10 years of Bayesian theories of autism: a systematic review

Ten years ago, Pellicano and Burr published one of the most influential articles in the study of autism spectrum disorders, linking them to aberrant Bayesian inference processes in the brain. In particular, they proposed that autistic individuals are less influenced by their prior beliefs about the environment. Since then, multiple studies have attempted to test this theory and its subsequent predictive coding formulations. In this comprehensive review, we collect all relevant studies from the past ten years which included comparisons between autistic and neurotypical individuals or measured the participants’ autistic traits. We categorize them based on the type of the investigated priors and synthesize their findings. Our results show mixed evidence overall, with a slight majority of studies finding no general impairment in the integration of Bayesian priors. We show that priors developed during the experiments were more frequently impaired than those that participants had acquired previously, with various studies providing evidence for learning differences between participant groups. Together these findings hint at a deficit in the development of priors in autism. We also focus on the methodological and computational aspects of the included studies, finding low statistical power and often inconsistent approaches. Based on our findings, we propose guidelines for future research.