forming impressions of personality - … impressions of personality * by s. e. asch ... jects in the...

FORMING IMPRESSIONS OF PERSONALITY *

BY S. E. ASCH

Graduate Faculty of Political and Social ScienceNew School for Social Research

WE look at a person and imme-diately a certain impression ofhis character forms itself in us.

A glance, a few spoken words are suf-ficient to tell us a story about a highlycomplex matter. We know that suchimpressions form with remarkablerapidity and with great ease. Subse-quent observation may enrich or upsetour first view, but we can no moreprevent its rapid growth than we canavoid perceiving a given visual objector hearing a melody. We also knowthat this process, though often imper-fect, is also at times extraordinarilysensitive.

This remarkable capacity we possessto understand something of the char-acter of another person, to form a con-ception of him as a human being, asa center of life and striving, with par-ticular characteristics forming a distinctindividuality, is a precondition of sociallife. In what manner are these impres-sions established? Are there lawfulprinciples regulating their formation?

One particular problem commandsour attention. Each person confrontsus with a large number of diversecharacteristics. This man is courageous,intelligent, with a ready sense of humor,quick in his movements, but he is alsoserious, energetic, patient under stress,not to mention his politeness and punc-tuality. These characteristics and many

* The present investigation was begun in 1943when the writer was a Fellow o£ the John SimonGuggenheim Memorial Foundation.

others enter into the formation of ourview. Yet our impression is from thestart unified; it is the impression of oneperson. We ask: How do the severalcharacteristics function together to pro-duce an impression of one person?What principles regulate this process?

We have mentioned earlier that theimpression of a person grows quicklyand easily. Yet our minds falter whenwe face the far simpler task of master-ing a series of disconnected numbersor words. We have apparently no needto commit to memory by repeated drillthe various characteristics we observein a person, nor do some of his traitsexert an observable retroactive inhibi-tion upon our grasp of the others.Indeed, they seem to support eachother. And it is quite hard to forgetour view of a person once it has formed.Similarly, we do not easily confuse thehalf of one person with the half ofanother. It should be of interest to thepsychologist that the far more complextask of grasping the nature of a personis so much less difficult.

There are a number ,of theoreticalpossibilities for describing the processof forming an impression, of which themajor ones are the following:

i. A trait is realized in its particularquality. The next trait is similarlyrealized, etc. Each trait produces itsparticular impression. The total im-pression of the person is the sum ofthe several independent impressions. Ifa person possesses traits a, b, c, d, e,

258

FORMING IMPRESSIONS OF PERSONALITY 259

then the impression of him may beexpressed as:

I. Impression = a + b +

II. Impression =

Few if any psychologists would at thepresent time apply this formulationstrictly. It would, however, be an errorto deny its importance for the presentproblem. That it controls in consider-able degree many of the procedures forarriving at a scientific, objective view ofa person (e.g., by means of question-naires, rating scales) is evident. Butmore pertinent to our present discussionis the modified form in which Proposi-tion I is applied to the actual formingof an impression. Some psychologistsassume, in addition to the factors ofProposition I, the operation of a "gen-eral impression." The latter is con-ceived as an affective force possessinga plus or minus direction which shiftsthe evaluation of the several traits inits direction. We may represent thisprocess as follows:

la. Impression =

To the sum of the traits there isnow added another factor, the generalimpression.

2. The second view asserts that weform an impression of the entire per-son. We see a person as consisting notof these and those independent traits(or of the sum of mutually modifiedtraits), but we try to get at the rootof the personality. This would involvethat the traits are perceived in relationto each other, in their proper placewithin the given personality. We mayexpress the final impression as

It may appear that psychologists gen-erally hold to some form of the latterformulation. The frequent reference tothe unity of the person, or to his"integration," implying that these quali-ties are also present in the impression,point in this direction. The generalityof these expressions is, however, notsuitable to exact treatment. Terms suchas unity of the person, while pointingto a problem, do not solve it. If wewish to become clear about the unity inpersons, or in the impression of persons,we must ask in what sense there is suchunity, and in what manner we come toobserve it. Secondly, these terms areoften applied interchangeably to Propo-sitions II and la. It is therefore impor-tant to state at this point a distinctionbetween them.

For Proposition II, the general im-pression is not a factor added to theparticular traits, but rather the percep-tion of a particular form of relationbetween the traits, a conception whichis wholly missing in la. Further, Propo-sition la conceives the process in termsof an imposed affective shift in theevaluation of separate traits, whereasProposition II deals in the first instancewith processes between the traits eachof which has a cognitive content.

Perhaps the central difference betweenthe two propositions becomes clearest

260 S. E. ASCH

when the accuracy of the impressionbecomes an issue. It is implicit inProposition II that the process it de-scribes is for the subject a necessary oneif he is to focus on a person with maxi-mum clarity. On the other hand,Proposition la permits a radically dif-ferent interpretation. It has beenasserted that the general impression"colors" the particular characteristics,the effect being to blur the clarity withwhich the latter are perceived. In con-sequence the conclusion is drawn thatthe general impression is a source oferror which should be supplanted bythe attitude of judging each trait inisolation, as described in Proposition I.This is the doctrine of the "haloeffect" (9).

With the latter remarks, which weintroduced only for purposes of illus-tration, we have passed beyond thescope of the present report. It mustbe made clear that we shall here dealwith certain processes involved in theforming of an impression, a problemlogically distinct from the actual rela-tion of traits' within a person. To besure, the manner in which an impres-sion is formed contains, as we shall see,definite assumptions concerning thestructure of personal traits. The validityof such assumptions must, however, beestablished in independent investigation.

The issues we shall consider havebeen largely neglected in investigation.Perhaps the main reason has been aone-sided stress on the subjectivity ofpersonal judgments. The preoccupa-tion with emotional factors and distor-tions of judgment has had two mainconsequences for the course investiga-tion has taken. First, it has induceda certain lack of perspective which hasdiverted interest from the study ofthose processes which do not involvesubjective distortions as the most deci-sive factor. Secondly, there has been

a tendency to neglect the fact that emo-tions too have a cognitive side, thatsomething must be perceived and dis-criminated in order that it may beloved or hated. On the other hand, theapproach of the more careful studiesin this region has centered mainly onquestions of validity in the final productof judgment. Neither of the mainapproaches has dealt explicitly with theprocess of forming an impression. Yetno argument should be needed to sup-port the statement that our view of aperson necessarily involves a certainorientation to, and ordering of, objec-tively given, observable characteristics.It is this aspect of the problem that wepropose to study.

Forming a Unified Impression:Procedure

The plan followed in the experimentsto be reported was to read to the sub-ject a number of discrete characteristics,said to belong to a person, with theinstruction to describe the impressionhe formed. The subjects were all col-lege students, mbst of whom werewomen.1 They were mostly beginnersin psychology. Though they expressedgenuine interest in the tasks, the sub-jects were not aware of the nature ofthe problem until is was explainedto them. We illustrate our procedurewith one concrete instance. The follow-ing list of terms was read: energetic— assured — talkative — cold — ironical— inquisitive — persuasive. The readingof the list was preceded by the follow-ing instructions:

I shall read to you a number of character-istics that belong to a particular person.

1 The writer wishes to express his gratitude tothe following colleagues for their help in the per-formance of these experiments in their classes:Drs. B. F. Riess, L. Welch, V. J. McGill, andA. Goldenson of Hunter College; Drs. M. Blumand A. Mintz of the College of the City of NewYork; Dr. Lois Adams, Mr. Michael Newman, andMr. Herbert Newman of Brooklyn College.


Please listen to them carefully and try toform an impression of the kind of persondescribed. You will later be asked to give abrief characterization of the person in just a ,few sentences. I will read the list slowly andwill repeat it once.

The list was read with an intervalof approximately five seconds betweenthe terms. When the first reading wascompleted, the experimenter said, "Iwill now read the list again," and pro-ceeded to do so. We reproduce belowa few typical sketches written by sub-jects after they heard read the list ofterms:

He seems to be the kind of person whowould make a great impression upon othersat a first meeting. However as time wentby, his acquaintances would easily come tosee through the mask. Underneath would berevealed his arrogance and selfishness.

He is the type of person you meet all toooften: sure of himself, talks too much, alwaystrying to bring you around to his way ofthinking, and with not much feeling for theother fellow.

He impresses people as being more capablethan he really is. He is popular and neverill at ease. Easily becomes the center ofattraction at any gathering. He is likely tobe a jack-of-all-trades. Although his interestsare varied, he is not necessarily well-versedin any of them. He possesses a sense ofhumor. His presence stimulates enthusiasmand very often he does arrive at a positionof importance.

Possibly he does not have any deep feeling.He would tend to be an opportunist. Likelyto succeed in things he intends to do. Hehas perhaps married a wife who would helphim in his purpose. He tends to be skeptical.

The following preliminary points areto be noted:

i. When a task of this kind is given,a normal adult is capable of respondingto the instruction by forming a uni-fied impression. Though he hears asequence of discrete terms, his result-ing impression is not discrete. In somemanner he shapes the separate qualitiesinto a single, consistent view. All sub-

jects in the following experiments, ofwhom there were over 1,000, fulfilledthe task in the manner described. Noone proceeded by reproducing the givenlist of terms, as one would in a rotememory experiment; nor did any ofthe subjects reply merely with syno-nyms of the given terms.

2. The characteristics seem to reachout beyond the merely given terms ofthe description. Starting from the bareterms, the final account is completedand rounded. Reference is made tocharacters and situations which are ap-parently not directly mentioned in thelist, but which are inferred from it.

3. The accounts of the subjectsdiverge from each other in importantrespects. This will not be surprisingin view of the variable content of theterms employed, which permits a con-siderable freedom in interpretation andweighting.

In the experiments to be reported thesubjects were given a group of traitson the basis of which they formed animpression. In view of the fact thatwe possess no principles in this regionto help in their systematic construction,it was necessary to invent groupings oftraits. xln this we were guided by aninformal sense of what traits were con-sistent with each other.

The procedure here employed isclearly different from the everydaysituation in which we follow the con-crete actions of an actual person. Wehave chosen to work with weak, in-cipient impressions, based on abbrevi-ated descriptions of personal qualities.Nevertheless, this procedure has somemerit for purposes of investigation,especially in observing the change ofimpressions, and is, we hope to show,relevant to more natural judgment.

More detailed features of the pro-cedure will be described subsequentlyin connection with the actual experi-

262 S. E. ASCH

ments. We shall now inquire intosome of the factors that determinethe content and alteration of suchimpressions.

.1. CENTRAL AND PERIPHERALCHARACTERISTICS

A. Variation of a Central QualityObservation suggests that not all

qualities have the same weight in estab-lishing the view of a person. Some arefelt to be basic, others secondary. Inthe following experiments we sought

consisting of pairs of traits, mostlyopposites. From each pair of terms inthis list, which the reader will findreproduced in Table i, the subject wasinstructed to select the one that wasmost in accordance with the view hehad formed. Terms were includedwhich were quite different from thoseappearing in the basic list, but whichcould be related to them. Of necessitywe were guided in the selection ofterms for the check list (as well as forthe experimental lists) by an informal

1. generous—ungenerous2. shrewd—wise3. unhappy—happy4. irritable—good-natured5. humorous—humorless6. sociable—unsociable

TABLE i

CHECK LIST I

7. popular—unpopular8. unreliable—reliable9. important—insignificant

10. ruthless—humane11. good-looking—unattractive12. persistent—unstable

13. frivolous—serious14. restrained—talkative15. self-centered—altruistic16. imaginative—hard-headed17. strong—weak18. dishonest—honest

for a demonstration of this process inthe course of the formation of animpression.

Experiment I

Two groups, A and B, heard reada list of character-qualities, identicalsave for one term. The list follows:A. intelligent—skillful—industrious—warm—

determined—practical—cautiousB. intelligent—skillful—industrious—cold—

determined—practical—cautious

Group A heard the person described as"warm"; Group B, as "cold."

Technique. The instructions were asdescribed above. Following the read-ing, each subject wrote a brief sketch.

The sketches furnish concrete evi-dence of the impressions formed. Theirexact analysis involves, however, serioustechnical difficulties. It seemed, there-fore, desirable to add a somewhatsimpler procedure for the determinationof the content of the impression andfor the purpose of group comparisons.To this end we constructed a check list

sense of what was fitting or relevant.Some of the terms were taken fromwritten sketches of subjects in prelimi-nary experiments. In the examinationof results we shall rely upon the writtensketches for evidence of the actualcharacter of the impressions, and weshall supplement these with the quanti-tative results from the check list.

There were 90 subjects in Group A(comprising four separate classroomgroups), 76 subjects in Group. B (com-prising four separate classroom groups).

Results. Are the impressions ofGroups A and B identical, with theexception that one has the addedquality of "warm," the other of "cold" ?This is one possible outcome. Anotherpossibility is that the differentiatingquality imparts a general plus or minusdirection to the resulting impression.We shall see that neither of theseformulations accurately describes theresults.

We note first that the characteristic"warm-cold" produces striking and con-


sistent differences of impression. Ingeneral, the A-impressions are far morepositive than the B-impressions. Wecite a. few representative examples:

Series A ("warm")A person who believes certain things to be

right, wants others to see his point, would besincere in an argument'and would like to seehis point won.

the frequency (in terms of percentages)with which each term in the check listwas selected. For the sake of brevityof presentation we state the results forthe positive term in each pair; thereader may determine the percentageof choices for the other term in eachpair by subtracting the given figurefrom 100. To illustrate, under Condition

TABLE 2

CHOICE OF FITTING QUALITIES (PERCENTAGES)

i. generous2. wise3- happy4. good-natured5. humorous6. sociable7. popular8. reliable9. important

ID. humaneii. good-looking12. persistent13. serious14. restrained15. altruistic1 6. imaginative17. strong1 8. honest

EXPERIMENT I

"WARM"N=9o

9165909477918494888677

IOO

IOO

7769519898

"COLD"N=76

82534171338289999316997998918199594

EXPERIMENT II

TOTALN=56

5549

69367i579688645898968244249595

"WARM"N=23

8773QI

QI

769183968791

969167684594

IOO

"COLD"N=33

3333585512

553997884553

IOO

IOO

94279

9692

EXPERIMENT III

"POLITE"N=20

5630758771839495945993

IOO

IOO

822933

IOO

87

"BLUNT"N=26

58506556486856

IOO

967779

IOO

IOO

774631

IOO

IOO

A scientist performing experiments and per-severing after many setbacks. He is drivenby the desire to accomplish something thatwould be of benefit.

Series B ("cold")A very ambitious and talented person who

would not let anyone or anything stand inthe way of achieving his goal. Wants hisown way, he is determined not to give in, nomatter what happens.

A rather snobbish person who feels that hissuccess and intelligence set him apart fromthe run-of-the-mill individual. Calculatingand unsympathetic.

This trend is fully confirmed in thecheck-list choices. In Table 2 we report

A of the present experiment, 91 percent of the subjects chose the designa-tion "generous"; the remaining 9 percent selected the designation "ungen-erous." Occasionally, a subject wouldnot state a choice for a particular pair.Therefore, the number of cases onwhich the figures are based is notalways identical; however, the fluc-tuations were minor, with the excep-tion of the category "good-looking—unattractive," which a larger proportionof subjects failed to answer.

We find:i. There are extreme reversals be-

264 S. E. ASCH

tween Groups A and B in the choiceof fitting characteristics. Certain quali-ties are preponderantly assigned to the"warm" person, while the opposingqualities are equally prominent in the"cold" person. This holds for thequalities of (i) generosity, (2) shrewd-ness, (3) happiness, (4) irritability,(5) humor, (6) sociability, (7) popu-larity, (10) ruthlessness, (15) self-cen-teredness, (16) imaginativeness.

2. There is another group of qualitieswhich is not affected by the transitionfrom "warm" to "cold," or only slightlyaffected. These are: (8) reliability,(9) importance, (u) physical attrac-tiveness, (12) persistence, (13) serious-ness, (14) restraint, (17) strength,(18) honesty.

These results show that a change inone character-quality has produced awidespread change in the entire impres-sion. Further, the written sketchesshow that the terms "warm-cold" didnot simply add a new quality, but tosome extent transformed the othercharacteristics. With this point we shalldeal more explicitly in the experimentsto follow.

That such transformations take placeis also a matter of everyday experience.If a man is intelligent, this has an effecton the way in which we perceive hisplayfulness, happiness, friendliness. Atthe same time, this extensive changedoes not function indiscriminately. The"warm" person is not seen more favor-ably in all respects. There is a rangeof qualities, among them a number thatare basic, which are not touched by thedistinction between "warm" and "cold."Both remain equally honest, strong,serious, reliable, etc.

The latter result is of interest withreference to one possible interpretationof the findings. It might be supposedthat the category "warm-cold" arouseda "mental set" or established a halo

tending toward a consistently plus orminus evaluation. We observe herethat this trend did not work in anindiscriminate manner, but was deci-sively limited at certain points. If weassume that the process of mutualinfluence took place in terms of theactual character of the qualities in ques-tion, it is not surprising that some will,by virtue of their content, remainunchanged.2

'The following will show that thesubjects generally felt the qualities"warm-cold" to be of primary impor-tance. We asked the subjects in certainof the groups to rank the terms ofLists A and B in order of their impor-tance for determining their impression.Table 3, containing the distribution ofrankings of "warm-cold," shows thatthese qualities ranked comparativelyhigh. At the same time a considerablenumber of subjects relegated "cold" tothe lowest position. That the rankingsare not higher is due to the fact thatthe lists contained other central traits.

These data, as well as the ranking ofthe other traits not here reproduced,point to the following conclusions:

1. The given characteristics do not allhave the same weight for the subject.He assigns to some a higher importancethan to others.

2. The weight of a given character-istic varies—within limits*—from subjectto subject.

Certain limitations of the check-listprocedure need to be considered:(i) The subject's reactions are forcedinto an appearance of discretenesswhich they do not actually possess, asthe written sketches show; (2) thecheck list requires the subject to choosebetween extreme characteristics, whichhe might prefer to avoid; (3) the quan-

2 This by no means excludes the possibility thatthe nuances of strength, honesty, etc., do changein relation to "warm-cold."

FORMING IMPRESSIONS OF PERSONALITY

titative data describe group trends; theydo not represent adequately the formof the individual impression. Generallythe individual responses exhibit muchstronger trends in a consistently posi-tive or negative direction. For thesereasons we employ the check-list resultsprimarily for the purpose of comparinggroup trends under different conditions.For this purpose the procedure is quiteadequate.

265

Under these conditions the selectionof fitting characteristics shows a signifi-cant change. The distribution of choicesfor the total group (see Table 2, columnlabeled "Total") now falls between the"warm" and "cold" variations of Experi-ment I. It appears that a more neutralimpression has formed.

The total group results are, however,largely a statistical artifact. An exami-nation of the check-list choices of the

TABLE 3

RANKINGS OF "WARM" AND "COLD": EXPERIMENT I

RANK

i2

34567

"WARM"

N

61544436

42

PERCENTAGE

M3510

10

10

714

100

"COLD"

N

12

8i2

32

13

41

PERCENTAGE

2721

2

57

. .533

100

B. Omission of a Central Quality

That the category "warm-cold" is sig-nificant for the total impression maybe demonstrated also by omitting it fromthe series. This we do in the followingexperiment.

Experiment II

The procedure was identical with thatof Experiment I, except that the terms"warm" and "cold" were omitted fromthe list read to the subject (intelligent-skillful - industrious - determined - prac-tical-cautious). Also the check list wasidentical with that of Experiment I,save that "warm-cold" was added as thelast pair. There were three groups,consisting of a total of 56 subjects.

subjects quickly revealed strong andconsistent individual differences. Theytended to be consistently positive ornegative in their evaluations. It willbe recalled that the terms "warm-cold"were added to the check list. Thispermitted us to subdivide the totalgroup according to whether they judgedthe described person on the check listas "warm" or "cold." Of the entiregroup, 23 subjects (or 41 per cent) fellinto the "warm" category. Our nextstep was to study the distribution ofchoices in the two subgroups. Theresults are clear: the two subgroupsdiverge consistently in the direction ofthe "warm" and the "cold" groups,respectively, of Experiment I. (SeeTable 2.) This is especially the case

266 S. E. ASCH

with the two "warm" series, which arevirtually identical.

It is of interest that the omission ofa term from the experimental list didnot function entirely as an omission.Instead, the subjects inferred the corre-sponding quality in either the positiveor negative direction. While notentirely conclusive, the results suggestthat a full impression of a person can-not remain indifferent to a category asfundamental as the one in question,and that a trend is set up to include itin the impression on the basis of the

distinguish them objectively. On thisassumption the addition or omission ofperipheral qualities should have smallereffects than those observed in Experi-ment I. We turn to this question inthe following experiment.

Experiment III

The following lists were read, eachto a different group:A,, intelligent—skillful—industrious—polite—

determined—practical—cautiousB. intelligent—skillful—industrious—blunt—

determined—practical—cautious

TABLE 4

RANKINGS OF "POLITE" AND "BLUNT": EXPERIMENT III

RANK

i2

34567

A: "POLITE"

N

0

0

0

2

34

rIO

19

PERCENTAGE

0

o0

10

1621

53

100

B: "BLUNT"

N

0

4356i7

26

PERCENTAGE

0

1512

19

23

427

100

given data. In later experiments toowe have found a strong trend to reachout toward evaluations which were notcontained in the original description.

C. Variation of a Peripheral Quality

Would a change of any character-quality produce an effect as strong asthat observed above? "Warm" and"cold" seem to be of special importancefor our conception of a person. Thiswas, in fact, the reason for selectingthem for study. If there are centralqualities, upon which the content ofother qualities depends, and dependentqualities which are secondarily de-termined, it should be possible to

The A group contained 20, the B group26 subjects.

The changes introduced into the selec-tion of fitting characteristics in thetransition from "polite" to "blunt" werefar weaker than those found in Experi-.ment I (see Table 2). There is furtherevidence that the subjects themselvesregarded these characteristics as rela-tively peripheral, especially the char-acteristic "polite." If we may take therankings as an index, then we mayconclude that a change in a peripheraltrait produces a weaker effect on thetotal impression than does a change ina central trait. (Though the changesproduced are weaker than those of


Experiment I, they are nevertheless sub-stantial. Possibly this is a consequenceof the thinness of the impression, whichresponds easily to slight changes.)

D. Transformation from a Central to aPeripheral Quality

The preceding experiments have dem-onstrated a process of discriminationbetween central and peripheral quali-ties. We ask: Are certain qualities

appearing in Table 5 shows. (CompareTable 3 of Experiment I.)

More enlightening are the subjects'comments. In Series A the quality"warm" is now seen as wholly de-pendent, dominated by others far moredecisive.

I think the warmth within this person is awarmth emanating from a follower to aleader.

TABLE 5

RANKINGS OF "WARM" AND "COLD": EXPERIMENT IV

RANK

i2

34567

"WARM"

SERIES A

N

i0

2

677.

23

PERCENTAGE

4o9273030

—

100

SERIES B

N

0

0

i4412

— -

21

PERCENTAGE

0o5191957

-

100

"COLD"

SERIES C

N

io32

I

2

II

20

PERCENTAGE

5o1510

51055

100

constantly central? Or is their func-tional value, too, dependent on theother characteristics?

Experiment IV

We selected for observation thequality "warm," which was demon-strated to exert a powerful effect onthe total impression (Experiments Iand II). The effect of the term wasstudied in the following two series:A. obedient—weak—shallow—warm—unam-

bitious—vainB. vain — shrewd — unscrupulous — warm —

shallow—envious

Immediately "warm" drops as a sig-nificant characteristic in relation to theothers, as the distribution of rankings

The term "warm" strikes one as being adog-like affection rather than a bright friend-liness. It is passive and without strength.

His submissiveness may lead people tothink he is kind and warm.

A more extreme transformation isobserved in Series B. In most instancesthe warmth of this person is felt tolack sincerity, as appears in the follow-ing protocols:

I assumed the person to appear warmrather than really to be warm.

He was warm only when it worked in withhis scheme to get others over to his side. Hiswarmth is not sincere.

A similar change was also observedin the content of "cold" in a further

268 S. E. ASCH

variation. The subject heard List B ofExperiment I followed by Series Cbelow, the task being to state whetherthe term "cold" had the same meaningin both lists.

C. intelligent—skillful—sincere—cold—con-scientious—helpful—modest

All subjects reported a difference.The quality "cold" became peripheralfor all in Series C. The following arerepresentative comments:

The coldness of i (Experiment I) borderson ruthlessness; 2 analyses coldly to differ-entiate between right and wrong.

i is cold inwardly and outwardly, while 2is cold only superficially.

i: cold means lack of sympathy and under-standing; a: cold means somewhat formal inmanner.

Coldness was the foremost characteristicof i. In 2 it seemed not very important, aquality that would disappear after you cameto know him.

That "cold" was transformed in thepresent series into a peripheral qualityis also confirmed by the rankings re-ported in Table 5.

We conclude that a quality, centralin one person, may undergo a changeof content in another person, and be-come subsidiary. When central, thequality has a different content andweight than when it is subsidiary.

Here we observe directly a processof grouping in the course of which thecontent of a trait changes in relationto its surroundings. Secondly, we ob-serve that the functional value of atrait, too—whether, for example, it be-comes central or not—is a consequenceof its relation to the set of surroundingtraits. At the same time we are ableto see more clearly the distinction be-tween central and peripheral traits. Itis inadequate to say that a central traitis more important, contributes morequantitatively to, or is more highly

correlated with, the final impressionthan a peripheral trait. The latter for-mulations are true, but they fail toconsider the qualitative process ofmutual determination between traits,namely, that a central trait determinesthe content and the functional place ofperipheral traits within the entire im-pression. In Series A, for example,the quality "warm" does not controlthe meaning of "weak," but is con-trolled by it.

The evidence may seem to supportthe conclusion that the same qualitywhich is central in one impression be-comes peripheral in another. Such aninterpretation would, however, containan ambiguity. While we may speak ofrelativity in the functional value of atrait within a person, in a deeper sensewe have here the opposite of relativity.For the sense of "warm" (or "cold")of Experiment I has not suffered achange of evaluation under the presentconditions. Quite the contrary; theterms in question change precisely be-cause the subject does not see the possi-bility of finding in this person thesame warmth he values so highly whenhe does meet it (correspondingly forcoldness).

Experiment V

The preceding experiments haveshown that the characteristics formingthe basis of an impression do not con-tribute each a fixed, independent mean-ing, but that their content is itselfpartly a function of the environmentof the other characteristics, of theirmutual relations. We propose now toinvestigate more directly the mannerin which the content of a given char-acteristic may undergo change.

Lists A and B were read to two sep-arate groups (including 38 and 41 sub-jects respectively). The first three


terms of ihe two lists are opposites; thefinal two terms are identical.A. kind—wise—honest—calm—strongB. cruel — shrewd — unscrupulous — calm —

strong

The instructions were to write downsynonyms for the given terms. Theinstructions read: "Suppose you had todescribe this person in the same man-ner, but without using the terms youheard, what other terms would youuse?" We are concerned with the

aspect of gentleness, while a grimmerside became prominent in Series B.3

Essentially the same may be said ofthe final term, "strong." Again, somesynonyms appear exclusively in one orthe other groups, and in the expecteddirections. Among these are:Series A: fearless—helpful—just—forceful—

courageous—reliableSeries B: ruthless—-overbearing — overpower-

ing—hard—inflexible—unbending—<•dominant

TABLE 6

SYNONYMS OP "CALM": EXPERIMENT V

serenecold, frigid, icy, cool, calculating, shrewd, nervy, schem-

ing, consciencelesssoothing, peaceful, gentle, tolerant, good-natured, mild-

manneredpoised, reserved, restful, unexcitable, unshakabledeliberate, silent, unperturbed, masterful, impassive,

collected, confident, relaxed, emotionless, steady,impassive, composed

"KIND" SERIES

18

oitii18

ii

"CRUEL" SERIES

3

20

O

0

7

26

synonyms given to the two final terms.In Table 6 we list those synonyms of

"calm" which occurred with differentfrequencies in the two groups. It willbe seen that terms appear in one groupwhich are not at all to be found in theother; further, some terms appear withconsiderably different frequencies underthe two conditions. These do not, how-ever, include the total group of syno-nyms; many scattered terms occurredequally in both groups.

We may conclude that the quality"calm" did not, at least in some cases,function as an independent, fixed trait,but that its content was determined byits relation to the other terms. As aconsequence, the quality "calm" was notthe same under the two experimentalconditions. In Series A it possessed an

The data of Table 6 provide evidenceof a tendency in the described direction,but its strength is probably underesti-mated. We have already mentionedthat certain synonyms appeared fre-quently in both series. But it is not tobe concluded that they therefore carriedthe same meaning. Doubtless the sameterms were at times applied in the twogroups with different meanings, pre-cisely because the subjects were underthe control of the factor being investi-gated. To mention one example: theterm "quiet" often occurred as a syno-nym of "calm" in both groups, but the

3 In an earlier investigation the writer (2) hasdealt with basically the same question though ina very different context. It was there shown thatcertain phenomena of judgment, which appearedto be due to changes of evaluation, were producedby a shift in the frame of reference.

270 S. E. ASCH

subjects may have intended a differentmeaning in the two cases. For thisreason Table 6 may not reveal the fullextent of the change introduced by thefactor of embedding.

The preceding experiments permit thefollowing conclusions:

1. There is a process of discriminationbetween central and peripheral traits.All traits do not have the same rank andvalue in the final impression. Thechange of a central trait may completelyalter the impression, while the changeof a peripheral trait has a far weakereffect (Experiments I, II, and III).

2. Both the cognitive content of a traitand its functional value are determinedin relation to its surroundings (Experi-ment IV).

3. Some traits determine both the con-tent and the function of other traits.The former we call central, the latterperipheral (Experiment IV).

II. THE FACTOR OF DIRECTIONIf impressions of the kind here in-

vestigated are a summation of the effectsof the separate characteristics, then anidentical set of characteristics shouldproduce a constant result. Is it possibleto alter the impression without chang-ing the particular characteristic? Weinvestigate this question below.

Experiment VI

The following series are read, each toa different group:A. intelligent—industrious—impulsive—criti-

cal—stubborn—enviousB. envious—stubborn—critical—impulsive—

industrious—intelligent

There were 34 subjects in Group A, 24in Group B.

The two series are identical withregard to their members, differing onlyin the order of succession of the latter.More particularly, Series A opens with

qualities of high merit (intelligent—industrious), proceeds to qualities thatpermit of a better or poorer evaluation(impulsive — critical — stubborn), andcloses with a dubious quality (envious).This order is reversed in Series B.

A considerable difference developsbetween the two groups taken as awhole. The impression produced by Ais predominantly that of an able personwho possesses certain shortcomingswhich do not, however, overshadow hismerits. On the other hand, B impressesthe majority as a "problem," whoseabilities are hampered by his seriousdifficulties. Further, some of the quali-ties (e.g., impulsiveness, criticalness) areinterpreted in a positive way underCondition A, while they take on, underCondition B, a negative color. Thistrend is not observed in all subjects, butit is found in the majority. A fewillustrative extracts follow:

Series AA person who knows what he wants and

goes after it. He is impatient at people whoare less gifted, and ambitious with those whostand in his way.

Is a forceful person, has his own convic-tions and is usually right about things. Isself-centered and desires his own way.

The person is intelligent and fortunately heputs his intelligence to work. That he isstubborn and impulsive may be due to thefact that he knows what he is saying andwhat he means and will not therefore give ineasily to someone else's idea which he dis-agrees with.

Series BThis person's good qualities such as indus-

try and intelligence are bound to be restrictedby jealousy and stubbornness. The person isemotional. He is unsuccessful because he isweak and allows his bad points to cover uphis good ones.

This individual is probably maladjustedbecause he is envious and impulsive.

In order to observe more directlythe transition in question, the writer


proceeded as follows. A new group(N=24) heard Series B, wrote thefree sketch, and immediately thereafterwrote the sketch in response to Series A.They were also asked to comment onthe relation between the two impres-sions. Under these conditions, with thetransition occurring in the same sub-jects, 14 out of 24 claimed that their

one. Somehow, he seems more intelligent,with his critical attitude helping that char-acteristic 6f intelligence, and he seems to beindustrious, perhaps because he is envious andwants to get ahead.

The check-list data appearing inTable 7 furnish quantitative support forthe conclusions drawn from the writtensketches.

TABLE 7

CHOICE OF FITTING QUALITIES (PERCENTAGES)

j. generous2. wise3. happy4. good-natured5. humorous6. sociable7. popular8. reliable9. important

10. humaneII. good-looking12. persistent13. serious14. restrained15. altruistic1 6. imaginative17. strong1 8. honest

EXPERIMENT VI

INTELLIGENT-* ENVIOUS(N=34)

24183*1852563584853674829764

6269480

ENVIOUS-* INTELLIGENT(N=24)

10

1750

21

27

14919021

3587

100

95

147379

EXPERIMENT VII

INTELLIGENT-* EVASIVE(N=46)

4235515453504496774959944491323774 ,66

EVASIVE-MNTSLLIGBNT(N=53>

23

19493729483994894653

100100

9125169681

impression suffered a change, while theremaining 10 subjects reported nochange. Some of the latter asserted thatthey had waited until the entire serieswas read before deciding upon theirimpression. The following are a fewcomments of the changing group:

You read the list in a different order andthereby caused a different type of person tocome to mind. This one is smarter, morelikeable, a go-getter, lively, headstrong, andwith a will of his own; he goes after what hewants.

The first individual seems to show his envyand criticism more than the second one.

This man does not seem so bad as the first

Under the given conditions the terms,the elements of the description, areidentical, but the resulting impressionsfrequently are not the same. Further,the relations of the terms to one anotherhave not been disturbed, as they mayhave been in Experiments I and II,with the addition and omission of parts.How can we understand the resultingdifference ?

The accounts of the subjects suggestthat the first terms set up in most sub-jects a direction which then exerts acontinuous effect on the latter terms.When the subject hears the first term,

272 S. E. ASCH

a broad, uncrystallized but directed im-pression is born. .The next jcharacter-istic comes not as a separate item, but isrelated to the established direction.Quickly the view formed acquires acertain stability, so that later character-istics are fitted—if conditions permit4—to the given direction.

Here we observe a factor of primacyguiding the development of an impres-

Some further evidence with regard tothis point is provided by the data withregard to ranking. We reproduce inTable 8 the rankings of the character-istic "envious" under the two conditions.

Experiment VII

It seemed desirable to repeat the pre-ceding experiment with a new series.As before, we reversed thie succession of

TABLE 8

RANKING OF "ENVIOUS": EXPERIMENT VI

RANK

i2

34

6

INTELLIGENT-»ENVIOUS

N

54

34

13

34

PERCENTAGE

15II159

it39

too

ENVIOUS-»!NTELLIGEN r

N

74522

4

24

PERCENTAGE

29i"21

88

100

sion. This factor is not, however, to beunderstood in the sense of Ebbinghaus,but rather in a structural sense. It isnot the sheer temporal position of theitem which is important as much as thefunctional relation of its content to thecontent of the items following it.5

4 For an instance in which the given conditionsmay destroy the established direction, see page 273.

6 In accordance with this interpretation theeffect of primacy should be abolished—or re-versed—if it does not stand in a fitting relationto the succeeding qualities, or if a certain qualitystands out as central despite its position. Thelatter was clearly the case for the quality "warm-cold" in Experiment I (see Table i) which,though occupying a middle position, ranked com-paratively high.

The distinction between the two senses ofprimacy could be studied experimentally bycomparing the recall of an identical series ofcharacter-qualities in two groups, one of whichreads them as a discrete list of terms, the otheras a set of characteristics describing a person.

terms. Unlike the preceding series,there is no gradual change in the meritof the given characteristics, but ratherthe abrupt introduction at the end (orat the beginning) of a highly dubioustrait. The series were:

A. intelligent— skillful — industrious—deter-mined—practical—cautious—evasive

B. evasive—cautious—practical—determined—industrious—skillful—intelligent

While the results are, for reasons tobe described, less clear than in the ex-periment preceding, there is still a defi-nite tendency for A to produce a morefavorable impression with greater fre-quency. We report below the moreextreme protocols in each series.

Series AHe seems to be a man of very excellent

FORMING IMPRESSIONS OF PERSONALITY 273character, though it is not unusual for oneperson to have all of those good qualities.

A scientist in an applied field, who doesnot like to discuss his work before it is com-pleted. Retiring and careful—but brilliant.Works alone, does not like to be annoyedwith questions. A very dynamic man.

A normal, intelligent person, who soundsas if he would be a good citizen, and of valueto all who know him.

He seems to have at least two traits whichare not consistent with the rest of his per-sonality. Being cautious and evasive contra-dicts his positive qualities. Altogether, he isa most unattractive person—the two above-mentioned traits overbalancing the others.

Series BThis is a man who has had to work for

everything he wanted—therefore he is eva-sive, cautious and practical. He is naturallyintelligent, but his struggles have made himhard.

He is out for himself, is very capable buttends to use his skill for his own benefit.

He is so determined to succeed that herelies on any means, making use of his cun-ning and evasive powers.

Questioning disclosed that, under thegiven conditions, the quality "evasive"produced unusual difficulty. Most sub-jects in both groups felt a contradictionbetween it and the series as a whole. Inresponse to the question, "Were thereany characteristics that did not fit withthe others?" n out of 27 in Group Amentioned "evasive" while it was men-tioned by ir out of a total of 30 inGroup B.

It is of interest to observe how thiscrucial term was dealt with by individ-ual subjects. Some in Group A feltunable to reconcile it with the view theyhad formed; consequently they rele-gated it to a subsidiary position and, inthe most extreme cases, completelyexcluded it. Others reported the oppo-site effect: the final term completelyundid their impression and forced a newview. The following comments areillustrative:

Series AI put this characteristic in the background

and said it may be a dependent characteristicof the person, which does not dominate hispersonality, and does not influence his actionsto a large extent.

I excluded it because the other characteris-tics which fitted together so well were somuch more predominant. In my first impres-sion it was left out completely.

It changed my entire idea of the person—changing his attitude toward others, the typeof position he'd be likely to hold, the amountof happiness he'd have—and it gave a certainamount of change of character (even for traitsnot mentioned), and a tendency to think ofthe person as somewhat sneaky or sly.

Similar reactions occur in Group B,but with changed frequencies.

The importance of the order of im-pressions of a person in daily experienceis a matter of general observation andis perhaps related to the process underinvestigation. It may be the basis forthe importance attached to first im-pressions. It is a matter of generalexperience that we may have a "wrongslant" on a person, because certain char-acteristics first observed are given a cen-tral position when they are actuallysubsidiary, or vice versa.

Experiment VIII

We studied the factor of direction inyet another way. Series A of ExperimentVI was divided in two parts and pre-sented to a new group as a descriptionof two persons. The new series were:A. intelligent—industrious—impulsiveB. critical—stubborn—envious

Procedure, (i) Series A was read tothis group (Group i), followed by thewritten sketch and the check list. (2)The subjects were instructed that theywould hear a new group of terms de-scribing a second person. Series B wasread and' the usual information wasobtained. (3) Upon completion of thesecond task the subjects were informed

274 S. E. ASCH

that the two lists described a single per-son. They were instructed to form animpression corresponding to the entirelist of terms. Certain questions weresubsequently asked concerning the laststep which will be described below. Acontrol group (Group 2) respondedonly to the entire list of six terms (asin Series A of Experiment VI), andanswered some of the final questions.

seemed to be a mass ofThe personcontradictions.

He seemed a dual personality. There aretwo directions in this person.

On the other hand, only a minority inGroup 2 (9 out of 24) report any diffi-culty. Further, the reasons given by thelatter are entirely different from thoseof Group i. These subjects speak invery general terms, as:

TABLE 9

CHOICE OP FITTING QUALITIES: EXPERIMENT VIII (PERCENTAGES)

i. generous2. wise3. happy4. good-natured5. humorous6. sociable7. popular8. reliable9. important

INTELLIGENT-INDUSTRIODS-

IMPULSIVE

(N=52)

874884748789948590

CRITICAI,-STUBBORN-

ENVIOUS

(N=52)

630

312

24

94724

10. humaneii. good-looking12. persistent13. serious14. restrained15. altruistic1 6. imaginative17. strong1 8. honest

INTELLIGENT-INDUSTRIOUS-

IMPULSIVE

(N=52)

8781858716666594

100

CRITICAL-STUBBORN-

F.NVIOUS

(N=52)

1936678337

o155058

We are concerned mainly to see howGroup i dealt with the final task, theestablishing of an impression based onthe two smaller series. That Lists A andB were widely different will be clear inthe check-list results of Table 9.

Most subjects of Group i expressedastonishment at the final information(of Step 3) and showed some reluctanceto proceed. In response to the question,"Did you experience difficulty in form-ing an impression on the basis of thesix terms," the majority of Group i (32out of 52) replied in the affirmative.The reasons given were highly uniform:the two sets of traits seemed entirelycontradictory.

I had seen the two sets of characteristics asopposing each other. It was hard to envisionall these contradictory traits in one person.

These characteristics are possessed by every-one in some degree or other. The terms donot give an inclusive picture.

Only two subjects in Group 2 mentioncontradiction between traits as a sourceof difficulty.

The formation of the complete im-pression proceeds differently in the twogroups. Series A and B are at firstreferred, in Group i, to entirely differ-ent persons. Each is completed in itsdirection, and the fact that they comesuccessively seems to enhance the con-trast between them. It is thereforedifficult for them to enter the newimpression. Some subjects are unableto reconcile the two directions com-pletely; in consequence their divergencebecomes the paramount fact, as the fol-lowing protocols illustrate:


The directions reacted on each other andwere modified, so that the pull in each direc-tion is now less strong. This gives a Jekylland Hyde appearance to this person.

I applied A to the business half of theman—as he appeared and acted during work-ing hours. B I referred to the man's sociallife.

The independent development of Aand B is on the other hand preventedin Group 2, where they function fromthe start as parts of one description.6

This conclusion is in general con-firmed by the following observation.To the question: "Did you proceed bycombining the two earlier impressionsor by forming a new impression?" thefollowing responses are obtained: (a)33 of 52 subjects answer that theyformed a new impression, differentfrom either A or B; 12 subjects speakof combining the two impressions, while7 subjects assert that they resorted toboth procedures. The following aretypical responses in the first subgroup:

I couldn't combine the personalities of Aand B. I formed an entirely new impression.

I can conceive of the two sets of character-istics in one person, but I cannot conceive ofmy impressions of them as belonging to oneperson.

As I have set down the impressions, one isexactly the opposite of the other. But I canfit the six characteristics to one person.

That the terms of Series A and Boften suffered considerable change whenthey were viewed as part of one seriesbecomes evident in the replies to anotherquestion. The subjects were asked,"Did the terms of the series A and Bretain for you their first meaning or didthey change?" Most subjects describe

6 The procedure of "successive impressions"here employed might be extended to the study ofthe effect of early upon later impressions. Forexample, the impression resulting from thesequence (A) + (B) might be compared with thereverse sequence (B) + (A), and each of thesewith the sequence (A+B) or (B+A).

a change in one or more of the traits, ofwhich the following are representative:

In A impulsive grew out of imaginative-ness; now it has more the quality of hastiness.

Industriousness becomes more self-centered.Critical is now not a derisive but rather a

constructive activity.Stubborn had an entirely personal mean-

ing; now it refers to being set in one's ideas.

The tenor of most replies is well rep-resented by the following comment:

When the two came together, a modifiestion occurred as well as a limiting boundaryto the qualities to which each was referred.

III. STRONGLY SIMPLIFIED IMPRESSIO>,STo a marked degree the impressions

here examined possess a strongly unifiedcharacter. At the same time they lackthe nuances and discriminations that afull-fledged understanding of anotherperson provides. Therefore they can beeasily dominated by a single direction.We propose now to observe in a moredirect and extreme manner the forma-tion of a global impression.

Experiment IXWe select from the series of Experi-

ment I three terms: intelligent—skill-ful—warm—all referring to-strong posi-tive characteristics. These form thebasis of judgment. The results appearin Table 10.

There develops a one-directed im-pression, far stronger than any observedin the preceding experiments. Thewritten sketches, too, are unanimouslyenthusiastic. The impression also de-velops effortlessly.

Negative characteristics hardly in-trude. That this fails to happen raisesa problem. Many negative qualitiescould quite understandably be livingtogether with those given. But the sub-jects do not as a rule complete them inthis direction. This, indeed, they seemto avoid.

276 S. E. ASCH

Experiment IXaThe next step was to observe an im-

pression based on a single trait. Thereare two groups; one group is instructedto select from the check list those char-acteristics wKich belong to a "warm"person, the second group those belong-ing to a "cold" person. The resultsappear in Table 10.

In order to show more clearly the

The differences between "warm" and"cold" are now even more considerablethan those observed in Experiment I.No qualities remain untouched. Buteven under these extreme conditionsthe characterizations do not becomeindiscriminately positive or negative."Warm" stands for very positive quali-ties, but it also carries the sense of acertain easy-goingness, of a lack of re-

TABLE 10

CHOICE OF FITTING QUALITIES: EXPERIMENT IX (PERCENTAGES)

i. generous2. wise3. happy4. good-natured5. humorous6. sociable7. popular8. reliable9. important

INTELLIGENT-SKILLFUL-

WARM

(N=34)

IOO

97IOO

IOO

IOO

IOO

IOO

IOO84

WARM(N=22)

IOO

95IOOIOOIOOIOOIOOIOO

68

COLD(N=33)

12II108

12

96

«754

ID. humaneii. good-looking12. persistent13. serious14. restrained15. altruistic1 6. imaginative17. strong1 8. honest

INTELLIGENT-SKILLFUL-

WARM

(N=34)

9772

IOO

IOO

66978297

IOO

WARM(N=22)

IOO

957868419i9574

IOO

COLD(N=33)

175797979739

8781

range of qualities affected by the giventerms we constructed a second checklist (Check List II) to which the sub-jects were to respond in the manneralready described. The results are re-ported in Table n.

A remarkably wide range of qualitiesis embraced in the dimension "warm-cold." It has reference to tempera-mental characteristics (e.g., optimism,humor, happiness), to basic relations tothe group (e.g., generosity, sociability,popularity), to strength of character(e.g., persistence, honesty). It evenincludes a reference to physical charac-teristics, evident in the virtually unani-mous characterizations of the warmperson as short, stout, and ruddy, andin the opposed characterizations of thecold person.

straint and persistence, qualities whichare eminently present in "cold." Asimplified impression is not to be simplyidentified with a failure to make distinc-tions or qualifications. Rather, what wefind is that in a global view the distinc-tions are drawn bluntly.

The consistent tendency for the dis-tribution of choices to be less extremein Experiment I requires the revision ofan earlier formulation. We have saidthat central qualities determine the con-tent and functional value of peripheralqualities. It can now be seen that thecentral characteristics, while imposingtheir direction upon the total impres-sion, were themselves affected by thesurrounding characteristics.

Upon the conclusion of the experi-ments, the subjects were asked to state


the reason for their choice of one pre-dominant direction in their characteri-zations. All agreed that they felt such atendency. Some cannot explain it, say-ing, in the words of one subject: "I donot know the reason; only that this isthe way it 'hit' me at the moment"; or:"I did not consciously mean to choosethe positive traits." Most subjects, how-ever, are explicit in stating that the

It is of interest for the theory of ourproblem that there are terms whichsimultaneously contain implications forwide regions of the person. Many termsdenoting personal characteristics showthe same property. They do not observea strict division of labor, each pointingneatly to one specific characteristic;rather, each sweeps over a wide area andaffects it in a definite manner.7

TABLE ii

CHECK LIST II: CHOICE OF FITTING QUALITIES: EXPERIMENT IXA (PERCENTAGES)

i. emotional2. practical3. optimistic4. informal5. cheerful6. short7. modest8. imaginative9. thin

10. intelligentn. brave12. pale

WARM(N=22)

100409595

100

91869515819i15

COLD(N=33>

12

73170

1889

2893967497

i

unemotionaltheoreticalpessimisticformalsadtallproudunimaginativestoutunintelligentcowardlyruddy

WARM(N=22)

0

60550

9M

585199

»5

COLD(N=33)

882783

10082929'72

74

263

given traits seemed to require comple-tion in one direction. The followingstatements are representative:

These qualities initiate other qualities. Aman who is warm would be friendly, conse-quently happy. If he is intelligent, he wouldbe honest.

The given characteristics, though very gen-eral, were good characteristics. Thereforeother good characteristics seemed to belong.When, for example, I think of a person aswarm, I mean that he couldn't be ugly.

This was the tenor of most statements.A few show factors at work of a some-what different kind, of interest to thestudent of personality, as:

I naturally picked the best trait because Ihoped the person would be that way.

I went in the positive direction because Iwould like to be all those things.

Some would say that this is a seman-tic problem. To do so would be, how-

7 On the basis of the last findings an objectionmight be advanced against our earlier account ofthe distinction between central and peripheraltraits. If, as has just been shown, "warm" refersto such a wide range of qualities, then the forceof the demonstration (see Experiment I) that itexerts a great effect on the final impressionseems to be endangered. Is it to be wandered atthat, this quality, which is single only in a lin-guistic sense, but psychologically plural, should beso effective? And should not the distinction bedrawn rather between qualities which containmany other qualities and qualities—such as"politeness"—that are much more specific inrange ?

The objection presupposes that a quantitativelylarger number of qualities will exert a greatereffect than a smaller number. But this assumptionis precisely what needs to be explained. Whydoes not the more inclusive term provide a greaternumber of occasions for being affected by otherterms? What the assertion fails to face is thatthere is a particular direction of forces.

278 S. E. ASCH

ever, to beg the question by disposingof the psychological process that givesrise to the semantic problem. Whatrequires explanation is how a term, anda highly "subjective" one at that, refersso consistently to so wide a region ofpersonal qualities. It seems similarlyunfruitful to call these judgments ster-eotypes. The meaning of stereotype isitself badly in need of psychologicalclarification. Indeed, in the light of our

IV. SIMILARITY AND DIFFERENCE OFIMPRESSIONS

The preceding discussion has definiteconsequences for the perception ofidentity and difference between thecharacteristics of different persons. Ofthese the most significant for theory isthe proposition that a given trait in twodifferent persons may not be the sametrait, and, contrariwise, that two differ-ent traits may be functionally identical

TABLE 12 ,RESEMBLANCE OF SETS: EXPERIMENT X

I2

34

SET I RESEMBLES

N

0

10

680

78

PERCENTAGE

o13870

100

SET 2 RESEMBLES

N

70

566

78

PERCENTAGE

90

685

1 00

observations, a stereotype appears (in afirst approximation) to be a centralquality belonging to an extremely sim-plified impression.

We propose that there is, under thegiven conditions, a tendency to graspthe characteristics in their most out-spoken, most unqualified sense, and onthat basis to complete the impression.The subject aims at a clear view; hetherefore takes the given terms in theirmost complete sense. (What is said herewith regard to the present experimentseems to apply also to the precedingexperiments. In each case the subject'simpression is a blunt, definite char-acterization. It lacks depth but notdefmiteness. Even when the view isof a mediocre character, it is out-spokenly so.) The comments of thesubjects are in agreement with thepresent interpretation.

in two different persons. We turn nowto an investigation of some conditionswhich determine similarity and differ-ence between personal qualities.

Experiment XI. The group has before it Sets i, 2, 3,

and 4 with instructions to state (i)which of the other three sets mostresembles Set i, and (2) which mostresembles Set 2.

SET i SET 2 SET 3 SET 4quick quick slow slowskillful clumsy skillful clumsyhelpful helpful helpful helpful

One quality—"helpful"—remains con-stant in all sets. The other two quali-ties appear in their positive form inSet i, and are changed to their oppo-sites singly and together in the threeother sets.

A remarkable uniformity appears inthe findings, reported in Table 12.


Set i is equated with Set 3 in 87 percent of the cases, while its similarity toSet 2 is reported in only 13 per cent ofthe cases. Similarly, Set 2 is asserted toresemble Set 4 in 85 per cent of thecases, while the resemblance to Set idrops to 9 per cent.

Set 2?" The results appear in Table 13.We see that qualities which, abstractly

taken, are identical, are infrequentlyequated, while qualities which areabstractly opposed are equated withgreater frequency. For example, thequality "quick" of Sets i and 2 is

TABLE 13

RESEMBLANCE OF QUALITIES: EXPERIMENT X

"QUICK" OF SET i RESEMBLES "QUICK" OF SET 2 RESEMBLES

"quick" of set 2"helpful" of set 2"slow" of set 3"skillful" of set 3"helpful" of set 3

N

iii

1621

I

50

PERCENTAGE

22

2

32422

100

"HELPFUL" OF SET i RESEMBLES

"helpful" of set 2"helpful" of set 4"helpful" of set 3"skillful" of set 3

N

72

336

48

PERCENTAGE

154

6813

i

100

i

"quick" of set I"slow" of set 4"clumsy" of set 4"slow" of set 3"helpful" of set 3

N

10

2:721

41

PERCENTAGE

245i1753

100

"HELPFUL" OF SET 2 RESEMBLES

! "helpful" of set I"quick" of set I"slow" of set 4"helpful" of set 4"clumsy" of set 4"slow" of set 3

N

522

3«

43

46

PERCEKTAGE

ii *44

6597

100

The choice of similar sets cannot inthis case be determined merely on thebasis of the number of "identical ele-ments," for on this criterion Sets 2 and3 are equally similar to i, while Sets iand 4 are equally similar to 2. Whatfactors may be said to determine thedecisions with regard to similarity anddifference ?

We come somewhat closer to ananswer in the replies to the followingquestion: "Which characteristics in theother sets resemble most closely (a)'quick' of Set i? (b) 'quick' of Set 2?(c) 'helpful' of Set i? (d) 'helpful' of

matched in only 22 and 25 per cent ofthe cases, respectively, while "quick" ofSet i is, in 32 per cent of the cases,matched with "slow" of Set 3, and"quick" of Set 2 with "slow" of Set 4in 51 per cent of the cases.8

At this point the reports of the sub-jects become very helpful. They wererequested at the conclusion to state inwriting whether the quality "quick" in

8 In a forthcoming publication the writer willdeal with theoretically similar issues in the con-text of a problem in social psychology. This willbe the report of an investigation of changes inthe content of identical social assertions when theyfunction as part of different frames of reference.

280 S. E. ASCH

Sets i and 2 was identical or different,together with their reasons, and simi-larly to compare the quality "slow" inSets 3 and 4. The written accountspermit of certain conclusions, which arestated below.

1. The content of the quality changeswith a change in its environment. Theprotocols Below, which are typical, willshow that the "quicks" of Sets i and 2are phenomenally different, and simi-larly for the "slows" of Sets 3 and 4.

The quickness of i is one of assurance, ofsmoothness of movement; that of 2 is a forcedquickness, in an effort to be helpful.

i is fast in a smooth, easy-flowing way; theother (2) is quick in a bustling way—thekind that rushes up immediately at yourrequest and tips over the lamps.

3 takes his time in a deliberate way; 4would like to work quickly, but cannot—there is something painful in his slowness.

3 is slow in a methodical, sure way, aimingtoward perfection; in 4 it implies a certainheaviness, torpor.

2. The dynamic sources of the qualityare relationally determined. In theprotocols we observe a process of mutualdetermination between traits. They aregrasped as not simply contiguous to oneanother but in dynamic relation, inwhich one is determined by, or springsfrom, the other.

i is quick because he is skillful; 2 is clumsybecause he is so fast.

Great skill gave rise to the speed of i,whereas 2 is clumsy because he does every-thing so quickly.

quick Skillful

In 3 slowness indicates care, pride in workwell-done. Slowness in 4 indicates sluggish-ness, poor motor coordination, some physicalretardation.

Speed and skill are not connected asare speed and clumsiness. Withoutexception, "quick" is perceived to springfrom skill (skillful—Kruick); but thevector in Set 2 is reversed, "clumsy"becoming a consequence of speed(clumsy«-quick). While Sets i and 3are identical with regard to the vectors,Set 2 is not equivalent to 4, the slow-ness and clumsiness of 4 being sensedas part of a single process, such assluggishness and general retardation(slow*=?clumsy).

3. Dynamic consequences are graspedin the interaction of qualities. "Quick"and "skillful" (as well as "slow" and"skillful") are felt as cooperating,whereas "quick" and "clumsy" cancelone another.

2 drops everything fast. He is fast butaccomplishes nothing. The clumsy manmight be better off if he were slow.

The second person is futile; he is quick tocome to your aid and also quick to get inyour way and under your hair.

i can afford to be quick; 2 would be farbetter off if he took things more slowly.9

In the light of these comments, whichare representative, we are able to formu-late the prevailing direction of the rela-tions within the sets.

In Sets i and 3 the prevailing struc-ture may be represented as:

slow skilful

The quality slow is, in person 3, somethingdeliberately cultivated, in order to attain ahigher order of skill.

8 Parallel experiments in which the last term ofthe sets was changed to "not helpful" gave resultsessentially identical with the above.


"Quick-slow" derive their concretecharacter from the quality "skillful";these in turn stand in a relation of har-mony to "helpful," in the sense thatthey form a proper basis for it and makeit possible.

In Sets 2 and 4 the characteristicstructures are as follows:

helpful

c|uick clumsy

similar processes. In view of the factthat such analyses have not been pre-viously reported, we select for briefdescription a few additional examples.

The task was to state whether theterm "aggressive" was alike or differentin Sets i and 2, and 3 and 4, respec-tively. This example will be of particu-

helforul

slow clumsy

But now these stand in a relation ofinherent contradiction to the quality"helpful," the fulfillment of which theynegate.

Our results contain a proportion ofcases (see Tables 12 and 13) that arecontrary to the described general trend.These do equate the characteristic of iand 2 and of 3 and 4. They requireexplanation. It is especially importantto decide whether the disagreements arecapricious or whether they have anunderstandable basis. As a rule we findin these cases that the given quality isviewed in a narrower, more limitedway. For example, these subjects view"quick" of Sets i and 2 in terms ofsheer tempo, deliberately excluding forthe moment considerations of fitness.The following protocols are illustrative:

These persons' reactions to stimuli are bothquick, even though the results of their actionsare in opposite directions.

They are both quick, but they differ in thesuccess of their actions.

The two terms are basically the same, forboth would execute their tasks with theirindividual maximum speed.

II. The reader will readily think ofother sets of characteristics involving

lar interest to psychologists, in view ofcurrent discussions of aggressiveness.

SET i SET 2 SET 3 SET 4active lazy weak stronghelpful unhelpful sensitive self-centeredaggressive aggressive aggressive aggressive

Nineteen out of 20 subjects judge theterm to be different in Sets i and 2; 17out of 20 judge it to be different in Sets3 and 4. Some representative reportsfollow:

The aggressiveness of i is friendly, open,and forceful; 2 will be aggressive whensomething offends him.

The aggressiveness of i is an expression ofconfidence in his abilities, of his strength ofwill and mind; in 2 it is a defensive measureto cover sensitivity.

3 will be aggressive to try to hide his weak-ness. The aggressiveness of 4 is a naturalresult of his strength and self-centeredness.

4 is aggressive because he has needs to besatisfied and wishes nothing to stand in hisway; 3 has the aggressiveness of self-pity andindecision.

In nearly all cases the sources ofaggression and its objects are sensed tobe different. In consequence, the formit takes and its very psychological con-tent become different in the seriescompared.

282 S. E. ASCH

Substantially the same results areobserved in another group in the com-parison of "unaggressive" in Sets i and2 below.

SET iactivehelpfulunaggressive

SET 2weaksensitiveunaggressive

Twenty-eight out of 30 subjects call"unaggressive" different in the twoseries. Some of their reasons follow:

Unaggressive in i might mean that he doesnot push or force his way into things. In thesecond case it may mean meekness or fear ofpeople.

1 does not care to be aggressive; 2 lacks thestamina for it.

2 does not fight back at the world nor tryto rise above his weaknesses.

The word "aggressive" must have the sameconnotations in both cases; otherwise why notuse different terms to express different things?

III. The second and third terms inSets i and 2 below were compared,respectively.

SET iintelligentcriticalstubborn

SET 2impulsivecriticalstubborn

The intelligent person might be stubbornabout important things, things that meansomething to him, that he knows somethingabout; whereas an impulsive person might bestubborn just to be contrary.

An intelligent person may be stubbornbecause he has a reason for it and thinks it'sthe best thing to do, while an impulsive per-son may be stubborn because at the momenthe feels like it.

Some representative statements defend-ing the identity of "stubborn" in the twoseries follow:

Stubbornness to me is the same in any lan-guage. Of course, an intelligent person mayhave a better reason for being stubborn thanan impulsive one, but that does not neces-sarily change the degree of stubbornness.

Both refuse to admit to anything that doesnot coincide with their opinion.

In my opinion there is only one kind ofstubbornness—an unswerving desire either todo or not to do a certain thing.

IV. In the following series the secondand third terms were to be compared:

SET i SET 2warm cold

All subjects in a group of 31 judgedthe term "critical" to be different in thetwo sets; while 19 (or 61 per cent)judged "stubborn" as different. A fewof the remarks follow:

Critical:i is critical because he is intelligent; 2

because he is impulsive.The intelligent individual is critical in a

constructive manner; the impulsive one prob-ably hurls criticism unthinkingly.

The intelligent person may be critical in acompletely impersonal way; 2 may be criticalof people, their actions, their dress, etc.

Stubborn:The stubbornness of an intelligent person

is more likely to be based on reason and itcan be affected by reasoning.

wittypersuasive

wittypersuasive

Twenty-seven of 30 subjects judged"persuasive" as different; all judged"witty" to be different. A few of thecomments follow:

Witty:i laughs with the audience; 2 is either

laughing at or trying to make others laugh atsome one. 2 is satirical, not humorous.

i has a jolly and happy-go-lucky wit. 2will use wit as one uses a bow and arrow—with precision. He will have a target whichwill not be missed.

The wit of the warm person touches theheart. The cold person's wit is touched withirony.

Persuasive:1 is persuasive in trying to help others; 2 in

trying to help himself.2 may persuade through fear.


2 would be detached in his arguments; iwould appeal more to the inner emotionalbeing of others.

V. The term "gay" was compared inthe following series:

SET i SET 2gayintelligentindustrious

Say _stupidlazy

Twenty-seven of 30 subjects call "gay"different. Some representative reasonsfollow:

They may both be equally gay, but theformer is different. The stupid person canbe gay over serious, sad matters, while theintelligent person is gay with reason.

The first person's gaiety comes from full-ness of life; 2 is gay because he knows nobelter.

i knows when to be gay and when notto be.

The gaiety of i is active and energetic; thegaiety of 2 is passive.

The intelligent person is gay in an intelli-gent way.

They are the same—gaiety has no relationto intelligence and industriousness.

The foregoing observations describe aprocess of relational determination ofcharacter-qualities. A given qualityderives its full concrete content from itsplace within the system formed by therelations of the qualities. Some quali-ties are seen as a dynamic outgrowth ofdetermining qualities. Qualities areseen to stand in a relation of harmonyor contradiction to others within thesystem. These processes set require-ments for the comparison of impres-sions. Identical qualities in differentstructures may cease to be identical: thevectors out of which they grow mayalter, with the consequence that theirvery content undergoes radical change.In the extreme case, the same quality intwo persons will have different, evenopposed, meanings, while two opposed

283

qualities will have the same functionwithin their respective structures.

DISCUSSION

IThe investigations here reported have

their starting-point in one problem andconverge on one basic conclusion. Indifferent ways the observations havedemonstrated that forming an impres-sion is an organized process; that char-acteristics are perceived in their dynamicrelations; that central qualities are dis-covered, leading to the distinctionbetween them and peripheral qualities;that relations of harmony and contra-diction are observed. To know a per-son is to have a grasp of a particularstructure.

Before proceeding it may be helpfulto note two preliminary points. First:For the sake of convenience of expres-sion we speak in this discussion offorming an impression of a person,though our observations are restrictedentirely to impressions based on descrip-tive materials. We do not intend toimply that observations of actual per-sons would not involve other processeswhich we have failed to find under thepresent conditions; we are certain thatthey would (see p. 288 ff.). But we seeno reason to doubt that the basic fea-tures we were able to observe are alsopresent in the judgment of actual per-sons. Secondly: We have not dealt inthis investigation with the role of indi-vidual differences, of which the mostobvious would be the effect of the sub-ject's own personal qualities on thenature of his impression. Though theissue of individual differences is unques-tionably important, it seemed desirableto turn first to those processes whichhold generally, despite individual differ-ences. A proper study of individualdifferences can best be pursued when a

284 S. E. ASCH

minimum theoretical clarification hasbeen reached.

Let us briefly reformulate the mainpoints in the procedure o£ our subjects:

i. There is an attempt to form animpression of the entire person. ( Thesubject can see the person only as aunit;10 he cannot form an impressionof one-half or of one-quarter of theperson. This is the case even when thefactual basis is meager; the impressionthen strives to become complete, reach-ing out toward other compatible quali-ties. The subject seeks to reach the coreof the person through the trait or traits.

2. As soon as two or more traits areunderstood to belong to one person,they cease to exist as isolated traits, andcome into immediate dynamic inter-action.11 The subject perceives not thisand that quality, but the two enteringinto a particular relation. There takesplace a process of organization in thecourse of which the traits order them-selves into a structure. It may be saidthat the traits lead an intensely sociallife, striving to join each other in aclosely organized system. The repre-sentation in us of the character ofanother person possesses in a strikingsense certain of the qualities of a system.

3. In the course of this process somecharacteristics are discovered to be cen-tral. The whole system of relationsdetermines which will become central.These set the direction for the furtherview of the person and for the concreti-

10 To be sure, we do often react to people in amore narrow manner, as when we have dealingswith the ticket-collector or bank teller. It cannothowever be said that in such instances we are pri-marily oriented to the other as a person. Themoment our special attitude would give way to agenuine interest in the other, the point statedabove would fully apply.

11 We cannot say on the basis of our observa-tions whether exceptions to this statement occur,e.g., whether some traits may be seen as acci-dental, having no relation to the rest of the per-son. It seems more likely that even insignificanttraits are seen as part of the person.

zation of the dependent traits. As arule the several traits do not have equalweight. And it is not until we havefound the center that we experience theassurance of having come near to anunderstanding of the person.

4. The single trait possesses the prop-erty of a part in a whole. A change ina single trait may alter not that aspectalone, but many others—at times all.As soon as we isolate a trait we not onlylose the distinctive organization of theperson; the trait itself becomes abstract.The trait develops its full content andweight only when it finds its placewithin the whole impression.

5. Each trait is a trait of the entireperson. It refers to a characteristicform of action or attitude which belongsto the person as a whole. In this sensewe may speak of traits as possessing theproperties of Ehrenfels-qualities. Traitsare not to be considerd as referring todifferent regions of the personality, onthe analogy of geographical regionswhich border on another.

6. Each trait functions as a repre-sentative of the person. We do notexperience anonymous traits the par-ticular organization of which constitutesthe identity of the person. Rather theentire person speaks through each of hisqualities, though not with the sameclearness.

7. In the process of mutual inter-action the concrete character of eachtrait is developed in accordance withthe dynamic requirements set for it byits environment. There is involvedan understanding of necessary conse-quences following from certain givencharacteristics for others. The envy ofa proud man is, for example, seen tohave a different basis from the envy ofa modest man.

8. On this basis consistencies andcontradictions are discovered. Certainqualities are seen to cooperate; others to


negate each other. But we are not con-tent simply to note inconsistencies or tolet them sit where they are. The con-tradiction is puzzling, and prompts usto look more deeply. Disturbing fac-tors arouse a trend to maintain the unityof the impression, to search for themost sensible way in which the charac-teristics could exist together,12 or todecide that we have not found the keyto the person. We feel that properunderstanding would eliminate, not thepresence of inner tensions and incon-sistencies, but of sheer contradiction.(It may be relevant to point out that thevery sense of one trait being in contra-diction to others would not arise if wewere not oriented to the entire person.Without the assumption of a unitaryperson there would be just differenttraits.)

9. It follows that the content andfunctional value of a trait changes withthe given context. This statement ex-presses for our problem a principleformulated in gestalt theory with regardto the identity of parts in different struc-tures (8, 10). A trait central in oneperson may be seen as secondary inanother. Or a quality which is nowreferred to the person may in anothercase be referred to outer conditions. (Inthe extreme case a quality may beneglected, because it does not touchwhat is important in the person.)

We conclude that the formation andchange of impressions consist of specificprocesses of organization. Further, itseems probable that these processes arenot specific to impressions of personsalone. It is a task for future investiga-tion to determine whether processes ofthis order are at work in other impor-

12 Indeed, the perception of such contradiction,or of the failure of a trait to fit to the others, maybe of fundamental importance for gaining aproper view. It may point to a critical regionin the person, in which things are not as theyshould be.

tant regions of psychology, such as informing the view of a group, or ofthe relations between one person andanother.

II

It may be of interest to relate theassumptions underlying the naive pro-cedure of our subjects to certain cus-tomary formulations, (i) It should nowbe clear that the subjects express certaindefinite assumptions concerning thestructure of a personality. The gainingof an impression is for them not aprocess of fixing each trait in isolationand noting its meaning. If they pro-ceeded in this way the traits wouldremain abstract, lacking just the contentand function which makes them livingtraits. In effect our subjects are inglaring disagreement with the elemen-taristic thesis which assumes independ-ent traits (or traits connected only ina statistical sense) of constant content.(2) At the same time the procedure ofour subjects departs from another cus-tomary formulation. It is equally farfrom the observed facts to describe theprocess as the forming of a homo-geneous, undifferentiated "general im-pression." The uriity perceived by theobserver contains groupings the partsof which are in more intimate connec-tion with each other than they are withparts of other groupings.13 Discrimina-tion of different aspects of the personand distinctions of a functional orderare essential parts of the process. Wemay even distinguish different degreesof unity in persons. Increasing clear-ness in understanding another dependson the increased articulation of thesedistinctions. But in the process these

13 If we may assume that the situation in theobserved person corresponds to this view, an im-portant conclusion follows for method, namely,that we can study characteristics of persons with-out an exhaustive knowledge of the entire person.

286 S. E. ASCH

continue to have the properties of partsin a single structure.

If we may for the purpose of discus-sion assume that the naive procedureis based on a sound conception of thestructure of personality, it would by nomeans follow that it is therefore freefrom misconceptions and distortions.But in that case the nature of errorsin judgment would have to be under-stood in a particular way. It wouldbe necessary to derive the errors fromcharacteristics of the organizationalprocesses in judgment. The presentinvestigation is not without some hintsfor this problem. It points to thedanger of forcing the subject to judgeartificially isolated traits—a procedurealmost universally followed in ratingstudies—and to the necessity of provid-ing optimal conditions for judging theplace and weight of a characteristicwithin the person (unless of course thejudgment of isolated traits is requiredby the particular problem). Under suchconditions we might discover an im-provement in the quality of judgmentand in agreement between judges. Atthe same time this investigation con-tains some suggestions for the study oferrors in factors such as oversimplifica-tion leading to "too good" an impres-sion, viewing a trait outside its contextor in an inappropriate context.

IllReturning to the main theoretical

conceptions described earlier (see pp.258-260) it is necessary to mention avariant of Proposition I, which we havefailed so far to consider and in rela-tion to which we will be able tostate more precisely a central feature ofProposition II. It would be a possiblehypothesis that in the course of form-ing an impression each trait interactswith one or more of the others, andthat the total impression is the summa-

tion of these effects. The impressionwould accordingly be derived from theseparate interaction of the components,which might be represented as follows:

\b. Impression =

etc.or

It is important to note that this for-mulation is in a fundamental regarddifferent from Proposition II. Thelatter proposition asserts that each traitis seen to stand in a particular rela-tion to the others as part of a completeview. The entire view possesses theformal properties of a structure, theform of which cannot be derived fromthe summation of the individual rela-tions.14 In the same manner that thecontent of each of a pair of traits canbe determined fully only by referenceto their mutual relation, so the contentof each relation can be determined fullyonly with reference to the structure ofrelations of which it is a part. Thiswe may illustrate with the example of

,a geometrical figure such as a pyramid,each part of which (e.g., the vertex)implicitly refers to the entire figure.We would propose that this is the basisfor the discovery of central and periph-eral traits and for assertions such asthat a given person is "integrated,"restricted, etc.

14 For a basic treatment of the concept of struc-ture the reader is referred to M. Wertheiraer (10).


On the the other hand, the notion ofstructure is denied in all propositionsof the form I, including Ib. In thelatter, an assumption is made concern-ing the interaction of qualities, whichhas the effect of altering the characterof the elements. Once we have takenaccount of this change, we have in thefinal formulation again a sum of (nowchanged) elements:

Ib. Impression = a + b-f -c + d + e

In still another regard there is a dif-ference between Propositions II and Ib.This has to do with the nature of theinteraction between the traits. In termsof Proposition II the character of inter-action is determined by the particularqualities that enter into the relation(e.g., "warm-witty" or "cold-witty"). Itis doubtful however whether a theorywhich refuses to admit relationalprocesses in the formation of a wholeimpression would admit the same rela-tional processes in the interaction ofone trait with another.

In view of the fact that PropositionIb has not, as far as we know, beenexplicitly formulated with reference tothe present problem, it becomes neces-sary to do so here, and especially tostate the process of interaction in sucha manner as to be consistent with it.This we might do best by applying cer-tain current conceptions. We couldspeak of traits as "conditioned verbalreactions," each of which possesses aparticular "strength" and range of gen-eralization. Interaction between traitswould accordingly be assimilated to theschema of differential conditioning tosingle stimuli and to stimuli in com-bination, perhaps after the manner ofthe recent treatment of "stimulus con-figurations" by Hull (4,5) .1B

16 Proceeding in the same manner, it would bepossible to restate some of our observations interms such as the following: (i) the distinction

How consistent would this inter-pretation be with the observations wehave reported? It seems to us thatthere are grave difficulties in the wayof such an interpretation. In so far asthe terms of conditioning are at all in-telligible with reference to our problem,the process of interaction can be under-stood only as a quantitative increase ordiminution in a response. This is not,however, the essential characteristic ofinteraction as we have observed it,which consists in a change of contentand function. The gaiety of an in-telligent man is not more or less thanthe gaiety of a stupid man; it is dif-ferent in quality. Further, the condi-tioning account seems to contain noprinciple that would make clear theparticular direction interaction takes.

Here we may mention a more gen-eral point. We have referred earlier tothe comparative ease with which com-plex situations in another person areperceived. If traits were perceived sepa-rately, we would expect to encounter thesame difficulties in forming a view ofa person that we meet in learning alist of unrelated words. That we areable to encompass the entire person inone sweep seems to be due to thestructured character of the impression.16

between central and peripheral traits would bereferred to a difference between conditioned reac-tions of greater and lesser strengths; (2) thechange from a central to a peripheral trait couldbe explained by the displacement of a response byother, stronger responses; (3) the factor of direc-tion might be dealt with in terms of changes inthe temporal appearance of stimuli; (4) stronglyunified impressions could be an expression ofhighly generalized reactions; etc. Such formula-tions would, however, fail to deal adequately withthe central feature of our findings, namely,changes in the quality of traits and the organizedform of the impression.

10 It should not, however, be concluded thatour views of persons are crystal clear. In fact,they lack the precision with which we grasp amathematical theorem. We rarely feel that wehave exhausted our understanding of another per-son. This has partly to do with the fact that theperson is in constant change.

288 S. E. ASCH

In terms of an interaction theory ofcomponent elements, the difficulty insurveying a person should be evengreater than in the formulation ofProposition I, since the former mustdeal with the elements of the latterplus a large number of added factors.

IV

In order to retain a necessary distinc-tion between the process of forming animpression and the actual organizationof traits in a person, we have spokenas if nothing were known of the latter.While we cannot deal with the latterproblem, one investigation is of par-ticular relevance to the present discus-sion. We refer to the famous investi-gation of Hartshorne and May (3), whostudied in a variety of situations thetendencies in groups of children to acthonestly in such widely varied mattersas copying, returning of money, cor-recting one's school work, etc. Therelations between the actions of childrenin the different situations were studiedby means of statistical correlations.These were generally low. On the basisof these results the important conclu-sion was drawn that qualities such ashonesty are not consistent character-istics of the child but specific habitsacquired in particular situations, that"neither deceit, nor its opposite, hon-esty, are unified character traits, butrather specific functions of life situa-tions." Having accepted this conclu-sion, equally fundamental consequenceswere drawn for character education ofchildren.

Abstracting from the many thingsthat might be said about this work,we point out only that its conclusionis not proven because of the failureto consider the structural characterof personality traits. As G. W. All-port (i, p. 250 ff.) has pointed out, wemay not assume that a particular act,

say the clandestine change by a pupilof an answer on a school test, hasthe same psychological meaning in allcases.17 Once this point is realized,its consequences for the thesis ofHartshorne and May become quitethreatening. Let us consider a few ofthe possibilities in the situation, whichwould be classified as follows by Hart-shorne and May:

Honest

1. The child wants to alter his answer on atest but fears he will be caught.

2. He does not change because he is indiffer-ent to the grade.

< Dishonest

1. The child changes his answer because heis devoted to his teacher and anxious notto lose her regard.

2. He cannot restrain the impulse to changethe wrong answer into the answer he nowknows to be correct.

Psychologically, none of these acts arecorrectly classified. Further, two ofthese are classified in precisely thewrong way. The child who wishes tocheat but is afraid does not belong inthe honest category, while the childwho cannot bear to leave the wronganswer uncorrected does not necessarilydeserve to be called dishonest. We donot intend to say that the psychologicalsignificance of the reactions was as arule misinterpreted; for the sake ofillustration we have chosen admittedlyextreme examples. But the failure toconsider the psychological content intro-duces a serious doubt concerning theconclusions reached by Hartshorne andMay.

A far richer field for the observationof the processes here considered wouldbe the impressions formed of actual

17 See also discussion by D. W. MacKinnon (7,p. 26 ff.).


people. Concrete experience with per-sons possesses a substantial quality andproduces a host of effects which haveno room for growth in the ephemeralimpressions of this investigation. Thefact that we are ourselves changed byliving people, that we observe them inmovement and growth, introduces fac-tors and forces of a new order. Incomparison with these, momentary im-pressions based on descriptions, or eventhe full view of the person at a givenmoment, are only partial aspects of abroader process.

In such investigation some of theproblems we have considered would re-appear and might gain a larger applica-tion. Other problems, which were ofnecessity excluded from the present in-vestigation, could be clarified in suchan approach. We mention one whichis of particular importance. It was aconstant feature of our procedure toprovide the subject with the traits ofa person; but in actual observation thediscovery of the traits in a person isa vital part of the process of establish-ing an impression. Since observationgives us only concrete acts and qualities,the application of a trait to a personbecomes itself a problem. Is character-ization by a trait for example a statisti-cal generalization from a number ofinstances? Or is it the consequence ofdiscovering a quality within the settingof the entire impression, which maytherefore be reached in a single in-stance? In the latter case, repeatedobservation would provide not simplyadditional instances for a statistical con-clusion, but rather a check on thegenuineness of the earlier observation,as well as a clarification of its limitingconditions. Proceeding in this manner,it should be possible to decide whetherthe discovery of a trait itself involvesprocesses of a strutural nature. Onlydirect investigation based on the ob-

289

servation of persons can furnish answersto these questions.

In still another regard did our in-vestigation limit the range of observa-tion. In the views formed of livingpersons past experience plays a greatrole. The impression itself has a his-tory and continuity as it extends overconsiderable periods of time, while fac-tors of motivation become importantin determining its stability and resist-ance to change.

Even within the limits of the presentstudy factors of past experience werehighly important. When the subjectformed a view on the basis of the givendescription, he as a rule referred to acontemporary, at no time to charactersthat may have lived in the past; helocated the person in this country, neverin other countries. Further, experimentswe have not here reported showed un-mistakably that an identical series oftraits produced distinct impressions de-pending on whether we identified theperson as a man or woman, as a childor adult. Distinctions of this orderclearly depend on a definite kindof knowledge obtained in the past.Indeed, the very possibility of graspingthe meaning of a trait presupposes thatit had been observed and understood.

That experience enters in these in-stances as a necessary factor seemsclear, but the statement would be mis-leading if we did not add that thepossibility of such experience itself pre-supposes a capacity to observe andrealize the qualities and dynamic rela-tions here described. The assertion thatthe properties of the impression dependon past experience can only mean thatthese were once directly perceived. Inthis connection we may refer to certainobservations of Kohler (6, p. 234 ff.)concerning our understanding of feel-ings in others which we have notobserved in ourselves, or in the absence

290 S. E. ASCH

of relevant previous experiences. In hiscomprehensive discussion of the ques-tion, G. W. Allport (i, p. 533 ff.) hasequally stressed the importance of directperception of a given structure inothers, of our capacity for perceivingin others dynamic tendencies.

Nor do we consider it adequate toassert that in the present investigationour subjects were merely reproducingpast observations of qualities and ofthe ways in which they modify eachother. When the subject selected a cer-tain trait as central (or when he de-posed a once central trait to a minorrole within a new context) it is by nomeans clear that he was guided by spe-cific, acquired rules prescribing whichtraits will be central in each of a greatnumber of constellations. It seemsmore in accordance with the evidenceto suppose that the system of the traitsitself points to a necessary center. Andas we have mentioned earlier, the inter-action between two traits already pre-supposes that we have discovered—whether in the past or in the present—the forces that work between them.Given the quality "quick" we cannotunequivocally infer the quality "skill-ful";18 but given "quick-skillful" wetry to see how one grows out of theother. We then discover a certain con-stancy in the relation between them,which is not that of a constant habitualconnection.

While an appeal to past experience18 That it is at times difficult to infer qualities

on the basis of central traits is due to such factorsas the lability of the person, the degree to whichthe actions of a person are directed by a singlecenter, as well as situational forces.

cannot supplant the direct grasping ofqualities and processes, the role of pastexperience is undoubtedly great whereimpressions of actual people extendingover a long period are concerned. Herethe important question for theory iswhether the factors of past experienceinvolve dynamic processes of the sameorder that we find at work in themomentary impression, or whetherthese are predominantly of the natureof associative bonds. It seems to usa useful hypothesis that when we relatea person's past to his present we areagain relying essentially on the compre-hension of dynamic processes.

REFERENCES

1. ALLPORT, G. W. Personality: a psychologicalinterpretation. New York: Holt, 1937.

2. ASCH, S. E. Studies in the principles ofjudgments and attitudes: II. Determinationof judgments by group and by ego stand-ards. /. soc. Psychol., 1940, 12, 433—465.

3. HARTSHORNE, H., & MAY, M. A. Vol. I,Studies in deceit, 1928; Vol. II, Studies inservice and self-control, 1939; Vol. Ill(with F. K. Shuttleworth), Studies in theorganization of character, 1930.

4. HULL, C. L. Principles of behavior. NewYork: Appleton-Century, 1943.

5. HULL, C. L. The discrimination of stimulusconfigurations and the hypothesis of affer-ent neural interaction. Psychol. Rev., 1945,52, 133-142-

6. KOHLER, W. Gestalt psychology. New York:Liveright, 1929.

7. MACKINNON, D. W. The structure of per-sonality. In Hunt, J. McV. (Ed.), Person-ality and the behavior disorders, Vol. I.New York: Ronald Press, 1944.

8. TERNUS, J. Experimentelle Untersuchungeniiber phanomenale Identitat. Psych,Forsch., 1926, 7, 81-136.

9. THORNDIKE, E. L. A constant error inpsychological rating. /. appl. Psychol.,1920, 4, 25-29.

10. WERTHEIMER, M. Productive thinking. NewYork: Harper, 1946.

forming impressions of personality - … impressions of personality * by s. e. asch ... jects in the...

Documents