Social interaction patterns and consensual categorization.

Abstract

We assume that individuals collectively construct category boundaries to interpret ambiguous events. The process by which this is achieved is called consensual categorization. The present research is concerned with the conditions under which consensual categorization is facilitated and how the existence of consensual categories affects judgments. In particular, it examines the impact of patterns of interactions among members, and the importance of the issue about which they are interacting, on the development and perception of consensual category boundaries. One hundred and three White and ninety-seven Asian-American subjects participated in an experiment and made a series of judgments about either representativeness or discrimination in university class situations. After judging individually, subjects in experimental conditions discussed one of the issues in an 8- (or 6-) person group. The discussion was broken up into six interaction episodes. Under localized interaction patterns, a subject alternated between the same two partners over episodes; whereas under dispersed conditions a member discussed the issue with a different partner in each episode. Subjects in control conditions were given a filler task instead of discussing the issue with others. After the discussion (or filler task), each subject judged the same materials individually. Consensual categories developed as a function of interaction patterns and issue importance: When the issue was relatively important (i.e., discrimination) subjects achieved a consensus with respect to category boundaries in both localized and dispersed interaction patterns; when the issue was relatively unimportant (i.e., representativeness), consensus developed only in localized patterns. However, to the extent that the issue was important, subjects were unlikely to accept the consensus privately. Moreover, subjects overestimated consensus more when their interaction patterns were localized. Individuals' judgments were analyzed using signal detection theory. In addition to subjects' explicit reports of their own category boundaries and perceived consensual boundaries, implicit boundaries were estimated from their judgments. Sensitivity based on implicit boundaries increased when there was consensus with respect to the less important issue; however, sensitivity did not change for the important issue, nor did it change when it was based on explicit or perceived boundaries. Response biases were not influenced by consensual categorization. Implications for the social construction of reality were discussed.