the psychological review - semantic scholar › 0874 › a64d60a23a20303877... · 2018-10-09 ·...

VOL. 55, No. 4 JULY, 1948

T H E PSYCHOLOGICAL REVIEW

COGNITIVE MAPS IN RATS AND MEN *

BY EDWARD C. TOLMAN

University of California

I shall devote the body of this paperto a description of experiments withrats. But I shall also attempt in a fewwords at the close to indicate the sig-nificance of these findings on rats forthe clinical behavior of men. Most ofthe rat investigations, which I shall re-port, were carried out in the Berkeleylaboratory. But I shall also include,occasionally, accounts of the behaviorof non-Berkeley rats who obviouslyhave misspent their lives in out-of-State laboratories. Furthermore, in re-porting our Berkeley experiments Ishall have to omit a very great many.The ones I shall talk about were car-ried out by graduate students (or un-derpaid research assistants) who, sup-posedly, got some of their ideas fromme. And a few, though a very few,were even carried out by me myself.

Let me begin by presenting diagramsfor a couple of typical mazes, an alleymaze and an elevated maze. In thetypical experiment a hungry rat is putat the entrance of the maze (alley orelevated), and wanders about throughthe various true path segments andblind alleys until he finally comes to

134th Annual Faculty Research Lecture, de-livered at the University of California, Berke-ley, March 17, 1947. Presented also on March26, 1947 as one in a series of lectures in Dy-namic Psychology sponsored by the divisionof psychology of Western Reserve University,Cleveland, Ohio.

the food box and eats. This is re-peated (again in the typical experi-ment) one trial every 24 hours and theanimal tends to make fewer and fewererrors (that is, blind-alley entrances)and to take less and less time betweenstart and goal-box until finally he is en-tering no blinds at all and running in avery few seconds from start to goal.The results are usually presented in theform of average curves of blind-en-trances, or of seconds from start 4:0finish, for groups of rats.

All students agree as to the facts.They disagree, however, on theory andexplanation.

(1) First, there is a school of ani-mal psychologists which believes thatthe maze behavior of rats is a matter ofmere simple stimulus-response connec-tions. Learning, according to them,consists in the strengthening of some ofthese connections and in the weakeningof others. According to this 'stimulus-response' school the rat in progressingdown the maze is helplessly respondingto a succession of external stimuli—sights, sounds, smells, pressures, etc.impinging upon his external sense or-gans—plus internal stimuli coming fromthe viscera and from the skeletal muscles.These external and internal stimuli callout the walkings, runnings, turnings, re-tracings, smellings, rearings, and thelike which appear. The rat's central

189

Dave

Note

Tolman, E.C. (1948). Cognitive maps in rats and men. Psychological Review, 55, 189-208.

190 EDWABD C. TOLMAN

START

Plan of maze14-Unit T-Alley Maze

Fro. 1(From M. H. Elliott, The effect of change of reward on the maze per-

formance of rats. Univ. Calif. Publ. Psychol., 1928, 4, p. 20.)

nervous system, according to this view,may be likened to a complicated tele-phone switchboard. There are the in-coming calls from sense-organs andthere are the outgoing messages tomuscles. Before the learning of a spe-cific maze, the connecting switches(synapses according to the physiologist)are closed in one set of ways and pro-duce the primarily exploratory responseswhich appear in the early trials. Learn-ing, according to this view, consists inthe respective strengthening and weak-ening of various of these connections;those connections which result in theanimal's going down the true path be-come relatively more open to the pas-

sage of nervous impulses, whereas thosewhich lead him into the blinds becomerelatively less open.

It must be noted in addition, how-ever, that this stimulus-response schooldivides further into two subgroups.

(a) There is a subgroup which holdsthat the mere mechanics involved in therunning of a maze is such that thecrucial stimuli from the maze get pre-sented simultaneously with the correctresponses more frequently than they dowith any of the incorrect responses.Hence, just on a basis of this greaterfrequency, the neural connections be-tween the crucial stimuli and the cor-rect responses will tend, it is said, to

COGNITIVE MAPS IN RATS AND MEN 191

get strengthened at the expense of theincorrect connections.

(b) There is a second subgroup inthis stimulus-response school whichholds that the reason the appropriateconnections get strengthened relativelyto the inappropriate ones is, rather, thefact that the responses resulting fromthe correct connections are followedmore closely in time by need-reductions.Thus a hungry rat in a maze tends toget to food and have his hunger re-duced sooner as a result of the true pathresponses than as a result of the blindalley responses. And such immediately

following need-reductions or, to use an-other term, such 'positive reinforce-ments' tend somehow, it is said, tostrengthen the connections which havemost closely preceded them. Thus itis as if—although this is certainly notthe way this subgroup would themselvesstate it—the satisfaction-receiving partof the rat telephoned back to Centraland said to the girl: "Hold that con-nection; it was good; and see to it thatyou blankety-blank well use it againthe next time these same stimuli comein." These theorists also assume (atleast some of them do some of the

14-Unit T-Elevated Mazes

FIG. 2(From C. H. Honzik, The sensory basis of maze learning in rats. Compar. Psychol. Monogr.,

1936, 13, No. 4, p. 4. These were two identical mazes placed side by side in the same room.)

192 EDWAED C. TOLMAN

!X

S30

190

OJ

S .501Uaa

c

90

70

50

30

ess hungry non-reward

Hungry

I 2 3 4 S 6 7 8 9 10 II II 13 1+ 15 16 17

Days

Error curves for four groups, 36 ratB.

Fie. 3

(From E. C. Tolman and C. H. Honzik, Degrees of hunger, reward and non-reward, and maze learning in rats. Univ. Calif. Publ. Psychol., 1930, 4, No.16, p. 246. A maze identical with the alley maze shown in Fig. 1 was used.)'

time) that, if bad results—'annoyances,''negative reinforcements'—follow, thenthis same satisfaction-and-annoyance-receiving part of the rat will telephoneback and say, "Break that connectionand don't you dare use it next timeeither."

So much for a brief summary of thetwo subvarieties of the 'stimulus-re-sponse,' or telephone switchboard school.

(2) Let us turn now to the secondmain school. This group (and I be-long to them) may be called the fieldtheorists. We believe that in the courseof learning something like a field mapof the environment gets established inthe rat's brain. We agree with theother school that the rat in running amaze is exposed to stimuli and is finallyled as a result of these stimuli to theresponses which actually occur. Wefeel, however, that the intervening brainprocesses are more complicated, more

patterned and often, pragmaticallyspeaking, more autonomous than do thestimulus-response psychologists. Al-though we admit that the rat is bom-barded by stimuli, we hold that hisnervous system is surprisingly selectiveas to which of these stimuli it will letin at any given time.

Secondly, we assert that the centraloffice itself is far more like a map con-trol room than it is like an old-fashionedtelephone exchange. The stimuli, whichare allowed in, are not connected byjust simple one-to-one switches to theoutgoing responses. Rather, the incom-ing impulses are usually worked overand elaborated in the central controlroom into a tentative, cognitive-likemap of the environment. And it is thistentative map, indicating routes andpaths and environmental relationships,which finally determines what responses,if any, the animal will finally release.


Finally, I, personally, would hold fur-ther that it is also important to dis-cover in how far these maps are rela-tively narrow and strip-like or relativelybroad and comprehensive. Both strip-maps and comprehensive-maps may beeither correct or incorrect in the sensethat they may (or may not), when actedupon, lead successfully to the animal'sgoal. The differences between suchstrip maps and such comprehensivemaps will appear only when the rat islater presented with some change withinthe given environment. Then, the nar-rower and more strip-like the originalmap, the less will it carry over success-fully to the new problem; whereas, thewider and the more comprehensive itwas, the more adequately it will servein the new set-up. In a strip-map thegiven position of the animal is con-nected by only a relatively simple andsingle path to the position of the goal.In a comprehensive-map a wider arc ofthe environment is represented, so that,if the starting position of the animal be

changed or variations in the specificroutes be introduced, this wider mapwill allow the animal still to behaverelatively correctly and to choose theappropriate new route.

But let us turn, now, to the actualexperiments. The ones, out of many,which I have selected to report aresimply ones which seem especially im-portant in reinforcing the theoreticalposition I have been presenting. Thisposition, I repeat, contains two assump-tions: First, that learning consists notin stimulus-response connections but inthe building up in the nervous system ofsets which function like cognitive maps,and second, that such cognitive mapsmay be usefully characterized as vary-ing from a narrow strip variety to abroader comprehensive variety.

The experiments fall under fiveheads: (1) "latent learning," (2) "vi-carious trial and error" or "VTE," (3)"searching for the stimulus," (4) "hy-potheses" and (5) "spatial orienta-tion."

6-TJnit Alley T-Maze

FIG. 4(From H. C. Blodgett, The effect of the introduction of reward upon

the maze performance of rats. Univ. Calif. Publ. Psychol, 1929, 4, No. 8,p. 117.)

194 EDWAKD C. TOLMAN

s.o

£ . 5

£ . 0

1.6

1.0

s °>s

_ _ _ _ _ Croup I_ _ _ _ _ Oroup II

Crop III

1 tlo. of «ay

FIG. 5

(From H. C. Blodgett, The effect of the introduction of reward upon the maze per-formance of rats. Univ. Calif. Publ. Psychol., 1929, 4, No. 8, p. 120.)

(1) "Latent Learning" Experiments.The first of the latent learning experi-ments was performed at Berkeley byBlodgett. It was published in 1929.Blodgett not only performed the experi-ments, he also originated the concept.He ran three groups of rats through asix-unit alley maze, shown in Fig. 4.He had a control group and two ex-perimental groups. The error curves forthese groups appear in Fig. 5. Thesolid line shows the error curve forGroup I, the control group. These ani-mals were run in orthodox fashion.That is, they were run one trial a dayand found food in the goal-box at theend of each trial. Groups II and IIIwere the experimental groups. The ani-mals of Group II, the dash line, werenot fed in the maze for the first sixdays but only in their home cages sometwo hours later. On the seventh day(indicated by the small cross) the ratsfound food at the end of the maze for

the first time and continued to find iton subsequent days. The animals ofGroup III were treated similarly exceptthat they first found food at the endof the maze on the third day and con-tinued to find it there on subsequentdays. It will be observed that the ex-perimental groups as long as they werenot finding food did not appear to learnmuch. (Their error curves did notdrop.) But on the days immediatelysucceeding their first finding of the foodtheir error curves did drop astound-ingly. It appeared, in short, that dur-ing the non-rewarded trials these ani-mals had been learning much more thanthey had exhibited. This learning,which did not manifest itself until afterthe food had been introduced, Blodgettcalled "latent learning." Interpretingthese results anthropomorphically, wewould say that as long as the animalswere not getting any food at the end ofthe maze they continued to take their


time in going through it—they con-tinued to enter many blinds. Once,however, they knew they were to getfood, they demonstrated that duringthese preceding non-rewarded trials theyhad learned where many of the blindswere. They had been building up a'map,' and could utilize the latter assoon as they were motivated to do so.

Honzik and myself repeated the ex-periments (or rather he did and I gotsome of the credit) with the 14-unitT-mazes shown in Fig. 1, and withlarger groups of animals, and got similarresults. The resulting curves are shownin Fig. 6. We used two control groups—one that never found food in themaze (HNR) and one that found itthroughout (HR). The experimentalgroup (HNR—R) found food at the endof the maze from the 11th day on andshowed the same sort of a sudden drop.

But probably the best experimentdemonstrating latent learning was, un-fortunately, done not in Berkeley but atthe University of Iowa, by Spence andLippitt. Only an abstract of this ex-periment has as yet been published.However, Spence has sent a preliminarymanuscript from which the followingaccount is summarized. A simple Y-maze (see Fig. 7) with two goal-boxeswas used. Water was at the end of theright arm of the Y and food at the endof the left arm. During the trainingperiod the rats were run neither hun-gry nor thirsty. They were satiated forboth food and water before each day'strials. However, they were willing torun because after each run they weretaken out of whichever end box theyhad got to and put into a living cage,with other animals in it. They weregiven four trials a day in this fashion

1 J. J « b 6 7 6 t U 11 l i U U 15 IS 1' IS 19 iO 31 33

DaysError curves for HR, HNR, and HNR-R

FIG. 6(From E. C. Tolman and C. H. Honzik, Introduction and removal of reward, and maze per-

formance in rats. Univ. Calif. Publ. Psychol., 1930, 4, No. 19, p. 267.)

196 EDWAKD C. TOLMAN

Ground plan of the apparatusFIG. 7

(Taken from K W. Spence and R. Lippitt,An experimental test of the sign-gestalt theoryof trial and error learning. J. exper. Psychol.,1946, 36, p. 494. In this article they were de-scribing another experiment but used the samemaze.)

for seven days, two trials to the rightand two to the left.

In the crucial test the animals weredivided into two subgroups one madesolely hungry and one solely thirsty. Itwas then found that on the first trialthe hungry group went at once to theleft, where the food had been, statisti-cally more frequently than to the right;and the thirsty group went to the right,where the water had been, statisticallymore frequently than to the left. Theseresults indicated that under the previ-ous non-differential and very mild re-warding conditions of merely being re-turned to the home cages the animalshad nevertheless been learning wherethe water was and where the food was.In short, they had acquired a cognitivemap to the effect that food was to theleft and water to the right, althoughduring the acquisition of this map they

had not exhibited any stimulus-responsepropensities to go more to the sidewhich became later the side of the ap-propriate goal.

There have been numerous other la-tent learning experiments done in theBerkeley laboratory and elsewhere. Ingeneral, they have for the most part allconfirmed the above sort of findings.

Let us turn now to the second groupof experiments.

(2) "Vicarious Trial and Error" or"VTE." The term Vicarious Trial andError (abbreviated as VTE) was in-vented by Prof. Muenzinger at Colo-rado 2 to designate the hesitating, look-

7Apparatus Used for Testing Discrimination of

Visual Patterns

FIG 8(From K. S Lashley, The mechanism of

vision. I. A method for rapid analyses ofpattern-vision in the rat. J. genet. Psychol.,1930, 37, p. 454.)

2 Vide: K. F. Muenzinger, Vicarious trialand error at a point of choice: I. A generalsurvey of its relation to learning efficiency.J. genet. Psychol., 1938, 53, 7S-86.

COGNITIVE MAPS IN RATS AND M E N 197

I.&V S"

FIG. 9

(From E. C Tolman, Prediction of vicarious trial and error by means of the schematic sow-bug. PSYCHOL. REV , 1939, 46, p 319.)

ing-back-and-forth, sort of behaviorwhich rats can often be observed to in-dulge in at a choice-point before actu-ally going one way or the other.

Quite a number of experiments uponVTEing have been carried out in ourlaboratory. I shall report only a few.In most of them what is called a dis-crimination set-up has been used. Inone characteristic type of visual dis-crimination apparatus designed by Lash-ley (shown in Fig. 8) the animal is puton a jumping stand and faced with twodoors which differ in some visual prop-erty say, as here shown, vertical stripesvs. horizontal stripes.

One of each such pair of visualstimuli is made always correct and theother wrong; and the two are inter-changed from side to side in randomfashion. The animal is required tolearn, say, that the vertically stripeddoor is always the correct one. If he

jumps to it, the door falls open and hegets to food on a platform behind. If,on the other hand, he jumps incorrectly,he finds the door locked and falls intoa net some two feet below from whichhe is picked up and started over again.

Using a similar set-up (see Fig. 9),but with landing platforms in front ofthe doors so that if the rat chose incor-rectly he could jump back again andstart over, I found that when the choicewas an easy one, say between a whitedoor and a black door, the animals notonly learned sooner but also did moreVTEing than when the choice was diffi-cult, say between a white door and agray door (see Fig. 10). It appearedfurther (see Fig. 11) that the VTEingbegan to appear just as (or just before)the rats began to learn. After thelearning had become established, how-ever, the VTE's began to go down.Further, in a study of individual dif-

198 EDWARD C. TOLMAN

LEARNING CURVESAVERAGE NO. CORRECT RUNS PER DAY

z<£t -O 3

Eoo . r -

/ N-ll

J/ /^v N-IO

- ^ / V.y

*» LIGHT GREY

IOO9A

BLACK

MEDIUM GREY

2 -

I -

I 2 3 4 S 6 7 e 9 IO II 12 13 M 15 16 17 18 19 2ODAYS

FIG. 10(From E. C. Tolman, Prediction of vicarious trial and error by means of the schematic sow-

bug. PSYCHOI. REV., 1939, 46, p. 319.)

ferences by myself, Geier and Levin3

(actually done by Geier and Levin)using this same visual discriminationapparatus, it was found that with oneand the same difficulty of problem thesmarter animal did the more VTEing.

To sum up, in visual discriminationexperiments the better the learning, themore the VTE's. But this seems con-trary to what we would perhaps haveexpected. We ourselves would expectto do more VTEing, more sampling ofthe two stimuli, when it is difficult tochoose between them than when it iseasy.

3 F. M. Geier, M. Levin & E. C. Tolman,Individual differences in emotionality, hy-pothesis formation, vicarious trial and errorand visual discrimination learning in rats.Compar. Psychoi. Monogr., 1941, 17, No. 3.

What is the explanation? The an-swer lies, I believe, in the fact that themanner in which we set the visual dis-crimination problems for the rats andthe manner in which we set similarproblems for ourselves are different.We already have our 'instructions.' Weknow beforehand what it is we are todo. We are told, or we tell ourselves,that it is the lighter of the two grays,the heavier of the two weights, or thelike, which is to be chosen. In such asetting we do more sampling, moreVTEing, when the stimulus-difference issmall. But for the rats the usual prob-lem in a discrimination apparatus isquite different. They do not knowwhat is wanted of them. The majorpart of their learning in most such ex-periments seems to consist in their dis-

COGNITIVE MAPS IN RATS AND M E N 199

covering the instructions. The ratshave to discover that it is the differ-ences in visual brightness, not the dif-ferences between left and right, whichthey are to pay attention to. TheirVTEing appears when they begin to'catch on.' The greater the differencebetween the two stimuli the more theanimals are attracted by this difference.Hence the sooner they catch on, andduring this catching on, the more theyVTE.

That this is a reasonable interpreta-tion appeared further, from an experi-ment by myself and Minium (the actualwork done, of course, by Minium) inwhich a group of six rats was firsttaught a white vs. black discrimination,then two successively more difficult gray

vs. black discriminations. For eachdifficulty the rats were given a long se-ries of further trials beyond the pointsat which they had learned. Comparingthe beginning of each of these threedifficulties the results were that the ratsdid more VTEing for the easy discrimi-nations than for the more difficult ones.When, however, it came to a compari-son of amounts of VTEing during thefinal performance after each learninghad reached a plateau, the opposite re-sults were obtained. In other words,after the rats had finally divined theirinstructions, then they, like humanbeings, did more VTEing, more sam-pling, the more difficult the discrimina-tion.

Finally, now let us note that it was

AVERAGE NO. OF VTE'SPER DAY

18

17

16

15

14

13

12

I I

K?6

5

432I

BLACK

MEDIUM GREY

LIGHT GREY

1 2 3 4 5 S 7 8 9 IO II 12 13 14 15 16 17 18 19 20DAYS

FIG. 11

(From E. C. Tolman, Prediction of vicarious trial and error by means of the schematic sow-bug. PSYCHOI.. REV., 1939, 46, p. 320.)


iec

(joe

uillC

iac

a5

I

FIG. 12

(From E C. Tolman and E. Minium, VTE in rats: overlearning and difficulty of discrimina-tion. / . comp. Psychol., 1942, 34, p. 303.)

also found at Berkeley by Jackson4

that in a maze the difficult maze unitsproduce more VTEing and also that themore stupid rats do the more VTEing.The explanation, as I see it, is that, inthe case of mazes, rats know their in-structions. For them it is natural toexpect that the same spatial path willalways lead to the same outcome. Ratsin mazes don't have to be told.

But what, now, is the final signifi-cance of all this VTEing? How dothese facts about VTEing affect ourtheoretical argument? My answer isthat these facts lend further support tothe doctrine of a building up of maps.VTEing, as I see it, is evidence that inthe critical stages—whether in the firstpicking up of the instructions or in thelater making sure of which stimulus iswhich—the animal's activity is not justone of responding passively to discretestimuli, but rather one of the active se-

4 L. L. Jackson, V. T. E. on an elevatedmaze. / . comp. Psychol, 1943, 36, 99-107.

lecting and comparing of stimuli. Thisbrings me then to the third type of ex-periment.

(3) "Searching for the Stimulus." Irefer to a recent, and it seems to me ex-tremely important experiment, done fora Ph.D. dissertation by Hudson. Hud-son was first interested in the questionof whether or not rats could learn anavoidance reaction in one trial. Hisanimals were tested one at a time in aliving cage (see Fig. 13) with a smallstriped visual pattern at the end, onwhich was mounted a food cup. Thehungry rat approached this food cupand ate. An electrical arrangement wasprovided so that when the rat touchedthe cup he could be given an electricshock. And one such shock did appearto be enough. For when the rat wasreplaced in this same cage days or evenweeks afterwards, he usually demon-strated immediately strong avoidancereactions to the visual pattern. Theanimal withdrew from that end of the


FIG 13(From Bradford Hudson. Ph.D. Thesis:

'One trial learning: A study of the avoidancebehavior of the rat' On deposit in the Li-brary of the University of California, Berke-ley, California )

cage, or piled up sawdust and coveredthe pattern, or showed various otheramusing responses all of which were inthe nature of withdrawing from thepattern or making it disappear.

But the particular finding which Iam interested in now appeared as a re-sult of a modification of this standardprocedure. Hudson noticed that theanimals, anthropomorphically speaking,often seemed to look around after theshock to see what it was that had hitthem. Hence it occurred to him that,if the pattern were made to disappearthe instant the shock occurred, the ratsmight not establish the association.And this indeed is what happened inthe case of many individuals. Hudsonadded further electrical connections sothat when the shock was received dur-ing the eating, the lights went out, the

pattern and the food cup dropped outof sight, and the lights came on againall within the matter of a second.When such animals were again put inthe cage 24 hours later, a large per-centage showed no avoidance of thepattern. Or to quote Hudson's ownwords:

'"Learning what object to avoid . . . mayoccur exclusively during the period afterthe shock. For if the object from whichthe shock was actually received is re-moved at the moment of the shock, a sig-nificant number of animals fail to learn toavoid it, some selecting other features inthe environment for avoidance, and othersavoiding nothing."

In other words, I feel that this ex-periment reinforces the notion of thelargely active selective character in therat's building up of his cognitive map.He often has to look actively for thesignificant stimuli in order to form hismap and does not merely passively re-ceive and react to all the stimuli whichare physically present.

Turn now to the fourth type of ex-periment.

(4) The "Hypothesis" Experiments.Both the notion of hypotheses in ratsand the design of the experiments todemonstrate such hypotheses are to becredited to Krech. Krech used a four-compartment discrimination-box. Insuch a four-choice box the correct doorat each choice-point may be determinedby the experimenter in terms of itsbeing lighted or dark, left or right, orvarious combinations of these. If all

FIG. 14(From I. Krechevsky (Now D. Krech), The genesis of "hypotheses" in rats

Calif. Publ. Psychol, 1932, 6, No. 4, p. 46.)Univ.


Apparatut uted in preliminary training

FIG. 15

(From E C. Tolman, B. F. Ritchie and D. Kalish, Studies inspatial learning. I. Orientation and the short-cut. / . exp. Psy-chol., 1946, 36, p. 16.) •

possibilities are randomized for the 40choices made in 10 runs of each day'stest, the problem could be made in-soluble.

When this was done, Krech foundthat the individual rat went through asuccession of systematic choices. Thatis, the individual animal might perhapsbegin by choosing practically all right-hand doors, then he might give this upfor choosing practically all left-handdoors, and then, for choosing all darkdoors, and so on. These relativelypersistent, and well-above-chance sys-tematic types of choice Krech called

"hypotheses." In using this term heobviously did not mean to imply verbalprocesses in the rat but merely referredto what I have been calling cognitivemaps which, it appears from his experi-ments, get set up in a tentative fashionto be tried out first one and then an-other until, if possible, one is foundwhich works.

Finally, it is to be noted that thesehypothesis experiments, like the latentlearning, VTE, and "looking for thestimulus" experiments, do not, as such,throw light upon the widths of the mapswhich are picked up but do indicate


the generally map-like and self-initiatedcharacter of learning.

For the beginning of an attack uponthe problem of the width of the mapslet me turn to the last group of experi-ments.

(5) "Spatial Orientation" Experi-ments. As early as 1929, Lashley re-ported incidentally the case of a coupleof his rats who, after having learned analley maze, pushed back the cover nearthe starting box, climbed out and randirectly across the top to the goal-boxwhere they climbed down in again andate. Other investigators have reportedrelated findings. All such observationssuggest that rats really develop widerspatial maps which include more thanthe mere trained-on specific paths. Inthe experiments now to be reported this

possibility has been subjected to fur-ther examination.

In the first experiment, Tolman,Ritchie and Kalish (actually Ritchieand Kalish) used the set-up shown inFig. 15.

This was an elevated maze. The ani-mals ran from A across the open circu-lar table through CD (which had alleywalls) and finally to G, the food box.H was a light which shone directly downthe path from G to F. After fournights, three trials per night, in whichthe rats learned to run directly andwithout hesitation from A to G, theapparatus was changed to the sun-burstshown in Fig. 16. The starting pathand the table remained the same but aseries of radiating paths was added.

The animals were again started at A

Apparatus wed in the tett trial

FIG. 16

(From E. C. Tolman, B. F. Ritchie and D. Kalish, Studies in spatiallearning. I. Orientation and short-cut. / . exp. Psychol., 1946, 36,P. 17.)


and ran across the circular table intothe alley and found themselves blocked.They then returned onto the table andbegan exploring practically all the radi-ating paths. After going out a fewinches only on any one path, each ratfinally chose to run all the way out onone. The percentages of rats finallychoosing each of the long paths from 1to 12 are shown in Fig. 17. It appears

Numbers o{ rtti which cho« eich of the p»thi

FIG. 17

(From E. C. Tolman, B. F. Ritchie and D.Kalish, Studies in spatial learning. I. Ori-entation and the short-cut J. exp. Psyckol.,1946, 36, p. 19.)

that there was a preponderant tendencyto choose path No. 6 which ran to apoint some four inches in front of wherethe entrance to the food-box had been.The only other path chosen with anyappreciable frequency was No. 1—thatis, the path which pointed perpendicu-larly to the food-side of the room.

These results seem to indicate that

the rats in this experiment had learnednot only to run rapidly down the origi-nal roundabout route but also, when thiswas blocked and radiating paths pre-sented, to select one pointing rather di-rectly towards the point where the foodhad been or else at least to select a pathrunning perpendicularly to the food-side of the room.

As a result of their original training,the rats had, it would seem, acquirednot merely a strip-map to the effect thatthe original specifically trained-on pathled to food but, rather, a wider compre-hensive map to the effect that food waslocated in such and such a direction inthe room.

Consider now a further experimentdone by Ritchie alone. This experimenttested still further the breadth of thespatial map which is acquired. In thisfurther experiment the rats were againrun across the table—this time to thearms of a simple T. (See Fig. 18.)

Twenty-five animals were trained forseven days, 20 trials in all, to find foodat Fi; and twenty-five animals weretrained to find it at F2. The L's in thediagram indicate lights. On the eighthday the starting path and table topwere rotated through 180 degrees sothat they were now in the positionshown in Fig. 19. The dotted linesrepresent the old position. And a se-ries of radiating paths was added.What happened? Again the rats ranacross the table into the central alley.When, however, they found themselvesblocked, they turned back onto thetable and this time also spent manyseconds touching and trying out foronly a few steps practically all thepaths. Finally, however, within sevenminutes, 42 of the SO rats chose onepath and ran all the way out on it.The paths finally chosen by the 19 ofthese animals that had been fed at Fx

and by the 23 that had been fed at F2

are shown in Fig. 20.


FIG. 18(From B. F. Ritchie. Ph.D. Thesis: 'Spatial learning in rats' On

deposit in the Library of the University of California, Berkeley, Cali-fornia.)

This time the rats tended to choose,not the paths which pointed directly tothe spots where the food had been, butrather paths which ran perpendicularlyto the corresponding sides of the room.The spatial maps of these rats, whenthe animals were started from the op-posite side of the room, were thus notcompletely adequate to the precise goalpositions but were adequate as to thecorrect sides of the room. The maps ofthese animals were, in short, not alto-gether strip-like and narrow.

This completes my report of experi-ments. There were the latent learningexperiments, the VTE experiments, thesearching for the stimulus experiment,the hypothesis experiments, and theselast spatial orientation experiments.

And now, at last, I come to the hu-manly significant and exciting problem:namely, what are the conditions whichfavor narrow strip-maps and what arethose which tend to favor broad com-prehensive maps not only in rats butalso in men?

There is considerable evidence scat-tered throughout the literature bearingon this question both for rats and formen. Some of this evidence was ob-tained in Berkeley and some of it else-where. I have not time to present it inany detail. I can merely summarize itby saying that narrow strip maps ratherthan broad comprehensive maps seemto be induced: (1) by a damaged brain,(2) by an inadequate array of environ-mentally presented cues, (3) by an

206 EDWAED C. TOLMAN

FIG. 19(From B. F. Ritchie. Ph.D. Thesis: 'Spatial learning in rats'

On deposit in the Library of the University of California, Berke-ley, California.)

1 5 -

10 -

5-

N-1915-

•0%

1I 2 3 4 5 6 7 8 9 10

, Chosen by Rots in the F, Group

10-

5 -

N-2J

I 2 3 4 5 6 7 8 9 10

Poths Chosen by Rots in the F, Group

FIG. 20

(From B. F. Ritchie. Ph.D. Thesis- 'Spatial learning in rats' On deposit in the Library ofthe University of California, Berkeley, California.)


overdose of repetitions on the originaltrained-on path and (4) by the pres-ence of too strongly motivational or oftoo strongly frustrating conditions.

It is this fourth factor which I wishto elaborate upon briefly in my conclud-ing remarks. For it is going to bemy contention that some, at least, ofthe so-called 'psychological mechanisms'which the clinical psychologists and theother students of personality have un-covered as the devils underlying manyof our individual and social maladjust-ments can be interpreted as narrowingsof our cognitive maps due to too strongmotivations or to too intense frustra-tion.

My argument will be brief, cavalier,and dogmatic. For I am not myself aclinician or a social psychologist. WhatI am going to say must be considered,therefore, simply as in the nature of arat psychologist's ratiocinations offeredfree.

By way of illustration, let me sug-gest that at least the three dynamismscalled, respectively, "regression," "fixa-tion," and "displacement of aggressiononto outgroups" are expressions of cog-nitive maps which are too narrow andwhich get built up in us as a result oftoo violent motivation or of too intensefrustration.

(a) Consider regression. This is theterm used for those cases in which anindividual, in the face of too difficult aproblem, returns to earlier more child-ish ways of behaving. Thus, to take anexample, the overprotected middle-agedwoman (reported a couple of years agoin Time Magazine) who, after losingher husband, regressed (much to thedistress of her growing daughters) intodressing in too youthful a fashion andinto competing for their beaux and thenfinally into behaving like a child re-quiring continuous care, would be anillustration of regression. I would notwish you to put too much confidence

in the reportorial accuracy of Time, butsuch an extreme case is not too differentfrom many actually to be found in ourmental hospitals or even sometimes inourselves. In all such instances myargument would be (1) that such re-gression results from too strong a pres-ent emotional situation and (2) that itconsists in going back to too narrow anearlier map, itself due to too much frus-tration or motivation in early childhood.Time's middle-aged woman was pre-sented by too frustrating an emotionalsituation at her husband's death andshe regressed, I would wager, to toonarrow adolescent and childhood mapssince these latter had been originallyexcessively impressed because of over-stressful experiences at the time she wasgrowing up.

(b) Consider fixation. Regressionand fixation tend to go hand in hand.For another way of stating the fact ofthe undue persistence of early maps isto say that they were fixated. This haseven been demonstrated in rats. Ifrats are too strongly motivated in theiroriginal learning, they find it very diffi-cult to relearn when the original path isno longer correct. Also after they haverelearned, if they are given an electricshock they, like Time's woman, tend toregress back again to choosing theearlier path.

(c) Finally, consider the "displace-ment oj aggressions onto outgroups."Adherence to one's own group is anever-present tendency among primates.It is found in chimpanzees and mon-keys as strongly as in men. We pri-mates operate in groups. And eachindividual in such a group tends toidentify with his whole group in thesense that the group's goals become hisgoals, the group's life and immortality,his life and immortality. Furthermore,each individual soon learns that, whenas an individual he is frustrated, hemust not take out his aggressions on


the other members of his own group.He learns instead to displace his ag-gressions onto outgroups. Such a dis-placement of aggression I would claimis also a narrowing of the cognitive map.The individual comes no longer to dis-tinguish the true locus of the cause ofhis frustration. The poor Southernwhites, who take it out on the Negroes,are displacing their aggressions fromthe landlords, the southern economicsystem, the northern capitalists, orwherever the true cause of their frus-tration may lie, onto a mere convenientoutgroup. The physicists on the Fa-culty who criticize the humanities, orwe psychologists who criticize all theother departments, or the University asa whole which criticizes the SecondarySchool system or, vice versa, the Sec-ondary School system which criticizesthe University—or, on a still larger andfar more dangerous scene—we Ameri-cans who criticize the Russians and theRussians who criticize us, are also en-gaging, at least in part, in nothing morethan such irrational displacements ofour aggressions onto outgroups.

I do not mean to imply that theremay not be some true interferences bythe one group with the goals of theother and hence that the aggressions ofthe members of the one group againstthe members of the other are neces-sarily wholly and merely displaced ag-gressions. But I do assert that oftenand in large part they are such meredisplacements.

Over and over again men are blindedby too violent motivations and too in-tense frustrations into blind and unin-telligent and in the end desperately dan-gerous hates of outsiders. And the ex-pression of these their displaced hatesranges all the way from discriminationagainst minorities to world conflagra-tions.

What in the name of Heaven andPsychology can we do about it? Myonly answer is to preach again the vir-tues of reason—of, that is, broad cog-nitive maps. And to suggest that thechild-trainers and the world-planners ofthe future can only, if at all, bringabout the presence of the required ra-tionality (i.e., comprehensive maps) ifthey see to it that nobody's children aretoo over-motivated or too frustrated.Only then can these children learn tolook before and after, learn to see thatthere are often round-about and saferpaths to their quite proper goals—learn,that is, to realize that the well-beings ofWhite and of Negro, of Catholic and ofProtestant, of Christian and of Jew, ofAmerican and of Russian (and even ofmales and females) are mutually inter-dependent.

We dare not let ourselves or othersbecome so over-emotional, so hungry, soill-clad, so over-motivated that onlynarrow strip-maps will be developed.All of us in Europe as well as in Amer-ica, in the Orient as well as in the Oc-cident, must be made calm enough andwell-fed enough to be able to developtruly comprehensive maps, or, as Freudwould have put it, to be able to learnto live according to the Reality Prin-ciple rather than according to the toonarrow and too immediate PleasurePrinciple.

We must, in short, subject our chil-dren and ourselves (as the kindly ex-perimenter would his rats) to the op-timal conditions of moderate motiva-tion and of an absence of unnecessaryfrustrations, whenever we put them andourselves before that great God-given'maze which is our human world. Icannot predict whether or not we willbe able, or be allowed, to do this; butI can say that, only insofar as we areable and are allowed, have we cause forhope.

the psychological review - semantic scholar › 0874 › a64d60a23a20303877... · 2018-10-09 ·...

Documents