zuckerman (1957) a reappraisal of the roles of past ...wexler.free.fr/library/files/zuckerman (1957)...

VOL. 54, No. 4 JULY,1957

Psychological Bulletin

A REAPPRAISAL OF THE ROLES OF PAST EXPERIENCE ANDINNATE ORGANIZING PROCESSES IN VISUAL PERCEPTION1

CARL B. ZUCKERMANBrooklyn College

AND IRVIN ROCKGraduate Faculty of Political and Social Science, New School for Social Research

Twenty years ago, in his Principlesof Gestalt Psychology, Kurt Koffkaposed the problem of visual percep-tion in the succinct question, "Whydo things look as they do?" He tookissue with the usual answer thatthings look as they do because of ourpast experience with them, arguingthat an empiristic theory not onlyfailed to account for many of thefacts of visual perception, but, inaddition, entailed a number of logicaldifficulties.

Few of Koffka's arguments havebeen satisfactorily met; nevertheless,two decades later, a survey wouldshow that his analysis has had littleor no impact on the psychologicalliterature dealing with this problem.For example, a widely used textbookstates: "With the few possible excep-tions provided by primitive organiza-tions, all perceiving is dependentupon past experience—the so-calledhabit factor" (45, p. 410). More-over, the empiristic theory has re-emerged in the currently popularassumption that perceptions are gov-erned by motivational and affectiveforces; in this remodeling, however,

1 We take this opportunity to state our in-debtedness to Dr. Hans Wallach who has sogreatly influenced our approach to the prob-lems discussed in this paper. We also wish toexpress our gratitude and appreciation to Dr.Evelyn Raskin for her invaluable editorial as-sistance.

critical questions involved in an ex-periential approach to perceptioncontinue to be overlooked.

The present paper attempts to re-consider the logic of the central prob-lem and to examine the evidence bear-ing, in particular, on the question ofwhether form perception is learned.We shall restrict our discussion to thecontroversy between theories whichemphasize the role of learning (em-piristic theory) and the theory whichstresses the role of innate organizingprocesses (which we shall briefly referto as the organization theory). Theconcept of organization is, of course,basic to Gestalt psychology. Withrespect to its relevance for the fieldof perception, however, it can beevaluated on its own merits quiteapart from the validity of otheraspects of Gestalt theory, particu-larly the physiological theories ad-vanced by Gestalt psychologists.

ANALYSIS OF THE Two THEORETICALAPPROACHES

It is difficult to find a clear, unam-biguous statement of an empiristicposition; moreover, many writersassume the validity of empiristichypotheses but do not offer an an-alysis of basic questions. The formu-lation of the problem by Ames and hisco-workers (the transactional ap-proach) may serve, however, to il-

269

270 CARL B. ZUCKERMAN AND IRVIN ROCK

lustrate a modern empiristic theorywhich has focused on some of theessential issues in the problem of per-ception.

The transactionalists (28) arguethat the percept cannot be derivedfrom the retinal image alone, sincean infinity of external objects cangive rise to the same pattern of stim-ulation on the retina. For example,a small object nearby and a large ob-ject at a greater distance can result inthe same sized retinal image; simi-larly, a circular retinal image maybe produced by a circle in the frontalparallel plane or by an ellipse tiltedfrom this plane. Or, once again, aretinal image of a specific intensitymay be produced by either a blackobject in bright illumination or awhite object in dim illumination.How then, in view of this equivalenceof outer configurations in producingidentical retinal images, can the or-ganism "know" which object to see?2

The answer given to this question isthat the explanation is to be soughtin the realm of past events; the ret-inal stimulus pattern must be inter-preted in the light of knowledge fromthe past.

The question of what the organismsees originally—before it is able tointerpret the retinal pattern—is notraised. It seems clear, however, thatan empiristic position of this kind

2 It is true, of course, that a given retinalform may be produced by an "infinity" of ex-ternal configurations. This statement must,however, be qualified. The same ellipticalretinal image can result from an elliptical ob-ject in the frontal parallel plane or from avariety of circles at different tilts from thisplane, etc., and, in this sense, the retinal imageis ambiguous. But, under no circumstances,could a retinal ellipse be produced by a rec-tangular or triangular object. There is a limi-tation, then, to the ambiguity of the retinalimage and, accordingly, there is no need forinvoking assumptions to explain why we seea rounded object and not a triangle.

must hold that the initial perceptualexperience would be ambiguous—agiven image could result in the per-ception of a small or large object, ablack or white object, etc. By meansof "purposive action" with respectto the object, we build up "assump-tions" which then determine thenature of the present perceptual ex-perience.

Empiricists in the past formulatedthe problem in a similar way, but in-stead of speaking about the ambig-uity of the stimulus they emphasizedthe fact that the stimulus is fre-quently such that it should lead to apercept different from the one whichactually occurs. They pointed out,for example, that the shrinking image-size of an object as it moves awayfrom the eye should result in the per-ception of diminishing size. Size con-stancy could not, therefore, be ex-plained by the retinal image alone;the latter had to be supplemented ormodified by the contributions of pre-vious learning. Moreover, the senseof touch, rather than purposive ac-tion, was thought to provide the basisfor the learning needed to attain thecorrect percept (especially, in thecase of form perception).

Kohler (32) has pointed out thatunderlying the empiristic concept isthe implicit assumption of the ex-istence of a one-to-one correspond-ence between local retinal stimulationand the resulting sensory experience.Any change in the local stimulus,therefore, should result in a corres-ponding change in the percept. Thefact that such a change does not al-ways occur (e.g., perceptual con-stancies) had to be explained.

The organization theory differsfrom empiristic theory in its concep-tion of the physiological correlate ofthe percept. Empiristic theoristshave assumed that the percept should

INNATE ORGANIZING PROCESSES IN VISUAL PERCEPTION 271

be correlated with the process initi-ated by the local retinal stimulus.Organization theorists have relatedthe percept to a more comprehensiveset of central processes initiated bycertain relationships in the stimuluspattern. When the stimulus is de-fined in relational terms, it is nolonger always necessary to considerthe retinal image either as inadequateor as ambiguous for the determina-tion of the percept and to invoke pastexperience as a way out.

The difference between the twotheories in this respect can be mostclearly illustrated by reference to theproblem of achromatic color percep-tion. The empiricist would say thatsince different intensities of reflectedlight may give rise to the same per-cept (an object in different illumina-tions appears the same color—i.e.,brightness constancy) or since thesame intensity may give rise to dif-ferent percepts (a piece of coal inbright illumination and a white paperin shadow which reflect equalamounts of light to the eye), theproximal stimulus is consequentlyeither ambiguous or inadequate.

According to the organizationtheory, however, what is seen in aparticular region of the visual fielddepends not only on the properties ofthe retinal image corresponding tothis region (the local stimulus) butalso on stimulation from adjacent orsurrounding areas. Wallach (72)has clearly shown that the stimulusfor achromatic surface color is not theabsolute intensity of light from regionA alone but is the ratio of light in-tensities from regions A and B. With-

FIG. 1

out changing the intensity from A,the perceived color in A can be madeto vary from black to white bychanging the intensity from B. Thespecific neutral color seen will de-pend on the ratio of the two lightintensities.3

The assumption of one-to-one cor-respondence between the local stimu-lus and perception is therefore in-valid. When the stimulus is consid-ered as a relational pattern, it is notambiguous as the determinant of per-ceived neutral colors. The coal inbright illumination and the paper inshadow do not give rise to the samepattern of retinal excitation. Theratio between the intensity of theobject and that of its surround wouldbe different in each case and thereforethe perceived colors would differ.Conversely, we can take an object ofa particular albedo, place it on abackground of some given color andvary the illumination. In spite of thechanging amount of light reflectedfrom the object, it will be seen as thesame neutral color (brightness con-stancy) because the ratio of light in-tensities from the object and itsbackground remains the same. Thereis, then, no necessity for assumingthat the organism has to learn to seethe object as black in one case or aswhite in the other (or as the samecolor in constancy situations).4

8 Reflection from two surfaces representsthe simplest stimulus for neutral color; ineveryday life, of course, the stimulus condi-tions are more complex.

4 As a matter of fact, careful considerationof Wallach's findings indicates that a learningtheory for perceived achromatic color (or forbrightness constancy) is impossible. Forlearning to occur, the organism would have totake into account the illumination in which aparticular gray surface is given and to correctfor changing illumination. There is, however,no way in which illumination can be regis-tered independently from the surface color;both are given by the same stimulus variable


The same approach is applicable tothe problem of size perception. Aparticular retinal image may cor-respond either to a large object faraway or a smaller object nearby. Thestimulus situation is ambiguous onlywhen distance cues are eliminated orare inaccurate. It is, therefore, en-tirely possible that the underlyingcorrelate for perceived size is the in-teraction of the area excited in thevisual cortex corresponding to thesize of the retinal image and thephysiological correlate of phenomenaldistance (whether the distance cuesthemselves are learned or not) .5 Suchan interaction process may be an out-come of learning but until this canbe proven, the alternative of innateorganization cannot be ruled out.

In certain cases, however, evenwhen the percept is considered to bebased upon stimulus relationships,ambiguity as to what will be per-ceived still persists. The followingexample illustrates the point. Sup-pose we have in a darkroom situationa luminous point A, surrounded by aluminous rectangle B. Duncker (14),in his investigation of the stimulusconditions for phenomenal move-ment, found that, if the rectangle Bis slowly moved to the right, point Ais seen to move to the left, while therectangle is perceived as stationary.Unless the stimulus situation is de-fined relationally, one would have topredict that B would be seen to move

•—the amount of light reflected from the ob-ject.

6 Size constancy (defined functionally as aprocess of interaction of retinal size and per-ceived distance) should be distinguished fromthe problem of distance perception per se.Evidence that distance cues are entirely orpartly learned would not prove that this inter-action process is learned. Conversely, if dis-tance cues are innate, it does not follow thatsize constancy is innate.

B •A

FIG. 2

since only the image of B is displacedon the retina. But even if defined inthis way, the stimulus condition isstill ambiguous, since in physicalterms the situation can be correctlydescribed either as A being displacedwith reference to B or B in relation toA. Thus there are two possibilities;phenomenally, however, one is re-alized—only point A is seen to move.

How is this "preference" to be ex-plained? Early empiristic theorywould have maintained that only Bshould appear to move. Transac-tionalists might argue that initiallyeither or both possibilities could beexperienced and we learn to see onlythe point move, since in real life it isthe smaller, surrounded object thatusually moves. Carr has offered asimilar explanation (8).

Duncker believed that seeing pointA move is a consequence of the opera-tion of a selective principle which maybe denned as follows: When an objectA is surrounded by a second object B,then no matter which one is actuallymoving, only the surrounded onewill be seen to move, the outer objecttaking on the character of a frame ofreference which tends to be perceivedas stationary. So strong is this princi-ple that, even if the surrounded ob-ject is the observer himself, he willfeel himself to be in movement al-though objectively it is the surround-ing object or scene which is moving(induced motion of the self). Accord-ing to this viewpoint, the law of sur-roundedness represents an outcomeof innate organizing factors in the


brain and not a product of learning.It may be useful at this point to

summarize the principal features ofthe organization theory:

1. The percept is considered to bebased on stimulus relationships.Some examples of phenomena explic-able in terms of relational stimulusconditions in addition to achromaticsurface color (ratio of light intensi-ties) and movement (relative dis-placement) are: phenomenal velocity(rate of figural change [6, 73]); geo-metrical illusions; chromatic colorcontrast; and those listed below asillustrating the operation of selectiveprinciples.

2. In many cases, it is necessary toassume, in addition to the relationalproperties of the stimulus, the opera-tion of selective principles accordingto which sensory data are organized.Thus, one perceptual experience arisesrather than another, although, on thebasis of stimulus conditions, both areequally possible.6 Some exampleswhere such principles are assumed tobe operating are: laws of grouping(80); apparent movement; figure-ground organization (58); laws of sur-roundedness and separation of sys-tems in movement (14); sound local-ization by head movements (74);depth based on retinal disparity (82);

6 Some writers, while opposing the empir-istic view, do not see any need for a concept oforganization. Thus, Gibson (19) and alsoPratt (49) argue that it is sufficient to correlatethe stimulus conditions with the resultant per-cept in accordance with traditional psycho-physical method. Gibson has also pointed upthe necessity for correlating the percept withmore complex aspects of the stimulus. Hedoes not see the need, however, to correlatethe percept with the central processes initiatedby the stimulus relationships, as does the or-ganization theory. The above examples of se-lective principles show very clearly that theproximal stimulus itself, however denned,does not contain all that is needed for an ex-planation of the percept.

kinetic depth effect (77); phenomenalidentity (41, 69).7'8

3. The further assumption is madethat the percept is innately deter-mined by such stimulus relationshipsand selective principles of organiza-tion. Lashley has described this posi-tion in the following statement:

"The nervous system is not a neu-tral medium on which learning im-poses any form of organization what-ever. On the contrary, it has definitepredilections for certain forms of or-ganization and imposes these uponthe sensory impulses which reach it"(35, p. 35). _

It is possible for an empiricist toagree with organization theory thatthe stimulus is a relational affair; inaddition, he might even agree to theassumption of selective principles.He would then have to argue thatthese principles are based upon pastexperience. These views, however,would represent a radical changefrom traditional empiristic thinking;there are indications that empiristictheory may now be moving in thisdirection (48, 71).

The contrast between the two the-oretical approaches can be furtherillustrated by the problem of formperception.

We assume that it is now generallyagreed that the relative position ofpoints in the visual field does not haveto be learned but is given by the rela-tive position of corresponding pointsof excitation in area 17 (although at

7 It is still too early to tell whether phe-nomenal casuality as investigated by Michotte(43), should be included in the above list.

8 It does not seem to the authors that theconcept of Pragnanz is either clear or helpfulin dealing with perceptual phenomena; more-over, there is very little unambiguous evidenceto support it. On the other hand, selectiveprinciples as described by Wallach do seem toimply some tendency toward preserving con-stancy in perceptual experience.


one time this was a debated issue).The orderly projection of the retinalpoints to the visual cortex in such away as to preserve the same positionof points relative to one another (top-ologically) seems a sufficient condi-tion for the explanation of the percep-tion of visual direction. Walls (79)refers to these anatomical facts as anadditional argument against an eni-piristic theory of visual direction.Furthermore, there is evidence againsta learning theory for radial direction—i.e., the direction of a point fromthe observer (see p. 282). A precondi-tion for radial direction must surelyinvolve the perception of correct posi-tion of points relative to one another.Consequently, the major unresolvedissue in the area of form perception iswhether organization of the field is aresult of learning.

As experienced, the visual field isnot a patchwork of various colors andbrightnesses but consists of circum-scribed units, certain areas belongingtogether and forming shaped regionswhich are segregated from other re-gions. Wertheimer (80) emphasizedthat segregation in the visual fieldwas not a fact to be taken for grantedbut one which presented a crucialproblem in the investigation of per-ceptual processes. One explanationof this problem has been given interms of the retinal image. But theexplanation that one sees a book be-cause the image of a book stimulatesthe retina is insufficient. Sometimesone may not see a segregated unitwhen its image is objectively present(e.g., camouflage) and at other times,one may see a unit where objectivelythere is none on the retina (e.g., astar constellation). An even morefundamental objection is the factthat, although the retinal image mayaccurately represent the external sit-uation, in that a homogeneously col-

ored form would give rise to an imagehaving the correct shape and repre-senting the color appropriately,9 thereis no reason why the percept shouldbe correctly organized (i.e., in agree-ment with the shape and segregationof that form in the external world).The mosaic of stimuli on the retinacould be organized in various ways.It is logically possible, for example,to see part of the form together withpart of the surrounding area as oneunit; the shape of this unit would bedetermined by the parts which areunited.

The tendency to attribute certainaspects of perceptual experience tothe retinal stimuli (the view that or-ganized shape is given by the image)has been called by Kohler (32) the"experience error." Many empiristicwriters, including Ames, do not ex-plicitly deal with the problem of formperception, apparently not realizingthat it is a problem. Similarly, S-Rtheorists generally speak of a form asa stimulus which is given and simplyassume that no explanation is neces-sary. Since, however, the organizedpercept is not directly given by theretinal image, a theory is needed toexplain the perception of forms.

The "correct" organization can beexplained in two ways. Empiristictheories assume that the sensory dataare structured only as a result oflearning. According to this view, ayoung child or animal would initiallynot experience a visual field withsegregated objects. Instead, it wouldsee a mosaic of different brightnessesand colors (or perhaps an incorrectlysegregated field). Murphy (46), forexample, states that the infant has tolearn to sort out his impressions andto learn that certain stimuli go with

8 We are limiting our discussion to two-dimensional forms presented in the frontalparallel plane.


others. From the initial blur, theregradually emerges a segregated visualobject.

The alternative explanation arguesthat the sensory impulses are struc-tured according to selective principlesof organization which do not dependon learning. Such selective principleswere first described by Max Wert-heimer (80), who referred to them aslaws of grouping. Using line and dotpatterns, he demonstrated that group-ing was not a random arbitrary affairbut occurred according to definiteprinciples such as proximity, similar-ity of color and size, good continua-tion, etc. Wertheimer described andillustrated these principles with dotsand lines but he did not in any wayimply that grouping factors operateonly with such stimuli. These factorsare intended as an explanation for ob-ject and form perception in general.Perhaps not all of these factors arenecessary to explain organization.Some may, as a matter of fact, proveto be incorrect; nevertheless, organiz-ing factors of this kind seem indis-pensable for the explanation of formperception. The following illustra-tion demonstrates how these princi-ples would operate to account for theperception of a black circle on a whitebackground.

1. Within the circular contour ofthe retinal image of the black circle,all points are similar in color as arethe points outside the contour (group-ing by similarity of color).

2. Within the contour, all pointsare nearer to each other than topoints outside the figure with the ex-ception of points on the contour(grouping by proximity). The follow-ing diagram illustrates how this fac-tor would work:

In A one sees two constellations oflines each because of proximity. InB the two constellations are even

further unified and segregated fromeach other. In C the grouping is stillbetter and now we have two distinctforms as in everyday life (modifiedfrom Kohler [30]).

Thus, similarity and proximity ex-plain why the points within the con-tour are grouped together and sep-arately from the points outside thecontour. As already noted, it is oftennot realized that if the retinal imageis an unstructured mosaic of stimula-tion, there is no reason why pointswithin a contour should not be seenas belonging with points outside thecontour. That they are not is pre-cisely what calls for explanation.

3. The principles of proximity andsimilarity are insufficient to explainwhy the circle instead of the sur-rounding area appears as a shapedentity. To account for this fact, wemust invoke another principle of or-ganization first described by Rubin(58). Physically, a contour line servesas a boundary for two areas; it, there-fore, can be described as belonging toboth. Phenomenally, however, thecontour belongs to only one area, giv-ing shape to that area which therebybecomes the figure. The other arearemains shapeless and is seen as theground. This biased belonging ofcontour must be due to a selectiveprinciple—figure-ground organiza-tion.10

10 In those cases where conditions are am-biguous, the figure-ground organization islabile and easily reverses itself. The phenome-non of reversible figures in general (i.e., in-cluding other types such as the Necker cubeand the Schroeder staircase) again shows veryclearly that the retinal stimulus does not con-tain all that is needed for an explanation ofthe percept.


In the above example, the contourbelongs to the black area (becausethe surrounded region is favored infigure-ground organization), therebygiving rise to the percept of a blackcircle on a white ground. Most in-stances of two-dimensional form per-ception would seem to be accountedfor by these laws of grouping (prox-imity and similarity) and the biasedbelonging of the contour (figure-ground) .

The principles of organization canbe considered as purely descriptivegeneralizations. One can employthese principles without invokingany theory of brain function; norwould any particular type of brainmodel be demanded. More specifi-cally, the value of the grouping fac-tors as explanatory concepts for formperception does not depend on thephysiological theory developed byKohler in connection with figuralaftereffects (34). This theory is de-signed to deal with the question ofhow the cortical pattern of excitationis related to the phenomenal size andshape of the percept. By relating thepercept to functional distance in thecortex (i.e., degree of interaction ofcurrent fields) rather than to geo-metrical distance, and by treatingsatiation as a changed resistance ofthe medium to interaction, Kohlerwas able to account for the facts offigural aftereffects. But even if thistheory is correct, it does not elim-inate the need for grouping principles,about which it says nothing.11

11 Recently, however, Kohler has suggestedthat the flow of direct current which he hasfound to accompany perception may also ex-plain why the figure appears as segregatedand distinct from the ground. The currentwithin the cortical correlate of the figure isconsidered to be highly concentrated andsharply segregated from that of the ground(33). Whether or not this particular idea iscorrect, it must be admitted that some physi-ological theory could also adequately explain

In the present paper, we have lim-ited the discussion of form perceptionto the problem of organization. Indoing so we simply assume that therelative position of points to one an-other in the visual cortex correspondsvery closely with the relative positionof points in the perceived scene, thusavoiding for the present the questionof whether some theory of functionaldistance is also necessary. Moreover,no attempt is made here to explainwhy the distortions of the retinalpattern which occur in its corticalprojection do not lead to distortionsin perception.

An additional problem in form per-ception arises in connection withwhole (Gestalt) qualities and withthe fact of transposition (i.e., thephenomenal equivalence of form trans-posed in size, color, position, etc.).There must be some aspect of or-ganization which underlies the wholequality and which distinguishes oneform from another. Underlying theperception of a circle, for example,there must be some characteristicpattern of interaction (one mightspeculate that it would be symmetri-cal in some way). This pattern yieldsthe whole quality of circularity,which can also be produced by otherpatterns of the same form but of adifferent size or color, since the samecharacteristic interaction occurs ineach case. Although this problem hasnot been solved, the fact of wholequalities and their transposabilityrepresents one of the strongest argu-ments for the existence of spontane-ous processes of organization.

Hebb (23) has outlined an em-piristic theory of form perceptiondirected toward answering the major

grouping. The important consideration is thatsome unlearned law of grouping is necessary,whether stated in purely descriptive terms orin terms of brain events.


arguments of Gestalt theory, manyof which are recapitulated in thispaper. Space prevents a detailed an-alysis of Hebb's theory, althoughmuch of the evidence he cites is eval-uated below. Hebb is not primarilyconcerned with phenomenal facts inperception but rather with the prob-lem of explaining the response tostimulation. Consequently, he dis-cusses only some of the problemsconsidered here.

Hebb grants that a form has "prim-itive unity" (Hebb's term for figure-ground organization) prior to learn-ing but (if we understand him cor-rectly) this unity does not suffice forform perception. The authors findthis point difficult to grasp. Wouldnot the primitive unity of an ex-tended figure (e.g., a straight line) bephenomenally different from that ofa compact one (e.g., a solid-color cir-cle) ? If so, the admission of primitiveunity implies that form perception isnot learned. Figure-ground organiza-tion means that the contour belongsto the figure, thereby giving it shape.Moreover, Hebb does not make clearhow the emergence of "cell assem-blies" (integrated networks of excita-tion in the visual areas) changes thephenomenal experience of a form inany way beyond its primitive unity.

Hebb believes that many facts ofmemory suggest that the memorytrace must entail a structural (i.e.,physical) change in a specific locusin the brain (a conclusion which heerroneously believes is denied byGestalt theory) (23, p. 12 ff.). Hisconcept of cell assembly as a particu-lar kind of neural change in a specificlocus, characteristic of the stimulus,is intended to explain recognition,i.e., how the same response can bemade to a transposed form. Presum-ably a multiplicity of such assembliesfor a given stimulus-pattern are estab-lished in all possible positions and to

each the same response is associated.To us, however, transposition sug-gests that the essential correlate ofphenomenal shape is the process towhich the stimulus gives rise and notthe place in the cortex where it oc-curs. But this does not imply thatthe memory trace is unlocalized. Ac-tually, those sympathetic to the or-ganization theory have often specu-lated that the trace may be localizedand recently some evidence for tracelocalization has been reported (76).They merely stress the fact that, forrecognition to occur, a later percep-tual process need not occur in thesame place as the trace.

Confusion arises when "appear-ance" is made synonymous with "re-sponse."12 Eventually, of course, atheory is needed to explain howstimulation of different cortical cellscan result in the same motor response.The authors believe, however, thatpremature preoccupation with thisproblem has led behavioristic psy-chologists in the wrong direction.13

It is possible for two similar forms(projected to different loci in thevisual areas) to look alike because ofsimilar cortical processes prior to thedevelopment of motor responses.Later on, motor development makespossible the association of a specificresponse (on the human level, the ap-propriate word) to these similar per-cepts.

LOGICAL DIFFICULTIES INHERENT INTHE EMPIRISTIC VIEW OF

FORM PERCEPTIONThe statement that the organiza-

tion of the visual field into shaped re-gions is learned must mean that at an

11 The widespread use of the term "percep-tual response" is a clear illustration of thisidentification.

13 For a lucid discussion of the necessity todeal with phenomenal data in perception, seeAllport (2, Chap. 2).

278 CARL B. ZVCKERMAN AND IRVIN ROCK

early period in the life history of anorganism the visual world does notconsist of segregated and unified ob-jects, but appears instead as a mosaicof sense impressions. In some waylearning and experience must thentransform the sensory data intoshaped visual areas. One argumentin support of this position is that ourvisual field contains segregated formsbecause of previous experience withthose particular forms. This of courseimplies that memory traces of previ-ous percepts play a causal role, i.e.,they serve to bring about the emerg-ence of those forms when the samestimulus situation occurs later. Buthow can a memory trace left by anunorganized mass of sensory datacreate a shaped visual object in thepresent field? If the initial percep-tions consist of amorphous sensa-tions, then how can the memory ofsuch perceptions organize subsequentprocesses? Instead of trying to ex-plain how the shaped object arisesfor the first time out of the chaos ofsensation, it would seem much sim-pler to admit some degree of visualsegregation resulting from innateorganizing processes. One might thensay that the influence of past experi-ence must be secondary to spontane-ous organization. This logical diffi-culty inherent in the empiristic the-ory can be expressed by asking, "Howcan we learn to see, if we must see inorder to learn?"

Empiricists in the past maintainedthat organization does not first arisein vision but comes about throughthe sense of touch. By means of tac-tual exploration of the environment,the child presumably becomes awareof forms and in some way the tactualform causes segregation in the visualfield as well. As Kohler has pointedout (32), however, this argumentmerely transfers the problem of or-

ganization from the visual modalityto that of touch. It is still necessaryto explain how the discrete tactilesensations can yield an experience ofa single object. Moreover, it is diffi-cult to understand how the tactile ex-perience can be transformed into ashaped visual object and why thereshould be such excellent correspond-ence between the two. Nor is thereadequate evidence that touch canyield the precision which we have invisual form perception.

These criticisms also apply to theconcept of purposive action as acreative agent of the percept. Thetransactionalists have not explainedhow the results of action (whichmust themselves be perceived) deter-mine the nature of subsequent per-ception.

A more sophisticated argument forthe empiristic theory of form percep-tion might be made by assuming thatthe principles of grouping are learned(cf. 48, p. 215). This would allow forthe transfer of effects of experienceto the perception of novel forms; theearlier argument cited above wouldnot, and is, therefore, of limited value.For example, perhaps the child learnsthat adjacent and similar stimuluselements belong to one object. Evenif this is granted, it is still necessaryto account for the first emergence ofa visual unit. How does the childlearn that these stimuli belong to-gether? Does such learning occurbecause the child sees the object moveas a whole when it is manipulated?If so, then "moving together" (Wert-heimer's law of common fate) is im-plicitly accepted as an unlearned or-ganizing principle. At some point,the assumption of innate organizingprinciples must be made in order toexplain how learning itself is possible.

Another logical difficulty involvedin the effort to explain how specific


past experience modifies subsequentperception relates to the problem oftrace selection. Even if it were as-sumed that previous learning has re-sulted in an organized memory tracefor a particular form, this trace can-not exert an influence when the samestimulus is presented again unless thistrace and no other is aroused. Oneway in which the trace of a previouspercept could be aroused and thusinfluence the unorganized sensory im-pulses would be for the latter to travelto the locus in the nervous systemwhere the relevant trace is "stored."Contact in this way might occur ifthe successive images of a given ob-ject always occurred in the sameplace on the retina, but this is rarely,if ever, the case. Consequently,as Kohler has argued in elabora-tion of a point made by Hoffdingmany years ago, appropriate tracearousal must depend on some kindof similarity between the presentperceptual process and the traceleft by the previous process (31).This means that the present percep-tual process must be organized beforeit can communicate with the trace,because only an organized process(i.e., resulting in a definite shape inthe case of form perception) can besimilar to the trace representing thepreviously seen form. If the sensorystimuli are unorganized, it is difficultto understand how the proper tracecan be selected from the multitude oftraces existing in the nervous system.In general, then, past experience can-not exert any influence until the sen-sory processes themselves are or-ganized.14

14 The same argument arises in connectionwith experiments purporting to show an influ-ence of motivation on form perception. If amotive is to affect a percept, it would have todo so via memory traces of need-related ob-jects. In one experiment (39), for example,

THE DISTINCTION BETWEEN PER-CEPTION AND RELATED PSYCHO-

LOGICAL PROCESSESThe term perception has suffered

an extension of meaning so broad asto include almost every psychologicalprocess. For example, an experiment(7) which obviously deals with recall—the subjects having to reproducepreviously seen figures from memory—is widely cited as an experiment inperception. Certain distinctions mustbe made, not to serve a theoreticalbias, but in order to understand theparticular process and its relation toother psychological functions.

Perception should, first of all, bedifferentiated from recognition. Rec-ognition implies a feeling of familiar-ity—I experience the present objectas something I have seen before. Thefirst time the object was seen, how-ever, the perceptual experience oc-curred without the element of famil-iarity. In terms of underlying func-tions, recognition implies that thememory trace of the object is arousedby the present perceptual process;activation of the trace is the basisfor the experienced familiarity. Bydefinition, therefore, recognition isdependent upon past experience.

One implication of this distinctionis that even if the same form werepresented repeatedly, the same per-ceptual experience could conceivablyoccur each time without recognition.

two profiles (one of which has been rewardedin training sessions and the other punished)are presented together to form an ambiguousfigure-ground pattern. If the subject is to seethe rewarded rather than the punished pro-file, the memory trace of the former must bethe one which has the greater influence. Buthow can this trace be selected prior to the oc-currence of figure-ground organization whenpresumably no shape is as yet seen which issimilar to the rewarded or punished face?(For a further discussion of this problem, seereferences 57 and 75.)


This point may be clarified by refer-ence to an imaginary experiment: Sviews a figure and describes it. Letus assume that the memory trace forthis form is destroyed. Later the fig-ure is presented again and 5 is askedto describe it. He is likely to give thesame description as he did on thefirst presentation, even though he isnot aware of having seen the figurebefore. The experience is the samebecause the stimulus gives rise to thesame process in the brain. Recogni-tion represents an additional step—the arousal of the appropriate trace.16

As stated above, it seems necessaryto assume that trace contact andarousal are mediated by the similar-ity of the present perceptual processto the trace left by a previous visualexperience of the particular object(although there is no explanationavailable at the present time as tohow such a process of trace contactcan occur). Even if the present per-cept is changed or attenuated to someextent, trace contact can still occuras long as there is some formal orstructural similarity between the per-cept and trace. This means that rec-ognition can occur even when mater-ial is exposed under unfavorable per-ceptual conditions (e.g., tachisto-scopic presentation, peripheral vision,or dim illumination); moreover, it isreasonable to suppose that it will oc-cur more readily in the case of fre-quently experienced forms (cf. refer-ence no. 24 and the recent work onthe tachistoscopic recognition ofwords [67]). Recognition of the ma-terial does not mean, however, thatthe percept qua form is affected. For

16 The same point applies as well to a trans-posed form. Gestalt psychologists oftenstressed the recognizability of a transposedstructure, such as a melody. But even if notrecognized upon repeated hearings, the melodymay give rise to a similar experience each time.

example, a nearsighted person mayrecognize a friend from a distance butthe recognition does not make thepercept any clearer; his visual experi-ence is still fuzzy and blurred.10

Closely related to the quality offamiliarity is the distinctiveness or"identifiability" of certain formswhich comes about only with re-peated experience. Hebb points out,for example, that, at first, all chim-panzees look alike; with continuedobservation one begins to recognizeindividual animals (a similar factconcerning difficulty in distinguishingfaces of members of a different racialgroup from one's own has been men-tioned by social psychologists). Fre-quently, differences among similarobjects are not phenomenally regis-tered in initial perceptions; withgreater experience, these differencesbecome manifest. This seems to betrue, however, only for complexforms. There is no evidence that re-current observation is necessary inorder for a circle and a triangle, forexample, to appear as distinct forms.The problem of the discriminabilityof similar complex patterns requiresfurther investigation,17 but it shouldnot be confused with the question ofform perception per se. Past experi-ence is involved in the former case:

16 A recent experiment by Engel (IS) onbinocular rivalry between an upright and aninverted face may be another instance wherea recognition effect is considered to be a per-ceptual one. The subjects in this experimentare reported as having seen the upright facemore frequently. This result may mean thatout of the array of superimposed stimuli, theymore readily recognized an upright ratherthan an inverted face. In our opinion, there isas yet no conclusive evidence that the stimu-lus elements of the inverted face are sup-pressed.

17 Gibson and Gibson (20) have recentlyperformed an interesting experiment to ex-plore this process, which they call "perceptuallearning."


repeated perceptions of the form mayserve to strengthen memory traces ofthe details and of the relation of partsto the over-all pattern. These tracesprovide the basis for an increasedawareness both of the internal struc-ture of the form and its differencefrom similar patterns.

It is also essential to distinguishperception from interpretation. Agood deal of the evidence concerningthe effects of past experience or moti-vation on perception actually refersto the process of interpretation.Form perception has been defined asthe experience of a segregated objectof a certain shape in the visual field;interpretation, on the other hand,refers to the meaning which the visualform has for the subject. Unlikeform, meaning is not an outcome ofthe present stimulus pattern; mean-ing consists of those qualities andproperties acquired by an objectthrough association and learning. Onthe functional level, meaning derivesfrom the memory traces which areassociated with the trace of the visualform itself (e.g., a hammer has mean-ing because on previous occasions wehave seen this particular form usedin a certain way; this use is preservedin traces which are associated withthe trace of the form percept.)18 Thisdistinction is difficult to make clearbecause, phenomenally, we perceive

18 There has been some confusion concern-ing the treatment of meaning in Gestalt psy-chology. Apparently, some of the earlier writ-ings of Gestalt psychologists created the im-pression that meaning was thought to begiven directly in the present percept. (Thecontact between Gestalt psychology andphilosophical phenomenology may have con-tributed to this impression.) It seems to thepresent writers that meaning must be ex-plained in terms of associated traces or tracesystems, and is, therefore, derived from pastexperience. Kohler has stated this positionvery clearly (32, p. 138 ff.).

meaningful objects; usually, we donot first experience a pure form per-cept and then become aware of itsmeaning. On the level of experience,the meaning is given in the percept,but functionally, two processes mustbe distinguished.19

In some sense modalities, the dis-tinction we are making does fre-quently appear in experience; e.g., Ihear a sound and then try to identifyit—the cry of a baby or meow of acat. Even in vision the separation ofprocesses may be experienced. Forexample, a nonsense form is seen as asegregated unit of a certain shape;nevertheless, it may have little or nomeaning, and one may strive to in-terpret it. (Moreover, after gainingmeaning, the form itself does notchange in my experience. At first,J" was a meaningless shape. Al-though I now see it as an eighth-notethe visual form has not altered in anyway.) The separation of processesmay be more evident in the child'sexperience than in the adult's. Itseems probable that the child sees ob-jects before he has any concept oftheir meaning.

The separation of the perceptualfrom the interpretive process is notan arbitary matter of definition; onthe contrary, it is necessary to makethis distinction in order to account forthe nature of our experience. It is im-portant also to keep this distinctionin mind when evaluating experi-mental studies of perceptual prob-lems. For example, if we should wantto describe correctly the initial per-ceptions of congenitally blind sub-jects whose vision had been restored,

" The same is true about the distinction be-tween perception and recognition. Phenom-enally, familiarity is in the object; function-ally, one must assume that familiarity derivesfrom trace reference after the perceptual proc-ess occurs.


we should not confuse their failure toidentify objects with an inability toperceive objects as segregated units.Much of the data collected by vonSenden (61) is vitiated because theinvestigators did not clearly distin-guish the two functions. We may bedistorting the experience of a hungrysubject who describes an ambiguousshape as a steak if we conclude thatthe hunger drive has affected his per-ception (cf. 39). Possibly, he seesthe same form as does the nonhungrysubject, but interprets it differently.(If asked to copy the form, both sub-jects might make fairly identicaldrawings.)

The Rorschach test, insofar as it isconcerned with the ways in whichshapes are described (leaving asidethe color, shading, and other aspects),is primarily a test of interpretation.Many meanings can be ascribed tothe blot as a whole or to a particularpart.

EXPERIMENTAL EVIDENCE

The major portion of the followingsection will be devoted to a criticalanalysis of some representative studiesdealing with the question of whetherform perception is innately deter-mined. To begin with, some evidencerelating to the determinants of otherperceptual processes will be brieflycited but there is no intention of mak-ing a comprehensive coverage of theliterature bearing on this issue.Many studies are inconclusive be-cause no attempt was made to controlthe effects of previous experience.

Visual direction. Schlodtmann (60)showed that congenitally blind sub-jects localized the direction of pres-sure phosphenes in the same way asdo normally sighted subjects. Morerecently, Hess (26) has confirmedearlier findings (e.g., 4, 11) thatchicks peck in directions innately de-

termined by retinal locus. Sperry'sexperiments (68) provide further sup-port of the thesis that visual directionis unlearned.20

Visual constancies. The constan-cies—size, color, and brightness-have been shown to exist in variousanimal species (cf. reference no. 40for a summary of the literature).Size constancy, for example, has beendemonstrated in a three-month-oldchicken (22) and in eleven-month-oldinfants (17). Although these studiesare not crucial for the issue of innate-ness, they would appear to conflictwith naive empiristic views whichaccount for constancy on the basis ofknowledge or unconscious inference.

Depth perception. There seems tobe little unequivocal evidence relat-ing to the problem of distance ordepth perception. Lashley and Rus-sell (36) concluded that visual depthwas innately determined in rats, andHess succeeded in showing that chick-ens with no previous visual experi-ence (or with prior alternating mo-nocular vision) utilized binoculardepth cues (26).

Visual reflexes. Observations oninfants reveal that some visual-motorcoordinations, such as eyelid responsesto intense light, pursuit movements,and fixation are present at birth, orsoon after (12, 50). These data, how-ever, are not entirely relevant to thestudy of visual experience, since theymay simply represent reflex responsesto stimulation by light without beingaccompanied by the perception ofdirection, color, form, or depth.

Form. Two major experimental ap-proaches have been employed to de-termine the effects of past experienceon form perception. The first groupof studies we shall discuss attempts

80 Caution is necessary in generalizing theresults of animal experimentation, in percep-tion as well as pther areas, to the human level.


to show how experimentally createdfamiliarity with specific forms affectssubsequent perception; the secondgroup attacks the problem more di-rectly by studying the consequencesof the deprivation of normal visualstimulation soon after birth on laterperceptual development. Includedin the latter approach are the ob-servations on congenitally blind hu-mans who gained vision in later life.

The classic experiment in the firstgroup is the investigation by Gott-schaldt (21). Gottschaldt wanted toshow that a novel geometrical figurewill be seen in accordance with thelaws of grouping rather than past ex-perience. He reasoned that if formperception were determined exclu-sively by experiential factors, a com-plex figure b, containing a simple forma which has been seen very frequentlyin the past, should be perceived asthe familiar unit a plus other parts.Gottschaldt designed some simpleoutline figures which were presentedrepeatedly to subjects for memoriza-tion. Later, complex figures in whichthe a figures were embedded wereshown and the subj ects were instructedto describe them. It was found thatonly in a negligible number of caseswas b spontaneously described as thea figure and additional lines. Despiteits great familiarity at the time ofthe test, a was not seen.

Gottschaldt's experiment has beencriticized on the ground that it merelyshows that familiar units can be cam-ouflaged by embedding them in largercontexts. This criticism misses thepoint completely because it fails tosee the necessity for explaining whythe physically present figure is phe-nomenally absent. The camouflageis successful because of the victory ofgrouping factors over past experi-ence. Not just any additional lineswill successfully camouflage the a

figure but only those which, becauseof the laws of grouping, produce newand compelling organizations. A fewwell-placed lines will achieve this ef-fect, whereas a complex array of linesmay not succeed in camouflaging thea figure (cf. 32, p. 193 ff.). Good con-tinuation is probably the strongestfactor in Gottschaldt's figures. Cam-ouflage in nature, which of course in-volves additional factors (e.g., coun-tershading, similarity of color, etc.),also demonstrates that familiar ob-jects will not be readily perceivedwhen they are in certain environ-mental backgrounds (10, 42).

According to Hebb, Gottschaldt'sconclusion "is valid only if the totalfigure is an unanalyzable whole,which it surely is not" (23, p. 24).One b diagram, for example, con-tained two parallelograms and a setof lines forming a Z. Hebb impliesthat the presence of these familiarunits in the b figure explains why thea figure was not seen. It is possible,however, to embed the a figure in ab diagram which contains familiar

FIG. 4. A. ONE OF GOTTSCHALDT'S SIMPLEFIGURES; B. A MODIFIED VERSION OF GOTT-SCHALDT'S COMPLEX FIGURE CONTAINING

"A,"1 TO WHICH HEBB REFERS

parts and still the simple form willstand out. It is the structure of thetotal figure which is crucial and notthe familiarity of any of its parts.Moreover, even if the subject dis-tinguishes such familiar parts in thecomplex diagram, the question still


remains whether this is due to famili-arity or to structure.21

According to some critics (28),Gottschaldt's thesis cannot be ac-cepted because it has been refuted bylater investigators. The experimentby Djang (13) is often cited in thisconnection.

Djang's results show a strong effectof past experience. Simple figureswhich had previously been learnedwere found in complex forms twentytimes more frequently than those notseen before. (All figures were com-posed of dotted lines; the subjectshad to learn to draw the figure cor-rectly from immediate memory andto associate a nonsense name to eachfigure. The task was described as oneof learning and memory.)

Careful examination of the condi-tions of Djang's study makes it ap-parent that her results do not invali-date or even challenge Gottschaldt'sconclusions. Of special significanceare the following aspects of the ex-periment.

1. The instructions encouraged thesubject to break up the complex fig-ure into individual sections or units."Try at once to reproduce . . . whatyou have seen . . . . Indicate by theadditional use of the yellow pencilthe individual units or sections intowhich you split up the figure" (13,p. 34). Evidence for seeing the simplefigure in the complex one was basedon the units which the subjects en-circled. The complex figure containedmany subunits so that, in addition toseeing the figure as a whole, the sub-ject with this set might be expected tosee now one part and now another as

81 Hebb also points out that if one looks forthe simple form, one can find it. Here, ofcourse, he is referring to the problem of theinfluence of attention or set and we agree thatno psychological theory has as yet provided asatisfactory answer to this problem.

a relatively separate entity. Whenthe subject recognizes a unit as onehe has previously seen, he is likely toencircle it. This would facilitate thelearning of the complex figure, sincethe subunit represents a substantialportion of the total figure. That sucha set was important is shown by theauthor's remark that "success in find-ing the simple figure in the camou-flage seems to bear a relation to theamount of interest displayed" (p.47). Enthusiastic subjects were themost successful. Djang does showthat her data cannot be explainedmerely as a result of a set to look forfamiliar units. But a set to break upthe figures into parts is an importantcondition for the effect.

2. Unlike Gottschaldt's simple fig-ures which were absorbed into thelarger structure, many of Djang's areisolable subunits because their con-tours are not destroyed by good con-tinuation. Since the camouflagingeffect of Gottschaldt's figures is anessential feature of his design, onemay question the construction ofDjang's figures. Her results whichare based on the use of these figurescan, therefore, in no way affect thevalidity of Gottschaldt's conclusions.

3. Even without an analytical set,the subjects in this experiment mighttake note of a simple form in a com-plex one because they recognize it;this is true because of the point madein paragraph 2 above. In Gott-schaldt's experiment, the a figure wasnot recognized because it was notseen. In Djang's study, however, boththe simple unit which had not beenseen in prior exposures and the onewhich is recognized may have beenperceived (if only briefly) in the com-plex figure with equal frequency; butif the subjects had not seen the simpleunit before, there would be no specialreason to notice it. In other words,


Djang has not proved that there is adifference in frequency of perceptionof the simple form but only thatthere is a difference in the utilizationof this form based on recognition.

4. The fact that some masked fig-ures were more easily found thanothers cannot be explained by previ-ous experience (which was equal forall simple forms) but must be under-stood on the basis of factors of or-ganization. Those masked figureswhose contours do not continue intothe other lines of the complex formshould be readily seen; on the otherhand, the use of good continuationshould lead to fewer successes. Fig-ures LAJ, ZIP, and GIW have theleast number of successes—in thesefigures the contour of the simple formis to some extent continued into thelarger structure. The influence ofpast experience is greatest with fig-ures XEH, QOW, POQ, KOJ—andthese are the simple forms which areeasily segregated from the largerform. It is possible to take one ofDjang's figures and, by strengthen-ing organizational factors, make itdifficult for the familiar simple figureto emerge (see Fig. 5).

This shows conclusively that it isnot the use of dot figures which dis-tinguishes Djang's experiment fromthat of Gottschaldt. It is the con-struction of the dot figures, togetherwith her procedure, which made herresults possible. As a matter of fact,this experiment supports Gott-schaldt's contention that strongstructural factors overcome the ef-fects of familiarity.

Braly (5) attempted to show thatthe perception of polygonal, dot fig-ures is influenced by the kind of fig-ures shown earlier. The test slides,however, contain several of these dotfigures and it is impossible to see allof them clearly in the very brief ex-

A B C

FIG. 5. A. ONE OF DJANG'S SIMPLE FIGURES(QEW). B. DJANG'S COMPLEX FIGURE CON-TAINING QEW (KOJ). C. A MODIFIED VER-

SION OF KOJ CONTAINING QEW

posure time. The experiment dem-onstrates only that given a set to per-ceive a certain form and subsequentlygiven inadequate perceptual condi-tions, 5s will tend to guess in accord-ance with that set.

Henle (24) posed the questionwhether a familiar form would bemore readily perceived than an un-familiar one when structure is heldconstant. A series of letters and num-bers and their mirror reversals (to-gether with obverse and reverse non-sense forms) was exposed peripherallyor tachistoscopically. The results,based on the >Ss' reproduction of theforms, show that the obverse letterswere reproduced correctly more fre-quently than the reverse letters.Does the experiment demonstrate aninfluence of familiarity on perception?Perhaps the obverse letters are notmore clearly perceived than theirmirror reversals, but are more easilyrecognized under difficult perceptualconditions because of their familiar-ity. Once recognized, a familiar let-ter is easy to draw. The reverse letterwould probably be seen as a nonsensefigure, and, consequently, the sub-ject is faced with the added difficultyof remembering its inadequately per-ceived shape in order to draw it afew moments later. Following theanalysis given above, we would arguethat the presence of stronger tracesystems for obverse letters allows


recognition to occur more readily.In his investigation of figure-

ground organization, Rubin (58)found that Ss who were instructed tosee one part of an ambiguous form asfigure (the other part then appearingas ground), would subsequently tendto see the same part as figure. IfRubin's results were valid, theywould certainly provide evidence thatmemory traces can organize percep-tual processes. Recently, however, acareful repetition of Rubin's experi-ment by Rock and Kremen (57)failed to demonstrate this effect.

Leeper (38) found that subjectswould generally see Street figures asmeaningless collections of fragmentsupon first presentation. After a briefperiod of observation (sometimes ac-companied by verbal hints from theexperimenter), the figures were re-organized and perceived as meaning-ful objects. Several weeks later,when the same Street figures were ex-posed tachistoscopically, they wereimmediately recognized in theirmeaningful form. Lceper's experi-ment does show a past experience ef-fect, and thus seems to contradict thelogical argument that traces cannotinfluence perceptual processes untilthe latter are organized. This spe-cific problem will be discussed below.

We turn now to a consideration ofthe more direct kind of evidence. Itwould appear that a crucial test of theempiristic and organization theoriescould be provided by a "deprivation"experiment in which no opportunityto learn form perception through vis-ual experience is permitted during theorganism's early life.

On the human level, the data con-sist of observations made on cases ofcongenital blindness (due to catar-acts) to whom vision was restored inlater life by surgical operation. Theliterature on such cases has been an-

alyzed by von Senden (61) and hisstudy is often cited in support of theempiristic theory. According toHebb, for example, these patientscould not immediately distinguishforms after vision was gained; a long,gradual, learning process was neces-sary to enable the patients to per-ceive. There are, however, seriousdeficiencies in this evidence (cf.Michael Wertheimer [81], who de-scribes some of the flaws and, in addi-tion, observes that von Senden hasoften been cited erroneously). Theconditions and the exact time afteroperation of the observations werenot adequately described; the extentof vision present before operationvaried from case to case; some of thecases were young children whose re-ports are difficult to evaluate. More-over, the patients, after operation,were faced with a strange new worldand often the investigator (usuallythe surgeon) did not know whatquestions to ask, or what tests to per-form, in order to elicit the subject'sexperience. In one case, for example,the patient "had great difficulty indescribing her sensations in such away as to convey any clear concep-tion of them to another" (37, p. 148).Much of this evidence, therefore, isinconclusive.

With respect to form perception, itappears that no distinction was madein these studies between perceptualand interpretive processes. In theeighteenth and nineteenth centuries(when most of the cases studied byvon Senden occurred) the problemwas posed by investigators in the fol-lowing way: Would a blind person,who can distinguish a sphere from acube by means of touch, be able toidentify these forms visually whenseen for the first time? Observationsof these newly sighted patients seemedto show that they could not. There is,


however, no reason to expect such aresult. The patient might see thesphere and cube as different formsbut would not know their appropri-ate names until permitted the use oftouch. Moreover, even if told whichwas which, he would have to remem-ber this information, so that furtherlearning would be required for cor-rect identification, although not forperceptual discrimination.

It is clear from some of these casesthat the visual field of the patientwas not an undifferentiated blur butdid consist of forms and shapes whichcould be perceived but, of course, notnamed. Frequently, the case reportdescribes the patient looking at some-thing and asking "what is that?"One intelligent patient, as a matterof fact, was able to identify a ball asround and a toy brick as square uponfirst presentation (37). In a more re-cent case (16), the report also sug-gests that the patient could see ob-jects but was not able to identifythem. The observations on newlysighted patients, therefore, in no waylend support to an empiristic theoryof form perception.

More carefully controlled observa-tions are, of course, possible with ani-mals. In recent years, there havebeen a number of investigations of theeffects of early visual deprivationupon subsequent perceptual behav-ior.

Siegel (63), in a carefully designedexperiment, raised a group of ringdoves with plastic head covers whichpermitted light stimulation but nopattern vision. The hoods were putinto place soon after the birds werehatched and were worn for a periodof from eight to twelve weeks. Acontrol group of birds was raised in anormal visual environment. At theend of this pretraining period, win-dow openings were cut in the hoods

of the experimental doves. Bothgroups were now trained on a visualform discrimination—jumping to atriangle vs. a circle. Thirty trialswere given on each training day. Thecriterion for successful discrimina-tion was nine out of ten consecutivejumps to the positive stimulus on>any given day of the training series,

Siegel found that the hood-reared!birds required an average of 126.$trials to reach the criterion, while thecontrols required 77.7 trials; the dif-ference between the groups was sta-tistically significant. These results,according to Siegel, tend to verifytheories which stress the crucial roleof past experience in perception. Ac-tually, they may be interpreted asfurnishing cogent evidence for a non-learning position. If form perceptionmust be learned, it is very surprisingthat after eight to twelve weeks ofhomogeneous light stimulation theexperimental birds required only 49.1additional trials for correct perform-ance (one and two-thirds additionaltraining days). Moreover, Siegel'spublished report gives only groupdata; the individual perforrmmcerecords (65) show that one or twohood-reared birds were able to re-spond correctly very soon after un-hooding. For example, experimentalbird no. 13 required only 58 trialsto reach the criterion; this perform-ance is better than that of eleven outof twelve controls and practically ona par with the best of the controls(no. 26) who required 50 trials. Itmust also be pointed out that thegroup difference obtained by Siegelrefers to the arbitrary criterion forsuccess of nine out of ten correcttrials. But eight out of ten correctjumps (or even seven of ten) for twoor more blocks of ten in succession iscertainly above chance performance.We do not know whether a significant


difference would be obtained for thiscriterion.

It must be remembered that formdiscrimination is not a test of percep-tion alone; cognitive factors are alsoinvolved. It is possible that the ani-mal perceptually distinguishes thetriangle and circle but requires train-ing in order to learn that response toone stimulus is followed by reward.It might take a human subject twoor three trials before he realizes thathe must respond to a triangle and notto a circle. Would anyone argue fromthis fact that on these first few trialsthe subject did not see two differentforms? It is not surprising that ananimal deprived of visual form ex-perience for the first several monthsof life would show some retardationin solving a discrimination problem.(We are not referring to any diffi-culty in motor performance; no ob-jection to the experiment on suchgrounds seems justified since Siegeltook the precaution of forcing theanimals to jump from a platform atotal of four hundred times beforethe hoods were removed.)

An interesting experiment by Mil-ler (44) is relevant for the interpreta-tion of visual deprivation studiessuch as those of Siegel and Riesen.Miller's hypothesis was that the firstvisual experience of an animal raisedin darkness may create "a negativedisturbance effect which inhibits in-stantaneous utilization of the newcues even though perception may beimmediate and accurate" (44, p.224). He raised a group of rats in alight-proof cage and a control groupin a normal visual environment. Atsixty-five days of age both groupswere trained to run an obstacle coursein the dark. (After each trial the ex-perimental animals were returned totheir dark cages.) When both groupshad learned to perform rapidly, the

lights were turned on for the run.The controls did not seem to be af-fected by the light. The experi-mental rats, for whom this was thefirst visual experience, showed a sig-nificantly longer mean running time,and an increase of inter-rat varia-ability (individual running timesranged from six to twenty-nine sec-onds). The experiment shows thatperformance on a task already learnedon the basis of other sensory cuesmay be disturbed by the new visualexperience. Therefore it is probablethat such a disturbance woulti bepresent in the learning of new tasks,as in Siegel's experiment. In addi-tion, Miller's results point up the im-portance of taking individual differ-ences into account in studies of thiskind.

The earlier work of Riesen (51, 52),in which chimpanzees were raised fora long period of time in total dark-ness, requires only brief mention forthe purposes of this paper. The vis-ual defects shown by these animalsmay have been due to optic atrophyrather than to the lack of opportu-nity for learning (79). The more re-cent investigations (53), on the otherhand, are very important for theproblem of learning in perception.

In the revised procedure, chim-panzees were placed in a dark roomfive days after birth. For 90 minuteseach day, the animal's head was en-closed in a Plexiglas dome which per-mitted stimulation by diffused light.This procedure was continued untilthe animal was seven and one-halfmonths old when gradually, over aperiod of ten days, it was given moreand more light (increased illumina-tion of the room). In addition to ob-serving the animal's behavior in rela-tion to visual objects, the followingexperiment was performed. Train-ing of an avoidance response was be-


gun; twice a day, a shock plaque (adisk painted with vertical yellow andblack stripes) was held in front of theanimal and brought slowly towardhim until an electrode made contactwith his face and delivered a shock.When an avoidance response hadbeen established, discrimination train-ing was started. The shock plaquewas shown, followed by shock if theanimal did not make an avoidanceresponse. Four other plaques werealways followed by the food bottle.These "positive" disks differed fromthe negative stimulus in either oneof the following characteristics: size,color, shape, and direction of stripes.Complete data are reported for onlyone such animal, Chow, and for twoother animals—Faik, reared nor-mally, and Lad, reared like Chow ex-cept that he received 90 minutes aday of patterned light stimulation.

Riesen reports that Chow andKora (another chimpanzee reared inthe same way as Chow) evinced dif-ficulty in learning to recognize ob-jects such as the food bottle. In theexperiment, Chow showed delay inavoiding the shock plaque as com-pared to the normal control. Also hisperformance for the discriminationseries as a whole was inferior to thatof Faik or Lad. There are, however,discrepancies in the data which makeinterpretation difficult. For example,although Chow made many moreerrors than Faik before reaching thecriterion for the shape discrimination,he was superior in learning the dis-crimination between the horizontaland vertical stripes. Certainly dis-crimination of the direction of stripesshows some degree of form percep-tion. Chow had the greatest diffi-culty in discriminating size and shapeand little difficulty with color as wellas direction of stripes. The difficultymay be a cognitive one—i.e., per-

haps it was difficult for Chow to ab-stract the size and shape character-istics from the more striking surfacefeatures of the plaque. The fact thatChow made more errors than Faikon the discrimination of size sup-ports this interpretation; empiristictheory does not imply that the per-ception of size differences in objectsof the same shape presented at thesame distance must be learned. An-other finding which is hard to under-stand is the fact that Lad, who hadonly 90 minutes a day of patternvision for the first seven months oflife, made fewer errors to all positiveplaques than Faik, the normallyreared animal. Yet Lad had manymore failures than either Faik orChow in reaction to the shock plaque.One animal, Mita, reared like Lad,but restricted in a supine position ina holder, apparently also had diffi-culty in learning to discriminate thebottle from other objects. This factis not easy to explain.

It is also worth mentioning that fora long time after being placed in anormal environment, chimpanzeeswho had been reared without patternvision had difficulty with pursuit ofmoving objects and with binocularfixation; they also manifested spon-taneous nystagmus. Although Ries-en's results cannot be fully accountedfor by the presence of these impair-ments, it is plausible that such visualanomalies contributed to difficulty inclearly perceiving a unified and stableworld of objects.

We do not wish to minimize theimportance and interest of thesestudies. It is essential, however, torecognize the problems involved inthe interpretation of the data. It isclear that no definite conclusions canbe reached on the basis of studies em-ploying so few subjects; the discrep-ancies mentioned above may simply


represent individual differences. Nordo we know what are the effects of arestricted sensory environment onthe cognitive maturation of the ani-mal—there is evidence that animalsreared under such conditions mani-fest some impairment of intelligence(70). Furthermore, the results of ex-periments in which subjects areraised in an abnormal or restrictedenvironment lend themselves to twodifferent interpretations. If a particu-lar function or capacity does not ap-pear or is retarded, this may meaneither that learning is necessary orthat the experimental conditionshave disrupted the normal matura-tion of a function which may be in-nate. For example, Nissen, Chow,and Semmes (47) have shown, that achimpanzee who had been rearedwith little opportunity for tactual ex-perience (his arms and legs were en-cased in cardboard tubes) was unableto solve a problem requiring tactualdiscrimination of two widely sep-arated stimulus points. But, the"neonate chimpanzee responds dif-ferentially according to the locationon his body of a tactual stimulus.Usually the principal movement isnear the region of stimulation" (47,p. 494). Similarly, if a pain stimulusis applied to the cheek of a human in-fant, the infant's hand is brought tothe cheek near the point stimulated(62). These facts suggest the possi-bility that some degree of tactual dis-crimination is innate and that the re-stricted experimental conditions havedisturbed or prevented normal de-velopment so that the chimpanzee isunable to solve the discriminationproblem. In spite of these reserva-tions, however, we have no doubtthat this type of experiment hasbrought us closer to a crucial test ofthe two theories of form perceptiondiscussed in this paper.

The effects of the deprivation ofpattern vision on interocular transferhave been recently investigated inbirds (64), cats (54, 55), and chim-panzees (9). It has been found thatif, in rearing, both eyes have been ex-posed to patterned light (either si-multaneously or alternately), the ani-mal later trained monocularly on avisual discrimination problem trans-fers immediately to the untrainedeye. If the animal is reared in dark-ness and then given diffuse light toone eye and patterned light to theother (or reared with both eyes stimu-lated by diffuse light), there is noimmediate transfer of the discrimina-tion to the untrained eye, regardlessof which eye is used in the training;the discriminations are re-learned,however, with considerable savings.These results, while highly significantand quite surprising, are not relevantto the perception of form. There is noevidence that the animals could notsee forms when the eye which hadbeen given diffuse light was exposedfor the transfer tests.22 In fact, someof the data permit the inference thatthe lack of immediate transfer couldnot have been due to any difficultyin perceiving with this eye. First ofall, even when an animal was trainedwith the eye which previously hadbeen stimulated only by diffuse light,there was no transfer to the other eye,which had received patterned light(9). Secondly, an animal trained onthree problems in succession failed ineach case to transfer to the untrainedeye (9). This animal re-learned thefirst problem with the untrained (dif-

22 Riesen et al. (54) report that when thediffuse-light eye was first exposed, the catsbumped into objects and moved about quiteslowly. But similar behavior occurred whenthe previously trained eye was re-exposed.Nissen et al. (9) do not report the initial be-havior of the chimpanzees in their experiment.


fuse-light) eye so that perceptionmust have been adequate when thesecond discrimination was begun.Furthermore, it would be difficult toargue that color differences are notperceived with the diffuse-light eye,yet color and brightness discrimina-tion problems show the same effectsas discriminations of form (9, 54).These experiments, therefore, relateto the problem of recognition (i.e.,accessibility to the trace) and reveala limitation of the process of tracecontact by similarity.23 We wouldargue, applying the analysis givenearlier (p. 280), that the stimulusseen with the untrained eye has thesame appearance for the subject butit is not recognized as one which leadsto reward. Why trace contact fromone eye to the other does not occur soreadily only in the case where botheyes have not had patterned light is,of course, a puzzling problem.

WHAT PAST EXPERIENCE CON-TRIBUTES TO PERCEPTION

Every theory must grant that someaspects of perception are spontaneousreactions to the stimulus situation;no one, for example, has argued thatwhen the retina is stimulated bylight of wavelength 700 mu., learningis required before the color red can beseen. But both logical analysis andempirical evidence support the con-

23 The fact that some perceptual phenomenadiminish in strength or magnitude when trans-ferred to the previously unstimulated eye re-veals a similar limitation of interaction. If arotating spiral is viewed with one eye, thenegative aftereffect is greater with the sameeye than with the other. Similarly, Gibson(18) found that the aftereffect of the inspec-tion of a curved line is stronger with the eyeused during the inspection period than withthe eye which had not been stimulated. Theinfluence of past experience on the perceptionof a wire cube is greater with the eye previ-ously used (1).

elusion that much more than colorexperience is immediately "given" asa result of innate organizing factors.Specifically, the organization of thevisual field into shaped areas is notan outcome of learning—past experi-ence cannot carve visual form out ofinitially formless perception. Otherphenomena in perception consideredto be innately determined were re-ferred to earlier (p. 273).

But this does not imply that per-ception is not affected by past experi-ence. On the contrary, it is only whensome degree of innate organization isgranted that the effects of learningcan be more clearly understood.

1. The role of past experience inlending familiarity, ease of recogni-tion, and discriminability, as well asmeaning, has already been discussed.Perceptual experience is greatly en-riched by the addition of these as-pects; in the light of the distinctionsmade above, however, form percep-tion as such is not affected.

2. In some cases, a memory tracecan reorganize or modify a percept.An experiment by Wallach, O'Con-nell, & Neisser (78) demonstratesthat a memory trace can impartthree-dimensionality to a figure whichat first was seen as two-dimensional.Wallach presented to a control groupthe shadow of a wire figure whichwas described by the subjects as flat.In the presentation to the experi-mental group, the wire figure wasrotated, giving rise to a constantlydistorting pattern on the shadowscreen. The figure was now seen asthree-dimensional (kinetic depth ef-fect). Sometime later, when the samefigure was presented in a stationaryposition (where it had been seen astwo-dimensional by the controlgroup), it was described by the ex-perimental subjects as three-dimen-sional. Certain controls ensured that


the effect was a perceptual and not acognitive one.24

To explain these results in terms ofunderlying functions we must assumethe following: As a result of the mov-ing presentation, a memory trace isleft of a three-dimensional form. Inthe later stationary presentation, con-tact with this "three-dimensional"trace occurs on the basis of similarityof form and three-dimensionality isthereby imparted to the perceptualexperience.

The experiment suggests the possi-bility that many of the purely figuralcues to three-dimensionality (per-spective, overlay, specific patternssuch as a trapezoid giving rise to thepercept of a rectangle at a slant, etc.)may be learned. Although empiricistshave assumed such cues to be learned,they have never offered a plausibleexplanation as to how the learningtakes place. In terms of the abovehypothesis, the following explanationbecomes possible.

Unlearned depth perception occurson the basis of certain cues such asretinal disparity or the kinetic deptheffect, and leaves visual traces. Thesevisual traces can impart three-dimen-sionality to new figures, which other-wise would be perceived as two-di-mensional (e.g., the Necker cube,perspective drawings, etc.). This as-sumption obviates reference to touchor purposive behavior as the sourceof the learned depth experience. Theevocation of past experience effectsoccurs only when relevant traces areselected by the presence of a stimuluswith some similarity to the previous

24 Informal repetition of this experiment atthe laboratory of the New School for SocialResearch has failed to confirm that the effectis as easily obtained as the original report sug-gests. But even if such a memory effect occursonly occasionally it remains of great im-portance. In the present discussion, the ex-periment is cited primarily to illustrate howpast experience might modify organization.

three-dimensional percepts. By ac-counting for the initial depth percep-tions and for the arousal of the ap-propriate traces, this hypothesis over-comes the logical difficulties in em-piristic theories.

One might explain Leeper's experi-ment (38) in a similar way. On ini-tial presentation the Street figuremay be experienced as a jumble offragments. After a period of inspec-tion, the figure may suddenly lookdifferent—it is now recognized as ameaningful object. How this initialrecognition occurs is not clear but infunctional terms we may assume thatthe memory trace of the meaningfulobject is aroused in some way by theStreet figure, and that this tracechanges the phenomenal appearanceof the figure. It will be recalled thatwhen the Street figures were re-ex-posed several weeks later, they wereinstantly seen in their meaningfulform. How is this effect to be ex-plained? Following Wallach's rea-soning, we may assume that the firstpresentation leaves behind twotraces—one corresponding to the per-ception of the figure as an unorgan-ized collection of fragments, and as-sociated with it a trace of the figurein its meaningful form. Meaningfulre-recognition of the figure meansthat the second trace is aroused. Inaccordance with the logical argumentstated earlier, arousal of the firsttrace must occur and only then canthe associated trace be activated inorder to restructure the percept.25

3. Past experience within the ex-perimental situation or experimental

26 Perhaps a similar process (i.e., the modi-fication of a percept by a trace which isaroused by some kind of partial similaritywith the still incompletely organized stimulus)could explain the selective influence of previ-ous experience in ambiguous situations. Thefactual basis for this type of effect, however,is still unclear (cf. 27, 56, 57, 58, 59, 66).


instructions may produce a set whichin turn influences the perceptual out-come. As noted above, there is asyet no explanation for the action of aset or attitude in modifying percep-tion.

4. Prior experience may changethe neural medium so that subse-quent percepts are modified. This isnot an effect of past experience in theusual sense because it is not specificto the contents of subsequent percep-tions; it affects more or less indis-criminately stimuli which impinge ata later time in a specific region. Ex-amples of this category might in-clude: adaptations of various kinds,figural aftereffects and the negativeaftereffect of movement. Evidenceconcerning the effects of long rangeadaptation to unusual stimulus con-ditions has recently appeared (25).

5. The studies of Ivo Kohler (29),however, suggest that adjustment toprism-produced distortions or to chro-matic lenses cannot be understoodmerely in terms of local adaptationssince the effects are dependent uponeye position. Thus, for example, 5swearing glasses, each lens of whichconsisted of a blue left half and yel-low right half, in time adapted sothat when the eyes were turned tothe left the scene appeared less blue,and when turned to the right, less yel-low, than at first. Furthermore, whenthe glasses were removed, 5s reportedaftereffects which are also dependenton eye position. With eyes to theleft, the scene looked yellowish; witheyes to the right, it looked bluish.Similar adaptations and "situationalaftereffects" occur with respect todistortions caused by the wearing ofprisms. It is difficult to assess the fullsignificance of these recently pub-lished findings, but clearly an im-portant effect of past experience onperception seems to have been dem-onstrated.

6. Past experience may have anindirect effect by determining condi-tions which make other processespossible, although these processesthemselves are not the results of ex-periential factors. The apparentoscillation of an objectively rotatingtrapezoidal window (3) may be un-derstood in this way. If the windowis seen as rectangular, other percep-tual effects must follow. The per-ceived rectangularity itself may bedue to previous experience26 (an as-sumption which could be challengedby those who accept the principle ofPragnanz). Seen as a rectangle, thewindow cannot come into the frontal-parallel plane and must, therefore,be perceived to oscillate through anangle of less than 180°.

Another possible example of an in-direct effect is furnished by the situa-tion where those cues to distancewhich may be learned give rise to sizeconstancy, which may be innately de-termined,

CONCLUSIONOne can hardly take a dogmatic

position in an area where, as yet,there exists so little decisive experi-mentation. Nevertheless, it is im-portant to determine the status of ascientific theory in relation to presentknowledge. On the basis of logicalanalysis and an examination of rele-vant evidence, we have argued forthe thesis that various aspects of thephenomenal world and, in particular,the segregation and shape of visualforms are given by innate organizingprocesses. Percepts may be modifiedand enriched by experiential factorsbut the effects of such factors presup-

M Explanation of the apparent rectangu-larity in terms of visual traces makes moreconcrete the hypothesis which the transac-tionalists imply by such terms as "assump-tions," "prognostic directives for future ac-tion," etc.


pose the prior existence of visualforms.

If the thesis defended in this paperis correct, perceptual organizationmust occur before experience (or per-sonality factors which depend on ex-perience, such as need, purpose, andvalue) can exert any influence. Ac-cording to holistic concepts, cur-rently so popular, psychological func-tions cannot be separated. But it isthe relative independence of the per-ceptual organizing processes which

makes possible an adequate phenom-enal representation of the externalworld. Despite changing motives andemotions, phenomenal color, form,and space remain remarkably stableand generally correspond to the ob-jective situation. Such correspond-ence is, of course, necessary for suc-cessful adaptation to the environ-ment and the innate neural processeswhich yield this correspondence mustthemselves represent the product ofadaptive evolution.

REFERENCES

1. ADAMS, PAULINE A. The effect of past ex-perience on the perspective reversal ofa tridimensional figure. Amer. J.Psychol., 1954, 67, 708-710.

2. ALLPORT, F. H. Theories of perception andthe concept of structure. New York:\Viley, 1955.

3. AMES, A. Visual perception and the ro-tating trapezoidal window. Psychol.Monogr., 1951, 65, No. 7 (Whole No.324).

4. BIRD, C. The effect of maturation uponthe pecking instinct of chicks. Fed.Sent., 1926, 33, 212-234.

5. BRALY, K. W. The influence of past ex-perience in visual perception. J. exp.Psychol., 1933, 16, 613-643.

6. BROWN, J. F. The visual perception ofvelocity. Psychol. Forsch., 1931, 14,199-232.

7. CARMICHAEL, L., HOGAN, H. P., &WALTER, A. A. An experimental studyof the effect of language on the repro-duction of visually perceived form. J.exp. Psychol., 1932, IS, 73-86.

8. CARR, H. A. Introduction to space percep-tion. New York: Longmans, Green,1935.

9. CHOW, K. L., &NISSKN, H. W. Interocu-lar transfer of learning in visually naiveand experienced infant chimpanzees. J.comp. physiol. Psychol., 1955, 48, 229-237.

10. COTT, H. B. Adaptive coloration in ani-mals. London: Methuen and Co., 1940.

11. CRUZE, W. W. Maturation and learningin chicks. /. comp. Psychol., 1935, 19,371-409.

12. DENNIS, W. Is the newborn infant's rep-ertoire learned or instinctive? Psychol.Rev., 1943, SO, 330-337.

13. DJANG, S. The role of past experience inthe visual apprehension of maskedforms. J. exp. Psychol., 1937, 20, 29-59.

14. DUNCKER, K. tjber induzierte Bewcgung.Psychol. Forsch., 1929, 12, 180-259.

15. ENGEL, E. The role of content in binocu-lar resolution. Amer. J. Psychol., 1956,69,87-91.

16. FISHER, J. H. Vision learning after suc-cessful operation at age 6. Ophthal.Rev., 1914, 33, 161-165.

17. FRANK, H. Untersuchungen tlber Seh-grossenkonstanz bei Kindern. Psychol.Forsch., 1926, 7, 137-145.

18. GIBSON, J.J. Adaptation, after-effect andcontrast in the perception of curvedlines. J. exp. Psychol., 1933, 16, 1-31.

19. GIBSON, J. J. The perception of the visualworld. Boston: Houghton Mifflin, 1950.

20. GIBSON, J. J., & GIBSON, ELEANOR J.Perceptual learning: differentiation orenrichment? Psychol. Rev., 1955, 62,32-41.

21. GOTTSCIIALDT, K. t)ber den Einfluss derErfahrung auf die Wahrnehmung vonFiguren, I. Psychol. Forsch., 1926, 8,261-317.

22. GOTZ, W. Experimented Untersuchungenzum Problem der Sehgrossenkonstanzbeim Haushuhn. Z. Psychol., 1926, 99,247-260.

23. HEBB, D. O. The organization of behavior.New York: Wiley, 1949.

24. HENLE, M. An experimental investiga-tion of past experience as a determinantof visual form perception. J. exp.Psychol., 1942, 30, 1-22.

25. HERON, W., & DOANE, B. K., & SCOTT,T. H. Visual disturbances after pro-longed perceptual isolation. Canad, J,Psychol., 1956, 10, 13-18,


26. HESS, E. H. Space perception in thechick. Sci. Amer., 1956, 195, 71-80.

27. JACKSON, D. N. A further examination ofthe role of autism in a visual figure-ground relationship. J. Psychol., 19S4,38, 339-357.

28. KILPATRICK, F. P. (Ed.) Human behaviorfrom the transactional point of view. Han-over, N. H.: Inst. for Associated Re-search, 1952.

29. KOHLEE, I. fiber Aufbau und Wandlungender Wahrnehmungsve.lt. Vienna: Rohrer,1951.

30. KOHLER, W. An aspect of gestalt psychol-ogy. In Mtirchison, C. (Ed.), Psychol-ogies of 1925. Worcester: Clark Univer.Press, 1926.

31. KOHLER, W. Dynamics in psychology.New York: Liveright, 1940.

32. KOHLER, W. Gestalt psychology. NewYork: Liveright, 1947.

33. KOHLER, W., HELD, R., & O'CONNELL,D. N. An investigation of cortical cur-rents. Proc. Amer. phil. Soc., 1952, 96,291-330.

34. KOHLER, W., & WALLACE, H. Figuralafter-effects. Proc. Amer. phil. Soc.,1944, 88, 269-357.

35. LASHLEY, K.S. Persistent problems in theevolution of mind. Quart. Rev. Biol.,1949, 24, 28-42.

36. LASHLEY, K. S., & RUSSELL, J. T. Themechanism of vision: XI. A prelim-inary test of innate organization. /.genet. Psychol., 1934, 45, 136-144.

37. LATTA, R. Notes on a case of successfuloperation for congenital cataract in anadult. Brit. J. Psychol., 1904, 1, 135-150.

38. LEEPER, R. A study of a neglected por-tion of the field of learning—the de-velopment of sensory organization. J.genet. Psychol., 1935, 46, 41-75.

39. LEVINE, R., CHEIN, I., & MURPHY, G.The relation of the intensity of a needto the amount of perceptual distortion:a preliminary report. .7. Psychol., 1942,13, 283-293.

40. LOCKE, N. M. Perception and intelli-gence: their phylogenetic relation.Psychol. Rev., 1938, 45, 335-345.

41. METZGER, W. Beobachtungen iiberphanomenale Identitat. Psychol.Forsch., 1934, 19, 1-60.

42. METZGER, W. Gesetze des Sehens. (2nded.) Frankfurt am Main: WaldemarKramer, 1953.

43. MICHOTTE, A. La perception de la causa-lite. (2nd ed.) Publications Universi-taires de Louvain, 1954.

44. MILLER, M. Observation of initial visual

experience in rats. /. Psychol., 1948,26, 223-228.

45. MUNN, N. L. Psychology. (2nd ed.)Boston: Houghton Mifflin, 1951.

46. MURPHY, G. Introduction to psychology.New York: Harper, 1951.

47. NISSEN, H. W., CHOW, K. L., & SEMMES,J. Effects of restricted opportunity fortactual, kinesthetic, and manipulativeexperience on the behavior of a chim-panzee. Amer. J. Psychol., 1951, 64,485-507.

48. OSGOOD, C. E. Method and theory in ex-perimental psychology. New York: Ox-ford, 1953.

49. PRATT, C. C. The role of past experiencein perception. /. Psychol., 1950, 30,85-107.

50. PRATT, K. C. The neonatc. In Car-michael, L. (Ed.), Manual of child psy-chology. (2nd ed.) New York: Wiley,1954.

51. RIESEN, A. H. Arrested vision. Sci.Amer., 1950, 183, 16-19.

52. RIESEN, A. H. The development of visualperception in man and chimpanzee.Science, 1947, 106, 107.

53. RIESEN, A. H. Plasticity of behavior:Psychological aspects. In Harlow,H. F., & Woolsey, C. N., Symposium onbiological and biochemical bases of be-havior. Madison, Wisconsin: Univer. ofWisconsin Press, in press.

54. RIESEN, A. H., KURKE, M. I., & MEL-LINGER, JEANNE C. Interocular trans-fer of habits learned monocularly invisually naive and visually experiencedcats. J. camp, physiol. Psychol., 1953,46, 166-172.

55. RIESEN, A. H., & MELLINGER, JEANNE C.Interocular transfer of habits in catsafter alternating monocular visual ex-perience. /. comp. physiol. Psychol.,1956, 49, 516-520.

56. ROCK, I., & FLECK, F. S. A re-examina-tion of the effect of monetary rewardand punishment in figure-ground per-ception. J. exp. Psychol., 1950, 40,766-776.

57. ROCK, I., & KREMEN, I. A re-examina-tion of Rubin's figural after-effect./. exp. Psychol, 1957, 53, 23-30.

58. RUBIN, E. Visuell viahrgenomene Figuren.Copenhagen: 1921.

59. SCHAFER, R., & MURPHY, G. The role ofautism in a figure-ground relationship./. exp. Psychol, 1943, 335-343.

60. SCHLODTMANN, W. Ein Beitrag zurLehre von der optischen Localisationbei Blindgeborenen. Arch. Opthal,1902, 54, 256-267.


61. VON SENDEN, M. Raum-und Gestaltauf-fassung bei operierten Blindgeborenenvor nnd nach der Operation. Leipzig:Barth, 1932.

62. SHERMAN, M. C., SHERMAN, I. C., &FLORY, C. D. Infant behavior. Comp.Psychol. Monogr., 1936, 12, No. 4.

63. SIEGEL, A. I. Deprivation of visual formdefinition in the ring dove: I. Discrim-inatory learning. J. comp. physiol.Psychol., 19S3, 46, 115-119.

64. SIEGEL, A. I. Deprivation of visual formdefinition in the ring dove: II. Percep-tual-motor transfer. /. comp. physiol.Psychol., 1953, 46, 249-252.

65. SIEGEL, A. I. The effects of visual formdefinition upon the development andtransfer of simple form discriminationin the ring dove. Ph.D. thesis, NewYork Univer., 1952.

66. SMITH, D. E., & HOCHBERG, J. E. Theeffect of "punishment" (electric shock)on figure-ground perception. /. Psy-chol., 1954, 38, 83-87.

67. SOLOMON, R. L., & HOWES, D. H. Wordfrequency, personal values, and visualduration thresholds. Psychol. Rev.,1951, 58, 256-270.

68. SPERRY, R. W. The eye and the brain.Sc-i. Amer., 1956, 194, 48-52.

69. TERNUS, J. ExperimentelleUntersuchungtiber phanomenale Identitat. Psychol.Forsch., 1926, 7, 81-136.

70. THOMPSON, W. R., & MELZACK, R. Earlyenvironment. Sci. Amer., 1956, 194,38-42.

71. TOCH, H. H., & ITTLESON, W. II. Therole of past experience in apparentmovement: a revaluation. Brit. J.Psychol., 1956, 47, 195-207.

72. WALLACE, H. Brightness constancy andthe nature of achromatic colors. J. exp.Psychol., 1948, 38, 310-324.

73. WALLACH, H. On constancy of visualspeed. Psychol. Rev., 1939, 46, 541-552.

74. WALLACH, H. The role of head move-ments and vestibular and visual cues insound localization. J. exp. Psychol.,1940, 27, 339-368.

75. WALLACE, H. Some considerations con-cerning the relation between perceptionand cognition. /. Pers., 1949, 18, 6-13.

76. WALLACE, H., & ADAMS, PAULINE. Rec-ognition and the localization of visualtraces. Amer. J. Psychol., 1954, 67,338-340.

77. WALLACE, H., & O'CONNELL, D. N. Thekinetic depth effect. J. exp. Psychol,,1953, 45, 205-217.

78. WALLACE, H., O'CONNELL, D. N., &NEISSER, U. The memory effect ofvisual perception of three-dimensionalform. /. exp. Psychol., 1953, 45, 360-368.

79. WALLS, G. The problem of visual direc-tion. Amer. J. Optom. and Arch. Amer.Acad. Optom., 1951, Monogr. 117.

80. WERTHEIMER, MAX. Untersuchungen zurLehre von der Gestalt, II. Psychol.Forsch., 1923, 4, 301-350.

81. WERTHEIMER, MICHAEL. Hebb andSenden on the role of learning in percep-tion. Amer. J. Psychol., 1951, 64, 133-137.

82. WBEATSTONE, C. Contributions to thephysiology of vision, I. Phil. Trans.,Ft. I, 1838, 371-394.

Received December 3, 1956.

zuckerman (1957) a reappraisal of the roles of past ...wexler.free.fr/library/files/zuckerman (1957)...

Documents