Delusions Emerge and Persist due to Generative Model Reorganisation: Insights from Hybrid Predictive Coding

Predictive coding has influenced many conceptual accounts of delusions, the impervious and often bizarre beliefs that accompany a range of neuropsychiatric conditions. However, these explanations remain incomplete and have rarely been tested using formal modelling. Here, we present a formal account of delusional beliefs based on hybrid predictive coding, which sheds new light on the computational mechanisms underpinning two core features of delusions: their thematic recurrence and imperviousness to contradictory evidence. In simulation experiments, we demonstrate that a combination of contextually inadequate initialisation of beliefs and excessive certainty in them (a hallmark of psychosis) triggers a reorganisation of the generative model relating observed events to hidden causes. This reorganisation enables the maintenance of delusional beliefs that are thematically stable, internally consistent with external inputs, and impervious to contradictory evidence, all without an increase in global prediction error and even after certainty has returned to normal levels. Overall, our results suggest that delusions emerge and are maintained by generative models learned under atypical conditions. These findings provide mechanistic insights into the computations that may underpin delusions and provide fresh ideas for potential therapeutic strategies that focus on re-training generative models.