VALUE AND NEED AS ORGANIZING FACTORS IN PERCEPTIONBY JEROME S. BRUNER AND CECILE C. GOODMAN •

Harvard University

THROUGHOUT the history of modernpsychology, until very recenttimes, perception has been treated

as though the perceiver were a passiverecording instrument of rather complexdesign. One might, in most experi-ments, describe him in much the samegraphical terms as one uses to describedie latest piece of recording apparatusobtainable from Stoelting or the Ameri-can Optical Company. Such psychol-ogy, practiced as it were in vitro, hasfallen short of clarifying the nature ofperception in everyday life much as didthe old nerve-muscle psychophysiologyfall short of explaining behavior ineveryday life. Both have been monu-mentally useful—in their place. Thenames of Weber, Fechner, Wundt,Titchener, Hecht, and Crozier aresafely ensconced in any respectablepsychological hall of fame. But theirwork, like the work of the nerve-muscle men, is only a beginning.

For, as Professor Thurstone (35) hasput it, "In these days when we insist sofrequently on the interdependence ofall aspects of personality, it would bedifficult to maintain that any of thesefunctions, such as perception, is isolatedfrom the rest of the dynamical systemthat constitutes the person." The prob-lem is, indeed, to understand how theprocess of perception is affected by otherconcurrent mental functions and howthese functions in their turn are affectedby the operation of perceptual proc-

• The writers are greatly indebted to Pauline B.Hahn and Dr. Leo J. Postman for invaluableassistance and advice.

esses. Given a dark room and a highlymotivated subject, one has no difficultyin demonstrating Korte's Laws of phe-nomenal movement. Lead the subjectfrom the dark room to the market placeand then find out what it is he seesmoving and under what conditions, andKorte's Laws, though still valid, de-scribe the situation about as well as theLaws of Color Mixture describe one'sfeelings before an El Greco canvas.

The discrepancy between the darkroom and the market place we have inthe past found it convenient to dismissby invoking various dei ex machina:Attention, Apperception, UnbewussterSchluss, Einstellung, Preparatory Set,etc, Like the vengeful and unan-nounced step-brother from Australia inthe poorer murder mysteries, they turnup at the crucial juncture to do thedirty work. Though such constructsare useful, perception itself must re-main the primary focus. To shiftattention away from it by invokingpoorly understood intervening variablesdoes little service. What we must studybefore invoking such variables are thevariations perception itself undergoeswhen one is hungry, in love, in pain,or solving a problem. These variationsare as much a part of the psychologyof perception as Korte's Laws.

It is the contention of this paper thatsuch perceptual phenomena are asscientifically measurable in terms ofappropriate metrics as such more hal-lowed phenomena as flicker fusion,constancy, or tonal attributes. But let

33

34 JEROME S. BRUNER AND CECILE C. GOODMAN

us pause first to construct a sketchyterminology. Let us, in what ensues,distinguish heuristically between twotypes of perceptual determinants. Thesewe shall call autochthonous and be-havioral. Under the former we groupthose properties of the nervous system,highly predictable, which account forphenomena like simple pair formation,closure, and contrast, or at anotherlevel, tonal masking, difference andsummation tones, flicker fusion, para-doxical cold, and binaural beats. Givenideal "dark-room" conditions and nocompelling distractions, the "average"organism responds to set physicalstimuli in these relatively fixed ways.Autochthonous determinants, in brief,reflect directly the characteristic electro-chemical properties of sensory endorgans and nervous tissue.

Under the category of behavioraldeterminants we group those active,adaptive functions of the organismwhich lead to the governance and con-trol of all higher-level functions, includ-ing perception: the laws of learningand motivation, such personality dy-namics as repression, the operation ofquasi-temperamental characteristics likeintroversion and extraversion, socialneeds and attitudes, and so on. Under-lying these behavioral determinants,doubtless, are a host of physiologicalmechanisms. But we can hardly waituntil we understand these before tack-ling experimentally the role of be-havioral determinants in perception.The physiology of Weber's Law is stillmore or less obscure, yet the enunciationof it has been recognizably useful—even to the physiologist for whom it hasbeen a challenge to discovery.

A paper of this kind dannot containany extensive review of the literatureon those perceptual dynamics which wehave called behavioral. Yet it is neces-sary to pass rapidly over some of the

notable facts and experiments whichhave forced us to draw certain distinc-tions and make bold claims about themensurability of behavioral determi-nants. First we have the facts of"sensory conditioning," a term firstused by Cason(8). Starting with thework of Perky in 1910 (24), it has beendemonstrated repeatedly by WarnerBrown (5), Ellson (u), Coffin (9), andothers that subjects can be conditionedto see and hear things in much -thesame way as they can be conditioned toperform such overt acts as knee jerk-ing, eye blinking, or salivating. Pair asound and a faint image frequentlyenough, fail to present the image, andthe subject sees it anyway when thesound is presented. Any student of sug-gestion, whether or not he has perusedBird's exhaustive bibliography (3) ofthe literature on the subject, knows that.Not perception? Why not? The sub-ject sees what he reports as vividly ashe sees the phi-phenomenon.

Closely related are such experimentsas those of Haggard and Rose (16),Proshansky and Murphy (27), andSchafer and Murphy (31) demonstrat-ing the role of reward and punishmentin altering perceptual organization.Haggard and Rose show that the extentof autokinetic movement can be alteredby a system of rewards; Proshanskyand Murphy that discriminable differ-ences in the perception of lines andweights can be similarly altered;Schafer and Murphy that, given anambiguous figure-ground configuration,what is seen as figure and what asground can be altered by a system ofreward and punishment.

Another group of researches hasdemonstrated that what is seen in acomplex configuration is not deter-mined solely by the laws of gestalt, butby practice. Among experimenterswho have confirmed this generalization

VALUE AND NEED AS ORGANIZING FACTORS IN PERCEPTION

are Henle(i7), Fehrer(i4), Braly (4),Leepcr(2o), and Djang(io). Closelyrelated are the experiments of Thou-less(34), showing that phenomenalconstancy or, as he calls it, "regressionto the real object," reflects the habits ofthe individual. Art students, forexample, see the "real" object—its color,shape, and brightness — less readily,show greater phenomenal constancy,than matched individuals with no arttraining. Indeed, v. Fieandt (15) hasshown that the appearance of a surfaceas light gray in shadow or dark grayin light can be controlled by simplePavlovian conditioning, the CS beinga sound or a button in the visualfield. And all of us are fond of citingthe work of Haddon in the TorresStraits (28) demonstrating that theseprimitive island spear-fishers are, mostlikely as a result of their experiencewith spears, considerably less susceptibleto the Muller-Lyer illusion.

Sherif's classic experiments (32) onsocial factors are too well known to needany elucidation here. Demonstratingfurther the role of social factors in per-ception are the experiments of Zuk-Kardos (36) and Fazil (13), students ofEgon Brunswik, who showed that thesubjective number equation for match-ing a standard cluster of stamps or coinsto a variable depended in part upon thevalue of the coins or stamps in thestandard and variable clusters. Withmany refinements and extensions, theseexperiments have been repeated inAmerica by Ansbacher (i).

One can go on to cite many moreexperiments, but in a very brief sum-mary review that would be impossible.Let us conclude then with two piecesof research, one French, the otherSwiss, indicating the possible connec-tion of general personality traits andperception. Binet (2) and Meili andTobler (22) have suggested that the

35

child is more susceptible to "illusions,"more a prey to those organizing factorswhich, as adults, we call distorting.Binet has shown that, as the child growsolder, his susceptibility to the Muller-Lyer illusion decreases. The contribu-tion of Meili and Tobler has been toshow that, as the child ages, histhreshold for seeing stroboscopic move-ment becomes higher. Whether fromthese two experiments, plus such inci-dental observations as Piaget's (25) tothe effect that the child sees the moonas following him, we can draw anyconclusions about increasing "percep-tual realism" as a function of age isopen to question. Yet the way hasbeen opened to those who wish to in-vestigate this area further.

So much for prior research. Thereexists a fruitful if slim body of litera-ture on behavioral factors in perception.Where does one go from here? Twoapproaches are open. Armed with ourslender reed of empirical proof, we canset about the task of systematization,indulging in S-R's, topology, or psycho-analytic constructs to suit the taste.There is already one brilliant theoreti-cal structure to account for many of thefacts we have been discussing, presentedin Egon Brunswik's Wahrnehmungund Gegenstandswelt (7). Or we maygo on to the empirical demonstration ofgeneral hypotheses concerning the rela-tion of behavior dynamics and percep-tion. Both are indispensable activities.The present paper, however, is con-cerned mainly with empirical hypothe-ses. But certain minimum systematicassumptions must first be made clear tobring these hypotheses into clear focus.

The organism exists in a world ofmore or less ambiguously organizedsensory stimulation. What the organ-ism sees, what is actually there percep-tually represents some sort of compro-mise between what is presented by

JEROME S. BRUNER AND CECILE C. GOODMAN

autochthonous processes and what isselected by behavioral ones. Suchselection, we know, is determined notonly by learning, as already indicated,but also by motivational factors such ashave been indicated for hunger bySanford (29, 30) and Levine, Chein,and Murphy (21). The selective proc-ess in perception we shall refer to as aperceptual hypothesis, using the term,with Krech (19), to denote a systematicresponse tendency. Such an hypothesismay be set into operation by a need, bythe requirements of learning a task, orby any internally or externally imposeddemands on the organism. If a givenperceptual hypothesis is rewarded byleading to food, water, love, fame, orwhat not, it will become fixated; andthe experimental literature, notably thework of Ellson (12) and Leeper(2o),indicates that the fixation of "sensoryconditioning" is very resistant to extinc-tion. As fixation takes place, the per-ceptual hypothesis grows stronger notonly in the sense of growing more fre-quent in the presence of certain typesof stimulation, but also more percep-tually accentuated. Perceptual objectswhich are habitually selected becomemore vivid, have greater clarity orgreater brightness or greater apparentsize.

Two other systematic matters mustconcern us before we turn to the experi-ments. One has to do with perceptualcompromise, the other with perceptualequivocality. Frequently, alternativehypotheses operate: a quick glimpse ofa man in gray on a European battlefieldmay leave us in doubt as to whether heis a civilian or a Wehrmacht infantry-man. Almost inevitably one or theother hypothesis prevails, and the fieldis perceived as either one or the other.But in spite of the dominance of asingle hypothesis in perception, compro-

mise also occurs. Using Ansbacher'sexperiments (i) as an example, a groupof small paper squares is seen both interms of number and in terms of valueas stamps. What results, if you will, isa perception of "number-value." Weknow precious little about such percep-tual compromises, although we shall bediscussing experiments demonstratingtheir operation.

As for equivocality, or ambiguity inthe perceptual field, it has generallybeen supposed that the greater theequivocality the greater the chance forbehavioral factors in perception tooperate, all other things being equal.Sheriff) chose the autokinetic phe-nomenon to work with for this reason.Proshansky and Murphy (27) workedclose to threshold illumination withsimilar intent. Within broad limits,which we shall discuss, the generaliza-tion is valid, in so far as equivocalityreduces the organizing capacity ofautochthonous perceptual determinants.How important this generalization iswe, who think so exclusively in termsof the well-controlled dark-room experi-ment, often forget. For in everydaylife, perception is, by and large, a seriesof quick looks, glances, inattentivelistenings, furtive touches. Save forwhat is at the very focus of interestedattention, the world of sense is moreequivocal than our textbook writersseem to think.

EMPIRICAL HYPOTHESESWe may turn now to the experiments

with which this paper is primarilyconcerned. Three general hypotheses,growing out of the systematic principlesjust presented, are under consideration,

i. The greater the social value of anobject, the more will it be susceptible toorganization by behavioral determi-nants. It will be selected perceptuallyfrom among alternative perceptual ob-jects, will become fixated as a perceptual

VALUE AND NEED AS ORGANIZING FACTORS IN PERCEPTION 37

response tendency, and will becomeperceptually accentuated.

2. The greater the individual needfor a socially valued object, the moremarked will be the operation of be-havioral determinants.

3. Perceptual equivocality will facili-tate the operation of behavioral deter-minants only in so jar as equivocalityreduces the operation of autochthonousdeterminants without reducing theeffectiveness of behavioral determinants.

In the experiments reported here, onlyone aspect of behavioral determinationwill be treated, what we have calledaccentuation—the tendency for sought-after perceptual objects to become morevivid. Perceptual selectivity and fixa-tion have already been demonstrated inother experiments, though they remainpoorly systematized. For purposes ofeconomy of exposition we omit con-sideration of them here, though theyconstitute important variables in thebroader research project of which thepresent experiments are a part.

THE SUBJECTS AND THE APPARATUS

The subjects were 30 ten-year-oldchildren of normal intelligence, divisibleaccording to certain characteristics tobe discussed shortly into three groups,two experimental and one control. Theapparatus consisted of a rectangularwooden box (9"x9"xi8") at one endof which was a 5" square ground-glassscreen and a knob at its lower right-hand corner. At the center • of theground-glass screen was an almost cir-cular patch of light (16.2 app. ft. cdls.)cast upon the back of the screen by a6o-watt incandescent light shiningthrough an iris diaphragm which couldbe varied in diameter from %" to 2"by turning the knob on the front endof the box. All that was visible to thesubject was the box with its ground-glass screen and the circle of light whose

diameter he could change by turningthe knob. The circle was not trulyround, containing the familiar nineelliptoid sides found in the Bausch &Lomb iris diaphragm. It was so closeto round, however, that subjects had nodifficulty making the subjective equa-tions required of them.

Subjects individually sat in a chair infront of the screen on the box with thelight circle slightly below eye level. Thebox rested on a table behind which satthe experimenter. The child was toldthat this was a game, and that he wasto make the circle of light on the boxthe same size as various objects he wasshown or told about. Before beginningjudgments, each child, with no urging,was encouraged to see how large andsmall the circle of light could be made.

The two experimental groups re-ceived the same treatment. Two serieswere run for these groups, comprising20 of the children in all. First the childwas asked to estimate the sizes of coinsfrom a penny through a half dollarfrom memory. He did the first inascending order of value, then in de-scending order, always making twojudgments for each coin named, onefrom the open, the other from theclosed position of the iris diaphragm.Four judgments were made for eachcoin by each child. No inkling wasgiven the child as to how "close" he hadcome.

Following the memory series, andusing the same order of presentation, asimilar series was then run with coinspresent. Coins, individually, were heldclose to the center of the palm of theleft hand, at a level with the light circleand six inches to its left. The subjectstook as much time as suited them.

A control group of ten subjects fol-lowed a procedure identical with theone just described. Instead of coins,medium gray cardboard discs of identi-

JEROME S. BRUNER AND CECILE C. GOODMAN

+55

© © coins0.—.0 discs

50<:FIG. i. SIZE ESTIMATIONS OF COINS AND Discs OF SAME SIZE MADE BY TEN-YEAR-OLDS

(Method of average error)

cal size were employed. No mentionof money was made to this group.

RESULTSLet us compare the difference be-,

tween judgments of size of coins andidentically sized cardboard discs. Twothings can be noted,in Figure i, whichpresents judgments of experimentalsand controls with coins present. First

off, coins, socially valued objects, arejudged larger in size than gray discs.Secondly, the greater the value of thecoin, the greater is the deviation ofapparent size from actual size. Theexception to this generalization is thehalf dollar, overestimation of which fallsoff below that of a quarter. By way ofthe sheerest guess one might explainthis reversal of the curve in terms of

VALUE AND NEED AS ORGANIZING FACTORS IN PERCEPTION 39the lesser reality-value of a half dollaras compared with a quarter for the ten-year-old. A half dollar at that age is,so to speak, almost too valuable to bereal! More likely there is some simpleautochthonous reason for the reversal.Yet, no such reversal is found in curvesplotted for adults.

The difference between experimen-tals and controls is, of course, highlysignificant. The variance in overesti-mation in the experimental groups in-troduced by using coins of differentvalue is similarly significant. Our re-sults, as handled by the Postman-Bruner (26) adaptation of the analysisof variance to psychophysical data, showthat variances due to coin value anddue to using discs versus coins yieldF-scores convertible to P-values of lessthan .01.l

So much for the first hypothesis, thatsocially valued objects are susceptible tobehavioral determinants in proportionto their value. Consider now the secondhypothesis, that the greater the subjec-tive need for a socially valued object,the greater will be the role of behavioraldeterminants of perception. In the sec-ond experimental variation, the experi-mental group was divided into twocomponent groups. One we call therich group, the other the poor group,each comprising ten subjects. Well-to-do subjects were drawn from aprogressive school in the Boston area,catering to the sons and daughters of

1 P-values at the .01 level were also found forconstant errors introduced by ascending anddescending value orders and for judgments madefrom the open and closed positions of thediaphragm. Since these parameters were con-trolled and balanced in the judgment data for thegroups discussed, nothing further need be saidof them here. They will be discussed in anotherplace (6). Analysis of variance was carried outboth with percentage scores representing deviationof individual judgments from actual size and withraw scores. Necessary corrections suggested bySnedecor (33) were used in the former method.The values presented here are applicable to bothraw and percentage scores.

prosperous business and professionalpeople. The poor subjects came froma settlement house in one of Boston'sslum areas. The reasonable assumptionis made that poor children have agreater subjective need for money thanrich ones. When the figures presentedin Figure i are broken down into scoresfor rich and poor groups, a striking dif-ference will be noted (Figure 2). Thepoor group overestimates the size ofcoins considerably more than does therich. Again there are some irregulari-ties in the curves. The drop-off for thehalf dollar we have already sought toexplain. As for the dip in the richgroup's curve at a dime, the explanationis problematical. All curves which wehave plotted for adults—and by now wehave collected more than two thousandjudgments (6)—show this dip. Perhapsit is due to the discrepancy between therelative size and value of the dime,perhaps to some inherent characteristicof the coin itself.2

The difference between rich and pooris highly significant, analysis of vari-ance showing that the source of varianceis significant beyond the P level of .01.Our second hypothesis cannot, then, berejected. It is notable too that the inter-action between the parameters of eco-nomic status and value of coins yieldsan F-score convertible to a P-value be-tween .05 and .01 which leads to asecondary hypothesis: given perceptualobjects of the same class but varying invalue, the effect of need for that classof objects will be to accentuate the mostvaluable objects most, the least valuableleast, etc.

What of ambiguity or perceptualequivocality? We have arbitrarily

2 If the reader is a smoker, let him ask himselfwhether a dime will cover the hump on the camelwhich appears as a trademark on Camel ciga-rettes. Hold the two six inches apart. In spiteof the apparently small size of the coin, it willcover the camel's hump with margin to spare.

JEROME S. BRUNER AND CECILE C. GOODMAN

assumed that a situation in which oneis judging size from memory is more"equivocal" than one in which the ob-ject being judged is in clear view sixinches away from the test patch. Theassumption is open to serious question,but let us examine what follows fromit experimentally. Compare first thejudgments of the rich group under con-ditions like those described: with coin

present as compared with coin as a merememory image. The curves are inFigure 3. It would seem that, for allvalues below a quarter, equivocality hasthe effect of making judgments con-form more to actual size, aiding, inother words, the operation of autoch-thonous determinants. For values overa quarter, equivocality favors behavioralfactors, making apparent size diverge

Rfch 10-yr. ddsPoor lO-yr olds

Fio. 2. SIZE ESTIMATIONS OF COINS MADE BY WELL-TO-DO AND POOR TEN-YEAR-OLDJ(Method of average error)

VALUE AND NEED AS ORGANIZING FACTORS IN PERCEPTION

+55

RICH GROUPCoin preserrt

—.0 Coin absent

FIG. 3. SIZE ESTIMATIONS op COINS WITH COINS PRESENT AND FROM MEMORY BYWELIXTO-DO TEN-YEAR-OLDS(Method of average error)

still more from actual size. For therich group, with coin present, a halfdollar is overjudged by 17.4 per cent;with coin absent, by 34.7 per cent.

This finding is difficult to interpretby itself. Consider now Figure 4, show-ing the discrepancy in "absent" and"present" judgments for the poor group.Here there is no crossing. Equivocality

seems, in this group, to have the ex-clusive effect of bringing judgmentsdown toward actual size. Equivocalityeven brings out the "dime dip" in thepoor group. How account for the dif-ference? Why does equivocality liberatebehavioral determinants among the richchildren for higher values, and depressthese factors for poor children? We can

JEROME S. BRUNER AND CECILE C. GOOOMAN

+55

POOR GROUPCoin presentCom absent

50CFIG. 4. SIZE ESTIMATIONS OF COINS WITH COINS PRESENT AND FROM MEMORY BY

POOR TEN-YEAR-OLDS(Method of average error)

offer nothing but a guess, one whichneeds confirmation by further research.Some years ago, Oeser (23) reportedthat in his study of children in Dundeehe found the fantasy life of the childrenof the unemployed strikingly chokedoff. Asked what they would like to bewhen grown, normal children of em-ployed parents gave such glamorous

replies as cowboy or film star, whilechildren of the unemployed named therather lowly occupations traditionallyfollowed by members of their class. Inthe figures just presented, it is our con-tention that we are witnessing the samephenomenon. In the case of the poorchildren, judging coin size from mem-ory, a weakened fantasy is substituted

VALUE AND NEED AS ORGANIZING FACTORS IN PERCEPTION 43for the compelling presence of a valuedcoin, while among rich children equivo-cality has the effect of liberating strongand active fantasy.3

Are any other explanations availableto account for the shape of the curveswe have been concerned with here?Weber's Law would predict in all casesa straight line plot parallel to the axisrepresenting actual size. DL should bea constant fraction of the stimulus,whatever its magnitude. If one were

thonous factor outweighed by the be-havioral determinants discussed in thecourse of this paper.

In conclusion, only one point needbe reiterated. For too long now, percep-tion has been virtually the exclusivedomain of the Experimental psycholo-gists with a capital E. If we are toreach an understanding of the way inwhich perception works in everydaylife, we social psychologists and stu-dents of personality will have to join

TABLE i

PERCENTAGE DEVIATION PROM ACTUAL SIZE OF JUDGMENTS op COINS AND Discs UNDERVARIOUS CONDITIONS

GROUP ANDCONDITION

20 O'scoin present

20 O'S

coin absent10 O's

disc present10 rich O's

coin present10 rich O's

coin absent10 poor O's

coin present10 poor O's

coin absent

PENNY

16.5

7.2

—5-4

10.3

2.6

22.7

i i . 8

NICKEL

23.9

19.6

— •9

20.4

19.8

27-3

19.4

DIME

29.1

11. 6

— i - 5

.6.3

7.8

41.8

15.4

QUARTER

37-0

32.8

1.8

22.4

28.3

51.6

37-3

HALF-DOLLAR

29.6

35.8

— .8

17.4

34-7

42.0

36.9

NUMBER JUDGMENTSPER COIN

80

80

40

40

40

40

40

to treat the slope of the curves by refer-ence to Hollingworth's central-tendencyeffect (18), one should find a negativerather than a positive slope. All valuessmaller than the center of the seriesshould appear larger in size; all largerthan the center of the series, smaller.Assuming that the Hollingworth effectis mediated by autochthonous factors,then it represents one more autoch-

3 The difference between rich and poor childrenin their size judgments of "absent" and "present"coins as here discussed is statistically significant.The interaction variance for these two parameters(economic status and presence-absence of coins) isat the .01 level of significance.

with the experimental psychologists andreexplore much of this ancient field ofperception whose laws for too long havebeen taken for granted.

BIBLIOGRAPHY

1. ANSBACHER, H. Perception of number asaffected by the monetary value of theobjects. Arch. Psychol., 1937, No. 215.

2. BINET, A. La mesure des illusions visuelleschez 1'enfant. Rev. phil., 1895, 40,11-25.

3. BIRD, C. Suggestion and suggestibility: abibliography. Psychol.' Bull., 1939, 36,264-283.

4. BUALY, K. W. The influence of past experi-ence in visual perception. /. exp. Psychol.,1933. 16. 613-643.

44 JEROME S. BRUNER AND CECILE C. GOODMAN

5. BROWN, W. Individual and sex differences insuggestibility. Univ. Calif. Publ. Psycho!.,1916, 2, 291-430.

6. BRUNFR, J. S., & POSTMAN, L. Perception andthe dynamics of behavior. (In preparation.)

7. BRUNSWIK, E. Wahrnehmung und Gegen-standswelt. Vienna, 1934.

8. CASON, H. Sensory conditioning. /. exp.Psychol., 1936, 19, 572-591-

9. COFFIN, T. E. Some conditions of suggestionand suggestibility: a study of some atti-tudinal and situational factors Influencingthe process of suggestion. Psychol.Monogr., 1941, 241.

10. DJANG, S. The role of past experience in thevisual apprehension of masked forms. /.exp. Psychol., 1937, so, 29-59.

11. ELLSON, D. G. Hallucinations produced bysensory conditioning. /. exp. Psychol.,1941, 28, 1-20.

12. ELLSON, D. G. Experimental extinction of anhallucination produced by sensory condi-tioning. /. exp. Psychol., 1941, 28, 350-361.

13. FAZIL, A. Miinzcnversuche iiber Anzahl-,Grossen-, und Wertwahrnehmung. Citedby Ansbacher (i).

14. FEHRER, E. V. An investigation of the learn-ing of visually perceived forms. Amer. /.Psychol., 1935, 47, 187-221.

15. v. FIEANDT, K. A new constancy phenome-non in color perception. Ann. Acad. Set.(Finland), 1938,41 (summary in English).See also, Dressurversuche an der Farben-wahrnehmung. Arch. ges. Psycho!., 1936,96, 467-495-

16. HAGGARD, E. R., & ROSE, G. J. Some effectsof mental set and active participation in theconditioning of the autokinetic phenome-non. /. exp. Psycho!., 1944, 34, 45-59-

17. HENLE, M. An experimental investigation ofpast experience as a determinant of visualform perception. /. exp. Psychol., 1942,30, I-2I.

18. HOLLINOWORTH, H. L. The inaccuracy ofmovement. Arch. Psychol., 1909, No. 13.

19. KRECHEVSKY, I. "Hypothesis" versus"chance" in the presolution period in sen-sory discrimination learning. Univ. Calif.Put!. Psychol., 1932, 6, 27-44.

20. LEEPER, R. A study of a neglected portionof the field of learning—the developmentof sensory organization. /. genet. Psychol.,1935. 46, 41-75- .

21. LEVINE, R., CHEIN, I., & MURPHY, G. Therelation of the intensity of a need to theamount of perceptual distortion: a pre-liminary report. /. Psychol., 1942, 13,283-293.

22. MEILI, R., & TOBLER, C. Les mouvementsstroboscopiques chez les enfants. Arch, dePsychol., 1931, 23, 131-156.

23. OESER, O. A. Personal communication, 1939.24. PERKY, C. W. An experimental study of

imagination. Amer. J. Psychol., 1910, 21,422-452.

25. PIACET, J. Language and thought in thechild. London and New York, 1932.

3.6. POSTMAN, L., & BRUNER, J. S. The relia-bility of constant errors in psychophysicalmeasurement. /. Psychol., 1946, 21, 293-299.

27. PROSHANSKY, H., & MURPHY, G. The effectsof reward and punishment on perception./. Psychol., 1942, 13, 295-305.

28. RIVERS, W. H. R. Report of the CambridgeAnthropological Expedition to the TorresStraits, Vol. II, 1901.

29. SANFORD, R. N. The effect of abstinencefrom food upon imaginal processes: a pre-liminary experiment. /. Psychol., 1936, 2,129-136.

30. SANFORD, R. N. The effect of abstinencefrom food upon imaginal processes: a fur-ther experiment. /. Psychol., 1937, 3,145-159.

31. SHAFER, R., & MURPHY, G. The role ofautism in a visual figure-ground relation-ship. /. exp. Psychol., 1943, 32, 335~343-

32. SHERIF, M. A study in some social factors inperception. Arch. Psychol., 1935, No. 187.

33. SNEDECOR, D. Statistical methods. Ames,Iowa, 1940.

34. THOULESS, R. H. Individual differences inphenomenal regression. Brit. /. Psychol.,1932, 22, 2I6-24I.

35. THURSTONE, L. L. A factorial study of per-ception. Chicago, 1944.

36. ZUK-KARDOS, I. Perzeptionale Zuganglichkeitvon Anzahl, Flache, und Wert unter ver-schiedenen Umstandskonstellationen. Citedby Ansbacher (i).