Proceedings of the Annual Meeting of the Cognitive Science Society

The social content of paranoid delusions has prompted theories that paranoia stems from deficits in social cognition. For instance, paranoid individuals are more likely to falsely attribute harmful intentions to randomly moving visual patterns—'social hallucinations.' However, paranoia also involves domain-general deficits in belief updating and learning. To investigate if domain-general deficits are sufficient to explain social hallucinations, participants with varying paranoia levels completed reversal learning tasks (social/nonsocial _ easy/hard), then judged intentional motion in ambiguous displays. High-paranoia participants showed increased win-switching and elevated social hallucinations, with the former predicting the latter—suggesting linked mechanisms. Hierarchical Gaussian Filter modeling revealed context-dependent mechanisms: when volatility was salient, prior volatility beliefs (__) linked win-switching, social hallucinations, and paranoia; when social context was salient, meta-volatility learning (__) mediated these relationships. These findings demonstrate that social hallucinations emerge from domain-general belief updating problems operating through context-sensitive pathways.