Persistent Belief Updating Despite Perfect Predictability – The Atypical Iterative Prior Updating Account Of Autism

Background: Within predictive coding accounts of autism spectrum disorder (ASD), two competing explanations locate perceptual atypicalities within inference: weak prior formation versus altered weighting of prediction errors. We propose the atypical iterative prior updating (AIPU) account, which shifts the locus from inference to what happens afterwards: sensory evidence is overweighted during iterative belief updating over time. The accounts predict distinct temporal signatures of attentional divergence. Methods: To distinguish among the accounts, we presented autistic and typically developing adults with a distractor suppression paradigm using a 100%-valid spatial cue and varying distractor prevalence, while eye movements tracked the temporal dynamics of attentional selection. Results: Both groups showed equivalent first-fixation capture, ruling out prior-formation differences. Distractor prevalence modulated interference differently across groups. Autistic participants produced larger reaction time interference, more fixations, and more target refixations, a pattern emerging after initial selection. These findings locate ASD-related differences specifically in iterative updating across time, not in perceptual inference. Conclusions: The mechanism points to sustained attentional engagement: rather than terminating sensory sampling once expectations are confirmed, autistic individuals continue accumulating evidence even under perfect predictability. The world might feel "too much" not because perceptual inference is altered, but because updating continues beyond resolution.