A Case Study in Computational Psychiatry

We review a new perspective on addiction as due to failure modes of decision-making networks. In a sense, this suggests that addiction is a symptom that can arise from any of a number of potential underlying vulnerabilities. We identify four primary action-selection systems and review how multiple deficits (or “failure modes”) of these systems can lead to continued harmful dysfunction typically identified as addiction. These methods have shaped a new generation of tools for studying the etiology of neuropsychiatric dysfunction. These tools are aimed at identifying specific failure modes so that treatments can be individualized for specific patients. Moving beyond dysfunction, we also review how a computational understanding of treatment paradigms can reveal their interaction with these multiple decision systems, which can suggest ways to identify the patients most likely to be helped by treatments and ways to improve the treatments themselves.