Sei sulla pagina 1di 9

Journal of Personality and Social Psychology Copyright 1986 by the American Psychological Association, Inc.

1986. Vol. 51, No. 4,858-866 0022-3514/86/S00.75

Stability and Malleability of the Self-Concept

Hazel Markus Ziva Kunda


University of Michigan Princeton University

The self-concept literature is characterized by a continuing controversy over whether the self-concept
is stable or malleable. In this article we suggest that it is both but that the stability observed for general
descriptions of the self may mask significant local variation. In this study the social environment was
varied by creating a situation in which subjects found themselves to be either very unique or very
similar to others. Following this manipulation, subjects responded to a series of self-concept mea-
sures. Although the uniqueness and similarity subjects did not differ in the trait terms they used to
describe themselves, they did differ systematically in their latency for these judgments, in positivity
and negativity of their word associations, and in their judgments of similarity to reference groups.
These findings imply that subjects made to feel unique recruited conceptions of themselves as similar
to others, whereas subjects made to feel similar to others recruited conceptions of themselves as
unique. The results suggest that very general self-descriptive measures are inadequate for revealing
how the individual adjusts and calibrates the self-concept in response to challenges from the social
environment.

Two seemingly contradictory aspects of the self have been laboratory research has, in fact, stressed only one aspect—sta-
emphasized in the empirical self-concept literature. The self bility. The most pervasive and least ambiguous finding to
has been regarded as a stable and enduring structure that pro- emerge from the recent surge of research on the self-concept is
tects itself against change (e.g., Greenwald, 1980; Markus, that individuals seek out consistency and stability and actively
1977; Mortimer & Lorence, 1981; Swann & Read, 1981). Yet, resist any information that challenges their prevailing view of
it is also acknowledged that in different social environments themselves. Swann and his colleagues (Swann, 1985; Swann &
different selves appear to emerge. People vary from one time to Hill, 1982; Swann & Read, 1981) found, for example, that per-
another in their self-relevant thoughts, feelings, and behavior ceivers will go to great lengths to verify their self-conceptions
(e.g., Gergen, 1967b; Savin-Williams & Demo, 1983). One's by attending most closely to information that fits their view of
feelings about the self when talking to the boss are different the self and by trying to arrange their environments so as to
from those one has when talking to a subordinate, and one's acquire further self-confirming evidence. Other studies reveal
feelings about the self when being asked for a date are different that individuals appear to ignore or reject those accounts of
from those one has when being stood up for one. The defense of their behavior that differ from their own (e.g., Greenwald, 1980;
an unpopular view engenders a very different set of self-relevant Greenwald & Pratkanis, 1984; Markus, 1977; Wurf & Markus,
thoughts than those engendered by a reluctant decision to go 1983; lesser & Campbell, 1983; Rosenberg, 1979). Together,
along with the group. these findings suggest that the self-concept is anything but mal-
Most comprehensive theories of the self have recognized leable and mutable. On the contrary, they imply a stable sense
these two apparently contradictory aspects of the self. Thus of self that is quite unresponsive to variations in the social situa-
Rogers (1951) described the self-concept as organized and con- tion and one that individuals are determined to preserve.
sistent but also as fluid. Similarly, Turner (1956) characterized
the self as a "stable set of evaluative standards" but also as quite
variable—"the picture the person has of himself or herself at
The Malleable Self-Concept
any given moment" (p. 231).
The dynamic and malleable properties of the self-concept are
In this article, we seek to examine the means by which the
thought to derive primarily from its essentially social nature. In
self-concept may remain relatively stable yet still vary with the
referring to the variable nature of the self, James wrote that
social environment.
the individual has "as many different social selves as there are
The Stable Self-Concept distinct groups of persons about whose opinion he cares"
The dual nature of the self-concept, that is, its stability and (James, 1910, p. 294). Although psychologists have extensively
malleability, has rarely been the focus of empirical work. Most analyzed self-presentation (Jones & Pittman, 1982) and impres-
sion management (Schlenker, 1980) and variation in self-
esteem (Jones, Rhodewalt, Berglas, & Skelton, 1981; Morse &
The research reported here was supported by National Institute of Gergen, 1970), there are only a few (cf. Fazio, Effrein, & Fal-
Mental Health grant MH29753. We would like to thank Nancy Cantor, lender, 1981; McGuire & McGuire, 1982) who have empirically
Richard Nisbett, R. B. Zajonc, and an anonymous reviewer for valuable
broached the social malleability of the self-concept. Instead, re-
comments on an earlier version of this article.
search on the social nature of the self has been primarily con-
Correspondence concerning this article should be addressed to Hazel
cerned with the process of reflected appraisal and with the con-
Markus, Institute for Social Research, University of Michigan, Ann Ar-
bor, Michigan 48106. ditions under which there is a correspondence between how

858
STABILITY AND MALLEABILITY OF THE SELF-CONCEPT 859

people see themselves and how others see them (e.g., Rosenberg, time, there are variations in which self-conceptions are active
1981; Shrauger & Schoeneman, 1979). These studies have suc- in thought and memory. The self-concept appears as malleable,
cessfully revealed the social origins of the stable and enduring then, because the contents of the working self-concept change.
aspects of the self-concept but they have been much less in- Some self-conceptions, because of their importance in denn-
volved with characterizing the self-concept as it is constructed ing the self and their extensive elaboration, are probably con-
in different social situations. stantly available for characterizing the self; they are what Hig-
Sociologists have been more concerned with the shifting dy- gins has called chronically accessible (Higgins et al., 1982).
namic nature of the self (Martindale, 1980;Zurcher, 1977). Al- These conceptions reflect one's behavior in domains of endur-
exander and Knight (1971), for example, developed the concept ing salience, investment, or concern, and they have been vari-
of situated identities to refer to the meanings of the self that ously labeled as core self-conceptions or as self-schemas (e.g.,
emerge during a particular social encounter. These identities Markus, 1977). Other self-conceptions, however, vary in then-
are thought to be created afresh each time and not to be carried accessibility depending on the individual's affective or motiva-
over from one situation to another. Sociological studies high- tional states, or on prevailing social conditions (e.g., I'm behind
lighting the social malleability of the self, however, are rarely in all my work, I'm the only one who is dressed up, I'm the
concerned with how the self-conceptions offered by the social best dancer here, I could win a scholarship). The working self-
environment are combined with the existing self-conceptions. concept consists, then, of one's core self-conceptions embedded
Moreover, these studies are not easily reconciled with the much in a context of more tentative self-conceptions that are tied to
larger volume of research demonstrating the impressive stabil- the immediate social circumstances.
ity of the self-concept. Self-conceptions become active in working memory when
they are triggered by significant self-relevant events, or they can
be tuned in by the individual in responses to an event or situa-
The Working Self-Concept
tion. Thus putting one's foot in one's mouth with an ill-consid-
In this article, we examine the stability of the self-concept as ered comment to a colleague may be accompanied by a flood
it confronts challenging information. Drawing on current of negative self-conceptions—one's self as foolish, awkward, or
thinking about cognitive structures, we propose that, although impulsive. Such challenges to one's self-esteem, however, even
the self-concept is in some respects quite stable, this stability minor or self-inflicted ones, will not be received passively (e.g.,
can mask significant local variations that arise when the indi- Swann & Hill, 1982; Tesser & Campbell, 1983). Instead, the
vidual responds systematically to events in the social environ- individual is likely to counter these newly activated negative
ment. Recent research in a variety of areas of psychology indi- self-conceptions by recruiting from the universe of self-concep-
cates that individuals possess a great diversity of complex tions those that offer a different view of the self—the self as tact-
knowledge structures from which different substructures can ful, controlled, and socially skilled. The result of this cognitive
be activated and then creatively combined and used to meet the work is that the content of the working self-concept becomes
demands of different occasions. This has been demonstrated for decidedly different from what it was prior to the embarrassing
social judgment (Higgins & King, 1981; Higgins, King, & Ma- remark. In fact, the consequences of an embarrassing or a chal-
vin, 1982), attitudes (Salancik & Conway, 1975;Snyder, 1982), lenging self-relevant event may actually be a momentary rise in
and inferential rules (Nisbett, Krantz, Jepson, & Kunda, 1983), self-esteem or a brief period of self-promotion caused by the
to list but a few examples. In the perspective developed here, the positive self-conceptions that are recruited to counteract the
self is not a unitary structure, or even a generalized average of initially negative thoughts about the self (see also Greenberg &
images and cognitions. Instead, the self-concept encompasses Pyszczynski, 1985; Steele & Liu, 1983). Such variations in the
within its scope a wide variety of self-conceptions—the good content of the working self-concept are significant. They can
selves, the bad selves, the hoped-for selves, the feared selves, have powerful consequences for one's mood, for temporary self-
the not-me selves, the ideal selves, the possible selves, the ought esteem, and for the immediately consequent thoughts and ac-
selves (cf. Gergen, 1967a; Greenwald & Pratkanis, 1984; Hig- tions.
gins, 1983; Jones & Pittman, 1982; McGuire, 1984; Sullivan,
1953; Tesser & Campbell, 1983). The self-concept at a given
Variation in the Self-Concept
moment—the working self-concept—is a subset of this universe
of self-conceptions. Malleability in the self-concept—variation in the working
The content of a particular working self-concept depends on self-concept—occurs as the context of self-conceptions sur-
what subset of selves was active just before, on what has been rounding the core elements changes. This mutability or fluidity
invoked by the individual as a result of an experience, event, or in the self-concept will be fairly subtle; it will not, under most
situation, and, very importantly, on what has been elicited by circumstances, involve a major revision or reorganization of
the social situation at the given time. Thus the working self- significant self-relevant thoughts and feelings. Indeed, many
concept is a temporary structure consisting of elements from typical assessments of the self-concept, such as adjective check-
the collection of self-conceptions, organized in a configuration lists, will not reflect these variations in self-conceptions.
determined by ongoing social events. The self-concept, then, Changes of this nature require measures that reveal the differ-
can be viewed as stable in that the universe of self-conceptions ential availability of self-conceptions and measures that reveal
is relatively stable. Certainly, new self-conceptions will be changes in the meaning or interpretation given to various self-
added, yet once a particular self-conception has been created, descriptions.
it is unlikely to disappear even if seldom elicited. At the same In developing a paradigm that would allow us to explore vari-
860 HAZEL MARKUS AND ZIVA KUNDA

aliens in the content and configuration of self-knowledge, we pus, and how they resemble, in some respects, members of the
selected a domain of general significance and investment such opposite sex.
that self-knowledge about it would constitute a core component We hypothesized that both groups would engage in self-veri-
of the self-concept. We chose to study self-knowledge of one's fication as a result of the information communicated by the ma-
uniqueness or similarity to others. It is vital to know that one nipulation but that the self-concepts being verified would be
is similar to others and that one can claim membership in some very different from each other. Thus subjects led to believe they
group or society, but it is equally essential that one be suffi- are extremely unique should attempt to verify their similarity
ciently different from others so as not to forsake one's own iden- to others, whereas those led to believe they are very similar to
tity. In this study we constructed situations designed to bring others should make every effort to affirm their uniqueness.
both self-conceptions of uniqueness and self-conceptions of
similarity into play. Measuring the Working Self-Concept
We reasoned that when people received information about
themselves that threatened a certain core self-conception, they We did not expect the two groups of subjects to show impor-
would make every effort to reaffirm that aspect of their selves. tant revisions in their core self-conceptions of uniqueness or
Thus people led to think of themselves as extremely similar to similarity as a consequence of the experimental manipulation.
others would experience a threat to their uniqueness and strive That is, we did not expect their general self-descriptions on an
to reaffirm their self-conceptions of uniqueness, whereas people adjective checklist to vary. We did, however, expect their work-
led to feel extremely unique would attempt to bolster their self- ing self-concepts to differ. So we anticipated that their responses
on a number of indirect measures of the self-concept would
conceptions as similar to others (see Maslach, 1974; Snyder &
vary systematically with condition. Thus, in comparison to sub-
Fromkin, 1980; and Ziller, 1964, for extensive empirical and
jects made to feel similar to others (similarity subjects), subjects
theoretical work in this area). Fromkin (1970; Snyder & From-
made to feel extremely unique (uniqueness subjects) would en-
kin, 1980), in fact, has shown that both extreme similarity and
hance the positivity of the desired similarity and deflate the pos-
extreme uniqueness are aversive states and lead to behavioral
itivity of the threatening uniqueness. To reveal the positivity or
attempts to reestablish the opposite state. In this article, we hy-
negativity associated with the manipulated uniqueness or simi-
pothesized that these behavioral reactions are mediated by
larity, subjects were given an opportunity to provide word asso-
changes in the working self-concept.
ciations to these concepts. Moreover, we expected that the two
To examine the nature of the variation in the content and
groups would differ in the accessibility of self-conceptions as
configuration of the working self-concept, we led college stu-
unique and as similar. As an indication of the accessibility of
dents to believe that they were either very unique and different
self-conceptions of uniqueness or similarity we recorded the la-
from their peers on some important dimensions or very ordi-
tency of subjects' self-categorization responses and required
nary and similar to a group of their peers. This was accom-
subjects to perform a social comparison task in which they were
plished by providing them with false feedback indicating that
asked to judge their similarity to a number of basic reference
their preferences were either very different from those of a num-
groups. In comparison to similarity subjects, uniqueness sub-
ber of confederates in the study or very similar to those of the
jects were expected to provide evidence that conceptions of the
confederates. In one case, we attempted to surround the core
self as similar to others were readily accessible and that concep-
self-conceptions of uniqueness and similarity with evidence
tions of the self as unique were much less accessible.
that the subject was nearly a clone of the other students, and in
the other case, we sought to embed these same core self-concep-
tions within indications of the subject's unusualness.
Method
We anticipated that the working self-concept would vary Subjects
markedly as a function of this uniqueness/similarity manipula-
tion. We did not expect that subjects would simply accept the Subjects were 40 University of Michigan female undergraduates, who
participated in the study to earn credit in an introductory psychology
view of themselves as extremely common or extremely unique,
course.
nor did we expect that they could simply reject the information
communicated by the social situation. Rather, we anticipated
that subjects made to feel extremely unique, perhaps a little odd Procedure
and peculiar, would attempt to recruit from their universe of Subjects were run one at a time, each with three female confederates.
self-conceptions some reassuring examples of past behaviors to The confederates were undergraduate students who were paid for their
reveal their similarity to others. Such subjects might think participation. The subject and confederates were all seated in a row be-
about how their jeans and sweatshirts made them indistinguish- hind a long table, with the subject always at the far left. Subjects were
able from their fellow students, about how different they were told that the study was concerned with attitudes and opinions. For the
first task, subjects were shown posters in a series of 18 trials. Each poster
from the foreign students on campus, and how they, like every-
had three items, such as three colors, three cartoons, or three greeting
one else, loved the newest popular movies. In contrast, students
cards, and the subject was to write down for each poster the number of
made to feel uncomfortably similar to others should recruit
the item she liked best. Following this the manipulation was delivered.
from their collection of self-conceptions those that were evi- The experimenter explained that she had to transfer the responses to
dence of their uniqueness. These individuals might think about computer coding sheets and that it would save her time and money if
how they sometimes laugh at things no one else finds amusing, the subjects would read out their responses so that she could transfer
about how bored they are with some of the latest fads on cam- them immediately to the coding sheets. To make the task more interest-
STABILITY AND MALLEABILITY OF THE SELF-CONCEPT 861

ing for them too, she said, she would show them each poster as they
went along. On each trial the real subject went first, and her responses Me Responses
determined those of the confederates.
Uniqueness condition. On all but 3 of the 18 trials the confederates Uniqueness Subjects
all disagreed with the subject and agreed with each other. Thus, if on a Similarity Subjects
given trial the subject said she had picked Color C as her favorite, each
of the confederates said she had picked Color A. On the remaining three
trials, included to increase credibility, the first confederate agreed with
the subject and the other two disagreed with her and with each other.
Similarity condition. On all but 3 of the 18 trials the confederates
all agreed with the subject. Thus, if on a given trial the subject said she
had picked Color C as her favorite, each of the three confederates also
said she had picked Color C. On the three remaining trials, none of the
confederates agreed with the subject and two of them agreed with each
other.
Following this manipulation, subjects responded to a series of depen-
dent measures. Similarity Uniqueness
Task 1: Judgments of similarity to reference groups. Subjects were Words Words
shown 26 slides, each containing the name of a group. Of the groups
presented, the first 15 were in-groups—groups that all or almost all of WORD TYPE
the subjects belonged to, for example, Americans or Women. The next Figure 1. Number of similarity and uniqueness words endorsed by
11 were out-groups—groups that all or almost all of the subjects did not similarity and uniqueness subjects as self-descriptive.
belong to, for example, men or UCLA students. For each slide, subjects
indicated by pressing one of six labeled buttons how similar they felt to
the presented group on a 6-point scale that ranged from not at all similar
(1) to extremely similar (6). A computer activated the slide projector ceiving the similarity manipulation) did not differ in either the
and collected both subjects' responses and reaction times. The slides number of uniqueness-related words endorsed as self-descrip-
were projected for 4 s with 2 s between slides. tive, the number of similarity-related words endorsed as self-
Task 2: Self-categorization judgments. Subjects were shown 31 descriptive, or in the number of control words endorsed as self-
slides, each containing a trait adjective. Of these, 8 were related to descriptive (all Fs < 1). The fact that both groups of subjects
uniqueness (e.g., original, independent, unique) and 9 to similarity (e.g., endorsed almost half of the uniqueness words and almost half of
average, normal, follower). These were the critical-manipulation rele- the similarity words indicates that indeed both their uniqueness
vant words. The uniqueness and the similarity words were approxi- and their similarity were of importance to them. On the basis
mately matched for word frequency and likability. Fourteen other words of these data alone, one might conclude that these groups did
from the dimension extroversion-introversion were included for com-
not differ in their self-concepts with respect to uniqueness and
parison. The order of the words was determined randomly and fixed for
all subjects. Subjects responded to each slide by pressing either a me similarity. They seem not to have been influenced by the infor-
button if the adjective described them or a not me button if the adjective mation about their similarity or uniqueness that was presented
did not describe them. Each slide was presented for 4 s, with 3 s until in the experimental manipulation.
the onset of the next trait slide. Once again, latencies were collected.
Task 3: Word association. Subjects were given a booklet containing Malleability of the Self-Concept
six pages. On the tap of each page was a word, and they were instructed
to write down as many words as came to mind. Subjects were given I Latency for self-categorization judgments. Although sub-
min to work on each page and 5 s between pages. The first two words jects' overall self-categorizations did not reveal any effects of
were conceptually related to similarity (ordinary, conforming), the next the uniqueness/similarity manipulation (see Figure 1), the la-
two were neutral (honest, polite), and the last two were conceptually tencies for these self-categorizations were markedly different for
related to uniqueness (unusual, individualistic). the two groups. These results imply that the two groups of sub-
Manipulation check. Subjects were asked what percentage of the jects may indeed have been recruiting different self-conceptions
time they thought the other participants had agreed with their prefer-
into their working self-concepts as a result of the challenge pro-
ence judgment in the first part of the study. Following the experiment,
each subject was debriefed at length. vided by the experimental feedback. The differences in laten-
cies between the two groups reflect the fact that different types
Results of self-conceptions were mediating their self-categorization
judgments.
Manipulation Check We expected that the attempts of uniqueness subjects to re-
As expected, subjects were very much aware of the extent to affirm their similarity to others would take the form of a search
which participants had agreed with them. The uniqueness sub- for pertinent self-conceptions. Such subjects would, therefore,
jects estimated that the others agreed with them 8% of the time activate similarity-related terms and the related behavioral evi-
and the similarity subjects estimated 77% of the time. In actual- dence known to be self-descriptive as well as uniqueness-related
ity, there was 17% and 83% agreement. terms and related behaviors known to be uncharacteristic of the
self. This activation would result in shorter latencies for terms
Stability of the Self-Concept conceptually related to similarity. The opposite should be true
As can be seen in Figure 1, the uniqueness (those receiving of similarity subjects. Uniqueness subjects would, therefore, be
the uniqueness manipulation) and similarity subjects (those re- expected to have faster me responses to similarity words and
862 HAZEL MARKUS AND ZIVA KUNDA

Similarity Words Uniqueness Words


3000 3000
E2 Me 2800 Me
2800 • Not Me Not Me

2600 2600
)
u
\ 2400 » 2400

2200 2200

2000 2000

1800 1800
Uniqueness Similarity Uniqueness Similarity
Subjects Subjects Subjects Subjects
CONDITION CONDITION
Figure 2. Latencies of me? and not me responses of similarity and uniqueness
subjects to similarity and uniqueness words.

faster not me responses to uniqueness words, whereas similarity gorization belies a great deal of variation in the contents of the
subjects should be quicker to endorse uniqueness words and to working self-concepts of these individuals.
reject similarity words. Word association. Another clear indication that the stu-
As can be seen in Figure 2, with respect to the uniqueness dents did not ignore the self-relevant information conveyed in
words, the similarity subjects were faster to respond me than the manipulation but were, in fact, made uncomfortable by it
were the uniqueness subjects. With respect to the similarity ad- came from their subsequent performance on the word associa-
jectives, the uniqueness subjects were faster to respond me than tion task. The manipulation was associated with marked
were the similarity subjects. Figure 2 also shows the not me re- differences in the affective connotations of the uniqueness-sim-
sponses to the two sets of words. For the similarity adjectives, ilarity dimension. Subjects' responses to each word were coded
the uniqueness subjects were markedly slower to respond not for total number of word associations, for number of unique-
me than were the similarity subjects. The pattern is not as clear, ness-relevant and similarity-relevant words, and for the number
however, for the uniqueness words. Here the two groups did not of positive (e.g., nice, happy) and negative (e.g., bad, ugly) ex-
differ in their not me latencies. pressions; agreement between coders was 87%.
The overall analysis of these data yielded a significant 2 X 2 X We expected that our manipulation would affect the way sub-
2 interaction (Experimental Condition X Word X Response), jects interpreted the concepts of uniqueness and similarity. We
f(S6) = 2.12, p < .05,' which indicates that similarity words anticipated that the uniqueness subjects would find this ex-
and uniqueness words yielded different patterns of interaction. treme uniqueness to be threatening and unpleasant and that
Planned comparisons showed that the Condition X Response related concepts would thus be colored negatively. In contrast,
interaction for similarity words was significant, /(28) = 2.61, concepts relevant to similarity would be colored positively. The
p < .01, but the interaction for uniqueness words was not, t < opposite was expected to hold for the similarity subjects. And
I. The control words—words related to extroverted and intro- it was the negativity attached to uniqueness or similarity as a
verted—did not yield a significant Condition X Response inter- result of the manipulation that we assumed would instigate the
action, which indicates that the effects of experimental manipu- search and elicitation of self-conceptions to counter those sug-
lation were restricted to the manipulation-relevant domain of gested by the experiment. For the word association task, the two
uniqueness-similarity. groups did not differ in number of associations to either unique-
ness, similarity, or control words, or in number of dimension-
The fact that the pattern was less evident for the uniqueness
related traits generated to them. However, as can be seen in Fig-
terms reflects the finding that uniqueness subjects were not
quicker than similarity subjects to reject uniqueness words. It
1
appears that the patterns for the similarity and uniqueness All contrasts used error terms obtained from a 2 X 3 X 2 repeated
words were not, as might be expected, mirror images of one measures ANOVA (Experimental Condition [uniqueness, similarity] X
another because the uniqueness terms were extremely positive, Word type [uniqueness, similarity, control] X Response [me, not me]).
even though every attempt was made to equate the uniqueness Experimental condition was a between-subjects factor, and the remain-
ing two factors were within-subjects. The weights for similarity words
and similarity words for positivity and desirability. All subjects
were +1, — 1, — 1, +1, respectively for uniqueness condition, me re-
seemed to have difficulty rejecting characterizations of them- sponses; uniqueness condition, not me responses; similarity condition,
selves as, for example, special. This result may indeed suggest me responses; similarity condition, not me responses. The weights for
the type of self-conception that is particularly resistant to in- uniqueness words were — I,+ !,+!, — I, respectively for the four groups.
fluence. In general, however, the results imply that the stability All p levels reported for planned comparisons are based on one-tailed
of self-conception suggested by the results of the initial self-cate- tests.
STABILITY AND MALLEABILITY OF THE SELF-CONCEPT 863

Similarity Words Uniqueness Words


1.2
1.0 E2 Positive
M • Negative
10.8

•S0.6
k-

| 0.4
3
Z
0.2

0
Uniqueness Similarity Uniqueness Similarity
Subjects Subjects Subjects Subjects
CONDITION CONDITION
Figure 3. Number of positive and negative associations generated by similarity and uniqueness
subjects in response to similarity and uniqueness words.

ure 3, when responding to the similarity stimulus words, conventional and ordinary and, therefore, may be subject to the
uniqueness subjects generated more positive and fewer negative same reinterpretation as are words that directly convey similar-
associations than did similarity subjects. The opposite pattern ity to others.
was found for uniqueness stimulus words—in response to these, Similarity to reference groups. The fact that the two groups
uniqueness subjects generated fewer positive and more negative did indeed recruit different self-conceptions in response to the
associations than did similarity subjects. A contrast showed this experimental manipulation is also indicated by the results of
three-way interaction (2 X 2 X 2) to be significant, #66) = 2.95, their judgments of similarity to basic reference groups. We
p<.01. 2 anticipated that this task would be subtle and nonreactive, re-
As may be seen in Figure 3, the terms conceptually related to vealing the accessibility of particular self-conceptions. Many
uniqueness appear to be colored more positively than the terms self-definitions are derived from social comparisons. And the
conceptually related to similarity—subjects generated more outcome of a particular social comparison depends on what
positive and fewer negative words in response to them than to self-conceptions are active in the working self-concept. Thus,
the similarity words. A contrast showed this two-way interac- if the two groups differed in their judgments of similarity to
tion to be significant, f(66) = 5.55, p < .001. This enhanced common reference groups, we assumed that it was because they
positivity of uniqueness words, also noted in the latency data, had different self-conceptions accessible. If, as expected, sub-
may explain why the effect of the manipulation was less pro- jects were seeking evidence to counter the self-conception con-
nounced for those words than it was for similarity words— veyed by the experimental manipulation, uniqueness subjects
planned comparisons showed the Condition X Valence interac- would want to affirm their similarity to others and would, there-
tion for similarity words to be significant, /(33) = 2.53, p < .01, fore, rate themselves as more similar to in-groups and as rela-
whereas the interaction for uniqueness words was marginal, tively less similar to out-groups than would similarity subjects,
f(33)= 1.37,0 <.10. who would want to affirm their uniqueness.
Control words showed a pattern of results similar to that Not surprisingly, all subjects felt more similar to in-groups
found for similarity words, although less pronounced, f(33) = than to out-groups, F(\, 36) = 236, p < .OOl, 3 yet, as can be
1.37, p < .10. In retrospect this is not surprising because the seen in Table 1, similarity subjects described themselves as
control words—honest and polite—are associated with being more similar to out-groups and less similar to in-groups than
did the uniqueness subjects. These ratings were analyzed by a
2 x 2 (Experimental Condition X Group) repeated measures
analysis of variance (ANOVA), which showed the obtained inter-
Table 1 action to be marginally significant, F(l, 36) = 2.75,p =.11.
Mean Judgments of Similarity to In-Groups and If, as we expected, one method of reaffirming uniqueness or
Out-Groups by Experimental Condition similarity is to search through one's collection of self-concep-
Experimental condition
2
Once again, contrasts used error terms obtained from a 2 X 3 X 2
Group Uniqueness Similarity repeated measures ANOVA (Experimental Condition [uniqueness, simi-
larity] X Word Type [uniqueness, similarity, control] x Valence [posi-
In-group 5.13 4.93
Out-group 1.82 2.26 tive, negative]).
3
Unlike the earlier analysis, these analyses involved all available re-
Note. The higher the mean, the greater the degree of similarity to sponses rather than a subset of responses and, therefore, used F tests
groups. For both groups, N = 19. from the ANOVA rather than planned comparisons.
864 HAZEL MARKUS AND ZIVA KUNDA

Table 2 relatively more negative associations to uniqueness-related


Mean Latencies (in Milliseconds) of High Similarity words and relatively more positive associations to similarity-
Judgments of Similarity to In-Group and Out-Group related words. As a consequence of the challenge to their simi-
by Experimental Condition larity to others provided by the experiment, uniqueness subjects
appeared to recruit conceptions of themselves as similar to oth-
Experimental condition
ers. They were relatively quick to endorse and slow to reject
Group Uniqueness" Similarity6 similarity-related words as self-descriptive. Moreover, they
viewed themselves as similar to in-groups and made these judg-
In-group 2,056 2,127 ments quickly. Individuals made to feel extremely similar to
Out-group 2,794 2,456 others responded in exactly the opposite way.
The fact that both groups of subjects described themselves in
= 14. the same way on the self-categorization task, regardless of the
manipulation, is significant because it suggests that conceptions
of one's self as similar to others and as different from others
are among those core self-conceptions or self-schemas that are
tions for relevant examples of similarity or dissimilarity to com- critical for self-definition and that are not easily altered. Such
mon reference groups, one's membership in these groups self-conceptions may indeed be enduringly salient to the indi-
should become accessible in the working self-concept, and this vidual and thus accessible regardless of prevailing social cir-
should be reflected in the time needed to make judgments of cumstances. Other studies (e.g., Markus, 1977) have shown that
similarity to in-groups and out-groups. The search of unique- when individuals do not have self-schemas for particular do-
ness subjects for evidence of their similarity to others would mains of behavior they will change their self-descriptions in the
therefore be reflected in shorter latencies to judgments about direction of the self-relevant information provided in the labo-
similarity to in-groups. Similarity subjects, on the other hand, ratory. Thus, if uniqueness and similarity were domains of only
searching their memories for evidence of their uniqueness and limited investment and concern, we would expect those pro-
seeking examples of their similarity to relatively unusual vided with uniqueness information to describe themselves as
groups, should be quicker to make judgments about their simi- unique and those provided with similarity information to char-
larity to out-groups. Uniqueness subjects would therefore be acterize themselves as similar to others.
quicker than similarity subjects to respond on in-group judg- The importance attached to conceptions of similarity and
ments and slower than similarity subjects to respond on out- uniqueness may be one reason why the similarity and unique-
group judgments.4 ness created in the lab produced negative associations and why
As may be seen in Table 2, that pattern was found. These these subjects were apparently unwilling to simply ignore the
latencies were analyzed using a 2 X 2 repeated measures self-relevant information. As the self-concept literature would
ANOVA, which showed the obtained interaction to be margin- predict (Greenwald, 1980; Snyder & Swann, 1978; Swann,
ally significant, F(l, 24) = 2.73, p = .11. A highly significant 1983), the individuals in this study, when challenged, did indeed
group effect was also found, F(l, 24) = 18.47, p < .001, reveal- attempt to affirm their self-conceptions. Yet, what is notable for
ing that subjects were faster to respond to in-groups than to out- the question of malleability in the self-concept is that the self-
groups. conceptions being verified were quite different for the two
groups. Moreover, these self-conceptions bore a systematic rela-
Discussion tion to the information received about themselves from the so-
cial environment. They were elicited in response to the chal-
These results demonstrate both the stability and malleability lenging event. It seems that the uniqueness subjects had working
of the self-concept in response to a challenging event. The self- self-concepts comprised, at least in part, of conceptions of their
concepts of these individuals seemed to remain quite stable in similarity to others, whereas the similarity subjects had working
that the experimental feedback did not influence the terms they self-concepts that included conceptions of their uniqueness.
used to describe themselves. Individuals receiving the unique- The self-concept, although resistant to the view of the self pre-
ness manipulation and individuals receiving the similarity ma- sented by the experimental manipulation, did indeed react to
nipulations were alike in characterizing themselves as average the self-relevant information and can thus be considered mal-
and similar to others and also as special and unique. If one col- leable and fluid.
lected only these observations, one would tend to infer that the What are the consequences of the self-concept's sensitivity to
self-conceptions of these individuals were relatively unrespon- important self-relevant events? As suggested earlier, variations
sive to the self-relevant information provided by the study. in the contents of the working self-concept are likely to have
At the same time, the results of the word association, latency, powerful consequences for temporary self-esteem, for mood,
and similarity tasks suggest that underlying these similar gen- and for the immediately consequent thoughts and actions.
eral self-descriptions were very different temporary self-concep- There is a large volume of research indicating that individuals
tions. Thus, individuals led to feel unique were apparently dis-
turbed by this knowledge and, following the preference manipu- 4
Because we were concerned with judgments of similarity, we in-
lation, viewed the state of uniqueness as negative and cluded only latencies of responses indicating similarity (i.e., ratings of
undesirable, whereas the state of similarity to others became 4, 5, and 6). Only subjects who felt similar to at least one in-group and
correspondingly positive and desirable. These subjects provided at least one out-group were included.
STABILITY AND MALLEABILITY OF THE SELF-CONCEPT 865

are heavily influenced in their self-evaluations, judgments, sert their similarity to others and vice versa. Yet these data ques-
memory, and behavior by their currently accessible pool of tion the nature of this stability and how this stability is achieved.
thoughts, attitudes, and beliefs (Higgins & King, 1981; Nisbett The general conclusion here is that the considerable stability
& Ross, 1980; Sherman, Skov, Hervitz, & Stock, 1981; Snyder, accorded the self-concept may, in fact, mask significant mallea-
1982). If the totalitarian ego (Greenwald, 1980) was completely bility or fluidity that results as individuals respond to the view
powerful and self-serving biases thoroughly pervasive, most in- of the self communicated by the situation. A challenging event
dividuals would enjoy more freedom from worry, sadness, anxi- appears to initiate a process whereby the individual evaluates
ety, guilt, and embarrassment. Yet, because the working self- the information and then responds by attempting to integrate
concept is quite responsive to the social environment, these re- the self-conceptions offered by the environment with existing
actions are common. self-conceptions. In this sense, the working self-concept can be
The malleability observed in this study resulted from a viewed as quite situation dependent.
change in the accessibility of particular self-conceptions. Indi- The malleability of the self-concept revealed in this article
viduals tuned in self-conceptions in response to the challenging suggests a need for measures of the self-concept that have the
self-relevant information, and thus the working self-concept capacity to reveal the entire range of behavior involved in self-
was different from what it was before the uniqueness-common- categorization, self-definition, or in the creation of an identity
ness manipulation. We can, however, inquire about a different for one's self. Measures that assume the self to be a static struc-
question of malleability; that is, whether the manipulation pro- ture and require individuals to respond to very general descrip-
duced changes not just in the relative accessibility of self-con- tions about the self or to simply label one's self are often not
ceptions but also in the content of self-conceptions. Although adequate for revealing how the individual adjusts and calibrates
the current study did not address this second type of malleabil- the working self-concept in response to the social situation.
ity, it appears that individuals were bothered by the manipula- Measures of latency and confidence of self-description, indica-
tion, and it is likely that they modified their self-conceptions to tions of social comparison efforts, evaluations of others, mea-
some extent. At the minimum, it can be assumed that each time sures of anticipated consensus with others, or predictions of the
a particular conception of the self is activated, it will change, evaluation of others, for example, all have the potential to re-
however slightly, by the conditions and circumstances of the flect the momentary fluctuations in the availability and organi-
processing. Moreover, these subjects have some new self-con- zation of self-conceptions that accompany self-relevant changes
ceptions to add to their universe. Those in the unique condition, in the social environment. For the most part, measures of the
for example, now have a conception of themselves as quite pe- self-concept have been unnecessarily constrained and may have
culiar and uncommon in many of their preferences. Even if this concealed its diversity and flexibility.
self-conception was later undermined by the experimental de- Finally, these findings imply that if we assume the self-con-
briefing that revealed the rigged nature of the information, a cept to be a significant regulator of behavior, it should be ex-
conception of what could be possible was created for the subject plored while the individual is engaged in a particular social situ-
(see Markus & Nurius, in press). Such changes were, of course, ation or responding to a task or an objective, a position that has
subtle and unlikely to cause a major revision of the self-concept, been argued repeatedly by Gergen (1967b, 1977). Measures of
yet they may constitute the process by which the self-concept the self-concept in completely unframed, unstructured circum-
changes progressively as it incorporates new conceptions of the stances may be useful for some general diagnostic purposes, but
self. If situations repeatedly arise that require the individual to they are unlikely to elicit the type of working self-concepts that
activate self-conceptions that are at variance with certain core are elicited in particular social situations and that mediate on-
self-conceptions, we can imagine that these core conceptions, going behavior. Placing the individual within a particular con-
too, would eventually change. text reveals how the self-concept, although resisting challenge
The findings of this study are consistent with the developing and disconfirmation, varies with the prevailing social situation
view that the self-concept should not be viewed as a monolithic and how it depends on the social context for its particular mani-
entity or even as a generalized average sense of the self that is festation and expression.
carried around from one situation to another. Instead, the self-
concept is more productively viewed as a space (McGuire &
References
McGuire, 1982; Markus & Wurf, in press), a confederation
(Greenwald & Pratkanis, 1984), or a system (Martindale, 1980) Alexander, N. C, & Knight, G. W. (1971). Situated identities and social
of self-conceptions. From this set of self-conceptions, the indi- psychological experimentation. Sociometry, 34,65-82.
vidual constructs a working self-concept that integrates the core Fazio, R. H., Effrein, E. A., & Falender, Y J. (1981). Self-perceptions
self-conceptions with those elicited by the immediate context. following social interactions. Journal of Personality and Social Psy-
In this sense, the self-concept becomes similar to that suggested chology, 41, 232-242.
by the symbolic interactionists. Thus, for Mead (1934) there Fromkin, H. L. (1970). Effects of experimentally aroused feelings of
undistinctiveness upon valuation of scarce and novel experiences.
was no fixed self-concept, only the current self-concept that was
Journal of Personality and Social Psychology, 16, 521-529.
negotiated from the available set of self-conceptions.
Gergen, K. J. (1967a). Multiple identity: The healthy, happy human
The findings of the responsivity of the self-concept presented being wears many masks. Psychology Today, 5, 31-35,64-66.
in this article do not contradict the possibility that the self re- Gergen, K. J. (1967b). To be or not to be the single self. In S. M. Journal
sists modification or change, or that it selectively distorts reality (Ed.), To be or not to be: Existential perspectives on the self (pp. 15-
to maintain a positive view. Indeed, those individuals who 26). Gainesville: University of Florida Press.
found themselves uncomfortably unique were struggling to as- Gergen, K. J. (1977). The social construction of self-knowledge. In T.
866 HAZEL MARKUS AND ZIVA KUNDA

Mischel (Ed.), The self: Psychological and philosophical issues (pp. Rogers, C. R. (1951). Client centered therapy. New York: Houghton-
139-169). Oxford, England: Blackwell. Mifflin.
Greenberg, J., & Pyszczynski, T. (1985). Compensatory self-inflation: Rosenberg, M. (1979). Conceiving the self. New York: Basic Books.
A response to the threat to self-regard of public failure. Journal of Rosenberg, M. (1981). The self-concept: Social product and social
Personality and Social Psychology, 49, 273-80. force. In M. Rosenberg & R. H. Turner (Eds.), Social psychology in
Creenwald, A. G. (1980). The totalitarian ego: Fabrication and revision sociological perspectives (pp. 593-624). New 'York: Basic Books.
of personal history. American Psychologist, 35, 603-618. Salancik, G. R., & Conway, M. (1975). Attitude inferences from salient
Greenwald, A. G., & Pratkanis, A. R. (1984). The self. In R. S. Wyer& and relevant cognitive content about behavior. Journal of Personality
T. K. Srull (Eds.), Handbook of social cognition (Vol. 3, pp. 129-178). and Social Psychology, 32, 829-840.
Hillsdale, NJ: Erlbaum. Savin-Williams, R. C., & Demo, P. (1983). Situational and transitional
Higgins, E. T. (1983). A theory of discrepant self-concepts. Unpublished determinants of adolescent self-feelings. Journal of Personality and
manuscript, New York University. Social Psychology, 44, 820-833.
Higgins, E. T., & King, G. (1981). Accessibility of social constructs: Schlenkei; B. R. (1980). Impression management. Monterey, CA:
Information-processing consequences of individual and contextual Brooks/Cole.
variability. In N. Cantor & J. F. Kihlstrom (Eds.), Personality, cogni- Schrauger, J. S., & Schoeneman, T. J. (1979). Symbolic interactionist
tion, and social interaction (pp. 69-121). Hillsdale, NJ: Erlbaum. view of self-concept: Through the looking glass darkly. Psychological
Higgins, E. T., King, G. A., & Mavin, G. H. (1982). Individual construct Bulletin, 86, 549-573.
accessibility and subjective impressions and recall. Journal oj'Person- Sherman, S. J., Skov, R. B., Hervitz, E. F., & Stock, C. B. (1981). The
ality and Social Psychology, 43, 35-47. effects of explaining hypothetical future events: From possibility to
James, W. (1910). Psychology: The briefer course. New York: Holt. actuality and beyond. Journal of Experimental Social Psychology, 17,
Jones, E. E., & Pittnian, T. S. (1982). Towards a general theory of strate- 142-158.
gic self-preservation. In J. Suls (Ed.), Psychological perspectives on Snyder, M. (1982). When believing means doing: Creating links be-
theself(V<A. 1, pp. 231-262). Hillsdale, NJ: Erlbaum. tween attitudes and behavior. In M. P. Zanna, E. T. Higgins, & C. P.
Jones, E. E., Rhodewalt, F., Berglas, S., & Skelton, J. A. (1981). Effects Herman (Eds.), The Ontario Symposium, Vol. 2: Consistency in so-
of strategic self-presentation on subsequent self-esteem. Journal of cial behavior (pp. 105-130). Hillsdale, NJ: Erlbaum.
Personality and Social Psychology, 41,407-421. Snyder, C. R., & Fromkin, H. L. (1980). Uniqueness: The human pur-
Markus, H. (1977). Self-schemata and processing information about suit of difference. New York: Plenum.
the self. Journal of 'Personality and Social Psychology, 35,63-78. Snyder, M., & Swann, W. B. (1978). Hypothesis testing in social interac-
Markus, H., & Nurius, P. (in press). Possible selves. American Psycholo- tion. Journal of Personality and Social Psychology, 36, 1202-1212.
gist. Steele, C. M., & Liu, T. J. (1983). Dissonance processes as self-affirma-
Markus, H., & Wurf, E. (in press). The dynamic self-concept: A social tion. Journal of Personality and Social Psychology, 45, 5-19.
psychological perspective. In M. R. Rosenszweig & L. W. Porter Sullivan, H.S. (1953). The interpersonal theory of psychiatry. New York:
(Eds.), Annual review of psychology (Vol. 38). Norton.
Martindale, C. (1980). Subselves: The internal representation of situa- Swann, W. B., Jr. (1983). Self-verification: Bringing social reality into
tional and personal dispositions. In L. Wheeler (Ed.), Review oj'per- harmony with the self. In J. Suls & A. G. Greenwald (Eds.), Social
sonality and social psychology (Vol. 1, pp. 193-218). Beverly Hills, psychological perspectives on the self(Vo\. 2, pp. 33-66). Hillsdale,
CA: Sage. NJ: Erlbaum.
Maslach, C. (1974). Social and personal bases of individuation. Journal Swann, W. B., Jr. (1985). The self as architect of social reality. In B.
of Personality and Social Psychology, 29,411-425. Schlenker (Ed.), The self and social life (pp. 100-125). New York:
McGuire, W. J. (1984). Search for the self: Going beyond self-esteem McGraw-Hill.
and the reactive self. In R. A. Zurcher, J. Amoff, & A. I. Rabin (Eds.), Swann, W. B., Jr., & Hill, C. A. (1982). When our identities are mis-
Personality and the prediction of behavior (pp. 73-120). New York: taken: Reaffirming self-conceptions through social interaction. Jour-
Academic Press. nal of Personality and Social Psychology, 43( 1), 59-66.
McGuire, W. J., & McGuire, C. V. (1982). Significant others in self Swann, W. B., Jr., & Read, S. J. (1981). Self-verification processes: How
space: Sex differences and developmental trends in social self. In J. we sustain our self-conceptions. Journal of Experimental Social Psy-
Suls (Ed.), Psychological perspectives of the self. (Vol. 1, pp. 71-96). chology, ; 7, 351-372.
Hillsdale, NJ: Erlbaum. Tessei; A., & Campbell, J. (1983). Self-definition and self-evaluation
Mead, G. H. (1934). Mind, self, and society. Chicago, IL: University of maintenance. In J. Suls & A. Greenwald (Eds.), Social psychological
Chicago Press. perspectives on the self (Vol. 1, pp. 1-31). Hillsdale, NJ: Erlbaum.
Morse, S., & Gergen, K. J. (1970). Social comparison, self-consistency, Turner, R. H. (1956). The real self: From institution to impulse. Ameri-
and the concept of the self. Journal of Personality and Social Psychol- can Journal of Sociology, 61, 316-328.
ogy, 16, 148-156. Wurf, E., & Markus, H. (1983). Cognitive consequences of the negative
Mortimer, J., & Lorence, J. (1981). Self-concept, stability and change self. Paper presented at the annual meeting of the American Psycho-
from late adolescence to early adulthood. Research in Community logical Association, Anaheim, CA.
and Mental Health. 2, 5-42. Ziller, R. C. (1964). Individuation and socialization. Human Relations,
Nisbett, R. E., Krantz, D. H., Jepson, C., & Kunda, Z. (1983). The use 17,341-360.
of statistical heuristics in everyday inductive reasoning. Psychological Zurcher, L. A. (1977). The mutable self: A self-concept for social change.
Review, 90. 339-363. Beverly Hills, CA: Sage.
Nisbett, R. E., & Ross, L. (1980). Human inference: Strategies and
shortcomings of social judgement. Englewood Cliffs, NJ: Prentice- Received July 12,1984
Hall. Revision received December 17, 1984 •

Potrebbero piacerti anche