Attenuated inferred–sensory mismatch during masked face recognition in schizophrenia

Introduction Predictive-processing accounts have become increasingly influential in understanding the pathophysiology of schizophrenia (SZ). Within this framework, mismatch negativity (MMN) paradigms provide a well-established and widely replicated assay of mismatch processing. However, most MMN designs operationalize mismatch as violations of externally defined regularities. In this study, we aimed to use magnetoencephalography and a masked-face paradigm to probe a complementary form of mismatch that arises when an internally inferred representation of an occluded face is confronted with newly available sensory evidence after unmasking (an inferred–sensory mismatch). Methods Participants first viewed the masked faces, followed by the same faces without masks. Two types of stimuli—individual and average—were used. Data from 18 individuals with SZ and 18 healthy controls (HCs) were analyzed using spatiotemporal cluster tests. Results Among HCs, mask removal from individual faces elicited a significant cluster at the left temporal sensors (234–511 ms), with source estimates suggesting involvement of the left insular cortex. No significant clusters were observed for average faces. In the SZ group, no significant mismatch responses were observed in either condition. Discussion Overall, HCs showed a reliable neural response to the unmasking of individual faces, consistent with a mismatch between an internally inferred representation and newly revealed sensory information. In contrast, no robust mismatch-related response was detected in the SZ group in the present dataset. This pattern is compatible with predictive-processing accounts that emphasize reduced precision of perceptual inference in SZ.