A generative framework for the study of delusions

Despite the ubiquity of delusional information processing in psychopathology and everyday life, formal charac-terizations of such inferences are lacking. In this article, we propose a generative framework that entails a com-putational mechanism which, when implemented in a virtual agent and given new information, generates belief updates (i.e., inferences about the hidden causes of the information) that resemble those seen in individuals with delusions. We introduce a particular form of Dirichlet process mixture model with a sampling-based Bayesian in-ference algorithm. This procedure, depending on the setting of a single parameter, preferentially generates highly precise (i.e. over-fitting) explanations, which are compartmentalized and thus can co-exist despite being incon-sistent with each other. Especially in ambiguous situations, this can provide the seed for delusional ideation. Fur-ther, we show by simulation how the excessive generation of such over-precise explanations leads to new information being integrated in away that does not lead to a revision of established beliefs. In all configurations, whether delusional or not, the inference generated by our algorithm corresponds to Bayesian inference. Further-more, the algorithm is fully compatible with hierarchical predictive coding. By virtue of these properties, the pro-posed model provides a basis for the empirical study and a step toward the characterization of the aberrant inferential processes underlying delusions.(c) 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Automated summary

This article proposes a generative framework that can generate belief updates resembling those seen in individuals with delusions. The framework can generate highly precise explanations that can co-exist despite being inconsistent with each other. In ambiguous situations, this mechanism can lead to delusional ideation. The excessive generation of such over-precise explanations does not lead to a revision of established beliefs. The inference generated by this algorithm corresponds to Bayesian inference and is fully compatible with hierarchical predictive coding. This model provides a basis for empirical study and understanding of the aberrant inferential processes underlying delusions.