EXPECTANCY CONFIRMATION PROCESSES ARISING IN THE SOCIAL INTERACTION SEQUENCE
by John M. Darley, Princeton University & Russell H. Fazio, Indiana University
NOTICE: THIS WORK MAY BE PROTECTED BY COPYRIGHT
YOU ARE REQUIRED TO READ THE COPYRIGHT NOTICE AT THIS LINK BEFORE YOU READ THE FOLLOWING WORK, THAT IS AVAILABLE SOLELY FOR PRIVATE STUDY, SCHOLARSHIP OR RESEARCH PURSUANT TO 17 U.S.C. SECTION 107 AND 108. IN THE EVENT THAT THE LIBRARY DETERMINES THAT UNLAWFUL COPYING OF THIS WORK HAS OCCURRED, THE LIBRARY HAS THE RIGHT TO BLOCK THE I.P. ADDRESS AT WHICH THE UNLAWFUL COPYING APPEARED TO HAVE OCCURRED. THANK YOU FOR RESPECTING THE RIGHTS OF COPYRIGHT OWNERS.
[Commission Counsel William B. Lytton, Esq.] Do you know a police officer by the name of Delores Thompson?
[Louise James, Former MOVE member] Yes, I do.
[Commission Counsel William B. Lytton, Esq.] Do you recall telling her that your son Frank was ordered to attack you, that you were beaten until you started to vomit violently ...
that Frank then placed a pillow over your face and asked John Africa if he wanted you to be cycled or killed, and that John Africa replied, “Not at this time”?
[Louise James, Former MOVE member] Did she tell you that I said that?
[Commission Counsel William B. Lytton, Esq.] I have her report of that conversation, Mrs. James.
[Louise James, Former MOVE member] Okay. Well, if she – were you satisfied with what she said?
[Commission Counsel William B. Lytton, Esq.] I’m asking you.
[Louise James, Former MOVE member] Okay.
I heard rumors to the effect that my son had beaten me. I’ve also heard rumors to the effect that Wilson Goode put his wife’s jaw on a pulley ...
that he beat the hell out of her.
Will you ask Mayor Good that when he comes in here if he in fact beat his wife?
And if not, why not? If it is relevant that my son beat me or whether or not he did beat me ...
then I would say it is just as relevant for you to ask Wilson Goode when he comes in here, “Did he beat his wife?”
I don’t think you’re gonna ask him that. I really don’t.
But you should, if you feel that it is relevant.
[Laverne Sims, Former MOVE member] Mr. Lytton, I just want to make it clear in my mind so that I understand.
Am I to assume that the bomb was dropped on MOVE people ...
because Frank beat his mother?
--Let the Fire Burn, directed by Jason Osder -- Illustrated Screenplay & Screencap Gallery
ABSTRACT: Psychologists have begun to use Merton's concept of the self-fulfilling prophecy to explain a wide variety of social phenomena and problems. The present article further develops the theory behind this and related concepts. The core of our argument is that self-fulfilling prophecy effects occur when any one of many possible forces distort the processes occurring in normal social interactions. To elucidate this argument, we describe a model of simple social interactions that involves (a) a perceiver's formation of an expectancy about a target person, (b) his or her behavior congruent with the expectancy, (c) the target's interpretation of this behavior, (d) the target's response, (e) the perceiver's interpretation of the response, (f) the target's interpretation of his or her own response. We discuss the biasing factors that may lead to self-fulfilling prophecy effects at each step of this sequence. In addition, we suggest several other forms of expectancy confirmation that may occur via the social interaction process. Our presentation has two major purposes: to increase the theoretical clarity of the self-fulfilling prophecy and related effects and to identify needs for future research on the topic.
The concept of the self-fulfilling prophecy continues to generate research (e.g., Crano & Mellon, 1978; Snyder & Swann,. 1978), theory (R. A. Jones, 1977), and controversy (Archibald, 1974; Wilkins, 1977). This activity is certainly a testimony to the importance of the construct, which is central to many of the practical implications of psychological knowledge for real-world settings (Rosenhan, 1973; Rosenthal & Jacobson, 1968). Partly because of this enormous practical importance, the majority of empirical investigations of the self-fulfilling prophecy have involved its occurrence in classroom settings. Rosenthal and Jacobson (1968) contended that the induction of a teacher's expectation of improvement in certain children's achievement performance caused such an improvement. Despite the controversy regarding this initial research (Elashoff & Snow, 1971; Jensen, 1969; Thorndike, 1968), subsequent experimental investigations (e.g., Meichenbaum, Bowers, & Ross, 1969; Zanna, Sheras, Cooper, & Shaw, 1975) have confirmed that teacher expectations affect student achievement.
Several recent studies, aimed at documenting the workings of the teacher-student expectancy confirmation in nonexperimental social interaction settings, have provided some estimates of the magnitude and generality of the effect. A series of correlational studies showed that naturally occurring teacher expectancies are related to students' achievement (e.g., Palardy, 1969; Rist, 1970; Seaver, 1973; Sutherland & Goldschmid, 1974). For example, Seaver (1973) suggested that teachers form expectations about a younger sibling on the basis of the older sibling's performance. He found a greater relationship between the achievement scores of two siblings taught by the same teacher than between the achievement scores of control siblings taught by different teachers. Such correlational findings are always open to question concerning causal direction. A recent cross-lagged panel analysis suggests, however, that it is the teacher expectations that affect student performance. Crano and Mellon (1978) found that teacher attitudes affected children's achievement to a greater degree than students' performance impinged on the teachers' attitudes. This is impressive testimony to the practical importance of expectancy effects.
The expectancy confirmation effect has also been investigated in contexts other than the classroom. Some of the earliest research demonstrated expectancy confirmation within the experimental situation. Under certain conditions experimenters may inadvertently bias the behavior of subjects. (See Rosenthal, 1976, for a summary of these research findings.) Furthermore, self fulfilling prophecies have bee shown to operate so as to confirm various social stereotypes. Expectancy effects have been implicated in the creation of apparent sex differences in behavior (Zanna & Pack, 1975), in the maintenance of racial stereotypes (Cooper & Fazio, 1979; Word, Zanna, & Cooper, 1974), in the perseverance of trait inferences regarding physically attractive women (Snyder, Tanke, & Berscheid, 1977), in the job performance of workers (King, 1971), and in the channeling of general social interactions (S. C. Jones & Panitch, 1971; Snyder & Swann, 1978).
Despite the practical importance of expectancy effects and despite the voluminous research literature documenting such effects, relatively little attention has been given to the process by which sell-fulfilling prophecies occur. We shall attempt to analyze this underlying social interaction process by presenting the steps in a general social interaction sequence and by examining the research documentation for each of these steps. 
The General Social Interaction Sequence
As recognized by researchers of widely differing orientations (e.g., signal detection theorists, Lindsay & Norman, 1977, and symbolic interaction theorists, Blumer, 1969), perception is a constructive, interpretative process. Such interpretation is particularly critical in the perception of other people. The actions of another person do not automatically convey meanings, but are given meanings by the perceiver. Therefore, a generally accepted account of a social interaction sequence between two people might be as follows: (1) Either because of past observations of the other or because of the categories into which he or she has encoded the other, a perceiver develops a set of expectancies about a target person. (2) The perceiver then acts toward the target person in a way that is in accord with his or her expectations of the target person. (3) Next, the target interprets the meaning of the perceiver's action. (4) Based on the interpretation, the target responds to the perceiver's action, and (5) the perceiver interprets the target's action. At this point, the perceiver again acts toward the target person and so can be regarded as reentering the interaction sequence loop at Step 2.
It is useful to regard the sequence as containing one more step: (6) After acting toward the perceiver, the target person interprets the meaning of his or her own action. Ordinarily, of course, the interpretation will be that the action was the appropriate one and was "caused" by the perceiver's action to which it was the response. However, other possibilities do exist. From his or her action, the individual may infer something new about himself or herself. As a result, the individual's self-concept may be modified.
This interaction sequence is arbitrary in at least two senses. First, the sequence can represent a cyclical process in that the perceiver's expectancy in Step 1 may have arisen from previous "passes" through the sequence, or it may actually mark the beginning of the interaction. Second, although we have arbitrarily designated one of the individuals involved in the interaction as the perceiver and the other as the target, it should be clear that the process can be symmetric in nature. Just as the perceiver has an impression of the target, the individual we have designated as target is likely to have expectancies concerning the perceiver. In another sense, of course, the identification of the perceiver in the self-fulfilling prophecy sequence is often nonarbitrary, and the interaction is nonsymmetric. The perceiver is the participant who generally has the power to impose his or her definition of the situation on, or affect the life course of, the other individual.
Having presented the general interaction sequence that organizes our analysis, we can now turn to the working definition of the term self-fulfilling prophecy and can examine our definition in the context of the interaction sequence. The concept of the self-fulfilling prophecy was first set forth by Robert Merton (1948, 1957):
Definitions of a situation (prophecies or predictions) become an integral part of the situation and thus affect subsequent developments .... 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. (Merton, 1957, p. 423)
Implicit in Merton's characterization is an image of an interaction in which one individual imposes an expectancy on a target individual in such a way as to make the second individual behave in a fashion which confirms that expectancy. Two related possibilities immediately suggest themselves: First, the behavior of the target person is not actually altered by the expectation of the perceiver and thus does not "objectively" confirm the expectation. But because of well-known perceptual distortion processes, it is interpreted by the perceiver as confirming his or her expectation. Second, the perceiver's expectation, communicated through his or her action, can actually alter the behavior of the target in such a way that observers judge the behavior to be in accordance with the definition implicit in the perceiver's expectancy. (Compare this with Rosenthal's, 1976, distinction between noninteractional and interactional experimenter expectancy effects.) The latter possibility seems most consistent with Merton's characterization of the concept, and it is the one we adopt in this article. The self-fulfilling prophecy, then, refers to the process by which the expectancies held by one individual about another alter the behaviors of that other in ways that observers would interpret as confirming the perceiver's expectancy.
Within the general interaction sequence outlined above, it is clear that Merton's definition and most research on the self-fulfilling prophecy involve the link between Step 1 and Steps 4 and 5. That is, the perceiver's expectancy evokes a new response from the target (Step 4), which fulfills the expectation. Step 5 represents the perceiver's "biased proof" that his or her expectancy regarding the target was correct. Clearly, several intermediary steps must be assumed to take place, but research on these linkages is relatively sparse. It is the purpose of this article to indicate what psychological theory might suggest about these intermediary linkages and to indicate what expectancy-related phenomena other than self-fulfilling prophecies might occur in the interaction sequence.
The Perceiver Forms An Expectancy
The first step in the interaction sequence is the perceiver's formation of an expectancy regarding the behavior of a target person. These expectancies can be about the specific behaviors predicted to be performed by the target person. However, attributional theories (Heider, 1944, 1958; E. E. Jones & Davis, 1965; Kelley, 1967) suggest that the expectancies are more likely to be about the intentions and dispositions of the target person, from which general classes of behavior are predicted. Whatever the exact nature of the expectancy, it can be inferred from any of several sources. For each kind of inference, various possible sources of bias are possible.
DIRECT OBSERVATION OF BEHAVIOR
First, the perceiver necessarily observes only a sample of behaviors of the target person. It is possible that this sample of behaviors was an unrepresentative one in that the target person was acting under some constraints of which the perceiver was unaware. An unlucky undergraduate may be suffering from an attack of flu and therefore do poorly on a test on which he or she would ordinarily do well. An observer who is unaware of the illness may form an expectation of future poor performance.
Second, whether or not the sequence of behavior observed is an unrepresentative one, the inferences that the observer draws from it may be unwarranted. The actor-observer effect postulated by E. E. Jones and Nisbett (1971) bas been confirmed by other studies (see E. E. Jones, 1976, for a review of these findings). Apparently, even if an observer is aware of the constraints limiting the freedom of action of an individual, he or she is likely to underestimate the force of these constraints and still make attributions to the target person (E. E. Jones, 1979; E. E. Jones & Harris, 1967; Ross, 1977; Snyder & Jones, 1974).
The above discussion has assumed a perceiver who is simply observing the behavior of a target individual. But often the perceiver may be involved in the sequence of actions from which he or she draws inferences about the target person. The perceiver may be involved because the outcomes produced by the target are personally relevant (cf. E. E. Jones & Davis's 1965, concept of hedonic relevance) or because he or she is a participant in the sequence. Both kinds of involvement have been shown to produce different evaluations of the target person than would otherwise be made (Chaikin & Cooper, 1973; E. E. Jones & DeCharms, 1957; Miller, Norman, & Wright, 1978).
Attribution theorists have discussed the conditions under which a specific behavior of a target will be judged correspondent to the underlying disposition of the actor and will lead to the expectation of disposition-related behaviors in the future. Implicit personality theorists (see Schneider, 1973, for a review) have pointed out the possibility of a related but different process that can create expectations for new classes of behavior to be exhibited by the target person. Following ordinary attribution processes, a perceiver may infer a trait or disposition of the target person. In the perceiver's naive theory of personality, this trait will be connected to other traits, and so the perceiver expects the target to exhibit behaviors dispositionally related to those inferred traits as well. Obviously these expectations of behavior stemming from inferred traits can be different depending on the implicit personality theory of the perceiver. Furthermore, depending on how appropriate the perceiver's implicit personality theory is for the target person, these inferences may lead to inaccurate expectancies regarding the future behavior of the target.
In the previous examples, direct observation of a sample of a single individual's behavior led to predictions about similar or inferentially related future behaviors on the part of that individual. But not all behavioral expectancies are derived from direct observation of an individual target. Frequently, initial expectancies about the behavior of an individual are drawn from evidence about the class of individuals to which that target individual is assumed to belong (cf. E. E. Jones & McGillis's 1976, related distinction between target and category-based expectancies). A perceiver's knowledge that a target individual belongs to some racial, ethnic, or gender category may trigger inferences about what actions he or she ought to display (see Hamilton, 1979, for a review). Thus stereotypes operate in a manner similar to the operation of implicit theories of personality. Both constructs involve the interconnection of various attributes. In fact, stereotypes may be regarded, as R. A. Jones (1977) has cogently argued, as regions of an individual's implicit personality theory. Like implicit personality theories, stereotypes can vary with regard to their appropriateness and accuracy. However, a great deal of the thrust of stereotyping research has been to demonstrate that these behavioral expectancies are overgeneralized and inaccurate predictors of actual behavior of the target individual. (For a review, see Brigham, 1971.)
One difference between implicit personality theories and stereotypes is in the cue that generates the inferential process. Generalization via an implicit personality theory occurs subsequent to the attribution of a particular trait to a target person. Stereotypes, on the other hand, are generated by the observation that an individual is a member of a particular group or category.
Finally, expectations about others can be derived from third parties. For instance, people's reputations may be known before their actual arrival and may set up expectations about their behavior. In experimental studies, expectations are frequently created in this fashion; for instance, in the well-known Rosenthal and Jacobson (1968) study, expectations about specific children who would be "late bloomers" were conveyed to the teachers by the psychologist.
The Perceiver Acts
Once the opportunity for a perceiver-target interaction arises, the perceiver's expectancy regarding the target (however that expectancy was formed) influences his or her behavior toward the target. By guiding the perceiver's actions, the expectancy channels the course of the perceiver-target interaction. This channeling can occur via different types of action on the part of the perceiver. It will be useful to distinguish between actions that terminate the interaction sequence and those that permit it to continue.
On the basis of his or her expectancy, the perceiver may actively avoid the initiation of any interaction with the target. Consistent with Newcomb's (1947) autistic hostility hypothesis, the perceiver may ignore any overtures on the part of the target and, if socially forced into the interaction, may seek to terminate it at the earliest possible moment. For example, a bigot is unlikely to seek out interactions with a black (DeFleur & Westie, 1958; Warner & DeFleur, 1969). Such a case is most apt to occur when the perceiver's expectancy concerning the target's behavior is so negative or ego threatening that he or she wishes to avoid contact with the target. The end result of such expectancy-guided avoidance behavior is the maintenance of the perceiver's impression of the target. By refusing to begin or by quickly terminating the interaction, the perceiver never allows the target an opportunity to correct any misperceptions. The perceiver's hostility has become "autistic" in the sense that it prevents him or her from being responsive to opportunities for interaction that might serve to provide additional information about the target person.
Of course, termination of the interaction by the perceiver may strongly affect the target's life if the perceiver holds some power over the target. On the basis of some quickly formed expectancy, an advisor may decide to dismiss a student or an employer may not hire a prospective employee.
Therefore, although the fact that a perceiver's categorizations may lead to a false diagnosis of the target is obvious enough, it is possible that its role in the production of self-fulfilling prophecies is underestimated. In the real world, high-power individuals often make treatment decisions on the basis of extraordinarily little evidence because they are under great time pressure. The psychiatrist, the teacher, and the police officer often need to make rapid assessments of persons. On the basis of those assessments, they may assign the individual to a specific treatment program. The interaction between the authority figure and the target has now terminated, but the authority's influence on the life of the target may be decisive.
Even if the perceiver's expectations do not lead to avoidance or premature termination of the interaction, they may still affect its course. In such cases the perceiver's expectation influences the manner in which he or she behaves toward the target, thus coloring the nature of the ensuing interaction. The link between perceiver expectancy and action has been widely demonstrated. In an observational study of a kindergarten class, Rist (1970) found that the teacher divided the children into rigid tracks very early in the year. This division appeared to he largely based on socioeconomic status and appearance qualities that the teacher considered to be indicative of academic promise. Not only did the teacher physically separate the different tracks, but she spent a disproportionate amount of time teaching the highest track and disciplining the lowest track. Thus teaching behavior was related to the teacher's expectations regarding the academic ability of the children.
Many experimental studies involving classroom or tutoring situations have demonstrated differential teacher behavior as a function of expectancy. Relative to controls, students labeled as bright have been found to receive a greater percentage of teacher interactions characterized by positive affect or a lesser percentage of interactions characterized by negative affect (e.g., Meichenbaum et al., 1969), more positive nonverbal behaviors  (e.g., smiling, direct eye gaze) from the teacher (e.g., Chaikin, Sigler, & Derlega, 1974), higher levels of positive reinforcement following correct responses, and higher levels of negative reinforcement following incorrect responses (e.g., Lanzetta & Hannah, 1969). In a review of many such classroom studies, Rosenthal (1974) bas cited evidence that teachers provide children labeled as bright with a warmer climate, more differentiated performance feedback, a greater amount of and more difficult material to learn, and mort opportunities to respond. (See also Cooper, 1979, for a review.)
The expectancy-action link has also been demonstrated by numerous studies involving interactions other than teacher-pupil ones. Subjects have been found to behave more competitively in a game situation when led to expect that their partner disliked them (S. C. Jones & Panitch, 1971) or that their partner was generally a hostile person (Snyder & Swann, 1978). Kelley (1950) found that participation in a class discussion varied as a function of whether the students expected a guest instructor to be a warm or cold person. In the context of an interracial job interview, Word et al. (1974) found that trained white and black confederates posing as job applicants were treated differently by white job interviewers. Relative to the white applicants, black applicants received shorter amounts of interview time, greater physical distancing, and higher rates of speech error (e.g., repetitions, stutters) from the white interviewers. These findings document the effect of stereotypic assessments of a target person on behaviors toward that person.
At this point it again becomes obvious that the interaction process is symmetrical in nature. The target, in effect, becomes a perceiver. Because the target's response is influenced by the interpretation given to the perceiver's action, it is important to consider how the action is apt to he perceived. First, the target is likely to have some perceptions of the perceiver that lead to expectations about the personality and actions of the perceiver. (These expectancies arise from the processes by which the perceiver categorized the target person, which we discussed earlier, and are subjected to the same potential biases.) In addition the target person has available the actions of the perceiver and will need to interpret their meaning.
The Target Interprets the Perceiver's Action
The perceiver has perceived and acted toward the target, and now the target may interpret the meaning of the perceiver's actions. Attribution theory gives a cognitive, decision-making account of a person's interpretations of the acts of another. Therefore an image arises of the perceiver (the target individual in this case) carrying out a conscious analysis that leads to a complete and explicitly formulated set of attributions about the other. Several elements of this image are frequently incorrect. First, we are frequently unaware of the attributional processes we employ. Like many other habitual processes, they are both very rapid and automatic, and they need not be very logical or rational in character. Second, they need not be complete; we frequently infer only one or two conclusions about another from his or her actions. At other times our interpretations are minimal, consisting largely of, for instance, an emotional discomfort provoked by the actions of the other. It is important to remember that the interpretations a person makes of another may be incomplete and fragmentary, and some of the future consequences that were discussed in the previous sections may not occur.
Still, as the symbolic interactionists remind us, actions always require interpretations before people can react to them, so some interpretation always occurs. And sometimes the interpretations are sufficiently detailed to generate future consequences. If the target has interpreted the perceiver's behavior as due to the dispositions of the perceiver, then the impression of the perceiver is apt to influence the target's behavior toward the perceiver in later encounters, possibly even in situations very different from that involved in the original interaction, For example, after a brief encounter with the perceiver in a seminar situation, the target may conclude that the perceiver is a warm person. Hence the target may now be more likely to greet the perceiver when encountered in other settings. One of the topics it will be important to address in future research is the conditions that lead to drawing complex attributions versus fragmentary and fleeting nonconscious or emotional reactions.
Four possible categories of attribution may be used by the target to account for the perceiver's actions.
DISPOSITIONAL CHARACTERISTICS OF THE PERCEIVER
Most obviously, an attribution may be made directly to the perceiver. For instance, Word et al. (1974) found that subjects rated an interviewer who behaved in a distant manner as less friendly than one who behaved in a more immediate manner. S. C. Jones and Panitch (1971) found that male targets eventually came to like male perceivers who had behaved cooperatively with them and to dislike perceivers who had bargained competitively with them.  This experiment is an interesting one in that it illustrates several steps in the general interaction sequence. A perceiver is originally led to believe that a target individual either likes or dislikes him or her. The perceiver then initiates cooperative or competitive behavior. As a result of the perceiver's behavior, the target in fact comes to like or to dislike the perceiver.
Once the target comes to this kind of generalization about the perceiver, that expectation can influence his or her behavior toward the perceiver in a variety of other future situations. Furthermore, the disposition attributed to the perceiver may be generalized so as to be considered characteristic of anyone who falls within the same category membership as the perceiver. For example, a child's first experience with a teacher may lead to generalizations about the ways in which all teachers behave.
ATTRIBUTIONS TO THE SITUATION
The target may make attributions to elements of the situation rather than to dispositions of the other actor. Such an attribution appears particularly likely to occur when the target is attempting to understand the nature of the situational forces. For instance, a subject in the Schachter (1959) fear and affiliation studies, on observing the fear reactions of another subject, may conclude not that the other subject is a fearful person but that the situation in which they both find themselves is generally a fear-provoking one. This kind of situation attribution will affect the target's present and future behavior in that setting, and if the specific setting is perceived as representative of a more general class of settings, then the target person develops expectations about all of those settings. A child whose first classroom is a formal and strictly disciplined one may conclude that this is the behavior required in all classrooms and may have trouble responding in less structured situations. As this example suggests, setting generalizations and role generalizations can be very similar. In one, all settings of a certain kind are seen as requiring certain behavior; in the other, all people of a certain category are expected to behave in a certain way.
THE TARGET'S SELF-ATTRIBUTIONS
The target can be aware that the perceiver's actions toward him or her may be based on some characteristic of himself or herself. The characteristic may be a personal one, or it may be a function of category membership. A person may be aware, for instance, of appearing formal in initial contact situations and may judge that the formal reaction received from the perceiver was elicited by his or her own formality. Or a black person may be aware that some whites may initially exhibit distant behavior because of his or her skin color. Of course, a good deal of research on actor-observer differences suggests that individuals too readily draw dispositional attributions about other individuals. This suggests that individuals are likely to minimize their own role as causal agents in generating the actions of other individuals. However, research (e.g., Arkin & Duval, 1975) has demonstrated that this process is by no means total or irreversible; people are sometimes aware that they cause other people's actions, or can be made to be so aware.
If the target regards something about himself or herself or his or her category membership as a cause of the perceiver's behavior, then this needs to be dealt with in future interactions with the perceiver. On one hand, the target may not accept the validity of that conclusion. Therefore, the target may attempt to counteract and change the perceiver's initial impressions. Or as Mead (1934) and Cooley (1900) have pointed out, the target may come to accept the self mirrored in the perceiver's actions.
More complex attributions about the mening of the perceiver's action are also possible. By and large they involve attributing the cause of the perceiver's action to interactions of the three determinants already discussed. A Person X Situation explanation could occur such that the target decides that the perceiver responds to situations of one type in a particular manner. For instance, the target may conclude that the perceiver reacts to unstructured situations with anxiety. Alternatively, the target may conclude that the perceiver's actions are a result of an interaction between the target and the situation. That is, the target decides that he or she, in the particular type of interaction setting, caused the perceiver to act a certain way. A black personnel manager may be aware that many individuals may react with anxiety when they meet him or her in an employment interview.
All of these interactional attributions for the perceiver's behavior create expectancies about the repetition of that behavior in at least some future settings. Because they are interactional rather than main effect expectancies, they apply to more limited classes of situations. Instead of expecting another individual to react to all situations with anxiety, for instance, we might expect him or her to react with anxiety to all situations in which he or she is confronted with a black authority figure. Insofar as the current scientific conceptualizations of human personality are more consistent with this personality-situation explanation of human action (Mischel, 1968, 1973), it might be argued that these interactionist generalizations are the more sophisticated ones for people to draw. But empirically, this remains to be seen. There is strikingly little research on the target's interpretation of the perceiver's action in an interaction setting.
The Target Responds
Up to this point, we have examined the perceiver's formation of an expectancy, subsequent behavior toward the target person, and the target's interpretation of that behavior. This interpretation is relevant to the manner in which the target will respond.
If the target believes that situational factors or dispositional attributes of the perceiver were responsible for the perceiver's action, the response may be quite different than if the target had decided that it was something about himself or herself that evoked the perceiver's action. The target's response is likely to be coordinated to that of the perceiver if situational forces or the perceiver's dispositions have been identified as causal. Both the original situational forces and the perceiver are still present. In these cases the target is likely to reciprocate the perceiver's action. Approach behavior from the perceiver will lead to reciprocal approach behavior from the target; avoidance behavior will produce avoidance response. If the perceiver has behaved competitively, as in the S.C. Jones and Panitch (1971) and Snyder and Swann (1978) studies involving a game situation, then the target responds competitively. If the perceiver behaves in a friendly, sociable manner, as did the male subjects who were led to believe that they were conversing with a physically attractive female did in a study by Snyder et al. (1977), then the target responds in a sociable manner. It is important to recognize that in each of the above studies, the target's reciprocation of the perceiver's behavior confirmed the perceiver's expectancy. In the first two studies, the target did act, as the perceiver expected, in either a hostile or a nonhostile manner. In the Snyder et al. study, the ostensibly physically attractive female did objectively behave (as scored by judges blind to experimental condition) in a more sociable manner than the ostensibly unattractive female -- just as the male perceiver's stereotype of "what is beautiful is good" led him to expect (cf. Dion, Berscheid, & Walster, 1972).
More subtle nonverbal behaviors are also apt to be reciprocated. For example, Rosenfeld (1967) found that subjects reciprocated with more smiles and positive head nods when treated in that way by the experimenter. Word et al. (1974) also found reciprocation of nonverbal behavior. They trained interviewers to behave in a manner that approximated the behavior exhibited toward the black or white applicants in their earlier experiment. That is, the interviewer sat closer to the applicant, made fewer speech errors, and spent more time with the applicant in an immediate condition than in a nonimmediate condition. Subjects reciprocated Interviewer behavior. Subjects facing the immediate interviewers took more proximate positions and made fewer speech errors than did nonimmediate subjects. Furthermore, raters blind to experimental condition judged the nonimmediate subjects as less calm and composed and less adequate for the job than the immediate subjects.
In most studies of interpersonal expectancy effect, the target's responses to the perceiver bear some reciprocal relationship to the perceiver's actions. On examination, this relationship proves to be of two sorts: first, categorically similar responses, for example, smiles to smiles or more precise speech as a response to a formal interview setting (both of the latter are thought to be indexes of the underlying category of social distance); second, responses that are different in kind from the initial actions that provoked them, for example, flight responses from aggressive actions or distancing responses from what are perceived as sexual advances. As the last example indicates, it is not simply the interpreted action of the perceiver that determines the target's response, but the perceiver's actions as they impact on the target's goals and motives. Approach behavior will lead to withdrawal, not approach, if the target does not want to be approached.
One valid criticism of the self-confirming prophecy literature is that this fact is frequently ignored, leading to an unnecessarily static account of dyadic interaction. In some areas of expectancy-confirmation research, the mediating mechanisms are beginning to be specified and researched. For instance, Rosnow (Rosenthal & Rosnow, 1975; Rosnow & Arkin, 1973) and Finkelstein (1976) have pointed out that experimenter expectancy effects require not only the experimenter's signal of expectancies but the subject's attention to and comprehension of those signals. Further, the subject may be motivated to go along with or do the opposite of the experimenter's demands and, finally, may or may not have the capability of doing what he or she is motivated to do. Similar complex mediating models will be needed in other areas of interpersonal interaction research.
Unfortunately, little research has been conducted to examine variables that might determine whether the target will accept or attempt to dispel the perceiver's impression (cf. Gurwitz & Topol, 1978, for one such attempt and Baumeister & Jones, 1978, for another). However, two factors would appear to be critical in determining the nature of that response: (1) the importance of the perceiver to the target and (2) the target's beliefs about the validity of the perceiver's impression. If the perceiver is someone whose opinion is simply not important to the target, then the target is unlikely to be concerned with attempting to dispel the impression. But if the perceiver's impression is important to the target, then the target may attempt to dispel the perceiver's impression if he or she does not believe it to be self-descriptive. Farina, Allen, and Saul (1968) empirically examined this hypothesis and found that subjects who believed that their partner considered them stigmatized in some manner (e.g., mentally ill) worked harder at a cooperative task. Similarly, Lambert, Libman, and Poser (1960) found that both Jewish and Christian subjects increased their pain tolerance after being told that their groups were typically inferior in regard to this attribute.
The target, on the other hand, may find the perceiver's impression to be relatively accurate. In such cases in which the perceiver's impression is congruent with the target's own self-image, the target is likely to behave in a manner that maintains the perceiver's assessment. For example, Baumeister, Cooper, and Skib (1979) found that a female subject who was led to believe that she possessed, and who was believed by a perceiver to possess, a positive (but bogus) trait performed very poorly on a subsequent anagram test if she was led to believe that such poor performance was consistent with this positive trait and if her performance was to be known by the perceiver. Thus given a favorable impression, the target sought to maintain that impression, even if it meant doing poorly on the task.
Further evidence consistent with this notion is provided by some research on the effects of social labeling. In an intriguing field experiment, Kraut (1973) also found evidence demonstrating effects of social labeling. Donors to a first door-to-door campaign were either labeled as charitable by the campaign worker or not labeled at all, whereas non- donors were either labeled as charitable or not labeled. The behavioral evidence on which the label was based apparently led subjects to accept the label as veridical. Subjects labeled charitable gave more, and subjects labeled uncharitable less, to a second charity than did their respective control groups. Thus, a label congruent with the target's own past behavior increased the extent to which later donation behavior was consistent with earlier behavior.
It is also interesting to speculate how the target might respond if uncertain about the fit of the perceiver's impression (cf. Festinger, 1954). Particularly if the perceiver is a significant other who has some expertise and familiarity with making judgments of the sort in question, the target may accept the perceiver's impression and come to believe it. For example, after repeated communications from a professor that he or she is not suited for graduate school, a student may come to believe that he or she in fact is not. If others have reacted to the student in the same manner, then the student is all the more apt to accept that self- definition (cf. Cooley, 1900; Kelley, 1967; Mead, 1934). In such cases, confirmatory behavior from the target may occur (Miller, Brickman, & Bolen, 1975).
The Perceiver Interprets the Target's Response
The target's response is likely to have implications for the perceiver's image of the target. One possibility is that without confirming or disconfirming the original expectancy, the target's action may lead the perceiver to arrive at a set of new expectancies regarding the target's behavior. On the other hand, the target's action may have direct implications specific to that original expectancy. In an objective sense, that is, as interpreted by people who hold no initial expectancy, the target's response may be congruent or incongruent with the perceiver's expectancy and, therefore, may confirm or disconfirm the perceiver's expectancy.
We have already considered some evidence regarding "objective" confirmation of the perceiver's expectancy. The Word et al. (1974) and Snyder et al. (1977) studies are instances in which naive raters judged target behavior to be characterized by the same qualities that the perceiver had expected. Many of the teacher expectancy studies (e.g., Meichenbaum et al., 1969) demonstrate that differential teacher behavior as a consequence of teacher expectancy produces expectancy confirmation on objective tests of achievement.
How might the perceiver interpret such objective confirmation? Rationally, the perceiver should recognize his or her own role in producing the target's behavior and should attribute the target's behavior to his or her own earlier action. However, a great deal of research suggests that this is not always the case. Observers appear to over-estimate the causal role of dispositional qualities of an actor (E. E. Jones, 1976; E. E. Jones & Nisbett, 1971; Ross, 1977). Even when explicitly made aware of some obvious situational pressures operating on an actor, observers draw dispositional inferences regarding the actor (e.g., E. E. Jones; 1979; E. E. Jones & Harris, 1967; Snyder & Jones, 1974). Basically, perceivers fail to correct adequately for the constraints that a social role or situation may impose on an actor (e.g., Ross, Amabile, & Steinmetz, 1977).
This fundamental attribution error, as Ross (1977) has termed observers' tendency to overestimate the role of dispositional factors and to underestimate the causal significance of situational forces, has implications for the manner in which our perceiver will view the target's response. The perceiver is likely to underestimate the role of his or her own earlier action (e.g., negative nonverbal behavior or friendly conversational demeanor) in determining the target's response. Instead, the perceiver is apt to conclude that the target's behavior was due to, and is an accurate reflection of, the target's disposition. The target is in fact incompetent for the job, friendly and sociable, aloof and hostile, or whatever in the eyes of the perceiver.