A Computational Perspective on Cognitive Behavioural Therapy

Cognitive Behavioural Therapy (CBT) contains multiple conceptions about the origin, nature, and besttreatment of psychopathology. In this Review, we synthesise insights from the rich field of neuralreinforcement learning to develop a principled theoretical framework that encompasses the interventionsthat were developed within CBT’s three waves, and links them to computational notions ofpsychopathology. We use a conception of meta-cognition as an internal theory-of-mind to develop threehierarchically interlinked computational layers that capture distinct aberrations in learning and decisionmaking in psychiatric diseases, and that collectively dissect psychotherapeutic interventional targets. Ourframework is intended to serve as connective tissue and bolster bidirectional translation between basicresearch and clinical practice.