Evidence for multiple kinds of belief in theory of mind

People routinely appeal to ‘beliefs’ in explaining behavior; psychologists do so as well (for instance, in explaining belief polarization and learning). Across three studies (N = 1,843, U.S-based adults), we challenge the assumption that ‘belief’ picks out a single construct in people’s theory of mind. Instead, laypeople attribute different kinds of beliefs depending on whether the beliefs play predominantly epistemic roles (such as truth-tracking) or non-epistemic roles (such as social signaling). We demonstrate that epistemic and non-epistemic beliefs are attributed under different circumstances (Study 1) and support different predictions about the believer’s values (Study 2) and behavior (Study 3). This differentiation emerges reliably across three distinct signatures of attributed belief, and even when the believed content and attributed level of certainty about that content are held constant across cases. Our findings call for a more fine-grained characterization of theory of mind and provide indirect support for the hypothesis that human cognition itself features multiple varieties of belief.