Understanding the Default Mode Network's Computational Role in Decision-Making

Understanding the neurobiological mechanisms of decision-making, especially in uncertain environments, remains a challenge in cognitive neuroscience. This study examines the role of the Default Mode Network (DMN) in decision-making, using fMRI data from the Human Connectome Project gambling (HCP) task. We modeled behavior with a reinforcement learning model, the Experience-Weighted Attraction (EWA) model, and a logistic regression model. While the EWA model showed inconsistent performance, the logistic regression model better captured behavioral patterns by predicting choices from past rewards and choices. Analysis of DMN activity during the pre-feedback period revealed its relationship with prior feedback and decisions. These findings suggest that the DMN potentially plays a role in processing past outcomes to inform future behavior. However, the limitations of the HCP gambling task highlight the need for more dynamic paradigms, such as probabilistic reversal learning, to further investigate adaptive decision-making.