2026/07/21 by Gabriel Braun, Einat Shetreet, Yaara Yeshurun · 1 voice
Neuroscience · Psychology · #Action Observation and Synchronization #Memory Processes and Influences #Memory and Neural Mechanisms
paper · pdf · doi:10.21203/rs.3.rs-10373454/v1
crossref issued 2026/07/21 · crossref published 2026/07/21 · openalex publication_date 2026/07/21 · crossref created 2026/07/21 · openalex created_date 2026/07/22 · openalex updated_date 2026/07/22 · crossref deposited 2026/07/31 · crossref indexed 2026/07/31
Abstract In an era of competing narratives and widespread misinformation, people must continuously evaluate what to believe and what not to. Here, we aimed to characterize the representational geometry of (dis)belief during narrative comprehension. We analyzed a naturalistic fMRI dataset in which participants listened to extended testimonies and rated their eventual belief. Rather than assuming belief is represented along a single axis, we compared multiple belief-similarity models, each capturing a distinct hypothesis about how belief guides neural activity. Across narratives, belief-related neural activity tended to organize around two axes: directional position along the belief-disbelief continuum and, unexpectedly, extremity of belief, irrespective of direction. We then used this neural result to motivate a behavioral prediction: if extremity reflects certainty in interpretation, certainty should increase toward both belief and disbelief extremes. Consistent with this prediction, certainty showed a quadratic relationship with belief, with higher certainty for strong belief or disbelief and lower certainty near the midpoint. Finally, despite the prominence of directionality and extremity, a ridge-based analyses showed that a belief-related neural representation was best described as a mixture of different axes. Together, our findings suggest that belief is represented as a multidimensional, context-sensitive geometry organized primarily around belief directionality and belief extremity.