretation. Local positive contrast was obtained with a zero
reinforcement rate in the adjacent component but was not clearly apparent with
equal reinforcement rates across components. Therefore, local contrast may be
related to reinforcement rate under an adjacent component. However, in a re-
cent experiment (B. A. Williams, 1976) local positive contrast systematically
depended upon the temporal pattern rather than the overall reinforcement rate
under an adjacent component. The pattern of reinforcement within components
may then be one of the variables controlling the close associations between
local contrast effects for operant and collateral responses.
Positive operant contrast has been said to come about when one schedule is
relatively more preferred than the schedule in an adjacent component (Bloom-
field, 1969; Premack, 1969). The present data are not easily compared to the
preference account because preference among schedule components was not
directly assessed in the present experiments. Extrapolating from previous ex-
periments, VI components would in general be expected to be preferred to
EXT components (Findley, 1958; Beale and Winton, 1970). However, positive
contrast was more dependent upon the pattern of collateral wheel running than
upon the schedule components per se with mult VI EXT (Experiment 11).
Therefore, the results may agree with previous findings that relatively less pref-
erence for an adjacent component may not be sufficient for positive contrast.
More formally, the preference account rests upon the validity of the assump-
tion that relative response rates in choice situations measure preference.
If positive operant contrast is dependent upon aversiveness generated by
emission of nonreinforced responses (errors), then the extent of positive con-
trast may be assumed to depend upon the probability of nonreinforced re-
sponses (Terrace, 1966). However, in Experiment III positive contrast was in
fact only obtained when the operant response probability was decreased to
zero under EXT. Moreover, in Experiment II positive contrast was dependent
upon the distribution of collateral wheel running, given the same low probabil-
ity of operant responding under EXT.
The present findings are consistent with reports of collateral response such
as wing flapping, aggression, and schedule escape generated by contrast proce-
dures. Such responses have been offered as "measures" of aversiveness or
frustration generated under an adjacent component in association with positive
operant contrast (Terrace, 1971, 1972; Rilling and Caplan, 1973; Rilling et al.,
1973). Nevertheless, a single collateral response, taken out of the collective be-
havioral context, is not unequivocally associated with positive contrast. In Ex-
periment III, for example, the increased probability of grooming under EXT
was not necessarily associated with positive operant contrast. Similarly, Ex-
244 Section II: Operant Conditioning Procedures

periment II found that positive operant contrast was not covariant with collat-
eral wheel running under EXT. The present results thus agree with previous
experiments (e.g., Terrace, 1971; Coughlin, 1973; Rilling and Caplan, 1973) that
individual collateral responses are not easily employed as indicants of aversive-
ness or frustration processes taking place between independent and dependent
variables; rather, collateral responses seem to be collectively interrelated with
operant responses and may actively participate in the development of positive
operant contrast.
The response additivity model holds that positive operant contrast results
when stimulus-directed responses, generated by differential reinforcement
rates across schedule components, add to the operant response by activating
the operant manipulandum. However, positive contrast was obtained in the
present experiments without the discriminative stimuli located on the operant
manipulandum, and with equal reinforcement rates in the schedule compo-
nents.
The present data agree better with a second aspect of response additivity-
that positive contrast may not occur when stimulus responses are directed to-
ward a second manipulandum. Experiments by Keller (1974), Schwartz (1975),
and Spealman (1976) clearly show extensive interactions between operant re-
sponses and topographically different stimulus-directed responses. These find-
ings are consistent with Experiment II in that an increase in collateral wheel
running during VI (produced by preventing wheel running in the adjacent com-
ponent with EXT) may decrease the concurrent probability of the operant re-
sponse and prevent positive contrast. Rilling and Caplan (1975) similarly found
that positive operant contrast may be prevented when the procedures generate
collateral attack responses concurrently with the operant response in VI of
mult VI EXT. Concurrent responses generated within one schedule component
may then compete with (or actually subtract from) the operant response in that
component and thereby block any operant response probability. Thus, al-
though positive contrast may not invariably result from an addition of stimulus-
directed responses to the operant manipulandum, the dislocation of the dis-
criminative stimuli away from the operant manipulandum may generate topo-
graphically different responses that compete with and prevent positive operant
contrast.

A Response-Displacement Model of Positive Contrast?

The experimental results might suggest a novel model emphasizing the dis-
placement of collateral responses in contrast experiments. Experiment II in
particular showed that positive contrast was associated with displacement of
collateral wheel running across schedule components. Positive operant con-
trast was then effectively eliminated by prevention of the displacement of col-
lateral wheel running across the different components.
Experiment III suggested a similar mechanism, but only more weakly. A
moderate decrease in operant wheel running under EXT allowed some dis-
placement of collateral grooming, but the displacement of grooming was nu-
merically small and did not significantly affect the concurrent probability of the
Chapter 4: Concurrent Responses with Multiple Schedules 245

operant response under VI. Similarly, exploration was clearly displaced into
the EXT component, allowing for positive contrast for operant responding,
only when the competing operant wheel running was prevented under EXT. In
combination, the interactions suggest that a decreased operant probability
under one component may allow for displacement of some collateral responses
toward that component and away from an alternate component with a relatively
higher probability of operant responding. As a result, the operant response may
further increase in the alternate component because of the reduced probability
of collateral responses. To speak loosely, a given collateral response may be
displaced from one to another component when the time available for respond-
ing is relatively less restricted during the second component.
In the present experiments, a prerequisite for displacement of collateral re-
sponses from one component to a second seemed to be a decrease in operant
responding during the second component. This is clearly a limiting point for an
eventual general model of contrast favoring displacement of collateral re-
sponses. First, previous experiments have shown that positive contrast may
occur when the operant probability is not appreciably decreased under an adja-
cent component, as with delayed reinforcement (Wilkie, 1971; Richards, 1972).
Second, positive contrast may not occur in spite of a decreased operant proba-
bility under an adjacent component, as with certain procedures of errorless dis-
crimination training (Terrace, 1%3), response-independent reinforcement (Hal-
liday and Boakes, 1971) or after drug injections or brain lesions (Bloomfield,
1972; Henke et aI., 1972). Therefore, compared across experiments, the mean
decrease in operant responses in an adjacent component appears to be neither
necessary nor sufficient for positive contrast (also Marcucella, 1976).
The available evidence on mUltiple schedules thus prevents response dis-
placement as yet another global formulation of the sequential interactions
found in all experiments. The present experiments do, however, indicate some
conditions under which collateral responses in one schedule component may
affect the response patterns in a second component, and thereby influence the
operant response within the second component. More firmly, the experiments
establish that operant and collateral responses concurrently interact under
each component of multiple schedules. This analysis may generalize, at least
potentially, across different topographies of collateral responses and also
across the more familiar interactions reported for multiple schedules. An analy-
sis of response interactions seems to provide a worthwhile methodology appli-
cable to prospective investigations of mUltiple schedules.

Collateral Responses in Multiple Schedules

Collateral responses may simply be defined as not-explicitly reinforced re-
sponses that concurrently occur with the specifically reinforced response. Col-
lateral responses may be divided into subcategories, depending upon their rela-
tionship to the experimental parameters and the operant response.
One category is responses that have an above zero baseline probability prior
to the introduction of conditioning procedures. Such collateral responses come
close to the normally understood "prepotent" responses and may include ex-
246 Section II: Operant Conditioning Procedures

ploration, grooming, rearing, or wheel running, for example. Prepotent collat-

eral responses appear to be most susceptible to displacement across compo-
nents of multiple schedules and may shift into the component with the
relatively lower restriction or interference from the operant response. On the
other hand, when the probability of such collateral responses is experimentally
forced to zero in one component, then the same collateral response may shift
into a second component and in tum restrict or reduce the operant response.
The particular pattern shifts and local changes would necessarily be dependent
upon specific parameters and the characteristics of the operant response. In
Experiment III, for example, collateral exploration reversed from positive to
negative local contrast when the operant responding was decreased to zero in
the same component. In addition, the distribution of grooming was lowered and
flattened by equalizing the reinforcement rates in the two schedule compo-
nents. Significantly, displacement of prepotent collateral responses originating
from a change in the operant probability in one component may in tum influ-
ence the operant response in a temporally remote schedule component.
A second category would be the collateral responses specifically induced by
the operant reinforcement procedures. Collateral responses that increase from
a low or zero baseline probability have also been called adjunctive or schedule-
induced responses. The category includes a variety of response topographies,
including water drinking, attack, mirror pecking, and schedule-escape re-
sponses (Falk, 1971). Such schedule induced responses can be further subdi-
vided, according to whether they occur primarily in the schedule component
with reinforcer delivery or the component without reinforcer delivery. Attack,
wing flapping, and schedule-escape responses generally occur during compo-
nents with relatively lower reinforcement rates (Azrin et al., 1966; Terrace,
1966,1971; Rilling et at., 1969). On the other hand, specific components of in-
duced responses may be common to both extinction and reinforcement compo-
nents. Thus Rilling and Caplan (1975) found that attack responses under EXT
generated preattack responses under VI. Water drinking may also occur in ei-
ther the EXT component or the reinforcement component depending upon spe-
cific procedures (compare Keehn and Colotla, 1971; Jacquet, 1972).
A third category, somewhat resembling the second, is collateral responses
generated by the reinforcement procedures but more explicitly "tied" to the
components with reinforcer delivery. This class may include sniffing and lick-
ing at reinforcer delivery systems. Stimulus-directed responses would appear
as "tied" only to schedule components with reinforcement (Schwartz and
Gamzu, 1977). As already mentioned, stimulus-directed responses may effec-
tively increase or decrease the probability of operant responses depending
upon whether they are directed toward or away from the operant manipulan-
dum.

Conclusion
In conclusion, subcategories and labels applied to collateral responses are
clearly related to procedures rather than a priori arguments over inherent re-
Chapter 4: Concurrent Responses with Multiple Schedules 247

sponse properties or values. The same physical response may be operant or

collateral and may enter different subcategories and interactions depending
upon specific parameters. Collateral responses, especially in multiple sched-
ules, is not a useful generic term allowing global generalizations to all similarly
labeled responses. The fact that a particular subcategory of collateral responses
is critically involved with operant responses in one procedural setting does not
necessarily suggest that the same collateral response is similarly involved in
different procedures. Simply, a specific variable may be neither necessary nor
sufficient for positive operant contrast in all experimental contexts. The distri-
bution of operant responses within different parametric contexts may very well
involve different subcategories of collateral responses, which could effectively
limit explanatory generalizations. The more concrete conclusion then is that
collateral response patterns might be more carefully analyzed in relation to
operant changes in multiple schedules.

References
Allen, J. D., and Porter, J. H.: Demonstration of behavioral contrast with adjunctive drinking. Phy-
siol. Behav., 15,511-515, 1975.
Azrin, N. H., Hutchinson, R. R., and Hake, D. F.: Extinction-induced aggression. J. Exp. Anal.
Behav., 9, 191-204, 1966.
Beale, I. L., and Winton, A. S. W.: Inhibitory control in concurrent schedules. J. Exp. Anal.
Behav., 14,133-137,1970.
Beecroft, R. S.: Within-session behavioral contrast? Psychonomic Sci., 17,286-287, 1969.
Bernheim, J. W., and Williams, D. R.: Time-dependent contrast effects in a multiple schedule of
food reinforcement. J. Exp. Anal. Behav., 10,243-249, 1967.
Bloomfield, T. M.: Behavioral contrast and relative reinforcement frequency in two multiple sched-
ules. J. Exp. Anal. Behav., 10, 151-158, 1967.
Bloomfield, T. M.: Behavioral contrast and the peak shift. In, R. M. Gilbert and N. S. Sutherland
(eds.): Animal Discrimination Learning. London and New York, Academic Press, 1969.
Bloomfield, T. M.: Contrast and inhibition in discrimination learning by the pigeon: analysis
through drug effects. Learning Motivation, 3, 162-178, 1972.
Boakes, R. A.: Frequency of houselight interruption as a dimension for inhibitory generalization
testing. Psychonomic Sci., 26,249-251, 1972.
Boneau, C. A., and Axelrod, S.: Work decrement and reminiscence in pigeon operant condition-
ing. J. Exp. Psychol., 64,352-354, 1962.
Bradshaw, C. W.: Behavioral contrast in albino rats. Psychol. Rep., 37,287-291, 1975.
Bradshaw, C. M., Szabadi, E., and Bevan, P.: Behaviour ofrats in mUltiple schedules of response-
contingent and response-independent food presentation. Q. J. Exp. Psychol., 30, 133-139, 1978.
Brownstein, A. J., and Hughes, R. G.: The role of response suppression in behavioral contrast:
signalled reinforcement. Psychonomic Sci., 18, 50-52, 1970.
Brownstein, A. J., and Newsom, C.: Behavioral contrast in mUltiple schedules with equal rein-
forcement rates. Psychonomic Sci., 18,25-26, 1970.
Buck, S. L., Rothstein, B., and Williams, B. A.: A re-examination of local contrast in multiple
schedules. J. Exp. Anal. Behav., 24,291-301,1975.
Carlisle, H. J.: Fixed-ratio polydipsia: termal effects of drinking, pausing, and responding. J.
Compo Physiol. Psychol., 75, 10-22, 1971.
Catania, A. C.: Concurrent operants. In, W. K. Honig (ed.): Operant Behavior: Areas of Research
and Application. New York, Appleton-Century-Crofts, 1966.
Catania, A. C.: Self-inhibiting effects of reinforcement. J. Exp. Anal. Behav., 19,517-526, 1973.
 248 Section II: Operant Conditioning Procedures

Catania, A. C., and Gill, C. A.: Inhibition and behavioral contrast. Psychonomic Sci., 1,257-258,
1964.
Chitwood, P. R., and Griffin, P.: The effects of response prevention via operandum removal in the
S- of a tonal discrimination. Psychonomic Sci., 27,37-38, 1972.
Coates, T. J.: The differential effects of punishment and extinction on behavioral contrast. Psy-
chonomic Sci., 27, 146-148, 1972.
Cohen, P. S., and Looney, T. A.: Schedule-induced mirror responding in the pigeon.J. Exp. Anal.
Behav., 19,395-408, 1973.
Coughlin, R. C.: Timeout from a stimulus correlated with the extinction component of a multiple
schedule. Learning Motivation, 4,294-304, 1973.
Dunham, P. J.: Contrasted conditions of reinforcement: a selective critique. Psychol. Bull., 69,
295-315, 1968.
Dunham, P. J.: Some effects of punishment upon unpunished responding. J. Exp. Anal. Behav., 17,
443-450, 1972.
Falk, J. L.: The nature and determinants of adjunctive behavior. Physiol. Behav., 6, 577 -588, 1971.
Farthing, G. W.: Behavioral contrast in pigeons learning an auditory discrimination. Bull. Psycho-
nomic Soc., 6, 123-125, 1975.
Felton, M., and Lyon, D.O.: The post-reinforcement pause. J. Exp. Anal. Behav., 9, 131-134,
1966.
Ferster, C. B.: Control of behavior in chimpanzees and pigeons by time-out from positive rein-
forcement. Psychol. Monogr., 72,461, 1958.
Ferster, C. B., and Skinner, B. F.: Schedules of Reinforcement. New York, Appleton-Century-
Crofts, 1957.
Findley, J. D.: Preference and switching under concurrent scheduling. J. Exp. Anal. Behav., 1,
123-144, 1958.
Flory, R. F.: Attack behavior in a multiple fixed-ratio schedule of reinforcement. Psychonomic
Sci., 16, 156-157, 1969.
Freeman, B. J.: Behavioral contrast: reinforcement frequency or response suppression? Psychol.
Bull., 75,347-356, 1971.
Friedman, H., and Guttman, N.: Further analysis of the various effects of discrimination training
on stimulus generalization gradients. In, D. I. Mostofsky (ed.): Stimulus Generalization. Stan-
ford, Calif., Stanford University Press, 1965.
Gaffan, D.: Inhibitory gradients and behavioral contrast in rats with lesions of the fornix. Physiol.
Behav., 11,215-220, 1973.
Gamzu, E., and Schwartz, B.: The maintenance of key pecking by stimulus-contingent and
response-independent food presentation. J. Exp. Anal. Behav., 19,65-72, 1973.
Gonzalez, R. C., and Champlin, G.: Positive behavioral contrast, negative simultaneous contrast
and their relation to frustration in pig~ons. J. Compo Physiol. Psychol., 87, 173-187, 1974.
Gutman, A.: Positive contrast, negative induction, and inhibitory stimulus control in the rat. J.
Exp. Anal. Behav., 27,219-233, 1977.
Gutman, A., and Minor, T.: Local positive behavioral contrast in the rat. Psychol. Record, 26,
349-354, 1976.
Gutman, A., Sutterer, J. R., and Brush, F. R.: Positive and negative behavioral contrast in the rat.
J. Exp. Anal. Behav., 23,377-383, 1975.
Halliday, M. S., and Boakes, R. A.: Behavioral contrast and response independent reinforcement.
J. Exp. Anal. Behav., 16,429-434, 1971.
Halliday, M. S., and BOakes, R. A.: Discrimination involving response-independent reinforce-
ment: implications for behavioral contrast. In, R. A. Boakes and M. S. Halliday (eds.): Inhibition
and Learning. London and New York, Academic Press, 1972.
Halliday, M. S., and Boakes, R. A.: Behavioral contrast without response-rate reduction. J. Exp.
Anal. Behav., 22,453-462, 1974.
Hemmes, N. S.: Behavioral contrast in pigeons depends upon the operant. J. Compo Physiol. Psy-
chol., 85, 171-178, 1973.
Henke, P. G.: Septal lesions and aversive nonreward. Physiol. Behav., 17,483-488, 1976.
Henke, P. G., Allen, J. D., and Davison, C.: Effect oflesions in the amygdala on behavioral con-
trast. Physiol. Behav., 8, 173-176, 1972.
Chapter 4: Concurrent Responses with Multiple Schedules 249

Herrnstein, R. J.: On the law of effect. J. Exp. Anal. Behav., 13,243-266, 1970.
Hughes, R. G.: Probability of signalled reinforcement in multiple variable-interval schedules. Psy-
chonomic Sci., 22,57-59, 1971.
Hutchinson, R. R., Azrin, N. H., and Hunt, G. M.: Attack produced by intermittent reinforcement
of a concurrent operant response. J. Exp. Anal. Behav., 11,489-495, 1%8.
Iversen, 1. H.: Behavioral contrast and collateral behavior. Paper presented at the Fourth Scandin-
avian Meeting on Physiological Psychology. Oslo, May 1975a.
Iversen, 1. H.: Concurrent responses during mUltiple schedules in rats. Scand. J. Psychol., 16,49-
54, 1975b.
Jacquet, Y. F.: Schedule-induced licking during multiple schedules. J. Exp. Anal. Behav., 17,413-
423, 1972.
Keehn, J. D., and Bratbak, R. B.: Limitations on environmental control of mUltiple schedule fixed-
ratio behavior. J. Exp. Anal. Behav., 10, 185-190, 1967.
Keehn, J. D., and Colotla, V. A.: Stimulus and subject control of schedule-induced drinking. J.
Exp. Anal. Behav., 16,257-262, 1971.
Keller, K.: The role of elicited responding in behavioral contrast. J. Exp. Anal. Behav., 21,249-
257,1974.
Lander, D. G.: Stimulus control following response reduction with signalled reinforcement. Psy-
chonomic Sci., 23,365-367, 1971.
Mackintosh, N. J., Little, L., and Lord, J.: Some determinants of behavioral contrast in pigeons
and rats. Learning Motivation, 3, 148-161, 1972.
Malone, J. C.: Local contrast and Pavlovian induction. J. Exp. Anal. Behav., 26,425-440, 1976.
Marcucella, H.: Signalled reinforcement and multiple schedules. J. Exp. Anal. Behav., 26, 199-
206, 1976.
Nevin, J. A.: Differential reinforcement and stimulus control of not responding. J. Exp. Anal.
Behav., 11,715-726, 1968.
Nevin, J. A., and Shettleworth, S. J.: An analysis of contrast effects in multiple schedules. J. Exp.
Anal. Behav., 9,305-315, 1966.
Premack, D.: Reinforcement theory. In, D. Levine (ed.): Nebraska Symposium on Motivation.
Lincoln, University of Nebraska Press, 1965.
Premack, D.: On some boundary conditions of contrast. In, J. T. Tapp (ed.): Reinforcement and
Behavior. New York, Academic Press, 1%9.
Rand, J. F.: Behaviors observed during S- in a simple discrimination learning task. J. Exp. Anal.
Behav., 27, 103-117, 1977.
Redford, M. E., and Perkins, C. C.: The role of autopecking in behavioral contrast. J. Exp. Anal.
Behav., 21, 145-150, 1974.
Reynolds, G. S.: An analysis of interactions in a multiple schedule. J. Exp. Anal. Behav., 4, 107-
117, 1%la.
Reynolds, G. S.: Behavioral contrast. J. Exp. Anal. Behav., 4,57-71, 1%lb.
Reynolds, G. S.: Relativity of response rate and reinforcement frequency in a mUltiple schedule. J.
Exp. Anal. Behav., 4, 179-184, 1961c.
Richards, R. W.: Reinforcement delay: some effects on behavioral contrast. J. Exp. Anal. Behav.,
17,381-394, 1972.
Rilling, M.: Stimulus control and inhibitory processes. In, W. K. Honig and J. E. R. Staddon
(eds.): Handbook of Operant Behavior. Englewood Cliffs, New Jersey, Prentice-Hall, 1977.
Rilling, M., and Caplan, H. J.: Extinction-induced aggression during errorless discrimination learn-
ing. J. Exp. Anal. Behav., 20,85-92, 1973.
Rilling, M., and Caplan, H. J.: Frequency of reinforcement as a determinant of extinction-induced
aggression during errorless discrimination learning. J. Exp. Anal. Behav., 23, 121-129,
1975.
Rilling, M., Askew, H. R., Ahlskog, J. E., and Kramer, T. J.: Aversive properties of the negative
stimulus in a successive discrimination. J. Exp. Anal. Behav., 12,917-932, 1969.
Rilling, M., Kramer, T. J., and Richards, R. W.: Aversive properties of the negative stimulus dur-
ing learning with and without errors. Learning Motivation, 4, 1-10, 1973.
Sadowsky, S.: Behavioral contrast and timeout, blackout, or extinction as the negative condition.
J. Exp. Anal. Behav., 19,499-507, 1973.
250 Section II: Operant Conditioning Procedures

Schaeffer, R. W., and Diehl, J. C.: Collateral water drinking in rats maintained on FR food rein-
forcement schedules. Psychonomic Sci., 4,257-258, 1%6.
Schwartz, B.: Discriminative stimulus location as a determinant of positive and negative behavioral
contrast in the pigeon. J. Exp. Anal. Behav., 23, 167-176, 1975.
Schwartz, B., and Gamzu, E.: Pavlovian control of operant behavior. In, W. K. Honig and J. E. R.
Staddon (eds.): Handbook of Operant Behavior. Englewood Cliffs, New Jersey, Prentice-Hall,
1977.
Schwartz, B., Hamilton, B., and Silberberg, A.: Behavioral contrast in the pigeon: a study of the
duration of key pecking maintained on multiple schedules of reinforcement. J. Exp. Anal.
Behav., 24, 199-206, 1975.
Skinner, B. F., and Morse, W. H.: Concurrent activity under fixed-interval reinforcement. J.
Compo Physiol. Psychol., 50,279-281, 1957.
Spealman, R. D.: Interactions in multiple schedul.es: the role of the stimulus-reinforcer contin-
gency. J. Exp. Anal. Behav., 26,79-93, 1976.
Staddon, J. E. R.: Multiple fixed-interval schedules: transient contrast and temporal inhibition. J.
Exp. Anal. Behav., 12,583-590, 1969.
Terrace, H. S.: Discrimination learning with and without "errors." J. Exp. Anal. Behav., 6, 1-27,
1963.
Terrace, H. S.: Stimulus control. In, W. K. Honig (ed.): Operant Behavior: Areas ofResearch and
Application. New York, Appleton-Century-Crofts, 1%6.
Terrace, H. S.: Escape from S-. Learning Motivation, 2, 148-163, 1971.
Terrace, H. S.: Conditioned inhibition in successive discrimination learning. In, R. A. Boakes and
M. S. Halliday (eds.): Inhibition and Learning. London, Academic Press, 1972.
Vieth, A., and Rilling, M.: Comparison of time-out and extinction as determinants of behavioral
contrast: an analysis of sequential effects. Psychonomic Sci., 27,281-282, 1972.
Wilkie, D. M.: Delayed reinforcement in a multiple schedule. J. Exp. Anal. Behav., 16,233-239,
1971.
Wilkie, D. M.: Signalled reinforcement in mUltiple and concurrent schedules. J. Exp. Anal. Behav.,
20,29-36, 1973.
Wilkie, D. M.: Behavioral contrast produced by a signalled decrease in local rate of reinforcement.
Learning Motivation, 8, 182-193, 1977.
Williams, B. A.: Behavioral contrast as a function of the temporal location of reinforcement. J.
Exp. Anal. Behav., 26,57-64, 1976.
Williams, D. R.: Negative induction in instrumental behavior reinforced by central stimulation.
Psychonomic Sci .. 2.341-342, 1%5.
Chapter 5

Collateral Responses with

Simple Schedules
Iver H. Iversen

Reinforced Responses Interacting with Collateral Responses

In the original formulation of an experimental analysis of behavior, Skinner
(1938) laid out the principal problem to be the establishment of laws for individ-
ual responses (reflexes). Skinner hastened to add that the remaining part of the
field was "the interaction of separate reflexes" (1938, p. 46). Models favoring
competition among behaviors, however, have not been particularly attractive,
which is somewhat surprising, given the awareness among the first experimen-
tal analysts of such a phenomenon. Initially, experimental subjects were
merely positioned on a table or in an open box with stimuli and objects deliv-
ered to the subjects more or less by hand. The investigatory reflexes elicited by
even slight noise, odor, light, or temperature changes were well known to "in-
hibit" the conditioned reflexes. Thus, behavior competition was a serious
everyday problem, which in fact must have prevented or at best delayed collec-
tion of replicable data. Thus, Pavlov wrote:
In our old laboratory the neglect to provide against external stimuli often led to a curi-
ous complication when I visited some of my co-workers. Having by himself estab-
lished a new conditioned reflex, working in the room with the dog, the experimenter
would invite me for a demonstration, and then everything would go wrong and he
would be unable to show anything at all. It was I who presented the extra stimulus:
the investigatory reflex was immediately brought into play: the dog gazed at me, and
smelled at me, and of course this was sufficient to inhibit every recently established
reflex. (pavlov, 1927, p. 45)
Eventually subjects and experimenters were situated in separate rooms with
stimuli and objects delivered by more remote control. Skinner pleaded for a
similar control in the establishment of laws of individual reflexes: "a first pre-
252 Section II: Operant Conditioning Procedures

caution is the removal of stimuli which elicit other reflexes . . . Not all such
stimuli can be removed, but a nearly maximal isolation can be achieved by con-
ducting the experiments in a sound-proof, dark, smooth-walled, and well-venti-
lated room" (1938, p. 55).
Until recent times, these arguments were not markedly altered. Thus, in the
introductory remarks to a first handbook of operant behavior, Honig wrote:
In operant work, control of the environment facilitates the concentration upon one
kind of response by removing the opportunity for strong competing behaviors. While
psychology is the science of behavior, it is naive to believe that "all" the behaviors in
a situation must be observed and recorded (Honig, 1966, p. 7).

Experimental equipment has been held tightly closed with an automatic rec-
ord of only the response required for reinforcer delivery. Although the investi-
gated single response was not really believed to occur in a behavioral vacuum,
"other" behaviors were assumed to remain constant or at best be homogen-
eously distributed between occurrences of the reference response. In the past
two decades, however, several reports have described significant changes in
various behaviors occurring with singly reinforced responses.
After observation of behaving organisms, Breland and Breland (1961) thus
reported several examples of how "instinctive" behaviors intrude and interfere
with particular conditioned responses. The simple response of merely drinking
water after food reward has recently received considerable attention, but other
behaviors have also been found to occur regularly with responses reinforced on
various schedules, the so-called adjunctive or schedule-induced behaviors
(Falk, 1969, 1971; Staddon and Simmelhag, 1971; Segal, 1972).
Moreover, adjunctive or collateral responses do not occur haphazardly
within reinforcement schedules; their rates and patterns seem to vary systemat-
ically with manipulated schedule parameters. Even in a simple environment
with only "one" experimental parameter, several topographically different be-
haviors may change rate and pattern simultaneously. Thus, an experimental en-
closure that excludes extraneous stimuli does not guarantee an elimination or
"stabilization" of competing behaviors. Perhaps collateral responses signifi-
cantly participate in the formation or patterning of the reinforced operant re-
sponse.
Recently several reports have provided simultaneous recordings of collat-
eral responses and reinforced responses. For example, the rate of a reinforced
response and the rate of collateral licking may change in opposite directions
after brain damage (Wayner and Greenberg, 1972), drug injections (Wuttke and
Innis, 1972), ambient temperature changes (Carlisle, 1973), or changes in re-
sponse contingencies (Dunham, 1971; Iversen, 1975b). Such reports have occa-
sionally been received with some contempt and have been described as
"chicken and egg" experiments. What affects what, the reinforced response or
the collateral response? However, if a particular problem of "causality" be-
tween collateral and reinforced responses is apparent within one operant
schedule, then similar problems must exist in similar applications.
Kuhn (1970) argues that a science develops in accordance with paradigm
Chapter 5: Collateral Responses with Simple Schedules 253

shifts. A paradigm shift occurs when a "new" or hitherto insufficiently empha-

sized variable proves to be a powerful determinant of familiar findings. Previ-
ous methodological strategies and theoretical models must then be revised to
accommodate the new functional relationships. For example, the discovery of
x-rays by Roentgen in 1895 led to a change in the apparatus previously used in
radiation studies. In the words of Kuhn:
Though x-rays were not prohibited by established theory, they violated deeply en-
trenched expectations. Those expectations, I suggest, were implicit in the design and
interpretation of established laboratory procedures. By the 1890's cathode ray equip-
ment was widely deployed in numerous European laboratories. If Roentgen's appa-
ratus had produced x-rays, then a number of other experimentalists must for some
time have been producing those rays without knowing it. Perhaps those rays, which
might well have other unacknowledged sources too, were implicated in behavior pre-
viously explained without reference to them. At the very least, several sorts of long
familiar apparatus would in the future have to be shielded with lead. Previously com-
pleted work on normal projects would now have to be done again because earlier sci-
entists had failed to recognize and control a relevant variable. (Kuhn, 1970, p. 59)

A novel behavioral paradigm might recognize competition among behaviors

as an important principle that should not be excluded from experimental evalu-
ation. Consequently, behavior would not be viewed in isolation but within a
behavioral context.
Inverse raltionships among simultaneously available behaviors have often
been seen as only indirect competition. For example, Hull suggested:
If the two stimulating situations are presented simultaneously or in close succession
in such a way that the acts of striving for one preclude the simultaneous performance
of the acts involved in striving for the other, there arises a competition within the
body of the organism and that reaction potential which is momentarily greater me-
diates the corresponding reaction. (Hall, 1952, p. 331)

Another interpretation offered that proprioceptive stimuli from one response

become a signal for nonreinforcement for other responses, and thereby exert
conditioned inhibitory influences on other responses (Berlyne, 1960). With re-
spect to conditioned inhibition, Hearst wrote:
In my opinion, some but by no means all of the decrements produced by CIs [condi-
tioned inhibitory stimuli] can be attributed to the development of specific motor re-
sponses that compete with or are antagonistic to the measured CR [conditioned re-
sponse]. In addition to this specific source of response reduction, CIs probably also
evoke more general expectancies or emotional states that mediate CR decrement.
(Hearst, 1972, p. 21)

Direct competition among responses however, has been considered in spe-

cific as well as general explanations of behavior. Admittedly, behavioral
models emphasizing response competition have most often been more or less
provocative discourses (Guthrie, 1935) without specific recordings and analysis
of competing behaviors. The heuristic character of competing response models
is perhaps responsible for the diverse references to responses that concurrently
occur with a recorded response. Such responses are thus said to be adjunctive,
254 Section II: Operant Conditioning Procedures

agitated, alternate, antagonistic, anticipatory, catalyzing, collateral, conflict-

ing, displaced, emotional, extraneous, irrelevant, intercurrent, interfering, in-
terim, intervening, mediating, schedule-induced, or superstitious.
Most recently, Schoenfeld and associates have developed a theoretical treat-
ment that emphasizes not only the temporal relationships between responses
and events, but also the fact that recorded responses are embedded in a "be-
havioral stream" (e.g., Schoenfeld and Farmer, 1970). In a further develop-
ment, at least one contemporary competing response model has explicitly for-
mulated a program for a multiresponse methodology. Thus Dunham (1971,
1972) demonstrated that prevention of one response may increase the probabil-
ity of a concurrent response. Dunham suggested that changes in one response
may be proportional to changes in concurrent responses, so that each specific
response occurs in a fixed proportion of the available time for that response.
Of equal interest, several experiments suggested that the patterning of re-
sponses reinforced on DRL schedules may be closely dependent upon the si-
multaneous pattern of specific collateral responses (Hodos et al., 1962; Laties
et al., 1965; Laties et al., 1969). In addition, collateral responses have been
manipUlated in FI, FR, and VI schedules with a resulting alteration in the rate
and pattern of reinforced responding (Skinner and Morse, 1957; Clark, 1962;
Segal and Bandt, 1966; Wayner and Greenberg, 1972; Gilbert, 1974; Colotla
and Keehn, 1975; Cook and Singer, 1976; Iversen, 1976). Thus the patterning of
reinforced responding appears to be related to collateral response characteris-
tics in addition to schedule variables.
The purpose of the following experiments was to systematically amplify the
previous analysis of response interactions within simple reinforcement sched-
ules. More inductively, the experiments described in the preceding chapters
predict that reinforced and collateral responses are also mutually dependent in
simple schedules. The present experiments therefore were undertaken to pro-
vide a quantitative analysis of the molecular patterns and interactions between
reinforced and collateral responses.

Pattern Reciprocity Between FR Responding and

Collateral Licking (Experiments I and ll)*
With periodic schedules of reinforcement, periods of inactivity typically appear
after reinforcer delivery. These so-called postreinforcement pauses have been
the subject of extensive investigation and explanation (Killeen, 1975). Under
FR schedules, the postreinforcement pause has been manipulated by parame-
ters such as ratio size, reinforcer deprivation, reinforcer size, drugs, and by
adding time-out, shock, or conditioned reinforcers during the ratio (Ferster and
Skinner, 1957; Dardano and Sauerbrunn, 1964; Findley and Brady, 1965;
Boren, 1966; Felton and Lyon, 1966; Powell, 1968, 1969; Barowsky and Mintz).

* Experiments were conducted at the University of Copenhagen, from October 1971 to March
1972, and reported by I. Iversen (1973, 1975a).
Chapter 5: Collateral Responses with Simple Schedules 255

A variety of collateral responses, including attack, escape, mirror pecking,

water drinking, and wheel running, have now been recorded under FR as well
as FI schedules (Falk, 1971; Staddon and Simmelhag, 1971; Segal, 1972; Cohen
and Looney, 1973) and often reveal closer relationships between the patterns of
reinforced and collateral responses.

Experiment I
In Experiment I (Iversen, 1975a), the FR response pattern was thus clearly as-
sociated with the patterning of a particular collateral licking response. Pressing
one lever by rats was reinforced with a 20% sucrose solution on FR 40, and
responses on a second lever were reinforced on FR 20 in the next component of
a mUltiple schedule. The components alternated every 3 min independent of re-
sponding.
Figure 5.1 shows sample event records oflever pressing, licking the sucrose
delivery tube, and sucrose delivery for four successive components of the mul-
tiple schedule for one rat. Similar records were obtained for two other rats. The
postreinforcement pauses in lever pressing were clearly longer under FR 40
than under FR 20, which agrees with previous reports that FR pauses are an
increasing function of the FR size (Ferster and Skinner, 1957; Felton and Lyon,
1966). In both schedule components, licking the sucrose delivery tube occurred
during consumption of the reinforcer, but also frequently extended well past
sucrose delivery. Such lick bursts were considerably longer following FR 40
reinforcement than following FR 20 reinforcement. This differentiation of col-
lateral licking is consistent with previous reports that the duration of specific
collateral responses is an increasing function of the FR size (Schaeffer and
Diehl, 1966; Hutchinson et al., 1968; Flory, 1969; Carlisle, 1971). Figure 5.1
also reveals that collateral licking alternated with presses on the other (nonrein-
forced) lever only under FR 40. Quite simply, the data demonstrate reciprocal
relations between reinforced and collateral responses. Similar interrelations be-
tween collateral attack and FR pauses were reported by Huston and Desisto
(1971). In addition, Carlisle (1973) found relatively longer FR pauses in associa-
tion with bursts of collateral grooming.

Experiment II
The precise relationship between FR pauses and bursts of collateral licking was
next examined as a function of the FR size (Iversen, 1973). Lever pressing by
rats was maintained with a 20% sucrose solution delivered on a progression of
FR schedules (FR 1 to FR 30). Figure 5.2 shows scatter plots between individ-
ual pauses and lick bursts after reinforcer delivery for each FR size for one rat.
Similar data were obtained for two other rats. The FR pause and duration of
collateral licking were clearly increasing functions of the response requirement.
For all FR sizes, relatively short lick bursts were associated with short pauses,
and relatively long lick bursts were correspondingly associated with long
pauses. Between 85% and 95% of all data points fall close to the line of equal-
 256 Section II: Operant Conditioning Procedures

FR 20

2.1 .1 i_
1------------------------------___________
- - _ Mia. 1M _ I.... _

FR 40
1-----------~lImllr-,--,----~M~'r-nfl"~I.~lnll~---.-,'Irr'...
~ln.m.r-----~••"nl.~lml.~d~
2----------~liri--------_mIii--,__nII--------._,I~--------_._rrIl-------

3~anr-'u~~~~~~r_--_,~~~----_, ~---

,------------,~r----------------,~r----------'~r-----------

I. •• - • - - 1- - -
FR 20
_..
--
1
2 Ii_I

,
3

FR 40
1 1111 'III 1"111 Iii 11 11 1111 1111. a. n
2 It iii
• Ii

3~~hN~--~~ . .1T_m1V_nm_._~~MW~~--~
,-----------------.~r---------------------,~r----------------

Figure 5.1. Response interactions shown as event records from four successive schedule changes
for one rat. In one component, lever A responding was reinforced with a sucrose solution on FR 20
and lever B responding was not reinforced. In the second component, lever B responding was rein-
forced on FR 40 and lever A responding was not reinforced. 1, lever B; 2, lever A; 3, licking; 4,
sucrose. Adapted from Iversen: Scand. J. Psychol., 16, 49-54, 1975.

ity. Relatively long pauses occasionally occurred, however, without a concom-

itant long lick burst. Visual observation indicated that these long interruptions
in lever pressing were associated with grooming, sniffing, biting of grill bars,
jumping, etc.

Discussion

These two experiments show that the postreinforcement pause in FR respond-

ing expands in close association with increased durations of a specific collateral
Chapter 5: Collateral Responses with Simple Schedules 257

FR 1 FR 3 FR 6 FR 10 FR 14 FR 18
12
...
4
~~~~~~~~~~~~~-.--~-r-r-.-
4 n 4 n 4 n 4 n ~ 4 n ~ 4 12 ~ 28

Z 4122028 44 60 76 4 12 20 28 44
g
~
a
II: 92

~76

~.... 60

44

28
20
12
4

4122028 44 60 76 92 108 120 136

POSTREINFORCEMENT PAUSE (SECONDS)

Figure 5.2. Relationship between collateral licking and FR pausing for each FR size. Each data
point show pause duration and licking for an individual pause. Data are presented for the last two
sessions of each FR size from FR 1 to FR 26 (98 pauses) and for the last 10 sessions of FR 30 (490
pauses).

response. A molecular analysis especially suggests that each individual FR

pause is closely associated with bursts of a specific collateral response across
different FR sizes. Therefore, isolated causal relationships between ratio size
and FR pause, on the one hand, and between the ratio size and collateral re-
sponding, on the other hand, cannot be determined unequivocally.
The present results then raise the following questions: How are FR pauses
related to bursts of collateral responses. Does the collateral response lengthen
the pause or does the pause permit the appearance of the collateral response, or
are both perhaps separately affected by FR size?
258 Section II: Operant Conditioning Procedures

Pattern Dependency Between FR Responding and

Collateral Licking (Experiment 111)*
If collateral responses affect the operantly reinforced response, then the operant
response must be dependent upon the collateral response as well as upon the
explicit schedule contingencies. Some previous experiments purposely mani-
pulated collateral response characteristics and found that the overall rate of
reinforced responding may be inversely related to the overall rate or duration of
collateral water drinking (Clark, 1962; Segal and Bandt, 1966; Wayner and
Greenberg, 1972).

Experiment III

Experiment III (Iversen, 1976) further analyzed interrelations between patterns

of reinforced and collateral responses by specifically manipulating the duration
of collateral licking. The experiment also sought to determine whether physical
compatibility is necessary for interactions between reinforced responses and
collateral licking by positioning the drinking tube close to the response lever.
Rats were reduced to 80% oftheir free-feeding body weights, and lever pressing
was then reinforced with 45-mg food pellets on FR schedules. The FR was 60,
60,90, and 40 for rats 1,4,7, and 8, respectively. The duration of licking from a
freely accessible drinking tube 2.5 cm above the lever was first increased by
sweetening the water (to a2.5% sucrose solution) and next decreased by empty-
ing the bottle.
Licking and lever pressing could easily occur simultaneously, as shown in
Figure 5.3 for rat 1. The lever was continuously held down with the left front
paw during a lick burst after pellet delivery. For all rats, the lever pressing rate
was inversely related to the licking duration (Table 5.1). The data thus show a
functional relationship between duration of collateral licking and operant re-
sponding. Changes in collateral licking evidently can modify operant respond-
ing.

Discussion
Sample cumulative records (Figure 5.4) indicate that the postreinforcement
pauses (PRP) in lever pressing were (1) shortest when the bottle was empty, (2)
intermediate when the bottle contained water, and (3) longest when the bottle
contained the sucrose solution. The local lever-pressing rate preceding pellet
delivery (i.e., the run rate) was relatively constant for each rat for all bottle
conditions. The inverse relationship between overall lever-pressing rate and
licking duration (Table 5.1) thus is obtained because the lever pressing pause is
directly related to the licking duration.
The relationships between lick bursts and individual pauses and between

* Experiment was conducted at the University of Copenhagen, from October to December

1972, and reported by I. Iversen (1973).
Chapter 5: Collateral Responses with Simple Schedules 259

Figure 5.3. Rat 1 lever pressing and licking the drinking tube simultaneously. The lever is contin-
uously held down during a lick burst after pellet delivery. Adapted from Iversen: Psycho!. Record,
26, 399-413 , 1976.

licking and individual run times are further analyzed in Figure 5.5 for each rat.
Individual pauses again were long for long licking durations but short for short
licking durations for all bottle conditions. The pause approached the lick bursts
for 85% to 90% of all individual pauses for all rats in spite of different pause
durations across rats.

Table 5.1 Medians over the means of overall licking duration, response rate,
reinforcer rate, postreinforcement pauses (PRP) , licking durations for PRP, run
times (Run), and licking durations for runs of last 10 sessions

Licking Response Reinforcer Licking Licking

Rat Bottle duration rate rate PRP for PRP Run for Run
no. condition (sec/min) (resp/min) (reinf/min) (sec) (sec) (sec) (sec)

empty 6.0 108.0 1.80 12.5 1.5 16.7 1.8

water 24.0 60.0 1.00 30.8 18.0 30.1 5.3
sucrose 33.5 33.0 0.55 82.4 50.1 46.6 12.2

4 empty 0.0 67.8 1.13 14.8 0.0 23.4 0.0

water 16.5 58 .6 0.98 35.5 17.7 22.8 0.0
sucrose 21.0 40.8 0.68 51.4 26.9 24.1 1.3

7 empty 0.0 144.0 1.60 7.0 0.0 25.1 0.0

water 2.4 130.2 1.45 7.5 1.4 27.2 0.6
sucrose 14.4 88.2 0.98 34.6 13.1 30.2 1.1

8 empty 9.0 69.0 1.75 16.2 3.1 17.0 3.0

water 28 .2 48.0 1.20 26.8 20.8 20.3 7.8
sucrose 37.8 40.2 1.01 30.3 23.7 29.4 16.9
260 Section II: Operant Conditioning Procedures

Empty Water Sucrose

R1

I
R4

R7

~
RS

6min

v
Figure 5.4. Relationship between licking and FR responding in segments of cumulative records
from the last sessions with each bottle condition. The pen was deflected downward by licks and
pellet deliveries. From Iversen: Psycho!. Record, 26,399-413, 1976.
Chapter 5: Collateral Responses with Simple Schedules 261

The same close correlation was obtained during the ratio run for rats 1 and 8
in the water and sucrose conditions: a very low lever-pressing rate was fre-
quently accompanied by prolonged licking after pellet delivery. Since the
fourth lever press defined onset of the run time, all subsequent licking was as-
signed to the run rather than to the pause. For rats 1 and 8, very brieflick bursts
also occurred simultaneously with the high local lever-pressing rate during the
run period. For rats 4 and 7, licking after pellet delivery was only infrequently
accompanied by lever pressing, and the high run rate was only infrequently ac-
companied by licking.
The mean pause and lick durations are given in Table 5.1. For all rats, the
pauses and run times followed the respective licking durations. Other behav-
iors, such as grooming and exploration, were also observed with lever pressing
and licking throughout the experiment. Consequently, long pauses or run times
were recorded when grooming or exploration behaviors occurred for relatively
long bursts. The relatively long pauses and run times without similarly long lick
durations were included in the calculation of mean durations presented in Table
5.1. Hence the pause considerably exceeded the lick duration for mean com-
parisons (Table 5.1) but approached the lick duration for most comparisons of
individual pauses and lick bursts (Figure 5.5). Because of these additional re-
sponses, the differences between the run time and the corresponding lick dura-
tion were similarly longer for mean comparisons than for individual compari-
sons.
These mutual interactions between lever pressing, licking, and other behav-
iors are more clearly illustrated in Figure 5.6. The event records were obtained
with access to water. Food tray entry was recorded by a switch attached to the
door to the food tray. Standing, grooming, and exploration (walking around,
biting or sniffing grill bars, and sniffing comers or walls) were recorded by ob-
servers who depressed approrpiate keys for as long as the responses occurred.
Licking after pellet delivery was clearly associated with the pause in lever
pressing (A). Note that licking occurred infrequently after pellet delivery for rat
7 (compare to Figures 5.4 and 5.5), and the lever-pressing pause was therefore
typically short (B). Prolonged pauses occurred when standing, grooming, or ex-
ploration followed the lick burst (C and D). The high lever-pressing rate preced-
ing pellet delivery was rarely interrupted by long bursts of either licking, food
tray entry, standing, grooming, or exploration (E). However, brief licks occa-
sionally occurred along with the high rate of lever pressing for rats 1 and 8 (F)
(compare to Figure 5.5). A high lever-pressing rate was also associated with
frequent brief openings of the food tray floor (G).

Pattern Interdependence Between FR Responding

and Collateral Licking (Experiment IV)*
Thus so far, the data argue that collateral licking may participate in the forma-
tion of the PRP in FR responding. The question then arises as to whether an
* Experiment was conducted at the University of Copenhagen, from January to December
1974, and reported by I. Iversen (l977a).
262 Section II: Operant Conditioning Procedures

SUCROSE
100
' ..
80

60

~ 40 '.
" .'
8 20 .:,,\ ..
! :it:..:., .
t' ".:

20 40 60 80 100 120 140 20 40 60 80 100 120

'fT
R4

10rr
EMPTY 30
50
7

40 "
WATER
~}:,::,:" 30
20 ~ '~:"' .. .. : .... : .. SO
70

40
80
.'. SUCROSE 20
60
. '..
40
" 40 60 80 100

20
A
20 40 60 80 100 120
POSTREINf'ORCEMENT PAUSE RUN TIME (second!!)
Figure 5.5. Scatter plots of licking duration during individual FR pauses and during individual FR
runs for each of four rats. Frequency distributions of pauses and run times without collateral licking
are given in (inverted) histograms, Data are presented for the last 200 ratio runs in each experimen·
tal condition, Adapted from Iversen: Psychol. Record, 26, 399-413, 1976.

experimental dislocation of collateral licking into the FR run period might pro-
duce a simultaneous dislocation of the pause of FR responding. Simply, will the
FR pause go with the burst of collateral licking? Experiment IV addressed two
questions, First, how is the FR response pattern affected by different locations
of collateral licking within the FR schedule? Second, how is a burst of collateral
licking affected by the location within the FR schedule? The experiment also
Chapter 5: Collateral Responses with Simple Schedules 263

R7

EMPTY
10 1 lp"
"i
~ I
I

WATER
2~ 01'
.• •

20j '.' .' •

SUCROSE ~.",:'
·io h '26....,.,
.-.40 . r: ·T-,·,60--r",8:1:'0-
'. 30& __

20 40 60 80 100 120
80~ ii
1......
10
,
,"W ';t'
I " i
"
I
, EMPTY
RS
i _!/;" , i i

SUCROSE
60
40
......::
:I:

20 .~:':'

B
20 40 60 80 100 20 40 60 80 100
POSTREINf'ORCEMENT PAUSE RUN TIME (secondS)
Figure 5.5. (Continued)

incorporated an analysis of food cup contact in relation to FR responding and

licking.

Experiment IV A
As the operant response, lever pressing by albino rats produced 45-mg food
pellets on a FR 60 schedule. The rats were experimentally naive and were
maintained at 80% of their free-feeding body weights. The apparatus was the
 264 Section II : Operant Conditioning Procedures

Rl ,.
2••
3
4
5
.,.
-- ail

.,.
II 1
a_a
I 11111,._
h.' •••
I •• ,
i ••
i.

6. A- C F E + Gt

.. iii _ ' i '"''

-
.,. .,. .,. o

R4
1" . . . . IIUIIIII" '11 _ _ I na jJ.II • • • • • 1IIIIIIIit • • • 1I
2 ~~•.r~---------------. ..~...
.w"'''~-------------'''''~''Ar-----------'''''''

2.~~====~~==~============~========~===
5
6 c G.,. E .,.

'1'@lllIllllilllil',. 10'1, ••1_ U8n lllll • • • I III

",II. 'Ii • Ii II i lill

c t

R7
"Ga.-••• ' ' __1

....... .......... ,••,•• '".- =P......

•• alli.lIlJiS.iii il!i 11 • • , . . ._
2'---,.
3 00N
4
..r-------------------~r-r--------------------------------
, ••, t"...... . ,f _.111 ••' "M •• ". ' " ,,... li • • '"

- ..._- -
~=~~.,.~A
;:::::E;=::~t;B;:::::::~G~t;:::~::::::::~.,.;B;:::::::::~1'B~::::::::~tB~::=
_. "iii. . . . . . . . .' .- '8

c +
RS
1 ..... j I Ihllihll' _ I Iii Ih i _ ' i 1 I t. . . . I 1111 li • • • 1 I
i.-a _
2 • _w ... . .. 'Pi .,
3,
4
5
6 t A E t ------.F--·t.---------~ t ,
,_.11&._.
I.,. j
_liPill_
hi ' .. lilli_rill
••
'11I111.i'

Ii '" Ii.
•
a •••
i
iii

::::::::~+~::::::::~G~t::::::::::::~t~::::~~==~D -- - ,
• 30 SECONDS ••

Figure 5.6. Event recorder segments illustrating mutual response interactions between lever
pressing (I), licking (2), opening the food tray door (3), standing (4), grooming (5), and exploration
(6). Pellet deliveries are indicated by arrows. From Iversen: Psychol. Record, 26,399-413, 1976.
Chapter 5: Collateral Responses with Simple Schedules 265

same as that described in Experiment II, Chapter 4, and included a response

lever, a retractable drinking tube, and a food cup. The running wheel was
locked throughout the experiment. The drinking tube was located 1.5 cm above
and 2.5 cm to the right of the lever; licking and lever pressing, and licking and
food cup contact were thus physically compatible response pairs. Licking and
food cup contact were recorded by separate drinkometer circuits.
The location of collateral licking was manipulated by inserting the drinking
tube at different locations within the FR schedule. For example, the drinking
tube was positioned between pellet delivery and the 20th lever press (0-20) or
between the 5th lever press and the 25th lever press (5-25). All experimental
conditions are given in Table 5.2. Free access to drinking is designated as the
0-60 condition. The drinking probe was presented once for each interreinforce-
ment interval, except in Phase H, in which the tube was continuously retracted.
The experimental manipulations established elaborate discriminative control
over collateral licking, which in tum systematically affected the entire pattern
of concurrent FR responding.
First, presentation of the drinking probe prompted a changeover to licking
and an interruption of lever pressing. When the probe was presented right after
pellet delivery, the response patterns were indistinguishable from free access to
drinking-a lick burst after pellet delivery and a subsequent changeover to a
high rate of lever pressing until the next pellet delivery. The dislocation of the
drinking burst into the run period entirely modified this typical "break and
run" pattern of FR responding. The PRP contracted, and a second pause ap-
peared in the run period at the exact location of the drinking probe. The dura-
tion of licking and the associated pause in lever pressing were parametrically
dependent upon the within-run location of the drinking probe. Figure 5.7 pre-
sents the relationships between the drink burst, the concomitant pause in lever
pressing, and the duration of food cup contact during pauses as functions of the
serial location of drinking probes.
Each of the three response measures were decreasing functions of the num-
ber of lever presses since pellet delivery. Almost identical pause and lick dura-
tions were obtained, with concurrent zero durations of food cup contact, when

Table 5.2 Experimental phases in Experiment IV A

Phase Probe location Sessions
A 0-60 8
B 0-20 6
C 20-40 10
D 0-20 3
E 40-60 10
F 0-20 3
G 5-25 10
H no probe 3
I 0-60 3
J 55-60 10
K 0-60 3
266 Section II: Operant Conditioning Procedures

iii
R5

:~ll:m~i
"0
c:
o
u
GI
$ 40

. !~l~~!
Z
Q
....
<t
a::
5 20
"080~ [J [;:I 8
\ \ 2~j0n-,8 I I I I I I I rh,
16 2t. 32 0 8 16 24 32
I I I I I ,
•
~
x.. DURATION
O~~-r_-- _ _·_·--
___··TX___··_-_'_'- __-~.--=-*-
__'_'--rX-_._--

40
R6

20

-
'.
"

O~r--r--"---~-'~~--------~-------f-
o 5 20 40 55
LEVER PRESSES PRECEDING
TUBE PRESENTATION
Figure 5.7. Relationships between FR pausing (black circles), licking (white circles), and food cup
contact (X) during drinking probes as a function of FR responses preceding tube presentation. FR
pause was defined from tube presentation to the second lever press after tube presentation. Data
are from the last three sessions under each condition, Inserted histograms for rat 5 illustrate the
bimodal response distributions when the probe was preceded by 20 or 40 lever presses,
Chapter 5: Collateral Responses with Simple Schedules 267

1
1
I

L . . 0:.
40-
.1

20~ "'* . -..,. . . . -.. ;C•• - ··X X1 X

-1/1 O~~~~~----.---~~---~I~

8 60- R5
'0 1

CII
.!!. -, .1
z I
o 40-
~
~
:::J -

-
0
20 ~ 1

.-
C

- ~J¥ ~~
z ··'!t-·'_-- XI
<l
is .. -" X
1 X
w 0~-'--~----4---~~_~1~
~

R6
40 I
-I
1
01
20
I·
X

o5 20 40 55 FREE NO
TUBE TUBE
LEVER PRESSES
PRECEDING TUBE
PRESENTATION
Figure 5.8. Relationships between the postreinforcement pause (black circles), licking (white cir·
cles), and food cup contact (X) as a function of lever presses preceding tube presentation. Data are
also shown for phases with the tube freely accessible and with the tube absent. Data are from the
last three sessions of each phase.

20 or more lever presses preceded the probe. For rat 5, pause and lick bursts
were bimodally distributed for the 20 to 40 and the 40 to 60 probes. Median
calculations were therefore not representative and are not presented. Instead,
histograms showing the exact distribution of pause, lick burst, and food cup
contact are inserted in Figure 5.7 for the 20 to 40 and 40 to 60 probes.
With little doubt, the well·known "break and run" pattern ofFR responding
may be modified by simply altering the collateral response characteristics. The
268 Section II: Operant Conditioning Procedures

"break and run" pattern was fractionated into a dual pause and run pattern,
with licking restricted within the run period: a brief pause associated with food
cup contact, a run of FR responses until presentation of the probe, a second
pause concomitant with licking, and then the last run of FR responding until the
next pellet delivery.
The manipulation of probe location not only affected the response pattern
during the probe, but also changed the patterns in the interprobe intervals. The
FR PRP (from pellet delivery to the second lever press) was thus shorter with
the drinking probe only available within the run period than with freely access-
ible drinking (Figure 5.8). The data therefore replicate Experiment III, which
also found that the FR PRP is shortened when collateral licking is prevented
after pellet delivery. However, the shortest pauses were obtained with the 5 to
25 probe. This probe early within the FR run had a clearly facilitating effect
upon FR responding compared to later or no probe presentations. Access to
collateral licking apparently not only has a suppressive effect upon FR re-
sponding during the probe, but may also have a facilitating effect upon FR re-
sponding preceding the probe.
The response interactions can also be followed in the event records in Figure
5.9. In general, these more "on the spot" response alternations confirm the im-
pressions revealed by the overall data. Note, for example, the typical FR re-
sponse pattern with the tube continuously accessible (0 to 60) and the seg-
mented FR pattern with pausing dislocated into the run period by collateral
licking (e.g., 20 to 40). The relatively briefPRPs with the 5 to 25 probe are also
clearly apparent along with the increased run times for the long lick bursts.
One outstanding feature is the very brief bursts of licks regularly and rapidly
alternating with the high rate of FR responding if the probe was scheduled dur-
ing the ratio run. The event strips provide some indication that the run time was
shorter (and hence the FR run rate higher) when the tube was not presented.
Evidently, even these brief lick bursts also suppressed FR responding to some
extent. Figure 5.10 presents a high-speed event record of lever pressing and
licks during two FR runs with free access to drinking. Licks occurred in clus-
ters of two or three tube contacts and the IRT in lever pressing tended to be
longer when licks occurred. Note that licks only rarely occurred in strict simul-
taneity with a lever press (a).

Experiment IVB
This experiment merely elaborated the experimental analysis to include a sys-
tematic replication with response-independent probes presented for fixed 10-
sec periods. Probes were distributed on a VT schedule with a mean of 1 min
(the individual interprobe intervals were 20, 36, 56, 76, and 96 sec). Probes
could therefore occur more than once within individual interreinforcement in-
tervals of the FR schedule. Experiment IVB was scheduled for six sessions be-
tween Phases H and I in Experiment IV A.
This systematic replication is described in Figure 5.11. The lick burst re-
mained at the maximal value of 10 sec for all probes for rat 4, but decreased as
Chapter 5: Collateral Responses with Simple Schedules 269

Rio R5 R6
TUBE. 0 - 60 (FREE)
t-, - ----------~,----- ___
~ , I
Ai Ii
l.-v---~_r----_r-_,~ • i_a
~~,,~"'~,--.,~"""~"~II~."'.~II~"~-~,--~ , 1i .' 4 ' '_Iii .. Ii .. _ • • • • '" ...... _

----
0-20

-- •• • .... _i,
_JM
_ _ ....IIT"....
" .,,.- -_ .......... . . - -._ _

!-~
l 0 _______ ___ _~ ~

_ _ • . _ _ a ..

lv---~--. .------~­
~~--~----~~~----

20--'0
ni." __ i._ -
~­
... '\........U.4 I i .,. i ....CW
.....-,T",- - - . -- -

l.
z;;. - _. __
1
__40-60
_

----~
_ ___
---;-;:.
__.ir_----___,_
. . . . _-._--- -"...-=---- ...- .. _--r--
---------
- -------- -
~--~-
~ -

~------------.--------
----- ------- -------------- --- --......-------;,.....r---

--------..
fi., __ • , •• _ ••
~-~-
.....-.rr- • • , ••_ , . __ li.
•• Ii _

~--~--
NOT PflESENT ED ~ ____--___.r- _ _
• ",_=_,.
I---r- ~ I r---.----~--- T"

~ ~ ........ r , ....... ,tnW .........

l,.~--~r---~~---' ..."....-. Ii .- " _ i
4. - ~ ---

Figure 5.9. Sample event records showing response interactions for each location of the drinking
tube. Numbers refer to drinking probe location (20-40, for example, refers to the presentation of
the drinking tube between the 20th and the 40th lever press since last pellet delivery). Brief deflec-
tions of pen 1 indicate pellet delivery and extended deflections indicate drinking probes. However,
pen I was not continuously deflected when the drinking tube was continuously accessible (0-60). I ,
pellet delivery or tube presentation; 2, lever pressing; 3, food cup contact; 4, licking.

---.-
1 'lee

Figure 5.10. High-speed event records of licking linked to lever pressing. Records show the last 40
lever presses of two FR runs terminating in reinforcement (rf). The lower record was selectively
chosen to show temporal overlap between a lever press and a lick (a).
270 Section II: Operant Conditioning Procedures

20 R4

10 \ . - .... _ !.-m.. ! .. " .. B -! -- .... <I .. .-.... !

0
Iii
6
u
R5
!
z
0

~ 10 C). ••• 0--- •• - - 0- - -_ •••• 0.

-.
a lC •• ,
"G--_ ••• -Q

'
z
~ ""-"-~- .. - .. - .. -II.. - .. _ .. _ '0
is
w 0
~

R6
20

O-S 6-10 11-20 21-30 31-40 41-50 51-60

CLASSES OF LEVER PRESSES
PRECEDING TUBE PRESENTATION

Figure 5.11. Relationships between FR pausing (black circles), licking (white circles), and food
cup contact (X) as a function of the number of lever presses preceding tube presentation.

the probe occurred later in the run period for rats 5 and 6. The duration offood
cup contact was zero for rats 4 and 6, and the lever-pressing pause therefore
exceeded the lick burst by not more than approximately 0.5 sec for probes in
the run period. For rat 5, the pause in lever pressing exceeded the lick burst by
an amount that closely approximated the added duration of food cup contact.
The pause in lever pressing then approached the sum of the durations of licking
Chapter 5: Collateral Responses with Simple Schedules 271

R4
1.- .....___ - -----...-----r
1 --r~,.----------r------;;; 3:;;-8 ~ I 34.-----Jr-r----,...-
2.--- • _..-_== t. . . . . . . _ ...-..rr ___ . - . - - ~ "'''--.-----...,
3- .. ~.r-----r--~,.---~rT----_.~---~_----~
4 -----~~---------~----~-------------

3.r..
_____
r
i __ ,I r ..: li__ .i -__.._
-,
~
r
.,
40:

,
_"1

•
4 i: ..:
~----r--Y""""--
1
II 8.... I ....
____ ""1-r"r_~
1"9 : 48"'. I-
iii'. ....

4-______ --- - -- - - - -;--------- - - - ----- - - - - ----.-.- - - ; - -

1.-~. ~- ---,---- 52:.) J~ ~ I 11--------

2..-..- ~1m. ..r~ . .__~_;. .__- ". . . .r_- •••~. ..-----~.- ••na~~_r_~.~
.~..
~ ~'---~~--;~--'~r---~'--~~~--'~,,-~'--------'
4- _

30 SEC. •

Figure 5.12. Event records showing response interactions when tube presentation occurred for 10
sec independent of lever pressing. Inserted numbers refer to the number of lever presses emitted
prior to tube presentations. Brief deflections of pen 1 indicate pellet deliveries and extended deflec-
tions indicate tube presentations. Pen 1 deflected upward for pellet deliveries during tube presenta-
tion. 1, pellet delivery or tube presentation; 2, lever pressing; 3, food cup contact; 4, licking.

and food cup contact. However, for rat 6 the summed durations of licking and
food cup contact exceeded the pause in lever pressing in the first class (0 to 5).
This resulted from the occasional emission of both licking and food cup con-
tact.
Similar to the findings in Experiment IV A, drinking probes also affected the
response pattern during the interprobe intervals, with a facilitating effect upon
lever pressing in the absence of licking. The lever pressing rates with no access
to drinking (Phase H in Experiment IVA) were 79.9, 66.6, and 83.7 re-
sponses/min for rats 4, 5, and 6, respectively. The rates of lever pressing for
interprobe intervals in Experiment IVB were 92.3, 85.7, and 124.1 re-
sponses/min for rats 4,5, and 6, respectively. The lever-pressing rate in the ab-
sence of licking was thus clearly higher with the response-independent probes
relative to sessions with no access to drinking.
The event records in Figure 5.12 graphically confirm the overall interactions
with an interruption of FR responding by licking during drinking probes .

Discussion
The patterning of FR responding was directly affected by the manipulated pat-
tern of collateral licking. The PRP decreased and a "new" pause appeared dur-
ing the run by simply displacing the burst of collateral licking to the run period.
The location of FR pauses may thus be determined by the location of collateral
licking, with FR pauses whenever and wherever bursts of licking are emitted.
At the same time, the licking burst is functionally determined by the location
within the FR schedule. Because licking was physically compatible with both
272 Section II: Operant Conditioning Procedures

lever pressing and food cup contact, the data further confirm the suggestion
that physical incompatibility is not necessary for response competition.
Some previous data would seem to be especially relevant to the present re-
sults. Using similar procedures, Flory and O'Boyle (1972) and Gilbert (1974)
established that collateral licking may occur at interreinforcement locations
other than the postreinforcement period. Gilbert (1974), in addition, found that
the overall rate of FI responding increased when collateral licking was trans-
ferred from the postreinforcement period to later in the FI. Collateral licking
was also changed by the FI response and increased at any given location when
the FI response was prevented.
The present experiment further establishes such reciprocal interactions be-
tween collateral responses such as food cup contact and licking. The food cup
contact was very brief after pellet delivery if licking occurred at that location.
However, if licking was restricted to the run period, then food cup contact in-
creased after pellet delivery. Recently, Penney and Schull (1977) reported a
similar reciprocal relationship between collateral water drinking and wheel run-
ning in rats responding on FI schedules. In the present experiments, however,
food cup contact did not expand to entirely replace licking. Ator (1976) simi-
larly found that collateral attack and escape responses are not interchangeable.
The tactic of manipulating experimental events during the run period has
been discussed by several investigators. Boren (1961), for example, reported
that the probability of a changeover to a second response during a discrete-trial
procedure was a decreasing function of the number of FR responses preceding
trial presentation. Lyon (1964) also found that suppression during a preshock
stimulus was a decreasing function of the number of FR responses emitted
since reinforcer delivery. In combination, the data suggest that a variety of dis-
crete events may disrupt an FR response as a function of the relative location
within the scheduled interreinforcement intervals.
The discrete-trial methodology has the experimental advantage of establish-
ing elaborate discriminative control over collateral responding. A changeover
from either FR responding or food cup contact to collateral licking was espe-
cially under the discriminative control of the drinking probe in the present
schedules. Since response preference may be expressed in terms of change-
overs from one response to another (Skinner, 1950), the results indicate that a
collateral response may be momentarily preferred relative to a concurrently
reinforced response. This preference analysis is also consistent with the finding
by Premack (1971) and Timberlake and Allison (1974) that a momentarily more
probable response may reinforce a momentarily less probable response.
In agreement with this formulation, collateral responses can be shown to
function as a reinforcer when contingent upon an additional response. Azrin et
al. (1965) thus demonstrated that squirrel monkeys would perform a chain-pull-
ing response to produce an object that could be attacked. Rats may similarly
respond on a lever to produce access to collateral licking during intermittent
food presentation (Falk, 1966; also see Cole and Parker, 1971). In the present
experiment, the FR response was facilitated by the scheduled access to collat-
erallicking when the drinking probe was contingent upon a low number of FR
responses (Experiment IV A). This facilitation was also apparent when access
Chapter 5: Collateral Responses with Simple Schedules 273

to collateral licking was independent of FR responding (Experiment IVB).

Note that the same event that suppressed FR responses during probes also faci-
litated the FR response during interprobe intervals. The establishment of dis-
criminative control of collateral responding may thus additionally establish a
reinforcement operation that can effectively alter response patterns.
Finally, we should note some semantic problems in the generalization that
two physically compatible responses may compete. Food cup contact, for ex-
ample appeared to comprise a broad class of different response topographies
such as sniffing, licking, or gnawing the cup, touching the cup during pellet re-
trieval, or passively holding the cup during licking. Many of these responses
involved orientation of the head to the cup (sniffing, licking, gnawing, and pel-
let retrieval) and were clearly incompatible with and reciprocally related to
tube licking. In this case, topographically different responses activated the
same manipulandum and were recorded as the same physical response. How-
ever, the different response topographies defined as one response class may not
necessarily enter the same relationship with a second response class. Essen-
tially, mere activation of a response manipulandum may not be sufficient to
identify the associated response topography. A generalization involving a given
response class may therefore not hold for all members of the class. Instead, a
more precise statement would be that sniffing, gnawing, or licking the food cup
may be reciprocally related to tube licking, whereas holding the food cup more
clearly accompanied than competed with licking.
Routine observations may also reveal different topographies of recorded
"licking", including tongue, nose, and paw contact. The extended burst of lick-
ing after pellet delivery clearly occurred as tongue contact whereas licking
linked with lever pressing consisted mainly of nose contacts during rapid up
and down movements of the head accompanying the high rate of lever pressing
(the tube was only 2.5 cm from the lever and nose contact was especially com-
patible with lever pressing). A recent analysis similarly indicated that drinking
tube contacts occur in two distinct patterns and topographies in a tandem FI
schedule (McLeod and Gollub, 1976). Tube contact occurred in a long burst
after pellet delivery, and paw rather than tongue contact apparently occurred
linked with FI responding.
In conclusion, our precision in description and generalizations would seem
to require an equal precision in the analysis of different response topographies
entering the same overall response class.

Pauses and Collateral Responses with

Reinforcement Omission (Experiment V)*
The purpose of Experiment V was to extend the response-response relation-
ship to a research area in which changes in pauses are critical; the point of in-
terest is the frustration effect (FE) (Amsel and Roussel, 1952). The FE his tori-
* Experiments VA and B were conducted at the University of Copenhagen, from June to De-
cember 1974, and reported by 1. Iversen (1977A). Experiment VC was conducted at the University
of Zurich, in February 1975, and reported by 1. Iversen (1977B).
274 Section II: Operant Conditioning Procedures

cally refers to an increase in running speed in the second alley of a double run-
way on those occasions in which reward is omitted in the first goal box. A simi-
lar effect can be found in a free-operant situation, in which a response rate
maintained on an intermittent schedule may increase following the occasional
omission of reinforcement (Staddon and Innis, 1966; Zimmerman, 1971). The
rate increase in the free-operant situation has been shown to be due mainly to a
shorter pause after reinforcement omission (Staddon and Innis, 1969; McMil-
lan, 1971) rather than to an increased local rate (Dews, 1966).
The implication of response interactions for the frustration effect was first
aroused by an equipment failure. In a preliminary experiment, lever pressing
by rats was maintained by food pellets on a FR 50 with water freely available.
Very brief PRPs suddenly appeared during an otherwise stable baseline of long
pauses associated with bursts of licking. An inspection of the apparatus re-
vealed that the pellet feeder had become defective in such a way that it either
operated with pellet delivery (as it should), operated without pellet delivery, or
did not operate at all. Since these different events appeared to have different
effects on lever pressing and licking, the malfunctioning of the feeder was put
under experimental control and a brief experiment was conducted.
The FR 50 was maintained continuously, and ordinary pellet delivery con-
sisted of three events: operation of the pellet feeder, pellet arrival in the food
tray, and a O.5-sec illumination of the food tray. For two sessions, 25% of the

1
2
3
n_ DII ...
,....------,'rr• ..---
~

.... t•
1-
4

n
- , - - - - - , - - - - - - - , r - - - T O 'r - - -
A
,
t
Ii,._ ...
1
2a
3--.--u
4
- 11_ ,
III_ i ji

t
r--,
II
. .M

".,
II

...---- ••••
8
'J1T\I

--v
• II
• Ii
..------,
.. jji )i
--
t I ..
30 SEC

Figure 5.13. Sample event records showing effect of reinforcement omission upon lever pressing,
licking, and food tray entry. A. Feeder illumination only. B. Feeder illumination and feeder opera-
tion. Arrows indicate pellet omission. 1, lever pressing; 2, licking; 3, food tray entry; 4, pellet deliv-
ery.
Chapter 5: Collateral Responses with Simple Schedules 275

pellet deliveries were replaced with only illumination of the food tray, and for
two subsequent sessions, with illumination of the food tray and operation of the
feeder without pellet delivery.
Figure 5.13 shows event record segments for one rat. Similar records were
obtained for two other rats. The rate of lever pressing remained high when the
food tray was only lit (arrows in record A). A changeover from lever pressing to
a brief food tray entry, and an immediate changeover back to lever pressing,
occurred when the food tray was lit and the feeder operated without pellet de-
livery (arrows in record B). With pellets also delivered, the changeover to food
tray entry was followed by licking and a pause in lever pressing. The tentative
results demonstrate that the typical pause in FR responding may be eliminated
when scheduled events do not control a changeover to a concurrent response
(no food tray entry), may be brief when the events control a brief changeover to
a concurrent response (food tray entry), and may be long when the schedule
controls long bursts or sequences of concurrent responses (both food tray entry
and licking).
A concurrent reponse analysis would suggest that part or all of the frustra-
tion effect may be related to alterations in the patterning of specific collateral
responses by changes in the events surrounding reinforcement omission. The
frustration effect of a decreased pause in reinforced behavior may thus result
from the decreased duration or virtual absence of collateral behaviors.
In the following three brief experiments, occasional reinforcement omission
and prevention of selected portions of the collateral behaviors were investigated
in a within-session design.

Experiment V A
The first experiment (V A) compared the frustration effect to the effect of pre-
vention of collateral licking after regular reinforcement. The rats and apparatus
were the same as in Experiment IV, with lever pressing reinforced on FR 60.
Pellet delivery with the drinking tube retracted, only feeder operation with the
tube present, and only feeder operation with the tube retracted were each sched-
uled three times in mixed order for each session. Normal pellet delivery with
the drinking tube present was in effect for the remaining completions of the FR
requirement. In the appropriate conditions, the tube was retracted 0.5 sec after
feeder operation and was first presented at the next feeder operation. The oper-
ations necessary for prevention of pellet delivery were performed manually by
the investigator. The soft and flexible rubber tube from the feeder to the food
cup was silently disconnected during the run period preceding feeder operation
without pellet delivery. The tube was reconnected during the next run period.
In no case was lever pressing disrupted during this operation. A predetermined
sequence of events was followed during sessions to avoid any bias in the man-
ual scheduling of feeder operations without pellet delivery. Experiment VA
was scheduled for six sessions of 25 min each after the last phase of Experiment
IVA.
All feeder operations produced a changeover from lever pressing to food cup
276 Section II: Operant Conditioning Procedures

iii • • ,_ _ •
.M'I_ _ ',.

,s.
1~----1O--

". '"
__"'---V--'r--r--~---r--,-~--~---,.----'r----r
11 i iii .. ii. '"'' i hilh i i 1111 . . i '_Ii Ii . . .1 nil In iii
'"
ab 'd

t. R,S
~ __~---..~,"~._m.....m_rn'Nn_~.rr'------ni_n.m._•••_r'-.'nN~_••_~_.I".~.
l . .+r--~~~Tr~--~-----.~n-~. .rn~-'~~~~nr~'----'---
4ri"-'__"1'H'.n,~, .--,.-,.__wr-..--.---------,__. .-r~~~----,._
5
6
h' 71 ab h e C II

2M_, i i W' .,.

1---r--------~iir_--~_,.~.--._--
-- ___ .
~_,i~'----,__W_,~r--_..~,--

",a .. Mdaili _
_
"'h""",
5
"A,mil t Ii • ;"iUiI iiii iii i ••I1....IWIII ., • • •

d II ab ec II d
- -.
30 SEC.

Figure 5.14. Event records showing effects of reinforcement omission and withdrawal of the
drinking tube. I, pellet delivery; 2, lever pressing; 3, food cup contact; 4, licking; 5, tube presenta-
tion; 6, empty feeder.

contact (a in Figure 5.14). If the drinking tube was available, a second

changeover to licking rapidly followed pellet delivery (b). Pellet delivery with
the tube retracted produced a longer burst of food cup contact, but the lever-
pressing pause was much shorter than with the additional collateral licking (c).
Again, FR pauses were clearly shortened by prevention of collateral licking.
After feeder operations without pellet delivery, the burst of food cup contact
was very brief and the burst of licking was entirely absent, leading to a very
brief lever-pressing pause (d) (the omission effect). Both food cup contact and
the lever-pressing pause were longer after pellet delivery with the tube re-
tracted relative to feeder operations with no pellet delivery (e).
Licking during the run period was, however, differently related to reinforce-
ment omission. Licking occurred in brief bursts in rapid alternation with the
high local rate of lever pressing (f) after pellet delivery as well as after rein-
forcement omission. The reinforcement omission thus removed the extended
lick burst that normally follows pellet delivery but did not affect the brief lick
bursts in alternation with lever pressing. The high rate of lever pressing tended
to be slightly increased if the tube was retracted (g).
In summary, the FR pause was equally brief after reinforcement omission or
after prevention of collateral licking with regular reinforcement. Reinforcement
omission differentially affected the licking pattern during the FR pause and run
periods.

Experiment VB
The second experiment (VB) investigated the relationship between extinction
of FR responding and collateral licking. Extinction was first scheduled with the
Chapter 5: Collateral Responses with Simple Schedules 277

drinking tube continuously available (after Phase K in Experiment IV A), and

then scheduled again with the drinking tube presented randomly for IO-sec
periods (after Experiment IVB). Feeder operation occurred during the_ extinc-
tion sessions, but food pellets were not delivered.
Figure 5.15 presents event records for rat 4 during the extinction session
with the drinking tube continuously accessible. Similar records were obtained
for rats 5 and 6. The usual licking burst was absent after the first two feeder
operations, which made the FR pause very brief (a and b). Subsequently,
longer bursts of food cup contact and longer pauses occurred after feeder oper-
ations (c). Licking still occurred in brief bursts, alternating with the high lever-
pressing rate preceding feeder operations (d). The frequency ofbrieflick bursts

TUBE FREELY ACCESSIBLE

~s.s,II. __II_'_"*'_ •••_ •• a_'I'. ___I' _ ••• _. ___ •• ,i . _•• __ •• -
I

~ i. r.Tn I
d
i. ' iii II • • • iI

-.
t

••
~.o~,-r------~~.--~----~.rr.-r----~--'i~.-----------r------'-----'
ur---"TO

,.
TUBE PRESENTED RANDOMLY
I j ~~~r------'---TI-~~.----'-'
_ _ _ _ i I Ni._im_iW,_ WHi_ , Wi.' _ .... _ •___

~,..u----r.-r-.-..J
,1".III_a,_I.. a, . . . . . .
A'
'i
.'i
._i'.'___.'.i.,.,iil
_ J 1,....--1 1 :...-r--r-~
it_iii . ."' ,iiU •••• i •• "
I • •"
I 1, ••••• _."
I ,
~

• i
' i '.,'
.,....,-

-----r---~~r-----r- ~----,~r-----~~---

Ii f i Ii
hi IW
f" h

l'~~~--,r---------.--,r----~
2. i i Ii

3.- " i i i Ii
',--.---------.--.m---------~.--------..,,---------------.m--------m.-------::-
n -

_R_6___~~~~. .------~~ :-------------~~-----

m m

Figure 5.15 Response interactions during FR extinction with the drinking tube present or with
random presentation of the tube for fixed IO-sec periods for rat 4. Records are also shown for rats 5
and 6 after lever pressing was extinguished to a low rate . I, empty feeder/tube presentation; 2, lever
pressing; 3, food cup contact; 4, licking; SS, session start.
278 Section II: Operant Conditioning Procedures

during the run period, however, rapidly decreased and the local rate of lever
pressing was concurrently increased during the extinction session (e). Gradu-
ally, longer periods without lever pressing, licking, and food cup contact regu-
larly appeared between bursts of lever pressing (t). These periods were asso-
ciated with unrecorded behaviors such as grooming and exploration. Thus, FR
extinction initiates with brief FR pauses in association with the absence of the
extended bursts of licking after regular reinforcement.
The relationship between collateral licking and FR extinction was somewhat
different when the drinking tube was presented randomly. Event records are
shown in three segments for rat 4 in Figure 5.15. Segments are also shown after
lever pressing was extinguished for rats 5 and 6. The drinking probe controlled
an immediate changeover to licking when presented early (g) but not late in the
extinction session (h). Although licking occurred irregularly rather than in an
extended burst throughout the probe G) and alternated with bursts of lever
pressing (k), extended bursts of licking did appear when lever pressing was
markedly decreased (1). For rats 5 and 6, licking occurred in a long burst in
some (m) but not all (n) drinking probes, even after complete extinction of lever
pressing. Thus the discriminative control over licking by tube presentation was
quite powerful in these experiments.
This experiment shows that the extended bursts of collateral licking may de-
pend upon the absence or presence of controlling stimuli; that is, bursts of lick-
ing controlled by pellet delivery were eliminated by FR extinction, whereas
licking continued to occur when under the discriminative control of the drink-
ing probe.

Experiment VC
The last experiment (VC) examined the response interactions in a schedule
using response-independent reinforcers. Water drinking, food cup contact, ex-
ploration, and grooming were recorded when food pellets were merely deliv-
ered every minute (FT 1 min) for 14 sessions. Two rats were maintained at 80%
of their free-feeding body weights. The apparatus was the same as in Experi-
ment IV and the wheel was locked throughout all sessions. Tube licking and
food cup contact were recorded automatically, and exploration (sniffing, climb-
ing in the locked wheel, and standing on hind legs) and grooming (scratching
and licking of the body) were recorded by the experimenter pressing appro-
priate keys for as long as the behaviors occurred. Experimental manipulations
were as in Experiment VA, except that the condition with the tube withdrawn
after pellet omission was not included.
Figure 5.16 shows the data for one rat. After pellet omission, licking re-
mained at zero probability and food-cup contact did not decrease to zero,
which is a reversed pattern relative to pellet delivery accompanied by collateral
licking. The reinforcement omission also altered the distribution of exploration
and grooming within the FT schedule, with both responses occurring earlier in
the interval than after pellet delivery with licking. The exploration probability
was therefore increased overall, whereas the grooming probability quickly re-
sumed to zero and was decreased overall. Essentially similar response distribu-
Chapter 5: Collateral Responses with Simple Schedules 279

PELLET NO PELLET PELLET

WATER NO WATER

\rv'
1.0 RAT 1
0.8
0.6
0.4
Q2
"
V - , . . FOOD-CUP
CONTACT

QO
>
....
::::i
iii LICKING
~
0
0::
a..
UJ 1.0
CI)
Z 0.8
~ 0.6
CI)
EXPLORING
0.4
UJ
a:::: 0.2
00
1D
0.8
O. GROOMING
0.4
0.2
0.01............H"r-n...-rl
1224 36 4860 12 24 364860 12 24 364860
TIME SINCE FEEDER OPERATION (seconds)
Figure 5.16. Mean response probabilities for rat 1 during successive 6-sec intervals since feeder
operation for each experimental condition. Food cup contact and licking probabilities are also
shown for the first three 2-sec intervals of the pellet plus water condition. From Iversen: Physiol.
Behav. 18,535-537, 1977.

tions were obtained when the drinking tube was removed and licking prevented
after ordinary pellet delivery.
The effect of reinforcement omission therefore again resembles the effect of
prevention of the licking burst after regular reinforcement.

Discussion
The reinforcement omission effect now appears to be directly related to simul-
taneous alterations in the patterning of collateral responses. Allen et al. (1975)
and McCoy and Christian (1976) also found an absence of licking and decreased
pauses in FI responding after reinforcement omission. In addition, Staddon and
Ayres (1975) reported an absence of licking and an increase in food cup ap-
proach after omission of response-independent reinforcement on an FT sched-
ule.
The results suggest that the reinforced response pattern produced by rein-
280 Section II: Operant Conditioning Procedures

forcement omission may be simulated after normal reinforcement by experi-

mentally interrupting the sequence of collateral responses. Thus, the pause in
FR responding decreased when the burst of licking was prevented after regular
reinforcement. A previous significant finding was that licking may be absent
after reinforcement omission if lever pressing is simultaneously increased or
experimentally prevented (porter et al., 1975). The absence of licking may thus
be a primary effect of reinforcement omission and may be partially a "cause"
rather than an exclusive "effect" of the simultaneous probability of other re-
sponses.
The present experiments have some implications for a related finding, the
FE. The FE of increased response "vigor" after reinforcement omission was
originally explained as an active emotional-frustrative motive state created fol-
lowing thwarted reward expectations (Amsel, 1958). A similar explanation was
applied to the increased response rates that may be seen during the first min-
utes of extinction and that has been said to measure the emotion aroused by
reinforcement omission (Thompson, 1961). The FE could also result from an
absence of short-term satiation or demotivation effects of reinforcer delivery
(Seward et al., 1957). By another account reinforcement is considered inhibi-
tory, and omission of reinforcement prevents inhibitory aftereffects and
thereby increases the local response rate relative to typical reinforcement
(Staddon, 1970; also see Catania and Gill, 1964).
The FE, however, could be examined without reference to intervening con-
cepts. The problem is not simply whether the FE "reflects frustration or in-
hibition" but exactly how the FE comes about. The accumulating results sug-
gest that the pattern of a reinforced response after reinforcement omission may
be related to the patterning of specific collateral responses. The data are thus
consistent with previous thoughts that the FE may be related to a decrease in
competing responses (Bolles, 1967; Wookey and Strongman, 1974). More pre-
cisely, the extent of the omission effect is directly related to the amount of col-
lateral responding under control of the scheduled events. Further, the FE could
be closely simulated in the absence of reinforcement omission. Simply, rein-
forced responding may then be increased or decreased to the extent that sched-
ules control the absence or presence of specific collateral responses.
In a traditional account we could say, but are not necessarily forced to do so,
that the changed collateral response pattern mediates the FE. We would rather
argue that reinforcement omission and prevention of collateral licking after reg-
ular reinforcement are merely two methods of increasing the probability of
reinforced responding-hence the above emphasis that the effect of reinforce-
ment omission is simulated by prevention of collateral licking after regular rein-
forcement.
In Experiments VA and VB, the same collateral licking response occurred in
two temporal locations and in two different patterns: a long burst after rein-
forcement and subsequent brief bursts in rapid alternation with the high rate of
FR responding. Rosenblith (1970), Wuttke and Innis (1972), Porter and Ken-
shalo (1974), and Corfield-Sumner et al. (1977) found licking after reinforce-
ment as well as after brief stimuli in some second-order FI schedules. The cu-
Chapter 5: Collateral Responses with Simple Schedules 281

mulative records presented by Rosenblith (1970) and Wuttke and Innis (1972)
seem to indicate that licking after brief stimuli was at least partly linked with FI
responding. Moreover, data by Segal (1969), Keehn and Colotla (1971), and
Allen and Porter (1977) provided some indication that short bursts of licking
may alternate with FI responding. Collateral licking may then be differently
maintained during the postreinforcement period and the run period. In terms of
discriminative control, pellet delivery maintained a long lick burst, whereas FR
responding controlled frequent and brief lick bursts. Reinforcement omission
or extinction merely remove the event controlling long lick bursts, leaving the
brief bursts linked with FR responding. This dualistic control of licking was fur-
ther evidenced by the presence of relatively long lick bursts during FR rextinc-
tion when the discriminative control was changed from pellet delivery to drink-
ing tube presentation. Significantly, this licking apparently was not linked with
FR responding since the discriminative control by drinking probes was main-
tained for some time after FR responding extinguished. Therefore, multiple
patterns of collateral responding may be dissociated, dependent upon the selec-
tive control of the appropriate discriminative events.
In conclusion, reinforcement omission is a procedure that at least roughly
separates different components of collateral responding. One component is the
responses that are linked with the reinforced response. The second component
is the responses under more direct discriminative control of reinforcer delivery.
Removal of the reinforcer then separates the different collateral response com-
ponents by eliminating only the stimulus control ofthe collateral response com-
ponent following reinforcer delivery.

Interresponse Times and

Collateral Responses (Experiment VI)*
A FR schedule provides a relatively clear distinction between pausing and re-
sponding by the respective pause and run periods within the interreinforcement
interval. Typical FR responding is thus described by a few very long IRTs (the
FR pauses) and many very brief IRTs (in the run period). On the other hand, a
schedule such as VI generates less dramatic differences between long and short
IRTs, with a characteristic mode at intermediate IRTs. If pauses in FR re-
sponding can be related to collateral responses, then pauses in VI responding
might too.
The relationship between collateral responses and IRTs in VI schedules has
been debated over the years. Anger (1956) provided an analysis of IRTs and
proposed that collateral responses probably do not mediate IRTs. However,
subsequent arguments have proposed that IRTs might be mediated by collat-
eral responses (Shimp, 1969). A very rough distinction has been between
whether a subject chooses (1) whether or not to respond at a given time, or (2)
how long to wait before emitting the next response (Shimp, 1969, 1973).

* Experiment was conducted at the University of Copenhagen, from January to February 1976.
282 Section II: Operant Conditioning Procedures

Our previous results suggest that a particular IRT (the PRP) becomes long if
a subject changes over to collateral licking. In fact, knowing that licking occurs
at the beginning of the PRP provides some information that the pause will defin-
itely be longer relative to pauses without collateral licking. Essentially, the data
suggest that the collateral response at the start of the pause may be a good pre-
dictor of pause duration in FR schedules. Perhaps similar predictions would
apply for pauses or IRTs in VI schedules. Experiment VI was an attempt to
examine such a model of IRTs in VI responding.

Experiment VI
Four experimentally naive Wistar albino rats were maintained at 80% of their
free-feeding body weights. The wheel-running equipment was used in this ex-
periment but modified to tum in only one direction (clockwise relative to the
panel with the lever and food tray). The drinking tube was not accessible in this
experiment, and the hole was covered with a metal plate. Contact with and ac-
tivity within 2.0 mm of the food cup and the lever were recorded by separate
body-capacitance systems. The IRT analysis included eight responses: lever
pressing, lever activity, food cup activity, wheel running, exploration, stand-
ing, face grooming, and body grooming.
After initial shaping, lever pressing was maintained on VI 1 min for 15 ses-
sions. One 45-mg food pellet was given as reinforcer delivery and each session
lasted 30 min.
Examples of the mutual interactions between the responses are shown in
Figure 5.17. Wheel running regularly alternated with lever presses for all sub-
jects (a). Most running bursts were of a fairly fixed duration, with the asso-
ciated IRT in lever pressing also approximately fixed (b). However, if explora-
tion or standing preceded wheel running, then the IRT was relatively longer (c).
In addition, relatively brief bursts of wheel running were occasionally inter-
spersed with longer bursts for rats 1 and 3; the associated IRTs were simulta-
neously relatively short and long, respectively (d). For all rats, a lever press
reliably followed a burst of wheel running, but wheel running did not reliably
follow a lever press.
The duration offood cup activity was usually shorter than the wheel-running
bursts, and the associated IRTs were also less than the IRTs associated with
wheel running (e). Prolonged food cup activity was usually obtained after pellet
delivery, with the associated IRT simultaneously increased (f). When food cup
activity was followed by a changeover to wheel running or exploration, the cor-
responding IRT was considerably increased (g and h).
More variable interactions were obtained with exploration, which occurred
with variable duration and occasionally was followed by standing (i), wheel
running G), or face grooming (k). The associated IRT also varied and were
especially long when exploration was followed by standing or grooming.
Activity around the lever obviously occurred with the emission of lever
pressing and was usually associated with brief IRTs (I). However, bursts of
lever activity occasionally occurred for rat 3 without actual lever pressing, with
the associated IRT varying with the burst duration of lever activity (m).
Chapter 5: Collateral Responses with Simple Schedules 283

RATt
21 Tr'T'T1I'IiIII, i I _ iiiO
11 .IIi iII Ii II Ai _i _11I i _ hl l i i _111111 1_i_i i a.I I I II Ii _.
II If ___
I I I Ii .0._ __
i i ' II i iii III II i i Ii., i I ,'i I I I iUll"ill II• "ill I i

4~ ......... ""W"" w" .ali 'Ii, •• h"'h i j i l l •••• " •• II Ii ...... Ii' '_H'. 1.4.11.",. , i .WiaIM .. '¥nrYV

~ r.RA.~,T~.2:.,~.~,,~.,~,,~.,~u~'::'~"~'~'~"~':"~:"~':"~"~":O;;'~';:::'~':";"~':"~"~.~I~':"'~'~~~.~~~~~~~~~~~:.:
_ .... ~ ~I ." . i i ,I all i .1i I .. I' .n ijj CD _;lIlli, 1M lill llil iii

4 ...... . . . . .... WI . . . ,.. 1 /1' i . . . . . . . . . . . . . . 'i'" ..... rt Ii _ WI Wi Wt ."" • • •• • • , ... ~~~~..............-

~==~==~~========;=============~=================
RAT 3
b b
11 111 11 ' ''1111111111 11 111111:1111111111_ , IJilli li , i iU tli lii i lin i Ii iii [iii II i iiilli lII i"'i' Iii Ii I I i liilliillliAli i "Wij

3,,, j"'j""".i'''ji~ij'I''''lhi''ill~'"'b''''''''''''.'iii'''i i

6~~ m h m d
2 RAT 4
lil i th' " i ii i Ii .811 11111 ir--Tf'J ii' . ii i I. i I Itl Il pi CUii " i IA1I iJ ill 1111.' IllIi III
____ ____
iii i I .----m

3. " I " " , -"""

" " ................,.. ' -~--'---""--"o-ro-.--.,....--on~'--''''-'~''''--'''''---'''----r--~
4~" .0. I n. 'N •____-r.--..-........._..-............ -M_.on.~ ...... __
~.. _..• .____.._~"'.._..~...__ ___........ ~~ .........
s ~~~~~~~~~~~-~~ ____ ~~~ __ ~-

6 b b .....,....~-b:-------:---:-----------g----------
30 ........

Figure 5.17. Event records showing mUltiple response interactions. Pellet deliveries are shown as
I-sec deflections of pen I. Pen 5 is continuously deflected for exploring responses and rapidly
pulsed for standing responses. Similarly, pen 6 is continuously deflected for grooming of the face
and rapidly pulsed for body grooming. I, lever/pellet; 2, wheel; 3, food cup activity; 4, lever activ-
ity; 5, explore/stand; 6, grooming: body/face.

Overall, different length IRTs appeared to be associated with emission of

different collateral responses; the collateral response that occurred right after a
lever press thus provided some clue as to the duration of the upcoming IRT.
Relative frequencies of IRTs are given in Figure 5.18 for each rat. The top
row gives the conventional calculation of the frequency of a given IRT divided
by the number of all IRTs. The subsequent rows give the relative IRT frequen-
cies conditional upon the collateral response that initiated the IRT (the first col-
lateral response within 0.5 sec of the previous lever press). This calculation re-
lates the frequency of a given IRT initiated by a particular collateral response to
the frequency of all IRTs initiated by that collateral response.
The IRT distribution without regard to the collateral response was about the
same for each rat, with the highest frequencies at relatively short IRTs, and
gradually decreasing frequencies for the longer IRTs. In comparison, the distri-
bution of IRTs initiated by lever activity was quite skewed toward relatively
briefIRTs. The mean ofIRTs initiated by lever activity was much shorter than
the mean of all IRTs.
The IRTs initiated by food cup acitivity resembled the overall IRT distribu-
tion for all rats. The peak frequency of IRTs initiated by wheel running was
skewed toward longer IRTs, with the mean IRT about twice as long as the col-
lective mean for all IRT. (For rat 3, a wheel-running distribution was not ob-
284 Section II: Operant Conditioning Procedures

Q4
Q3

0.1
Q2
L RAT 1
,900
M01.S6
RAT 2
No1125
RAT 3
~
RAT 4

i 11" ..... ,..

No602
Mo22S

II .. "",

LI
0.6 Nom M=3"
~~ 05 r M°o.71
>
u "'~ 04 I
z ~t; 0.3
W ...... 02
:::J 0.1
o iii
w

~ B; Q3~
0:
N.. S8

~
Ii. N=31'
1 M°1.91 Mo227

: e~ lit"
w .... Q> 02 ,
~ Ql • tI, ..... II ....
~ !!;
W
0:

i "lllJir. ....
Ntl82 N·131 Nom

~ g~J 1 ::~s,~ ~ M=4.72

II [!fIr. ,"
M·2.29 M=2.36

~ 0.11Ilh"' .. 1" 1 ,
1 2 3 4 5 6 7 1 910 12345&71910
.. ~ dUh! , , ,
12345171910
iii
i'l'dI'II",1""",
12345678910

INTERRESPONSE TIME (seconds)

Figure 5.18. Relative frequencies ofIRTs. The top distribution includes all IRTs. Each ofthe sub-
sequent distributions shows the relative frequencies of IRTs initiated by lever activity, food cup
activity, wheel running, or exploration. The collateral response initiating an IRT was defined as the
response that followed a lever press within 0.5 sec. Iflever activity occurred throughout the 0.5-sec
period, then the IRT was initiated by lever activity. Data were calculated from event records from
the last session. The number (N) of IRT and the mean (M) IRT initiated by that collateral response
are shown in each panel.

tained because wheel running never occurred right after a lever press.) Finally,
IRTs initiated by exploration were relatively short, but with a gradually de-
creasing frequency for longer IRTs. The mean IRTs initiated by exploration
was therefore higher than the overall mean for all IRTs. Standing and grooming
never initiated IRTs.
The mean IRT then changes with different collateral responses. Therefore,
the collateral response changed over to provides an estimate of the associated
IRT. An IRT calculation based upon the initiating collateral response may thus
provide a rather useful tool for a molecular analysis. Simply, a given IRT could
be predicted far more accurately by analyzing the initiating collateral response
than by knowing only the average of all IRT.
The data were also analyzed to determine the probability of a changeover to
a lever press conditional upon the momentarily occurring collateral response.
The changeover probability within a given bin, conditional upon the particular
collateral response, was calculated as follows: within a given bin, the number
of times a particular collateral response was followed by a lever press was di-
vided by the total number of times that collateral response occurred in that bin.
The calculation of the probability of a changeover to a lever press irrespective
of a collateral responding is the ordinary calculation of IRTs per opportunity.
Chapter 5: Collateral Responses with Simple Schedules 285

Table 5.3 Rat 1: Probabilities of terminating an IRT within a given bin conditional
upon occurring collateral response or irrespective of collateral response

Conditional upon collateral response

Irrespective of Lever Food cup Wheel

Bin collateral response activity activity running Exploring Standing
0.0-0.5 0.245 0.577 0.131 0.000 0.000
0.5-1.0 0.325 0.809 0.408 0.000 0.156
1.0-1.5 0.419 1.000 0.677 0.000 0.386 0.200
1.5-2.0 0.277 0.500 0.123 0.359 0.625
2.0-2.5 0.419 0.769 0.228 0.625 0.666
2.5-3.0 0.279 0.666 0.159 0.765 0.000
3.0-3.5 0.288 1.000 0.257 0.250 0.000
3.5-4.0 0.298 0.291 1.000 1.000
4.0-4.5 0.500 0.538
4.5-5.0 0.750 0.888
5.0-5.5 0.400 1.000

The probabilities of terminating IRTs are shown in Tables 5.3 through 5.6. In
each table, the first column gives the probability of a lever press as a function of
the time since the last lever press irrespective of collateral responding (the ordi-
nary IRTs per opportunity distribution). The subsequent columns show the
probability of a lever press conditional upon the collateral response emitted
preceding the lever press. (Note that the probability of a lever press conditional
upon a collateral response is only defined for actual occurrence of that collat-
eral response in that bin. For example, the probability of a lever press condi-
tional upon standing was not defined for the first bins, because standing never
occurred in these bins. In contrast, the probability of lever pressing conditional

Table 5.4 Rat 2: Probabilities of terminating an IRT within a given bin conditional
upon occurring collateral response or irrespective of collateral response
Conditional upon collateral response
Irrespective of Lever Food cup Wheel
Bin collateral response activity activity running Exploring Standing
0.0-0.5 0.200 0.232 0.282 0.000 0.000
0.5-1.0 0.525 0.906 0.453 0.000 0.392
1.0-1.5 0.371 1.000 0.497 0.005 0.398 0.000
1.5-2.0 0.249 0.412 0.000 0.571 0.000
2.0-2.5 0.327 0.560 0.127 0.700 0.375
2.5-3.0 0.507 0.360 0.429 OAOO 0.406
3.0-3.5 0.567 0.500 0.492 0.666 0.333
3.5-4.0 0.600 0.170 0.636 1.000 0.500
4.0-4.5 0.417 0.200 1.000 0.000
4.5-5.0 0.500 1.000 0.000
5.0-5.5 0.500 0.500
5.5-6.0 0.000 0.000
6.0-6.5 0.666 1.000
286 Section II: Operant Conditioning Procedures

Table 5.5 Rat 3: Probabilities of terminating an IRT within a given bin conditional
upon occurring collateral response or irrespective of collateral response

Conditional upon collateral response

Irrespective of Lever Food cup Wheel

Bin collateral response activity activity running Exploring Standing

0.0-0.5 0.141 0.166 0.199 0.000

0.5-1.0 0.397 0.543 0.410 0.000 0.098
1.0-1.5 0.466 0.769 0.453 0.025 0.209 0.000
1.5-2.0 0.286 0.676 0.595 0.013 0.222 0.000
2.0-2.5 0.297 0.808 0.666 0.038 0.415 0.180
2.5-3.0 0.314 0.714 0.600 0.147 0.480 0.620
3.0-3.5 0.415 0.857 0.500 0.406 0.333 0.333
3.5-4.0 0.800 1.000 0.000 0.816 0.800 1.000
4.0-4.5 0.818 0.000 1.000 1.000
4.5-5.0 0.666 1.000

upon wheel running was zero in the first bins for rat 1 (Table 5.3) because a
lever press never occurred in this bin. Note also that the number of bins in each
table is less than in Figure 5.18. IRT above 5 or 6 sec were generally few and
are not shown.)
Between 0.0 and 1.5 sec (the first three bins) the probability of a lever press
conditional upon lever activity was much higher than the probability of a lever
press irrespective of the occurring collateral response. The probability of a
lever press conditional upon wheel running or exploration was relatively low or

Table 5.6 Rat 4: Probabilities of terminating an IRT within a given bin conditional
upon occurring collateral response or irrespective of collateral response

Conditional upon collateral response

Irrespective of Lever Food cup Wheel

Bin collateral response activity activity running Exploring Standing
0.0-0.5 0.273 0.436 0.241 0.000 0.000
0.5-1.0 0.444 0.876 0.415 0.000 0.054
1.0-1.5 0.145 1.000 0.267 0.000 0.060 0.000
1.5-2.0 0.087 0.666 0.022 0.110 0.143
2.0-2.5 0.111 0.500 0.011 0.197 0.333
2.5-3.0 0.077 1.000 0.045 0.178 0.000
3.0-3.5 0.155 0.188 0.154 0.250
3.5-4.0 0.183 0.188 0.269 0.276
4.0-4.5 0.177 0.214 0.333 0.100
4.5-5.0 0.352 0.455 0.360 0.143
5.0-5.5 0.316 0.500 0.500 0.500
5.5-6.0 0.128 0.250 0.000 0.333
6.0-6.5 0.215 0.333 0.000 0.500
6.5-7.0 0.121 0.500 1.000 0.000
7.0-7.5 0.207 0.500 1.000
7.5-8.0 0.217 1.000
Chapter 5: Collateral Responses with Simple Schedules 287

zero in the first bins. Therefore, for all rats, a changeover to lever pressing
within the first three bins was far more likely given lever activity than given
wheel running or exploration.
In subsequent bins, the probability of a lever press conditional upon food
cup activity was typically higher than the corresponding probability conditional
upon either wheel running, exploration, or standing. Simply, the probability of
a lever press conditional upon a particular collateral response was substantially
different from the probability calculated irrespective of collateral responses.
The data were complex however, and might be illustrated by a few examples.
For rat 1 (Table 5.3), the probability of a lever press within the 2.5 to 3.0 bin
was relatively high given the simultaneous occurrence of either food cup activ-
ity or exploration within that bin, but was relatively low given wheel running or
standing. For rat 3 (Table 5.5), the lever-pressing probability within the 2.0 to
2.5 bin was relatively high given either lever activity, food cup activity, or ex-
ploration, but was low given either wheel running or standing.
Although grooming occurred only rarely, the data were entirely consistent
with the analysis in Tables 5.3 to 5.6. Body grooming usually occurred in rela-
tively long bursts for all rats, and then resulted in very long IRTs and a low
probability of a changeover to a lever press.
In retrospect, the time since the last lever press and the collateral response
occurring at any given time yielded a fine-grained estimate of the temporal
probability of lever pressing.

Discussion
Pauses in one response apparently depend upon the particular concurrent re-
sponse engaged at the onset of that pause. Pause termination may then be pre-
dicted well by the initiating collateral response rather than the average pause
duration. However, the initiating collateral response does not solely determine
the total IRT since a changeover from one collateral to a second frequently
occurs within an IRT. Consequently, the collateral response occurring at a
given time since the IRT onset provides a further clue to the probability of ter-
minating that IRT.
One view holds that the probability of a given response is determined by the
stimulus complex at any given moment (Estes, 1950; Bush and Mosteller,
1955). The present data also indicate that the collateral response at IRT onset is
a powerful determinant of the IRT distribution. The lever-pressing probability
at any given timex since IRT onset would thus depend upon the duration ofthe
initiating collateral response and all subsequent collateral responses. However,
at this stage of the sequential analysis, the relationship between IRT termina-
tion and collateral responses is perhaps best described as a two-link depen-
dency. The first-order determinant ofIRT termination is the collateral response
emitted at IRT onset. The second-order determinant is the collateral response
emitted at any time x since IRT onset. Clearly, estimates based simply on time
per se irrespective of concurrent behaviors would not seem to be equally accu-
rate in the estimation of IRT termination.
288 Section II: Operant Conditioning Procedures

The data are more akin to a second model, which holds that the length of an
IRT is determined at IRT onset. Shimp (1969) suggested that a subject chooses
which IRT to emit rather than whether or not to respond at a given time. The
present data would further argue that a particular IRT is probably more of a
consequence than a "cause" of the collateral response emitted during the IRT.
Shimp (1969) originally suggested that the subject chooses which "mediating
behavioral chain to initiate," and the IRT then follows as a consequence of the
collateral response selected at IRT onset. Reynolds and McLeod (1970) also
suggested that relatively briefIRTs may be associated with one mode of collat-
eral responses (such as standing in front of the response key), whereas rela-
tively long IRTs may be associated with different modes of collateral responses
(such as exploring and grooming).
The present data certainly agrees with the view that the examination of mo-
lecular behavior patterns is required for a comprehensive behavioral analysis.
In support of this argument, Gray (1976) reported differential stimulus control
of different IRT lengths ofDRL responding during generalization testing, while
at the same time stimulus control was obscured by averaging different IRT
lengths. Also the results from Experiment VIII in Chapter 3 clearly showed
that a response latency may be predicted on the basis of the concurrent re-
sponse occurring at latency onset. The progressive development of the analysis
of response sequences and patterns would emphasize a more conditional ap-
proach including the collateral responses that initiate IRTs or latencies.
From whatever perspective, the accruing data strongly indicate that the
operant IRT is closely related to collateral responses. Fundamentally, one
"choice" occurs when the subject changes over to a collateral response at IRT
onset; the next "choice" occurs with the changeover from a collateral response
to terminate the IRT. This analysis is consistent with the view that preference
or "choice" is the changeover from one response to another (Skinner, 1950).

Concluding Comments
Patterns of responses maintained by simple operant conditioning seem to be
systematically interrelated with collateral response characteristics. The re-
sponse interactions generalized across many different procedures and consist-
ently replicated the behavioral interactions reported with concurrent operant
schedules and multiple schedules, as well as classical-operant combinations.
Clearly, the present results strengthen the argument that "other" collateral re-
sponses may affect reinforced responses and hence serve as independent vari-
ables.
The results are first of all consistent with many recent reports of interactions
between reinforced and collateral responses within simple operant conditioning
(Clark, 1962; Segal and Bandt, 1966; Laties et al., 1969; Dunham, 1971; Wayner
and Greenberg, 1972; Cohen and Looney, 1973; Gilbert, 1974; Colotla and
Keehn, 1975; Cook and Singer, 1976; Anderson and Shettleworth, 1977).
In particular, the present results are in close accord with the recent emphasis
Chapter 5: Collateral Responses with Simple Schedules 289

on response competition as penetrating many conditioning procedures. Wong

(1977), for example, suggested a "behavioral field approach" of combining nat-
uralistic observation and experimentation. According to Wong, learning and
extinction may be viewed as representing narrowing and broadening ofthe be-
havioral field that surrounds the instrumental response. Ray and Brown (1975),
Anderson and Shettleworth (1977), Rand (1977), and Staddon (1977) have sug-
gested a similar study of the behavioral interrelationships in conditioning situa-
tions. The mutual reinforcement of consistent data within and across labora-
tories supports the generalization that collateral responses may be fundamental
determinants within conditioning procedures.
Response interactions have also been of some advantage to applied research
areas such as the study of physiological and pharmacological processes.
Changes in reinforced behavior after central nervous system lesions, for exam-
ple, are clearly associated with changes in specific collateral responses (Slon-
aker and Hothersall, 1972; Wayner and Greenberg, 1972). Ellen et al. (1977)
also suggested that such lesion-induced changes might be critically affected by
competing behaviors hitherto unexamined. In the field of behavioral pharma-
cology there is a growing emphasis on the incorporation of concurrent behav-
iors into the functional analysis of pharmacological agents (Wuttke and Innis,
1972; Smith and Clark, 1975; Wallace and Singer, 1976; Sanger, 1977).
Our stress on increased attention to interacting responses also ties in with
recent considerations that simple operant schedules may be viewed as concur-
rent schedules (Hermstein, 1970, 1974). Similarly, Staddon (1977) considers
reinforced responses to be in a reciprocal relationship with other behaviors.
These models, however, consider response interactions as coming about by
means of changes in underlying reinforcement values or behavioral states. This
emphasis would retain rather molar constructs where response averages are ei-
ther synonymous or isomorphic with the explanatory values and states. The
many experimental findings in this and previous chapters in tum argue that mo-
lecular interactions are also important, and in fact appear to be rather system-
atic and controllable. The more specific implications of interacting response
patterns will be discussed next.
One critical feature of these interactions is that the same collateral response
may both reinforce and inhibit the ongoing operant response. Collateral licking
thus suppressed FR responding during the drinking probe but facilitated re-
sponding in the absence of the probe, provided that access to drinking was sched-
uled early in the FR run or response independently. The reinforcing and sup-
pressive effects of changeovers to collateral responses hence refer to separate
effects within different temporal intervals and describe the sequential effects of
operant-collateral pairings or the concurrent effects of simultaneous access to
operant and collateral responses. Changeovers to collateral responses may
then be arranged to serve both as sequential reinforcers or concurrent inhibi-
tors of operant responding. Consequently, collateral responses do not compete
"because" they are reinforcers, since competition and reinforcement describe
different response interactions in different temporal intervals.
The present experiments were primarily concerned with the relationships
290 Section II: Operant Conditioning Procedures

between pausing in reinforced responding and concurrent emission of collateral

responses. Postreinforcement pause (pRP) describes the pause between rein-
forcer delivery and the beginning of the run period of responding. This usage
was intended to be descriptive. Recently, the termPRP was characterized as a
"misnomer" because the pause is considered to be largely a function of the
upcoming ratio size or run time (Griffiths and Thompson, 1973). Neuringer and
Schneider (1968) and Barowsky and Mintz (1975) reported that the PRP was
increased by inserting time-out periods to lengthen the run time of FR respond-
ing. In the present experiments, the PRP was altered by either manipulating the
duration of collateral licking during the pause, without changing the run time,
or by increasing the run time by inserting a drinking probe within the run period
(Experiment IVA). In this case, however, the PRP was decreased by lengthen-
ing the run time. So increasing the run time by time-out is apparently not func-
tionally equivalent to lengthening the run time by inserting collateral licking.
These findings suggest that the PRP depends upon other parameters in addition
to the run time [also see Dardano and Sauerbrunn (1964), Findley and Brady
(1965) and Boren (1961) for manipulations of the PRP by alterations in the run
period of FR responding.] In summary, the PRP appears to be critically de-
pendent upon the particular events occurring during the run period as well as
during the PRP, rather than only upon the duration of the run period per se.
Of equal interest, pausing in FR responding is not invariably located immedi-
ately after reinforcement; rather, multiple pauses may be effected by locating
experimental events, such as drinking probes, during the run period. Such
pauses seem to be associated with different response interactions, with pauses
after pellet delivery associated with food-cup contact and pauses during the run
period associated with drinking. The exact location of the experimental probe
further determines the relative duration of each of these pauses. Pauses in sim-
ple FR schedules may thus be fruitfully considered to be multiplicatively deter-
mined.
Different pause terminologies have been suggested to reflect the different
functional relations: pause after reinforcement, postrun pause, prerun pause,
and between-run pause (e.g., Griffiths and Thompson, 1973). The descriptive
term PRP seems at least to serve the purpose of referring to the location of a
pause after reinforcement as opposed to a pause within the run period.
In any case, the present experiments amplify the variety of functional rela-
tions between bursts of collateral responses and FR pauses. The manipulated
access to a collateral response was a precise predictor of the simultaneous
pause in the ongoing operant response. The opposite interpretation, that the FR
pause predicts the simultaneous duration of a particular collateral response,
however, provides a much less clear-cut account. One particular collateral re-
sponse is not clearly assessed in retrospect from a particular pause, since a
pause may be generated by various collateral responses and changeovers be-
tween them, which is not evidenced by the operant pause per se. However, the
operant pause is more immediately known at the moment of the changeover to
one or another collateral response. Therefore, pausing or absence of one re-
sponse does not invariably discriminate between the occurrence of one or more
Chapter 5: Collateral Responses with Simple Schedules 291

collateral responses. More clearly, the data show that the absence of an
operant response is not associated with a homogeneous distribution of inter-
changeable collateral responses.
The relationship between pauses and collateral responses, however, is not
merely a one-way interaction. The interdependence between reinforced and
collateral responses was probably more clearly apparent in Experiment IV, in
which collateral licking served both as an independent and a dependent vari-
able. The present results would then be misread to suggest that all pauses in
reinforced responding are caused by collateral responses. The observed mutual
interdependencies among end terms jeopardize the fixing of simple causative
relations among overt behaviors as well as physiological responses (Horridge,
1969).
The analysis offered in Part II was originally initiated by observations of an
inverse relationship between one reinforced and one other concurrently avail-
able but nonreinforced response (Iversen, 1973). The experimental analysis
quickly becomes problematic when more than one or two responses are re-
corded. However, recording only one response does not so much prevent as
avoid the analytic problems. For example, the addition of an LH contingency
to a VI schedule not only increased the rate of VI responding, but also de-
creased the rate of collateral licking and decreased the rate of VI reinforcement
(Iversen, 1975b). If only the VI response had been recorded, the interpretation
would have been superficially straightforward-the LH contingency presum-
ably reinforced IRTs shorter than the LH duration. Had only the licking re-
sponse been recorded, the interpretation could have been that the LH reduced
either the reinforcement rate or the opportunity for adventitious reinforcement
of licking. However, with both responses recognized, the increase in VI re-
sponding may be interpreted as responsible for the decrease in licking, and vice
versa. The analysis was complicated by the fact that only two responses were
recorded and would have been even more complicated by incorporating still
more responses.
The now more firmly established relationship between collateral and rein-
forced responses recognizes the possibility that an independent variable might
have an indirect effect brought about by means of affected collateral responses.
For example, a particular brain lesion might directly affect the pattern of a par-
ticular collateral response and thereby change the pattern of the reinforced re-
sponse (Wayner and Greenberg, 1972). Conversely, yet another lesion might
directly affect the reinforced response and only indirectly affect collateral re-
sponses. The analysis would be complicated, but necessarily so since the basic
question remains unsolved by recording only one response of theoretical inter-
est.
Clearly, a functional relationship can be established between an independent
variable and a dependent variable without any concern for the molecular nu-
ances of the relationship. However, a science has been said to progress when a
once established relationship is subdivided into evermore detailed functional
relationships (Kuhn, 1970). A previous suggestion by Findley thus pointed out
that' 'the way to increase our understanding of behavior is not to analyze a par-
292 Section II: Operant Conditioning Procedures

ticular bit of behavior exhaustively, but rather to complicate the sample of be-
havior under investigation as rapidly as good experimental and technological
procedure permit" (1962, p. 114). Various recording devices are now available
for the simultaneous monitoring and recording of several behaviors. The pres-
ent series of experiments suggests that the analysis of such behavioral interac-
tions may provide a fruitful starting point for a more comprehensive analysis of
previously established laboratory facts.
A long tradition in the experimental analysis of behavior has favored overall
averages sampled over minutes, hours, or days as dependent variables, with fre-
quent arguments that behavior is better represented at molar than molecular
levels. Molar averages are easy to deal with. They smoothly fit hand in hand
with theoretical constructs in a simple one on one fashion (motive up-response
up, motive down-response down). This straight isomorphism between con-
struct and molar behavior does not, however, strike the observer of performing
animals as being partiCUlarly useful. A subject emits one response and then
soon another, but at which moment is the underlying conceptual determinant in
force? Molecular changes in behavior have been rendered "chaotic" and not
consistent with theory, and eventually have been disregarded as data. How-
ever, an experimental analysis must eventually cope with precisely such molec-
ular behavior patterns, especially since averages obviously are made up of mo-
lecular changes. Consider a previous argument that molecular patterns are not
important because a molar measure may remain unchanged in spite of altera-
tions in molecular patterns. We may reverse the argument: if the molecular pat-
tern changes but the molar pattern does not, then the molar pattern is not suffi-
ciently sensitive or relevant to the experimental manipUlation. Molecular
changes are significant, and worthy of study in and of themselves, in spite of
the possibility that they may have no effect on an overall molar measure.
Molecular interactions of this nature are perhaps best illustrated by the last
experiment in this series. The distribution of operant IRTs was related to differ-
ent collateral responses initiating each IRT. Essentially, the collateral response
initiating an IRT proved to be a powerful determinant of the duration of the
IRT. Knowledge of the collateral response and the previous distribution of cor-
responding IRTs yields a prediction of the ensuing IRT duration.
The ongoing collateral response at any given moment also provides informa-
tion on the immediate probability of terminating the IRT. Collateral responses
such as lever activity and food cup activity, which occurred in relatively short
bursts, allow the prediction that the probability of IRT termination is relatively
high. Wheel running, exploration, standing, and grooming, however, which
occur in relatively long bursts, all indicate a much lower probability ofIRT ter-
mination. The collateral response at any given time therefore provides defini-
tive information on the characteristic distribution of operant behavior in time.
Overall, the results are entirely consistent with the more general conclusion
from previous chapters that scheduled events "inhibit" one response to the ex-
tent that they control emission of concurrent responses. A significant step in
the control and prediction of response probability at a given time is therefore
provided by incorporation of concurrent responses into the behavioral analy-
Chapter 5: Collateral Responses with Simple Schedules 293

sis. The precise control of molecular response patterns might then be well
within the domain of operant conditioning and analysis.

References
Allen, J. D., and Porter, J. H.: Sources of control over schedule-induced drinking produced by
second-order schedules of reinforcement. Physiol. Behav., 18,853-863,1977.
Allen, J. D., Porter, J. H., and Arazie, R.: Schedule-induced drinking as a function of percentage
reinforcement. J. Exp. Analy. Behav., 23,223-232, 1975.
Amse1, A.: The role offrustrative nonreward in noncontinuous reward situations. Psychol. Bull.,
55, 102-119, 1958.
Amsel, A., and Roussel, J.: Motivational properties of frustration: I. Effect on a running response
of the addition of frustration to the motivational complex. J. Exp. Psychol., 43, 363-368, 1952.
Anderson, M. C., and Shetdeworth, S. J.: Behavioral adaptation to fixed-interval and fixed-time
food delivery in golden hamsters. J. Exp. Anal. Behav., 27,33-49, 1977.
Anger, D.: The dependence of interresponse times upon the relative reinforcement of different in-
terresponse times. J. Exp. Psychol., 52, 145-161, 1956.
Ator, N.: Interchangeability of mirror attack and response-produced time out in pigeons. Paper
presented at the meeting of the Americal Psychological Association, Washington, D. C., Sep-
tember 1976.
Azrin, N. H., Hutchinson, R. R., and McLaughlin, R.: The opportunity for aggression as an
operant reinforcer during aversive stimulation. J. Exp. Anal. Behav., 8, 171-180, 1965.
Barowsky, E. I., and Mintz, D. E.: The effects of time-out locus during fixed-ratio reinforcement.
Bull. Psychonomic Soc., 5, 137-140, 1975.
Berlyne, D. E.: Conflict, Arousal, and Curiosity. New York, McGraw-Hill, 1960.
Bolles, R. C.: Theory of Motivation. New York, Harper & Row, 1967.
Boren, J. J.: Stimulus probes of the fixed ratio run. Paper presented at Eastern Psychological Asso-
ciation, Philadelphia, 1961.
Boren, 1. 1.: The study of drugs with operant techniques. In, W. K. Honig (ed.): Operant Behavior:
Areas of Research and Application. New York, Appleton-Century-Crofts, 1966.
Breland, K., and Breland, M.: The misbehavior of organisms. Am. Psychol., 16,681-684, 1961.
Bush, R. R., and Mosteller, F.: Stochastic Models for Learning. New York, Wiley, 1955.
Carlisle, H. J.: Fixed-ratio polydipsia: termal effects of drinking, pausing, and responding. J. Compo
Physiol. Psychol., 75, 10-22, 1971.
Carlisle, H. 1.: Schedule-induced polydipsia: effects of water termperature, ambient temperature,
and hypothalamic cooling. J. Compo Physiol. Psychol., 83,208-220, 1973.
Catania, A. C., and Gill, C. A.: Inhibition and behavioral contrast. Psychonomic Sci., 1,257-258,
1964.
Clark, F. C.: Some observations on the adventitious reinforcement of drinking under food rein-
forcement. J. Exp. Anal. Behav., 5,61-63, 1962.
Cohen, P. S., and Looney, T. A.: Schedule-induced mirror responding in the pigeon. J. Exp. Anal.
Behav., 19,395-408, 1973.
Cole, J. M., and Parker, B. K.: Schedule-induced aggression: access to an attackable target bird as
a positive reinforcer. Psychonomic Sci., 22, 33-35, 1971.
Coloda, V. A., and Keehn, J. D.: Effects of reinforcer-pellet composition on schedule-induced pol-
ydipsia with alcohol, water, and saccharin. Psychol. Record, 25, 91-98, 1975.
Cook, P., and Singer, G.: Effects of stimulus displacement on adjunctive behavior. Physiol.
Behav., 16, 79-82, 1976.
Corfield-Sumner, P. K., Blackman, D. E., and Stainer, G.: Polydipsia induced in rats by second-
order schedules of reinforcement. J. Exp. Anal. Behav., 27,265-273,1977.
Dardano, J. F., and Sauerbrunn, D.: Selective punishment of fixed-ratio performance. J. Exp.
Anal. Behav., 7,255-260, 1964.
Dews, P. B.: The effect of multiple S4 periods on responding on a fixed-interval schedule: V. Effect
 294 Section II: Operant Conditioning Procedures

of periods of complete darkness and of occasional omissions offood presentations. J. Exp. Anal.
Behav., 9, 573-578, 1%6.
Dunham, P. J.: Punishment: method and theory. Psychol. Rev., 78,58-70, 1971.
Dunham, P. J.: Some effects of punishment upon unpunished responding. J. Exp. Anal. Behav., 17,
443-450, 1972.
Ellen, P., Gillenwater, G., and Richardson, W. K.: Extinction responding by septal and normal rats
following acquisition under four schedules of reinforcement. Physiol. Behav., 18,609-615, 1977.
Estes, W. K.: Toward a statistical theory oflearning. Psychol. Rev., 57,94-107,1950.
Falk, J. L.: The motivational properties of schedule-induced polydipsia. J. Exp. Anal. Behav., 9,
19-25, 1966.
Falk, J. L.: Conditions producing psychogenic polydipsia in animals. Ann. N. Y. Acad. Sci., 157,
569-593, 1969.
Falk, J. L.: The nature and determinants of adjunctive behavior. Physiol. Behav., 6,577-588,1971.
Felton, M., and Lyon, D.O.: The post-reinforcement pause. J. Exp. Anal. Behav., 9, 131-134,
1966.
Ferster, C. B., and Skinner, B. F.: Schedules of Reinforcement. New York, Appleton-Century-
Crofts, 1957.
Findley, J. D.: An experimental outline for building and exploring multi-operant behavior reper-
toires. J. Exp. Anal. Behav., 5, 113-166, 1962.
Findley, J. D., and Brady, J. V.: Facilitation of large ratio performance by use of conditioned rein-
forcement. J. Exp. Anal. Behav., 8, 125-129, 1965.
Flory, R. K.: Attack behavior in a multiple fixed-ratio schedule of reinforcement. Psychonomic
Sci., 16, 15615-157, 1%9.
Flory, R. K., and O'Boyle, M. K.: The effect of limited water availability on schedule-induced
polydipsia. Physiol. Behav., 8, 147-149, 1972.
Gilbert, R. M.: Ubiquity of schedule-induced polydipsia. J. Exp. Anal. Behav., 21,277-284, 1974.
Gray, V. A.: Stimulus control of differential-reinforcement oflow-rate Iesponding. J. Exp. Anal.
Behav., 25, 199-207, 1976.
Griffiths, R. R., and Thompson, T.: The post-reinforcement pause: a misnomer. Psychol. Record,
23,229-235, 1973.
Guthrie, E. R.: The Psychology of Learning. New York, Harper, 1935.
Hearst, E.: Some persistent problems in the analysis of conditioned inhibition. In, R. A. Boakes
and M. S. Halliday (ed.): Inhibition and Learning. London and New York, Academic Press,
1972.
Herrnstein, R. J.: On the law of effect. J. Exp. Anal. Behav., 13,243-266, 1970.
Herrnstein, R. J.: Formal properties of the matching law. J. Exp. Anal. Behav., 21, 159-164, 1974.
Hodos, W., Ross, G. S., and Brady, J. V.: Complex response patterns during temporally spaced
responding. J. Exp. Anal. Behav., 5,473-479, 1962.
Honig, W. K.: Introductory remarks. In, W. K. Honig (ed.): Operant Behavior: Areas ofResearch
and Application. New York, Appleton-Century-Crofts, 1966.
Horridge, G. A.: The interpretation of behavior in terms of interneurons. In, M. A. B. Brazier
(ed.): The Interneuron. Los Angeles, University of California Press, 1%9.
Hull, C. L.: A Behavior System. New Haven, Yale University Press, 1952.
Huston, J. P., and Desisto, M. J.: Interspecies aggression during fixed-ratio hypothalamic
self-stimulation in rats. Physiol. Behav., 7,353-357, 1971.
Hutchinson, R. R., Azrin, N. H., and Hunt, G. M.: Attack produced by intermittent reinforcement
of a concurrent operant response. J. Exp. Anal. Behav., 11,489-495, 1968.
Iversen, I. H.: Interactions between reinforced and non-reinforced responses. Paper presented at
the Third Scandinavian Meeting on Physiological Psychology, Copenhagen, May 1973.
Iversen, I. H.: Concurrent responses during multiple schedules in rats. Scand. J. Psychol., 16,49-
54, 1975a.
Iversen, I. H.: Interactions between lever pressing and collateral drinking during VI with limited
hold. Psychol. Record, 25,47-50, 1975b.
Iversen, I. H.: Interactions between reinforced responses and collateral responses. Psychol. Rec-
ord, 26,399-413, 1976.
Chapter 5: Collateral Responses with Simple Schedules 295

Iversen, I. H.: Behavioral pauses and their relation to concurrent behaviors. Paper presented at the
Easter Conference of the English Experimental Analysis of Behavior Group. Exeter, March
1977a.
Iversen, I. H.: Reinforcement omission and schedule-induced drinking in a response-independent
schedule in rats. Physiol. Behav., 18,535-537, 1977b.
Keehn, J. D., and Colotla, V. A.: Schedule-induced drinking as a function of interpellet interval.
Psychonomic Sci., 23,69-71, 1971.
Killeen, P.: On the temporal control of behavior. Psychol. Rev., 82,89-115,1975.
Kuhn, T. S.: The Structure of Scientific Revolutions. International Encyclopedia of Unified Sci-
ence, vol. 2, No.2, 2nd ed. Chicago, University of Chicago Press, 1970.
Laties, V. G., Weiss, B., Clark, R. L., and Reynolds, M. D.: Overt "mediating" behavior during
temporally spaced responding. J. Exp. Anal. Behav., 8, 107-116, 1965.
Laties, V. G., Weiss, B., and Weiss, A. B.: Further observations on overt "mediating" behavior
and the discrimination of time. J. Exp. Anal. Behav., 12,43-57, 1969.
Lyon, D.O.: Some notes on conditioned suppression and reinforcement schedules. J. Exp. Anal.
Behav., 7,289-291, 1964.
McCoy, J. F., and Christian, W. P.: Schedule-induced drinking and reinforcement omission. Phy-
siol. Behav., 17,537-539, 1976.
McLeod, D. R., and Gollub, L. R.: An analysis of rats' drinking-tube contacts under tandem and
fixed-interval schedules of food presentation. J. Exp. Anal. Behav., 25,361-370,1976.
McMillan, J. C.: Percentage reinforcement of fixed-ratio and variable-interval performances. J.
Exp. Anal. Behav., 15,297-302, 1971.
Neuringer, A. J., and Schneider, B. A.: Separating the effects ofinterreinforcement time and num-
ber of interreinforcement responses. J. Exp. Anal. Behav., 11,661-667, 1968.
Pavlov, I. P.: Conditioned Reflexes (Translated by G. V. Anrep). London, Oxford, 1927 (Re-
printed, New York, Dover, 1960).
Penney, J., and Schull, J.: Functional differentiation of adjunctive drinking and wheel running in
rats. Anim. Learning Behav., 5,272-280, 1977.
Porter, J. H., and Kenshalo, D. R.: Schedule-induced drinking following omission of reinforcement
in the rhesus monkey. Physiol. Behav., 12,1075-1077,1974.
Porter, J. H., Arazie, R., Holbrook, J. W., Cheek, M. S., and Allen, J. D.: Effects of variable and
fixed second-order schedules on schedule-induced polydipsia in the rat. Physiol. Behav., 14,
143-149, 1975.
Powell, R. W.: The effect of small sequential changes in fixed-ratio size upon the post-reinforce-
ment pause. J. Exp. Anal. Behav., 11, 589-593, 1968.
Powell, R. W.: The effect of reinforcement magnitude upon responding under fixed-ratio sched-
ules. J. Exp. Anal. Behav., 12,605-608, 1%9.
Premack, D.: Catching up with common sense or two sides of a generalization: reinforcement and
punishment. In, R. Glaser (ed.): The Nature of Reinforcement. New York, Academic Press,
1971.
Rand, J. F.: Behaviors observed during S- in a simple discrimination learning task. J. Exp. Anal.
Behav., 27, 103-117, 1977.
Ray, R. D., and Brown, D. A.: A systems approach to behavior. Psychol. Record, 25,459-478,
1975.
Reynolds, G. S., and McLeod, A.: On the theory of interresponse-time reinforcement. In, G. H.
Bower (ed.): The Psychology of Learning and Motivation. New York, Academic Press, 1970.
Rosenblith, J. Z.: Polydipsia induced in the rat by a second-order schedule. J. Exp. Anal. Behav.,
14, 139-144, 1970.
Sanger, D.: Schedule-induced drinking of chlordiazepoxide solutions by rats. Pharmacol. Bio-
chem. Behav., 7, 1-6, 1977.
Schaeffer, R. W., and Diehl, J. C.: Collateral water drinking in rats maintained on FR food rein-
forcement schedules. Psychonomic Sci., 4,257-258, 1966.
Schoenfeld, W. N., and Farmer, J.: Reinforcement schedules and the "behavior stream." In, W.
N. Schoenfeld (ed.): The Theory of Reinforcement Schedules. New York, Appleton-Century-
Crofts, 1970.
 296 Section II: Operant Conditioning Procedures

Segal, E. F.: Transformation of polydipsic drinking into operant drinking: A paradigm? Psycho-
nomic Sci., 16, 133-135, 1969.
Segal, E. F.: Induction and the provenance of operants. In, R. M. Gilbert and J. R. Millenson
(eds.): Reinforcement: Behavioral Analysis. New York, Academic Press, 1972.
Segal, E. F., and Bandt, W. M.: Influence of collateral water drinking on bar pressing under com-
plex reinforcement contingencies. Psychonomic Sci., 4,377-378, 1966.
Seward, J. P., Pereboom, A. c., Butler, B., and Jones, R. B.: The role of prefeeding in an apparent
frustration effect. J. Exp. Psychol., 54,445-450, 1957.
Shimp, C. P.: Optimal behavior in free-operant experiments. Psychol. Rev., 76,97-112, 1969.
Shimp, C. P.: Synthetic variable-interval schedules of reinforcement. J. Exp. Anal. Behav., 19,
311-330, 1973.
Skinner, B. F.: The Behavior of Organisms, New York, Appleton-Century-Crofts, 1938.
Skinner, B. F.: Are theories ofleaming necessary? Psychol. Rev., 57, 193-216, 1950.
Skinner, B. F., and Morse, W. H.: Concurrent activity under fixed-interval reinforcement. J.
Compo Physiol. Psychol., 50,279-281, 1957.
Slonaker, R. L., and Hothersall, D.: Collateral behaviors and the DRL deficit of rats with septal
lesions. J. Compo Physiol. Psychol., 80,91-%, 1972.
Smith, J. B., and Clark, F. C.: Effects of d-amphetamine, chlorpromazine, and chlordiazepoxide
on intercurrent behavior during spaced-responding schedules. J. Exp. Anal. Behav., 24, 241-
248, 1975.
Staddon, J. E. R.: Temporal effects of reinforcement: a negative "frustration" effect. Learning
Motivation, 1,227-247, 1970.
Staddon, J. E. R.: Schedule-induced behavior. In, W. K. Honig and J. E. R. Staddon (eds.): Hand-
book of Operant Behavior. Englewood Cliffs, New Jersey: Prentice-Hall, 1977.
Staddon, J. E. R., and Ayres, S. L.: Sequential and temporal properties of behavior induced by a
schedule of periodic food delivery. Behaviour, 54,26-49, 1975.
Staddon, J. E. R., and Innis, N. K.: An effect analogous to "frustration" on interval reinforcement
schedules. Psychonomic Sci., 4,287-288, 1966.
Staddon, J. E. R., and Innis, N. K.: Reinforcement omission on fixed-interval schedules. J. Exp.
Anal. Behav., 12, 689-700, 1969.
Staddon, J. E. R., and Simmelhag, V. L.: The "superstition" experiment: a reexamination of its
implications for the principles of adaptive behavior. Psychol. Rev., 78,3-43, 1971.
Thompson, T.: Effect of chlorpromazine on "aggressive" responding in the rat. J. Compo Physiol.
Psychol., 54,398-400, 1961.
Timberlake, W., and Allison, J.: Response deprivation: an empirical approach to instrumental per-
formance. Psychol. Rev., 81, 146-164, 1974.
Wallace, M., and Singer, G.: Schedule-induced behavior: a review of its generality, determinants
and pharmacological data. Pharmacol. Biochem. Behav., 5,483-490, 1976.
Wayner, M. J., and Greenberg, I.: Effects of septal lesions on palatability modulation and sched-
ule-induced behavior. Physiol. Behav., 9,663-665, 1972.
Wong, P. T.: A behavioral field approach to instrumentalleaming in the rat: I. Partial reinforce-
ment effects and sex differences. Anim. Learning Behavior,S, 5-13, 1977.
Wookey, P. E., and Strongman, K. T.: Reward shift and general activity in the rat.Br. J. Psychol.,
65, 103-110, 1974.
Wuttke, W., and Innis, N. K.: Drug effects upon behavior induced by second-order shcedules of
reinforcement: the relevance of ethological analysis. In, R. M. Gilbert and J. D. Keehn (eds.):
Schedule Effects: Drugs, Drinking, and Aggression. Toronto, University of Toronto Press, 1972.
Zimmerman, D. W.: Rate changes after unscheduled omission and presentation of reinforcement.
J. Exp. Anal. Behav., 15,261-270, 1971.
SECTION III

CLASSICAL CONDITIONING
PROCEDURES
Chapter 6

Response Patterning in
Classical Conditioning
Wendon W. Henton

Review of Conditioning and Competing Responses

This chapter describes an initial extension of the concurrent responses-con-
current schedules analysis to include standard and not so standard classical
conditioning procedures. Classical conditioning surely enjoys an extensive his-
tory within experimental psychology, with precursor associationistic formula-
tions apparent throughout the history of philosophy and metaphysics (e.g.,
Descartes, Locke, Hobbes, J. S. Mill, etc.; excellent reviews by Jones, 1952;
Boring, 1957). The physiology and psychology of reflexive behavior separated
from the philosophy of mentalistic associations with the early work on uncondi-
tioned reflexes by Sherrington (1906) and Sechenov (1935; also Creed et aI.,
1932). The extraordinary accomplishment of Pavlov was the systematic elabo-
ration of the unconditioned reflex model (unconditioned stimulus-uncondi-
tioned response) into the now familiar model of conditioned reflexes (condi-
tioned stimulus-conditioned response; unconditioned stimulus-unconditioned
response). The conditioned or acquired reflexes were optimistically offered as
factual alternatives to "the fantastic speculations as to the existence of any
possible subjective state in the animal which may be conjectured on analogy
with ourselves" (Pavlov, 1927, p. 161). As noted by Razran (1957), the Pav-
lovian conception of ~onditioned reflexes was pointedly aimed at scrapping the
mentalistic states and faculties then serving as explanations of behavior. Al-
though Pavlov's repeated warnings (Pavlov, 1906, 1927, 1932) against philo-
sophical explanations are frequently ignored by Western psychologists, the
Sherringtonian reflex model remains the basic analytic unit for classical
conditioning by Soviet and Eastern European investigators (Konorski, 1948,
1967; Bykov, 1958; Sokolov, 1960).
In only brief review, a CR is acquired when a "neutral" stimulus (CS) is
300 Section III: Classical Conditioning Procedures

paired with the presentation of a second stimulus that invariably elicits a spe-
cific response (UCS-UCR). Several pairings of the CS with the UCS-UCR re-
sult in the elicitation of a CR during the previously ineffective CS. Responses
recorded as conditioned reflexes include salivation, heart rate, respiration, leg
flexion, eyeblink, and many other responses elicited by unconditioned stimuli
such as food (alimentary conditioning) or electric shock (defensive condition-
ing) (review by Osgood, 1953). The CR has been variously described as identi-
cal with the UCR, a fractional component of the UCR, or a preparatory re-
sponse adjusting the organism for UCS onset (Kimble, 1961).
Classical conditioning may be arbitrarily divided into excitatory and inhibi-
tory procedures. In general, excitatory conditioning refers to procedures that
increase the frequency or intensity of the CR. Among the more common excita-
tory procedures are (1) simultaneous conditioning, in which the onset ofthe CS
is immediately followed by the UCS, (2) delay conditioning, in which the onset
of the CS is followed by the UCS only after a specific temporal interval, (3)
trace conditioning, in which both CS onset and offset precede UCS onset, and
(4) compound conditioning, in which two CS are combined and followed by the
UCS. Inhibitory conditioning refers to a withholding or diminution of a refer-
ence response, and therefore occurs within the context of previous excitatory
conditioning. Inhibitory procedures are subdivided into internal inhibition,
which refers to an acquired or conditioned withholding of the CR, and external
inhibition, which refers to the elicitation of inborn, unconditioned reflexes that
are antagonistic to the CR. The gradual acquisition of internal inhibition occurs
(1) during the initial portion of relatively long CS-UCS delay conditioning (in-
hibition of delay) (2) when the CS is no longer paired with the UCS, (inhibition
of extinction) of (3) when a CS paired with a UCS is occasionally combined
with a second neutral stimulus, but the two-stimulus complex is never followed
by the UCS (differential or conditioned inhibition). Unconditioned inhibition
immediately occurs when a novel stimulus is (1) superimposed upon an excita-
tory conditioning procedure and the reference CR is diminished (external in-
hibition) or (2) superimposed upon an internal inhibition procedure and the ref-
erence inhibition of the CR is disrupted (inhibition of inhibition, or
disinhibition) .
Although classical conditioning is occasionally described as a simple method
of conditioning involuntary and stupid behaviors, Pavlov described a multitude
of variables that would modify the probability of the CR. Fundamental require-
ments were that the subject be alert and healthy, and that the to-be-conditioned
stimulus be presented prior to the UCS, overlap the UCS, and be neither "too
strong nor too weak" (Pavlov, 1927). The fourth point concerning the intensity
of the CS introduced a fundamental interaction that substantially complicates
the analysis of classical conditioned reflexes. Virtually all "neutral" or to-be-
conditioned stimuli are not in fact neutral but are UCSs in their own right.
The initial presentation of a novel stimulus unconditionally elicits an ori-
enting reflex toward the stimulus source (Pavlov, 1927). The unconditioned ori-
enting reflex is an integrated set of reactions (receptor orientation, somatomo-
Chapter 6: Response Patterning in Classical Conditioning 301

tor responses, and changes in respiration, electroencephalogram, galvanic skin

response, and cardiovascular responses) that are biologically adaptative and
exclude or inhibit all other reflex systems (Sokolov, 1960, 1963). With repeti-
tive stimulation, the stimulus-orienting reflex is replaced by the adaptation re-
flex. With es-ues pairings, however, the orienting reflex is replaced by con-
ditioned reflexes (Bykov, 1958) or is transformed into a conditioned orienting
reflex (Biriukov, 1958). Yet, an intense "neutral" stimulus unconditionally
elicits defense reflexes that never adapt and are antagonistic to the develop-
ment of conditioned reflexes during subsequent es-ues pairings. The inter-
play of stimulus-orienting, adaptation, and defense reflexes with conditioned
reflexes has become a major element in the analysis of classically conditioned
effects (review by Razran, 1961). "Simple" classical conditioning thus results
in a multifaceted interaction between unconditioned orienting, adaptation, and
defense reactions elicited by the es, conditioned orienting and consummatory
reactions generated by es-ues pairings, and unconditioned reactions elicited
by the ues.
Undoubtedly, a most comprehensive analysis of orienting reflexes has been
provided by Sokolov and associates (Sokolov, 1958, 1959, 1960, 1963), result-
ing in a detailed Pavlovian model of perception and attention (reviews by Lynn,
1966; Graham, 1973). Sokolov emphasized that orienting reflexes actually em-
brace two distinct reactions: a narrowly defined unconditioned orienting to
stimulus novelty, resulting in a nonspecific "tuning" of sensory analysers, and
a conditioned reflex chain of investigatory and exploratory behaviors. Simi-
larly, Biriukov (1958) described the orienting reflex as a complex uncondi-
tioned-conditioned reaction consisting of two phases: an unconditioned reflex
that rapidly dissipates with stimulus repetition, overlaid by a conditioned ori-
enting reaction to the es that is "practically unextinguishable". The latter
conditioned component is approximately analogous to the approach and con-
tact responses toward the es described by Pavlov. Orienting and investigatory
responses toward stimuli paired with reinforcers had not been described in the
Western literature, with the notable early exception of Zener (1937), until quite
recently. Patton and Rudy (1967), for example, reported an increased fre-
quency of approach toward a light paired with delivery of water to rats. The
approach or "autoshaping" responses toward stimuli paired with response-in-
dependent reinforcers have also been described by Brown and Jenkins (1968),
Williams and Williams (1969), among many others (review by Schwartz and
Gamzu, 1977).
Pavlovian conditioning is complicated still further by the multiple response
or polyeffector nature of all conditioned and unconditioned reflexes. "It is es-
sential to realize that each of these two reflexes-the alimentary reflex and the
mild defense reflex to rejectable substances-consists of two distinct compo-
nents, a motor and a secretory" (Pavlov, 1927, p. 17). In a similar manner, we
have just seen that orienting and exploratory reflexes are also described as
polyeffector reactions with both autonomic and skeletal components.
One of the most detailed treatments of the interactions between polyeffector
302 Section III: Classical Conditioning Procedures

reflexes is the analysis by Konorski (1967). In the Konorski system, classical

conditioning involves not one but two unconditioned reflexes elicited by the
DCS, described as consummatory reflexes and drive (or preparatory) reflexes.
Therefore, a stimulus paired with a DCS will generate two corresponding
sets of conditioned consummatory and conditioned drive responses. At this
point, classical conditioning is an interaction between four different reflexes
(conditioned and unconditioned consummatory responses, conditioned and
unconditioned drive responses). Next, each unconditioned reflex is opposed
by an antagonistic reflex. The opposing sets of consummatory versus anti-
consummatory reflxes and drive versus antidrive reflexes are somewhat
analogous to the "on" and "off" neural systems responding to the presence
and absence of stimulation (Hartline, 1938). For example, preparatory drive re-
flexes, such as "hunger," consist of motor activity, sensitivity to stimuli, and
arousal of the sympathetic autonomic system, whereas the antidrive or "satia-
tion" reflexes involve the opposite reactions of general motor relaxation and
parasympathetic activation. A still further complication is that the conditioning
environment also includes stimuli paired with the anticonsummatory and anti-
drive reflexes and will necessarily result in the acquisition of conditioned anti-
consummatory and conditioned antidrive reflexes. Thus, Konorski reports that
classical conditioning is a complex interweaving of eight different polyeffector
reflexes (conditioned and unconditioned consummatory reflexes versus condi-
tioned and unconditioned anticonsummatory reflexes, plus conditioned and un-
conditioned drive reflexes versus conditioned and unconditioned antidrive re-
flexes). Classical conditioning might then be only deceptively simple and may
result in a literal mosaic of distinct sets of covert and overt responses contin-
uously interacting with other conditioned and unconditioned reflex systems.
A different view of classical conditioning, however, had become common to
Western theories of psychology. A traditional belief among American psychol-
ogists, for example, holds that classical conditioning only involves covert re-
sponses, not skeletal responses (Skinner, 1937; Mowrer, 1960). The covert-
overt and autonomic-skeletal distinctions have generated a variety of two-
process theories proposing that classical conditioning generates states, ex-
pectancies, motivations, etc. (Mowrer, 1960; Solomon and Turner, 1962). In
general, state theories propose that the sensory link mediates perception, the
efferent link controls responses, and an intermediate connecting link serves in-
ternal states of emotion and motivation (Arnold, 1969). The mediational inter-
pretations frequently propose that these internal states are indexed by auto-
nomic measures and motivate or energize skeletal responses (reviews by
Rescorla and Solomon, 1967; Black, 1971). Furthermore, the effects of classi-
cally conditioned motives and expectances are determined solely by the char-
acteristics of the central state and should not be attributed to interactions be-
tween conditioned skeletal responses (Seligman et aI., 1971). The classical
conditioning of internal states remains the prevalent view among contemporary
theorists and continues to generate substantial experimental research (e.g.,
Prokasy, 1965; Black and Prokasy, 1972).
Chapter 6: Response Patterning in Classical Conditioning 303

Given the extensive and occasionally argumentative history of classical con-

ditioning, the extension of the concurrent response analyses to include condi-
tioned reflexes should be decidedly easy and simply involve some days collat-
ing previous data, followed by a straightforward analysis of the reported
results. The initial impression, however, is incorrect. The behavioral data un-
fortunately do not exist in the previous American literature, with the particular
exceptions of Wendt (1936), Zener (1937), Bindra and Palfai (1967), and Palfai
and Cornell (1968). On one hand, central state interpretations have relegated
overt behaviors to minor and trivial roles in the analysis of classically condi-
tioned motivational processes. At best, skeletal responses are only used as in-
direct measures of covert, unrecordable states. In general, skeletal responses
are viewed as a nuisance, to be literally eliminated with curare or figuratively
eliminated by deductive logic and control procedures. The mediational view
has thus tended to become a self-fulfilling argument, with the lack of empirical
data on overt behaviors in turn bolstering the proposition that classically condi-
tioned effects are uniquely due to covert states. Within the central state inter-
pretations, recording and analyzing skeletal behaviors during classical condi-
tioning would be little more than a fool's errand, as our critics delightedly
inform us.
The mediational interpretations however, are not free of criticism. Arnold
noted that "over and over again, it is said that emotion motivates; but hardly
ever does anyone come out and say unambiguously just how this is done" (1969,
p. 1041). In a critical review, Black (1971) similarly concluded that the simpli-
fied view of classical conditioning common to two-process theories may be in-
adequate and may provoke erroneous conclusions. The reluctant conclusions
of Black, Myer (1971), and other investigators (Brady et al., 1969; Hineline,
1973) are rather similar to previous critiques of Anohkin, Biriukov, and other
Russian investigators. Anohkin (1958), for example, pointedly criticized the
fractionalization of polyeffector reflexes into seemingly independent compo-
nents, with the resultant erroneous interpretation that one covert component
mediates other covert or overt components. Rather, Anohkin insisted that re-
flexes are integrated, holistic sets of overt and covert responses collectively
elicited and conditioned by the CS-UCS pairings. Moreover, Pavlov irremedi-
ably rejected emotions, motivations, and other "fantastic states" as meta-
physical faculties. Instead, "playfulness," "fear," "anger," and other "states"
were viewed as nothing more nor less than reflexes, not psychological states
intervening between stimuli and responses. Konorski's (1967) system is sim-
ilarly rich in the use of state and drive terminology such as "hunger" and' 'fear."
Konorski, however, clearly defined states and drives as polyeffector reflexes,
not disembodied mental processes ambiguously energizing locomotor activities.
The interaction between reflex systems may thus be an interaction between one
set of covert and overt responses and a second set of covert and overt re-
sponses, rather than a mediation of skeletal behaviors by covert responses.
On the other hand, the quantitative analysis of overt behaviors is also una-
vailable in the Russian classical conditioning literature, but for different rea-
304 Section III: Classical Conditioning Procedures

sons. Initially, Russian investigators tended to follow Pavlov's earlier descrip-

tion of skeletal responses as essentially important but difficult to quantify
relative to autonomic responses. In addition, Brozek (1963) previously noted
that Soviet psychology is occasionally weak in the quantitative analysis of
data. With occasional exceptions (e.g., Bykov, 1958), the Russian investigators
remain predominantly biased toward qualitative or observational analysis
rather than automated behavioral measurements. The Russian literature dating
back to Pavlov, however, is replete with brief, qualitative descriptions of the
interactions between skeletal responses in classical conditioning. Pavlov de-
scribed the pitting of one reflex against a second reflex, as in the suppression of
restless escape behaviors ("the freedom reflex") by an increase in the motor
and secretory responses of appetitive conditioned reflexes (' 'the hunger re-
flexes") (Pavlov, 1927, pp. 11-12). The pitting of one reflex against a second is
in fact a "concurrent reflex" analysis. Similarly, Pavlov (1927) described the
disruption of alimentary reflexes by orienting reflexes (pp. 44-45), the disrup-
tion of defensive reflexes by alimentary reflexes (pp. 29-30), the augmentation
of conditioned reflexes by the addition of similar conditioned reflexes (p. 79),
and the augmentation of extinguished reflexes by the addition of orienting re-
flexes (pp. 62-67). At a behavioral level, the interactions described by Pavlov
provide perhaps the earliest account of the inhibition and facilitation effects of
concurrent responses.
One exception to the suppression of one reflex by the excitation of a second
competing reflex was raised by Pavlov's account of internal inhibition as an an-
tagonism between excitatory and inhibitory processes. The inhibitory proc-
esses were localized in the center of the inhibitory CS, with an irradiation of
inhibition spreading outward across the cortex. As described by Konorski and
other investigators, the Pavlovian assumption of internal inhibition may not be
consistent with more recent physiological and behavioral evidence. Konorski
concluded that the "inhibitory CR is no more and no less inhibitory that the
excitatory food CR," and that the theoretical terms of' 'internal inhibition" and
"inhibitory CR" should be abandoned as inadequate (1967, p. 327). Instead,
the inhibition of one reflex is always a consequence of the concurrent excitation
of a second reflex system. Specifically, the inhibition of all motor acts is a result
of the excitation of antagonistic reflexes (Konorski, 1972).
Anohkin (1958, 1968) has similarly proposed an "exclusion principle" based
upon the dominant interpretation of Ukhtomsky (1954). The principle of exclu-
sion proposes that the organism cannot simultaneously combine two or three
sets ofreflexes: "Every holistic activity of the organism has a tendency to be
the only one present at a given time and to exclude all other acts" (Anohkin,
1958, p. 4). For Anohkin, complimentary inhibition is the first and foremost in-
teraction between reflexes. Special exceptions to the exclusion principle are
the summation or enhancement of one reflex by a second reflex, and the trans-
formation of a current reflex by latent or previously conditioned reflexes. A
concurrent response analysis would therefore seem to be not only applicable
but already qualitatively applied to classically conditioned reflexes (also see
Ukhtomsky, 1954; Kupalov, 1955; Russell, 1966).
Chapter 6: Response Patterning in Classical Conditioning 305

Simple Classical Conditioning (Experiment 1)*

The first group of experiments was designed to analyze the behavioral interac-
tions previously observed within standard classical conditioning procedures
(e.g., Zener, 1937), but used the more quantitative methods described by
Bykov (1958), Bindra and Palfai (1967), and Palfai and Cornell (1968). Initially,
Wendt (1936) reported that external inhibition, internal inhibition, inhibition of
extinction, and inhibition of delay were all examples of a redirection of activity
into other response systems, a "competition between reaction systems." Com-
peting responses that inhibited food-retrieval responses of monkeys were rela-
tive specific to individual subjects and included retreat to the back of the cage,
repetitive circling of the cage, climbing and hanging responses, etc. Wendt con-
cluded that all inhibitory effects were more appropriately described as the re-
ciprocal or complimentary inhibition produced by the elicitation of more domi-
nant behaviors. Excitatory classical conditioning effects were then described
by Zener (1937). A bell was sounded for 25 sec, with the response-independent
delivery of food into a pan at the 15th sec.
After conditioning, the dog may look at either the bell or the food pan during the
entire delay period. A few ofthe older, long-trained and stable dogs may look at the
bell immediately and continuously until the food drops or until a second or two be-
fore . . . More frequent, however, are transitional cases in which the bell is first fix-
ated, for varying number of seconds, and then the food pan for the remainder of the
delay period. More frequent still is oscillation between bell and food pan. Occasion-
ally there may be 6 or more successive glances from bell to food pan and back during
the interval. (Zener, 1937, pp. 390-391)

Still more pronounced orienting and approach behaviors were observed

when the restraints of the conditioning harness were removed from the sub-
jects. During extinction, Zener, like Wendt, reported a decrease in positive ap-
proach responses and a corresponding increase in restless movements and neg-
ative responses away from the food pan. Zener noted that the conditioned
orienting responses to the CS were neither identical with nor components of the
UCR elicited by the food UCS, and were therefore inconsistent with the tradi-
tional stimulus substitution or S-S contiguity conception of classical condition-
ing. Zener instead proposed a sign-urge interpretation, in which a perceptual
reorganization of the psychological field interacts with tension systems to re-
lease activity toward or away from signified goals.
More recently, Bykov (1958) measured the time spent in orienting reflexes
elicited by the CS (motor and respiratory responses) and conditioned reactions
elicited by CS-UCS pairings (motor and secretory responses) during appetitive
conditioning with dogs. The simple presentation of a 15-sec novel stimulus ini-
tially elicited prolonged orienting behaviors accompanied by a complete in-
hibition of all other responses. Stimulus orienting progressively decreased over
subsequent trials, with a resumption of ongoing activities throughout the re-

* Experiments were conducted by W. Henton, Washington, D. C., from December 1974 to

April 1975.
306 Section III: Classical Conditioning Procedures

mainder of the stimulus duration. Classical conditioning with the stimulus

paired with a food DCS again increased orienting reflexes to 12 to 15 sec, with a
virtually zero duration of conditioned consummatory reflexes throughout the
first six trials. The conditioned food reflexes then systematically increased to
nearly 14 sec/trial, with a reduction of CS orienting reflexes to approximately 1
sec over conditioning trials. Bykov further reported that differential classical
conditioning generated a complex, three-stage interaction between orienting
and conditioned reflexes within and across the positive CS paired with food and
a negative CS explicitly unpaired with food. Within all Pavlovian procedures,
Bykov found a continuous interaction between CS orienting and conditioned
consummatory reflexes, with concomitant inhibition as the "root of the strug-
gle" between the reflex systems.
The pattern of activity during defensive as well as appetitive classical condi-
tioning has been described by Bindra and Palfai (1967) and Palfai and Cornell
(1968). Briefly, perambulation and grooming decreased and freezing increased
during a stimulus previously paired with shock. On the other hand, a CS paired
with food resulted in decreased grooming and a slight increase in perambulation
with an unchanged duration of sitting responses. Bindra and Palfai (1967) sug-
gested that the basic effects of appetitive and defensive conditioning were an
increase in specific environment-orienting and environment-rejecting behav-
iors, respectively, with compensatory changes in other response categories.
The environment-orienting and -rejecting behaviors were in tum proposed to
result from a central motive state (Bindra, 1969, 1974).
The specific purpose of the present studies was to further quantify the be-
havioral interactions in Pavlovian conditioning by combining the analysis of CS
and DCS orienting behaviors described by Zener and Bykov with the analysis
of general activity reported by Bindra and associates. The time spent in individ-
ual responses was recorded when unrestrained rats were exposed to various
appetitive classical conditioning procedures. The absolute duration of each re-
sponse was measured during a 10-sec CS, either a red light or a white light, and
during the immediately preceding control period. The number of recorded re-
sponses was dependent upon the specific experimental procedure and progres-
sively increased from three to seven behavioral categories across successive
phases of the study. (A preliminary analysis of the present experiments was
reported by Henton, 1976.)
A standard Lehigh-Valley rodent test chamber was used as the conditioning
apparatus. A red jewel lamp mounted on the left side of the response panel
served as CS 1, and a white lamp mounted on the right side of the panel served
as CS 2 or as a neutral stimulus. Food cup 1 and food cup 2 were placed on the
grid floor and attached to the front wall, 10.1 cm from the left and right cage
walls, respectively. Each food cup was 3.8 x 2.5 x 1.3 cm. The two food cups
were separated by a Plexiglas divider (6.4 x 0.6 x 6.4 cm) in order to more
easily identify the separate orienting and approach responses to each food cup.
Feeder 1 and feeder 2 were balanced to deliver single 45-mg rat food pellets into
the respective food cups with an approximate latency of 0.3 sec. The back wall
of the chamber was painted flat black to reduce reflected light from the CS. The
Chapter 6: Response Patterning in Classical Conditioning 307

chamber was placed in a semidarkened room and diffusely illuminated by ap-

proximately 25 foot-candles of ambient lighting.
The time spent in selected behaviors was recorded by the method described
by Bindra and Palfai (1967) and Palfai and Cornell (1968). Each subject was ob-
served through the right Plexiglas wall of the chamber, and the duration of each
behavior was measured by completing individual electronic circuits connected
to a 0.33-sec recycling timer. The separate circuits were activated by touching a
stylus to metal contacts on a recording sheet. This stylus device was used
rather than microswitches to eliminate the mechanical latency of switch clo-
sure. Behaviors recorded in different phases of the experiment were approach
of the head within 3.8 cm of (1) stimulus 1, (2) food cup 1, (3) stimulus 2, or (4)
food cup 2, plus collateral behaviors of (5) standing up on hind legs, and (6)
sniffing the grid floor/sawdust waste tray (Nadel, 1968). In addition, the time
spent away from the stimuli and food cups was separately recorded as time
spent in the back half of the chamber. The configuration of the data-recording
contacts approximated the physical location of the lights, food cups, floor, and
ceiling of the subject's chamber. This relative distribution was an attempt to
balance the unrecorded time spent by the subject changing over from one to
another behavior by a corresponding time spent by the observer changing over
from one to another recording contact. The number of0.33-sec time units spent
in each behavior was separately accumulated on individual counters for the 10-
sec pre-CS (control) period and the 10-sec CS-DCS interval of each trial. All
trials were initiated by the observer, but with a 5-sec delay to prevent biasing or
selection of the data sample during the pre-CS period. The subjects received
five conditioning trials per 30-min session, with irregular intertrial intervals
varying between 5 sec and 20 min.

Delay Conditioning (Experiment IA)

The purpose of the first study was to record the relative durations of es ori-
enting and DeS approach responses as a replication of the Zener and Bykov
experiments. In addition, standing in the back of the chamber away from the
es and DeS was recorded in each session. The study was actually divided into
three phases: general habituation to the experimental chamber (5 sessions),
food hopper or DeS training (5 sessions), and delay classical conditioning (10
sessions). The time spent in each behavior and experimental phase is shown in
Figure 6.1. The ordinate gives the temporal probability expressed as the abso-
lute fraction of each second allocated to each behavioral class. In the general
habituation sessions, the subjects were simply placed in the chamber, and the
three behaviors were recorded during five 10-sec sample periods. Orienting to
both the unilluminated red light and the empty food tray was virtually zero for
each subject. The time spent in the back of the chamber varied between 300 and
650 millisec/sec within and across subjects.
During the DeS training sessions, each 10-sec sample period was ended with
the delivery of a single food pellet. The unsignaled delivery of the DCS in-
creased the responses to the food cup, with a range of75 to 300 millisec/sec for
308 Section III: Classical Conditioning Procedures

S1 S2

o
O.6
1 ® ©
®®
Back of
chamber

0.0 ~~
_ _ __ V'Vv
0.6
UCS
approach

0.0

0.6
CS
orienting

Sessions
Figure 6.1. Response patterns during habituation to apparatus (A), UCS training (B), and CS-
UCS pairings (C). Each column gives the data for a different subject, and each row gives the mean
time spent in the back of the chamber, in ues approach responses, or in es orienting responses.
Each data point is the mean of five trials per session.

subjects 1 and 2, and 30 to 90 millisec/sec for subject 3. The increase in ap-

proach behaviors was relatively specific to the food cup and did not generalize
to increased orienting to other environmental features. The duration of ori-
enting to the unilluminated red light, for example, remained at approximately
zero for each subject. The ues training, however, did substantially reduce the
time spent in the back of the chamber from perhaps 50% to approximately 15%
of the session time. This altered distribution is somewhat surprising, in that the
food hopper procedure is also a standard "ues only" control procedure used
in classical conditioning, with typical reports of no effects of unsignaled ues
presentation (review by Kimble, 1961). The decreased time spent in the back of
the chamber and the increased duration of food cup responses seem to be con-
ditioned effects and may facilitate the subsequent effects of es-ues pairings.
The subjects were then trained with delay classical conditioning in each of
the next 10 sessions, a total of 50 trials. The classical conditioning consisted of
the red light es presented throughout the 10-sec interval preceding ues deliv-
ery. The three behaviors were also recorded during a to-sec pre-eS or control
period immediately preceding es onset. During the es-ues interval, ori-
Chapter 6: Response Patterning in Classical Conditioning 309

enting to the es systematically increased across sessions, stabilizing at approx-

imately 200 to 300 millisec/sec over the final 25 conditioning trials. ues ap-
proach responses also ranged between 200 and 300 millisec/sec over the last 5
sessions. Relative to the previous ues only training, however, the ues re-
sponses were only slightly increased, whereas the es orienting responses were
more substantially increased for each subject. The time spent in the back of the
chamber was generally less than 100 millis eel sec throughout the classical con-
ditioning sessions. We should also point out that the total duration of the three
recorded responses sum to 450 to 500 millisec/sec, or approximately 45% to
50% of the es. The remaining portion of the es duration was allocated to
changing over from one recorded behavior to another, as well as unrecorded
collateral behaviors of sniffing the grid floor and standing up on hind legs pre-
viously described by Nadel (1968).
The primary conditioning effect was thus a sustained increase in es ori-
enting over successive trials, as reported by Zener (1937; also see Brown and
Jenkins, 1968), rather than the biphasic increase and decrease in es responses
described by Bykov (1958). For each subject, 1- to 3-sec bursts of stimulus ori-
enting alternated with equally brief bursts of preparatory ues responses; that
is, neither es nor ues responses were positively accelerated during the lO-sec
es. The regular oscillation between responses has been characterized by Zener
as the most frequent behavioral pattern during classical conditioning. How-
ever, Zener further noted that the duration of ues responses could be in-
creased, with a virtual elimination of es orienting responses, by additional pro-
cedural manipulations. Although the parameters are far from clear, the
divergent patterns of high and low rates of es orienting reported by Western
and Russian investigators might then be dependent upon additional constraints
and procedural nuances (also see Konorski, 1967).

External Inhibition (Experiment IB)

The orienting-investigatory reflexes to novel stimuli were initially described by
Pavlov as the' 'what is it" reflex. The phasic elicitation and extinction of stimu-
lus-orienting reflexes have been described in some detail by Bykov (1958). Pav-
lov and all subsequent investigators have consistently reported the transient in-
hibition of other conditioned reflexes by the superimposed elicitation of
orienting to novel stimuli (i.e., external inhibition). If complimentary interde-
pendence of reflexes is the Pavlovian mechanism of external inhibition, then we
might anticipate a reciprocal, quantitative interaction between conditioned re-
sponses and imposed orienting responses during external inhibition proce-
dures. The first purpose of the present study was to record these interactions
when a novel stimulus was superimposed upon es-ues pairings. The second
purpose was to expand the number of recorded behaviors to include the sniffing
and standing behaviors observed in Experiment IA. The study was conducted
in two phases of25 trials each, with an external inhibition procedure in Phase A
and a return to the classical conditioning baseline in Phase B.
As before, all behavior in the back of the chamber was merely recorded in
310 Section III: Classical Conditioning Procedures

S1 S2 S3

.3[
fl.
A(Q.

o L 1.1 .I d.. 10

.3[ EXT.

tLL .. ,1!1.1
INH.

L1I I. ,I ,I !I
1

o 0 ,I.I .1
w
~.3[
o
-
I-
«
0
:,
L
!

~
I .1
!. !1 tLL!.;LI .1 H .1
2

~ .3[
Z
«
0 i. n ~ 1 ,I il [L:.I.:LI 3

~ .3[ !

o !. II .. :.
[I.. i! : -:ilil:
4

J;.1IL,I;1 [L,.,.ld .. [l.!!L,I!1 5

111II1 /lllli 1111/1

~
;;;
WI
:=

BEHAVIORS
Figure 6.2. External inhibition: distribution of responses during delay conditioning and five subse-
quent sessions of external inhibition (rows) for each of three subjects (columns). Each histogram
gives the mean duration of each response during the to-sec control interval (hatched bars) and the
10-sec CS-UCS interval (solid bars).

one category. Behaviors in the front half of the chamber were further subdi-
vided into orienting to the es, approach to the ues, orienting to the novel
stimulus, sniffing the grid floor, and standing or rearing on hind legs. The exter-
nal inhibition procedure simply consisted of the white (right) light presented as
a novel stimulus during the first 5 sec of the to-sec delay conditioning proce-
dure. Figure 6.2 presents the response profile of each subject during the five
sessions of external inhibition. The last session of delay conditioning from Ex-
periment IA is also presented for comparison. Each behavior during the es and
the immediately preceding control period is represented by solid and hatched
histograms, respectively, for each subject (columns) and each session (rows) in
Figure 6.2. In the first external inhibition session, orienting to the neutral stimu-
lus occurred for approximately 100 millisec/sec during the es, with a low or
zero duration during the control periods. es orienting and preparatory ues
responses also increased relative to the control periods, but decreased relative
Chapter 6: Response Patterning in Classical Conditioning 311

to the previous delay conditioning sessions. This relative attenuation of condi-

tioned behaviors during the es is the external inhibition effect described by
Pavlov. Orienting elicited by the novel stimulus quickly habituated and de-
creased to zero by the second session (subject 2) or the third session (subjects 1
and 3), as previously reported by Bykov. This rapid habituation of orienting to
the external inhibitor was paralleled by a progressive increase in es and/or
ues responses over the first three sessions. The additional behaviors in the
back of the chamber or sniffing and standing in the front of the chamber were
reduced during the es relative to pre-eS control rates, as previously observed
during simple delay conditioning. Thus, the net effect of superimposed ori-
enting to a novel stimulus was a change in the pattern of es and ues respond-
ing, rather than a change in the patterning of background collateral responses.
The reciprocal interactions between orienting and conditioned responses would
be consistent with the reflex antagonism model of external inhibition.
A comparison of the control and es intervals further suggests that responses
elicited by the es, ues, or novel stimulus were associated with matching com-
pensatory decreases in the background collateral activities. For example, for

51 52 53

()
w
]1. H ,. :I :I t!. !I ~ I ;1 II tI, II .1 ;I ..

"- tLH
VJ .3
z
0
l-
e:(
o J t!. · . · . :I ti, ! H .1 II
; I .1 I
2

t
a:: .3

LH .1 :1 tI, · ,I;I ti J I
::J
c 3
Z o 51 . · . ,. ; I E

:t~ .
e:(

:I .I L ,I til H
w
~

II . dd
4
iI ·. ·. ;1 :

.3

o :
t n:I ;I t;
5I :. : .,I I i tI, h . . .1 i I
5

1III1 1111/ IIII1

I
I
z
....'"
0
iii

BEHAVIOR 5
Figure 6.3. Delay conditioning: distribution of collateral and conditioned responses during five
successive sessions of delay conditioning with es I-UeS I pairings. See Figure 6.2 legend.
312 Section III: Classical Conditioning Procedures

subject 1 in session five, responses to the es and the DeS increased a total of
360 milliseclsec compared to the control period, and the collective time spent in
all collateral responses decreased 380 millisec/sec (with an unchanged zero du-
ration of responses to the novel stimulus). Figure 6.3 similarly presents the re-
sponse profiles for each subject during simple delay conditioning in the five
baseline sessions of Phase B. The increased durations of conditioned es and
DeS responses were again intimately associated with complimentary de-
creases in the time spent in other behaviors. To continue the example of subject
1 and session five, the total duration of es and DeS responses increased 330
millisec/sec and the combined durations of collateral responses decreased 363
millisec/sec during the es relative to the control interval. The quantitative re-
sults simply demonstrate that background activities are inhibited by superim-
posed es and Des responses in the same reciprocal manner that the condi-
tioned responses are themselves externally inhibited by imposed orienting
responses.

Extinction (Experiment Ie)

The elimination of conditioned behaviors by the repeated presentation of the
es without the DeS is frequently viewed as a prototypical model of internal
inhibition. The gradual decrement in responses was described by Pavlov as an
interaction between inhibitory and excitatory processes within the eS-DeS
bond, rather than as an antagonism toward other conditioned reflexes. The neu-
rophysiological evidence for the internal inhibition model has recently been re-
viewed by Asratian (1969) and Molnar and Grastyan (1972). Alternatively,
other investigators have reported behavioral effects of extinction that do not
wholly support the internal inhibition model nor the usual dichotomy of internal
versus external inhibition. Konorski (1967), for example, used a repeated ac-
quisition -extinction procedure and found rapid sequential changes in the con-
ditioned responses that are inconsistent with the gradual accumulation of inter-
nal inhibition. Wendt (1936) and Zener (1937) also reported qualitative
observations of complimentary behavioral interactions during extinction, not a
simple decrement in conditioned responses. The present study was an attempt
to systematically replicate the observations of Wendt and Zener by measuring
the actual time spent in competing behaviors during classically conditioned ex-
tinction.
The subjects and delay conditioning procedures of the preceding experi-
ments were used again, with the exception that the 10-sec es was no longer
terminated by the Des food pellet. Extinction was continued for 10 sessions
(50 trials). Figure 6.4 presents the response distributions for each subject dur-
ing sessions 1, 3, 5, 7, 9, and 10. eS-orienting and Des approach responses
progressively decreased across the successive extinction sessions for all sub-
jects. In general, approach to the ues food cup appeared to extinguish margin-
ally faster than orienting to the es. By the seventh session, the response distri-
butions were relatively equivalent during the control and es intervals, with
virtually zero durations of es and ues responses. This extinction of condi-
Chapter 6: Response Patterning in Classical Conditioning 313

S1 52 53

[ .
[!. i . .LI h. il .LI
h.It iI .I ,I
•
[n ~ I II d I. 3

[!III~I .1.1 [II ;1111

;. :;. [ iI •!: . - . til II H
[i.IIII . =-
9

:td'lll filii!, L [1,11.1 . d

10

11/llllllf /1111
BEHAVIORS
Figure 6.4. Extinction: distribution of responses during extinction of CS I-UCS I conditioning.
See Figure 6.2 legend.

tioned responses was associated with a corresponding increase in the collateral

responses during the CS. In the latter sessions, the response interactions were
primarily corresponding increases and decreases in the durations of individual
collateral responses, rather than the previous interactions between conditioned
and collateral behaviors.
The quantitative interactions are especially consistent with the previous ob-
servations of complimentary interactions during extinction procedures (Wendt,
1936; Zener, 1937). Extinction procedures thus do not simply inhibit condi-
tioned responses, but instead generate complex, reciprocal interactions involv-
ing increasing as well as decreasing response probabilities. Traditionally, ex-
tinction has been argued to be only a response decrementing procedure, and
therefore would be most consistent with an internal inhibition model of Pav-
lovian extinction. In turn, the recorded behavioral decrements seemed to ob-
viously oppose external inhibition accounts, which required increments in
antagonistic behaviors. Wendt, however, insisted that a simple behavioral
314 Section III: Classical Conditioning Procedures

decrement interpretation is more of an accurate description of limited behav-

ioral analysis than an analysis of the subject's actual behavior. The present
results amplify Wendt's concern that CR inhibition should not be generalized
as an unbiased description of all behaviors during Pavlovian extinction.
A basic argument against extinction as reciprocal inhibition is the apparent
requirement for an acquisition of competing responses in the absence of any
acquisition reinforcement procedure. The present results would suggest, how-
ever, that the acquisition of additional competing behaviors is not basic to an
external inhibition account of extinction. The competing behaviors were in fact
intrinsic to the classical conditioning procedure in the form of ongoing collat-
eral responses during the intertrial interval. During acquisition, the ongoing
stream of behaviors is disrupted or suppressed by conditioned responses elic-
ited by the CS-UCS pairings (as in Experiments IA and IB). Extinction may
only lower the frequency of changing over from ongoing behaviors to pre-
viously conditioned responses at CS onset. In this system, extinction may be a
passive failure of formerly conditioned responses to inhibit ongoing intertrial
interval behaviors, rather than an active inhibition of conditioned responses by
the acquisition of still more competing behaviors.
However, as noted by Wendt and throughout this volume, response elimina-
tion procedures involve special difficulties in identifying the initial or primal
cause of the altered behavioral interactions. On one hand, the initial effect may
be a decrement in one overt response allowing increased durations of other be-
haviors. This decremental interpretation would not necessarily be embarrass-
ing to an internal inhibition model expanded to predict complimentary incre-
mental effects. Alternatively, the first effect may be the extinction of
changeover responses, with the high and unchanged rate of ongoing collateral
responses mediating the low rates of CR. This latter interpretation would be
more akin to the complimentary inhibition account proposed by Konorski
(1967, 1972). Moreover, pure instances of internal and external inhibition may
only be problematic and perhaps an illusory theoretical dichotomy. In either
theoretical case, extinction procedures clearly involve a progression of com-
plex behavioral interactions that seem to be directly amenable to a systematic
empirical analysis.

Delay Conditioning plus Inhibition

of Extinction (Experiment ID)
One derivative of extinction as only an inhibitory procedure is the proposition
that internal inhibition will spread to combine with and reduce the excitation of
other reflexive behaviors (Pavlov, 1927). Alternatively, a complimentary re-
sponse inhibition would deny a generalized suppression by a superimposed ex-
tinguished CS, and instead propose that any disruption of the reference reflex is
related to specific competing responses maintained by the extinction proce-
dure. In the present experiment, the extinquished CS 1 used in Experiment IC
was superimposed upon a delay conditioning baseline. The fundamental ques-
tion was whether the inhibition of extinction would generalize to decrement the
Chapter 6: Response Patterning in Classical Conditioning 315

time spent in all other behaviors or would differentially suppress baseline re-
sponses by eliciting antagonistic behaviors.
Initially, es 2 (the white, right light) was illuminated for 10 sec and paired
with ues 2 (delivery of one food pellet into food cup 2). Next, the previously
extinguished es 1 (the red, left light of Experiment IC) was also illuminated
throughout the 10-sec es 2-UeS 2 delay interval. Each experimental phase
lasted 10 sessions, with 5 trials per session. Recorded responses were orienting
and approach to es 1, es 2, ues 1, ues 2, standing up, or floor sniffing in
the front half of the chamber, and the collective time spent in the back of the
chamber.
Figure 6.5 presents the mean duration of each response during the acquisi-
tion ofes 2-UeS 2 delay conditioning. Orienting to es 2 and approach to ues
2 increased over the initial sessions and stabilized at approximate asymptotic
levels by the fifth acquisition session for all subjects. Relative to the intertrial
control periods, the increments in es 2-UeS 2 behaviors were associated with
compensatory decrements in the time spent sniffing, standing, or in the back of

Sl S2 S3

.3[ ;
oillill.. iI . . . . [!. !I d " II .. n 1

.3[
o onll h .. !I.I.I [u Id 1 .. ,.alil [illll. .. ,.a,d 3

.! i
"-~ 3[oiln
IU

1/1

h .. .. .1 .aI [I. I. i . .1.1 fl.! iI I .1 ,1 5

t! ·i I til .. I. .... I tI.!ai. __ J'

I-

~ .3
Q
z 0 iI .. ;1 . . . . . 1• I ,L ,I
'~" .3r
oCa .11 .I .1L. 1.11. ... J tIl III. __ ... 1.1
a
,1
9

.3r
0[11 II L.. .1
r
I
[.
,I J U I. I. .' .' ;1.1 i. L I. '. ...1.
I 10

11/lllllllllll/llllll !
BEHAVIORS
Figure 6.5. Delay conditioning with CS 2-UCS 2: distribution of responses during the acquisition
of lO·sec delay conditioning using CS 2-UCS 2 pairings. See Figure 6.2 legend.
 316 Section III: Classical Conditioning Procedures

the chamber. For each subject, the final response pattern stabilized more rap-
idly during this es 2-UeS 2 conditioning than during the original es I-UeS 1
conditioning in Experiment IA. However, es orienting was markedly less than
ues approach responses in the present study, in contrast to the regular oscilla-
tion and approximately equal durations of es and ues responses with the es
l-UeS 1 pairings in Experiment IA. This change to a predominance of ues
responses is analogous to the patterns described by Bykov (1958) and reported
for "high hunger states" by Zener (1937).
In addition, the present es 2-UeS 2 pairings also induced a transient in-
crease in the extinguished approach responses to the ues 1 food cup (sessions
1 to 3). The ues 1 responses not only were elicited by the onset of es 2, but
also increased during the intertrial control periods. The increments in ues 1
responses within sessions I to 3 were followed by a return to the near zero,
extinquished baselines by session 5 for each subject. In contrast, responses to
the unilluminated former es I remained at unchanged zero duration throughout
the es 2-UeS 2 conditioning. This transient "spontaneous recovery" of pre-
viously extinquished reflexes was noted by Pavlov when his subjects were re-
turned to the conditioning chamber or exposed to unsignaled delivery of the
former ues. The increased ues I responses may be related to the spontane-
ous recovery phenomenon, especially since the ues 2 food pellets were identi-
cal to the former ues 1 and perhaps were functionally unsignaled during the
early es 2-UeS 2 pairings.
Second, the usual definition of excitatory conditioning as only response-in-
crementing techniques also seems to be somewhat inaccurate. The disruption
of normal activities by a es previously paired with shock was described by
Bindra and Palfai (1967), with the proposal that the response disruption may be
identical to the conditioned suppression of operant responses by superimposed
classical conditioning (Brady and Hunt, 1955). This disruption or suppression

(I:) (ll)
+ .50
..c
o :
:. ~
:
_I _2 '"'I A
\ ..
Ii:
_3

:~ : l :
"'z :1 \ ~ :\

2...
.00 -- --I: - ~ --------------
..
--r------------------·--
: ...
- - :~- -

,
.: 1 ••~ : 'I I :
:1 \
... Ii. r.
\ A

•......
III

.50
:::I
VI

10 10 10
SESSIONS

Figure 6.6. Suppression of three collateral responses during appetitive Pavlovian conditioning of
three subjects. Each data point is the percent change in response duration during the CS relative to
the control period (Hunt et aI., 1952). Complete response suppression is - 1.00. Each data point is
the mean of five trials per session. Circled points indicate back of box; triangles: sniffing; solid
squares: standing up.
Chapter 6: Response Patterning in Classical Conditioning 317

of each collateral response is given in Figure 6.6, expressed as a ratio of the

response times during the control period (i.e., the "inflection" ratio of Hunt et
al., 1952; also see Chapter 1, this volume). With the apparent exceptions of ses-
sions 1 to 3, the probability of each response was clearly decreased during the
CS. The magnitude of disruption was relatively variable across sessions and
response classes. Experimenter-imposed events, such as CS-UCS pairings,
may then result in the disruption of the subject's current behavior as well as the
acquisition of additional behaviors. The apparent interactions between collat-
eral behaviors and CS 2-UCS 2 responses argue that excitatory conditioning,
like inhibitory conditioning, may be a mixed amalgam of incremental and de-
cremental effects.
The further effects of adding the previously extinguished CS 1 are described
in Figure 6.7. The behavioral changes were quite marginal and were limited to a
transient increase in CS 1 and UCS 1 responses for subject 3 and, to a lesser
extent, subject 1. The small increments in orienting and approach to CS 1 and
UCS 1 resulted in equally small decrements in both CS 2 and UCS 2 responses.
The final response patterns over the latter sessions were limited to the usual

Sl S2 S3

.3
tI P.
oLii
i j
i •.• .. .1 ! I h !. i! 1 ._ z • .1 .1 tl. I, !. 1 .1 .•. 1
t I. i. .. .. ;I tio ;II.
•3

I tI.
u
w
o •. iI ! 1i i. I. .1 d iI ,I 3

"- t II I .. J tl. !I i
til .3
,
;1 tI. t
z ;
0
~
o .• iI .. :I I d I II .1 ,I .1.1
. I L. ; I
'"
I til I.
II:
::;)
.3f
I .1,1
Q
~ I 7

z o ! III !. -. , d i • I , I .. .1 i !I
'w"
::I
.3

o ;I
tI I,I 5•.• .1 .1 t.. I I
! I ! •.. ,I J t !II. .LI
i.
i i 9

.3
tI
I. .. .. ~ I tl. .1 !I tl. I. .. .. ,I .I
II .: i
; 10

o '. ! "I ! I ;. .. !I

1IIIIIr11111111111111 I..
:I

BEHAVIORS
Figure 6.7. Superimposed inhibition of extinction: mean duration of responses during CS 2-UCS
2 delay conditioning when combined with previously extinquished CS 1. See Figure 6.2 legend.
318 Section III: Classical Conditioning Procedures

interactions between the three collateral behaviors and es 2-UeS 2 responses

and were indistinquishable from the response patterns during the previous
delay conditioning sessions for each subject. The effects of an extinquished es
upon Pavlovian baselines are apparently marginal and temporary and may be
dependent upon the spontaneous recovery offormerly extinguished responses.
The marginal effects of the inhibitory es 1, however, provide only an in-
verted description of the more robust stimulus overshadowing effects de-
scribed by Pavlov. That is, es 1 and es 2 were simultaneously presented and
terminated by the same ues food pellet, yet orienting responses were consist-
ently elicited only byeS 2, not es 1, throughout the 50 compound-stimulus
conditioning trials. Pavlov (1927) described similar procedures that resulted in
responses predominantly conditioned to one stimulus element rather than
equally conditioned to both elements of a compound es. The asymmetrical
stimulus-response bias may be dependent upon a variety of variables and has
been extensively described by Wagner and Rescorla (1972) and Kamin (1969).
Similar to most tests of conditioned inhibition, the inhibitory procedure in the
present study was a compound classical conditioning method combining two
stimuli with divergent conditioning histories. The slight inhibitory effects of es
1 could be equally well described as an overshadowing of es 1 byeS 2 during
the compound trials. The biased responding within each compound trial is con-
sistent with the overshadowing effects usually assessed by separate test trials
with each es element (Kamin, 1969; Rescorla and Wagner, 1972; Wagner and
Rescorla, 1972). Inhibition by an imposed es and overshadowing by the base-
line es might then be opposing results defining the continuum of possible inter-
actions between the different responses elicited by each es element.

Delay Conditioning: Replication (Experiment IE)

Sidman's review of scientific research methods (1960) described a variety of
experimental techniques to establish the generality and reliability of empirical
relationships. A systematic replication procedure was adopted throughout the
present experiments, for example, and the collective results suggest that be-
havioral interactions commonly occur over a wide range of classical condition-
ing situations. The reliability of experimental findings may also be assessed by
a second technique of direct replication, or duplication of previous procedures.
The purpose of the last study was to use the direct replication method to further
examine the internal consistency and reliability of the observed response pat-
terns. The 10-sec delay conditioning of Experiment IA was repeated, but with
the collateral responses as well as es and ues responses recorded during each
trial. Figure 6.8 presents the duration of each response during the five es l-
ues 1 trials of sessions 1,3,5,7,9, and 10. In session 1, es orienting and ues
approach increased up to 140 millisec/sec relative to intertrial control rates.
The es l-UeS 1 pairings also generated a transient increase in approach re-
sponses to the empty ues 2 food cup (subjects 1 and 2) or to the unilluminated
es 2 (subject 3) used in the previous study. Similar to the original conditioning
in Experiment lA, es 1 and ues 1 responses again increased to asymptotic
levels over successive conditioning sessions, with a further inhibition of collat-
Chapter 6: Response Patterning in Classical Conditioning 319

Sl S2 S3

;, .. .1 ..• 1 t.. !.!I.I!! .. !! t,!dl.I.I:lH

()
W
III
.. !I tll~I~I.LI
"'-
tlll,L.LI ....
Z
o
l-
e(
IIII,I.L ...
a:

.. I
:::I
C

z
e(
~d I.I ; tilil!..I.1
w
~

;. tLI.I..I.1 tLlil .. I.I .. ,. 9

]!,hL.LI.. tLI.L;LL ... tlli,!,.I,I ... 10

111111/11111111111111 BEHAVIORS
Figure 6.8. Reacquisition of delay conditioning: distribution of responses during the reacquisition
of delay conditioning with the previously extinguished CS 1 paired with UCS 1. See Figure 6.2
legend.

eral responses and an unchanged zero duration of CS 2 and UCS 2 responses.

For each subject, the time allocated to UCS 1 responses was consistently
greater than the time spent in CS 1 orienting responses. These results substan-
tially replicate the behavioral patterns observed in the comparable delay condi-
tioning in Experiment lA, using CS 1-UCS 1 pairings, and Experiment ID,
using CS 2-UCS 2 pairings. Moreover, many experiments in the present series
are in fact direct or systematic replications of the conditioning effects described
by previous investigators (Wendt, 1936; Zener, 1937; Palfai and Cornell, 1968).
The data within and across replications strongly suggests that classical condi-
tioning methods consistently elicit broad changes in the distribution and tem-
poral integration of mutually antagonistic response systems (Anohkin, 1958;
Konorski, 1967).

Summary
From the outset, a major finding of the concurrent responses analysis has been
the seemingly intimate relationships between the probability of any given re-
320 Section III: Classical Conditioning Procedures

sponse and the concurrent probabilities of other respondents, operants, and

collateral behaviors in the subject's repertoire (Brady and Hunt, 1955; Henton
and Brady, 1970; Henton, 1972; Henton and Iversen, 1973; Iversen, 1975). We
have noted throughout the present experiment, for example, that the absolute
duration of anyone response is closely associated with complimentary altera-
tions in the obtained durations of other responses. The conditional probability
of one response upon the simultaneous probabilities of concurrent responses
might then be used to formally describe the behavioral patterns generated by
conditioning schedules:
Time RSa, ... ,n(cs) = Time RSa, ... ,n(pre-cs) = 1000 millisec/sec. (1)

This descriptive equation represents the exhaustive case, when the total time of
all recorded responses (Rs a, ... n) equals the total available time during the CS
and the pre-CS control intervals. The absolute time spent in any given response
may be partialed out of the total response time, yielding
Ra(cs) = Ra(pre-cS) + Rs b, ... ,n(pre.CS) - Rs b, .... n(CS) (2)
or
Ra(cs) - Ra(pre-CS) = - [Rs b, ... ,n(CS) - Rs b, ... ,n(pre.CS)] (3)

Equation 2 specifies that the time spent in any response a during the CS is de-
pendent upon the base duration of that response during the pre-CS and the con-
current changes in the durations of all other responses b, . . . ,n. Equation 3 sug-
gests that any alteration in the duration of a single response a will exactly
balance the collective changes in all other responses over the two time samples.
Unfortunately, the equations must be immediately scrapped and are little
more than concise descriptions of our long-sought goal of complete behavioral
analysis. Contemporary experimental techniques most frequently record only
one response, accounting for a few hundred milliseconds per second. The pres-
ent procedures are only slightly less removed from the ideal; and recorded
seven responses summing to no more than 700 millisec/sec. Equation 1 may be
reformulated, however, to describe more realistically our current capabilities
by replacing the equal symbol with an approximation. Equation 1 would then
recognize our less than complete behavioral analysis and propose that the ac-
tually recorded durations during the CS could only approximate the actually
recorded response durations during the pre-CS. The accuracy of the approxi-
mation would be systematically increased as more and more responses are ana-
lyzed and the total durations approach 1000 millisec/sec in each time interval.
More simply, the accuracy of behavioral analyses is directly dependent upon
recording all relevant behaviors that are present, rather than conjectured re-
sponses that are nevertheless absent. Alternatively, the inaccuracy of the ap-
proximation is the discrepancy between the summed durations recorded during
the CS and the pre-CS time samples. Similarly, Equations 2 and 3 could be re-
written as approximations to propose that the recorded changes in one re-
sponse will approximately balance the simultaneous changes in other recorded
responses.
Chapter 6: Response Patterning in Classical Conditioning 321

Figure 6.9 presents a comparison of the obtained duration of each response

during the CS in Experiment IE relative to the duration expected by Equation
3. The data for each subject in Figure 6.9 correspond to the results presented in
Figure 6.8. The obtained durations during the CS closely approximated the net
changes in the concurrent probabilities of other responses for each subject. The
maximal discrepancy between calculated and obtained durations ranged from
an underestimation of 87 millisec/sec (subject 2, session 2, not shown) to an
overestimation of 100 milliseclsec (subject 3, sessions 1 and 3). In general, the
error was greater during the acquisition of new behavioral patterns in the initial
sessions than for the stabilized response patterns within the final sessions. The
mean error for each of the final five sessions ranged from - 60 to + 13, -75 to
+ 13, and -33 to +47 millisec/sec for subjects 1,2, and 3, respectively. The
quantitative interactions thus quite firmly describe a substantial interdepen-
dence in the temporal patterning of concurrent responses. Indeed, a literal ap-
plication of the response interaction analysis rather accurately describes the

Sl S2 S3

'] 0
0

• • • • •
0

•
• 0
0
II
0

f • •
• • •
0
0 0

•
•
0

0
0

f •" • • • • • •
0 0 0 0
0 0

"]
0
0
~
• "
0 0
•
f f
0

"• • • •
(;

•
0 3
• • •
0 0 0

u e ! 8 0

! •
0 • •
0 0
• •
W

]
III
"- \I • c
0
•
f• •
Z 0 0
8
•
5

• f
• • • •
0 0 0
0
I-
C>
II •
0 0 8 0 0
8 •
0
• • 0
• •
0 0

]
a: 0
•
t• t
;:)
0
•" • • • • 0

• • • •
0 7
0
• •
0
,
"
0
z ! 8 • •c 0
• •, • •
0

c " 0 0
0

]
w
2 0
•
0

• •
• t t•
c
~
i , 0
• •
0
9

• "• •"
0

• • •
0
•
0 0 0 0 0 0

• •
0 0

•
0
• •

T
0
• •
•
0
II
t • • • • •
0 0 10
• • f
<) 0
0 0

• •
0 0
II • II • 0• 0• • •
0

1IIIIrl 1111111 1111111

0 0 0

I
I
BEHAVIORS
Figure 6.9. Obtained duration of each response during the CS (filled circles) compared to expected
duration (open circles). Expected duration is the complement of the summed changes in all other
responses. Each data point is the mean of five delay conditioning trials per session.
322 Section III: Classical Conditioning Procedures

absolute duration of each of seven recorded responses within and across indi-
vidual sessions and subjects, with a maximum error of about tOO millisec/sec.
Moreover, the accuracy was neither generated by massive statistical averaging
nor numerous statistical assumptions.
The individual analysis of each session and subject was limited to a compari-
son of five to-sec pre-CS periods with five to-sec CS-UCS intervals, or a total
of two 50-sec samples. Presumably, sophisticated statistical treatments across
a large number of trials or sessions would create at least the appearance of
greater precision. Second, the analysis describes the duration of collateral and
conditioned responses with equivalent accuracy. This result suggests that the
quantitative analysis is relatively free of underlying assumptions concerning
the emitted, elicited, or adjunctive nature of individual responses; that is, the
concurrent responses analysis is behaviorally atheoretical as well as statisti-
cally nonparametric, and thus equally applicable to various combinations of
operants, respondents, and collateral behaviors. Such an equivalence princi-
ple, however, is only a more formal restatement of the empirical generality of
response interactions previously noted within various operant conditioning
procedures (Part II) and classical-operant combinations (Part I).
Although gratifyingly accurate, we should nevertheless hope that the pres-
ent analysis may be further improved by more complete data recording. As
noted above, the seven responses recorded in the present procedures ac-
counted for perhaps 700 millisec/sec, with a resultant discrepancy between
available and obtained response durations of at least 300 millisec/sec. This dis-
crepancy is in fact the so-called behavioral pause and is the magnitude of error
within a concurrent response analysis. The error in Equation 3 reduces to the
discrepancy between the amount of behavior analyzed during the pre-CS and
CS intervals. This bias may be illustrated with the two sets of data presented in
Table 6.1. For the data from session 8, the total response times during the CS
and the pre-CS are equal. Therefore, the changes in the absolute duration of
any arbitrary response would precisely equal the net change in the remaining
six responses. On the other hand, the response times during the pre-CS
summed to 55 millisec/sec more than the CS response times for the data from
session 9 (Table 6.1). As a result ofthe recording discrepancy, a change in any
one response cannot precisely balance the net change in the other recorded re-
sponses. Thus, the relationship between any given response and recorded con-
current responses would necessarily be in error by 55 millisec/sec (the "be-
havioral pause"). This recording of different amounts of behavior in two time
samples is the usual disproof of complimentary behavioral changes. Yet, the
apparent deviation from true reciprocity is a constant, dependent upon the total
amount of behavior recorded, and not a variable dependent upon the specific
characteristics of any response. The divergent response totals are a direct con-
sequence of the failure to analyze behavior during 432 millisec/sec of the pre-
CS but 487 millisec/sec of the CS.
A behavioral analysis may then be unintentionally altered by yet unanalyzed
behaviors such as grooming, immobility, etc., as well as the inherent differ-
ences within the set of required changeover reSponses. For example, the
Chapter 6: Response Patterning in Classical Conditioning 323

Table 6.1 Relative changes in response duration (millisec/sec): subject 2

Change Deviation
Net other from exact
Responses Pre-CS CS change responses reciprocity

Session 8
Back of Chamber 330 26 -304 304 0
Front of Chamber
Sniffing 140 33 -107 107 0
Standing 60 66 6 -6 0
Orient CS 1 0 178 178 -178 0
Orient UCS 1 90 317 227 -227 0
Orient CS 2 0 0 0 0 0
Orient UCS 2 0 0 0 0 0
Total 620 620

Session 9
Back of Chamber 139 20 -119 64 55
Front of Chamber
Sniffing 191 66 -125 70 55
Standing 205 59 -146 91 55
Orient CS 1 20 151 131 -186 55
Orient UCS 1 13 217 204 -259 55
Orient CS 2 0 0 0 0 55
Orient UCS 2 0 0 0 0 55
Total 568 513

change-over time from spatially separated responses (such as standing in the

back of the chamber to es orienting in the front of the chamber) is substantially
greater than the changeover time between spatially close responses (such as
orienting to the es and approach to the neighboring UeS). The unrecorded du-
rations and patterns of changeover responses therefore should affect the distri-
bution of reference responses. Thus, the absolute duration of a recorded re-
sponse is determined by alterations in any and all concurrent responses,
independent of whether the behaviors happen to be recorded by the experi-
menter.
Equating inaccuracy with unrecorded events is admittedly a begging of as-
sumptions, no matter how logically derived. The same point may be empirically
derived, however, by analyzing the successive higher order relationships be-
tween specifically recorded responses. Figure 6.10 presents the relative incre-
ments in one response during the es compared to the simultaneous changes in
one response, or the combined changes in two, three, and four concurrently
available responses. Figure 6.10A shows that the increase in ues responses
cannot be fully explained by simultaneous changes in the time spent in the back
of the chamber. For each subject, the time in the back of the chamber generally
underestimates the concurrent changes in ues responses. The seeming ab-
sence of any close relationship might then be logically generalized to suggest a
comparable lack of association with all other responses. Alternatively, the ap-
324 Section III: Classical Conditioning Procedures

X increase in duration of UC5 R's

51 52 53
o .3 o .3 o .3

".
, .. ..'. .•
Back of box

~
. '\.

"
\, ~

"" ." '\.

.~

A .3
"
., .. ."
":." .
" Back of box
" +

.. .' .," .
" sniffing r's

'\..
..
u ... '\.
'\.
!
B f. .3 " '. "
• 1.3
'\.

"
Back of box
+
" '\.. " '\. " '\.
sniffing r's
+
.... .'\. '\.
standing r's

.' "
. ."..
\

C
.'....
o...-------r---,
." Back of box

..
'\. +

. ...
sniffing r's
'\.
~ +
~ • • '\. standing r's
'\. • ~ +
o .3
" "
Figure 6.10. Changes in ues responses relative to the simultaneous changes in other responses.
." cs r's

A. Change in 1 response. B. Net change in 2 concurrent responses. e. Net change in 3 concurrent

responses. D. Net change in 4 concurrent responses. Each data point is the mean of 5 trials for each
of 10 delay conditioning sessions.

parent discrepancy may be a failure to describe other relevant responses. The

underestimation of UCS response durations is substantially altered by expand-
ing the analysis to include a second response (Figure 6.10B) and a third re-
sponse (Figure 6.10C). The collective changes in the three collateral responses
would then result in an overestimation of UCS responses for each subject.
Again, the relationship between responses might be offered as a disproof of be-
havioral interactions, or merely a failure to record actual behaviors.
A further expansion of the analysis to include CS orienting responses would
yield the close relationship between responses in Figure 6.10D. Conversely, re-
Chapter 6: Response Patterning in Classical Conditioning 325

sponses to CS 2 and UCS 2 were at virtually zero durations throughout both the
pre-CS and the CS for each subject, as shown in Figure 6.8. The absence of
interactions involving CS 2 and UCS 2 responses therefore do not alter the ob-
tained response relationships and were not included in Figure 6.10. That is, the
potential behaviors that do not occur during an experiment could be listed ad
infinitum but are quite irrelevant to the actual response patterns generated by
particular conditioning schedules. In each case, the discrepancy within arbi-
trary comparisons is systematically related to unanalyzed occurrences of actual
behavior, and thus is systematically eliminated by the additional analysis of
concurrent responses. A more complete description of classical conditioned ef-
fects might then be profitably based upon a comprehensive analysis of be-
havioral patterns, rather than an isolated analysis of individual responses ab-
stracted out of the behavioral matrix.

Concurrent Classical Conditioning

(Experiments II, III, and IV)*
Some years ago, we summarized our data in a 16-fold table comparing classical
and operant procedures superimposed upon other classical or operant condi-
tioning baselines (Henton and Iversen, 1973). In the case of classical condition-
ing, clearly similar response patterns occur when (1) classical conditioning is
superimposed upon an operant baseline, or (2) operant procedures are superim-
posed upon a classical conditioning baseline (Chapter 2, this volume). The pur-
pose of the final set of experiments was to offer a preliminary analysis of the
third and remaining combination, classical conditioning superimposed upon
classical conditioning.
The concurrent classical conditioning methods used in Experiment II are de-
scribed in Figure 6.11. Initially, all subjects were trained with es 1 and ues 1

A B c o E

C51 ..J L L L
UC 5 I

C 5 2 L L
UC52 _ _ __

time
Figure 6.11. Classical conditioning used in Phases A to E of Experiment II. A. Delay conditioning
with CS I-DCS 1. B. Trace conditioning with CS 2-DCS 2. C. Concurrent delay-trace condition-
ing. D. Delay conditioniing with CS2-DCS 2. E. Concurrent delay-delay conditioning. Top to bot-
tom traces: onset of CS 1, DCS I, CS 2, DCS 2, and time in I-sec intervals, respectively.

* Experiments were conducted by W. Henton, Washington, D. C., from May 1975 to June 1976.
326 Section III: Classical Conditioning Procedures

in a simple delay conditioning procedure (Phase A), then with es 2 and ues 2
in a trace conditioning procedure (Phase B), and finally with the delay and trace
conditioning components combined into a concurrent schedule (Phase e, con-
current delay-trace conditioning). Next, the es 2-UeS 2 trace conditioning
was altered to simple delay conditioning (Phase D), which was later superim-
posed upon the es I-UeS 1 delay conditioning (Phase E, concurrent delay-
delay conditioning). The concurrent delay-trace schedules of Phase e and the
concurrent delay -delay schedules of Phase E are thus related to traditional
compound stimulus procedures, but with the addition that each es element is
individually terminated by an independent ues.

Concurrent Delay-Trace Conditioning (Experiments IIA,B, and C)

Three experimentally naive Long-Evans rats were trained in a three-phase ex-
periment, each phase consisting of 5 trials in each of 10 sessions. The apparatus
and data recording methods described in Experiment I were used throughout
the present study. In Phase A, the left, red light (eS 1) was presented for 10 sec
and immediately terminated by one food pellet delivered into food cup 1 (UeS
O. In Phase B, the right, white light (eS 2) was illuminated for 3 sec, followed
by a 7-sec interval, and then the delivery of one food pellet into food cup 2
(UeS 2). In Phase e, the onsets of the 10-sec es 1 and the 3-sec es 2 were
simultaneous; both ues 1 and ues 2 were delivered 10 sec after es onset.
The duration of es orienting and ues approach responses for the acquisi-
tion ofeS l-UeS 1 delay conditioning is shown in Figure 6.12A, es 2-UeS 2
trace conditioning in Figure 6.12B, and concurrent delay-trace conditioning in
Figure 6.12e. The delay conditioning in Phase A generated the typical increase
in es 1 and ues 1 responses across sessions, as described in Experiment I. In
contrast, the trace conditioning in Phase B resulted in lower response durations
for all subjects. Also, the relative proportions of es orienting and ues ap-
proach responses were variable across subjects, with es 2 orienting most prob-
able for subject I, ues 2 approach responses most probable for subject 3, and
a more intermixed response pattern for subject 2. Trace conditioning therefore
seems to support lower frequencies and shorter durations of both es orienting
and ues approach responses compared to simple delay conditioning. The
lowered probability of conditioned ues consummatory responses has fre-
quently been noted in trace conditioning by Pavlov and many other workers.
Although tentative, the present results imply that es orienting is also sharply
reduced in trace conditioning relative to delay procedures.
The concurrent conditioning in Phase e elicited slightly different interac-
tions between the delay and trace conditioned responses across SUbjects. For
subject I, es I and ues 1 responses were clearly inhibited relative to the pre-
vious delay conditioning baselines of Phase A, while es 2 and ues 2 responses
were less disrupted relative to the trace conditioning baselines of Phase B. For
subjects 2 and 3, however, the concurrent procedures resulted in a more mutual
inhibition of the es I-UeS I and es 2-UeS 2 responses relative to the previ-
ous baselines. The trace conditioned responses were slightly more affected
Chapter 6: Response Patterning in Classical Conditioning 327

A B C
Delay conditioning Trace conditioning Concurrent conditioning

CS 1_UCS 1 CS 2_UCS2 CS 1-UCS1 + CS2 _UCS2

e __ e 0--0

51

.3

.3 53

o ~ ______ ______
~ ~~ ______ L -_ _ _ _ _ _- U_ _ _ _ ~~~ ______ ~

46-50 96-100 146-150

Trials
Figure 6.12. Mean duration of CS orienting and UCS approach responses. A. CS I-UCS 1 delay
conditioning. B. CS 2-UCS 2 trace conditioning. C. Concurrent delay (CS I-UCS I)-trace (CS
2-UCS 2) conditioning. Each row gives the data for a different subject, and each data point is the
mean of five trials.
 328 Section III: Classical Conditioning Procedures

than the delay conditioned responses for both of these subjects. For all sub-
jects, es orienting tended to be more disrupted than Des approach responses
within each component of the concurrent schedules. In all comparisons, re-
sponse durations were unequally distributed between es 1 and es 2 responses,
between DeS I and DeS 2 responses, and between the collective time spent in
delay conditioned es I-DeS I responses and trace conditioned es 2-DeS 2
responses. The unequal distributions of concurrent responses therefore did not
"match" the equal rates of obtained Des reinforcers and would not support a
response-reinforcer matching theory for concurrent classical conditioning
schedules.
Response profiles for delay, trace, and concurrent delay-trace schedules are
given in Figure 6.13. As in Experiment lA, the delay procedure increased es I
and DeS I responding, with a concomitant inhibition of the three collateral re-
sponses relative to control periods. In a like fashion, the trace conditioning in-
creased es 2 and DeS 2 responses with simultaneous decrements in collateral
responding. The responses elicited byeS I and DeS I were markedly atten-
uated by the concurrent conditioning in Phase e compared to the delay condi-
tioning in Phase A. Similarly, the conditioned increments in es 2 and DeS 2
responses were attenuated during the concurrent conditioning relative to the
previous trace conditioning in Phase B. The three collateral responses, how-
ever, were approximately equally suppressed during the concurrent ess as
during the previous baseline phases for each subject. Thus, the concurrent

A B C
Delay conditioning Trace conditioning Concurrent delay-trace

CS 1 _UCS 1 CS 2 _UCS 2 (CS 1-UCS 1 ) + (CS 2 -UCS 2 )

u
W

.1.1 tL l. h.l;r .I.I

III

"'-
Z
SI
o
-
I-
!
"a:
~
:

!I:!I . . ._ .1,1 t,_ il i. ,1,1 ... 1"

I '
Q !

z
"w::I! .3, I
[I. I, ~ I: II !. ,. L ; I d [i.!I i I d d II
i
I [11o.
o
---- ;
- ------- ------- :I 53

IIII1I1 IIIIII! IIIIII! BEHAVIORS

Figure 6.13. Response distribution during control and CS-UCS intervals. A. Last session of delay
conditioning. B. Last session of trace conditioning. C. Last session of concurrent delay -trace con-
ditioning. See Figure 6.2 legend.
Chapter 6: Response Patterning in Classical Conditioning 329

schedule did not result in an additive inhibition of collateral responses, but a

mutual inhibition of delay and trace conditioned responses.
Parallel interactions were also observed during the intertrial interval, with
corresponding interactions between the two sets of CS orienting and DCS ap-
proach responses. Initially, CS 1 and DCS 1 responses were most probable dur-
ing the control intervals of the delay conditioning in Phase A, but CS 2 and
DCS 2 responses were more frequent during the control intervals of the trace
conditioning in Phase B. Combining the two sets of responses resulted in lower
frequencies of each response during the control intervals of the concurrent
schedules relative to the previous baselines. Concurrent classical-classical
conditioning therefore systematically alters the characteristic response pattern
of each schedule component, with both subtle and obvious pattern changes
throughout the conditioning sessions.

Concurrent Delay-Delay Conditioning (Experiments IID,E)

The es 2-DeS 2 procedure was changed to a simple 10-sec delay conditioning
for the next 10 sessions (Phase D), which was in tum superimposed upon the
10-sec es I-DeS 1 delay conditioning baseline (Phase E). The effects of this
concurrent delay-delay conditioning are described in Figure 6.14. The last four
sessions ofthe previous concurrent delay-trace conditioning are also presented
for comparison in Figure 6.14A. The schedule change to simple delay condi-
tioning in Phase D resulted in an immediate increase in orienting to es 2 for
each subject. Responses to DeS 2 also increased, but more gradually, across
the 10 es 2-DCS 2 delay conditioning sessions. The proportion of es orienting
and Des approach responses over the final sessions closely replicated the rela-
tive response distributions during the comparable es I-DeS 1 delay condi-
tioning in Phase A, with subjects 1 and 3 spending more time in es orienting,
and subject 2 spending marginally more time in DeS approach responses.
Combining the two delay components into a concurrent schedule again gen-
erated different interaction patterns across subjects. For subjects I and 2, the
delay -delay conditioned patterns were relatively comparable to the interac-
tions with the delay-trace schedules of Phase e (comparison of Figures 6.14e
and 6.14A). Subject 1 spent relatively more time responding to the es 2-DeS 2
component, whereas subject 2 responded more to the es I-DeS 1 component.
For subject 3, however, orienting to es 2 was clearly greated during delay-
delay conditioning than delay-trace conditioning. This predominance of es 2
responses resulted in a further inhibition of concurrent es 1 and DeS 1 re-
sponses.
A more detailed analysis of the individual response profiles is presented in
Figure 6.15 for the last session of concurrent delay-delay conditioning (Figure
6.15B) compared to the last session of concurrent delay-trace conditioning
(Figure 6.l5A). For all subjects, response times were asymmetrically distrib-
uted across the two delay conditioning components, and again did not corre-
spond to the matched parameters and obtained rates of Des reinforcers. The
relative distributions of both es and DeS responses therefore do not simply
330 Section III: Classical Conditioning Procedures

B
Responses
C51 ~- ..
A UC51 .to---.a.
C52 0--'0 C
Concurrent UC52 tr--6 Concurrent
delay C51 -UC5 1 delay C5 1-UC5 1
2 2
trace C5 2 _UC5 2 delay C5 -UC5

o ,0
.3 51 " '0 ,0 \\
o
II
I' , 0 I ,
I ' ,
o, I
I 0
\ I
0 I
I ,
,
I ,
o \/ I
o
,
,
o

~
0'0,

A/

o ))_e____~+_----~------~~~--~----~
5!"
~
) 52
'c:"
Co

o
.;::
E
"c:
'0

''""
~
~
, , I
/

o ))
N
0-0-0 I

.3 53

e,
,

o
)) _~_·:._ L:~_ l. ._~_-o_'AJ_ '/....J.-1----!!..----1-_--J
146-150
_---L

196-200
1

246-250
Trials
Figure 6.14. Duration of CS orienting and UCS approach responses. A. Last 4 sessions of concur-
rent delay-trace conditioning. B. Next 10 sessions of simple delay conditioning. C. Next 10 ses-
sions of concurrent delay-delay conditioning. Each row gives data for a different subject.
Chapter 6: Response Patterning in Classical Conditioning 331

A B
Concurrent delay-trace Concurrent delay-delay
1 1 1 1
delay C5 -UC5 delay C5 -UC5
2 2
trace C5 2 _UC5 2 delay C5 -UC5

.3r
oU I L. !I. . I : I .1 ;I 51

.3r
oL _! I ~I i I d; I d t !. ~ I . :I d - I d
52

Behaviors
Figure 6.15. Response distribution during control and CS-UCS intervals. A. Last session of con-
current delay-trace conditioning. B. Last session of concurrent delay-delay conditioning. Each
row gives the data for a different subject.

reflect the relative strength of the ues. The response patterns seem to be more
adequately described as the complimentary inhibition of concurrent responses,
in the reciprocal inhibition model, rather than reinforcer inhibition, in the rein-
forcer value model.
The idiosyncratic patterns of responding across subjects emphasize that a
uniform experimenter manipulation does not necessarily result in a uniform
conditioned effect. Individual differences, although commonly reported in the
operant literature, are frequently washed out in the averaging of classical con-
ditioning data over groups of subjects. Instead, individual differences may be
as common in respondent as in operant conditioning, with the effects of a given
environmental manipUlation importantly conditional and constrained by dis-
tinct interaction patterns within the response matrix.
The obtained deviation from exact complimentary response interaction is
presented in Figure 6.16 for each subject, phase, and session of Experiment II.
Each data point represents the discrepancy between the recorded response du-
rations during the es and control intervals. The deviation from exact reciproc-
ity was relatively equal within and across the various experimental phases,
with similar deviations obtained with delay conditioning (Phases A and D),
332 Section III: Classical Conditioning Procedures

A 8 c D E
z + ISO 5I
o

'"
V)

z
oco. - 150
V)

~ v + ISO S2

1)( V> '"

...00...
~

z~
Ov>
;: E -ISO
V
Q S3
+ I so

'"'"
Q.

~ 0

'"0
'"'"
'" - ISO

46- '6- 146- 196- 246-

SO 100 150 200 250
TRIALS
Figure 6.16. Deviation of responses during the CS from true complementary interactions for each
session of each phase of Experiment II. Each data point is the difference between obtained re-
sponse durations and the exact complement of all changes in other recorded responses in milli-
seconds per second. Each data point is the mean of five trials.

trace conditioning (Phase B), and the more complex schedules of delay-trace
conditioning (Phase C) and delay-delay conditioning (Phase E). The difference
between obtained and expected response durations was less than 100 milli-
sec/sec for 146 of the 150 session means in Figure 6.16. (The exceptions are
subject 1, Phase E, session 1; subject 2, Phase D, session 1; and subject 3,
Phase D, session 1 and Phase E, session 2.) The balancing pattern of simultane-
ous response probabilities clearly suggests an interdependence rather than an
independence of concurrent responses in a variety of classical conditioning
procedures. In summary, the quantitative nature of the present results affirm
the previous competing response and complimentary inhibition interpretations
of classical conditioned effects (Zener, 1937; Anohkin, 1958; Konorski, 1967).

Concurrent Delay-Delay Conditioning:

Systematic Replication (Experiment III)
To some extent, the final behavior patterns noted in the previous study could
have been influenced by the sequential order of the delay and trace condition-
ing components across the successive experimental phases. The purpose of the
present experiment was to replicate the effects of concurrent delay-delay con-
ditioning with four additional subjects by using slightly different procedures.
Chapter 6: Response Patterning in Classical Conditioning 333

A B C
Concurrent
Delay Delay delay-delay
CS1_UCS 1 CS2 _UCS2 CS1-UCS1 CS2 -UCS2

.1 d
.. . [.!
[..! I !! I ::. ; . ! III i. I.
!. I '.! I .... •151 b.LL .. d .1. I- 53

Behaviors
Figure 6.17. Replication of response distributions. A. Simple CS l-UCS 1 delay conditioning. B.
Simple CS 2-UCS 2 delay conditioning. C. Concurrent delay-delay conditioning. Each row gives
the data for a different subject. See Figure 6.2 legend.

Each 30-min session was initially divided into two equal parts, with one block
of five trials using es l-UeS 1 pairings and a second block of five trials using
es 2-UeS 2 pairings. In each case, a simple 10-sec delay procedure was used,
with intertrial intervals ranging between 20 sec and 10 min. The sequential
order of conditioning trials was alternated within daily sessions. The two delay
conditioning components were then simultaneously scheduled in 10 additional
sessions in Phase B, with 10 trials scheduled at intervals ranging between 20 sec
and 5 min.
The initial acquisition and terminal pattern of delay conditioned responses
were qualitatively comparable to the data described in previous studies in this
section. Figure 6.17 presents a summary comparison of the response profiles
during the last session of es l-UeS 1 delay conditioning (Figure 6.17A), es
2-UeS 2 delay conditioning (Figure 6.17B), and concurrent delay-delay con-
ditioning (Figure 6.l7e). For a given subject, the response distributions were
quite similar within each of the two delay conditioning components. The es
l-UeS 1 pairings and the es 2-UeS 2 pairings elicited virtually equal dura-
tions of es orienting and ues approach responses for subject 1. In contrast,
subject 2 spent marginally more time in es orienting, and subject 3 spent mar-
ginally more time in ues approach responses in each ofthe separate delay con-
ditioning components. The single exception is that subject 4 spent relatively
334 Section III: Classical Conditioning Procedures

more time in food cup approach responses in the es I-UeS I pairings, but
more time in es orienting in the es 2-UeS 2 pairings. For all subjects, re-
sponses in the back of the chamber or sniffing and standing in the front of the
chamber were inhibited during the es relative to the control intervals.
The distribution of responses within the concurrent conditioning phase were
consistent within individual subjects but again variable cross subjects, as in the
preceding Experiment II. es I-UeS 1 responses were most common for sub-
ject 1, es 2-UeS 2 responses were most common for subject 3, and there was
a more intermixed distribution across the two conditioning components for
subjects 2 and 4. As before, the concurrent delay-delay procedure did not re-
sult in an increased suppression of the collateral responses relative to previous
baselines, but instead generated reciprocal decrements in the es and ues re-
sponses conditioned within each component. As one example, es 1 plus ues 1
responses by subject 4 decreased a total of 252 millisec/sec during the concur-
rent ess compared to the previous es l-UeS 1 delay conditioning, with an
approximately balancing increase of 235 millisec/sec in es 2 and ues 2 re-
sponses (comparison of bottom rows of Figures 6.17 A and 6.17C). Relative to
the es 2-UeS 2 delay conditioning, es 2 plus ues 2 responses decreased a
total of 228 millisec/sec during the concurrent ess, with a simultaneous in-
crease of 211 millisec/sec in es 1 and ues 1 responses (bottom row of Figure
6.17B versus 6.17e). Substantially similar patterns of mutual increments and
decrements in concurrent responses were obtained with subjects 1,2, and 3.
The interlocking distributions of responses, both within and across condition-
ing components, systematically replicate the patterns obtained with the delay-
delay conditioning in Experiment II. The complimentary inhibition of re-
sponses, relative to both control periods and previous baselines, would be yet
again consistent with the suggestion that overt responses are disrupted to the
degree that other behaviors are increased.

Concurrent Trace-Trace Conditioning (Experiment IV)

Given concurrent delay-delay and delay-trace conditioning, one last experi-
ment remains-an experiment to examine concurrent trace-trace classical
conditioning. The next study therefore examined the concurrent response pat-
terns when two separately established trace conditioning procedures were
scheduled simultaneously. The study was also designed with several additional
and secondary purposes.
The data analysis to this point has involved an interaction analysis of seven
responses manually recorded by the investigator. This procedure introduces
two attendant problems. First, such observation procedures are open to ob-
server bias, with at least the potential for selectively recording and misrecord-
ing data. The problem is not immediately solved by electronic instrumentation,
however, since data displayed on counters and recorders can also be easily se-
lected and misreported. For this reason, sciences rich in theory and inference
such as psychology especially require cross-laboratory verification of experi-
mental results. Indeed, the traditional defense against continuing prejudicial
Chapter 6: Response Patterning in Classical Conditioning 335

experimental descriptions has been the success or failure of direct and system-
atic replication by the scientific community. We can only note here that the
present observer recordings are fundamentally consistent with the previous re-
sults obtained in other laboratories (e.g., Zener, 1937; Konorski, 1967, 1972;
Patton and Rudy, 1967; Palfai and Cornell, 1968), and we can only expect that
the results will also be consistent with future experimental findings.
A second problem of observer recordings is the time required daily to con-
duct one study at the expense of other experimenter activities. This is, if you
will, a competing response analysis of the experimental behavior of the investi-
gator rather than the subject and is more immediately solved by automated data
recording. The secondary purpose of the experiment was then to instrument
the classical conditioning chamber and thereby attenuate the potential problem
of observer bias and reduce the experimental time devoted to data recording.
The conditioning chamber was rebuilt with transparent Plexiglas, with light-
sensitive photocells (Archer 276-176, 1.3 cm diameter) placed 2.5 cm immedi-
ately above each es and each DeS food cup. A directional lamp holder was
fashioned from a miniature in-line phone jack and surgically fixed to the midline
of the subject's skull. A white 12-V minilamp was attached to the screw base of
the phone jack, with the forward intensity set to activate each photocell when
the subject was within 5.0 cm and facing the corresponding es or food cup. To
minimize light reflection within the chamber, the front wall was painted flat
black, except for a 1 cm2 area in front of each photocell, and the entire back
wall was also blackened. The side walls remained transparent for observation
of the subjects with a closed-circuit television camera. To avoid masking of the
ess by the white directional light, es 1 and es 2 were changed respectively to
yellow and green jewel lamps that blinked on and off 6 times/sec. A 5 x 5 ma-
trix of photocells was placed beneath the Plexiglas floor to record downward
orientation of the head, approximating the previous manually recorded floor
sniffing responses. Standing up on hind legs was also recorded as the relative
position of a counterbalanced arm connected to the wires and lamp mounted on
the subject's skull. Standing responses were defined as the vertical position of
the subject's head 12.5 cm or more above the floor. However, the time spent in
the back half of the chamber was not separately recorded in the present study.
The experiment thus additionally served as a systematic replication of the pre-
vious experiments by using an automated apparatus and six rather than seven
recorded response categories. [Blough (1977) has recently reported a similar al-
though reverse technique for recording stimulus-orienting behaviors. In the
Blough apparatus, a single photocell is mounted on the subject and various light
sources are placed within the environment, with the coincidence of photocell
activation matched to the unique flicker rate of each stimulus by an on-line
computer.]
Three experimentally naive Long-Evans rats were initially trained with
separate 10-trial blocks ofes I-DeS 1 pairings and es 2-DeS 2 pairings within
each session. A 10-sec delay procedure was used in each block of trials during
the first five sessions (Phase A), but was then altered to trace conditioning for
10 additional baseline conditioning sessions (Phase B). Trace conditioning
336 Section III: Classical Conditioning Procedures

consisted of a 5-sec CS, followed by a 5-sec trace interval, and then the response-
independent delivery of one food pellet into the appropriate food cup. The 50
trials of Phase A and the 100 trials of Phase B with each conditioning component
thus served as an automated replication of simple delay conditioning and
simple trace conditioning. The response distributions during the last sessions
of delay conditioning are given in Figure 6.18A and those of trace conditioning
are described in Figure 6.18B for subjects 1,2, and 3.

A B
Delay Trace

I hi d d
51

At
L.. d ·-
C5 1 _U5 1

o i : I .- !! I • •

At
o L iI : . 51 d til .. . I :I . =I !
C5 2 _U5 2

t 52

til
.4
.
;I .. - .
u
CD C5 1_U5 1

. . i I 51 ·. 51
.!!!
c:
.;;0
...::lca
o 5I i • I
.4 f
"C

!I t~
c:
m

I.
~ C52 _U5 2

01 : I n ; I :I I • ; I -I .1

til
53
.4f
. I :I . . •
C5 1 _U5 1

o i. : I 51 d :! • ; I : i
•

.Behaviors
Figure 6.18. Replication of delay and trace conditioning using automated apparatus. A. Delay con-
ditioning. B. Trace conditioning. Each pair of rows shows data for different subject. See Figure 6.2
legend.
Chapter 6: Response Patterning in Classical Conditioning 337

Subjects 1 and 2 developed a consistent pattern of responses across both

delay conditioning components, with relatively more es orienting in each com-
ponent for subject 1 and relatively more ues approach responses for subject 2.
In comparison, subject 3 spent more time in es orienting responses within the
es l-UeS 1 pairings but more time in ues approach responses within the es
2-UeS 2 pairings. This comparison of response patterns across different sub-
jects is quite similar to the individual distributions of delay conditioned re-
sponses manually recorded in the previous experiments (e.g., Figure 6.17).
With the trace conditioning, ues responses were consistently more proba-
ble than es orienting in each component for all subjects. The relatively greater
probability of ues responses was independent of whether es responses or
ues responses had been more probable during the previous delay conditioning
of each subject. Also, the combined durations of conditioned es and ues re-
sponses were somewhat decreased during trace conditioning relative to the
comparable durations in the delay conditioning phase. As a consequence, floor
sniffing was somewhat increased during the trace pairings compared to the pre-
vious delay procedure; that is, floor sniffing was relatively more suppressed by
the higher durations of delay conditioned responses and relatively less sup-
pressed by the lower durations of trace conditioned responses.
(Standing responses were at relatively low durations during the intertrial in-
terval as well as during the es-ues interval of the trace conditioning for all
subjects. This low probability of standing responses during the intertrial inter-
val is not in agreement with the probabilities recorded in all previous experi-
ments. A session by session examination of the data revealed that standing ini-
tially occurred at relatively long durations in the first two or three delay
conditioning sessions, with mean durations ranging up to 289, 476, and 323 mil-
lisec/sec for subjects 1,2, and 3, respectively. These mean durations decreased
to no more than 83 millisec/sec during the latter trace conditioning sessions.
Perhaps one possibility is that the lamp holder and attached wires, the smooth
Plexiglas floor, or some other equipment modification might have reduced the
probability of standing responses over successive sessions, although this now
remains only a speculation.)
In the final Phase e, the trace conditioned es l-UeS 1 responses and the
trace conditioned es 2-UeS 2 responses were concurrently scheduled for 10
sessions. The concurrent trace-trace conditioning consisted of 10 trials per
session, with intertrial intervals ranging between 20 sec and 5 min. Figure 6.19
presents the response distributions during the control and es intervals for con-
current sessions 1, 3, 5, 7, 9, and 10 for each subject. The simultaneous presen-
tation of the two trace components produced distinctive response increments
and decrements for each subject. The concurrent procedure resulted in a mu-
tual inhibition of all conditioned responses for subject 1, with responding to
each es and each ues decreased relative to the previous individual baselines.
The response patterns were more asymmetric for subjects 2 and 3, with a differ-
ential inhibition of the behaviors within one of the two trace conditioning com-
ponents. Initially, for subject 2, the es 2-UeS 2 responses were relatively un-
changed compared to simple trace conditioning, with a concurrent suppression
of es 1 and ues 1 responses. A transition to a reversed interaction began with
338 Section III: Classical Conditioning Procedures

Sl S2 S3

.4
tI i
.1 tI ... : :1.1 al t!I
o !I !I ,. 5 I .1 I

U
III
1/1
t!. '. !. d n II ti. 3

"-
tll..ald ... 1 tl. •.. I.L,51
Z
o
I-

.4r
0(

til .. il ..
~
::::I

t· ·I1 .1.1
C

z 0[1 I.. if .1 .1 ..
:.
iI 5. 5 I
:
:1 i-
.4r
0(
III

tII I. !I ...
:::E

JII ..
9
.1 1

tIIL!I§I'ld b,L ...1 !I

10

II1111 111!1! IIIII! BEHAVIORS

Figure 6.19. Response distributions during concurrent trace-trace classical conditioning. Col-
umns give data for individual subjects and rows give data for different sessions. See Figure 6.2
legend.

session 5, with a resultant predominance of es l-UeS 1 responses and in-

hibition of the formerly dominant es 2-UeS 2responses across the final ses-
sions. A similar pattern of differential responding to es 1 and ues 1 occurred
consistently throughout all concurrent sessions for subject 3. For each subject,
ues approach responses remained relatively more frequent that the corre-
sponding CS orienting responses within each component, independent of the
overall distribution between the two components. The relative increments and
decrements in anyone response during the es-ues interval were counter-
balanced by reciprocal changes in one or more concurrent responses to within
100 millisec/sec for 104 of the 120 blocks of trials throughout the study.
The interactions recorded electronically in the present trace-trace condi-
tioning thus systematically replicate the asymmetric distributions of responses
across components in the previous delay-delay and delay-trace conditioning.
Apparently, all types of concurrent classical conditioning tend to differentially
inhibit conditioned responses in one component, conditional upon relative and
reciprocal increments in other conditioned responses maintained by the con-
current component.
Chapter 6: Response Patterning in Classical Conditioning 339

Lastly, an Esterline-Angus strip chart recorder was added to the equipment

for this experiment, and we might appropriately end this chapter with a more
molecular trial by trial analysis of the responses within concurrent classical
conditioning. Figure 6.20 presents the temporal pattern of concurrent re-
sponses within selected trials for subject 2. The records in Figure 6.20A are
from the last session with isolated es I-UeS I pairings and es 2-UeS 2 pair-
ings. In each simple trace conditioning procedure, floor sniffing intermittently
occurred throughout the pre-eS control interval, with occasional orienting to
the unilluminated es or approach to the empty U es food cup. The onset of the
trace es elicited immediate if relatively brief orientation to the appropriate es,
followed by a changeover to the corresponding ues food cup. Offset of the
5-sec es and the beginning of the 5-sec trace interval frequently elicited a
changeover to a concurrent response, such as orienting to the unilluminated es
2 in the trace interval ofthe es I-UeS I component, or approach to the empty
ues 1 food cup during the es 2-UeS 2 trace interval. Preparatory responses

A
, ,
Trace CS - US

t "--,
StimulUS
UCS2 F:r ;:..:~~;;,;~~~
CS2 ~
UCS1 ~ ------------~~
p
CS 1
Stand
Sniff ~---;.=n I I ~ i =+
B Concurrent trace- trace

~
I

1--_ _---ln.t\Jl . •
3 ~al
~ ~--1-~
~

9$' ~ 10 ~~
~
Ul ah R.- ~

4J~~;I 1 *
~

='Ti9.4 i 11 . CE
Figure 6.20. Concurrent response patterns within individual trials. A. Simple trace conditioning
with CS I-UCS I and CS 2-UCS 2 pairings. B. Concurrent trace-trace conditioning from sessions
1,3,5,7,9, and 10. Individual tracings within each strip chart are, from top to bottom, (1) onset of
the 5·sec CS plus the 5-sec trace interval; (2) UCS 2 approach responses; (3) CS 2 orienting re-
sponses; (4) UCS 1 approach responses; (5) CS 1 orienting responses; (6) standing on hind legs; and
(7) sniffing/orienting toward the floor.
340 Section III: Classical Conditioning Procedures

to the appropriate ues food cup were the most frequently elicited response
throughout the last 2 or 3 sec of the trace interval of each component.
Concurrent trace-trace conditioning generated the response patterns shown
in Figure 6.20B. The successive changes and pattern transitions described in
the previous figure for subject 2 are also apparent in the individual trials taken
from sessions 1, 3, 5, 7, 9, and 10. The concurrent ess initially elicited re-
sponses within the es 2-UeS 2 component, with a low or zero concurrent
probability of es 1 and ues 1 responses. This response pattern mainly con-
sisted of rapid alternation between brief bursts of es 2 orienting and ues 2
approach responses. Over subsequent sessions, onset of the two ess progres-
sively elicited more and more responses to the es I-UeS I component, with a
growing inhibition of the formerly dominant es 2-UeS 2 responses. The tem-
poral patterning in the latter concurrent trials was then similar to the interac-
tions with simple es I-UeS 1 trace conditioning, virtually as if the es 2-UeS
2 component had not been concurrently scheduled. In the final sessions, pre-
paratory ues I responses were most frequent during both the es and the sub-
sequent trace intervals, with occasional changeovers to es I responses but a
continuing inhibition of the concurrent es 2-UeS 2 responses. Additionally,
the collateral responses of floor sniffing and standing that occurred during the
intertrial interval were also suppressed to near zero probabilities during the es-
ues interval. Similar interactions between the six recorded responses were
obtained for subjects I and 3.

Summary
The reciprocal frequencies and patterns of responses within concurrent classi-
cal conditioning share some common characteristics with the behavioral pat-
terns in concurrent classical-operant conditioning (Part I). In each case, the set
of responses controlled by a classical conditioning component is markedly al-
tered by the additional set of responses maintained by the concurrent compo-
nent, independent of whether that alternate component is a classical or an
operant conditioning procedure. The changes in anyone response seem to be
dependent upon the actual elicitation or emission of concurrent responses and
occur in both the presence and absence of any specifically delivered reinforcer.
Indeed, the local and averaged response probabilities are frequently asymmet-
ric across components with equated rates of reinforcers in concurrent delay-
delay conditioning, concurrent delay -trace conditioning, and concurrent trace-
trace conditioning. The data therefore suggest that the obtained response
patterns do not simply reflect the value of the reinforcer or the valence of a state
isomorphic with reinforcer value. All too frequently such discrepancies are re-
solved by argument for the primacy of the assumed explanatory state and the
irrelevancy of peripheral responses in classical conditioning. An uncomfortable
relationship between explanation and observation may be differently resolved,
however, by appealing to the primacy of observed events and the irrelevancy of
as summed causality. Although starting from different perspectives, we can at
least agree with Holland's recent concerns that the relationship between as-
Chapter 6: Response Patterning in Classical Conditioning 341

sumed associative states and conditioned responses "may well be looser than
is usually assumed" and may reflect "differences in the nature of responding,
which were perhaps unrelated to the strength of association" (Holland, 1977, p.
102).
The results of concurrent schedules involving one or more classical condi-
tioning components are also consistent with the response interactions obtained
when two operant conditioning procedures are concurrently scheduled (Part
11). The combined results again indicate that the effects of concurrent sched-
ules are related to actual probabilities of concurrent responses instead of molar
states or predispositions to respond. This dissociation of actual from potential
functional relationships is most readily apparent in the analysis of local proba-
bilities and trial to trial response patterns. The resultant pattern analysis of con-
currently available responses is then at least broadly consistent with the princi-
ple that similar response interactions may, or indeed must, occur in concurrent
classical-classical schedules, classical-operant schedules, and operant-
operant schedules.
Concurrent classical conditioning is then but one of many procedures that
simultaneously control the rates and patterns of interacting sets of responses.
As in Konorski's system, the experimental evidence suggests that even "sim-
ple" classical conditioning procedures may be viewed as temporally concur-
rent schedules of experimental events, with the CS and associated responses
variously combined with a UCS and associated preparatory or consummatory
responses, and the entire lot imposed upon ongoing behaviors maintained by
still other environmental variables. Such an analysis is not far removed from
the suggestion that even "simple" operant schedules also have many of the
functional properties of concurrent schedules controlling reinforced, collateral,
and ongoing behaviors. This response pattern analysis of both classical and
operant conditioning has therefore been both the title and substance of this
volume.
Response patterns as basic characteristics of conditioning schedules unfor-
tunately run headlong into the prevalent belief that overt responses are irrele-
vant to the analysis of schedule effects. The supposedly trivial and irrelevant
nature of response interactions is especially entrenched in classical condition-
ing-so much so that many authors insist that overt conditioned responses are
of little interest and useless to the' 'proper" interpretation of Pavlovian condi-
tioning. In reply, however, Razran (1965) some years ago pointed out that
American interpretations are not necessarily consistent with the analysis of
Pavlov and subsequent Russian investigators. Razran's thesis was that Pav-
lovian principles were frequently altered and revised to suit the needs of then
dominant American theory ("passing Pavlov through Hull's wringer"), with
current theories being ideological revisions of still previous revisions of Pavlov.
The debate over skeletal responses in classical conditioning is one such case.
The initial reporting in the American literature of motor and skeletal responses
directed to the CS (Zener, 1937; Patton and Rudy, 1967) plausibly argued for an
elimination of the autonomic-skeletal distinction between classical and
operant conditioning, and, more recently, some reconsideration of whether
342 Section III: Classical Conditioning Procedures

such overt responses violate stimulus substitution theory of classical condition-

ing (e.g., Mackintosh, 1974). In contrast, Russian investigators have all along
insisted upon overt responses as one of the integral components of conditioned
reflexes, and the reporting of skeletal responses therefore did not require the
overthrow or revision of Pavlovian theory. Furthermore, many East European
and Russian investigators deny that sustained es orienting ("autoshaping") re-
sponses are conditioned reflexes, and therefore do not require any redefinition
of stimulus substitution. Konorski (1967), for example, described the mainte-
nance of eS-directed responses as the adventitious instrumental reinforcement
of orienting reflexes by the delivery of the ues reinforcer. Although such ad-
ventitious reinforcement effects were initially dismissed by American authors,
more recent studies have implicated response-reinforcer contingencies in the
maintenance of eS-directed responses (Wessells, 1974; Deich and Wasserman,
1977). Ironically then, Western psychologists have argued that es autoshaping
responses are true examples of Pavlovian conditioning principles, yet Pav-
lovian investigators argue that such responses may be examples of instrumental
or Skinnerian conditioning.
A more fundamental distinction, however, is that Pavlovian analysis is not
restricted to the examination of cognitive states or autonomic responses, and is
certainly not based on the assumption that skeletal responses are irrelevant to
the study of classical conditioning. Indeed, overt responses may be seen to be a
part of rather than a threat to Pavlovian analysis. Bykov (1958), for example,
insisted that motor responses are especially advantageous to the study of the
dynamic interactions between orienting-exploratory reactions and conditioned
consummatory reflexes, and their mutual interactions with still other activities
which serve as the background to conditioned reflexes. The interactions be-
tween es orienting responses, ues approach responses, and ongoing intertrial
interval behaviors in the present series of experiments then firmly commit us to
a similar view of overt responses as fundamental rather than trivial characteris-
tics of classical conditioning.
We therefore have good reason to believe that concurrent response patterns
may be used to some advantage in the analysis of Pavlovian conditioning. More
particularly, the response patterns verify a central principle offered by a broad
spectrum of previous researchers. A common thread shared by investigators
from Anohkin to Zener has been the analysis of response interactions, and has
been summarized with a simple clarity by Konorski: "Inhibition of the mala-
daptive motor acts occurs owing to excitation of neurons eliciting antagonistic
motor acts" (1972, p. 354). The present concurrent response analysis of classi-
cal conditioning, as our previous analyses of other schedules, is therefore nei-
ther new nor unique, but simply fits within a more extensive context of previ-
ous work.
We would surely be remiss, however, if we at all implied that a concurrent
response analysis is finished or complete at this point. The examination of re-
sponse patterns has too often been foreclosed by denials of skeletal behaviors
in classical conditioning to be yet complete. Moreover, Anohkin (1958) has set
out three general types of interactions between polyeffector reflexes: conflict or
Chapter 6: Response Patterning in Classical Conditioning 343

mutual exclusion, assimilation, and transformation of one reflex by another.

The data presently in hand are primarily concerned with the first and most com-
mon type of interaction between overt responses. The latter two interactions of
assimilation and transformation may well be involved in summation and facili-
tation of overt responses, and are especially in need oflaboratory investigation.
Furthermore, the concomitant relationships between covert and overt re-
sponses are.not yet understood, nor are the higher order relationships between
sets of covert-overt responses interacting with other polyeffector response
sets. Currently, we can only prejudicially guess whether skeletal, autonomic,
endocrine, and cardiovascular response patterns are intrinsically fixed or are
composed of interacting yet separable (Brady, 1971) or dissociable (Black,
1972) responses influenced by myriad variables and contingencies. The recent
work by Powell and associates, Obrist, and other investigators is fortunately
directed toward this much needed analysis (Obrist et aI., 1970; Howard et aI.,
1974; Powell and Joseph, 1974; Powell et aI., 1976).
Still more demanding and difficult is the eventual analysis of sequential as
well as concurrent response patterns. The large but finite number of concurrent
interactions is enormously expanded when the analysis is extended to the se-
quential permutations of responses ordered across time. Such a detailed analy-
sis of response chains and sequences has now been initiated by Ray, whose
timely report (Ray, 1977) is an elegant example of sequential analysis of overt
responses in Pavlovian conditioning. Additional examples of response se-
quences in classical conditioning are presented in Figure 6.20. Moreover, these
sequential changes do not differ notably from the sequential changeovers be-
tween responses in operant conditioning schedules (Part II) or classical-
operant combinations (Part I). Again, our argument is not that cognitive-based
theory cannot accurately specify response patterning, but only that explana-
tions of behavior should do so. We must then conclude, perhaps reluctantly,
that concurrent and sequential pattern analyses are only beginning rather than
finishing, but nevertheless offer the promisory note of a more complete under-
standing of conditioning and behavior.

References
Anohkin, P. K.: The role of the orienting-exploratory reflex in the formation of the conditioned
reflex. In, L. G. Voronin, A. N. Leontiev, A. R. Luria, E. N. Sokolov, and O. S. Vinogradova
(eds.): Orienting Reflex and Exploratory Behavior. Moscow, Academy of Pedagogical Sciences,
1958.
Anohkin, P. K.: The Biology and Neurophysiology of the Conditioned Reflex. Moscow, Medihsina,
1%8.
Arnold, M. B.: Emotion, motivation and the limbic system. Ann. N. Y. Acad. Sci., 159, 1041-1056,
1969.
Asratian, E. A.: Mechanism and localization of conditioned inhibition. Acta Bioi. Exp. 29,271-
291, 1969.
Bindra, D.: A unified interpretation of emotion and motivation. Ann. N. Y. Acad. Sci., 159, 1071-
1083, 1969.
Bindra, D.: A motivational view of learning, performance, and behavior modification. Psychol.
Rev., 81, 199-214, 1974.
344 Section III: Classical Conditioning Procedures

Bindra, D., and Palfai, T.: Nature of positive and negative incentive motivational effects on general
activity. J. Compo Physiol. Psychol., 63,288-297, 1967.
Biriukov, D. A.: On the nature of the orienting reaction. In, L. G. Voronin, A. N. Leontiev, A. R.
Luria, E. N. Sokolov, and o. S. Vinogradova (eds.): Orienting Reflex and Exploratory Behavior.
Moscow, Academy of Pedagogical Sciences, 1958.
Black, A. H.: Autonomic aversive conditioning in infrahuman subjects. In, R. F. Brush (ed.): Aver-
sive Conditioning and Learning. New York, Academic Press, 1971.
Black, A. H.: The operant conditioning of central nervous system electrical activity. In, G. H.
Bower (ed.): The Psychology of Learning and Motivation. New York, Academic Press, 1972.
Black, A. H., and Prokasy, W. F. (eds.): Classical Conditioning II. New York, Appleton-Century-
Crofts, 1972.
Blough, D. S.: Visual search in the pigeon: hunt and peck method. Science, 196, 1013-1014, 1977.
Boring, E. G.: A History of Experimental Psychology. New York, Appleton-Century-Crofts, 1957.
Brady, J. V.: Emotion revisited. J. Psychiatr. Res., 8,363-384, 1971.
Brady, J. V., and Hunt, H. F.: An experimental approach to the analysis of emotional behavior. J.
Psychol., 40,313-325, 1955.
Brady, J. V., Kelly, D. D., and Plumlee, L.: Autonomic and behavioral responses of the rhesus
monkey to emotional conditioning. Ann. N. Y. Acad. Sci., 150, 959-975, 1%9.
Brown, P. L., and Jenkins, H. M.: Auto-shaping of the pigeon's key peck. J. Exp. Anal. Behav.,
11, 1-8, 1968.
Brozek, J.: Soviet psychology. Appendix Bin M. N. Marx and W. A. Hillix: Systems and Theories
in Psychology. New York, McGraw-Hill, 1%3.
Brozek, J.: Recent developments in Soviet psychology. Annu. Rev. Psychol., 15,493-594, 1964.
Bykov, V. D.: On the dynamics of the orienting-exploratory reaction during the formation ofposi-
tive and inhibitory conditioned reflexes and their alterations. In, L. G. Voronin, A. N. Leontiev,
A. R. Luria, E. N. Sokolov, and o. S. Vinogradova (eds.): Orienting Reflex and Exploratory
Behavior. Moscow, Academy of Pedagogical Sciences, 1958.
Creed, R. S., Denny-Brown, D., Eccles, J. C., Lidell, E. G. T., and Sherrington, C. S.: Reflex
Activity of the Spinal Cord. London, Oxford University Press, 1932.
Deich, J. D., and Wasserman, E. A.: Rate and temporal pattern of key pecking under autoshaping
and omission schedules of reinforcement. J. Exp. Anal. Behav., 27,399-405, 1977.
Graham, F.: Habituation and dishabituation of responses inneverated by the autonomic nervous
system. In, H. V. S. Peeke and M. J. Hertz (eds.): Habituation, vol. 1. New York, Academic
Press, 1973.
Hartline, H. K.: The response of single optic nerve fibers of the vertebrate eye to illumination of
the retina. Am. J. Physiol., 121,400-415, 1938.
Henton, W. W.: Avoidance response rates during a pre-food stimulus in monkeys. J. Exp. Anal.
Behav., 17,269-275, 1972.
Henton, W. W.: Concurrent response patterns in simple and concurrent classical conditioning fol-
lowing exposure to microwave radiation. Paper read at Southeastern Psychological Association,
New Orleans, March 1976.
Henton, W. W., and Brady, J. V.: Operant acceleration during a pre-reward stimulus.J. Exp. Anal.
Behav., 13,205-209, 1970.
Henton, W. W., and Iversen, I. H.: Concurrent response rates during pre-event stimuli. Paper read
at the Easter Conference, Cambridge, March 1973.
Hineline, P. N.: Varied approaches to aversion: a review of aversive conditioning and learning. J.
Exp. Anal. Behav., 19,531-540, 1973.
Holland, P. C.: Conditioned stimulus as a determinant of the form of the Pavlovian conditioned
response. J. Exp. Psychol. [Anim. Behav. Processes], 3, 77-104, 1977.
Howard, J. L., Obrist, P. A., Gaebelein, C. J., and Galosy, R. A.: Multiple somatic measures and
heart rate during classical aversive conditioning in the cat. J. Compo Physiol. Psychol., 87,228-
236, 1974.
Hunt, H. F., Jernberg, P., and Brady, J. V.: The effect of electroconvulsive shock (ECS) on a
conditioned emotional response: the effect of post-ECS extinction on the reappearance of the
response. J. Compo Physiol. Psychol., 45, 589-599, 1952.
Chapter 6: Response Patterning in Classical Conditioning 345

Iversen, I. H.: Reciprocal response interactions in concurrent variable interval and discrete trial
fixed ratio schedules. Scand. J. Psychol., 16, 280-284, 1975.
Jones, W. T.: A History of Western Philosophy. New York, Harcourt, Brace, 1952.
Kamin, L. J.: Predictability, surprise, attention and conditioning. In, B. A. Campbell and R. M.
Church (eds.): Punishment and Aversive Behavior. New York, Appleton-Century-Crofts, 1969.
Kimble, G. A.: Hilgard and Marquis' Conditioning and Learning. New York, Appleton-Century-
Crofts, 1961.
Konorski, J.: Conditioned Reflexes and Neuron Organization. London, Cambridge University
Press, 1948.
Konorski, J.: Integrative Activity of the Brain. Chicago, University of Chicago Press, 1967.
Konorski, J.: Some ideas concerning physiological mechanisms of so-called internal inhibition. In,
R. A. Boakes and M. S. Halliday (eds.): Inhibition and Learning. New York, Academic Press,
1972.
Kupalov, P. S.: General results of the study of cerebral inhibition. Zh. Vyssh. Nerv. Deiat., 5,
157-172, 1955.
Lynn, R.: Attention, Arousal and the Orientation Reaction. London, Pergamon Press, 1966.
Mackintosh, N. J.: The Psychology of Animal Learning. New York, Academic Press, 1974.
Molnar, P., and Grastyan, E.: The significance of inhibition in motivation and reinforcement. In,
R. A. Boakes and M. S. Halliday (eds.): Inhibition and Learning. New York, Academic Press,
1972.
Mowrer, o. H.: Learning Theory and Behavior. New York, Wiley, 1960.
Myer, J. S.: Some effects of noncontingent aversive stimulation. In, R. F. Brush (ed.): Aversive
Conditioning and Learning. New York, Academic Press, 1971.
Nadel, L.: Dorsal and ventral hippocampal lesions and behavior. Physiol. Behav., 3, 891-900,
1968.
Obrist, P. A., Webb, R. A., Sutterer, J. R., and Holland, J. L.: The cardiac-somatic relationship:
some reformulations. Psychophysiology, 6,569-587, 1970.
Osgood, C. E.: Method and Theory in Experimental Psychology. New York, Oxford University
Press, 1953.
Palfai, T., and Cornell, J. M.: Effect of drugs on consolidation of classically conditioned fear. J.
Compo Physiol. Psychol., 66, 584-589, 1968.
Patton, R. L., and Rudy, J. W.: Orienting during classical conditoning: acquired vs. unconditioned
responding. Psychonomic Sci., 7,27-28, 1967.
Pavlov, I. P.: Scientific study of the so-called psychical processes in higher animals. Paper read at
Charing Cross Medical School, London, 1906, reprinted in W. Dennis (ed.): Readings in the His-
tory of Psychology. New York, Appleton-Century-Crofts, 1948.
Pavlov, I. P.: Conditioned Reflexes. London, Oxford University Press, 1927.
Pavlov, I. P.: The reply of a physiologist to psychologists. Psychol. Rev., 39, 91-127, 1932.
Powell, D. A., and Joseph, J. A.: Autonomic-somatic interaction and hippocampal theta activity.
J. Compo Physiol. Psychol., 87, 978-986, 1974.
Powell, D. A., Milligan, W. L., and Buchanan, S. L.: Orienting and classical conditioning in the
rabbit (Oryctolagus cuniculus): effects of septal lesions. Physiol. Behav., 17, 955-962, 1976.
Prokasy, W. F. (ed.): Classical Conditioning: A Symposium. New York, Appleton-Century-Crofts,
1965.
Ray, R. D.: Psychology experiments as interbehavioral systems-a case study from the Soviet
Union. Psychol. Record, 2,279-306, 1977.
Razran, G.: Recent Russian psychology: 1950-1956. Contemp. Psychol., 2,93-101, 1957.
Razran, G.: The observable unconscious and the inferable conscious in current Soviet psychophys-
iology: interoceptive conditioning, semantic conditioning, and the orienting reflex. Psychol.
Rev., 68,81-147,1961.
Razran, G.: Russian physiologists' psychology and American experimental psychology: a histori-
cal and a systematic collation and a look into the future. Psychol. Bull., 63, 42-64, 1965.
Rescoria, R. A., and Solomon, R.: Two-process learning theory: relationships between Pavlovian
conditioning and instrumental learning. Psychol. Rev., 74, 151-182, 1967.
Rescorla, R. A., and Wagner, A. R.: A theory of Pavlovian conditioning: variations in the effective-
346 Section III: Classical Conditioning Procedures

ness of reinforcement and nonreinforcement. In, A. H. Black and W. F. Prokasy (eds.): Classi-
cal Conditioning II. New York, Appleton-Century-Crofts, 1972.
Russell, R. W.: Biochemical substrates of behavior. In, R. W. Russell (ed.): Frontiers in Physiolog-
ical Psychology. New York, Academic Press, 1966.
Schwartz, B., and Garnzu, E.: Pavlovian control of operant behavior: an analysis of auto shaping
and of interactions between multiple schedules of reinforcement. In, W. K. Honig and J. E. R.
Staddon (eds.): Handbook of Operant Behavior. Englewood Cliffs, New Jersey: Prentice-Hall,
1977.
Sechenov, 1.: Selected Works. Moscow, State Publishing Houses, 1935.
Seligman, M. E. P., Maier, S. F., and Solomon, R. L.: Unsignalled and uncontrollable aversive
events. In, R. F. Brush (ed.): Aversive Conditioning and Learning. New York, Academic Press,
1971.
Sherrington, C. S.: The Integrative Action of the Nervous System. New Haven, Yale University
Press, 1906.
Sidman, M.: Tactics of Scientific Research: Evaluating Experimental Data in Psychology. New
York, Basic Books, 1960.
Skinner, B. F.: Two types of conditioned reflex: a reply to Konorski and Miller. J. Genet. Psychol.,
16,272-279, 1937.
Sokolov, E. N.: The orienting reflex, its structure and mechanisms. In, L. G. Voronin, A. N. Leon-
tiev, A. R. Luria, E. N. Sokolov, and O. S. Vinogradova (eds.): Orienting Reflex and Explor-
atory Behavior. Moscow, Academy of Pedagogical Sciences, 1958.
Sokolov, E. N.: Orienting Reflex and Problems of the Higher Nervous Activity in Normal and Ab-
normal Children. Moscow, Academy of Pedagogical Sciences, 1959.
Sokolov, E. N.: Neuronal models and the orienting influence. In, M. A. B. Brazier (ed.): Central
Nervous System and Behavior. New York, Josiah Macy, 1960.
Sokolov, E. N.: Perception and the Conditioned Reflex. New York, MacMillan, 1963.
Solomon, R. L., and Turner, L. H.: Discriminative classical conditioning in dogs paralyzed by
curare can later control discriminative avoidance responses in the normal state. Psychol. Rev.,
69, 202-219, 1962.
Ukhtomsky, A. A.: Complete Works, vol. 5. Leningrad, State Publishing, 1954.
Wagner, A. R., and Rescorla, R. A.: Inhibition in Pavlovian conditioning: application of a theory.
In, R. S. Boakes and M. S. Halliday (eds.): Inhibition and Learning. New York, Academic
Press, 1972.
Wendt, G. R.: An interpretation of inhibition of conditioned reflexes as competition between reac-
tion systems. Psychol. Rev., 43, 258-281, 1936.
Wessells, M. G.: The effects of reinforcement upon the prepecking behaviors of pigeons in the
auto shaping experiment. J. Exp. Anal. Behav., 21, 125-144, 1974.
Williams, D. R., and Williams, H.: Auto-maintenance in the pigeon: sustained pecking despite con-
tingent nonreinforcement. J. Exp. Anal. Behav., 12,511-520, 1969.
Zener, K.: The significance of behavior accompanying conditioned salivary secretion for theories
of the conditioned response. Am. J. Psychol., 50,384-403, 1937.
Index

A Behavioral
analysis, 319-325
Abstracta, substantialization of, 5-6 changes, molecular, 11
Acquired drives, 24 contrast, see Contrast, behavioral
Action potential, general, 9 pauses, 155, 322; see also Pauses
Adjunctive responses, 246 stream, 254
Age in negative classical-positive operant con- Behaviors, see also Reflexes; Responses
ditioning, 52 adjunctive, 252
Amnesiac effects of localized brain stimula- competing, 251
tion, 53 stabilization of, 252
d-Amphetamine, 55, 66, 72 concurrent, 72
Amygdalectomy, 52 avoidance rates and, 64-65
Amygdaloid stimulation, 53 conditioned suppression and, 50
Analysis consummatory, 47-48
behavioral, 319-325 emotional, brain aspects and, 22
experimental, 1-2 instinctive, 252
Anticipation, 67 isolated, 217
Anxiety, 21, 22 as measures of sensations, 4
conditioned, 155 molar versus molecular, 292
Assimilation, 343 omitted by single response, 134
Attack, collateral, 235 schedule-induced, 252
Attenuation of conditioned suppression, 43-44 Benzodiazepines, 55, 72
Autoshaping, 148, 154-155,301,342 Blood pressure, systolic and diastolic, 56-59
Aversiveness, 243 Brain
Avoidance, shock aspects and emotional behaviors, 22
behavior, 97-110 function, isomorphism between skull shape
procedures, discrete-trial, 151 and, 2-3
rates, concurrent behaviors and, 64-65 stimulation, amnesiac effects of localized,
responses 53
partial reinforcement of, 64 structure and mental function, 3
suppression during stimuli associated Break and run pattern, 267-268
with, 49 Butyrophenones, 55
schedule parameters in negative classical-
negative operant conditioning, 64 c
Sidman, see Sidman avoidance
Cannabis, 56
B Cardiovascular responses
conditioned suppression and, 56-59
Barbiturates, 55 in negative classical-negative operant condi-
Behavior, 2 tioning,66
local changes in, 11 in positive classical-positive operant condi-
science of, requirements of, 12 tioning,72

347
348 Index

Central nervous system, see Nervous system, summary, 319-325

central UCS, 20
CER, see Conditioned emotional response variables
Cerebratonia, 3 in negative classical-negative operant
Changeover conditioning, 61-63
delay (COD), 167 in negative classical-positive operant con-
rates and response rates, 180 ditioning, 34-45
Changeovers in positive classical-positive operant con-
discriminative control of, 174 ditioning, 68-70
from ongoing to concurrent response, 174- views on, 301-304
186 Classical-operant schedules, experimental
Chlordiazepoxide, 66 analysis of, 29-73
p-Chlorophenylalanine, 56 negative classical conditioning scheduled
Chlorpromazine, 54, 56 with
CIs, see Conditioned inhibitory stimuli negative operant conditioning, 59-66
Classical conditioning, 22-24 positive operant conditioning, 32-59
appetitive, 306 positive classical conditioning scheduled
Sidman avoidance and, 97-110 with
concurrent, 325-343 negative operant conditioning, 73
delay-delay conditioning, 329-332 positive operant conditioning, 67-72
delay-delay conditioning, replication of, COD, see Changeover delay
332-334 Competition
delay-trace conditioning, 326-329 among behaviors, 251
operant and, see Conditioning proce- indirect, 253
dures, concurrent classical and operant Compound conditioning, 300
summary, 340-343 Conditioned
trace-trace conditioning, 334-340 anxiety, 155
under curare, 153 emotional response (CER) paradigm, 34
defensive, 306 inhibitory stimuli (CIs), 253
fear and, 24 response (CR), 20n, 253, 299-300
negative, 25 consummatory, 27
scheduled with negative operant condi- inhibition, 314
tioning, 59-66 stimulus, see Stimulus, conditioned
scheduled with positive operant condi- suppression, 26, 75
tioning, 32-59, 134-138 attenuation of, 43-44
operant conditioning interactions with, 24- cardiovascular responses and, 56-59
25 concurrent behaviors and, 50
peripheral responses in, 340 operant punishment in, 48-49
positive second-order, 41-42
concurrent response rates during, 110-118 summation of, 42-43
scheduled with negative operant condi- Conditioning
tioning,73 chamber, 335
scheduled with positive operant condi- classifications
tioning, 67-72 backward,44
primary characteristic of, 20 backward in negative classical-negative
procedures, 19, 300 operant conditioning, 63
reinforcement in, 20n classical, see Classical conditioning
response patterning in, 297-346 delay, see Delay conditioning
simple, 305-325 discrimination, 5
delay conditioning, 307-309 fear, 43
delay conditioning, replication of, 318-319 hope, 43
delay conditioning plus inhibition of ex~ operant, see Operant conditioning
tinction, 314-318 relief,43
external inhibition, 309-312 review of, and competing responses, 299-
extinction, 312-314 304
Index 349

trace, see Trace conditioning Drugs

procedures, concurrent classical and in negative classical-negative operant condi-
operant, 17-159 tioning,66
empirical analysis of, 97-155 in negative classical-positive operant condi-
history of, 19-25 tioning, 54-56
review of, 17-% Duration, component, collateral responses
summary of, 146-155 and,224-227
Constancy, 164
Consummatory behaviors, 47-48 E
Continuous reinforcement (CRF), 46
Contrast, behavioral, 164 ECS, see Electroconvulsive shock
effects, 121 Ectomorphy, 3
experiments, collateral responses in, 244 Electroconvulsive shock (ECS), generalized
explanations of, 221-224 seizures by, 53-54
interactions, 221 EMG activity, 57-58
local, 240-242 Emotional
multi-response interactions and, 236-242 behaviors, brain aspects and, 22
negative, 242 effect, 165
negative, for a specific collateral response, state, 5, 21-22
230 general,28
overall, 237-240 Emotions, 26, 222
positive, 221-223 fleeting, 11
collateral responses and, 227-235 interpretation of, 21-22
local reinforcement and, 223 Endomorphy, 3
for operant response, 230, 236 Environment, control of, 252
reinforcement rate and, 243 Equivalence principle, 322
response-displacement model of, 244-245 Estes-Skinner procedure, see Classical-
Control of environment, 252 operant schedules
CR, see Conditioned response Excitation, 9
CRF, see Continuous reinforcement Excitatory conditioning, 300
CS, see Stimulus, conditioned External inhibition, 300, 309-312
Curare, classical conditioning under, 153 Extinction, 312-314
inhibition of, delay conditioning plus, 314-
318
D
F
Data revolution, 10 FE, see Frustration effect
Delay conditioning, 300, 307-309 Fear
concurrent delay conditioning and, 329-332 classical conditioned, 24
replication of, 332-334 conditioning, 43
concurrent trace conditioning and, 326-329 inhibition of, 36
Dependent variables in classical-operant con- Fechner's law of psychophysics, 4
ditioning, 30-32 Feedback loops, 9
Differential reinforcement, see Reinforcement, Feelings, 26; see also Emotions
differential Fixed ratios (FR), 46
Discrimination conditioning, 5 pause, 169-170, 173
Discriminative control of changeovers, 174 PRP,268
Disinhibition, 300 responding, collateral responses and
Dissociation procedure, double, 6 pattern dependency between, 258-261
Double dissociation procedure, 6 pattern interdependence between, 261-
DRH, see Reinforcement, differential, of high 273
rate contingency pattern reciprocity between, 254-257
Drives, acquired, 24 responding, pausing in, 290
DRL, see Reinforcement, differential, of low responding, VI responding and, 175
rate contingency response, VI reinforcers and, 173
350 Index

FR, see Fixed ratios K

Freedom reflex, 304
Freezing, 50 Key pecking and shock intensity, 135-138
CS and, 143-145
shock intensities and, 138-146 L
UCS intensity and, 142, 145
Frustration, 222, 243 Lesions, 52-53
effect, (FE), 273-276, 280 in negative classical-negative operant condi-
Functional relationships, 10--12 tioning, 66
Lever pressing and wheel running, 229-235
H Limited hold (LH) procedure, 169
Lysergic acid diethylamide (LSD), 56
Heart rate, 56-59
Hidden reinforcement, 216
M
Hippocampal
lesions, 52-53
stimulation, 53 Marihuana, 56
Hope conditioning, 43 Matching law, reinforcer, 7-8
5-HTP, see 5-Hydroxytryptamine Mental function and brain structure, 3
Hunger reflexes, 304 Meprobamate, 55
5-Hydroxytryptamine (5-HTP), 56 a-Methyl-p-tyrosine, 66
Modulation effects, mutual, 118
I Motivational
state, general, 9
ICS, see Intracranial stimulation theory of punishment, 28
Induction interactions, 221
Inferential models, 9 N
Inhibition, 9
concept of, 215 Nalorphine, 56
CR,314 Nervous system, central
external, 300, 309-312 in negative classical-negative operant condi-
of extinction, delay conditioning plus, 314- tioning, 66
318 in negative classical-positive operant condi-
of fear, 36 tioning, 52-56
of inhibition, 300 in positive classical-positive operant condi-
internal, 300, 304 tioning,72
latent, see CS preexposure
Inhibitory o
conditioning, 300
effect of concurrent reinforcement, 164 Operant
Interactions avoidance schedules as aversive UCS, 65
induction, 221 baseline, 30
reciprocal, 113, 146 conditioning
Interference hypothesis, 26 classical conditioning interactions with,
Internal inhibition, 300, 304 24-25
Interresponse times (IRT), 188, 281-288; see concurrent classical and, procedures, see
also Postreinforcement pauses Conditioning procedures, concurrent
analysis, responses in, 282 classical and operant
collateral responses and, 281-288 history, 48
terminating, 285-286, 292 negative, scheduled with negative classi-
Intracranial stimulation (ICS), 68 cal conditioning, 59-66
IRT, see Interresponse times negative, scheduled with positive classical
Isolation, 252 conditioning, 73
Isomorphism between skull shape and brain positive, concurrent response rates dur-
function, 2-3 ing, 110--118
Index 351

positive, scheduled with negative classical Phrenology, 1-10

conditioning, 134-138 behavioral, 4-7
positive, scheduled with positive classical dynamic, 4
conditioning, 67-72 physiological, 3
procedures, 161-296 static, 4
schedule, 19 structural, 2-3
variables in negative classical-negative Polyeffector reflexes, 301-302
operant conditioning, 64-65 interactions between, 342-343
variables in negative classical-positive Postreinforcement pauses (PRP), 226-227,
operant conditioning, 45-59 254-257, 258-259, 290; see also Interre-
variables in positive classical-positive sponse times
operant conditioning, 70-72 FR,268
procedures, discrete-trial, 151 Preconditioning, sensory, 41
punishment in conditioned suppression, 48- Preference, relative, 222
49 Prefood stimuli, 154
rates, alteration of, 26 Preparedness, heightened, 28
response Preshock stimuli, 154
positive contrast for, 230 Programmed and superstitious operants, 134-
rates, 5 138
responses, suppression of classically condi- PRP, see Postreinforcement pauses
tioned responses by superimposed, Psychology, different views of, 1-13
138-146 Psychophysics, Fechner's law of, 4
schedule, random ratio, see Random ratio Punishment
operant schedule hypothesis, 26
suppression, 27-28,47-48,75 motivational theory of, 28
CS duration and, 114 operant, in conditioned suppression, 48-49
Operants Pytahexyl, marihuana homologue, 56
superstitious and programmed, 134-138
suppression of, 27-28, 47-48, 75
R
Operational definitions, 5
Overshadowing effect, 41, 43
Random ratio (RR) operant schedule, 34
p Reciprocal interactions, 113, 146
Recovery, spontaneous, 44-45,316
Paradigm shift, 10, 252-253 Recovery time measure, 32
Pattern Reflexes, 251; see also Behaviors; Responses
break and run, 267-268 concurrent, 304
dependency between FR responding and consummatory, 302
collateral responses, 258-261 freedom, 304
interdependence between FR responding hunger, 304
and collateral responses, 261-273 polyeffector, see Polyeffector reflexes
reciprocity between FR responding and col- preparatory, 302
lateral responses, 254-257 Reinforcement, 165n; see also Reinforcer
Patterning, response, 108 in classical conditioning, 20n
Pause, behavioral, 155, 322 concurrent
Pauses and collateral responses with rein- concurrent responses and, 217
forcement omission, 273-281 as inhibitory, 215
Pausing in FR responding, 290 context, 7
Pavlovian conditioning, see Classical condi- continuous, see Continuous reinforcement
tioning defined, 19n
Performances, concurrent, 165n delayed, 245
analysis of, 217 differential
reinforcement model and, 214 of high rate (DRH) contingency, 195, 199
Phenothiazines, 54-55 oflow rate (DRL) contingency, 46-47,
Phrene, 2 174,254
352 Index

distribution, response distribution and, 167 of concurrent delay-delay conditioning,

effect, partial, 35 332-334
inhibitory effect of concurrent, 164 of delay conditioning, 318-319
interaction, 164 Reserpine, 54, 55
demonstration of, 203 Response
direct or indirect, 167-168 accuracy, 71
response independence and, 164-167,203 measures, 46-47
response interaction and, 186-188 additivity
response versus, 163-217 model,244
response versus, issue of, 163-168 theory, 222
local, positive contrast and, 223 analysis, concurrent, 152, 342-343
model, 164, 173, 176, 215-216 changes, molecular, 176-177
concurrent performances and, 214 competition, 201
response model versus, 182 conditioned, see Conditioned response
omission, pauses and collateral responses -contingent UCS, 23
with, 273-281 defined, 19n
operant, positive contrast and, 236 -displacement model of positive contrast,
of other behavior contingency, variable-time 244-245
schedule with a differential (VT DRO) , distribution, reinforcement distribution and,
182-184 167
partial, of avoidance responses, 64 independence, 164
positive, time-out from, as aversive UCS, reinforcement interaction and, 164-167,
50-51 203
procedure, signaled, 201 response interaction versus, 178
rate, 7, 10 -independent
increased, as positive UCS, 72 reinforcement, superstitious, responding
positive contrast and, 243 with, 203-213
response rate and, 163 interactions, 164, 216
relative rate of, 163 reinforcement interactions and,
response-independent, superstitious re- 186-188
sponding with, 203-213 response independence versus, 178
schedules interdependencies, 229
concurrent, 164 latency, signal, 200
in negative classical-positive operant con- model versus reinforcement model, 182
ditioning, 46-48 multi-, interactions and local behavioral
signaled, responses with, 186-203 contrast, 236-242
spontaneous or hidden, 216 pattern reciprocity in concurrent FR VI
value, 7-10 schedules, 168-173
Reinforcer, see also Reinforcement patterning, 108
defined, 19n-20n in classical conditioning, 297-346
delivery, 20 patterns, interacting, 289-293
magnitude in negative classical-positive rate measures, 46, 70
operant conditioning, 45-46 rates
matching law, 7-8 as basic data, 8
parameters in negative classical-positive changeover rates and, 180
operant conditioning, 45-46 concurrent, during positive classical-
rates, response rates and, 129 positive operant conditioning, 110-118
relation, stimulus- and response-, 222 concurrent, during preevent stimuli, 118-
removal of, 176 134
types of, in negative classical-positive operant, 5
operant conditioning, 45 reinforcement rates and, 163
VI,173 reinforcer rates and, 129
Relief conditioning, 43 versus reinforcement interaction, 163-217
Replication issue of, 163-168
Index 353

-reinforcer with signaled reinforcement, 186-203

relation, 222 single, behaviors omitted by, 134
value matching law, 7-8 skeletal, 302-304, 342
sequences, 343 suppression of classically conditioned, by
-shock interval, 64 superimposed operant responses, 138-
systems, interacting, 25-26 146
unconditioned, see Unconditioned response Retention of suppression, 44-45
Responses, see also Behaviors; Reflexes RR, see Random ratio
adjunctive or schedule-induced, 246
autoshaped, see Autoshaping S
avoidance, see Avoidance responses
changeover, 149-150 Schedule
collateral, 247 effects, 5
component duration and, 224-227 -induced responses, 246
in contrast experiments, 244 parameters in positive classical-positive
interresponse times and, 281-288 operant conditioning, 70--72
in multiple schedules, 228-229, 245-246 Schedules, 19
negative contrast for, 230 classical-operant, see Classical-operant
pattern dependency between FR respond- schedules
ing and, 258-261 concurrent, 20, 118-119
pattern interdependence between FR re- reinforcement, 164
sponding and, 261-273 response versus reinforcement inter-
pattern reciprocity between FR respond- action, 163-217
ing and, 254-257 FR VI, response pattern reciprocity in con-
pauses and, with reinforcement omission, current, 168-173
273-281 multiple, 118-119, 221
positive contrast and, 227-235 collateral responses in, 228-229, 245-246
reinforced responses interacting with, concurrent responses with, 221-247
251-254 operant conditioning, 19
with simple schedules, 251-293 simple, collateral responses with, 251-293
competing, 26, 133 single-response, 164-165
review of conditioning and, 299-304 variable-time, with differential reinforce-
concurrent, 165n, 254 ment of other behavior contingency
changeover from ongoing response to, (VT DRO), 182-184
174-186 Scopolamine, 55, 66, 72
concurrent reinforcements and, 217 Seizures, generalized, by electroconvulsive
with multiple schedules, 221-247 shock, 53-54
conditioned, see Conditioned response Sensations, behaviors as measures of, 4
covert and overt, 343 Sensory preconditioning, 41
drive-conditioned, 27 Septal lesions, 52
FR,173 Sequences, response, 343
incompatible, 7 Serotonin, 56
in IRT analysis, 282 Shock
ongoing, l65n. 213 avoidance, see Avoidance, shock
changeover from, to concurrent response, electroconvulsive, see Electroconvulsive
174-186 shock
enhancement of, 186 intensity
operant, see Operant response of avoidable, 64
overt, 341-343 freezing and, 138-146
peripheral, in classical conditioning, 340 key pecking and, 135-138
prepotent, 245 Sidman avoidance, appetitive classically con-
reinforced, interacting with collateral re- ditioned responses and, 97-110
sponses, 251-254 Signal-detection theory, 4-5
sequential alternation between, 174 Simultaneous conditioning, 300
354 Index

Skeletal responses, 302-304, 342 random presentation of, in negative classi-

Skinnerian conditioning, see Operant condi- cal-positive operant conditioning, 36-37
tioning Subject variables
Skull shape, isomorphism between brain func- in negative classical-negative operant condi-
tion and, 2-3 tioning,66
Somatonia, 3 in negative classical-positive operant condi-
Somatotypes, 3 tioning, 52-59
Spontaneous in positive classical-positive operant condi-
recovery, 316 tioning,72
reinforcement, 216 Subjects in classical-operant conditioning, 30
Stabilization of competing behaviors, 252 Substantialization of abstracta, 5-6
Stimulation, intracranial, see Intracranial stim- Summation of conditioned suppression, 42-43
ulation Superstitious
Stimuli and programmed operants, 134-138
conditioned, see Stimulus, conditioned responding with response-independent rein-
discrimination of, 5 forcement, 203-213
pairs of, 23 Suppression
preevent, concurrent response rates during, of classically conditioned responses by su-
118-134 perimposed operant responses, 138-146
prefood and preshock, 154 conditioned, see Conditioned suppression
reinforcer relations to, 222 operant, see Operant suppression
unconditioned, see Unconditioned stimulus ratio, 30-32
Stimulus, conditioned (CS), 19,299-300 retention of, 44-45
compound during stimuli associated with shock avoid-
in negative classical-negative operant ance responses, 49
conditioning, 63
in negative classical-positive operant con- T
ditioning, 42-44
defined, 20n TEA, see Tetraethylammonium chloride
duration Temporal correlations, 6
in negative classical-positive operant con- Tetraethylammonium chloride (TEA), 54
ditioning, 37-38 Thalamic lesions, 53
operant suppression and, 114 Thioxanthenes, 55
in positive classical-positive operant con- Time-out (TO)
ditioning, 69-70 component, 119-122
relative, 62-63 from positive reinforcement as aversive
freezing and, 143-145 USC, 50-51
generalization Times, interresponse, see Interresponse times
in negative classical-negative operant TO, see Time-out
conditioning, 63 Trace conditioning, 300
in negative classical-positive operant con- concurrent delay conditioning and, 326-329
ditioning, 39-40 concurrent trace conditioning and, 334-340
inhibitory, 304 "Transfer of training" design, 61, 73
intensity in negative classical-positive Transformation, 343
operant conditioning, 38-39 Turning motions, collateral, 235
parameters
in negative classical-negative operant u
conditioning, 62-63
in negative classical-positive operant UCR, see Unconditioned response
conditioning, 37-45 UCS, see Unconditioned stimulus
in positive classical-positive operant con- Unconditioned response (UCR), 19, 300
ditioning, 69-70 Unconditioned stimulus (UCS), 19, 300-303
preexposure in negative classical-positive aversive
operant conditioning, 40-41 operant avoidance schedules as, 65
Index 355

time-out from positive reinforcement as, response-contingent, 23

50-51 training, 307-309
classical conditioning, 20 types of
defined, 20n in negative classical-negative operant
drugs as, 56 conditioning, 62
duration in positive classical-positive operant con-
in negative classical-negative operant ditioning, 68-69
conditioning, 61
in negative classical-positive operant con-
ditioning, 35
v
in positive classical-positive operant con-
Value matching law, response-reinforcer, 7-8
ditioning, 68
Variable
intensity
interval (VI)
freezing and, 142, 145
reinforcers, FR response and, 173
in negative classical-negative operant
responding, FR responding and, 175
conditioning, 61
schedules, 46
in negative classical-positive operant con-
ratios (VR), 46
ditioning, 34-35
-time schedule with differential reinforce-
parameters
ment of other behavior contingency
in negative classical-negative operant
(VTDRO), 182-184
conditioning, 61-62
VI, see Variable interval
in negative classical-positive operant con-
Visceral responses, 22
ditioning, 34-37
Visceratonia, 3
in positive classical-positive operant con-
VR, see Variable ratios
ditioning, 68-69
VTDRO (variable-time schedule with differential
positive, increased reinforcement rate as, 72
reinforcement of other behavior contin-
probability
gency), 182-184
in negative classical-negative operant
conditioning, 61-62
in negative classical-positive operant con- W
ditioning, 35-36
random presentation of, in negative classi- Weber fraction, 4
cal-positive operant conditioning, 36-37 Wheel running and lever pressing, 229-235