Unveiling the Aha! moment: a computational account of insight in active inference

Insight, or the 'Aha!' moment, is a cognitive phenomenon where a solution suddenly enters consciousness. This paper formalizes a computational model of insight using active inference, simulating it as Bayesian model reduction in a modified Wisconsin Card Sorting Task. The model reproduces insight's characteristic burst of confidence and discontinuous nature in an artificial agent. The authors propose that restructuring corresponds to Bayesian model reduction, impasse is a state of heightened uncertainty over actions, and problem-solving involves generative replay to create and evaluate alternative models for later reduction.

Study at a glance

Abstract

Insight, also referred to as the “Aha!” moment, is a distinctive cognitive phenomenon in which an idea or solution to a problem suddenly emerges into consciousness. Although extensively described in the literature, the nature of its distinctive cognitive and affective mechanisms is still poorly understood. In this paper, we formalize a computational approach to insight grounded in the active inference framework. By modeling insight as Bayesian model reduction in a modified Wisconsin Card Sorting Task, we succeed in simulating its distinct affective signature (i.e., its discontinuous nature and burst of confidence) in an artificial agent. In light of those simulations, we regard restructuring as Bayesian model reduction and impasse as a transient state of heightened uncertainty over possible actions. Finally, we propose that, during problem-solving, one regularly engages in phases of (generative) replay to generate and evaluate alternative models for subsequent model reduction.