Linking Subclinical Autistic Traits and Perceptual Category Learning

Abstract

Autism spectrum condition is a neurodevelopmental condition with difficulties in social interaction, communication and repetitive behaviours. Autistic individuals often exhibit difficulties in non-social cognitive processing, such as grouping items into meaningful categories based on their holistic visual appearance. Underlying mechanisms might be a deficit in abstracting a category's central tendency (i.e., the prototype) or more general atypicalities in visual category learning processes. Milder autistic traits often also extend to a broader autism phenotype in neurotypical individuals. Our study compared adult neurotypical individuals with high or low autistic traits on behavioural performances and neural correlates measured by event-related functional magnetic resonance imaging (fMRI) during a single-category perceptual categorization task, based on the well-known dot-pattern paradigm. Bayesian computational modelling was used to investigate links between autistic traits and representing category knowledge by the prototype or memorizing single exemplars. We found that a high degree of autistic traits was linked to worse accuracy for endorsing category members. Autistic trait groups also differed in neural correlates during the training phase related to visual processing in occipital regions, decision-making in midfrontal regions and the posterior cingulate, and feedback processing in the posterior cingulate and the ventral striatum. Model-based analyses did not support deficits in prototype abstraction but yielded a link between autistic traits and stricter decision policies. In sum, we found no relationship between high autistic traits and difficulties with the prototype strategy but more general atypicalities in visual category learning processes, namely, visual processing, decision-making and feedback processing.