Abstract
BACKGROUND: Addiction is characterized by compulsive drug seeking and substance use, yet many individuals break free of these patterns and change their behavior. Traditional candidate predictors of behavior change/persistence rely on self-reports of factors such as readiness to change. However, explicit measures only characterize top-down influences on behavior. The incentive sensitization model of addition suggests that more implicit, automatic processes, such as the tendency to approach substance cues, play a major role in behavior. METHODS: We examined implicit alcohol approach and avoidance tendencies using a reaction time (RT) task in a sample of problem drinkers with alcohol use disorder (AUD) seeking to reduce heavy drinking. We measured alcohol approach and avoidance tendencies at baseline and at outcome, 12 weeks later. We asked whether alcohol approach and avoidance tendencies (i) changed over time, (ii) related to current drinking, and (iii) predicted changes in drinking from baseline to outcome. RESULTS: Approach and avoidance tendencies did not significantly change over time, nor did they correlate with current drinking, but these tendencies at baseline did predict drinking weeks later. Faster alcohol approach was associated with greater overall drinking at outcome, and faster alcohol avoidance predicted fewer drinking days per week at outcome. Exploratory analyses examined the relationship between approach and avoidance and traditional explicit measures including appraisals of alcohol and motivation to change. Implicit approach tendencies were largely distinct from explicit measures, and approach and avoidance tendencies explained unique variance in outcome drinking. CONCLUSIONS: The current findings suggest that implicit alcohol approach and avoidance tendencies assessed via a simple reaction time task can predict relative changes in drinking weeks later. Given that many explicit measures typically used in treatment studies fail to predict who will change, approach and avoidance tendencies are promising candidates to understand individual differences in treatment responses.