Memory biases in alcohol use disorder: enhanced memory for contexts associated with alcohol prospectively predicts alcohol use outcomes.

Memory for prior drinking experiences may powerfully drive later alcohol use in familiar drinking contexts, yet we know little about what patients with alcohol use disorder (AUD) remember of alcohol-related episodes. Although animal and theoretical models of addiction emphasize the importance of different memory systems for understanding maladaptive use, clinical research parsing what AUD patients remember from alcohol-related episodes is lacking. The current study applied a novel memory task in which moderate drinkers (N = 30) and treatment-seeking individuals with alcohol use disorder (AUD: N = 29) encoded associations between photographs of objects (alcoholic beverages and neutral items) and photographs of neutral scenes. At least 24 h later, two types of memory were assessed: item memory (object recognition) and associative memory (cued recognition of scenes associated with objects). To assess which memories predicted drinking, real-world behavior was assessed in patients with AUD at baseline and for 4 weeks following memory tests. Despite demographic differences, the results showed broadly impaired item memory in AUD compared with moderate drinkers (p < 0.001), but enhanced associative memory for scenes paired with alcohol (p = 0.015). These associative memory biases were especially pronounced for stimuli rated as more affectively salient. Furthermore, stronger but less detailed memory for alcohol-related associations (i.e., choosing the correct scene but the incorrect photograph) significantly predicted heavier baseline (p = 0.002) and higher subsequent (p = 0.01) drinking in patients with AUD. These findings reveal a novel alcohol-related memory bias in AUD, and uncover the importance of associative memory for understanding real-world heavy alcohol use.