ABSTRACT
Although the predictive coding theory is not brand new, it has only recently been picked up by the autism academic community. This chapter discusses the research about predictive processing in autism and the hypotheses evolved from these studies. Although there are minor differences in these hypotheses, most researchers seem to agree that the process of predicting the world and dealing with prediction errors is much less context-sensitive in people with autism than in people without autism. The models created and used by the autistic brain to predict the world are absolute and therefore insufficiently contextual. It seems like an autistic brain also regards sensory input as more informative than the models of the world that the brain has at its disposal. An autistic brain processes prediction errors in absolute terms, rather than seeing them as being relative. All deviations from what is expected are regarded as being important, at all times and in all places. As a result, the number of prediction errors systematically increases, creating a vicious circle of hypervigilance for a world full of volatility and unpredictability.
