Abstract Research has shown individuals with autism spectrum disorder (ASD) display atypical patterns in predictive processing, yet it remains controversial regarding what causes these atypical behaviors because findings have been inconsistent. Our study employed a two-state Bayesian model to decide between two suggested factors: an aberrant long-term prior and atypical short-term iterative belief updating. Both ASD individuals and typically developing (TD) counterparts participated in a task where they reproduced time durations over two sessions, one characterized by high volatility and the other by predictable sequences. Both sessions involved the same time durations, but the sequences differed in volatility. While both groups adjusted their prior belief to the level of volatility in a similar manner, they diverged in their short-term iterative updating of beliefs. Individuals with ASD iteratively adjusted their beliefs with each new stimulus as if it were less connected to previous ones compared to TD individuals, thereby assigning greater weight to incoming sensory information. Our findings highlight that the distinctive signature of cognitive processing in ASD lies in a mechanism of overreliance on sensory input specific to short-term iterative belief updating.. This unique characteristic may mirror an experience among individuals with ASD of temporal discontinuity rendering the world overly unpredictable.

Beyond Prior Belief and Volatility: The Distinct Iterative Prior Updating Process in ASD