"The lnotitute of Mind and Behavior, Inc.

The Journal of Mind and Behavior

Spring, 1984, Volume S, Number 2

171

Pages 171-210
ISSN 0271-0137

A Biofunctional Model of Distributed Mental Content,

Mental Structures, Awareness, and Attention
Wholetheme Graphic Abstract

Asghar lran-Nejad and Andrew Ortony

Unit1ersir:y of Illinois at Urbana-Champaign

Central to current cognitive theories is the belief that knowledge is an organized

collection of long-term structures upon which various information processing mecha
nisms operate. Consequently much research has been dc\l'Oted to investigating the
organizational and processing aspects of knowledge representations. This paper pro
poses a shift in the locus of theoretical analysis. Following Bartlett, we argue that mental
functioning may be more readily characterized if the idea of abstract long-term associations and structures is abandoned. An account of cognition is proposed in which mental
relations are transient functional relations, and in which psychological permanence is a
functional characteristic of the neuronal system. Cognition and other aspects of mental
life are explained in terms of the activity of anatomically distributed constellations of
neuronal elements. These elements are conceived of as physiological microsystems
which are capable of generating specialized awareness experiences. The overall mental
counterpart of the combined activity of these elements we call the schema-of-themoment. We hope that the model we are proposing can contribute tO bridging the gap
between cognitive psychology and the neurosciences.

When psychological theories employ theoretical terms like memory, representation, and structure, they often do so because the descriptions and explanations of psychologically interesting phenomena that result are at a sufficiently
abstract level to be informative and intelligible. As a first step in theory
construction the use of theoretical terms at a level of description close to the
phenomenological level is helpful, and probably indispensable. However, the
theories that result are often rather vague and ill-constrained, and tend to lack
predictive power (see, for example, reviews of schema theory by Alba and
Hasher, 1983, and Thorndyke and Yekovitch, 1980). If this is true, then a
sensible next step would be to try to account for the phenomena of interest in
terms of more concrete constructs. In this paper we offer some proposals for
taking this step. To do this, we maintain that it is necessar~ to reconsider the
traditional notion of knowledge representations as long-term (essentially)
static structures. We shall argue that the nature of mental content, schemata,
The research reported herein was supported in part by the National Institute of Education under
Contract No. HEW-NIE-C-400-76-0116, and in part by a Spencer Fellowship awarded to the
second author by the National Academy of Education. Request for reprints should be sent to
Asghar lran-Nejad, Ph.D., Center for the Study of Reading, University of Illinois at Urbana
Champaign, 51 Gerty Drive, Champaign, lllinois 6 1820.

.17:t.
.' r:

IRAN-NEJAD AND OR TONY

awareness, and attention may be more readily specifiable in terms of concrete

neurophysiologically-realistic constructs if one abandons the notion oflongterm static knowledge representations in favor of transient dynamic patterns.
An interesting aspect of attempting to employ more concrete theoretical
constructs in psychological explanations is that it sharpens the distinction
between artificial intelligence and cognitive psychology. The two cease to be
simply methodological variants of one another. Artificial intelligence is concerned with characterizing cognition and intelligence in abstracto; its goal is a
"system-independent" specification of the cognitive software. Theories in
cognitive psychology must be more constrained. They need to take into
account not only the constraints imposed by people's behavior, but also the
sort of constraints likely to be imposed by the biological hardware, since it is
presumably these that give cognition its unique! y human quality. Thus, while
it might seem reasonable to start by assuming that cognition can be explained
solely in terms of the formal characteristics of psychological software, it may
well be that this assumption cannot be upheld. Certainly, arguments have
been presented to this effect. For instance, Eliashberg (1981) examined the
properties of hypothetical machines and argued that "the popular thesis that
the problem of the algorithms performed by the brain . . . has but little to do
with the problem of brain hardware" is inadequate. Similarly, Kugler, Kelso,
and Turvey ( 1980) claimed that "abstract automata formally equivalent to the
turing machine do not satisfy the natural constraints that must be met by any
actual, evolved epistemic agent .... The cost variables imposed on organisms by the laws of physics and biology are quite different from those formally
placed on the workings of abstract automata" (p. 5). We share these views,
believing that more attention needs to be devoted to the functional characteristics of the physical systems that exhibit the phenomena of interest. In fact,
we believe that the ultimate goal of cognitive psychology ought to be the
specification of the way in which the functional properties of the nervous
system make cognition possible.
There are other reasons for attempting to base an account of cognition on
relatively concrete constructs. One is that neuroscientific models in general
tend to be parsimonious. A striking example is provided by the recent
advances (e.g., Berlin and Kay, 1969; Kay and McDaniel, 1978) in understanding the relationship between the perception of color and the meaning of color
terms in different languages. It now seems that "all the basic color categories
of the languages of the world are based on . . . six fundamental neural
response categories, whose structures are determined by the firing patterns of
... cells in the visual pathway'' (Kay, 1981, p. 64 ). Only after taking account
of the physiology of color perception did it become possible to give a unified
explanation of the principles governing the way in which people in different
cultures speaking different languages talk about the world of color.
For us, the most compelling reason for using relatively concrete constructs

A BIOFUNCTIONAL MODEL

173

in psychological theories is that their use a~oids some oft~: problems related
to the metaphorical nature of the theorencal terms tradmonally employed.
Terms like memory and knowledgMepresentation are complex abstractions, and
it is not at all clear to what they refer. Many psychologists (e.g., Bartlett, 1932;
Bransford, McCarrell, Franks, and Nitsch, 1977; Jenkins, 1977; Pylyshyn,
1973) have objected to the heavy theoretical burden imposed upon such
terms. The problem is that often the essentially metaphorical use of these
concepts can give rise to misleading implications. For example, we ordinarily
talk about mental representations being stored in memory, searched for, and
retrieved. It is easy to see how, if taken literally, such ideas can lead to the
conclusion that people's heads are populated with a huge number of pre~
packaged permanent structures corresponding to everything they know. We
will refer to the view that postulates permanent knowledge structures as the
"structural" approach and to the alternative view, that treats mental phenomena as resulting from transient patterns directly crented by t~e functioning
of the biological hardware, as the "biofunctional" approach.
Not only does the biofunctional approach differ from the structural
approach in its rejection of long-term mental structures, but it also differs in
the way it views the dynamic aspect of cognition. The structural view deals
with the dynamic aspects of the mind in terms of searches for and changes to
permanently-stored knowledge structures. Since we question the need to
postulate such structures, we try to avoid this way of dealing with the
problem: If there are no permanent cognitive structures, then they cannot be
found or changed. In biofunctional terms, cognitive patterns are viewed as
transient dynamic structures. In short, along with Bartlett, Bransford and his
colleagues (e.g., Bransford, Nitsch, and Franks, 1977), and Dennett (1983),
we argue that cognition does not involve the selection of pre-existing cognitive
structures; rather, it involves the creation and re-creation of transient ones.
We wish to emphasize, however, that our arguments against the use of
structural concepts must not be interpreted as an attempt to free all psychological exposition from structural terms. As Freeman ( 1975) points out, even
at the more concrete levels of exposition "it is reasonable and perhaps
necessary to describe the manipulations of the central state with concepts that
are both generalized and familiar from common experience ... [although]
there is not and cannot be an a priori relation between those concepts and the
dynamics of the central neural mechanisms" (p. 414 ). Our view is that as long
as the prevailing theoretical context clarifies the meaning of "structural"
terms, their use should cause no problems. When, on the other hand, the
terminology itself determines the underlying theoretical context, as is sometimes the case when long,term memory metaphors are used, we believe an
inappropriate picture of the nature of mental functioning arises.
Over the years cognitive psychologists have gathered a great deal of empirical data that seem to support various aspects of the structural approach.

. 4

f74

IRAN-NEJAD AND ORTONY

However, there is also a great deal of indirect evidence in the neuroscientific

literature that contradicts it (see, e.g., Freeman, 1975; Uttal, 1978). While
there is currently no direct psychological evidence supporting a biofunctional
approach (for a discussion of some indirect evidence, see Schallert, 1982;
Shanklin, 1981 ), the major assumption that differentiates the structural and
the biofunctional perspectives (i.e., mental structures are transient biofunctional patterns) is shared by many current psychobiologists and neuroscien,
tists (see Edelman, 1978; Freeman, 1975; John, 1967, 1972; Katchalsky,
Rowland, and Blumenthal, 1974; Uttal, 1978).
Our account of cognition attempts to bridge the conceptual gap that results
from the absence of a common language between cognitive psychology and
neurophysiology. Although one might think that compatibility of psycholog,
ical theories with what is known about the human nervous system is an
obvious minimal requirement, such compatibility is frequently conspicuously absent. For example, Schmitt (1978) noted that "many theories of
higher brain function (learning, memory, perception, self-awareness, con,
sciousness) have been proposed; but in general they lack cogency with respect
to established anatomical and physiological facts and are without biophysical
and biochemical plausibility" (p. l ). Similarly, Gallistel ( 1980) in discussing a
psychological model of the control oflimb movement (Adams, 1977) claims
that modern neurophysiological work on the mechanisms of co,ordination
renders the theory untenable. The message is clear: psychologists need to
attend more closely to neuroscientific research.
The problem as it relates to psychological research, therefore, does not
seem to be the absence of biologically plausible theories. Such theories exist in
the work of authors like Donchin (1981), Freeman (e.g., 1975), Grossberg
(e.g., 1982), John (e.g., John, 1972; John and Schwartz, 1978), Maturana
(1978), O'Keefe and Nadel (1978, 1979), and Uttal (1978). Rather, in some
subtle way, the problem seems to relate to the deeprseatedness of the influ,
ence of the structural paradigm on cognitive psychology. The structural bias,
we believe, has drawn attention away from existing neuroscientific theories
many of which are in essence biofunctional. For instance, Jenkins (1981)
noted that structural psychology frequently cites William James' treatment of
habit formation and ignores his "true functionalism."
The view we are proposing is based on theoretical constructs at three
interacting levels: (a) the (neuroanatomic) micro-0rganizational level, (b) the
biofunctional macro-organizational level, and (c) the psychological level. At
the micro,organizational level, we will attempt to characterize a physically
unitary and functionally autonomous microsystem as the most elementary
biofunctional unit. Consistent with the current trend in neurophysiology, we
will assume that neural microsystems correspond to neurons, and will often
refer to them as (neuronal) elements. At the macrar0rganizational level, we will
..:

!~

A BIOFUNCTIONAL MODEL

175

attempt to characterize what we refer to as macroaaive structures. These are

patterns of activity resulting from simultaneous functioning of physically
distributed elements. Following Freeman ( 1975), we will refer to the totality
of active elements in the nervous system as the mass action system. Finally, at
the mental level, we will attempt to characterize the central concept in the
biofunctional model, namely what we call the schema-of-the'100ment. This is a
transient mental structure that arises from activity in the mass action system.
We use the term schema,-of-the-moment for two main reasons. First, we argue
that it is in terms of this' 'functioning mass of the moment," as Bartlett ( 1932)
and Head ( 1920) called it, that all cognitive phenomena (comprehension,
learning, remembering, awareness, attention, etc.) take place. Second, the
schema-0f-the-moment is assumed to be the only mental pattern in existence
in a given individual at a particular time-everything else is neuroanatomic or
neuroph ysiological.
The discussions in this paper are organized in thee main sections around
the notion of the schema-0f-the-moment. The first discusses a number of
important background questions including that of how it might be possible
for widely distributed elements to intercommunicate. The second section
discusses the pre,subjective foundations of the schema-0f-the,moment:
exactly what neuronal microsystems are, how they function, how they generate psychological qualities, how they are distributed, and how they functionally relate to one another. In the third section we describe the main characteristics and functions of the schema-0f,the-moment.

The Schema-of-the-Moment: Some Preliminary Issues

The Foundations of the Structural View

There are two fundamental assumptions upon which the structural view of
cognition is based. One is that mental life can be characterized in terms of
various kinds of cognitive processes. The other is that these processes are
performed on long-term-knowledge representations. Neither of these assumpr
tions are part of the biofunctional view.
Most structural theories (e.g., schema theories) assume that the dynamic
aspects of cognition can be accommodated in information processing terms.
For example, the schema selection process is assumed to be the result of some
kind of search or retrieval mechanism. The central concept employed in
information processing models to capture the dynamic (i.e., the processing)
aspect is that of an input-transformation-output sequence-the system
accepts inputs either from memory or from outside, performs transformations on them, and produces resultant outputs, that get stored in memory or

..-

:i 76

IRAN-NEJAD AND ORTONY

are manifested in verbal or nonverbal behavior. 1 The inputs and outputs

themselves are essentially static. Typically, they are knowledge representations-data structures that exist independently of the dynamic component. In
general, information processing models are concerned with "what happens to
information about a stimulus from the 'real world' as it passes through the
system" (Klattky, 1975, p. 11).
We believe that many of the questions addressed by information processing theories arise only as a result of the assumption that the "objects" of
processing possess some independent existential status. In the biofunctional
model, the system is dynamic but it does not process anything; there is no
object of processing. Knowledge is considered to be a transient phenomenon
created and re-created by the functioning of the biological hardware. What is
created lasts only while the underlying biological system that creates it continues to be active. Nothing nonbiological is stored, and apart from its
potential to be re-recreated, knowledge has no permanent existence.
An analogy based on the functioning of the endocrine glandular system
may serve to clarify the contrast between the re-creation and inputtransformation-output views. There is a group of cells located in the cortical
part of the adrenal glands. These cells, when activated, produce the hormone,
cortisol. The cells themselves get activated by another hormone, ACTI-l
( adrenocorticotropic hormone), released in the anterior part of the pituitary
gland. The crucial point is that there is absolutely no input-ttansformationoutput relationship between the stimulator ACTI-l and the produced cortisol.
Adrenal cortical cells, once activated, create the cortisol through, for example,
biochemical operations based on substances other than those contained in
Acrn. It is this dissociation between the input and the output-mediated by
the intrinsic functional properties of the specialized organismic system-that
renders any system-independent ACTI-l-to-cortisol transformation rules, or
any precise formal description of the product based on them, inappropriate.
The qualitative properties of the output are determined entirely by the
biofunctional properties of the specialized cells in the adrenal glands, even
though there may sometimes be a (linear or nonlinear) quantitative relation
between the input and the output. This qualitative dissociation between the
input and the output means that, in principle, cortisol can be produced in the
total absence of ACill, and that cortisol might fail to be produced in the
1There is a striking parallel between the approach taken by the Roman physiologist Galen ( c.AD

130-201) and current information processing psychology-information processing psychology

is based on the same type of industrial plant metaphor that haunted Galenian physiology (see
Miller, 1978). Galen was concerned with how Inanimate matter, as the input to the body via
foodstuff, .is transformed to animate matter. Internal organs (e.g., the heart, the liver, the lungs)
were considered relevant to the extent that they helped carry out such transformations. In
Galen's physiology, as in information processing psychology, "the most notable feature of the
system is the emphasis on manufacrure and ttansformation ... processes which convert ...
substances" (Miller, 1978, p. 187).

A BIOFUNCTIONAL MODEL

177

presence of ACill. In much the same way, neuronal mechanisms active at a

given time combine functionally to create a transient cognitive structure. Such
a dynamic functional organization is input-independent in the sense that there
exists a qualitative dissociation between the characteristics inherent in the
external stimulation and the functional properties of the neuronal system.
Conceptualizing the dynamic aspect of cognition in this way eliminates the
need to postulate the preservation of long-term knowledge structures.
Ironically, Bartlett ( 1932), who is often cited in the context of structural
views of cognition (especially with respect to schema theory), was strongly
opposed to the idea of long-term mental representations and favored some
kind of functional account. He made this point explicitly when he stated that
his approach was based on the "study of the conditions of organic and mental
functions, rather than ... [on] an analysis of mental structures" and that "it
was ... the latter standpoint which developed the traditional principles of
association" (p. 304 ). Our distinction between structure and function is
similar to that made by Bartlett. He believed that a functional approach was
necessary to explain a number of observations that puzzled him. For example,
he found it curious that although incoming information is learned only if it is
incorporated into what he called the "organized mass of the moment," later
recollections of such information, in recall or in thinking, do not always occur
"en masse." What happens to the strong tie established between the input
information and the schema in terms of which it was learned? According to
Bartlett, "In remembering, we appear to be dominated by particular past
events," as opposed to past schemata in their intact original form. He stated
that what was once an "active organized setting looks as if it has somehow
undergone a change, making it possible for parts of it which are remote in time
to have a leading role to play" (p. 202) ..He was also puzzled by the fact that
incoming information is learned in a chronological sequence in which every
new item is strongly influenced by the one before it. However, later recall of
an item does not seem to favor recitation of the entire sequence and would be
highly inefficient and inappropriate if it did (see p. 219). Thus, Bartlett
concluded that thinking, for instance, is only "possible when a way has been
found of breaking up the 'massed' influence of past stimuli and situations,
only when a device has already been discovered for conquering the sequential
tyranny of past reactions" (p. 225). This, according to Bartlett, would be
possible if schema relations were conceived of as transient (i.e., functional)
relations.
Bartlett also favored the functional approach over the structural approach
because he found unacceptable the idea that in an associative structure each
element "retains its essential individuality." He preferred to think of elements
as combining into an organized mass. In an organized mass, the components
are not related by association. Rather, each element loses its identity and
becomes an integrated part of the combination in the same way that, when

---- - ------------- -

'
.l7B

IRAN-NEJAD AND OR TONY

oxygen and hydrogen combine to produce water, the properties of these

elements are no longer evident. Furthermore, the resultant structure comes to
possess emergent properties that are not present in any of the component
elements in the same way that water possesses properties not possessed by its
component elements.
Thus, far from taking mental structures as given, Bartlett was concerned
with two complementary problems: how elements combine into a schema,
and how schema elements manage to free themselves from the shackles of past
combinations (i.e., how they "re-individualize" themselves). While he appar~
ently believed that this was possible only if schema elements, when combin
ing, entered into functional relations, he could not decide what sort of
elements would make this possible, reluctantly picking the image as his
candidate. His reluctance seems to have been based in part on a realization that
images are overly subjective and insufficiently biological (seep. 220). lmages
are themselves cognitive structures, and Bartlett apparently felt that they
lacked the appropriate combinatorial properties that the true elements in a.
functioning system would require. Our solution, to be discussed later, is to
specify, at the pre-subjective neuronal level, elements that are biofunctionally
(and, only by extrapolation, psychologically) primitive.

Mental Relations and Brain Connections

Once the notion of long-term cognitive structures is abandoned, the question arises as to the relation between the neuroanatomic network and the
cognitive organization. In particular, it becomes necessary to consider the
extent to which, if any, there is structural conformity between the two
systems. Minsky ( 1980) and Norman ( 1980) draw attention to the problem of
specifying this relation and refer to it as the "crossbar" and "address"
problem respectively. According to Minsky, "this problem confronts every
brain theory that tries to explain how the mind is capable of any great range of
'associations"' (p. 124). According to Norman, "associations among memory
concepts ... [imply] much too much knowledge of the wire (or of its
biological equivalent) that is to snake its way among the already existing stuff'
(p. 22).
In theory, there are at least three types of solutions all of which have been
proposed at one time or another. The first possibility is to postulate a
particular (pre-existing or, rather, pre-functional) neuroanatomic pattern,
partial or complete, corresponding to every cognitive pattern. This essentially
amounts to mapping the structural cognitive network into an isomorphic
neuroanatomic network. Such isomorphism was a major psychobiological
premise in Gestalt psychology (see Uttal, 1978, for a discussion of this).
Isomorphism is also implicit in the connectionist approach to neural modeling of semantic networks, whether these models represent concepts as partic-

A BIOFUNCTIONAL MODEL
:---

179

ular hardware units (Fahlman, 1981; Feldman, 1979) or as patterns of activity

in localized cell assemblies (Hinton, 1981).
The seco nd possibility is that the neuronal network is analogous to some
sort of sophisticated telephone network. By allowing directional hard-wired
routes between elements, the nervous system would somehow generate twounit or multi-unit (transient) communication patterns. A telephone netwo rk
is directional because the initiating unit must know the "address" of the target
unit{s ). Directional connectionist models imply "that 'remembering' requires
the discharge of those particular cells which constitute the new line, and those
of the cells to which the line is directed" Oohn, 1972 ). John and his colleagues
(e.g., John, 1967, 1972; John and Schwartz, 1978; Thatcher and John, 1977)
have been among the most outspoken critics of connectionist models, arguing
that, for example, responses to even the most elementary stimuli (e.g., a flash
oflight or a click) are made by cells distributed throughout the brain and that
a given cell participates in many functional patterns. Although few psychologists and neuroscientists still subscribe to the type of connectionism that John
criticized, connectionism in some form or another is still widely embraced
(see, for example, the essays in Hinton and Anderson, 1981). It is now
generally recognized that mental relations are variable. But modern connectionists attempt to accommodate such variability in terms of synaptic plasticity. As Uttal ( 1978) has convincingly argued, the large conceptual gap between
synaptic plasticity (defined in terms of synaptic weights, facilitation levels,
etc.) and complex mental phenomena renders synaptic connectionism
implausible (pp. 540-541).
Another line of argument against connectionist mo dels comes from recent
developments in theoretical chemistry and their application to activity in
masses ofneurons (Freeman, 1975; Katchalsky et al., 1974 ). The thrust of the
argument, which is incidentally highly reminiscent of Bartlett's criticism of '
associative connectionism cited earlier, is that neural activity, far from occurring in terms of independent hardware units joined by neuroanatomic connections, tends towards organization and self-consistency in a fashion analogous to that occurring in diffusion-coupled chemical reactions. According to
Freeman (1975), these ideas "lead to expectations of neural activity quite
different from the discrete characteristics of activity in networks" (p. 8).
Freeman (p. 8) claims that given this perspective, what emerges "from the
study of neural mass action is not merely an extension of current understanding; it is revolutionary in the sense defined by Kuhn (1970)."
Incidentally, it must be noted that anticonnectionists do not reject the
existence of precise neuroanatomic connections. Rather, they maintain that
although precise neural connections exist at the anatomic level, it is necessary
to distinguish between anatomical connections and functional relations, and
that a set or constellation of neurons having fixed anatomical connections may
admit of many functional patterns (Freeman, 1975).

18\.,.

J:

IRAN-NEJAD AND OR TONY

A BIOFUNCTIONAL MODEL

Note: Pages 182-186 have been inserted from Rich Text and reveal occasional font displacements

Ifsynaptic plasticity or modifiability does indeed fail to adequately explain

functional connections among elements, the nervous system must accomodate variable functional relations among elements in some other way. The fact
that neuronal elements are capable of interacting with other distant neuronal
elements that are distributed throughout the brain suggests that there might
exist within the nervous system some sort of relational medium to make such
interaction-at-a-distance possible, The third possibility, therefore, and the
one we find most plausible, is that in addition to synaptic modifiability, the
neuronal network also communicates through an all-spreading nondirectional relational medium. Such a medium would allow (within amplitude,
etc., constraints) nondirectional conductance of electrical or chemical energy
(signals) in addition to directional element-to-element interactions,
In a totally nondirectional network every signal can potentially reach all
specialized units and no signal is aimed at any particular unit directly. Thus,
the initiating unit does not need to know the address of the target unit. Rather,
target units are specialized to get activated in response to (or "to recognize")
the functioning of the initiating unit that produces the signal, and to remain
indifferent to the functioning (and, thereby, to the signal) of any other.
Particularly relevant examples of specialized systems functioning in a nondirectional environment are the auditory and visual systems of animals. While
both of these mechanisms function in the same environment-filled with
sound and light waves-the ears respond to sounds but are deaf to light while
the eyes perceive light but are blind to sounds. One can imagine a similar
principle to hold for the neural network. In other words, it is possible that
neuronal elements interact, not merely because source elements are connected or have the addresses (or phone numbers) of target elements, but
because the elements themselves are uniquely specialized. This specialization
permits target elements to "hear" impulse patterns that spread, like sound
waves, throughout the neural network if those patterns happen to be signals in
their "language." We believe this shift of "responsibility" from source units
to target units solves the problem of address. Thus, in this particular sense, the
act of communication is more like a (radio) broadcast than a (phone) call
because in a broadcast the source unit emits the signal indisciminately whereas
in a phone call a decision must be made as to who is going to receive the call
and the phone number of the target unit must be known,
The assumption of non-directional communication among neuronal elements critically depends on the nature oflocalization and distribution in the
nervous system. By localization we mean that specialized elements that are
functionally highly selective are fixed with respect to their physical location.
Disr:ribution, on the other hand, means that the elements that are simultaneously functioning can be physically widely spread apart. Thus, not only does
the nondirectional hypothesis solve the problem of address but it also
resolves the apparent imcompatibility between localization and distribution.

181

Early distributed models were explicitly nonlocalizationist. Lashley's (1950)

original formulation of distribution stated that "the same neurons which
retain the memory traces of one experience must also participate in countless
other activities" (p. 479 ): Somehow every neuron learned (or stored) everything that many other neurons learned. Nonlocalized storage is also an
assumption in holographic models of distributed memory (Pribram, 1981;
Wess and Roder, 1977). However, evidence gathered by Hubel and Wiesel
(1959, 1962, 1965) and others has shown thatthe brain is not a homogeneous
mass. Whereas traditional models of distributed memory considered localization and distribution to be antithetical, more recent evidence suggests that
distribution and localization are not incompatible (e.g., Freeman, 1975; John
and Killam, 1960; John and Schwartz, 1978; Uttal, 1978). For instance,
Freeman argued that "the behavior ofanimals depends both on the properties
of neurons and on the ways in which they are functionally connected or
interconnected" (Freeman, 1975, p. 4, italics adde_d). Similarly, Uttal (1978)
pointed out that while one must "emphasize the concept of interacting
systems and the premise of nonunique localization of each psychological
function ... it also appears that there is a considerable degree of differentiation of function of the various areas of the brain and the brain stem; that is,
they are not equipotential" (p. 354 ). Variable functional relations among
interacting systems can be accommodated by a system consisting of specialized units that can communicate in terms of a nondirectional, all-spreading
environment.
We have hypothesized an all-spreading medium in order to clarify, at least
conceptually, the problem of interaction-at-a-distance in a mass action system
with distributed elements. An all-spreading environment, however, does not
mean that electrical or chemical conductance takes place in a vacuum, even
though some sort of extracellular propagation may play an important role (see
Nicholson, 1979). Neither does it mean nonspecificity or imprecision in the
pattern of actual neural connections. The neural network as a whole may
serve as a common network.
The two types of physical relational vehicles (directional and nondirectional) may be illustrated by an analogy to the functioning of exocrine
and endocrine glandular systems. Exocrine glands (e.g., the salivary glands)
release their products into specific ducts which direct them to target organs.
These would correspond to directional element-to-element (neuroanatomic)
connections. Endocrine glands, on the other hand, secrete their products into
the extracellular fluid surrounding capillaries. The hormones they produce
enter the blood circulation system, which is itself an all-spreading environment. This make.s it possible, for example, for the ACTH released in the
anterior pituitary gland, located on the lower surface of the brain, to stimulate
(activate) cortical adrenal cells located above the kidneys. It is conceivable, in
principle, that a direct point-to-point duct could have been physically availa-

182

IRAN-NEJAD AND ORTONY

ble to carry ACTH from the pituitary to the adrenal glands. However, if a tube
were to be available from every endocrine gland to its target organ, organisms
would become monstrously complex. Instead, ACTH enters the blood circulation system. This, of course, takes the hormone to other irrelevant organs as
well (hence, all-spreading and nondirectional), but it is also sure to reach die
adrenal cells which are speciali.7.ed to get activated by it, because these cells like
everything else are connected to die blood circulation network.
The possibility that the nervous system is also, in part, an all-spreading
environment analogous to die blood circulation system cannot be ruled out.
As early as the 19201s, Paul Weiss argued against the conncctionist view and
' concluded, based on die then existing evidence, that "the central nervous
system and the non-nervous periphery entertain dieir mutual correspondence
by means of some sort of sending-receiving mechanism, specific for each
individual muscle." According to this view, die central nervous system is
"endowed with die capacity for discharging as many different modes or forms
of impulses as there are different muscles in the limb." There is a specific
impulse for every muscle receptor. Every muscle receptor, on the other hand,
"would possess the power to respond selectively" to its proper impulse.
Consequently, if "die central impulses for a limb muscle were circularized in
die whole limb" die mechanism of selectivity of function "would ensure that
every call be answered by the correct muscle, even diough die latter may have
been displaced, re-innervated by strange nerves, and prevented from sending
informative messages back to the centers" (Weiss, 1936, pp. 511-512).
Weiss's resonance principle is no longer generally accepted by developmental
neuroscientists, but we believe his ideas concerning indiscriminate synapdc
connectivity, successfully challenged by Sperry and his associates (see Attardi
and Sperry, 1960,1963; Meyer and Sperry, 1976), must be distinguished from
his suggestive element-impulse specificity hypothesis, which has yet to be
directly tested.
An all-spreading functional environment implies that, regardless of its
place of origin in the nervous system, die signal that a given functional pattern
generates can stimulate clements diat "recognize" it wherever dicy may be
located in the nervous system. There are definite indications diat diis may be
die case. Consider a letter recognition (identification) task. Images ordinarily
begin on the retina and presumably stimulate corresponding centers or
elements somewhere in the brain. It is conceivable that specific "image--tocenter" connections as well as long-term graphemic patterns could mediate
recognition. However, recognition need not depend on particular hard-wired
connections or on pre-existing long-term associations. Blindfolded subjects
are capable of correctly identifying letters (finger-written) on dieir skin.
White, Saunders, Scadden, Bach-Y-Rita, and Collins (1970) used a visual
substitution apparatus which converted optical images into tactile displays
which blind or blindfolded subjects were able to "sec widi dieir skin." It was

A BlOFUNCTIONAL MODEL

183

shown that "subjects are able to perceive certain simple displays...almost as

soon as they have been introduced" (p. 23) and diat widi minimal amounts of
training diey are able "to identify familiar objects and to describe their
arrangement in depth" (p. 25).
The hypodiesis of functional communication between distinctively specialized neuronal elements also finds support in the evidence diat the reestablishment of original functional relations is possible even after specialized
cells are surgically removed from dieir original site and arc regrown at a
different part of die body. If a piece of skin is removed from the belly region of
a salamander and planted on its back and if, after regeneration, diis skin, now
on the back, is stimulated, the animal proceeds to scratch its belly, the orii,,>inal
site. Such seemingly maladaptive responses, extensively studied by Sperry
and others, arc often discussed in the light of die nature/nurture Issue (see,
e.g., Rose, 1976). However, more basic than whedicr somediing is innate or
acquired is die problem of how it works. One may simply assume diat
regeneration only connects die pre-specialized skin receptor cells to an allspreading neural network. There is no need for the re-establishment of
particular nerve fibers to wind dieir path, dirough some mysterious innate
guiding mechanism, all the way to the related central cells. Once specialized
receptor cells arc merely connected to die neural network (or perhaps to the
particular brain region), diey can activate die individual target cells dirough
generation of unique energy patterns. The energy patterns, generated by die
central cells, can, in turn, activate the muscles involved in die scratching of die
belly. Because die belly receptor cells function in die same unique fashion
regardless of where they arc located, and because this functioning is recognized by the related central cells as "belly" stimulation, the animal responds
maladaptively. Sperry (e.g., 1943) explained diese results "in terms of reestablishment of specific anatomical associations rather than in terms of
specific nerve energy and resonance phenomena." But he also emphasized
diat "die latter possibility is by no means excluded" (p. 47). In fact, it is still
not dear that a resolution between Sperry's directional connectionism and
Weiss's element-impulse specificity model has yet been achieved (see Meyer
and Sperry, 1976; Sperry, 1966; Wall, 1966; Weiss, 1966). As Wall (1966)
argued, "the so-called specificity of neuronal function ... may mean dint
specificity of function can be attained widiout a microscopic determination of
die exact morphological structure of some parts of die nervous system''
(p. 230).
Given the concept of an all-spreading relational vehicle, die most efficient
way of relating die cognitive system and the neuronal system seems to be to
assume that (a) die cognitive system is comprised of transient functional
relations, and diat (b) post-functional patterns are independent of any isomorphic pre-functional neural associations. Independence of post-functional
(mental) relations also resolves Minsky's crossbar problem. As Minsky

IRAN-NEJAD AND OR TONY

( 1980) put it "if the mechanisms of thought can be divided into specialists that
intercommunicate only sparsely, then the crossbar problem may need no
general solution. For then, most pairs of agents will have no real need to talk to
one another; indeed, since they speak ... different languages, they could not
even understand each other" (p. 125). And, to continue the metaphor, if they
can understand each other, they will do so regardless of where they are located
or whether they are connected directly so long as they can "hear" each other
(i.e., so long as they are part of the overall neural network).
Pre-Subjective Foundations of the Schema-of-the-Moment
This section discusses some key assumptions pertaining to the microbiofunctional and macrobiofunctional properties of the brain in an attempt to
show how these properties can explain the pre-subjective foundations of the
schema-of-the-moment. These assumptions, while speculative in detail, are
generally compatible with existing neurophysiological literature bearing on
localized and distributed functional properties of the nervous system. For
clarity of exposition, we shall not interrupt to substantiate every claim or to
discuss the controversies that might be involved. The interested reader is
encouraged to consult Freeman (1975, chap. 1) on the nature of dynamic
patterns, Hubel and Wiesel (1980) on localization in the primary visual
cortex, Lynch (1980) on distribution of function in the posterior parietal
cortex, Selverston (1980) on the central pattern generators underlying rhythmic behavior, O'Keefe and Nadel (1979) on localization of "place-coded"
neurons in the hippocampus, Puccetti and Dykes (1978) on the problem of
accounting for qualitative differences between subjective experiences of
touch, hearing, and vision, and Uttal (1978) on the role of microscopic
structures in the formation of dynamic representations.

Are Pre-Existing Long-Term Pattem.s Necessary?

As we have said, current approaches to cognition and comprehension
presuppose the existence oflong-term relatively static knowledge representations that underlie cognitive functioning. One motivation for hypothesizing
long-term mental structures is the fact that ideas seem to come to mind (to be
recalled, etc.) together, or in relation to one another. It is then assumed that
they stay together in some cognitive warehouse, even when they are not
operative. Thus, the structural approach maintains that ideas are related before
they become active, and that remembering involves activation of static
knowledge representations.
By contrast, from the biofunctional perspective, there are no pre-existing
mental patterns. Cognitive relations are established only after distributed

A BIOFUNCTIONAL MODEL

185

neuronal elements become active. Like chemical elements, neuronal elements

have properties that determine their combinatorial potentials. Only when two
or more elements with appropriate combinatorial properties (see the section
on types of combinatorial relations below) are in a state of simultaneous
functioning do they combine to generate a cognitive pattern. Consider, as
another analogy, a collection of colored lightbulbs. When a subset of them is
on, a unique pattern oflight and color is generated. It does not matter whether
individual bulbs are physically connected or when each bulb goes on. The
characteristics of the pattern are determined by the participating elements and
not by how the bulbs are connected physically or by the history of their
participation. A given pattern could result from a long sequence of events in
which some bulbs would go on and some would go off. Once some or all of
the bulbs go off, the particular pattern no longer exists.
Similarly, a constellation of active neuronal elements generates an idea or
concept as the component elements combine intt> a biofunctional pattern.
This can be illustrated in terms of what might happen when a word like dog is
understood after it is heard. Hearing the word activates a constellation of
independent neuronal elements. This happens, not because the elements are
preconnected as a neuroanatomic pattern, but because the stimulus itself (in
this case the pattern of sound waves that reach the ear) has independent signal
components that activate each of the independent elements of a constellationelements that can be physically distributed throughout the brain. Once the
elements are active, they combine into a biofunctional pattern and, in doing
so, they generate the concept corresponding to the word. This notion can be
clarified in terms of another analogy. When a handful of pebbles is thrown
into a pond, each pebble hits the water at a different spot and creates a wave
pattern. Then, the wave patterns of all the pebbles combine to form a global
pattern of waves. This global pattern, which results from the combined effects
of the individual pebbles, corresponds to the dynamic pattern created by the
active elements of the constellation underlying a concept. The concept will
arise in the psychological experience of the moment whenever the particular
elements that produce it change their activity, functioning as a coherent
combination. It does not matter how each element comes to be active (i.e.,
one at a time, all at once, etc.) or what causes them to be active (the sight of the
word, the sound of the word, a dog, etc.). In fact, according to this view,
elements that create a particular concept can come to be active as a result of
the combination of elements from two or more unrelated constellations
whose corresponding concepts have no apparent relation to one another. In
sum, while in a structural pattern it is the long-term "relations among
elements that count" (Piaget, 1970, p. 9 ), for a functional pattern what counts
is the elements themselves-their characteristics, how they function, and how
they functionally combine.

. 1186.

.'

_,-

IRAN-NEJAD AND OR TONY

Neuronal Elements as Biofunctional Micro.ryscems

;.
! :

-::

:i"

The model we are proposing places a heavy theoretical burden on the

notion of neuronal elements. It is, therefore, necessary to specify exactly what
sorts of biofunctional properties these elements should have in order to
combine into macroactive structures and in order to generate subjective
experiences with sufficient qualitative diversity.
While there is general agreement that mental structures are complex and
that they consist of more elementary components, there is less agreement as to
what these components, or elements, are. Depending on the situation or level
of analysis, theorists have used semantic features, perceptual features, images,
ideas, etc. as primitive units of analysis. However, in an approach such as
Bartlett's or ours, where it is critical that the elements be capable of combining
while remaining; in principle, individualizable, the choice of an appropriate
standard element is much more constrained. The biofunctional model
assumes that the most elementary theoretical construct is a physically unitary
(although not physically elemental) and functionally autonomous microsystem. The assumption that the elements are physically unitary systems means
that they are taken to be neurophysiological "atoms." The assumption that
the elements are functionally autonomous is compatible with their being
viewed as cognitive "atoms." In other words, we view these elements as being
the most elementary biofunctional units in the nervous system and the locus
of the link between the mind and the brain.
It is important to point out that our use of the term neuronal element is not
meant to imply that single neurons actually represent mental structures.
Traditionally, such concepts as pontifical neurons (Fessard, 1954; Sherrington, 1947), cardinal cells (Barlow, 1969), command neurons (Wiersma and
Ikeda, 1964), and feature detectors (Hubel and Wiesel, 1980) have been
postulated to represent complex psychological patterns. In contrast to these
views, we assume that mental structures are "molecular" rather than
"atomic"; more than one neuronal element must be involved in the creation
of any mental entity. In other words, individual neuronal elements do not
represent mental structures (i.e., features, images, etc.); only distributed
constellations of them do.
The claim that it is distributed constellations of neurons, and not individual
ones, that are responsible for the creation of cognitive structures could be
interpreted as meaning that individual neuronal elements are equipotential.
This is certainly not our view. Rather, like atoms of particular simple substances, neuronal microsystems are assumed to have unique functional properties.
As with chemical elements, these properties constrain the types of combinations in which an element can participate, and they constrain the conditions
under which such participation can occur. However, within these constraints,
a particular element can participate in an indefinite number of combinations.

A BIOFUNCTIONAL MODEL

187

The biofunctional perspective also means that the neurons are not equipotential in the sense used by nonlocalization theorists such as Lashley. While in
the biofunctional model, each neuronal element can be considered a
"memory," or rather a "potential memory" element, it is not assumed that
neurons retain memory traces; in particular it is not necessary to assume that
neurons retain memory traces for a large number of experiences and behaviors (Lashley, 1950, p. 479). Such a claim would make the activity of neurons
completely dependent on past traces by implying that they could only participate in combinations they already had traces for, and that they would only
respond to stimuli for which they had a matching trace. In the biofunctional
model, past experiences are re-created because the same constellations of
elements recur, not because something is stored in each element. The system
is completely generative, creating new experiences entirely through new
combinations. At the micro-organizational level, each element is physically
located in some area of the brain. However, since t:he elements can functionally coact or interact at a distance through an all-spreading medium, the
system can also be fully distributed.
At the macro-organizational level, the totality of active neuronal elements
constitutes a mass action system (Freeman, 1975). In recent years, the nature
of macroactivity in the nervous system has been the subject of an interesting
new theoretical approach derived from the field of dynamics (Freeman, 1975;
Katchalsky et al., 1974; Prigogine and Nicholis, 1971; Kugler, Kelso, and
Turvey, 1980). According to this view, macroactive structures are dynamic
patterns that (a) consist of discrete elements, (b) are self.-organizing, and (c)
consume energy to maintain stability (see, e.g., Freeman, 1975). Katchalsky
and his colleagues, who pioneered the work in this area, state that one possible
. consequence of "considering discrete systems embedded in a continuous
system would be the subordination of obvious structural discreteness to a
functional one: the spatially discrete elements could be brought to functional
continuity ... or the structurally continuous medium to functional (dynamic)
discontinuity" (Rowland, reported in Katchalskyetal., 1974, p. 78). Clearly,
the biofunctional model is compatible with these views. Rewording the above
quotation in terms of the present view: one ad vantage of considering neuronal
microsysterns embedded in dynamic macroactive structures that they themselves create would be the subordination of the microsystems to their overall
functional organization. A constellation of physically unitary microsystems
could be brought to functional continuity or the continuous macroactive
organization can be subordinated to the independent functioning of individual
microsystems. It seems to us that it was shifts of this sort in the relative
subordination of individual components to the global structure and vice
versa, that Bartlett (1932) deemed necessary in mental functioning when he
argued that not only must individual elements combine into an organized
mass but that after they do, they must once again be individualizable in the
context of the global structure.

. 18&- .

A BIOFUNCTIONAL MODEL

IRAN-NEJAD AND OR TONY

Microsystems as Elementary Loci of Psychological Qualities

,1

;.
<'

<:

It seems to us that the key to bridging the gap between cognitive psychology
and the neurosciences lies in the specification of the manner in which the
nervous system might generate qualitatively diverse psychological experiences. Concern for this problem does not seem to be a characteristic of current
practice in cognitive psychology, which tends to focus on unconscious mental
structures and processes. However, there has been some concern with this
issue in the neurosciences (see, e.g., Eccles, 1953; Sperry, 1952, 1969, 1977).
Recently, the problem has been brought into sharp focus by Puccetti and
Dykes ( 1978) who emphasized the apparent structural ( cytotechtonic) similarity of the primary visual, auditory, and somesthetic sensory areas. Noting
that these similarities afford little opportunity for explaining differences in
subjective quality, they concluded that "not only is present-day neuroscience
unable to account for the subjective differences [between vision, hearing, and
touch] in terms of discrete neural mechanisms, but there is no good indication
that it ever will be able to do so" (p. 337). While some of the commentators
did not find Puccetti and Dykes's conclusions very'convincing, others agreed
with their assessment of the state of the art. For example, Szentagothai ( 1978)
went so far as to say that "the spectacular developments in the last quarter of a
century ... have widened ... the gulf between the brain and the mind"
(p. 367).
Our views on subjective quality are in general agreement with those
advanced over the years by Sperry (e.g., 1952, 1976, 1977), who claimed (a)
that subjective qualities emerge from the activity of the brain, and (b) that
these qualitities, in tum, exert a causal influence on brain activity. However, in
searching for an appropriate level of analysis to explain the causes of subjective qualities, Sperry ( 1978) excluded the microbiofunctional level and argued
that "it is our bet that (the proper level] is not at the atomic, molecular,
cellular, or nerve-relay levels, nor even at the sensory cortical levels, but rather
at a somewhat higher level that involves ... centralized adjustments of the
brain as a unit" (p. 366). In this respect, we disagree with Sperry. While
Sperry's theory might account for the undifferentiated conscious experience
(i.e., global awareness), we believe it fails to explain finer qualitative discriminations (i.e., focal awareness). By fixing the locus of subjective quality at the
level of neuronal microsystems, the biofunctional model can not only
account for people's competence in making fine subjective discriminations
(by individualizing components of the whole), but am also explain global
differences in subjective experiences (because microsystems can combine into
macrobiofunctional organizations that involve the activity of the entire brain).
In order to demonstrate how psychological qualities can arise as a consequence of macroactivity in distributed constellations of elements, we must
take a closer look at the key concepts of specialization, constellation, and

189

combination. In order to understand what we mean by element specialization, it

.,i
I

may be helpful to again use the lightbulb analogy. Let us suppose that our
array of lightbulbs contains two broad categories of bulbs, namely, colored
and uncolored ones. We will call the colored bulbs "specialized elements."
The uncolored bulbs we will call "raw elements," implying that they can
become specialized by getting painted a particular color. In this array, each
bulb can "perform" a few feats always in the same unique fashion: it can go on
or off, it can become brighter or dimmer, and if it is not already specialized it
can become so. Similarly (and now we are out of the analogy), the neuronal
network can be assumed to consist of a great number of elements, each of
which is or can become specialized and each of which can get activated or
inhibited. In addition, each specialized element can (a) change its rate of
activity, (b) produce a unique pattern of energy (i.e., a signal), ( c) initiate
functioning consistently in the presence of some unique pattern of internal or
external energy (i.e., a signal), and (d) generate, when functioning, a unique
feeling of awareness. In this sense, a specialized element is a discrete unit with
quite specific but very limited properties. This assumption is consistent with
the view that "neurons, in the course of differentiation and development and
in processing of information over the span of the organism's lifetime, develop
unique identities: generically and experientially determined individualities"
(Schmitt, 1970, p. 208); the assumption is also consistent with the evidence
that the relative number of highly specific cells seems to vary drastically with
experience (Imbert and Buisseret, 1975).
When two or more specialized elements function simultaneously they can
combine to create a macroactive organization or combination (see the section
on types of combinatorial relations below). To emphasize the physical distribution of elements participating in a combination, we refer to the physical
counterpart of a combination as a constellation. Constellations differ from
elements in several respects. First, unlike elements, they cannot be considered
specialized. This is because they contain autonomous elements which can
participate in other constellations. Second, elements are assumed to be
localized and physically unitary while constellations can have elements scattered throughout the nervous system. Third, while individual elements possess element-specific biofunctional properties that theoretically are unambiguously traceable to some unitary physical entity-the element itself-constellations have nonspecific (i.e., emergent) properties which result from the
functional combination of the elements involved and which cannot be traced
to any unitary physical entity because they are different from those possessed
by any one of the participating elements.
The notion of combination encompasses four major biofunctional aspects
that result from the activity of the corresponding constellation. First, combination is the establishment of a transient dynamic pattern involving anatomically distributed elements-the combinat.orial aspect. Second, combination

190 ...

IRAN-NEJAD AND OR TONY

A BIOFUNCTIONAL MODEL

191

involves the merging of element-specific energy patterns resulting in nonspecific (i.e., emergent) energy patterns (i.e., signals)-the relational aspect. In the
case of energy, pattern combinations can be conceptualized in terms of
interference patterns (Lashley, 1950). Third, combination involves blending
of element-specific awareness patterns into emergent awareness patterns-the
qualitative aspect. Finally, combination involves the merging of elementspecific activity into a global intensity dimension-the quantitative aspect.
Thus, not only does the biofunctional approach specify how "potential
memory" elements (i.e., the microsystems) might be distributed but it also
implies that these elements are distributed loci of potential subjective
qualities.
It appears, therefore, that the concepts of specialization, constellation, and
combination can provide a foundation in terms of which one can account for
the origination of psychological quality within the biofunctional mode. However, a fuller understanding of this account requires an examination of three
specific issues. The first of these, discussed in the next section, pertains to the
physical locus of subjective qualities. The second, has to do with the kinds of
constraints that exist on possible combinations, and the third is concerned
with singling out the activity of components of combinations.

Localization and Distribution of Subjective Qualities

As already mentioned, the relationship between the brain hardware and
subjective qualities has received a provocative treatment in Puccetti and
Dykes ( 1978) and the associated commentaries. Puccetti and Dykes started by
pointing out that vision and hearing, for instance, are qualitatively different
subjective experiences. They then assumed that if subjective qualities are
localized in the brain, one would expect to find corresponding differences in
the physical structure of those areas of the brain ordinarily associated with
vision and hearing. However, in reviewing the relevant evidence, they could
find no support for such structural differences, and concluded that perhaps
dualism was the only way out of the dilemma. Sperry ( 1978), on the other
hand, questioned the validity of Puccetti and Dykes's assumption and argued
that qualitative differences need not be reflected in "activity in the primary
sensory fields of the cortex" (p. 336). Rather, they can emerge from the
activity of the brain as a whole. Like Puccetti and Dykes, we assume that
differences in psychological quality presuppose differences at the neuronal
level, even though we disagree with their corollary assumption that qualitative
differences must be evident in the anatomic strucrure of various brain areas.
Rather, we assume that different brain areas differ in biofunctional properties
and not necessarily in cytostructural characteristics. We also question Sperry's,
and Puccetti and Dykes's, claim that the causal loci of subjective qualities
cannot exist in particular areas of the brain. In fact, a major goal of the

biofunctional theory is to explain differences in subjective quality in terms of

microbiofunctional properties of brain areas. According to this view, differences must still exist even if they are not evident in the cytoarchitectural make
up of brain tissues.
In terms o f the biofunctional model, the solution to the problem of the
locus of psychological quality lies in distinguishing between two types of
localization. An important implication of Puccetti and Dykes's argument is
that it suggests the need for just such a distinction. Localization ordinarily
refers to the physical locus of qualitatively complex psychological phenomena
(e.g., semantic features, concepts, the self), or to the physical locus of mechanisms that deal with qualitatively complex input (e.g., short-term memory,
pattern analyzers, etc.). Puccetti and Dykes's approach, on the other hand,
s~ggests that localization occurs according to qualitatively similar inputs (i.e.,
visual and auditory) or qualitatively similar subjective experiences (e.g., visual
imagery and auditory imagery). It is this type oflocalization that is compatible
with the notion of distribution discussed earlier. Neuronal elements are
localized in particular areas of the brain according to their qualitative functional affinities, elements with similar qualitative properties (e.g., those generating spatial qualities) tend to be physically close together and those with
dissimilar qualitative properties (i.e., spatial vs. affective elements) tend to be
removed from one another. In other words, the biofunctional theory implies
that, in principle, at the microbiofunctional (i.e., elemental} level, quality is
homogeneously localized. At the macrobiofunctional level, on the other hand,
constellating elements that generate complex and varied conceptual categories
cannot form physically localized groups. It appears that at this level the
complex nature of mental categories and functions necessarily requires hetero-

geneous distribution.
A clearer illustration oflocalization according to biofunctional affinities of
neuronal elements can be found in the work of O'Keefe and Nadel (1978,
1979) on localization of function in the hippocampus. These authors have
postulated that in order to find their way around an environment, organisms
make use of two partially independent systems called the locale and the ta.xon
systems. The locale system, containing (qualitatively homogeneous} "placecoded" neurons, is responsible for the generation of absolute, nonegocentric,
spatial maps. This cognitive mapping system, they claimed, is localized in the
hippocampus. The taxon system, on the other hand, is responsible for
(qualitatively heterogeneous) taxonomic or categorical information, comprises the rest of the brain, and consists of a number of separate subsystems.
O'Keefe and Nadel's approach may be contrasted with that of Olton, Becker,
and Handelmann ( 1979) who have argued that the hippocampus is the seat of
the working memory. These two approaches to localization are based on very
different beliefs about the functional properties of the brain. O'Keefe and
Nadel's approach is consistent with the biofunctional model in that it implies

'L.
. ,..

IRAN-NEJAD AND OR TONY

that place-neurons, as a qualitatively homogeneous class, form a localized

biofunctional set.

Types of Combinatorial Relations

~.. '.

...
!.<

::

; -

" i

; ;:

-, '

A system comprising a large number of partially-independent subsystems

must possess combinatorial properties so that coactivation (or interaction)
among the subsystems is possible. For instance, it can be assumed that only a
subset of elements within each brain subsystem and throughout the entire
mass action system, constellate and reconstellate from moment to moment.
Given that neuronal elements are specialized, and given the assumption that
they can interact at a distance, it is possible to consider biofunctional relations
among distributed elements indepe11dently of the actual neuroanatomic
connections.
To characterize the combinatorial relations (i.e., potentials) of neuronal
elements, we adopt a system of relations that was postulated for somewhat
different purposes by Festinger ( 1957). In Festinger's system, three types of
relations were assumed to exist among cognitions (cognitive units): consonance, dissonance, and irrelevance. According to Festinger, cognitions X and Y
are consonant if one follows from the other. When two cognitions have
nothing to do with each other, the relation is irrelevance. And, finally, two
elements are in a dissonant relation, "if considering these two alone, the
obverse of one element would follow from the o ther" (p. 13).
Festinger's system of relations can be re-conceptualized as dynamic combinatorial properties of neuronal elements, rather than as relations among
complex mental units. Recall that an important consequence of element
specialization is that neuronal elements can generate characteristic energy (or
interference) patterns that provide the necessary and sufficient conditions
(i.e., the signals) for the activation of other elements. This means that the
functional relation between any two elements does not require that they be
connected to one ano ther directly . Rather, in the same fashion that squaredancers respond to the sound pattern in the air, which is, loosely speaking, a
"complementary combination" of the sound patterns generated by the voice
of the caller and that generated by the music, the neuronal elements also
"dance" to the "sound" of the dominant interference pattern of the movement that spreads indiscriminately through the neuronal network. The difference is that in the case of the neuro nal elements these microsystems are each
simultaneously a caller, a musician, and a dancer- it is a caller/ m usician/dancer. Consider, for instance, three elements, A, B, and C. Suppose that element
A is specialized to generate a unique energy pattern, E{A). Element B is
specialized to become activated in the presence of E(A); and both A and Bare
specialized to coact (i.e., engage, for instance, in synchronous rhythmic
activity) in the presence of E(AB), where E(AB) is a consonant (or comple-

----------------- - - - - - -- ----- .

A BIOFUNCTIONAL MODEL

193

mentary) combination of E(A) and E(B). This means that functionall

E(AB) = E(A) = E(B), in much the same way as when different instrumen~
playing in unison are producing the same tune, both individually and as a
group. To continue the analogy, when B gets activated it joins the band, and
adopts the tune of the moment. In this sense, there is an A-to-B consonant
ac~vity initiation relationship and an A-B synchronous coactivation relationship. On the other hand, specialization of elements other than B would be
such that they could not "hear" E{A) or E{AB). This would mean an A-toNON-B irrelevant biofuntional relationship. Similarly, a C-to-B activation~
inhibition consonant relationship might imply a C-to-NON-B irrelevant biofunctional relation.
Now suppose that A and Care active at the same time. E(A) will tend to
activate B while E(C) will tend to inhibit it. This would constitute a dissonant
biofunctional relation among A, B, and C as a constellation . Such dissonance
:'ould cause a momentary state of dissolution,, by perturbing the prevailing
tnterfe~enc.e ~attern (.hence, tending to change its nature from signal to noise),
rendermg it 10effect1ve, and thereby breaking the dynamic combinatorial
relations among A, B, C. By analogy, dissonance is caused by participation of
individuals who can hear the music but do not know how to call/playI dance
with it. Dissonance is caused because these participants continue to engage in
activities incompatible with the ongoing tune and so tend to disrupt the
operation, locally or globally. Resolution can be achieved if a new interference
pattern emerges to support a surviving and/ or novel constellation of elements. This can happen if expert dancers begin to ignore the unskilled ones if
unskilled dancers drop out, if new experts join in and act as "tu~e
translators" for unskilled d;mcers, o r if the unskilled dancers manage to make
their own tune predominant, in which case those who now cannot tune-in
drop o ut.
Another aspect of the combinatorial potentials of neuronal elements can be
clarified in terms of the analogy to chemical combination. So far we have
assumed that neuronal elements combine as long as they are consonant. This
might seem to suggest that any number of consonant elements, once active,
would combine into a unified whole. However, it appears more appropriate
to conceptualize consonant elements as forming complementary sets, in the
same way that oxygen and hydrogen form a complementary combinatorial
set, when they combine into water. In other words, we assume that constellations of elements capable of simultaneous activity in the context of particular
interference patterns form complementary o r coherence sets such that in a
particular instance, if some elements are not yet active, the constellation will
remain incomplete. Once all the elements come to be active, the set (and the
macroactive structure) reaches closure. According to this view, irrelevant
elements are defined by their inability to join the active coherence set of the
moment when they get activated. Dissonant elements, on the other hand, can

IRAN,NEJAD AND OR TONY

join in, but unlike consonant elements, they tend to dissolve or disintegrate the
active pattern by disrupting the prevailing interference pattern. It should be
evident that irrelevance is a biofunctional relation that is different from
consonance or dissonance because it does not by itself affect the functioning
of macroactive patterns. It is possible, therefore, to consider the quality o f
activity in the nervous system as a dichotomous factor (consonance vs. disson,
ance) as opposed to a trichotomous one (consonance vs. dissonance vs.
irrelevance).

Simultaneous

and Independent Functioning

The notion of consonance provides a way of conceptualizing how constella,

tions of neuronal elements that are physically distributed across many brain
subsystems can combine via simultaneous (rhythmic) activity. However, if
consonant activity were restricted to simultaneous activity, the biofunctional
system would not work. Simultaneous activity tends to unify all consonant
elements in the mass action system into a global combination. As mentioned
earlier, in biofunctional combinations, like in chemical combinations, indi,
vidual components (i.e., elements or constellations localized within particular
subsystems) tend to lose their qualitative individualistic properties as they
become part of the larger combination. However, since the activity of the
individual components always occurs in the context of the mass action system
and never in isolation, it follows that these components should be unable to
manifest their individual qualitative properties. Therefore, a system allowing
only simultaneous activity would be capable of manifesting only global
subjective experiences but would be unable to manifest localized (or finer)
,qualitative experiences. It was perhaps for this reason that Bartlett (1932)
postulated that people must have a way of "turning round upon" their
schema,of,the,moment so as to individualize its components.
In order to deal with the problem of component individualization in the
context of the mass action system, we believe a second type of consonant
functioning must be postulated which we refer to as component independent
functioning. Independent functioning of a component of a larger combination
occurs if the component changes its rate of activity in relation to that of the
combination as a whole. When a component does this, it manifests its
individualistic qualitative properties. In terms of the light constellation anal,
ogy, when a constellation oflightbulbs is on, it generates a global pattern of
light. A bulb, or a subconstellation of bulbs, can be said to function independ,
ently if it becomes brighter or dimmer than the rest of the bulbs in the
constellation. At the time that the subconstellation is doing this, its character,
istic pattern of light (or color) becomes more evident.
It must be noted that any element or any consonant constellation of
elements in the mass action system is, in principle, capable of functioning

A BIOFUNCTIONAL MODEL

195

indep~ndently. It m~st also ~e noted that in a system of the type proposed

here ~i.e., one that 1s exclusively comprised of physically distributed and
funct10nally autonomous elements without containing all~purpose executo
or homunculi), independent functioning is the only possible mechanism r~
. componen: individualization. Thus, the present account, while claiming th:t
the system 1s homunculus,free, does not specify how individual components
come to function independently without a homunculus. However the bi'
functional mo~el does transform the homunculus problem into ilie more
concrete question of how components of the mass action system come to
function independently. Clearly, it is conceivable, in terms of the lightbulb
analogy, that in a constellation of burning bulbs a subconstellation of them
n:anifests its particular characteristics by growing momentarily brighter or
dimmer, :hat is, by functioning independently of the rest of the larger
constellation. More difficult is the question of why this should happen.
P~es~mably, the bulbs _do not change their b'tightness spontaneously-"at
will. They cannot manifest spontaneous initiation of activity. But organismic
~ubsystems appear to manifest initiation of spontaneous (or "willful") activ,
1ty. ln other words, organisms are somehow capable of exerting control over
~rganismic subsystems that comprise them (e.g., their limbs). The biofunc,
ttonal model implies that this control does not take place because of the
control exerted by some single all,purpose homunculus embedded some,
~here in the system that is somehow capable of controlling other qualitatively
diverse subsystems. Rather, spontaneous control is possible because of the
influence of multiple causes all of which, however, exert their influence in
terms of component in~e~~dent functioning. While this way of viewing
apparent~y s~ntaneous mttatJ.on of organismic activity concretizes the pro~
lem, our mtumons as to the cause of spontaneous independent functioning of
components add little to those of Bartlett. Bartlett ( 1932) believed that the
problem of component individualization was unanswerable at the time but he
insisted that it had to somehow occur. He also maintained that whatever form a
satisfactory answer to the problem turned out to take, subjective determina,
tion (i.e., awareness mediation) would have to be involved. One way in which
~wareness~mediated component independent functioning might work can be
illustrated by considering the tip-0f, the,tongue phenomenon. Perhaps this
phenomenon occurs when a person is implicitly aware of a particular comPD'
nent of the mass action system but is not sufficiently so to make it function
independent!Y and, thereby, explicit. There are also more automatic instances
~f component independent functioning. For example, independent function,
mg can occur in direct response to external energy patterns, as when one looks
at a flashing light or when one encounters surprising information.
In addition to individualization, component independent functioning
s~rves tw~ o.ther basic functions. First, it is an awareness~nhancing mecha,
msm. ThIS 1s, of course, a restatement of an earlier claim that when a

'196.'
. ' ,,..

IRAN-NEJAD AND OR TONY

component functions independently, it manifests its characteristic qualitative

properties. Secondly, independent functioning is an attention mechanism;
that is, an independently functioning component becomes the center of focal
attention for the duration that it is functioning independently. Thus, according to the functional theory, component individualization, awareness, and
attention are mediated by a single mechanism-component independent
functioning.
It is possible that the two types of consonant functioning postulated
here-independent and simultaneous-are responsible for the two types of
brain wave activity often observed in EEG records. One type of brain wave, the
synchronized slow electrical activity, is more evident when the cortex is
relatively idle. Since these slow electrical rhythms also occur during slowwave sleep, many investigators have concluded that synchronizing activity is
totally passive, that slow electrical activity is only epiphenomenal to the
activity of the brain, and that no active synchronization is involved in psychological functioning. On the other hand, psychological activity has been
assumed to occur when slow-waves are less evident and when desynchronized
activation becomes prominent (see, e.g., Jasper, 1981). In terms of the biofunctional theory, it may be argued that slow-wave synchronizing activation
occurs as a result of simultaneous functioning, and that desynchronized
electrical activity occurs as a consequence of component independent
functioning.

The Scherna..-of-the-Mornent: Principal Characteristics and Functions

..

We have now specified the biofunctional properties of neuronal microsystems, how they are physically distributed, how they intercommunicate, how
they generate psychological qualities, and how they engage in simultaneous or
independent functioning. We have characterized the neuronal system as a
dynamic mass action system consisting of a large population of specialized
neuronal elements which can combine in activity to form functional constellations. We have proposed that specialized neuronal microsystems, as elementary loci of subjective qualities, constitute the basis not only for distributed
(potential) memory, but also for distributed awareness and distributed attention. Simultaneous functioning was proposed as the mechanism for both
implicit and global awareness, as well as for broad attention. Component
independent functioning, on the other hand, was postulated as a mechanism
for focal awareness and focal attention.
With these concepts, it is now possible to present a rather explicit account
of the schema-of-the-moment, which is, loosely speaking, the subjective
counterpart of the activity in the mass action system. In this section we shall
discuss the main characteristics of the schema-of-the-moment. ln particular,
we will discuss the stability of different components of the schema-of-the-

A BlOFUNCTIONAL MODEL

197

moment and its overall continuity; we will argue that the organizing forces of
the schema-of-the-moment are content-based rather than structure-based
and we will try to specify the nature of changes that occur in the schema-of~
the-moment in response to incoming information. Since we believe that our
concept of the schema-of-the-moment is similar to that suggested by Bartlett
(1932), much of our discussion will involve elaborations or clarifications of
his ideas.
Stability and Continuity
The schema-of-the-moment is a constantly changing phenomenon involving both global and focal experiences. With respect to stability and change,
the totality of the schema-of-the-moment may be viewed as comprising three
theoretically distinguishable, but not actually separate, components. We will
refer to these as the background component of the schema-of-the-moment
(Background-SOM), the dominant component of the schema-of-the-moment
(Dominant-SOM), and the independently functioning component of the
schema-of-the-moment (Independently -Functioning-SOM). The BackgroundSOM is a slowly-functioning loosely-integrated component in which elements
with consonant, dissonant, and irrelevant functional properties can coexist. It
involves the major portion of the schema-of-the-moment and the majority of
the elements in the mass action system. Because of the slow rate of activity in
the Background-SOM, it remains the closest component to the microbiofunctional level. This is because at low levels of activity, there is less functional
integration and, consequently, the active elements will tend to preserve their
localized individualistic functional properties. This component is responsible
for the background or peripheral awareness of such things as time, space, self,
and various other "active" content domains. The Background~SOM is ordinarily a stable component of the schema-of-the-moment and most of its
elements maintain an activity rhythm that can last for hours, weeks, months,
or even years without undergoing significant change. Major shifts in this
component, however, do occur especially during landmark occasions such as
unusual personal successes or failures, personal tragedies, and, to some
extent, during less dramatic changes in normal life patterns such as travel and
vacations. More subtle changes in the Background-SOM occur as a function of
interaction with other components of the schema-of-the-moment. The
Background-SOM remains stable to the extent that its elements fail to participate, because of their irrelevance, in other components of the schema-of-themoment.
The second major component, the Dominant-SOM, results from simultaneous macroactivity in consonant elements of the moment. This compo. nent depends for stability on its incompleteness and, occasionally, on rehearsal. More specifically, an incomplete schema tends to remain dominant longer

~.

..r;.i,.
1

I .

,.

t.98

IRAN-NEJAD AND ORTONY

than a complete one, because an incomplete schema remains active through

development while a complete schema can remain dominant merely through
effortful rehearsal. A person is only globally aware of activity in the
Dominant-SOM and has only implicit awareness of its components. With
respect to the nature of ongoing activity, the Dominant-SOM may be either
resolt1ing or dissolving. A resolving Dominant-SOM consists of an incomplete
set of consonant elements and tends to remain active until closure is reached
through element enrichment (i.e., through activation of other consonant
members of the coherence set), after which time it can only remain dominant
through rehearsal. A dissolving Dominant-SOM consists of consonant and
dissonant elements and tends toward resolution through disintegration. A
dissolving Dominant-SOM may end up in total disintegration through what
might be called element shedding (i.e., the loss of active consonant elements).
Element shedding may also result in partial disintegration when "dissonanceinfected" consonant elements drop out until no such elements are involved, at
which time the remaining consonant elements may initiate a resolving
Dominant-SOM. Experientially, resolving Dominant-SOM activity manifests
itself as feelings of consistency, curiosity, suspense, understanding, interestingness, and closure. Dissolving Dominant-SOM activity, on the other hand,
gives rise to experiences of conflict, fear, anxiety, confusion, aversion, and
lack of closure.
The third major component is the independently functioning schema-ofthe-moment, the Independently-Functioning-SOM. This is the most transitory component of the schema-of-the-moment, since it soon joins either the
Dominant-SOM, if it is consonant or dissonant with it, or the BackgroundSOM, if it is irrelevant to the Dominant-SOM.
This tripartite characterization of the schema-of-the-moment is not meant
to suggest that the three components are actually distinct. First, the initial
creation of the Dominant-SOM occurs when some elements in the Background-SOM come to function independently, under the influence of external
stimulation, for instance. Subsequently, the Background-SOM serves as the
context for the development of the Dominant-SOM and as the only source of
element enrichment for it. Subsequent Dominant-SOM enrichment is
mediated by the activity of the Independently-Functioning-SOM. It occurs
when elements in the Background-SOM consonant with those in the
Dominant-SOM come to function independently, either as a result of changes
in the internal relational environment which is, in turn, caused by the activity
in Dominant-SOM. In this way, the three components of the schema-of-thernoment continue to interact and to create the constantly changing phenomenal experience of the moment. So, if the biofunctional model is correct, there
are no individual mental entities-there are no cognitive building blocks.
There is total continuity, not only with the immediate external or internal
context, but also in time, in space, and with respect to personal history. In

A BIOFUNCTIONAL MODEL

199

spite of this total continuity, it is often possible to single out particular

components of the schema-of-the-moment. But, even when focussing on a
single "distant" component, the continuity is never lost. A quick excursion to
a remote childhood experience does not destroy the experience of the
moment. It seems that it is always the past that "visits" the present (by getting
re-created when the conditions are suitable) and not the present that searches
for the past. Transitions are almost always smooth and continuous.

The Primacy of Content OtJer Structure

Even though we claim that there exist no long-term static structures in the
head, we still have to explain the origin of transient structures. Our explanation here is essentially the same as Bartlett's (1932), although Bartlett did not
make his theory very explicit. In the model we are proposing, neuronal
microsystems are the biofunctional generators of primitive psychological
qualities. At a slightly less biological level, Bartlett (1932) referred to these
primitive qualities as "images" (including what he called percepts, appetites,
instincts, ideals), and claimed that images are the basic ingredients of the
schema-of-the-moment. According to functional models of this sort, the only
type of structure that can exist is the structure of organized contentstructure cannot exist independently of content (Shanklin, 1981 ). Thus, if our
interpretation of Bartlett is correct, his functional schema theory is very
different from the kinds of structural schema theories that it has spawned.
Structural theories are based on the assumption that relatively content-free
abstract structures serve to organize content. Bartlett's theory, on the other
hand, seems to suggest that content possesses intrinsic organizing properties
that constant!y produce and reproduce structure thus eliminating the need to
postulate abstract organizing structures. Furthermore, while in his theory,
Bartlett stressed that schema bias, tir "determination by schemata" as he
called it, is a critical factor in cognitive functioning, he also insisted that there
is an even more potent bias, namely, the bias inherent in the qualitative
properties of specific content elements. Element bias is more potent because,
for instance, it makes it possible to skip directly to events that occurred in the
remote past despite the determinism of the current schema-of-the-moment:
In the experiments on perceiving, or imaging, and on all the various modes of recall,
while there was a sense in which subjects could accurate! y be said to have reacted to
whatever material was presented 'as a whole,' yet in that whole some special features
were invariably dominant. In many cases, when the material had to be dealt with at a
distance, as in remembering, the dominant features were the first to appear, either in
image form, or descriptively through the use oflanguage. In fact, this is one of the great
functions of images in mental life: to pick items out of 'schemata,' and to rid the
organism ofover-determination by the last preceding mern her of a given series. (Bartlett,
1932, p. 209)

IRAN-NEJAD AND OR TONY

Bartlett illust;:ated how the reappearance of some key content elements
enabled one of his subjects to remember a story after more than ten years.
Bartlett maintained that remembering begins with a global impression built
around a few dominant details from the original experience, an impression
which is primarily of the nature of affective quality. After the establishment of
this global impression comes the immediate return of other details that may
contain "some inventions and transformations, [but] seem clearly to be
derived from some of the events of the original story" {1932, p. 209). Bartlett
maintained that in any learning or remembering situation, the "dominant, or
over-weighted, elements [that] stand out from the rest ... together with their
determining tendencies, are apt to set the meaning of that situation" {p. 234 ).
Thus, according to Bartlett, the qualitative properties of a few content elements cause a global impression that sets the stage for the recall of past
experiences.
Similarly, the biofunctional model assumes that at the most primitive level
content elements are created by neuronal elements. When a subconstellation
of neuronal elements representative of those that were active at the time of the
original experience get activated (e.g., as a function of the stimuli pro~ided by
seeing and talking to an experimenter and by the recall probes provided), it
generates element-specific awareness patterns that combine to create the
global impresion. The elements also generate element-specific energy patterns
that combine to create the relational environment that existed at the time the
material was originally learned. The relational environment then sets the stage
for the activation of other consonant elements that enrich the global impression. The result is a schema-of-the-moment that approximates an earlier
experience.
The square-dancing analogy used earlier can illustrate how this might
happen. Recall that individual neuronal elements were likened to individual
square-dancers with the difference that neuronal elements not only served as
dancers, but, at the same time, as caller/ musicians. One can imagine how a
few caller/ musician/dancers might initiate a performance in a large crowd.
Soon the sound of their music pervades the air and more and more individuals
join in. Similarly, once they come to be active, a few neuronal elements that
participated in an earlier experience can re-create a relational environment (a
tune) uniquely representative of that experience.
Acquiring New Information; Combination OT Slot Filling

One of the most widely studied aspects of conventional schema theories is

the slot filling thesis. According to this, a schema is an abstract frame that
contains slots which are filled by incoming schema-related information. A
corollary of the slot filling thesis is that people only learn what they have
schemata for and ignore everything else (Neisser, 1976). The thesis has

A BIOFUNCTIONAL MODEL

201

trouble with the fact that people can and do remember incongruous informa~
tion (see Schallert, 1982; Thorndyke and Yekovich, 1980).
It must be acknowledged that there have been attempts to deal with the
processing and recall of incongruous information within conventional
schema theories (see, e.g., Schank, 1982, on expectation failures; Schank and
Abelso n, 1977). One approach, studied extensively by Graesser and his
associates (Graesser, 1981; Graesser, Gordon, and Sawyer, 1979; Graesser,
Woll, Kowalski, and Smith, 1980; Smith and Graesser, 1981; Woll and
Graesser, 1982) defines schema-relatedness in terms of typicality-the more
typical an item of information the more likely it is to be in the schema. To the
extent that an item is schema-atypical, it is to be considered u nrelated or
incongruous. In t his approach, an atypical item is recalled becai:;.s~ at the time
of learning it is indexed as such, that is, it "is encoded with a distinctive,
unique tag and stored as a separate unit" (Woll and Graesser, 1982, p. 290).
Even if salvaged through some kind of indexing scheme, the slot filling
thesis suffers, we think, from another problem related to the fact that it
implies that new information or content fills the slots provided by the schema
passively. However, if the biofunctional model is correct in claiming that
content elements possess their own functional properties, then these elements
must play an active combinatorial role. In fact, the claim that content elements
exert their own "active biases" was one of the recurrent themes in Bartlett's
theory. He argued that the "active biases" caused by new incoming information play a dominant role in the comprehension of both congruous and
incongruous information. Furthermore, Bartlett cautioned against a passive
slot filling interpretation of his theory:
The process is not merely a question of relating the newly presented material to o,ld
acquirements of knowledge. Primarily, it depends upon the active bias, or special
reaction tendencies, that are awakened in the observer by the new material, and it is these
tendencies which then set the new into relation to the o ld. To speak as if what is accepted
and given a place in mental life is always simply a question o f what fits into already
formed apperception systems is to miss the obvious point that the process offitting is an
active process. ( 1932, p. 85).

For Bartlett, therefore, incoming information does not passively fill slots that
are made available by the operative schema. Rather, it is the potential of the
new information to awaken qualitative "active biases" that sets "the new into
relation to the old." In other words, what is newly acquired actively combines
with what is old. The word active must be interpreted with caution here. It
means that what is "awakened" by the new information does not surrender
itself passively to the shackles of an active schema. Rather, the new information imposes "active biases" of its own that often override the biases imposed
by the schema-of-the-moment, as when incongruous information spontaneously draws attention away from the operative schema. Furthermore, if our
interpretation of Bartlett is correct, in his theory, and certainly in the biofunc~

IRAN-NEJAD AND OR TONY

:"~

.."

; ~ ,- ;
: . 1

'.!

. ..

tional model presented here, the potential for content elements to play an
active role exists after learning as much as it does at the time of learning. In
other words, being functionally autonomous, these elements do not remain
chained passively to a structure after they combine with it until that structure
is reactivated, any more than dancers freeze into a "solid" frame as soon as the
tune to which they are dancing stops. Being autonomous individuals, each
dancer can participate in a different activity in the meantime. Dynamic
combination is not a long-term bond. It is some sort of momentary cooperative activity (see Freeman, 1975), a cooperation to create something novel.
Bartlett's observations about the nature of learning and remembering are
completely compatible with those implied by the biofunctional model. The
biofunctional model explicitly rules out the preservation of static relations
and of abstract structures. The only option open, therefore, is to explain
remembering in terms of the functional properties of autonomo us neuronal
elements and not in terms of static mental relations. In the biofunctional
model, the only relarions that can be preserved are transient functional
relations-active, concrete, and particular functional relations-where active,
concrete, and particular mean on-going biological activity in a particular
organismic system. As Bartlett put, "what is essential to the whole notion" of
a schema is tha t it is "actively doing something all the time . .. [it is J carried
along with us, complete, though developing, from mo ment to moment"
( 1932, p. 201 ).
The term transient also needs some qualification. There is a sense in which
transient functional relations could last a long time, that is, if the activity
involved continues in the manner postulated by Bartlett and specified in this
paper. A square dancing session is inherently transient, but it could, in
principle, last for days, weeks, or even years. Therefore, if our interpretation
of Bartlett is correct, his theory was not based on the preservation of abstract
long-term relations underlying generic information, as has been suggested by
some authors (see, e.g., Woll and Graesser, 1982). On the contrary, he held
that every piece of generic or abstract information, or any other complex
mental structure, had to be re,created afresh based on the qualitative proper,
ties of active elements. What is permanent are the elements themselves (for
Bartlett "image-like" content elements and in our model neuronal microsystems ). This is probably why Bartlett emphasized the tendency of subjects to
preserve the concrete. For instance, he stated that

!_.:

[in folk-tales and) in other types of material, every general opinion, every argument,
every piece of reasoning, and every deduction, is speeqily transformed and then omitted.
The greatest efforts in this direction were achieved by subjects who reported a visual
method of recall, as if this method carries with it an inevitable bias towards the concrete.
The tendency observable in several instances for a narrative, a description, or an
argument to take on a personal form, seems to be due in part to the same factor , .. . It
may, at first, seem that the mass of folk-proverbs which are traditionally preserved
among every people contradicts the tendency toward the concrete. But the strength of

A BIOFU NCTIONAL MODEL

203

the folk-proverbs lies in its applicability to the individual instances As am

1

ld h

ere genera 1ty

it never wou
ave been preserved and, except in a literary sense it is practically
used.(!932,pp. 172-173)
'
never

Bartlett maintained that acquisition of new information involves tw b

F'
h
o as1c
fu nct1ons
. .1rst, t ere is an immediate physiological function made possible
b y the reaction of a sensory mechanism to external stimuli. This he belie d
. I d 1 . . l
'
ve '
is a re.a Y~~ ect1ve; its se ectivity is determined by the qualitative properties of
the stimuli mvolved, and it approximates what is generally meant b h

d
Y earmg,
seem~, an so on. The second function has to do with the reaction of the
orgamsm a~ a whol~ to the immediate physiological pattern ofactivity. This is
also sel~ct1ve but tts selectivity is made possible, not by some localized
mechanism, but by the global qualitative properties of the active mass of the
~om~nt. This, Bartlett maintained, approximates what is generally called
hst~nmg as opposed to hearing, or looking as opposed to seeing. Bartlett
believe~ that ~he type of selectivity that is directly based on "qualitative
factors is dominant over any other type in all the higher mental processess"
(1932, p. 190). This ~electivir: makes it possible to gather, from among
elements present both m the active sensory pattern or in the active mass of the
moment, those elements that are "most relevant to the needs of the moment"
a~d so to construct an updated schema. He maintained that construction is
either spontaneous and immediate, or that it is mediated by what he called
effort after meaning, effort to relate "what is given to something else" or to
understand what is not immediately obvious.
Bartl~tt's th~ory can be readily specified in biofunctional terms. According
~o the b1ofunct1onal model, sensory stimulation causes independent functioning of a constellati~n .of neuronal elements and creates a momentary
Independently,Funct1onmg-SOM which then interacts with the DominantSOM in the f?llowi~g fashion: If the Independently-Functioning-SOM, or a
subconstellan~n of its elements, is consonant with (but not necessarily typical
of) the Dommant-SOM, it will combine with it. Those elements in the
Independently-Functioning-SOM that are irrelevant to the Dominant-SOM
~eco~e p.art of.the Background-SOM, even if they happen to be typical of the
situation m which the Dominant~SOM is active (e.g., the waitress serving in a
restaurant h~s brown hair). If the Independently-Functioning-SOM, or a
subconstella~on of its elements, is dissonant with (but not necessarily atypical
o~ the Dommant-SOM (e.g., long waiting lines are annoying in restaurants), it
will cause a temporary state of dissolution in the Dominant-SOM by causing a
perturbation in the internal relational environment and by changing the
nature of the energy p attern of the moment from signal to noise. Dissonance is
resolved if the Dominant-SOM undergoes spontaneous element enrichment
~r eleme~t shedding. The diners may be relieved to see an acquaintance in the
ltne who 1~ fun to talk to while waiting, or they may give up waiting and go to
an otherwise less preferred restaurant. If resolution cannot occur, the disso-

_,-

'. 204

IRAN-NEJAD AND ORTONY

nant Independently-Functioning-SOM becomes part of the BackgroundSOM. The diners might decide to wait, move their thoughts to a different
topic, but, at the fringe of their awareness, they might still remain troubled by
the long wait. Resolution or lack of it is caused by localized element bias
effects and by global schema bias effects, both of which are caused by the
functional properties of elements and both of which together manifest themselves in terms of effort after meaning.
According to the biofunctional model, therefore, to the extent that the
Independently-Functioning-SOM combines with the Dominant-SOM, it will
lose its distinctive qualitative characteristics, just as oxygen and hydrogen lose
their combustible properties when they combine to form water. This is how
the combination hypothesis explains the fact that in recognition memory new
items can be difficult to discriminate from similar old items. More specifically, it is impossible to discriminate already integrated old information from
new information that differs from it only by distinctive properties that are lost
as a result of the act of combination.

Sources of Functional Initiation

:';

:.(

The assumption that long-term static structures do not exist, and the
complementary claim that mental relations are established only after neuronal
elements are already active, raises the problem of how neuronal elements
come to be in a state of functioning to begin with. This problem seems
particularly urgent in relation to remembering. If mental structures are transient, how can people remember anything? How can they recall together what
they have learned together if they have not stored it together? That these
questions appear to be so challenging seems to us to be a reflection of the
deep-seatedness of the permanent~storage metaphor, which also seems to be
responsible for widespread rejections or misinterpretations of Bartlett's
reconstructive theory of remembering.
Bartlett (1932) rejected the long-term storage metaphor and proposed that
remembering is reconstructive or re-creative. In support of his claims, he
showed that recall is often inaccurate. Some researchers (e.g., Zangwill, 1972)
have treated reconstruction as if it were equivalent to inaccuracy in recall and
have considered the fact that recall is often accurate as evidence against
Bartlett's theory. Although Spiro ( 1977) argued against this interpretation of
the notion of reconstruction, authors continue to fail to distinguish between
reconstruction and the mere occurrence of recall errors. For instance, a recent
review of research on schema theory concluded that "the consensus is that
reconstruction [i.e., as evidence by the incidence of recall errors] is quite rare
and occurs only under special circumstances" (Alba and Hasher, 1983).
Several other researchers, on the other hand, have followed Bartlett, as we
have, in calling into question the long-term storage metaphor and in maintain-

A BIOFUNCTIONAL MODEL

205

ing that re~embering is re-creative (see, e.g., Bransford, McCarrell, et al.,

1977; Jenkms, 1977). However, the issue of how accurate recall is po 'bl
.
nl
. fun
SS! e
given o Y transient
ctional relations has yet to be resolved.
. Acc~rding the biofunctional model, in order to demonstrate how recall
IS possible. without long-term storage of static structures, the problem of
remembering must be considered in terms of two separate problems, namel
the problem of specifying the sources of functional initiation in neuron~
e.leme~~ .an.cl the problem of specifying what happens following such functional 1mnanon.

:o

The causes of initiation of (or changes in) activity in elements can only be
understood by rec~gnizing that the nervous system is a multiple-source
dependent system with respect to functional initiation. First, there are endogenous sources of functional initiation that arise within the organism. Endo~
g~nous sources m~y be biowgical, biofuncticnal, or mental. That hungry individuals are more likely to seek food has perhaps more to do with biol I
. . . . th
og1ca
sour~es o f mltlanon an with other endogenous sources. While we cannot
s~cify. the re~ative contribution of biological factors and their interaction
with b1ofuncn?n~l, psychological, or environmental sources, the fact that
such factors ex15t 1s well.-established (see e g Colquhoun 1971) Th

b' fu
. .. .
' "

e maier
10 nct1~nal source of minanon of functioning, according to the present
model, might be called the combinatorial source. As elements combine they
create novel energy {or signal) patterns that set the stage for the initiadon of
~nctionin~ in other elements through the establishment of emergent functl~~al relanons. There are also more subtle biofunctional sources that play a
critical ~ol~. A large .number of neuronal elements in the Background,SOM
are specialued to maintain a particular biofunctional rhythm or cycle. Endogenous sour.ces responsible for awakening organisms from sleep might be
largely of thIS type. The main p~howgical source of functional initiation in
neuronal elements is assumed to be the Dominant-SOM. How the Dominant,
SOM acts as a source of initiation of functioning, or whether it is the only
component of the schema-of-the-moment through which the mind influences
the acti~ity of the brain, is a question that we cannot yet answer.
The final but perhaps the most important source of functional initiation as
far as the stability and the development of the schema-of-the-moment are
concerned, are exogenous sources-those external energy patterns (or signals)
that constantly influence the neuronal system through several independent
sense organs. It seems as though nature has found it profitable to relate
?rganisms to the world through more than one sense organ, each serving as an
independent source of functional initiation.
The multiplicity of sources of functional initiation means that after initiation. of activity, the development of the schema-of,the-moment is not a
~tra1g~tforwar~ c,ombina:ion of functioning elements. Rather, "the complex,
lty of schematic formation means that many objects, many stimuli, many

. 206

IRAN-NEJAD AND ORTONY

reactions, get organized simultaneously into different cross-streams of o:g~n

ized influences" (Bartlett, 1932, p. 302). Thus, the qualitative charactensncs
of autonomous elements play a vital role in the re-creation of past experiences
and in the creation of new ones. Consequently, after functional initiation,
there is a more critical phase in the development of the schema-of-themoment that must be taken into account. According to the biofunctional
model, the nature of activity in this phase is solely determined by the
qualitative functional properties of neuronal elements, both at a global level
(as manifested in schema bias) and at a local level (manifested in terms of
element biases). Earlier, we used the term enrichment to refer to the development toward closure of the Dominant-SOM. However, since multiple-source
functioning means activation of dissonant elements, activity in the DominantSOM during the enrichment phase, might also be viewed as an act of problemsolving. According to this view, post-initiation enrichment is guided by two
basic types of subjective qualities, which are determined by dissonant a.nd
consonant biofunctional properties of active neuronal elements and which
tend to manifest themselves in terms of what might be called problem recogni
rion and resolution recognition capacities. If the biofunctional theory is correct,
problem-solving during recall, and problem-solving in general, must operate
toward justification of these two types of subjective qualities. It is this
problem-solving nature of remembering that renders recall inaccurate ~r
accurate as far as "the actual facts of the learning situation," as Bartlett put 1t,
are con~erned, since post-initiation problem-solving makes rememberi~g
totally dependent on the Dominant-SOM at the time of recall. Accuracy m
recall is determined by the degree to which the actual facts of the recall
situation, especially those that serve as the sources of functional initiatio n,
approximate those of the learning situation.

Conclusion
We have attempted to sketch a model of the mind that we hope is compati
ble with what is known about the brain and the nervous system. Our primary
goal has been to address questions relevant to PSJC~Wgy. as op.pos~d to
artificial intetligence. We have tried to show how cognltlon 1s possible m an
animate system having the kind of biological constraints that humans have,
rather than how cognition might be possible in some more abstract "systemindependent" manner. This choice was made because we believe that the
nature of human cognition and experience is necessarily determined by the
way in which the individual components of the system function.
A second goal was to bridge the gap between cognitive psychology and the
neurosciences. To the extent that we have succeeded, the result is a model that
strictly speaking does not conform to the standards of either neuroscientific
models or of psychological ones. We have drawn upon what we judged to be

A BIOFUNCTIONAL MODEL

207

relevant literature in both the neurosciences and cognitive psychology. G iven

the diversity of this literature, and given our own particular goals, it is likely
that some of the authors of the ideas we have used will find our approach
difficult to accept. In employing the proposals of others, we have taken care to
specify clearly those aspects of their work we have found attractive. Thus, for
example, our endorsement of Sperry's proposals about consciousness in no
way entails a commitment to his general philosophy.
There are doubtless many problems with the model we have proposed, and
perhaps with the way we have presented it. We hope that these problems are
no more serious than those facing conventional models. On the positive side
we think that a model of the type we have presented might be able to provide~
solution to some of the more complex philosophical problems having to do
with mental representations discussed, for example, by Dennett (1983). Like
Dennett, the central claim we have made is that it is neither necessary nor is it
ultimately fruitful to conceive of knowledge repre8entations a8 stored abstractions upon which various kinds of cognitive processes operate.
In a paper of this kind, it is not possible to do all that one would like. We
have resisted trying to propose detailed accounts of the huge range of aspects
of mental life-each would take a book. We have also not discussed the
empirical consequences of the view we have proposed because that would
have necessitated just such detailed discussions of the individual aspects of
cognition. We have preferred to present an impressionistic sketch of our
account from which, hopefully, the big picture emerges even if some of the
details are absent or do not accurately portray the way things really are.

References
Adams,J .A. ( 1977). Feedback theory of how joint receptors regulate the timing and positioning
of a limb. Prychologic.a! Review, 84, 504-523.
Alba, J:W.. and Hasher, L. (1983). Is memory schematic? P.rychologic.al Bulletin, 93, 203-23 1.
Attard1, D.G., and Sperry, R.W. (1960). Central routes taken by regenerating optic fibers.
PhJsiologisc, 3, 12.
Attardi, D.G., and Sperry, R. W. ( 1963 ). Preferential selection ofcentral pathways by regenerating optic fibers. Experimental Neurology, 7, 46-64.
Barlow, H.B. ( 1969). Pattern rccongition and responses of sensory neurons. Annals of the New
York Aaulemy of Science, 156, 872.
Bartlett, F.C. (1932). Remembering: A sooh in experimental and social prychology. Cambridge,
England: Cambridge University Press.
Berlin, B., and Kay, P. (1969). Basic color terms: Their uni11eTsaliry and evolution. Berkeley:
University of California Press.
Bransford, J.D., McCarrell, N.S., Franks, J.J., and Nitsch, K.E. (1977). Toward unexplaining
memory. In R.E. Shaw and J.D. Bransford (Eds.), Perceiving, acting, and knowledge: Tou.oard
an ecologic.al psychology (pp. 431-'166). Hillsdale, New Jersey: Erlbaum.
Bransford, J .D., Nitsch, K.E., and Franks,J.J. ( 1977). Schooling and the facilitation of knowing.
In R.C. Anderson, R.J. Spiro, and W.E. Montague (Eds.), Schooling and the acquisition of
knowledge (pp. 31-55). Hillsdale, New Jersey: Erlbaum.
Colquhoun, W.P. (1971). Biologic.al rhythms and human perfmmance. New York: Academic
Press.

208

D ennett, D. ( 1983, March). St;tle.s of menial represencation. Paper presented at the University of
Illinois, Urbana.Champ aign.
Donchin, E. (1981). Surprise! ... Surprise? Ps,chophysiowg:y, 18, 493-513.
Eccles, J.C; (1953 ). The neurophysiological basis of mind: The p-rinciple.s of neurophysiolog,. Oxford:
Clarendon.
Edelman, G.M. (1978). Group selection and phasic reentrant signaling: A theory of higher
brain function. In G .M. Edelman and V.B. Mountcasde (Eds.), The mindful brain (pp.
51-100). Cambridge: MIT Press.
Eliashberg, V. (1981). The concept of E-Machine: On brain hardware and the algorithms of
thinking. In Proceedings of the Third Annual O:mference of the Cogniti11e Science Societ;t (pp.
289-291 ). Berkeley.
Fahlman, S.E. (1981 ). Representing implicit knowledge. In G.H. Hinton and J.A. Anderson
(Eds.), Parallel models of associative memory (pp. 145- l 59). Hillsdale, New Jersey: Erlbaum.
Feldman,] .A. ( 1979 ). A dislTibuied information (J'roeeJsing model of 11~1 memory (Tech. Rep. No.
52 ). Rochester: U niversity of Rochester, Computer Science Department.
Feldman, J.A. ( 1981 ). A connectionist model of visual memo ry. In G.H. Hinton and J.A.
Anderson (Eds.), Parallel models of associative memory (pp. 49-81 ). Hillsdale, New Jersey:
Erlbaum.
Fessard, A.E. ( 1954 ). Mechanisms of nervous integration and conscious experience. In E.D.
Adrian, F. Bremer, and H. Jasper (Eds.), Brain mechanisms and coruciousne.si (pp. 200-236) . .
Springfield, Illinois: Thomas.
Festinger, L. (1957). A theory of cognitive dissonanc.e. Stanford, California: Stanford University
Press.
Freeman, W. (1975). Mass action in the nen.oou.s rystem. New York: Academic Press.
Gallistel, C.R. (1980). The organization of action: A new S)'llthesis. Hillsdale, New Jersey:
Erlbaum.
Graesser, A.C. (1981). Prose romp-rehension beyond the word. New York: Springer-Verlag.
Graesser, A .C., G ordon, S.E., and Sawyer, J.D. ( 1979). Recognition memory for typical and
atypical actions in scripted activities: Tests of a script pointer + tag hypothesis. journal of

Verbal Leaming and Verbal Behovior, z8, 319-332.

G raesser, A.C., Woll, S.B., Kowalski, D.J., and Smith, D .A. (1980). Memory for typical and
atypical actions in scripted activities. journal of Experimental Psychology, 6, 503-515.
Grossberg, S. (1982). Studies of mind and brain. Boston: D. Reidel.
Head, H. ( 1920). Studies in neurology (Vols. 1-2 ). London: Frowde.
Hinton, G.H. (1981). Implementing semantic netwo rks in parallel hardware. In O.H. Hinto n
and J.A. Anderson (Eds.), Parallel models of associative memory (pp. 161-187). Hillsdale, New
Jersey: Erlbaum.
Hin ton, G .H., and Anderson, J .A. (Eds.). ( 1981 ). Parallel models ofassociative memroy. Hillsdale,
New Jersey: Erlbaum.
Hubel, D.H., and Wiesel, T.N. ( 1959). Receptive fields of single neurons in the cat's striate
cortex. Journal of Ph"Ysiology (London), r48, 574-591.
Hubel, D.H., and Wiesel, T.N. (1962). Receptive fields, binocular Interaction and functional
architecture in the cat's visual cortex. Journal of Ph:isiology (London), r6o, 106-154.
Hubel, D.H., and Wiesel, T.N. (1965). Receptive field s and functio nal architecture in two
non-striate visual areas ( 18 and 19) of the cat. Journal of Neuroph'Ysiology, 28, 229-289.
Hubel, D.H., and Wiesel, T.N. (1980). Brain mechanisms of vision. In Readings from Scientific
American: Mind and behatrior (pp. 32-44). San Francisco: Freeman.
Imbert, M., and Buisseret, Y. ( 197 5 ). Receptive field characteristics and plastic properties of
visual cortical cells in kittens reared with or without visual experience. Experimental Brain

Research,

A BIOFUNCTIONAL MODEL

IRAN-NEJAD AND OR TONY

22,

2-36.

James, W. ( 1890). Principles of psycholog-y. New York: Ho lt.

Jasper, H.H. ( 1981 ). Problems of relating cellular or modular specificity to cognitive functions:
Importance of state-dependent reactions. In F.O. Schmitt, F.G. Worden, G. Adelman, and
S.G. Dennis (Eds.), Proceedings of a Neuroscience.s Research Program Colloquim: The organiza
tion of the cerebral rorrex (pp. 375-393 ). Cambridge: MIT Press.
Jenkins, J.J. ( 1977). Remember that old theory of memory? Well, forger it. In R .E. Shaw and
J.D. Bransford (Eds.), Perceiving, acting, and knowledge: Toward an ecological prychology (pp.
413-429). Hillsdale, New Jersey: Erlbaum.

209

Jenkins,}.~. (198 1). C?an ":'e have a fruitful cognitive psychology? lnJ.H. Flowers (Ed.), Nebraska
symposium on mot1mnon, 198<> (pp. 211-238). Lincoln: U niversity of Nebraska p
John, E.R. (1967). Mechanisms of memory. New York: Academic Press.
ress.
John, E.R. ( 1972). Switchboard versus statistical theories oflearningand memo ,,_, __ c
850-864.
ry. "'-== 177,
John, E.R., and Killam, K.F. ( 1960). Electrophysiological correlates of avoidance co d
in the cat. Journal of Nervous and Mental Disease, 131, 183.
n ltlonmg
John, E.~." and Schwartz, E.L. (1978). The neurophysiology of information process
d
cognition. Annual Review of Psycholog:y, ~ 9 1-29.
mg an
Katchalsk~, A.K., R~wland, V., and Blumenthal, R. (1974). D)onamic pauerns of hr ll
assemblies. Cambridge: MIT Press.
am ce
Kay, P. (1981). Color perception and the meaning of color words. In Proceedings of the Th ' d
Annual O:mferenc: of the Cognitiue Science Sociecy (pp. 289-291 ). Berkeley.
tr
Kay, P., and M cDaniel, C.K. ( 1978). The linguistic significance of the m eanings of basic ol
terms. Language, 54 , 610-646.
c or
Klanky, R .L. (1975). Human memory. San Francisco: Freeman.
Kugler ._P.~., ~elso, J .A .S., and Turvey, M.T. ( 1980). O n the concept of coordinate structures
as d1ss1panve structures. In G.E. Stelmach and J. Requin (Eds.), Tutorials in mocor behavior
(pp. 3-47). Amsterdam: North-Holland.
Kuhn, T.S. ( 1970). The stnlaUre of scientific rewludons. Chieago: University of Chicago Press
Lashley, ~.S. (_195~). ln scare? of ~he e~gram. In S:ymposia ofthe Societ;t far Experimental Biolo
No. 4 Phys101ogu.al mechan'.'57'15 m anun~I behavior (pp. 454-483). New York: Academic Pr~'.
Lyn.ch, J.C. ( 19~0). The functional o rganuation of posterior parietal association cortex Behav.

ioral and Bram Sciences, 3 , 485-534.

Maturana, H.R. ( 1978 ). Biology oflanguage: The epistemology ofreality. ln G.A. Miller and E.
Lenneberg (Eds.), Prychology and biology of language and thought: Essays in honor of Eric
Lermeberg (pp. 27-63). New York: Academic Press.
Meyer, R.L., and Sperry, R.W. ( 19?6). Re~n.o~ectal specificity: Chemoaffinity theory. In G.
Gottlieb (Ed.), Neural and behavioral sflificil':J: Studies on che development of behavior and che
. nervous S)'srem (Vol. 3, pp. 111149). New York: Academic Press.
Miller, J. (1978). The bod'Y in question. New York: Random House
Mi~ky, M. (1980). K-Hnes: A theory of memory. Cognitiw ~. 4 , 117-133.
N~tsser, U . ( 1976). Oignition and realiry. San Francisco: Freeman.
N1cholso~ , C. ( 1979). Brain-cell microenvironment as a communication channel. In F.O.
Schmitt and F.G. Worden (~ds.), The neurosciences (pp. 457-476). Oimbridge: MIT Press.
Norman,.D.A. ( 1980). Twelve 1SSues for cognitive science. Cognitiue Science
I 32
o:Keefe, J., and Nadel, L. (1978). The hippocampus as a cognitive map. Oxfo.rd; Clar~ndon.
0 Keefe,]., and Nadel, L. (1979). Precis of O'Keefe and Nadel's "The hi'ppoca

,, D - L-.c
I
mpus as a
cogmnve map. oc:n<wiora and Brain Sciences, 2, 487-533.
Olton, D._s., Becker, J !.,and Handelman, 0.E. ( 1979). Hippocampus, space, and memory. The

Behwioral and Bram Sciences,

2,

313-365.

Piaget, J ( 197.0). S1TUCtUralism (C. Maschler, Trans.). New Yo rk: Harper and Row. (Original
wo rk pubhshed 1968)
Pribram, K.H. (19~.1). The brain, the telephone, the thermostat, the computer, and the
hologram. Cognmon and Brain Thehry, 4, 105-119.
Prigogine, I., and Nicholis, G. ( 1971 ). Biological order structure and instabilities Quarterly

Review of Biophysics, 4 , 107148.

'
.
Dy~cs, ~.D. (1978). Sensory cortex and the mind-brain problem. The
Behooimal and Bram Sciences, 3, 337-375.
Pyl~shyn, Z.W. (1973). What the mind 's eye tells the mind's brain: A critique of mental
lmagery. Prychol.ogical Bulletin, Bo, 1-24.
Rose, S. ( 1976). The conscious brain. New York: Vintage Books.
Schallert, D.L. ( 1982 ). The significance of knowledge: A synthesis of research related to schema
theory. I_n W. Otto and S. White (Eds.), Reading ex[>osicory material (pp. 13-48). New York:
Puccetti, .R .. and

A cademic Press.
Schank, R.C. ( 1982 ). D)onamic memory: A thehryof reminding and learning in computers and people.
London: Cambridge.

IRAN-NEJAD AND ORTONY

Schank, R.C., and Abelson, R.P. ( 1977). Saipu, plans, goals, and undersianding: An inquiry into
human knowledge stTUCtUres. Hillsdale, New Jersey: Erlbaum.
Schmitt, F.O. ( 1978). Introduction. In G.M. Edelman and V.B. Mountcastle (Eds.), The
mindful brain (pp. 1.Q). Cambridge: MIT Press.
Schmitt, F.O., and Samson, F.E. ( 1970). Brain cell membranes and their microenvironment. In
F.O. Schmitt, T. Melnechuk, O.C. Quarton, and G. Adelman (Eds.), Neurosci.enceHesearch
S)'ITlposium summaries (Vol. 4, pp. 191-325). Cambridge: MIT Press.
Selversron, A.I. ( 1980). Are central pattern generators undersrandable? The Beha.Jioral and

Brain Scienas, 3 535-571.

( ,

Shanklin, N.K.L. ( 1981 ). Relating reading and writing: Developing a transactional theory of the
writing process. Monographs in language and reading srudies (3, US ISSN 0193-4740).
Bloomington: Indiana University, School of Education.
Sherrington, C. { 1947 ). The inregrntfoe action of the ner'JOUS rystem. New Haven: Yale University
Press.
Smith, D.A., and Graesser, A.C. ( 1981 ). Memory for actions in scripted activities as a function
of typicality, retention interval, and retrieval task. Memory and Cogni.cion, 9, 550-559.
Sperry, R.W. ( 1943). Visuomotor coordim1tlon in the newt ( triturus viridesccns) after regeneration of the optic nerve. ]0!4mal of Comparative Neurology, 79, 33-55.
Sperry, R. W. ( 1952 ). Neurology and the mind-brain problem. American Scientist, 40, 291-312.
Sperry, R.W. ( 1966). Selective communication in nerve nets: Impulse specificity vs. connection
specificity. ln F.O. Schmitt and T. Melnechuk, (Eds.), Neuroscienas re.search S)'!llposium
summaries (Vol. 1, pp. 213-219). Cambridge: MIT Press.
Sperry, R. W. ( 1969). A modified concept of consciousness. Psychological Retiiew, 761 532-536.
Sperry, R.W. (1976). Mental phenomena as causal determinants in brain function. In G.G.
Globus, G. Maxwell, and I. Savodnik (Eds.), Consciousness and the brain (pp. 163-177 ). New
York: Plenum Press.
Sperry, R. W. ( 1977). Forebrain commissurotomy and conscious awareness. ]0!4mal of Mediail

Philosophy,

..ij
,};.
1:
;

2,

101-126.

Sperry, R.W. ( 1978). Mentalist Monism: Consciousness as a causal emergent of brain process.
The Behavioral and Brain Sciences, 3, 365-366.
Spiro, R.J. ( 1977). Remembering information from text: The "state of schema" approach. In
R.C. Anderson, R.J. Spiro, and W.E. Montague (Eds.), Schooling and the acquisition of
knowledge (pp. 137-165). Hillsdale, New Jersey: Erlbaum.
Szentagothai, J. (1978). A false alternative. The Behavioral and Brain Sciences, 3, 367-368.
Thatcher, R .W., and John, E.R. (1977). FOl4ndation.i of cognitive proce.sses. New York: Hillsdale.
Thorndyke, P.W., and Yekovich, F.R. ( 1980). A critique of schema-based theories of human
story memory. Poetics, 9, 23-49.
Utral, W.R. ( 1978). The psychobio!og)' of mind. New York: Wiley.
Wall, P.D. (1966). Functional specificity. In F.O. Schmitt and T. Melnechuk (Eds.), Neuro.seiences research S)'!llposium summaries (Vol. 1, pp. 229-230). Cambridge: MIT Press.
Weiss, P. ( 1936). Selectivity controlling the central-peripheral relations in the nervous system.

Biologic.al Review,

11,

494-531.

Weiss, P.A. ( 1966). Specificity in neurosciences. In F.O. Schmitt and T. Melnechuk (Eds.),
Neurosciences research rymposium summaries (Vol. 1, pp. 179-212 ). Cambridge: MIT Press.
Wess, 0., and Roder, U. ( 1977). A holographic model for associative memory chains.
Biological Cybernetics, 27, 89-98.
White, B.W., Saunders, F.A., Scadden, L., Bach-Y-Rita, P., and Collins, C.C. (1970). Seeing
with the skin. Perception and Psychophysics, 7, 23-27.
Wiersma, C.A. , and Ikeda, K. ( 1964 ). lnterneucons commandingswimmeret movements in the
crayfish. Comparative Biochemistry and Ph:tsiology, 1 2 , 509-525.
Woll, S.B., and Graesser, A.C. (1982). Memory discrimination for information typical or
atypical of person schemara. Social Cognition, 1 , 287-310.
Zangwill, O .L. ( 1972 ). Remembering revisited. Quarterly Journal of Experimental Psychology, 24,

123-138.