Using computation to understand change in psychological therapy: a reinforcement learning framework of behavioural activation for depression
Year: 2023
Type: dissertation
Abstract: Behavioural activation (BA) is an effective learning-based psychological therapy for depression. BA is underpinned by the behavioural (learning) theory of depression which conceptualises depression as a recursive cycle of inactivity which both limits opportunities for positive reinforcement and, through avoidance, also elicits negative reinforcement from short-term relief of distress. BA aims to increase approach of reward and reduce avoidance by supporting clients in re-engaging with meaningful activities, whilst also monitoring the positive reinforcement from these behaviours impacting back on their mood. However, these mechanisms are not well specified meaning reduced clarity as to whether we are targeting them optimally in therapy and an imprecise understanding of how they effect change. The relatively nascent field of computational psychiatry provides a highly mechanistic approach across multiple levels of understanding [brain/affect/behaviour processes], by formalising these processes in explicit terms. Reinforcement learning in particular provides a rigorous mechanistic framework for the breakdown of goal-directed behaviour in depression. This conceptual introduction will review both fields and present an example therapy case, applying the core concepts of reinforcement learning to therapy concepts. It then reviews the empirical work at the intersection of these fields and attempts to lay down a roadmap for greater synergy between computational framework and BA for depression.
Author: Crawley, Daisy Victoria
Language: English
Subfield: Experimental and Cognitive Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7053316741