BACKGROUND: Impulsive decision making relates to problematic substance use. Specifically, altered delay discounting (DD) has been suggested as a behavioral marker for addiction, while other relevant facets of choice impulsivity such as probability discounting (PD) or loss aversion are clearly understudied.

METHODS: Two studies were performed collecting behavioral data on choice impulsivity with a value-based decision-making battery providing estimates of DD, PD for gains and losses, and loss aversion. Study (1): In a sample of 198 male 18-year-old social drinkers, we analyzed impulsive choice behavior and its association with alcohol consumption and self-report measures of substance use-related personality traits on a cross-sectional level. Additionally, the predictive value of baseline choice behavior for the trajectories of alcohol consumption over a 12-month follow-up period was evaluated. Study (2): Behavioral data on choice impulsivity were collected for 114 detoxified patients with alcohol use disorder (AUD) and 98 control participants. We analyzed group differences at baseline and assessed the predictive value of choice impulsivity for relapse to heavy alcohol use in patients during a follow-up period of 48 weeks.

RESULTS: Study (1): Only DD was associated with baseline alcohol use, but no measure of choice impulsivity predicted the drinking trajectories over the following 12 months. Study (2): Compared to the control group, AUD patients showed higher DD, lower risk aversion regarding probabilistic gains, lower risk seeking regarding probabilistic losses, and lower loss aversion facing mixed prospects. Further, shallow discounting of probabilistic losses at baseline was predictive for relapse in patients.

CONCLUSIONS: All 4 domains of impulsive decision making were considerably altered in AUD patients though mostly not related to alcohol use in young adult social drinkers. This suggests that these facets of impulsive behavior may develop as consequences of chronic alcohol consumption. Furthermore, discounting of probabilistic losses might prove valuable in identifying patients vulnerable for relapse.