[What Connects Bayes's Rule to Mental Health?]

A branch of computational psychiatry aims to handle the problems which arose due to the heterogeneity and comorbidity of mental illnesses by capturing phenotypic differences behind symptoms. In order to more accurately describe the mechanisms of the mind and its disturbances, it uses the tools and methods of computational cognitive science. In this paper, we introduce the reader to the (mathematical) language in which (Bayesian) predictive coding was written, which holds one of the possible explanations of the perceptual differences found on the autism or the psychosis spectrum. By representing prior knowledge and sampled sensory information as (prior and likelihood, respectively) distributions, the Bayes-rule allows us to calculate the inferred posterior distribution. The aforementioned terms are introduced through simple examples, such as coin tosses (inferring how biased it is) or making a diagnosis, however, this framework might also help us understand the more complex mechanisms of the mind. Studies show that visual perception can be examined following Bayesian formalization. The likelihood information can be manipulated with bistable or noisy stimuli, whereas through instructions and/or cues, prior expectations are formed. With the use of perceptual and response models fitted to cognitive task performance, it becomes possible to identify parameters which shed light on the features of information processing that characterize the perceptual alterations in certain clinical populations. At the end of the paper we draw attention to the limitations of computational modelling.