Decomposing decision mechanisms in female substance use disorder: drift diffusion modeling of context-dependent...

Decision-making impairments are central to substance use disorder (SUD), particularly in evaluating immediate versus delayed outcomes. However, conventional behavioral analyses provide limited insight into underlying cognitive mechanisms. This study applies the Drift Diffusion Model (DDM) to investigate intertemporal decision-making in female SUD across both gain and loss contexts, addressing a significant gap in understanding context-dependent decision processes.

The study compared 100 females with opioid use disorder to 86 female controls using intertemporal choice tasks in both gain and loss contexts. Participants made choices between smaller-immediate and larger-delayed monetary options across varying magnitudes, delay lengths, and reward differences. Behavioral preferences were analyzed using delay discounting models, while cognitive mechanisms were examined using hierarchical drift diffusion modeling to extract decision parameters (drift rates, thresholds, bias, non-decision time).

Behaviorally, the SUD group showed stronger preferences for immediate rewards in gain scenarios and stronger avoidance of immediate losses in loss scenarios compared to controls. Delay discounting analysis revealed significantly lower discount rates in the SUD group in loss contexts (p <.001). DDM analysis demonstrated that the SUD group exhibited lower decision thresholds across contexts, reflecting impulsive decision characteristics. Additionally, they showed lower drift rates in gain scenarios, indicating reduced sensitivity to non-substance rewards, but higher drift rates in loss scenarios, suggesting heightened sensitivity to negative outcomes. These decision patterns varied systematically with monetary and temporal parameters.

This study reveals distinct context-dependent decision biases in female SUD, characterized by computational signatures that differ markedly between gain and loss domains. These findings enhance our understanding of SUD-related decision mechanisms beyond traditional behavioral measures and suggest potential computational targets for individualized assessment and intervention approaches, though these clinical implications remain exploratory and require extensive validation before practical implementation.