IOURNAL OF EXPERIMENTAL SOCIAL PSYCHOLOGY
14, 148-162 (1978)
Behavioral Confirmation in Social Interaction: From Social
Perception to Social Reality
MARK SNYDER AND WILLIAM
University of Minnesota
Received January 3, 1977
A perceiver’s actions, although based upon initially erroneous beliefs about a
target individual may channel social interaction in ways that cause the behavior of
the target to confirm the perceiver’s beliefs. To chart this process of behavioral
confirmation, we observed successive interactions between one target and two
perceivers. In the first interaction, targets who interacted with perceivers who
anticipated hostile partners displayed greater behavioral hostility than targets
whose perceivers expected nonhostile partners. Only when targets regarded their
actions as reflections of personal dispositions did these behavioral differences in
hostility persevere into their subsequent interactions with naive perceivers who had
no prior knowledge about them. Theoretical implications of the behavioral
confirmation construct for social perception processes are discussed.
If men define situations as real,
they are real in their consequences.
W. I. Thomas
It has been fashionable for many years to view deviance as a process of
social definition and social creation (e.g., Becker, 1963; Schur, 1971;
Tannenbaum, 1938). The essence of this “labeling” orientation is that
“ . . . social groups create deviance by making the rules whose infraction
constitute deviance. . . .” (Becker, 1963, p. 9). Of critical importance to
this process is the influence of labels on the dynamics of social interaction.
Having once been tagged with a label that implies deviance, one’s
behavioral options may be constrained in ways that actually force one to
become deviant. Consider the observations of Tannenbaum on the social
creation of crime and delinquency:
This research and the preparation of this manuscript were supported in part by National
Science Foundation Grant No. SOC 75-13872, “Cognition and Behavior: When Belief Creates
Reality,” to Mark Snyder. We thank Debra Polinsky, Gail Gaebe, and Stephanie Suther, who
assisted in the empirical phases of this investigation. We appreciate the constructive
commentary of Nancy Cantor, Bill DeJong, Albert H. Hastorf, E.
Jones, and Walter
Mischel on an earlier version of this manuscript.
Requests for reprints should be sent to Mark Snyder, Laboratory for Research in Social
Relations, Department of Psychology, University of Minnesota, 75 East River Road,
Minneapolis, Minnesota 55455.
Copyright 0 1978 by Academic Press, Inc.
AU rights of reproduction in any form reserved.
BEHAVIORAL CONFIRMATION 149
The process . . becomes a way of stimulating . . and evoking the very
traits that are complained of. . . The person becomes the thing he is described
as being. . The community expects him to live up to his [deviant] reputation,
and will not credit him if he does not live up to it (Tannenbaum, 1938, pp. 19-20;
Similar themes emerge from descriptions of the effects of stigmatization
(e.g., Goffman, 1963; Scott, 1969; Szasz, 1961). Most notably, Goffman
(1963) has argued that we may unintentionally “force” the stigmatized to
play the roles we prescribe for them. Stigmatized individuals may thus fall
“victim” to cultural stereotypes and expectations. Ex-mental patients or
former prisoners may be provoked by the prejudicial treatment of others
into angry reactions that may then be interpreted as manifestations of their
disturbance, and in turn justify treating them as socially dangerous
The case histories and descriptive prose of the labeling theorists sensitize
us to the possible impact of labels on subsequent social interaction.
However, within social psychology, this concern with the interpersonal
consequences of social perception has been largely neglected. Researchers
and theorists concerned with social perception and attribution processes
have focused their attention on the information-processing “machinery”
by which individuals attempt to understand an actor’s behavior (e-g.?
Jones, Kanouse, Kelley, Nisbett, Valins, & Wiener, 1972). Conspicuously
neglected within this cognitive social psychology is a consideration of the
effects of interpersonal perceptions (including social labels) on
interaction. It is this reality-constructing function of social perception with
which we are concerned. Accordingly, we ask: Can
perceptions of another individual channel social interaction in ways that
actually cause the target individual’s actions to provide behavioral
confirmation of the perceiver’s beliefs? The descriptions of the labeling
theorists tentatively suggest an affirmative answer. So too does the
research literature of empirical social psychology.
In certain specific contexts, a perceiver’s expectations may influence
another individual’s behavior (e.g., Jones & Panitch, 1971; Kelley
Stahelski, 1970: Kuhlman & Wimberley, 1976; Miller & Holmes, 1
Rosenthal, 1974). Thus, Rosenthal and his colleagues have documented
ways in which teachers and experimenters who expect particular patterns
performance from their students and subjects actually can and do elicit
performances that confirm these expectations (for a review, see Rosenthal,
1974). As impressive as these demonstrations are, certain constraints ofthe
situations studied may have produced a process quite different from the
hypothesized behavioral confirmation process, Interactions between
“trainers” (teachers and experimenters) and “performers” (students and
subjects) are highly structured and focused on selected criterion behaviors
(the dependent variable in the psychological experiment, academic
SNYDER AND SWANN
performance in the classroom). Expectancies are generally explicitly
communicated to the trainers and usually have high credibility. Teachers
and experimenters may feel that their own competence is being put to test.
the extent that their students and subjects behave as expected, their
competence as teachers and experimenters is confirmed in both their own
eyes and the eyes of the source of the expectancies. Indeed, in studies of
experimenter effects, experimenters often are told that the purpose oftheir
participation is to see how well they can replicate well-established
experimental studies (e.g., Rosenthal & Fode, 1963). Thus, teachers and
experimenters may be acting to validate or bolster their own self-
conceptions of competence and efficacy (cf. Secord & Backman’s, 1965,
analysis of response-evocation as one means of maintaining self-
However, research in other contexts does suggest that actions based on
interpersonal perceptions may produce behavioral confirmation. Correla-
tional analyses by Kelley and Stahelski (1970) indicate that individuals with
competitive orientations to social relationships believe the world to be
composed homogeneously of competitive individuals. By contrast, those
with cooperative orientations construe the world to be more heterogene-
ously composed of both competitive and cooperative people. One
consequence of these stereotypes is that competitive individuals are highly
likely to elicit competitive responses from their partners in the Prisoner’s
Dilemma situation, whether these partners have cooperative or competi-
tive dispositions. Thus, the feedback that competitors receive validates
their stereotype that all people are competitive, even though it was their
own behavior that determined their partner’s competitive behavior.
However, this effect may be specific to the structural characteristics of the
particular Prisoner’s Dilemma situation used by Kelley and Stahelski
(Miller & Holmes, 1975). To this extent, this and other laboratory
investigations of the interpersonal consequences of social perception that
have relied on the Prisoner’s Dilemma game (e.g., Jones & Panitch, 1971;
Kuhlman & Wimberley, 1976) may be of limited relevance to behavioral
Despite the interpretive ambiguities associated with some of the
procedures and results reviewed above, this evidence seems to indicate
that perceivers’ initial beliefs about another (target) individual may lead
them to channel their subsequent interaction with this individual in ways
that cause the target’s behavior to confirm these beliefs. But how stable and
enduring are the effects of this behavioral confirmation process? If
behavioral confirmation were limited to the confines of the specific
interaction between perceiver and target, it would be an inconsequential
effect of interpersonal perception.
If the “new” behaviors displayed by the target are not overly discrepant
from his or her own self-image, these new behaviors may be internalized.
BEHAVIORAL CONFIRMATION 151
The “latitude of acceptance” for internalization or accommodation of
self-perceptions to new behaviors may be rather wide (cf. Secord &
Backman, 1965). Most people believe that they possess a wide variety of
traits; witness their readiness to accept bogus personality descriptions in
demonstrations of the P. T. Barnum effect (e.g., Meehl, 1956; Stagner,
19.58; Ulrich, Stachnik, & Stainton, 1963). Moreover, most people believe
that they often express apparently contradictory traits in different
situations (Nisbett, Caputo, Legant & Marecek, 1973). Accordingly, a
wide variety of perceivers’ beliefs may be made to “stick” as a
consequence of behavioral confirmation processes.
If the target’s new behaviors generated by the behavioral confirmation
process are internalized, then both the target and the perceiver will share
perceptions of the target. What began in the mind of the perceiver will have
become reality not only in the behavior of the target but also in the mind of
the target. The target may then be prepared to act on this new
self-perception in the contexts beyond those that include the original
perceiver. Then the target may provide other perceivers with behavioral
evidence consistent with the original perceiver’s expectations and these
new perceivers will treat the target accordingly.
Contemporary attribution theories (e.g., Jones et al., 1972) suggest that
the internalization and perseveration of behavior are most likely to occur
when targets accept their new behaviors as representative of their
underlying personality traits (that is, when targets make dispositional
self-attributions). By contrast, when targets attribute their new behavior to
the specific influences of the perceiver and the context of their interaction
(that is, make situational self-attributions), behavioral confirmation will be
limited to interactions with that perceiver.
To investigate the process of behavioral confirmation and its persevera-
tion over situations, we conducted an experiment in which one male
participant (the “target”) interacted on successive occasions with two
other male participants (the “labeling perceiver” and the ““naive
perceiver”) in a context designed to permit assessment of behaviorai
manifestations of hostility. Prior to the first interaction, the labeling
perceiver was given access to information that led him to believe the
to be a hostile individual (Hostile Label condition) or a nonhostile
individual (Nonhostile Label condition). This first interaction was
as to induce the target to regard his actions either as
representative of corresponding personal dispositions (Dispositional
Attribution condition) or as a reflection of transitory influences ofthe other
participant’s behavior (Situational Attribution condition). In the second
interaction, the target interacted with a naive perceiver who had no prior
expectations about the target’s dispositions and no knowledge of his prior
history. We expected that:
(1) In the first interaction, between the labeling perceiver and the target,
SNYDER AND SWANN
the labeling perceiver’s beliefs would receive behavioral confirmation:
Targets believed to be hostile would actually come to display more
behavioral manifestations of hostility than would targets regarded as
(2) Behavioral confirmation would persevere into the second interac-
tion, between the naive perceiver and the target, only under Dispositional
Attribution conditions: Those targets who had originally been labeled as
hostile would continue to manifest more behavioral hostility than would
those targets who had been labeled as nonhostile.
We first present our empirical demonstration of behavioral confirmation
in social interaction, and then outline a theoretical account of the processes
that we believe to underlie behavioral confirmation.
One hundred and eight male undergraduates at the University of Minnesota participated in
this experiment for extra credit in their introductory psychology course. Groups of three
previously unacquainted participants were scheduled to report to separate waiting rooms.
Each participant was randomly assigned to one ofthree “roles”: labeling perceiver, target, or
The First Interaction: Labeling Perceiver and Target
The experimenter escorted the labeling perceiver and the target to adjacent experimental
rooms, leaving the naive perceiver in his waiting room. Each room was equipped with
headphones, a telegraph key, and a signaling system that fed into a control room. Before
learning about the experimental tasks, the labeling perceiver and the target each completed a
“Trait Survey” questionnaire that consisted of 15 self-descriptive items. Five of these items
set the stage for the soon-to-be-enacted label manipulation: sensitive-insensitive,
submissive-self-assertive, kind-unkind, passive-aggressive, cooperative-competitive.
The experimenter then explained separately to each participant that he would compete in a
reaction time task with an opponent in the next room. The competition was to consist of 24
trials grouped into eight blocks, each with three trials. On each trial, both players would
respond as quickly as possible to a signal. The player with the faster reaction time would be the
winner. Each player’s wins and losses would be determined not only by his speed, but also by
his clever and strategic use ofa “noise weapon.” The availability of this noise weapon was to
alternate, block by block, between the players. At the onset ofeach block, the user ofthe noise
weapon could adjust it to deliver one of six intensities of noise to his opponent for the duration
of that block. A “1” level noise was generally regarded as inoffensive, a “3” level noise was
typically perceived as distracting, and a “6” level noise was almost uniformly felt as
offensively irritating and annoying, but not physically painful. Each player then experienced
the relative intensity of each noise level. Use of the noise weapon constituted our dependent
measure of the extent of behavioral hostility displayed by each participant.
The attribution manipulation-Part one. The first part of the attribution manipulation
channeled the perspective that the targets would use in observing, encoding, and interpreting
their behavior during the ongoing interaction (cf. the successful use of observational sets to
influence attributional processes, e.g., Regan & Totten, 1975; Taylor & Fiske, 1975). To
encourage targets in the Dispositional Attribution conditions to regard their noise weapon
BEHAVIORAL CONFIRMATION 153
usage as a reflection oftheir own personal characteristics and individual reactions to the task,
the instructions stressed these points: (a) Previous research bad indicated that “the loudness
of the noise bursts most people choose to deliver to their opponent depends on the tYPe of
person they are and what they think is the best way of winning competitive reaction time tasks
like this one”; (b) typically, a person performs best “if he uses the strategy that uniquely fits
his capabilities and personality;” and (c) in planning his strategy, he should ask himself
several questions, including: “Am I the kind of person who likes to use a competitive or
cooperative strategy in situations like this?”
encourage targets in the
conditions to construe their use of the
noise weapon as a reaction to their opponent’s treatment of them, the instructions stressed
these points: (a) Previous research had indicated that “the loudness of the noise bursts most
people choose to deliver to their opponent depends on how their opponent uses his noise
weapon against them”; (b) typically, people perform best when they “devise their strategy by
taking into account the strategy that their opponent has used and might use in
(c) in planning his strategy he should ask himself several questions, including: “Is my
opponent the kind of person who will try to slow down my reaction time QT intimidate me by
delivering very loud noise bursts to me?“.
After hearing a description of the reaction time task, each
learned that, in order to help him plan his strategy, he would have access to his
opponent’s “Trait Survey” questionnaire. He received one of two qnestionnaires, prepared
in advance, to induce him to view his opponent either as a hostile individual or as a nonhostile
individual. Labeling perceivers who had been randomly assigned to the
conditions learned that they were pitted against an opponent who loved contact sports and
who described himself as relatively insensitive, self-assertive, cruel. aggressive, and
competitive. By contrast, labeling perceivers in the
Label conditions discovered
that their opponents enjoyed writing poetry and sailing and thought of themselves as rather
submissive, sensitive, passive, kind, and cooperative types.
Competition and use ofthe noise weapon.
The experimenter then signaled both participants
to put on their headphones for the first trial of the competition. He signaled the labeling
perceiver to set his noise weapon for the first block of three trials.
Each trial consisted of four discrete events: (I) A noise weapon tone of the designated
intensity was sounded in the headphones of either the target or the labeling perceiver; (2) the
“ready light” flashed, telling both players to depress the telegraph key: (3) the release light
flashed, signaling players to release the key as quickly as possible; and (4) the noise weapon
tone was turned off. After each block of three trials, the noise weapon changed bands.
After the 24 trials, the labeling perceiver was escorted ont of the experimental room
asked to record his final impressions of his opponent. The experimenter then invited the target
to play a new opponent in a second session. He did not provide the target with any explicit
feedback about his performance.
The attribution manipulation
-Part two. To complete the cognitive work begun by the first
part of the manipulation, the experimenter used verbal communications and specially-
constructed questionnaires to channel the target’s
rerrospective observation and interprela-
tion of his previous behavior in the first interaction.
Attribution conditions, the communication stressed that people usually
interpret their use of the noise weapon as a reflection of “the type of person they are and also
what they think is the best way to win in competitive tasks like this one.” Moreover, the
questionnaire contained questions written to encourage the targets to emphasize the intluence
of his own character and the nature of the task itself on his strategic use ofthe noise weapon;
e.g., the final question asked, “How much do you think the intensity of noise bursts you
delivered to your partner was a function of the way you as a person react to tasks of this
type?“. These items constituted an attempt to induce targets to formulate and accept the
desired dispositional attributions (cf. biased questioning as a technique for ~~flue~~~~~
attitudes [DiIlehay & Jernigan, 19701 and attributions [Salancik. 1976]).
SNYDER AND SWANN
Instructions in the Situational
conditions fostered quite different attributions.
These targets learned that individuals often pattern their use ofthe noise weapon to mirror and
reflect “how their opponent uses his noise weapon against them.” Again, a subtly-biased
questionnaire aided them to bolster and consolidate their situational self-attributions; e.g., the
final question asked, “How much do you think the number of noise bursts you delivered to
your opponent was a function of his competitiveness?”
From a methodological standpoint, it might have been desirable to include, at this point in
the procedure, a check on the effectiveness of the attribution manipulation. However, because
the second part ofthe attribution manipulation consisted of a self-rating
we did not want
to risk sensitizing targets to the fact that this procedure was a manipulation by introducing
more measures at that point. Moreover, as research on attributional processes makes
painfully clear, it is all too often impossible to elicit accurate self-reports of attributional
processes, even when participants behave precisely in accord with the dictates of these
hypothesized attributional processes.
The Second Interaction: Naive Perceiver and Target
The experimenter escorted the naive perceiver into the room that had been occupied
previously by the labeling perceiver and explained the reaction-time task to him. Naive
perceivers were given PZO information about either the target’s personal characteristics (that is,
no label) or his previous interaction with the labeling perceiver.
Once again, there were three trials in each of the eight blocks of the interaction. This time,
the target had first access to the noise weapon. It was, therefore, possible to assess the extent
to which “hostile” targets would actually initiate hostility to the naive perceivers on the first
block of trials. After the second session, all three participants were thoroughly educated about
the purposes of the experiment.
We examined the effects of the label and attribution manipulations on: (a)
behavioral confirmation in the first interaction; and (b) perseveration of
behavioral confirmation in the second interaction.
Did the labeling perceivers initiate a chain of events that ended in the
behavioral confirmation of their beliefs about their targets? We first
examined the manner in which labeling perceivers translated their
information about the personalities of their targets into actual behavioral
strategies for coping with that person as an opponent in the competitive
task. Their strategy was to use higher intensity levels of the noise weapon
when confronted with a reputedly hostile individual. When the noise
weapon was available to them, 61.1% of labeling perceivers who
anticipated hostile partners used intensities that averaged in the “4”, “S’,
and “6” range. By contrast, only 27.7% of labeling perceivers who
expected nonhostile partners adopted this upper-range strategy. This
difference is highly reliable, z = 2.02, p < .03.
Our faith that the use of the use of the upper noise levels was the
defining characteristic of the labeling perceivers’ strategies was bolstered
BEHAVIORAL CONFIRMATION 155
by the perceivers’ accounts of their actions offered during debriefing; for
example, “When the man says on the questionnaire that he’s a
cruel, you know he means to play tough. ‘Course I’m gonna give him the
loud ones” ; “My old roommate was a boxer, and he was always
aggressive. He was always getting into arguments. I figured this guy was
like him, so I gave him some loud ones”; “Aggressive people play hard. I
was just playing his game.”
As a consequence of the labeling perceivers’ treatment of them, the
targets soon began to behave in a hostile or nonhostile fashion. For when
the labeling perceivers initiated hostility, the targets reciprocated these
hostile overtures. To assess this
confirmation effect, we
computed for each
an index of behavioral hostility by averaging the
intensity of noises he delivered to his opponent throughout the first
interaction. An examination of the means displayed in the first row of Table
1 reveals that targets whose perceivers believed them to be hostile
individuals delivered higher levels of noise than did targets whose perceivers
viewed them as nonhostile persons. A least squares analysis of variance
yielded a reliable main effect of the label manipulation, F(1 ,X2) = 5.82, p
= .02. There was neither a reliable main effect of the attribution
manipulation nor any interaction between the two factors (both Fs < 1).
Therefore, changes in the targets’ behavior during the first interaction
cannot be explained in terms of prior changes in self-perception. It was not
BEHAVIORAL CONFORMATION IN SOCIAL INTERACTION
Hostile Non-hostile Hostile Non-hostile
label label label label
Behavioral confirmation of hostility:
the first interactiona
Labeling perceivers’ final impres-
sions of targets”
Perseveration of behavioral confirma-
tion: the second interaction
Naive perceivers’ final impressions
4.12 3.17 3.92 3.02
3.83 2.64 3.39 2.83
4.30 2.70 3.35 3.77
3.66 2.75 3.12 3.34
a Range = 1,6. Higher mean indicates greater behavioral display of hostility as assessed by
intensity of “noise weapon” usage by targets during the first interaction.
b Range = 1.5, 4.66. Higher means indicate greater perceived aggressiveness.
c Range = 1, 6. Higher means indicate greater behavioral display of hostility as assessed
by intensity of “noise weapon” usage by targets during the second interaction.
rt Range = 1.33, 4.66. Higher means indicate greater perceived aggressiveness.
SNYDER AND SWANN
that targets decided that they were hostile or nonhostile persons and then
acted upon these new self-perceptions. Rather, it seems that targets
behaviorally reciprocated the hostile or nonhostile overtures of the labeling
perceivers. Accordingly, the first interaction produced behavioral
confirmation of the labeling perceivers’ initially erroneous expectations
about the hostile or nonhostile natures of their targets. Nonetheless, as we
shall soon see, subsequent shifts in the targets’ self-perceptions, induced
by the attribution manipulation, did affect their behavior in their later
interaction with the naive perceivers.
Clearly, targets came to behave in accord with the labels with which they
had been “tagged” by the labeling perceivers. Moreover, the labeling
perceivers seemed willing to interpret the behavior that they themselves
had generated in their targets in terms of the dispositions of the targets.
After their interaction, labeling perceivers in the Hostile Label conditions
perceived their opponents to have much more aggressive natures (see
Table 1, second row) than did perceivers in the Nonhostile Label
conditions, F( 1,32) = 9.97, p = .004. Neither the main effect of attribution
conditions, F < 1, nor the interaction between label and attribution
condition, F = 1.25, were reliable. At the very least, these results confirm
the effectiveness of the label manipulation. More intriguing, yet, is the
possibility that the labeling perceivers may have committed a “fundamen-
tal error” of attribution (Ross, 1977); they may have attributed their
opponents’ hostile or nonhostile behavior to their opponents’ general
dispositions and not to the actual impact of their interaction with the
Perseveration of Behavioral Confirmation over Situations
Behavioral confirmation persevered into the second interaction between
the target and the naive perceiver only under Dispositional Attribution
conditions (see Table 1, third row). As predicted, there was a reliable
interaction between the label and attribution conditions, F(1,32) = 6.67, p
= .Ol. Planned comparisons revealed that under the Dispositional
Attribution conditions, targets who had once been labeled as hostile now
continued to behave in more hostile fashion than targets who had originally
been labeled as nonhostile, F(1,32) = 8.31, p < .Ol; in contrast, under
Situational Attribution conditions, the original labels no longer had any
impact on the behavior of the targets, F < 1. Moreover, these effects were
apparent even on the first block of trials, interaction F(1,32) = 4.85,~ = .03.
Thus, even before the targets (who had access to the noise weapon on the
first block) could observe the naive perceivers’ use of the noise weapon,
they initiated hostile or nonhostile treatment of them.
Having no reason to suspect otherwise, the naive perceivers regarded
their partners’ actions as representative of their true natures. As the pattern
of means presented in the fourth row of Table I indicates, there was a
substantial interaction between label and attribution, F(l,32) = 6.42, p
= .02. Under Dispositional Attribution conditions, naive perceivers viewed
targets who had once been labeled as hostile as much more aggressive
individuals than targets who had originally been tagged with the nonhostile
label, planned comparison F(1,32) = 6.70, p = -01. In contrast, naive
perceivers under Situational Attribution conditions did not form differe
impressions of targets as a function of their original labels, planned
comparison F < 1. These differences
course parallel the actual
behavioral differences of the targets in
Of what importance are our impressions and perceptions of others? Our
that social perceptions can and do exert
powerful channeling effects on subsequent social interaction such that
actual behavioral confirmation of these beliefs is produced. Even
initially erroneous impressions may become real. Social perceptions may
truly function as self-fulfilling prophecies:
The self-fulfilling prophecy is, in the beginning, a false definition of the situation
evoking a new behavior which makes the originally false conception come true. The
specious validity of the self-fulfilling prophecy perpetuates a reign of error. For the
prophet will cite the actual course of events as proof that he was right from the very
beginning. Such are the perversities of social logic (Merton, 1957, p. 423).
True to Merton’s (1957) script, our “prophets,” in the beginning, created
false definitions of their situations: they erroneously (as a result of the
experimental manipulation) believed their
be hostile or
nonhostile persons. But these attributional errors became self-erasing
errors. For the perceivers’ false definitions evoked new behaviors that
made their originally false conceptions come true: They treated their
targets as hostile or nonhostile persons and, indeed, these tar
responded in kind and began to behave in hostile or nonhostile fashion.
Unlike Merton’s scenario, ours did not end. Some of our targets who
regarded their actions as reflections of personal dispositions actually came
to believe that they were generally of hostile
nonhostile character. For
these targets, the process of behavioral confirmation extended and
persevered beyond the bounds of the original confirmation interaction.
They behaved in a hostile manner not only toward the perceivers who first
had regarded them as hostile, but also toward others in new and different
social contexts. They had become truly hostile persons, whose behavior
reflected the cross-situational consistency and temporal stability that are
hallmarks of personality traits and dispositions. The product-moment
correlation between hostile behavior in the two interaction situations for
targets in the Dispositional Attribution conditions (Y = .9O) was
SNYDER AND SWANN
substantially larger than that for targets in the Situational Attribution
conditions (r = .44), z = 2.74,~ = .003. We had apparently succeeded in
socializing, within the confines of our laboratory, a “trait” of hostility.
Behavioral Confirmation: A Theoretical Perspective
Our empirical investigation has demonstrated that a perceiver”s initially
erroneous beliefs about a target individual may initiate a chain of events
that channel subsequent social interaction in ways that cause the behavior
of the target to confirm the perceiver’s beliefs. It is now time to consider,
from a theoretical perspective, the processes that may underlie and
generate behavioral confirmation.
We view the unfolding over time of the events of the behavioral
confirmation process in terms of those critical cognitive activities of
perceiver and target by which each formulates strategies of action. The first
“link” in the “chain” of behavioral confirmation is that between the
labeling perceiver’s beliefs about his partner (e.g., “he is a hostile person”)
and the actions generated by those beliefs (e,g., “I will use my noise
weapon against him”). We view this link between thought and action as a
form of “reality-testing.” Social labels and attributions may serve as
grounds for predictions and generate behaviors designed to validate or
invalidate these beliefs (cf. Kelly, 19.55). The formation and testing of these
hypotheses may be guided by “scripts” (cf. Abelson, 1976) or “rules of
thumb.” These rules of thumb are scenarios involving sequences of events
and consequences and reflect implicit theories of the interplay between
persons and their situations. Thus, a perceiver in the Hostile Label
condition of our experiment might (literally or metaphorically) say to
himself: “If he is as mean and nasty as I think he is, then he will seize the
first attempt to attack me with his noise weapon. Perhaps I can forestall that
with a show of force. One good blast of a “6” level of my noise weapon and
he will know not to play tough with me. Anyway, better to get to him before
he gets to me.” He may bolster this line of thought with instances from his
own life experiences or those of acquaintances where such a strategy has
been successful. Moreover, he may remind himself of specific individuals,
similar in personality to the target, who would clearly deserve the
treatment he now plans for the target (cf. Abelson’s, 1976, discussion of the
use of scripts in decision-making and behavior-planning). Having
symbolically formulated his strategy, the perceiver proceeds to test his
But the hypothesis-testing process itself may generate behaviors that
erroneously confirm the prediction and validate the attribution-based
hypothesis. For the targets themselves, no doubt, formulate their strategies
of coping with their opponents using similar rules of thumb (e.g., “If this
guy, for no reason apparent to me, starts off with so much hostility, he
BEHAVIORAL CONFIRMATION 159
leaves me no option but to respond in kind with equally intense blasts of the
noise weapon. When attacked, defend youself; fight fire with fire.“) and
assimilate their behavior to that of the perceiver.
have created for themselves a situation not unlike
that of Kelly’s (1955) example:
A man construes his neighbor’s behavior as hostile. By that he means that his
neighbor, given the proper opportunity will do him harm. He tries out this
construction of his neighbor’s attitude by throwing rocks at his neighbor’s dog. His
neighbor responds with an angry rebuke. The man may then believe that he has
validated his construction of his neighbor as a hostile person (Kelly, 1955, pp.
Perhaps both Kelly’s hypothetical actor and our labeling perceivers may
commit the classic attribution error (cf. Ross, 1977): They may attribute the
targets’ behavior to corresponding inner dispositions rather than to the
constraints of the reality-testing procedure. They seem blissfully unaware,
as are Merton’s prophets, of the causal role that their own activities play in
generating the behavioral evidence that erroneously confirms their
expectations, inferences, and attributional labels. Unbeknownst to them,
the reality that they perceive to exist “out there” in the social world has in
fact been constructed by their own transactions with the social world.
Reality-testing has become reality-construction. It is not that the
perceivers are unaware of their beliefs or their actions based upon those
beliefs. It is that they seem to be unaware of their impact on the behavior of
others; that how others treat them is partially a reflection of how they first
treated those others.
According to the theoretical analysis of behavioral confirmation as
reality-testing, the perceiver’s activity is conceptualized as the cognitive
formulation and the behavioral testing of hypotheses. Behavioral
confirmation is seen as the unintended reality constructing consequence of
reality-testing: Perceivers as reality-testers unknowingly fail to take
adequate account of the biased nature of their hypothesis-generating an
hypotbesis-testing procedures. But how appropriate is it to regard
perceivers in this investigation as reality-testers? After all, they were not
explicitly instructed to test the accuracy of their beliefs about the targets’
natures. Perhaps, rather than testing reality, perceivers were simply taping
with the reality of their targets’ natures. However, from our theoretical
perspective, reality-testing is in practice no different than reality-coping.
When perceivers are in doubt about the accuracy of their beliefs about
targets, they may test the reality of these beliefs by treating targets ‘ ‘ as if’ ’
these beliefs were accurate. When perceivers have no uncertainty about
the reality of their beliefs, they may (quite reasonably) cope by treating
targets “‘as if” they were the persons they are reputed to be. In either
case- whether reality-testing or reality-coping-perceivers use their
beliefs about the target to formulate interaction strategies oftreating targets
SNYDER AND SWANN
“as if” their initial beliefs were accurate. In either case, behavioral
confirmation may be the outcome of such “as if” strategies.
Behavioral Confirmation and the Nature of Social Perception
Whatever the ultimate fate of our admittedly speculative analysis of
behavioral confirmation as reality-testing, the theoretical
the behavioral confirmation process itself cannot be ignored or minimized.
Researchers in social perception and the attribution process have focused
almost exclusively on the manner in which individuals process information
provided to them as they form impressions of other people. This
information processing is typically studied in static circumstances of
minimal personal involvement for the perceiver (cf. Taylor, Note 1; Taylor
& Fiske, Note 2). Such an approach may unfortunately blind us to the
intimate interplay between social perception and social interaction in
ongoing interpersonal relationships. Our investigation of behavioral
confirmation suggests that traditional information processing approaches
may seriously underestimate the extent to which the information that
perceivers process in actual social interaction is a product of the
perceiver’s own actions toward their targets, actions that may be based
upon and guided by their beliefs about those targets.
From our perspective, the perceiver’s knowledge of the target may be
seen as active, initiatory cognitive structures or conceptual schemas that
guide and influence the processing of information about the target, the
search for new information about the target, and the course of social
interaction between perceiver and target. The perceiver’s knowledge of the
target includes anticipations of what events are to appear as the interaction
unfolds. It may be easier to construct mental scenarios in which the target
acts in accord with the perceiver’s beliefs. Accordingly, it is these “as if”
scenarios (rather than “as if not” scenarios, in which the target violates the
perceiver’s expectations) that the perceiver may use to guide his or her
actions toward the target. As a consequence of this process, the target’s
behavior may be constrained in ways that generate confirming evidence for
the perceiver’s anticipations. Behavioral confirmation is then an end
product of the chain of events first initiated by the perceiver’s beliefs.
Our investigation suggests that a cognitive social psychology must pay
explicit attention to the ways by which perceivers
that they process in additon to probing the machinery of information
processing itself. Not only are our images of the social world a reflection of
events in the social world, but the very events of the social world
themselves may be reflections and products of our images of the social
Contemporary viewpoints in cognitive and perceptual psychology have
emphasized the active, integrative, and constructive aspects of human
information processing (e.g., Bower, 197.5; Klatzky, 1975; Neisser, 1976).
Our viewpoint, although clearly compatible with this constructivist
perspective on the formation of knowledge, goes at least one important ste
beyond this approach. Not only is knowledge (at least in the domain of
social cognition) the product of active, constructive processes, but the very
events that serve as the “raw materials” for this information-processing
are themselves the product of active, constructive processes generated by
the individual’s beliefs. It is in this sense that beliefs can and do create
Abelson. R. P. Script processing in attitude formation and decision making. In J. S. Carroll &
J. W. Payne (Eds.),
Cognifion and social behavior.
Potomac, Md.: Erlbaum, 1976.
Becker, H. W.
Outsiders: Studies in the sociology ofdeviance. New
York: The Free Press
Bower, G. H. Cognitive psychology: An introduction,
W. K. Estes (Ed.),
cognition and /earning.
Hillsdale, N. .I.: Erlbaum, 1975.
Dillehay, R. C., & Jernigan, L. R. The biased questionnaire as an instrument of opinion
Journal of Personality and Social Psychology,
1970, 19, 144-150.
E. Stigma: Notes on the management of spoiled identity.
Jones, E. E., Kanouse, D., Kelley, H. H., Nisbett, 12. E., Valins, S., cst Wiener, B. (Eds.),
Attribution: Perceiving the causes of behavior. New
York: General Learning Press,
Jones, S. C., & Panitch, D. The self-fulfilling prophecy and interpersonal attraction. Journal
of Experimental Social Psychology,
1971, I, 356-366.
Kelley, H. H., & Stahelski, A. J. The social interaction basis ofcooperators’ and competitors’
beliefs about others.
Journal of Personality and Social Psychology, 1970, 16, 66-91.
The psychology of personal constructs. New
York: Norton, 1955.
Klatzky, R. L.
Human memory: Structure and processes.
San Francisco: W. H. Freeman &
Kuhlman, D. M., & Wimberley, D. L. Expectations of choice behavior held by cooperators,
competitors, and individualists across four classes of experimental game.
Personality and Social Psychology,
1976, 34, 69-81,
Meehl, P. E. Wanted-A good cookbook.
1956, 11, 263-272.
Social theory and social structure.
Glencoe, Ill.: The Free Press, 1957.
Miller, D. T., & Holmes, J. G. The role of situational restrictiveness on self-fulfilhng
prophecies: A theoretical and empirical extension of Kelley and Stahelski’s tria@e
Journal of Personality and Social Psychology, 1975, 31,
U. Cognition and realify: Principles and implications of cognitive psychology. San
Francisco: W. H. Freeman, 1976.
Nisbett, R. E., Caputo, C., Legant, P., & Marecek, J. Behavior as seen by the actor and as
seen by the observer.
Journal of Personality and Social Psychology, 1973, 27,
Regan, D. T., & Totten, J. Empathy and attribution: Turning observers into actors. Journal oj
Personality and Social Psychology,
1975, 32, 850-856.
On the social psychology of the self-fulfilling prophecy: Further evidence for
Pygmalion effects and their mediating mechanisms. M. S. S.
SNYDER AND S WANN
Rosenthal, R., & Fode, K. L. Three experiments in experimenter bias. &‘sy&&gical
Reports, 1963, 12, 491-511.
Ross, L. The intuitive psychologist and his shortcomings: Distortions in the attribution
process. In L. Berkowitz (Ed.),Advances
in experimental socialpsychology
New York: Academic Press, 1977.
Salancik, G. R. Extrinsic attribution and the use of behavioral information to infer attitudes.
Journal of Personality and Social Psychology, 1976, 34,
Secord, P. F., & Backman, C. W. An interpersonal approach to personality. In B. Maher
Progress in experimental personality research
(Vol. 2). New York: Academic
Schur, E. M.
Labeling deviant behavior: Its sociological implications. New
York: Harper &
Scott, R. A.
The making of blind men. New
York: Russell Sage, 1969.
Stagner, R. The gullibility of personnel managers.
The myth of mental illness.
New York: Hoeber, 1961.
Crime and the community.
Boston: Ginn, 1938.
Taylor, S. E., & Fiske, S. T. Point ofview and perceptions ofcausality.
and Social Psychology,
1975, 32, 439-445.
Ulrich, R. E., Stachnik, T. J., & Stainton, N. R. Student acceptance of generalized
1963, 13, 831-834.
A socialpsychology of salience: Perspective in attribution processes.
presented at the New England Social Psychological Association, 1975.
2. Taylor, S. E., & Fiske,
Salience, attention, and attribution: Top
Unpublished manuscript, Harvard University, 1977.