Social Perception and Phenomenal Causality

SOCIAL PERCEPTION AND PHENOMENAL CAUSALITY

BY FRITZ HEIDER

Smith College

I. INTRODUCTION

In recent years a great many studieshave been made of the processes of or-ganization in the perceptual field. It isthe thesis of this paper that the prin-ciples involved in these studies can beapplied profitably to the perception ofother persons and their behavior, andthat one of the features of the organiza-tion of the social field is the attributionof a change to a perceptual unit. Whenwe see a moving object A, we, can at-tribute the movement either to A itselfor to another object B. In the first casewe see the movement as a spontaneousact of A, in the second as passive move-ment induced by B. When we have adisagreeable experience, or a pleasantone, we may locate its origin in anotherperson, in ourselves, or in fate. An at-tempt to investigate the role of these at-tributions is described in a study of 'ap-parent movement' in which a motionpicture of geometrical figures was pre-sented to groups of subjects (16). Thisstudy showed how causal attributionsinfluence the perception of the behaviorof other persons. The relatively ab-stract nature of the material used in thisexperiment made it possible to studysome of the more elementary factors ofcausal attribution in social perception;in the present paper an attempt is madeto analyze these processes further andto show their influence in a variety ofmore concrete cases of social behavior.The discussion will not be restricted topurely perceptual processes but will in-clude processes like 'apperception' or'cognition.' Since the organization ofthe social environment seems to followsimilar laws whether or not the data aregiven perceptually it seems, in this case,

profitable to treat perception and apper-ception from a common point of view.

Making a first approach to this analy-sis we may say that the origin and thechange which is attributed to the originform a unit; that is to say, the change'belongs' to the origin. The connectionbetween origin and change is in manycases manifest and appears phenome-nally as causal dependence: "The origincauses the change." Therefore, originattribution is a case of 'phenomenalcausality' (11, pp. 78 ff.), or 'causalintegration' (46, pp. 679 ff.).

Ryan, in discussing the case in whichan object is perceived as the source ofa sound, says as follows:

Such an attribute of 'being a source ofsound' is certainly neither a pattern ofqualities (either visual or auditory) nor aspatial attribute. It is a property whichbelongs among what we shall call 'dynamic'characteristics of an object. Such proper-ties can be as directly perceived as colorand play a great part in integrating thetotal sensory mechanism (46, p. 679).

The difference between what Ryanhere probably means by the term 'pat-tern of qualities' and causal integrationcan be demonstrated in the field ofvision. When we see a color as be-longing to an object, the color is per-ceived as a property of the object, andthis object unit with its properties couldbe called a 'pattern of qualities.' Butwhen we see a color patch as a spot ofsunlight, the color patch is not a prop-erty of the sun. In both cases we findan integration, although this integrationis of different degree. The pattern ofqualities which corresponds to an objectunit is more closely knit than the unitproduced by causal integration in which

358

SOCIAL PERCEPTION AND PHENOMENAL CAUSALITY 359

origin and effect are relatively inde-pendent parts.

Such causal effects often play the roleof data and can be thought of as proxi-mal stimuli through which are mediatedto us properties of the origins which be-long to the stable relevant psychologicalenvironment. Through causal integra-tion changes or passing experiences arerelated to the framework of the invari-ant environment which gives the changeits meaning. Thus, origin attribution isa case of the organization of experi-ences and has its roots in the 'individ-ual's pursuit of meaning' which has beendiscussed as a problem of social cog-nition in relation to standards of judg-ments and social values (8, pp. 53 ff.).1

II. PERSONS AS ORIGINS

That animate beings, especially per-sons, are the prototype of origins hasbeen amply demonstrated by manywriters.

In earlier and simpler ages men knewlittle and frankly referred the explanationof what they could not account for on thebasis of concrete experience, to the worldof spirits which, as they, conceived of it,surrounded and permeated their existence.In so doing they regularly sought for anagent rather than a cause (54, p. 654).

In the development of the child'scausal thought "the original union ofdoer and deed forms the schema accord-ing to which causal thought can de-velop" (51, p. 409; this sentence hasbeen omitted from the English transla-tion). The child has the convictionthat "all changes that concern me (ac-tively or passively) are 'made to hap-pen '" (51, p. 410).

The tendency to attribute changes topersonal origins is, of course, related to

1 Some authors talk of a general tendencytowards causal explanation, a 'causal drive.'Oppenheimer (42, pp. 238 ff.) considers it asa third basic drive beside the drives for selfconservation and for the conservation of thespecies.

what Piaget calls animism, that is, thetendency to attribute life to inanimateobjects.

Animism, or personification, has beenexplained in different ways. The mostcommon explanation refers to an in-ference based on analogy and is similarto those theories which contend that weunderstand other persons' minds be-cause we conceive them as analogous toour own mind. Similarly, it is said thatin personification the child attributes to "things properties which it has experi-enced as belonging to his own person,or to persons of his environment.2

Werner rejects this theory. He be-lieves that "personification in a primi-tive interpretation of the world is de-rived from the physiognomic mode ofexpression and perception" (61, p. 75).By the term 'physiognomic perception'he understands a preference for inter-pretation in terms of dynamic ratherthan static properties.

It may be that the child apprehends per-sons physiognomically more readily thanother objects in his surrounding world.This fact might give rise to the erroneousimpression that the child first discoversphysiognomic characteristics in human indi-viduals and then transfers them to non-human objects. The more direct assump-tion, however, and one which is in greateraccordance with the facts, is that the child,grasping the world as he does through hismotor-affective activity, will understand theworld in terms of physiognomies beforepersonifying (61, pp. 75 f.).

Personification arises out of thisphysiognomic perception when "thechild's need to handle objects in hiseveryday life as if they were personscomes into being. For instance, hewants something from the inanimate ob~

2 Dennis (9) offers quotations from authorswho held this view. For arguments againstthis kind of theory in regard to the under-standing of other persons, cf. Kohler (28, pp.233 ff.) and Koffka (27, pp. 655 ff.).

360 FRITZ HEIDER

ject, and as a result of this desire it in-evitably comes to personal life. Thechild can become angry with it, orsympathize with it, or punish it. Per-sonification is engendered principally bya specific attitude which the child mustadopt in dealing with the objects ofhis world in a 'social' fashion" (61, pp.77 f.).

In this explanation it is a need, forinstance, a need for revenge, which re-quires an object and thus leads to per-sonification. That this is not alwaysthe case is shown by experiments on ap-parent behavior. Personification canbe compelling in the absence of suchneeds.

Another explanation besides those interms of experience or needs seems to bepossible. The prevalence of personifi-cation in an imperfectly structured en-vironment might be caused, at least inpart, by the simplicity of origin-or-ganization. The changes are attributedto a single concrete unit as source,which is certainly a simpler organiza-tion than that to which an analysis interms of a causal nexus with many con-ditions would lead to.8

Discussions of the concept of re-sponsibility would tend to support thispoint of view. We find in them re-peatedly stressed the difference be-tween a more differentiated causalanalysis and personal causation inwhich an effect is attributed in an ab-solute way to one origin, the causalchain terminating in it. Fauconnet, inpresenting the views of the classicalschool of indeterminism, says:

There exists a causality peculiar to man,different from the causality which connectsnatural phenomena. Man is, in a certainsense, a first cause, if not of the physicalmovements which constitute his acts, atleast of their moral quality. . . . From

8 Cf. J. M. Baldwin's description of the de-velopment of a child's idea of person (2, p.124).

this perfect causality originates his re-sponsibility. . . . Be it good or bad, theact is mine exclusively and absolutely (12,pp. 177 f.).

A true antagonism exists between thisidea [the idea of human causality] and thescientific notion of causality. First of all,science knows only of secondary causes:every cause, in itself, is also an effect, andcausal explanation is a regression whichhas no conceivable end; secondly, thecause is for science a phenomenon like theeffect, the laws of causality only describenecessary relations between phenomena.On the other hand, the person conceivedas cause is a first cause; the act, saidAristotle, has its beginning in itself;secondly, the two terms are here heteroge-neous, between them exists, not the rela-tion of phenomenon to phenomenon whicha law expresses, but the relation of pro-ducer to product, of workman to finishedwork (12, pp. 277 f.).4

If we conceive of the idea of 'person'in this way, light is also thrown on thetheory that personification originates inneeds. Interesting in this connection isFauconnet's description of the develop-ment of the concept of responsibility.He says, following to some degree Durk-heim whose pupil he is, that crime vio-lates society and menaces its life. Itmust be annihilated so that the moralorder can be rehabilitated. But thecrime is a fait accompli; it cannot beannulled. Therefore, a substitute hasto be found, a symbol of the crimewhose destruction replaces the destruc-tion of the crime, and the beings whichare taken as substitutes are consideredresponsible for the crime.

4 Cf. Stok (52, p. 357): "Things appear ashomogeneous entities. . . . The person, on theother hand, is a final, heterogeneous causalcenter (Wirkzentrum); everything that occursin its region is thought of as coming from it,and only from it. All lines which one followsbackwards intersect in one point, no linepoints beyond. The world of relations endswith the other person. The field of explana-tions is, in regard to personal events, con-vergent." Also Maclver (34, pp. 225 ff.).


Is a first and personal cause anythingelse but a cause conceived in such a waythat it can be held responsible, that it canfurnish something fixed and constant towhich sanction can be applied? (12, p.278).

These theories, whether or not en-tirely correct, seem to illuminate themeaning of the unit 'person' in our lifespace. Through these units the organ-ism is enabled to reinstate an equi-librium even when otherwise irreversiblechanges have disturbed it. Persons, asabsolute causal origins, transform ir-reversible changes into reversible ones.We understand now why the need forrevenge, a special case of a need for thereinstatement of an equilibrium, re-quires a person as object: because ofthe simplicity of organization, the per-son can represent the disturbing changein its entirety.

That this tendency to perceive per-sons as origins influences social percep-tion can be shown by many examples.It can lead, for instance, to an under-estimation of other factors responsiblefor the effect. Changes in the environ-ment are almost always caused by actsof persons in combination with otherfactors. The tendency exists to ascribethe changes entirely to persons. Thisis shown by the reaction to other per-sons' successes or failures. The inter-pretation will, of course, depend also onforces acting on the 'alter level,' that isto say, on the value level of the otherperson within our life space. If we areinclined to disparage him we shall at-tribute his failures to his own person,his successes to his good luck or un-fair practices. But disregarding theseforces, there probably exists a tendencyto be 'intropunitive' in regard to otherpersons, that is, to see the cause of theirsuccesses and failures in their personalcharacteristics and not in other condi-tions.

When Nietzsche says, "Success is thegreatest liar," he refers to this error inattribution. Klages, in his book onNietzsche, devotes a whole chapter tothe suasion of success, in which hetreats the fact that success convinces usof the worth of a person even if the suc-cess is largely due to chance circum-stances (26, pp. 85 ff.). Ichheisermakes the same point in several of bispapers (21, 22).

Stern's concept of 'personal causality'also refers to this exaggeration of the in-fluence of personal causal factors:

"By this term we understand [withStern] the causal connection betweenthe doing subject and his deed. WhenA performs an act by which the deathof B is caused, we say: 'A has killedB.' We say that, even if there are in-termediate links between the act of Aand the succeeding death of B . . .which are outside the sphere of A (en-vironmental factors). It still is the actof A which led to the death of B. Thisconception is certainly much more pri-mary than the analysis of an actioninto intention of the actor, environ-mental factors, and final effect" (56,p. 97).5 In the same way the laymanis apt to exaggerate the importance ofthe personal factor in the interpreta-tion of historical events though withsome historians we can also find one-sided attribution to impersonal forces(CJ. 20).

The same tendency can be seen atwork in our perception of the conditionswhich are responsible for an act. Oftenthe momentary situation which, at leastin part, determines the behavior of aperson is disregarded and the behavioris taken as a manifestation of personalcharacteristics. Argelander discussesthe different ways in which this fact canlead to wrong judgments of personality.

o Cf. also 63, Ch. IX: "The will as the sub-ject of moral judgment and the influence ofexternal events.'

362 FRITZ HEIDEE

The trait of character which will beascribed to a person depends on the situa-tion in which the person's behavior isbeing observed (1, p. 285).

In She Stoops To Conquer Goldsmithpresents a young man whose behaviorvaries extremely with the momentarysituation. The other characters at-tribute the temporary behavior to thepermanent personality and this errorproduces many of the comic situationsof this play.

III. T H E RELATIONS BETWEEN CAUSALUNITS AND THE PROPERTIES

OF THEIR PARTS

In the following sections we shall dis-cuss the role causal attribution plays inthe cognitive structure of the environ-ment. We have to consider the factorswhich affect attribution, and the influ-ence which causal integration exerts onthe appearance of the parts. Followinga proved procedure (Cf. Koffka, 27;Lewin, 33), we shall first consider theforces resulting from the structure ofthe cognitive field itself, and secondlythe forces which relate to motivationsand needs.

A. The relation between the parts as afactor in causal unit formation

Origin and effect, or a person and hisact, are the parts of a causal unit. Theproperties of the act may be similar orin contrast to those of the person. Thisis possible since we often judge bothpersons and acts in reference to thesame dimension. Many trait nameswhich we use in describing persons canalso be used in describing acts. Per-sons or acts may be altruistic, vicious,or brutal. This is especially true forvalue properties: a person or an actmay be good morally. Applied to aperson the adjective refers to a dis-positional concept, for instance, to apermanent faculty; applied to an act itrefers to a temporary event.

Therefore, the relation between anact and a person can be one of varyingdegrees of similarity. We know thatsimilarity is a factor in figural unit for-mation (62). The same holds true forcausal unit formation. In general onemay say that factors making for figuralunit formation also make for causal in-tegration. Duncker (11) and Huang(20) have pointed out the importanceof similarity and proximity in phe-nomenal causality. If two events aresimilar to each other, or near eachother, one is likely to be perceived asbeing the cause of the other. Theliterature on magic practices is alsorelevant here since magic actions arebased on primitive causal ideas aboutthe environment. Tylor (57) andFrazer (13) have shown what role simi-larity and proximity play in this re-spect (Cf. also Kohler, 29).

That similarity is a factor in socialcausal integration is demonstrated byZillig's experiments (64). She showedthat poor performances will be attrib-uted to unpopular children by theirclassmates. A bad act is easily con-nected with a bad person.

Fauconnet says that all situationswhich give rise to the idea of responsi-bility (the attribution of a crime to aperson) can be classed in two groups:responsibility by contiguity and re-sponsibility by similarity (12, p. 261).The latter includes several varieties.A crime can be blamed on a person be-cause of physical similarity "he looksas if he could have committed thiscrime." 6 Or he can be held responsible

«Cf. the Toscanini anecdote related byOscar Levant in Harper's, 1939, 179, p. S89:"One such unfortunate, a violinist, was the in-variable target of his criticism, because of hismottled complexion. During a visit by theorchestra to Hartford, someone in the violinsection made a false entrance, and Toscanini,in a rage, placed the blame squarely on hisbete noir—even though the musician had re-mained in New York, ill." (Cf. also 4, PP-165 f.)


for it because of 'spiritual' similarity,that is, a similarity between a 'crime asa moral event and the natural disposi-tion of the responsible.' "Can one dis-cover in the accused a basic immoralitycomparable to that of which the act isheld to be a symptom, does one find hiscrime in him, his character in his crime,in short, does he resemble his crime?His responsibility is not complete un-less the answer is affirmative" (12, p.264).

The responsibility can also be estab-lished by 'affective similarity.'

It is easy to assume that persons are re-sponsible who are not loved. In a generalway, 'antipathy' arouses suspicion; and aspecial 'antipathy' arouses a special sus-picion. Persons dreaded for their brutalityare the first ones to be suspected of a vio-lent crime; despised persons, of a meanact; and those who arouse disgust, of anunclean act. People with bad reputationsare accused and convicted on the basis ofevidence which one would consider insuffi-cient if an unfavorable prejudice did notrelate them to the crime in advance. Onthe contrary, if the accused has won ourfavor we demand irrefutable proof beforewe impute to him the crime (12, p. 266).

The property which makes a personin our eyes 'similar' to a crime or an actcan be created by past experience: weknow that he has committed such anact before. In causal attribution thefactor of past experience cannot beseparated from the factor of similaritybecause a person, once having been theorigin of an act, takes over its quality.The person is then connected with theact also by the relation of similarity.The difficult rehabilitation of peoplewho once were inmates of a prison or amental hospital, or the fable of the boywho cries 'Wolf are relevant examples.

That a crime can be judged as 'be-longing to' a person on the basis of ir-relevant circumstances is shown byincidences related by Gross:

So for example, an intelligent man toldme that he suspected another of a murderbecause the latter's mother died a violentdeath. The witness stuck to his state-ment: "the man who had once had some-thing to do with killing must have hadsomething to do with this killing." In asimilar manner, a whole village accused aman of arson because he was born on thenight on which a neighbouring villageburned down. Here, however, there wasno additional argument in the belief thathis mother had absorbed the influence ofthe fire inasmuch as the latter was toldthat there had been a fire only after thechild was born. "He once had somethingto do with fire," was the basis of the judg-ment, also in this case (15, p. 122).

In these cases the person takes over,by assimilation, a 'whole-quality' of acrime with which he was in some wayconnected. Then, the factor of simi-larity causes a new crime of the samesort to be attributed to him. Thus thefactor of proximity can make for anassimilation of a person to an event.It is not so easy to find a clear case inwhich this factor is directly the causeof an unjustified attribution of an actto a person. The cases which Faucon-net cites as examples of 'responsibilityby contiguity' cannot always be callederrors of attribution,—for instance, thefact that the presence at the place ofcrime speaks against the accused in theeyes of the jury; . . . nor are they al-ways clear cases of attribution,—for in-stance, the fact that a person is held tobe contaminated by a contact with thecriminal (pp. 261, 262).

B. The influence of causal integrationon the relations between parts

1. Assimilation and contrast.—It isknown, from the study of elementaryperception, that the parts of an inte-grated structure will tend either to lookas much alike as possible, or to look asmuch unlike each other as possible.Werner, who treats these phenomena inregard to intensity, says:

364 FRITZ HEIDEK

This organization occurs in the most im-portant cases either in such a way that apart or whole property of the structure in-fluences the intensity directly by assimila-tion, or so that it affects the intensity by arelation of opposition {beziehend.es Gegen-uberstellen). In the first case . . . wespeak of a dependence of mixture or as-similation, in the second of a dependenceof segregation or dissimilation (59, pp.47 f.).

Gestalt psychologists talk in this con-nection of the maximum-minimum prin-ciple, that is, the tendency towardsmaximum or minimum s impl ic i ty .". . . w e must remember that there ex-ist always two possibilities, correspond-ing to the minimum and the maximum;either as little or as much as possiblewill happen" (27, p. 109). "Roughlyspeaking, a minimum simplicity will bethe simplicity of uniformity, a maxi-mum simplicity that of perfect articula-tion" (27, p. 171). Werner (59) andKoffka (27, pp. 171 ff.) see in assimila-tion and contrast two different expres-sions of the tendency towards Pragnanz.Examples are given by Werner (60).T

2. The influence of the origin on theeffect.—In social perception, the act isin many cases assimilated to the origin.Acts or products are colored by thequalities of the person to whom theyare ascribed. A joke made by a personconsidered silly will usually seem silly,while the same joke made by a personwith the reputation of being witty willarouse laughter.

7 For an experimental demonstration of as-similation see also Fuchs (14) and Campbell(7). For conditions which determine whetherassimilation or contrast will be effective, cf.Koffka (27, p. 173, explanation in terms ofdisposable energy of organism), and Werner(59, p. 115; 60, p. 252; assimilation occurswith high degrees, contrast with lower degreesof integration). Usnadze's findings with ex-periments on sets corroborate Werner's opin-ion: assimilation appears when the differencebetween members is small, contrast when it isgreat (58, p. 332).

When the worship of Mammon is wide-spread the millionaire is a high authority,not only in the technique of money-making,but in everything else as well. . . . Hiscrassest stupidities have the indescribablecharm of coming from a 'solid' man (44,pp. 175 f.).

The phenomenon of 'prestige sugges-tion' is familiar to social psychologists:the acceptance of statements and theevaluation of artistic products dependson the prestige of the person to whomthey are attributed (47, 49).

Contrast or dissimilation occurs whenthe origin is taken as the standard forjudging the quality of the act. We canassume that in these cases the fusionbetween act and origin is less completeand the act is perceived, so to speak,with the origin as background. Our re-action to praise or any personal com-ment depends greatly upon whether itcomes from a reserved or a loquaciousperson. Shakespeare makes use of thiskind of contrast when he describesOthello as a person to whom jealousy isforeign. If he had introduced Othelloas a man inclined to be jealous, his actsof jealousy would have lost much oftheir dramatic force (Cf. 53). Sopho-cles' Philoktetes offers a similar ex-ample. Philoktetes is presented as pos-sessing a high degree of self-control.That such a man cries out loud in painis very impressive.

It has been remarked that visual con-trast phenomena are related to colortransformation (usually called, in amisleading way, 'color constancy').8

A proximal stimulus of medium in-tensity seen in dim illumination signi-fies a very white object. This relationis similar to that in the examples takenfrom social perception. A reaction ofmedium intensity coming from a person

8 Cf. Jaensch's thesis: Laws of contrast be-come laws of transformation if the term 'sur-rounding field' is replaced by the term 'illu-mined space' (24, p. 272). Also Werner (59,pp. 120 ff.).


whose usual reactions are known to befeeble, will signify a strong environ-mental 'press,' to use Murray's term.This relation is similar to that in theexamples just mentioned of Sophocles'portrayal of Philoktetes' pain.

3. Influences of effects on origins.—Changes will influence the appearanceof an origin if they are combined withit in a causal unit. Effects often playthe role of data through which we learnto know about the origins, and it isobvious that our idea of a person can beinfluenced by the attribution of acts.But here, again, we have to note a dif-ference between a causal unit and asimple object unit. We saw above thatin the latter a color is directly ascribedto the object as a property. In the lesshighly integrated causal unit the changedoes not appear as a property of theorigin, only the possibility or faculty ofproducing this change is ascribed to theorigin as a property.

At the same time, the origin is as-similated to the produced effect. Ef-fects can be the determining elementsin the whole-qualities of origins whichare permeated with the color of the ef-fects. As examples may be mentionedpoisons, 'hot' pepper, alcohol, or ex-plosives. In die same way a personwho is seen as the origin of a bad act isusually judged to be bad himself. Mac-Iver mentions as one stage in the im-putation of responsibility "the imputa-tion to the conscious agent of an atti-tude or disposition, perhaps we mightsay an animus, which makes not thedoing only but the doer susceptible toa value judgment" (34, p. 226)?

8 The influence of causal attribution on ourattitudes is also referred to in Spinoza's sys-tem: "If we remove a disturbance of the spirit,or emotion, from the thought of an externalcause, and unite it to the other thoughts, thenwill the love or hatred toward that externalcause, and also the vacillations of spirit whicharise from these emotions, be destroyed" (50,p. 2S4).

Contrast does not seem to occur asinfluence of act on person. The unit oforigin and effect is a hierarchical unit:the origin is a superordinate part, theeffect a subordinate one. Therefore, theeffect can be seen with the origin asbackground and the latter can providea standard according to which the mean-ing of the effect appears, but the re-ciprocal relation (origin with effect asbackground) is impossible.

However, the effect can be evaluatedin reference to an environmental 'press,'and thereby something like a contrastphenomenon can arise. If we perceivethat a person stays moderately calm un-der extremely exciting circumstances weshall judge him as very quiet. Just asour perceptual mechanism 'knows' thatthe light reflected from a surface is de-pendent on two conditions, and repre-sents it as a combined effect of illumina-tion and object color, so we 'know,'in many cases, that the behavior of aperson is dependent on situation andpersonal characteristics. Sometimes be-havior and personal characteristics aregiven and the press value of the situa-tion is inferred (on the perceptual, notrational level), as in the above men-tioned example of Philoktetes; some-times behavior and situation are givenand we infer personal characteristics.In these cases we find dissimilation be-tween personal origins and effects.

Often, the change that our idea of aperson undergoes when we attribute anact to him cannot be described so simplyin terms of assimilation or contrast.When we see a person lift a heavyweight we perceive at the same timethat he is strong. 'Lifting a heavyweight' and 'being strong' belong to-gether as manifestation and faculty butthey can hardly be called similar.These faculty concepts play an impor-tant role in social apperception. Forinstance, we often have a definite ideaof what a person—our own person or

366 FRITZ HEIDER

another person—is able to do, or not todo.10 This 'being able to do some-thing' is ascribed in part to the environ-ment; for instance, the possibility ofmoving around freely in a region isascribed to the fact that the region iswithout barriers. But it is also ascribedin part to the person, especially, whenachievement or failure to do somethingdeviates greatly from that of averagepeople.

Ichheiser says: "It may be said withassurance that the knowledge aboutkind and degree of the personal basisof our own capability usually consti-tutes itself by way of real acts. Onlyin acting and in attaining goals, achieve-ments, and successes do we learn toknow what we can do" (22, p. 372).

Thus the change or act, somethingtemporary, is correlated to somethingrelatively stable and invariant, a pos-sibility or capability. This is true inall cases of attribution to persons. Dy-namic characteristics, too, can be as-cribed in this way to persons as stableproperties. If an act is seen as di-rected against us, we shall ascribe to theperson whom we see as origin of thisact a hostile intention which may be ofa much longer duration than the tempo-rary act. On the other hand, if we areconvinced that the act does not origi-nate in constant characteristics of theperson we shall not hold him responsi-ble.11

Essentially, these cases are not sovery different from simple assimilation.The origin is changed by the effect in

10 Cf. Lewin's concept of 'space of freemovement' (31, pp. 42 ff.), or Ichheiser's'characters of attainability' (Errekkbarkeits-ckaraktcre, 22, p. 376).

1 1 The interesting question to what parts ofa person acts are attributed, and the errors ofattribution in this respect, will not be treatedhere. Cf. Ichheiser (23) who, for instance,discusses the fact that we tend to overrate theinfluence of the 'rational' parts of our ownperson.

such a way that it 'fits' the effect, or, ifone wants to use an analogy taken fromfigural units, in such a way that the ef-fect is a good continuation of the origin.We can express this dependence in moregeneral terms: If we are, for a certainreason X, convinced that A is responsi-ble for act a, then we see A in such away that A and a would be perceivedas belonging together even if reason Xwere absent.

IV. CAUSAL UNITS AND DYNAMIC

FACTORS

A. The influence of attribution on thedynamic state of the person

Emotions and tensions in the personare closely linked to facts of attribu-tion. As soon as one attempts to deter-mine the cognitive structure of the en-vironment which is coordinated to emo-tional reaction one faces the fact ofcausal attribution. Spinoza's treatise'On the origin and nature of emotions'in his Ethics is the most thoroughlyworked out system in this regard. Agreat number of correct observationsare treated in a theoretically unifiedway and attribution plays a prominentrole in this system.12

This is understandable since many ofthese dynamic states do not refer toproximal stimuli qua proximal stimuli,that is to say, not to the effects as theyreach our person but to effects referredto the framework of the permamentpsychological environment.18

12 Cf. also the classifications of emotions bypsychologists who were influenced by Spinoza(Merrier, 38; Hbffding, 17; Calkins, 5. Mer-der's classification can be found in Ruckmick,45).

18 "It is necessary to be remarked, respectingthe causes of our pleasurable and painful sen-sations, that they are apt to become greaterobjects of concern to us, to rank higher inimportance, than the sensations themselves. . . the mind is not much interested in at-tending to the sensation. . . . The mind, how-ever, is deeply interested in attending to the


In treating these phenomena we haveto consider first the influence of attribu-tion on the dynamic state of the person;in the next section we shall discuss theinfluence of the state of the person onattribution.

One can say that attribution influ-ences the needs or emotions because thelatter are correlated to the meanings ofeffects, and the meanings of effects aredetermined by attribution to an origin.

That the meaning of a datum is de-termined by the reference to an origincan be shown by many examples. Ourreaction to a disagreeable experience,for instance of pain, is greatly influ-enced by the attribution to a source,which we may see in another person, inthe workings of chance, or in ourselves.The same datum may mean aggression,misfortune, or a stupid mistake. Whenan injury is attributed to a personalagent, it is more likely to lead to an ag-gressive reaction. If we were convincedthat the pain we feel at the dentist'shad its ultimate source in the malevo-lence of a person instead of in the poorhealth of our teeth, we would react toit quite differently.

The same is true of the meaning ofan experience of frustration. Usually,frustration leads to aggression only (andnot always even then) when the originof the frustration is attributed not toone's own person, or to impersonalcauses, but to another person. Severalof the critics of the book Frustrationand aggression by the Yale group (10)point out the importance of attributionin this respect.

It is a question, however, as to whethernumerous situations in which the indi-vidual does not deal directly with frustra-

cause; that we may prevent, or remove it, ifthe sensation is painful; provide, or detain it,if the sensation is pleasurable. This createsa habit of passing rapidly from the sensation,to fix our attention upon its cause" (39, p.188).

tion readily translatable into terms of anaggressor, typically stimulate the aggres-sive rejoinder, for example, frustrationsarising out of one's own inability to solvea mathematical problem and the like (30,pp. 357 f.).

Maslow has something similar inmind when he makes the important dis-tinction between deprivation and threatto personality:

Thus a certain child deprived of an ice-cream cone which he wanted may have lostsimply an ice-cream cone. A second child,however, deprived of an ice-cream cone,may have lost not only a sensory gratifica-tion, but may also feel deprived of the loveof his mother because she refused to buyit for him. . . . It is only when a goal ob-ject represents love, prestige, respect, orachievement that being deprived of it willhave the bad effects ordinarily attributedto frustration in general (36, pp. 364 f.).

One could add that a frustration is ex-perienced as a threat to the personalityonly when it is interpreted in terms ofthe social environment, that is to say,when persons are perceived as theorigins of the frustration, and the tem-porary 'deprivation' is related to dur-able environmental relevancies.14

The influence of causal integrationon the experience of success or failureoffers another example of the deter-mination of the meaning of a datum byorigin attribution.

Not merely success or failure as such isdynamically of importance but above allthe experience of belongingness of theseeffects to the own person, the fact that the

14 That origin attribution has something todo with the reaction to frustration is con-ceded also by the Yale group: ". . . thestrongest instigation, aroused by a frustration,is to acts of aggression directed against theagent perceived to be the source of the frustra-tion . . ." (10, p. 39). Cf. also Koffka'sbrief description of experiments conducted byDembo (27, p. 673) in which she comparedanger directed against a personal cause and'impersonal' anger. The first one was muchstronger and its after-effects lasted longer.

368 FRITZ HEIDER

achievement is considered as an expres-sion of the value of the own person (19,p. 33).

In general it may be said that successor failure is experienced only when thesource of the achievement (or, in thecase of failure, the source of the in-ability to achieve a goal) is located inthe own person. One of the devicesused to lift morale is to restructure thefield in such a way that a defeat is notattributed to one's own inferiority. De-feat undermines the morale of a nationor a person only if the cause of the de-feat is attributed to its own weakness.When, on the other hand, the blame islaid on a 'stab in the back' or someother factor which is not connected withthe relationship between the own powerand that of the enemy, the self-evalua-tion which is basic for morale is not af-fected.

These examples show that causal at-tribution is of great importance in cog-nitive structures which give rise to ten-sions in the person since many vitalequilibria concern relationships of theown person to other persons; relation-ships of power, of value, of benevolenceor hostility. Most of the 'sensory' ex-periences of changes having a positiveor negative value for the person becomerelevant to these equilibria only whenthey are, by attribution, related to thestable social environment consisting ofother persons.

But this is not entirely true of allsensory experiences, just as it would notbe entirely true in the field of objectperception. There exists a 'stimulusdirected' perception and the phenomenalrepresentations of local proximal stimuli,in some cases, retain their identity inspite of their being organized into dif-ferent object units. This is also true inregard to the experience of ego-relevantfacts. Pain is disagreeable in itself and,though its coloration can be widelychanged by different attribution, its core

of negative value persists. The mean-ing of frustration is changed by causalintegration to a greater degree than thatof pain. However, even the effect offrustration is to some degree independ-ent of how it is embedded in the socialenvironment. Frustration caused by aperson typically leads to anger; butalso one not caused by a person canlead to anger (27, p. 673). Therefore,though not being universally correct, the'constancy hypothesis' which lies hid-den in the original theory of the Yalegroup has some limited justification.

B. The influence of attitudes onattribution

We have seen that the similarity be-tween origin and effect, which makesfor, or is brought about by, attribution,often refers to value properties whichin turn are closely linked with dynamicattitudes. There exist tendencies toperceive ourselves or other persons ina certain way, and these tendencies af-fect origin attribution.

The tendency to keep the ego levelhigh is an example. Hoppe showedhow this tendency can influence attribu-tion. His subjects frequently tried toclear themselves of the responsibilityfor failures. "Usually the blame is thenput on the material. . . . One evades,as far as possible, the effect of the fail-ure by looking for the cause of it not inthe own person but in the neutralsphere of objects" (19, p. 33).

It is obvious that this tendency tokeep the ego level high must play arole in attribution. Since origins areassimilated to acts attributed to them,an act of low value, when attributed tothe ego, will lower the ego level, and anact of high value will raise it. How-ever, this will happen -only when thestimulus conditions are so strong as toenforce the attribution, that is, if therecan be no doubt that the own person isthe source. Often, the possibility of


different organizations will exist. Thenthe tendency to raise the ego level willstructure the causal units in such away that only good acts and not badones are attributed to the own person.

Examples demonstrating these effectsare sometimes treated as cases of ra-tionalization,15 or projection. Claparedetalks of excuse (denying the attributionof a bad act to oneself) and imputation(ascribing bad acts to others).

When we have spilled ink on a lady'sdress, or when we have committed a moregrave sin, we consider calmly that it wasnot our fault, since the course of the worldis determined inevitably, and this accidentwas already virtually contained in the struc-ture of the primordial nebula.

Imputation appears frequently in mentalpathology in the form of imaginary ac-cusation, etc. These facts are well known,and psychoanalysis has interpreted them asa means to end an interior conflict painfulto the subject, by the projection on an-other person of sentiments condemned byconscience (6, pp. 281, 282).

Scapegoat behavior often is not simplyrelease of aggression but includes blam-ing others for changes which, if at-tributed to the person, would lower theself esteem. Cantril says that one ofthe functions of scapegoat behavior isthat "it exonerates the people themselvesfrom any blame for their conditions"(8, p. 245).

Maslow's concept of 'dominance feel-ing' is similar to that of ego levelthough it seems to refer to a morepermanent characteristic of personality(Cf. for instance, 35). Dominance feel-ing is an 'evaluation of the self and highdominance feeling involves 'good self-confidence, self-assurance, high evalua-tion of the self, feelings of general capa-bility or superiority,' etc. If a personhas high dominance feeling and his ownperson within his life space has theproperties of high value or strength, the

15 Cf. the examples given by Taylor (55).

perceptual data will be organized in sucha way that they conform to that picture.If environmental changes threaten themaintenance of this cognitive representa-tion of the own person, needs will ariseto make these changes harmless by ac-tion.

With some persons there exists atendency to keep the ego level low.They structure the attribution accord-ingly, as is shown by an example Maslowgives:

For instance low dominance women cannot 'take' compliments in spite of the factof their need and hunger for them. Thisis because they are apt at once to discountthe compliment as untrue and seek sus-piciously for other motives. Often, forexample, they may think the compliment ismaking fun of them, holding them up toridicule, or else trying to get something outof them (35, pp. 21 f.).

Rosenzweig maintains that there existcharacteristic reactions to frustrationwhich are distinguished by the kind ofattribution (43). He distinguishes be-tween three typical reactions: in the'extrapunitive' one the outer world isblamed for the frustration; in the *in-tropunitive' one the own person isblamed, and in the 'impunitive' one thefrustration is glossed over. Rosenzweigsays that the kind of reaction is typicalfor a personality.

He relates these reactions to frustra-tion to psychoanalytical concepts. Ex-trapunitive and intropunitive reactionsare manifestations of outwardly or in-wardly directed aggressive needs; eroticneeds lead to impunitive reactions.1*

Though these reactions are expressedin terms of needs, they can also be statedin terms of attitudes, and they are obvi-ously related to tendencies to see the re-lation between the own person to theenvironment in a certain way.

16 In later publications Rosenzweig relatesthe impunitive reaction to personality dis-sociation.

Documents

Social Perception and Phenomenal Causality