Enacting recovery: Virtual reality, active inference, and cognitive behavioural therapy for depression

Attempts to use virtual reality technology in the treatment of depression have yielded disappointing results, lacking foundational theoretical principles. In this article, we aim to provide the theoretical foundation and guidelines for the principled use of virtual reality in cognitive behavioral therapy treatments for depression. We leverage the active inference framework, a neuroscientific theory grounded in probabilistic belief updating and action, to furnish us with an understanding of action and learning as uncertainty minimization. This approach through active inference is well justified, as recent work in active inference has provided models of both depression itself, and of the mechanisms underpinning the efficacy of cognitive behavioral therapy. In short, we argue that the aim of virtually assisted cognitive behavioral therapy should be to introduce a degree of flexibility into an individual’s action-based world models. Where depression is characterized on the computational level by overly rigid beliefs, we can use the curation of virtual affordances to disrupt its characterizing patterns, primarily through curating affordances that maximize the learning rate in relation to desired outcomes.