Typical use of prior object knowledge in perceptual decisions in autism

A prominent theoretical framework in cognitive neuroscience proposes that the brain functions as a probabilistic prediction machine, continuously generating and updating hypotheses about sensory input based on prior experience. Within this framework, perception is the result of dynamic interactions between sensory inputs and prior knowledge. It has been hypothesised that perception in autism spectrum conditions (ASCs) is characterised by a reduced influence of prior knowledge, leading to a more data-driven or less contextually modulated processing of sensory input. While some evidence supports this predictive processing account of ASCs, findings are inconsistent. In the present study, we examined the role of prior object knowledge in perceptual inference using a two-tone image paradigm. Two-tone images are uninterpretable without prior knowledge of image content but result in coherent object percepts once prior knowledge has been acquired. We tested whether incorporation of prior knowledge into perceptual inference differs between adults with and without autism. Our results show that both autistic and non-autistic participants exhibited comparable improvements in image recognition following acquisition of prior knowledge. These findings challenge the hypothesis of a reduced influence of prior knowledge on visual perception in autism, and argue against a domain-general predictive processing deficit.