Journal ol Abnormal and Social Psychology 1961, Vol. 63, No. 2, 302-310 ROLE PLAYING VARIATIONS AND THEIR INFORMATIONAL VALUE FOR PERSON PERCEPTION 1 EDWARD E. JONES, KEITH E. DAVIS, AND KENNETH J. GERGEN Duke University E IGELY under the impetus of Heider's (1944, 1958) persistent concern with phenomenological analysis, much of the recent research in social perception has addressed itself to the naive psychology of the individual perceiver. How do individuals use the behavior of others to infer the probable existence of more enduring personal charac- teristics? What are the bases for social evalu- ation that in turn color the impressions one forms of another? What information is ignored and what information is made central in the formation of an impression? A number of investigators have sought a partial answer to these questions by assuming that a basic feature of naive phenomenology is the assign- ment of observed behavior to psychological causes. It seems logical to propose, for example, that behavior whose locus of causation lies within the person is more relevant to inferences about his particular characteristics than behavior that is induced or constrained by external events. The present investigation was designed to demonstrate this proposition with specific reference to the adoption and per- formance of social roles. The concept of role has had a lively and controversial history in the literature of social science. It is often treated as a crucial bridging concept since it concerns the relations between social requirements and normative expecta- tions on the one hand, and individual percep- tions and behavior on the other. Controversy has surrounded the many attempts to define role, as Levinson (1959) notes, because these attempts have vacillated between viewing role as an aspect of social structure and viewing it as a description of socially relevant individual behavior. In the present paper, the concept of role refers to role demands rather than actual behavior. Role is herein treated as a set of expected behaviors implicit in the instructions to a stimulus person. These 1 This study was carried out under support from the National Science Foundation, G88S7. We are much in- debted to Barbara Chapman who served as experi- menter throughout the study. instructions define the impression the stimulus person should attempt to create in presenting himself to an interviewer, and variations in behavior given this role definition represent the major independent variable. The present treatment of role is quite consistent with any other treatment that stresses the shaping of individual responses by social expectations or externally imposed norms. The point has often been made that general adherence to relevant sets of social norms is very important in facilitating social interaction. Particularly in organizational contexts, but by no means exclusively there, many social interactions can be effectively described in terms of the interplay of appro- priate role behaviors. Jones and Thibaut (1958) have emphasized the economic signifi- cance of such interactions between roles as reducing the need for inferences about idiosyn- cratic personal characteristics. The comple- ment of this point is that behavior appropriate to role expectations has little informational value in highlighting these individual charac- teristics. To follow this line of reasoning a little further, roles facilitate interaction and the social cognitions that support it. The naive person has his own repertory of role constructs that help to anchor his perceptions of the social environment and to endow it with the necessary stability for planful action. On the other hand, the performance of social roles tends to mask information about individual characteristics because the person reveals only that he is responsive to normative requirements. If these requirements are un- clear or conflicting, of course, he may re- veal something about himself by the way in which he defines and displays appropriate behavior. The stronger and more unequivocal the role demands, however, the less informa- tion is provided by behavior appropriate to the role. Following our introductory comments, this conclusion may be derived from con- sidering probable differences in the attribution of phenomenal causality. When a person's 302