Revealing dichotomous prior biases in social anxiety through a social prism model

Social situations can be overwhelming for some people, triggering avoidance and social anxiety (SA). However, it remains unknown why identical situations lead to different interpretations. Here, we developed a Social Prism Model to systematically address mechanisms underlying atypical social cognition in SA within a Bayesian cognitive framework. Through eight social valence judgment experiments (N = 541), we demonstrated that social anxiety may be shaped not by how sensory evidence is processed, but by robust dichotomous prior biases depending on social cues. The dichotomous prior biases indicated two parallel mental shortcuts that predispose individuals toward social fear, where negative prior biases were associated with an intolerance to uncertainty and a fear of social evaluations, and over-positive prior biases were associated with negative social learning. Our Bayesian simulations further demonstrated how variations in prior expectations parameters can give rise to biased social judgments, providing mechanistic support for the proposed framework. Collectively, these findings showed that the negative interpretations of the social situations can be understood within the Bayesian framework, highlighting a key role of prior expectation in shaping human social cognition.