Slower Learning Rates from Negative Outcomes in Substance Use Disorder over a 1-Year Period and their Potential Predictive Utility

Computational modelling is a promising approach to parse dysfunctional cognitive processes in substance use disorders (SUDs), but it is unclear how much these processes change during the recovery period. We assessed 1-year follow-up data on a sample of treatment-seeking individuals with one or more SUDs (alcohol, cannabis, sedatives, stimulants, hallucinogens, and/or opioids; N = 83) that were previously assessed at baseline within a prior computational modelling study. Relative to healthy controls (HCs; N = 48), these participants were found at baseline to show altered learning rates and less precise action selection while completing an explore-exploit decision-making task. Here we replicate these analyses when these individuals returned and re-performed the task 1 year later to assess the stability of these baseline differences. We also examine whether baseline modelling measures can predict symptoms at follow-up. Bayesian analyses indicate that: (a) group differences in learning rates were stable over time (posterior probability = 1); (b) intra-class correlations (ICCs) between model parameters at baseline and follow-up were significant and ranged from small to moderate (.25 < ICCs < .54); and (c) learning rates and/or information-seeking values at baseline were associated with substance use severity at 1-year follow-up in stimulant and opioid users (.36 < rs < .43, .002 < ps < .02). These findings suggest that learning dysfunctions are moderately stable during recovery and could correspond to trait-like vulnerability factors. In addition, computational measures at baseline had some predictive value for changes in substance use severity over time and could be clinically informative.