Explore-exploit behaviors predict broad autism social phenotypes in general population

Autism spectrum disorder (ASD) is most often defined by social communication challenges and behavioral rigidity, but executive function deficits have long been considered potential contributors to autism-related impairments across these domains. Autism is a spectrum, with a broad range of phenotypes presenting below the diagnostic threshold, raising the possibility that variability in executive function may contribute to individual differences in autistic traits in the general population. Value based decision making tasks access aspects of executive function, and critically are amenable to computational approaches to dissect the latent variables that most contribute to individual differences in cognition. We capitalize on this approach to uncover the relationship between autistic traits as measured by the Broad Autism Phenotype Questionnaire (BAPQ) and explore-exploit balance in a three-armed restless bandit decision making task in a large (1001 participants) sample. We find that the BAPQ aloof subscale, which primarily describes social behavior related phenotypes, most strongly explains changes in choice behaviors, including sensitivity to outcomes, changes in choice flexibility, and level of exploration as inferred from a Hidden Markov Model. Canonical correlation analysis reveals that the strongest loading for these non-social reward related measures are in fact socially coded items. These findings suggest that different aspects of executive function challenges may be related to social and nonsocial autism-related behaviors in the general population, and that social components of behavior produce measurable differences in nonsocial tasks.