A neurocomputational model of delusion

Abstract

Delusions are currently characterised as false beliefs produced by incorrect inference about external reality (DSM IV, 1994). This inferential account has proved hard to link to explanations pitched at the level of neurobiology and neuroanatomy. This paper provides that link via a neurocomputational theory, based on evolutionary considerations, of the role of the prefrontal cortex (PFC) in managing responses to experience. Neural network theory distinguishes between weight-based and activation based-processing. This distinction maps (roughly) to a distinction between modularised online cognition conducted by perceptual and sensory systems and offline cognition under the control of the PFC. The PFC regulates activation-based processing in transient neural networks constructed to deal with experiences produced as outputs of weight-based sensory and perceptual systems. The advantages of this implementation-level account are explored via a comparison with Shitij Kapur’s influential attempt to link dopamine dysregulation to the phenomenology of schizophrenia..