Higher alcohol use is associated with subsequent increased risk seeking toward gains: A longitudinal cohort study in young men

Abstract

Background: A higher propensity for impulsive and risky choices has often been reported in individuals with addiction, such as alcohol use disorder (AUD). Although domains of choice impulsivity, for example, temporal discounting, have been identified to predispose the development of hazardous alcohol use, research on altered decision making as a consequence of drinking is scarce. These may be particularly pronounced during periods of progressive brain development, such as young adulthood.

Methods: This 3-year prospective study investigated the effects of alcohol use on changes in four decision-making domains in 18-year-old men (N = 130). We assessed temporal changes in discounting of delayed rewards, risk aversion for gains, risk seeking for losses, and loss aversion. By correlating three-year cumulative alcohol consumption and total binge drinking frequency with respective change scores, we aimed to explore the influences of drinking on altered development in different impulsive choice tendencies.

Results: From ages 18 to 21, choice impulsivity in our moderately drinking cohort decreased significantly with respect to temporal discounting and risk aversion for gains, while risk seeking for losses and loss aversion did not change significantly. Importantly, higher cumulative alcohol intake and more binge drinking occasions over 3 years were associated with slower increases in risk aversion for gains, that is, the general trend for lower choice impulsivity was diminished. Such relationships were not found for temporal discounting, risk seeking for losses, or loss aversion.

Conclusions: Alcohol consumption in young adulthood is linked to stunted development of risk aversion for gains. Given that risk seeking for gains was previously identified as a risk factor for increasing alcohol intake, this relationship may reinforce a spiral of escalating consumption over time. The absence of similar findings in other decision-making domains suggests that drinking behavior and modifying factors interact differently across domains, rather than universally enhancing impulsive choice behavior.

Keywords: alcohol use; decision making; delay discounting; loss aversion; risk taking.

FIGURE 1

Value‐based decision‐making (VBDM) battery comprised four tasks: delay discounting (DD), probability discounting for gains (PDG), probability discounting for losses (PDL), and mixed gambles (MG); for details, see Pooseh et al. (2018). Every task included forced choice trials for one of two offers. Within the DD task, a smaller immediate versus a higher delayed reward was offered. The PD tasks offered a certain smaller versus a probabilistic larger gain (PDG) or loss (PDL). The MG task incorporated rejecting or accepting a 50/50 gamble between a monetary gain versus loss.

FIGURE 2

Alcohol use from age 18 until age 21. (A–C) Individual drinking trajectories. Blue lines refer to the median. (D) Distribution of cumulative consumption, which represents the sum of all assessments over 3 years, excluding the year before baseline. Number at the top right refers to Spearman's correlation.

FIGURE 3

Development in different domains of value‐based decision making from ages 18 to 21. (A) Boxplots of raw parameter estimates and (C, left) choice consistency at both ages. Only individuals with non‐missing values are included. (B) Distribution of change scores of the parameters and (C, right) choice consistency, calculated by subtracting log‐transformed values at age 18 from those at age 21. Blue lines refer to the median. DD, delay discounting; MG, mixed gambles (loss aversion); PDG, probability discounting for gains; PDL, probability discounting for losses. Asterisks refer to the Wilcoxon signed‐rank test for paired samples: **p < 0.01, *p < 0.05.

FIGURE 4

Associations between cumulative alcohol consumption and change scores of (A, B) different decision‐making dimensions as well as (C) choice consistency from ages 18 to 21. Change scores were calculated by subtracting log‐transformed values at age 18 from those at age 21. Values of choice consistency represent the average of log‐transformed and mean‐centered β‐values of DD, PDG, PDL, and MG. Spearman correlations (ρ) are shown at the bottom. DD, delay discounting; MG, mixed gambles (loss aversion); PDG, probability discounting for gains; PDL, probability discounting for losses.