Proceedings of the Annual Meeting of the Cognitive Science Society

Multi-person conversation is a crucible for social organization and human ingenuity. But not everybody gets equal access.Members of minority and marginalized groups can struggle to participate. Why? Explanations have focused on institu-tional factors, socialization (e.g., feminine communication styles), or ubiquitous prejudice. Here, we propose that it maybe a pernicious consequence of otherwise rational processes: namely, the role of experienced-based prediction in nego-tiating turn-taking during communication (e.g., through gaze allocation). Using an agent-based model, we demonstratethat this mechanism suffices to explain phenomena that have been reported empirically, but without a unified treatment:members of minority or marginalized groups talk less; this is more pronounced in larger groups; despite talking less, theyare perceived to talk more; they are more likely to be interrupted. Besides practical implications for increasing partici-pation by underrepresented groups, we discuss theoretical implications for the emergence of group-level inequality fromindividual cognitive processes.