Rationale: Both from a clinical and theoretical standpoint, it's essential to understand how the evaluative reinforcement learning mechanisms (Model-Free, MF and Model-Based, MB) operate under challenging conditions like stress, especially in Alcohol Use Disorder (AUD). Yet, this area remains scarcely studied. Objective: To investigate whether AUD individuals tend to favor habitual behaviors (MF) over prospective and deliberative simulations (MB) more than their controls and to explore the modulation role of social stress on their respective computational estimates. Methods: A cohort of 145 participants, including 67 with AUD (not undergoing treatment) and 78 controls, was analyzed. Acute social stress was introduced via the socially evaluated cold pressure task (SECPT). Subsequent engagement in a two-step Markov task allowed the evaluation of MB and MF learning tendencies using a Hybrid-RL 7-parameter model. We also evaluated biological and subjective stress levels, verbal working memory, and fluid intelligence. Results: Both groups showed similar baseline cortisol levels and responses to SECPT. In the absence of stress, AUD subjects relied less on MB learning than controls. Notably, under stress, the AUD group mirrored the MB levels of controls. Regression analysis showed that, for controls, cortisol increases due to stress shifted the MB/MF ratio towards MF, a trend not observed in AUD participants. The results were unrelated to the AUD participants' compromised verbal working memory. Conclusions: While individuals with AUD generally display decreased MB reliance, they exhibit resilience in this deliberative mechanism during an acute social stressor, underscoring the intricate dynamics within AUD and hinting at heightened goal-driven.

Authors:
Xavier Noel, The Free University of Brussels, Belgium

About the Presenter(s)
Professor Xavier NOEL is a University Associate Professor/Senior Lecturer at Université Libre de Bruxelles (ULB) in Belgium

See this presentation on the full schedule – Wednesday Schedule

Bookmark