VOL. 62, No.

5 SEPTEMBER, 1955

THE PSYCHOLOGICAL REVIEW

PRINCIPLES OF PERFORMANCE
EDWARD C. TOLMAN
University of California

Apparently it behooves a so-called Secondly, I assume three types of

"cognitive theorist" such as myself to intervening variables and intervening-
explain how his assumed intervening variable systems: (a) discriminations
cognitive and motivational variables within a discrimination system; (b) be-
ever issue into actual behavior. I must, liefs and expectancies within a belief-
it seems, provide some rules or princi- expectancy system; and (c) drives and
ples of performance according to which incentive-values within a drive-incen-
these variables—namely, discrimina- tive-value system. The three interven-
tions, expectancies, drives, incentive ing variable systems are inferred sets
values—add to or multiply with one an- of dispositions (in the sense of pro-
other (or whatever the appropriate com- pensities of the organism to perform in
bining operations may be) to produce certain prescribable ways under speci-
behavior. Otherwise, as Guthrie (5) fied test conditions). And I assume
has suggested, I seem to leave my ani- that the momentary activated discrimi-
mal (whether rat or human being) nations, beliefs, expectations, drives,
merely "buried in thought" (though and incentive values are determined by
why this should be considered so shock- the momentary values of the presented
ing in the case of a human being I have independent variables in the contexts of
never quite been able to fathom). these three innate and acquired dispo-
Let me begin by summarizing briefly sitional systems.2 Let me elaborate.
my basic assumptions. First, I assume 1. The Discrimination System. This
four distinctive types of independent is to be specified by (a) categorizing or
variables: (a) present stimulus-units; nominal dimensions, that is, types of
(b) previous presentations of stimulus-
general object-classification, of which
unit-response-stimulus-unit sequences,
the given organism has become capable
abbreviated as past (S-R-S)'s; (c) (e.g., most human beings have acquired
present drive disturbances and currently
2
presented incentive objects; and (d) 1 wish to express my great indebtedness to
genetic or early training differences be- Professor Leo Postman for his unstinting and
crucial help in the development of these con-
tween individual subjects (as parame- cepts concerning the independent and the in-
ters affecting the results from the other tervening variables and variable systems. The
three types of independent variable). performance principles which I shall propose
later in this article, though in part Professor
1
Presented, with some modification, as Postman's, are at present largely mine—at
presidential address to the Division of General least insofar as the pictorial language is con-
Psychology of the American Psychological As- cerned. Postman does not find this language
sociation, Sept. 7, 1953. helpful.
315
316 EDWARD C. TOLMAN

such categories as colors, pitches, smells, possession by the organism of such a

chairs, tables, trees, other human be- set I would call an expectancy disposi-
ings, etc., etc.)) and by (b) the inten- tion. An "expectancy disposition" is
sitive dimensions along which the given thus a small acquired disposition of
organism has, in addition, become ca- greater or less certainty and perma-
pable of making intensity discriminations nence, which will tend to lead to an
(e.g., specific hues, heights of pitches, activated expectation whenever an in-
intensity of odors, comfortableness of stance of the given sign stimulus-unit
chairs, sturdiness of tables, shadiness of is presented. Such an expectancy dis-
trees, likeableness of human beings, position will also tend to be activated
etc.). Such categorizing and intensi- by other stimulus-units which are near
tive dimensions are uncovered in stand- the sign stimulus on some common di-
ard situations in which the organism mension of generalization. Thus, there
is presented with selected arrays of may be groups or families of expectancy
stimulus-units and which it is forced dispositions in the sense that an array
(or tricked) into categorizing and/or of them related by different signs may
discriminating intensitively. In other all have attached to them similar ex-
words, the discrimination system is to pectancy dispositions as to the type of
be discovered by the methods of a com- significate to be reached, if a certain
prehensive and up-to-date, as well as general type of means-manipulatory re-
a mathematically sophisticated, psycho- sponse be carried out. Such a group or
physics. Knowing this discrimination envelope of related expectancy disposi-
system, we can thereafter state the spe- tions I shall call a "belief disposition."
cific categorizations and intensity dis- An individual's expectancy system is
criminations which are likely to be acti- constituted by the sum of all his ex-
vated when such and such stimulus- pectancy and belief dispositions.3
units are actually presented on any 3. The Drive-Incentive-Value System.
given occasion. This system is to be conceived as a col-
2. The Belief-Expectancy System. lection of dispositions to approach and/
This is to be specified by the sum of or consume specific types of positive in-
the organism's expectancy dispositions. centive objects or to escape from and/or
For example, a subject may show that avoid specific types of negative incen-
upon the sounding of a given tone, Si, tive objects. These approach (consum-
he is prepared, provided he waits, RI, mation) and avoidance (and/or escape)
in a given place, for the coming of a dispositions depend upon the interaction
shock, S2. Or again, he may show that of two distinguishable independent vari-
when presented by the choice point of ables: namely, (a) types and amounts
a maze, Si, he is prepared for a result- of drive-disturbance, and (b) types of
ant food, S2, provided he turns, RI, to stimulus-units presented as incentive ob-
the right. Or he may show that upon jects.
the presentation of a first syllable, Si, a. Drive-disturbances would be: hours
on a memory drum he is prepared, pro- of specific food-deprivation, hours of
vided he continues to look at the drum, sex-deprivation in males, time-position
R2, for the coming of a second syllable, of the female in the oestrus cycle, per-
S2. Any one such set in the animal, 3
which (to distinguish it from the objec- I would like to emphasize that my notions
tive occurrences) may be symbolized as about expectancies have been clarified by the
recent discussion by MacCorquodale and
an (Si-.Ri-Si), I would designate as an Meehl. I wish here to acknowledge my very
activated expectation. The continued great indebtedness to them (13).
 PRINCIPLES OF PERFORMANCE 317

centage decrement in bodily weight, PERFORMANCES AND THE

hours of water-deprivation, amounts and BEHAVIOR SPACE
prior time of saline injections, intensi- I shall illustrate these principles or
ties of just-preceding electric shocks, rules in large part by means of what I
hours of social deprivation, specific rais- shall call behavior-space diagrams or
ings and lowerings of bodily health, spe-
pictures. This notion of a behavior
cial maintenance-schedule with respect
space stems, of course, from Kurt
to aesthetic objects or with respect to
exercise, etc., etc. Lewin's concept of the life space (12).
b. Incentive objects would be: spe- It differs from the latter, however, first,
cific foods, liquids, sex objects, groups in that the structures and processes
of individuals of the same or different which Lewin put inside his "psychologi-
species, barriers, and distances to be cal person" as part of the life space
overcome, enemies, friends, present elec- have been, in the present analysis, taken
tric shocks, poems, paintings, sympho- care of by the three dispositional sys-
nies, open windows to look through in tems of discriminations, belief-expect-
the case of monkeys (see Harlow [6]), ancies, and drives and incentive values,
etc. together with the momentary processes
The character and strength of an activated within these systems.
activated approach (consummation) and Secondly, my behavior space is more
avoidance (escape) disposition will re- frankly a merely pictorial way of repre-
sult from the combination of the par- senting certain relationships and less a
ticular drive-disturbance and the par- would-be true mathematics or logic.
ticular incentive object presented on Thirdly, I differ perhaps from Lewin
the given occasion. The sum total of in that what I am trying to illustrate or
drive-disturbance and incentive object explain by the behavior space are not
relations obtaining for a given indi-
concrete responses but certain general-
vidual make up his drive-incentive-
ized performances. A "performance,"
value system.
as I here use the term, is not something
Assuming now that the above brief
that can be observed on a single occa-
discussion of discrimination dispositions,
sion but something that can be speci-
expectancy dispositions, and drive-in-
fied only from observing two or more
centive-value dispositions has succeeded
concrete behaviors. For example, Tol-
in presenting the basic theoretical posi-
tion, I wish to pass on to the main man, Ritchie, and Kalish in one of their
topic of this paper. How on particular experiments on spatial learning (25)
occasions—given the independent vari- used a single-unit T maze with two
ables of presented sign stimulus-units, vertically opposite, alternatively usable,
previously presented (S-R-S)'s, present starting stems. And they discovered
drive-disturbances and present positive that only by observing the responses of
or negative incentive objects—do the the animals from both stems could they
to-be-inferred resulting then-and-there say whether a rat's acquired behavior—
activated discriminations, expectancies, what I am here calling his "perform-
drives, and incentive values interact to ance"—was that of "taking right-hand
produce actual performance? What, in turns" or that of "going to the lighted
short, are the Rules or Principles of side." But, whichever it was, such an
Performance which must be added to "always going to the right or "always
the system? going towards the lighted side," would
318 EDWARD C. TOLMAN
Expectancy of food OB a result
of pressing lever 190% certainty)
food, F. Also, the expected food is
depicted as having a positive valence vf,
indicated by the plus sign. And the
actor is shown as having an activated
drive or need-push for food, indicated
by the minus sign, labeled «/, inside
him. Further, the length of the dashed
FIG. 1. Behavior-space: rat pressing lever for arrow, labeled "pressing," indicates the
food in a Skinner box. degree of effort or work which, accord-
ing to the expectancy, will be involved
be an example of what is here meant by in the pressing. And the actor is shown
a performance.4 as having a drive or need-push against
Fourthly, it is to be emphasized that work indicated by the second minus
a behavior-space diagram will be neither sign, labeled «_«, within the actor. Fi-
a picture of the actual geographical en- nally, the arrow at the left, labeled Pv,
vironment nor a picture of the supposed is a vector representing the magnitude
phenomenological experience of the or- and direction of the resulting to-be-pre-
ganism. It is rather a merely pictorial dicted performance of pressing.
way of trying to represent the interac- Let us consider further for a moment
tions between the inferred activated dis- need-pushes and valences. I conceive
positions and their consequent determi- a need-push as, so to speak, that por-
nation of the to-be-predicted perform- tion of a drive which, under the con-
ance. For example, my diagram for the crete stimulus conditions of the mo-
case of the rat pressing the lever in a ment, gets into the behavior space. The
Skinner box (17) at a certain rate is magnitude of a need-push is, in short,
shown in Fig. 1. assumed to be determined not only by
This diagram represents the assump- the magnitude of the actual activated
tion that the rat discriminates himself drive but also by the goodness of the
as "actor" in the presence of the dis- discrimination of drive stimuli under
criminated lever, Lr. The pair of dotted the given experimental conditions.
lines connecting actor to Lr symbolizes Consider, for example, that type of
"immediate presence of the lever." And latent-learning experiment which has
the diagram represents further the fact been called by Thistlethwaite (19) the
that this discriminated lever, Lr, has irrelevant-incentive experiment. In such
become the sign term in the activated experiments, during the initial training
expectancy. The whole expectancy is period an irrelevant goal object (say
indicated by the contents of the dash- food) is presented in one alley of a T
dot ellipse. This expectancy has ac- or Y maze, and water, a relevant goal
quired, it is assumed, a certainty of object (since the animals are in this pe-
about 90 per cent and is to the effect riod run thirsty), is presented in the
that if the actor responds by the "per- other alley. Then the rats are made
formance" of pressing, he will get to hungry and it appears from the original
4
For this notion of a performance as a be- experiment of Spence and Lippitt (18)
havioral disposition which can be identified and from various repeat experiments
only by observing at least two somewhat dif- (see, for example, Kendler [11]) that
ferent test situations I am indebted primarily rats often do not then tend to go to
to Professor Benbow F. Ritchie. (Private the food side when first made hungry.
communication.) But see also a quite recent
statement of what seems to be essentially the However, it has also appeared from an
same position by Campbell (2). experiment by Thistlethwaite (20) that,
 PRINCIPLES or PERFORMANCE 319

if the animals have had some previous food pellets will have in the behavior-
training in discriminating their drive space diagram corresponding to the
stimuli, they are much more likely to Skinner box will depend not only upon
show some initial tendency to go at the incentive value of such pellets (as
once to the food side. These two con- measured in some standard situation)
trasting results would be interpreted, by but also upon the discriminability of
me, as meaning that, although the rats the pellets in the concrete stimulus con-
have been made hungry (as would be text of a Skinner box. And this dis-
shown by their readiness to approach criminability may, hypothetically at
or consume food in the standard situa- least, be less than in the standard situa-
tion), this hunger drive does not tend tion. Hence the valence of the food, in
to be discriminated (i.e., to produce a the behavior space corresponding to the
corresponding need-push in the behav- Skinner box, could be less (though I
ior space corresponding to the maze) doubt it) than the basic incentive value
unless there has been previous practice of such pellets, as measured in a stand-
in drive discrimination in a mazelike ard situation. Similarly, the negative
situation. Further, some recent find- valence of the work involved in pressing
ings by Heyer and O'Kelly (7) indi- might, in the behavior space correspond-
cate that tissue dehydration due to sub- ing to the Skinner box, be less than the
cutaneous saline injection will not cause actual negative incentive value of such
appropriate performance in a maze to work. But this also seems doubtful.
get to water, unless this dehydration Look again at Fig. 1. My final argu-
has been experienced over a consider- ment is that as a result of (i) the need-
able preceding period of time. I would push for food, (ii) the positive valence
interpret this also as lack of thirst need- of the expected food, (iii) the need-
push in the behavior space (except in push against work, and (iv) the nega-
the actual presence of water itself) un- tive valence of the expected work (this
less there has been plenty of preceding expected work is indicated by the length
time to experience and discriminate the of the dashed arrow), there results the
special drive stimuli resulting from the performance vector Pv. It is this vector
saline-produced dehydration. which specifies the direction and magni-
Return now to the Skinner box. We tude of the to-be-predicted actual per-
may probably assume that in it, be- formance. The greater the valence of
cause of preliminary training, the hun- the expected food, the greater the food
ger drive stimuli and the nonwork drive need-push, and the greater the expect-
stimuli have become completely dis- ancy that the food will result, the
criminated. We may assume in short greater the magnitude of the perform-
that the magnitude of the food need- ance vector toward actually pressing the
push corresponds directly to the magni- lever. On the other hand, the greater
tude of the actual food hunger drive. the need-push against work and the
And finally we may probably also as- greater the negative valence of the ex-
sume that the magnitude of the non- pected work and the stronger the ex-
work need-push corresponds directly to pectation of this work, the less the per-
the actual nonwork drive. formance vector toward pressing the
What now about the valences? Val- lever.
ences are conceived as bearing nearly For the sake, now, of the critic who
the same relation to incentive-values dislikes such a pictorial mumbo jumbo,
that need-pushes bear to drives. That it may be pointed out that these rela-
is, the amount of valence which the tions probably can be equally well ex-
320 EDWARD C. TOLMAN

pressed by an algebraic equation (Equa- an activated (Si-R^-S^) expectation.

tion 1). Further, however, it is to be noted that
Hull's equation does not contain the
Pv = I® (»/, Vf, exp/) second part of the right-hand side of
— }v (n^,,, v-w, expw) [1] my equation. It may be recalled, how-
Here Pv stands for the magnitude of ever, that he does introduce, separately,
the performance vector (i.e., the tend- the construct of conditioned inhibition
ency to press the lever). And this mag- sia which, if I interpret him correctly,
nitude is shown as increasing with the is somewhat related to what I am here
values of «/, v/, and exp/, combined ac- saying about w or expected work. SIR
cording to some function, fa, and as be- he apparently subtracts from aER just
ing decreased by the values of M_W, v.*,, as I subtract /„ (»_„,, v-u,, exp w ). It
and expw, combined according to some must be pointed out, however, that for
function, /„. The exact forms of these Hull inhibition is somewhat different
two functions, /„ and /„, I have not at- from my concept of expected work and
tempted to specify. It may be that need against work. But in any event,
these functions both have the form of both he and I assume that the "ex-
simple multiplication. But in any event, pected work" or the "conditioned in-
it is assumed for both of these functions hibition" gets established through repe-
that the component variables all operate titions of the activity involved.
in a positive direction such that the in- Turn now to escape behavior. I
crease in any one of these component would feel that, in the usual case, es-
variables tends to increase the value of cape is not as has been suggested by
the function. the Hullians primarily a getting-to a
Still further, the critic will note that positive valence attached to the reduc-
the first half of the right-hand side of tion of pain but is rather simply a
this equation is rather similar to Hull's getting-away from the negative valence
equation for reaction potential, or SER of the shock itself. Such an interpreta-
(9), (which is presented here as Equa- tion is represented by Fig. 2. In Fig. 2
tion 2). the jumping is represented as a mere
leading away from the grid. And the
(Hull) 8ER = } ( D X V 1 animal will merely jump in any direc-
X K X SHR) [2] tion easily available.
Hull's D, or drive, is closely similar Turn next to avoidance behavior.
to my nf, or hunger need-push. His V\, Avoidance behavior has always been
or stimulus intensity dynamism, is, how- something of a problem for S-R psy-
ever, a term I do not have. For me the chologists. This point was made some
effect of any intensity or a presented years ago by Hilgard and Marquis (8)
stimulus-unit such as the stimulus-pat- and has recently been re-emphasized
tern resulting from the lever would al- and subjected to a further analysis by
ready have been evaluated by an analy-
sis of the discriminatory dispositional Enpectoncy of continued shock
from being on grid (90% certomty]
system and of the effects in it of in-
tense or nonintense stimulus-units upon
the discriminations of Lr and of F.
Hull's K, or incentive motivation, cor-
responds to my valence, vf. And his
SHR, or habit, is an S-R connection FIG. 2. Behavior-space: rat escaping from
whereas my expectancy is conceived as charged grid.
 PRINCIPLES OF PERFORMANCE 321

Ritchie (16). In avoidance behavior Expectancy of coming Ihock from

going 10 grid (90% certainly)
the performance vector results, accord-
ing to my language, from the expecta-
tion that going to the grid is going to
lead to shock. And this performance
vector pushes or pulls the animal away
from the grid even when he is not on it.
The negative need-push against shock FIG. 3. Behavior-space: rat avoiding grid
will lead, not as in the escape case from which shock is expected.
merely to getting off the grid after the
shock has come, but rather to the per- The greater the need-push w_s/t against
formance of avoiding the grid alto- shock, the greater the negative valence
gether. See Fig. 3. In such a case, t>_rt of the particular shock and the
pain never occurs and hence there is no more certain the expectation exps;j (or
pain to be reduced. Mowrer (14), of immediate discrimination of shock), the
course, is the one among the stimulus- stronger will be the performance vector
response-oriented psychologists who has leading to behavior away from the sign
most clearly seen this. And he has of pain (in the case of avoidance) or
solved it by postulating a conditioned leading to the leaving of the pain-ob-
anxiety to the whole grid box, the re- ject (in the case of escape). And the
duction of which constitutes positive re- greater the need against work »_„,, the
inforcement. Now while I would ac- greater the negative valence TJ_«, of the
cept something like Mowrer 's inter- particular work involved in the escape
pretation for many instances, I do not or the avoidance and the greater the ex-
believe that anxiety is always or neces- pectancy of this work, the smaller will
sarily involved. The skilled human sub- be the Pv toward escape or avoidance.
ject who has learned to lift his finger in It is to be noted further, however,
time to avoid a minor shock is not usu- that, in the avoidance case, w, or the
ally, I believe, suffused with even a expected work involved between sign
minimum of anxiety. He responds so and significate, can be large and lead
as to avoid the shock, quite unemotion- to relatively weak avoidance of the sign
ally and purely on the basis of his ex- as a result of a number of relatively
pectancy, his need-push, and the nega- different physical reasons. The w may
tive valence of the shock. He expects be waiting a long time for the shock.
shock in a certain place at a certain Or it may involve doing some special
time, provided he goes to or stays in further act such as pressing a treadle
that place, and his performance is one for the shock to appear. Or, to take a
of avoiding that place before that time, quite different example, where the sig-
in a quite relaxed, nonanxious fashion. nificate is not pain from shock but pain
Finally, however, it is to be noted from a rat-hungry cat, w might involve
that although my diagrams for escape the breaking down of a tough barrier by
and avoidance are different from that the rat between it and the cat. In all
for approach, my algebraic equation for three cases the avoidance response of
either escape or avoidance would be of the rat will be relatively weaker than it
the same form as that for approach would have been if these intervening
(Equation 3). w's had been small.
But up to now you may well feel that
Pv = fa, (n.,1, W- my formula and Hull's are not signifi-
,,, v-w, expw) [3] cantly different. And, if you do, so
322 EDWARD C. TOLMAN

much the better. It would indicate that ance which I would explain with the
there has been really much common help of angles, namely, VTEing (vicari-
agreement between us in spite of dif- ous trial and error). The explanation
ferences of words, phrases, and slogans. will be closely related to, but in some
However, when it comes to cases ways different from, that which Clark
where not one but two or more alterna- Crannell and I evolved some years ago
tive performance tendencies are involved (21, 22) in connection with what I
and result in some quite new perform- then called the "schematic sowbug."
ance—such, for example, as taking a Actually, anyone who read the earlier
compromise path between two original papers may have already recognized
ones, taking a shortcut, taking a round- that the actor as I have presented him
about route, VTEing, or the like—then here is no more than a modern, and I
depicting the situation by means of the hope a less complicated, version of my
behavior space becomes (to me) more old friend, the schematic sowbug.
helpful and suggestive. (And, of course, There are two sets of empirical find-
not to renege on my basic position, my ings as to VTEing to be emphasized:
own "private style of thinking," to (a) Early in learning, just before (or
quote Kendler [11], is such that I can- just as) the rats begin to react to the
not easily think about the interaction crucial sign dimension, they tend to do
of a single sign, expectancy, need-push, more VTEing when there is a large dis-
and valence without first seeing these in criminable difference between the two
some sort of spatial terms.) But turn cards than when there is a small one.
now to the new problem of combining They VTE more for a white card vs. a
two expectancies. I shall introduce black card than for a white card vs. a
here a new concept, that of the angular light gray card (21). (b) Later, how-
ever, when they have had much preced-
dispersion between expectancies.
ing experience in discriminating along
This concept of angles is tricky and the white-gray-black dimension, they
I know I am not wholly clear on it. It tend to VTE more when the discrimi-
is one that Lewin also bogged down on. nable difference between the two cards
Nevertheless let me present, as an illus- is small (24). They VTE more for
tration, one type of compound perform- white vs. light gray cards than for white
vs. black cards.
WHITE vs. BLACK
Consider first the behavior-space pic-
tures early in learning when the rats
VTE more for the white-black than for
4 i the white-gray. (See Fig. 4a and Fig.
4b.)
In Fig. 4a the actor (the sowbug) is
in the white-black situation and in Fig.
4b in the white-gray one. And I have
\v^>' •--. .A-^X* assumed that white vs. black corre-
©K ^ sponds to a 90° angle relative to the
actor's (sowbug's) nose, whereas the
white vs. gray corresponds to a smaller
angle. I have also assumed that in this
FIG. 4a. Behavior-space, early in discrimi-
early part of learning there operates, in
nation learning. White card versus black addition to the hunger need-push nf, a
card. Much VTE. more purely cognitive sort of need-push
 PRINCIPLES OF PERFORMANCE 323

to examine new stimuli, nmg. The need- VTEing for white vs. black (Fig. 4a)
push corresponding to this cognitive-ex- than for white vs. gray (Fig. 4b) I
amination drive I have indicated by a shall also assume that since in the case
minus sign in the fore part of the actor of white vs. black the two cognitive-
or bug. I have also assumed that both examination vectors are at right angles,
the stimulus cards have for a while, the reduction of one of these vectors
as new stimuli, positive valences for through satiation has no effect on the
this cognitive-examining drive. That is, other. The rat gets temporarily sati-
there will be performance vectors push- ated for white but he still wants to look
ing the actor to "look at" each stimulus at black, and vice versa. In the case
card. These vectors are those indicated of the white vs. gray, on the other hand,
at the rear of the bug. In addition, I the two examining vectors are at a nar-
have indicated, by putting in F (for rower, acute, angle which means that
food) with a plus valence on it on both the shortening of one vector through
sides, that at this early stage of learn- satiation projects upon and tends to
ing there have developed expectancies shorten the other. Looking at white
that going toward either card may lead reduces temporarily not only the vector
to food. And for the diagram to be to look at white but also the one to look
complete there should also be another at gray, and vice versa. Hence early in
pair of shortish vectors pushing the bug, learning or just before learning there
as a result of these expectancies, to jump will be less VTEing for white vs. gray
towards each card. I have omitted than for white vs. black.
these latter, jumping vectors, so as not Consider now the situation later in
to complicate the picture. The vectors learning. The original cognitive-exami-
shown are "looking towards and examin- nation drive is assumed now to have be-
ing" vectors. come wholly satiated. The vectors cor-
I shall assume further (and here I responding to it will have disappeared.
am assuming very much what Glanzer But expectations have begun to build
[3, 4] has proposed in two recent ar-
ticles in which he was concerned with WHITE vs. GRAY
alternation behavior in a symmetrical
maze) that in the course of looking at
one card the valence of this card as
"something to look at and examine"
tends to become temporarily discharged
or satiated. This then weakens, for the
moment, the vector for looking at this
first card. The actor then turns to look
at the other card. The vector toward
this latter card, however, then becomes
satiated in its turn; whereas the vector
for looking at the first card becomes
stronger again. So the rat looks back
and forth, or VTE's, until satiation for
both cards, as far as looking or examin-
ing is concerned, has become complete.
In short, the rat VTE's until jumping
Flo. 4b. Behavior-space, early in discrimi-
takes over. nation learning. White card versus gray card.
To explain further why there is more Little VTE.
324 EDWARD C. TOLMAN

up to the effect that, in the one case, Now, for the sake of the nonvisually
white leads to food and black to no minded critic, let me here suggest that
food, and, in the other, that white leads perhaps a more respectable and conven-
to food and gray to no food. These tional way of talking would be to say
situations are represented in Figs. Sa merely that early in learning there is
and Sb. more generalization of satiation of the
The vectors are no longer ones for cognitive examining drive from white to
looking, but ones for jumping to white gray than from white to black; whereas,
(as sign of food) and for not-jumping later in learning, there is more generali-
to black (as sign of nonfood). In the zation of acquired approach and avoid-
white-black case (Fig. Sa), because the ance tendencies from white to gray than
two vectors are at right angles the from white to black. More generaliza-
negative vector "away from the black" tion of satiation for "looking at" early
does not affect the positive one "to- in learning means less VTEing between
ward the white," and vice versa. In white and gray, whereas more gener-
the white-gray case (Fig. Sb), on the alization of approach and avoidance
contrary, since the two arrows are at an tendencies later in learning means more
acute angle, the negative vector "away VTEing between white and gray. Actu-
from the gray" does tend to project ally, however, I suspect that such a
upon and reduce the positive vector "to- more "respectable" way of talking is
ward the white"; and the positive one really just as confused and as full of
"toward the white" projects upon and hidden assumptions as is my pictorial
decreases the negative one "away from way. In any event, the pictorial terms
the gray." And this means in the white- suggest to me (and I emphasize to me)
gray case that until the expectancies be- a greater number of further experiments
come very certain and the vectors very to try out than do the purely verbal
long, the reduction of the positive vec- ones.
tor by the negative one and vice versa The topic of this paper was "Prin-
will be sufficient to keep the rat in un- ciples of Performance." And by a per-
stable equilibrium so that VTEing will
tend to continue much longer. WHITE v». GRAY

WHITE vs. BLACK

FIG. Sa. Behavior-space, later in discrimi- FIG. Sb. Behavior-space, later in discrimi-
nation learning. White card versus black nation learning. White card versus gray card.
card. Little VTE. Much VTE.
 PRINCIPLES or PERFORMANCE 325
formance was meant, as has been said, they will turn out to be describable in
a generalized way of behaving to be dis- terms of interesting complications and
covered and specified only by observing interactions between the performance
the responses in at least two and per- vectors, as these complications and in-
haps more concrete test situations. Fur- teractions appear in particular behavior-
ther, a performance when thus discov- space setups.
ered is identifiable only in general terms,
such as approach, consummation, es- REFERENCES
cape, avoidance, vicarious trial-and-er- 1. BRUNSWIK, E. Organismic achievement
and environmental probability. Psy-
ror examination, or the like. The mo- chol. Rev., 1943, 50, 255-272.
tor skill characteristics involved in such 2. CAMPBELL, D. T. Operational delineation
a performance are not included in its of "what is learned" via the transposi-
definition. Only in a very general way tion experiment. Psychol. Rev., 1954,
may a performance be further specified 61, 167-174.
3. GLANZER, M. The role of stimulus satia-
as a "looking at," a "jumping to," a tion in spontaneous alternation. J. exp.
"running down," a "pressing," an "eat- Psychol., 1953, 45, 387-393.
ing," or a "drinking." The point of 4. GLANZER, M. Stimulus satiation: an ex-
view being argued for is that perform- planation of spontaneous alternation
ances and not movements or skills are and related phenomena. Psychol. Rev.,
1953, 60, 257-268.
all that we ought now, at the present 5. GUTHRIE, E. R. The psychology of learn-
time, to try to explain or to predict in ing (Rev. Ed.), New York: Harper,
detail. Under certain controlled con- 1952.
ditions we can use movements as pointer 6. HARLOW, H. F. Motivation as a factor in
readings for performances but we can- the acquisition of new responses. In
Current theory and research in motiva-
not as yet predict movements independ- tion, a symposium. Lincoln: Univer.
ently of the performances within which of Nebraska Press, 1953, pp. 24-49.
they occur. As Brunswik has pointed 7. HEYER, A. W., JR., & O'KELLY, L. I.
out (1), all kinds of different move- Studies in motivation and retention.
Comp. Psychol. Monogr., 19S1, 20,
ments and skills may vicariously substi- Serial No. 106.
tute for one another and produce one 8. HILGARD, E. R., & MARQUIS, D. G. Con-
and the same achievement or perform- ditioning and learning. New York:
ance. Movements and skills in his terms Appleton-Century-Crofts, 1940.
are mere mediational, peripheral phe- 9. HULL, C. L. A behavior system. New
Haven: Yale Univer. Press, 1952.
nomena. It is the "distal," not the 10. KENDLEH, H. H. An investigation of la-
"proximal," characters of S's and R's to tent learning in a T-maze. J. comp.
which I am suggesting that we devote physiol. Psychol., 1947, 40, 265-270.
ourselves for the time being. This is 11. KENDLER, H. H. "What is learned?" A
my interest and my bias. theoretical blind alley. Psychol. Rev.,
1952, 59, 269-277.
One last word. Most of the really in- 12. LEWIN, K. A dynamic theory of person-
teresting problems of psychology—such ality. New York: McGraw-Hill, 1935.
as the distortion effects of motivation 13. MACCORQUODALE, K., & MEEHL, P. E.
on perception (and I still believe there Section 2 in Estes, W. K. et al. Mod-
are some) or again those facts which ern learning theory. New York: Ap-
pleton-Century-Crofts, 1954.
have to do with what I have called the 14. MOWRER, O. H. On the dual nature of
widening or narrowing of cognitive maps learning—a reinterpretation of "condi-
(23)—will eventually find their place, tioning" and "problem-solving." Har-
I believe, under the heading of compli- vard Educ. Rev., 1947, 17, 102-148.
15. MURRAY, H. A. Explorations in person-
cations in the principles of performance. ality. New York: Oxford Univer. Press,
Or, to use the language of this paper, 1938.
326 EDWARD C. TOLMAN
16. RITCHIE, B. F. Can reinforcement theory matic sowbug. Psychol. Rev., 1939,
account for avoidance? Psychol, Rev., 46, 318-336.
1951, 58, 382-386. 22. TOLMAN, E. C. Discrimination vs. learn-
17. SKINNER, B. F. The behavior of organ- ing and the schematic sowbug. Psy-
isms: an experimental analysis. New chol. Rev., 1941, 48, 367-382.
York: Appleton-Century-Crofts, 1938. 23. TOLMAN, E. C. Cognitive maps in rats
18. SPENCE, K. W., & LIPPITT, R. An experi- and men. Psychol. Rev., 1948, 55, 189-
mental test of the sign-gestalt theory of 208.
trial-and-error learning. /. exp. Psy- 24. TOLMAN, E. C., & MINIUM, E. VTE in
chol., 1946, 36, 491-502. rats: overlearning and difficulty of dis-
19. THISTLETHWAITE, D. A critical review of crimination. J. comp. Psychol., 1943,
latent learning and related experiments. 36, 91-98.
Psychol. Bull., 1951, 48, 97-129. 25. TOLMAN, E. C., RITCHIE, B. F., & KALISH,
20. THISTLETHWAITE, D. Conditions of ir- D. Studies in spatial learning: II. Place
relevant-incentive learning. /. comp. learning versus response learning. /.
pkysiol. Psychol., 1952, 45, 517-525. exp. Psychol., 1946, 36, 221-229.
21. TOLMAN, E. C. Prediction of vicarious
trial and error by means of the sche- (Received November 22, 1954)