Is Schizophrenia a deficit in how the brain performs Bayesian inference?
Current theories suggest that schizophrenia could be described as a deficit in how the brain learns and uses internal models of its environment. In a collaboration with the Royal Edinburgh Hospital, Dr Peggy Seriès and her team tested those theories by modelling the behaviour of patients with schizophrenia performing a visual statistical learning task.
This work builds on our previous work showing that humans are very quick to learn an internal model of our environment (e.g. what objects we are likely to encounter or not) based on our experience. We form expectations that can bias our perception, to the point that we will tend to perceive reality as being more similar to what we expect than it really is. Sometimes we will even hallucinate what we expect, even if it's not there.
Current theories suggest that this process is generally disrupted in schizophrenia. Patients would form a distorted internal model of the world, and would either trust it too much or - on the contrary - would sometimes fail to access it. This would lead them to hallucinating and sticking to irrational beliefs.
V. Valton, P. Karvelis, K. L. Richards, A. R. Seitz, S. Lawrie and P. Seriès (2019). Acquisition of visual priors and induced hallucinations in chronic schizophrenia. Brain, in press.