Reflection impulsivity in binge drinking: behavioural and volumetric correlates

Abstract

The degree to which an individual accumulates evidence prior to making a decision, also known as reflection impulsivity, can be affected in psychiatric disorders. Here, we study decisional impulsivity in binge drinkers, a group at elevated risk for developing alcohol use disorders, comparing two tasks assessing reflection impulsivity and a delay discounting task, hypothesizing impairments in both subtypes of impulsivity. We also assess volumetric correlates of reflection impulsivity focusing on regions previously implicated in functional magnetic resonance imaging studies. Sixty binge drinkers and healthy volunteers were tested using two different information-gathering paradigms: the beads task and the Information Sampling Task (IST). The beads task was analysed using a behavioural approach and a Bayesian model of decision making. Delay discounting was assessed using the Monetary Choice Questionnaire. Regression analyses of primary outcomes were conducted with voxel-based morphometry analyses. Binge drinkers sought less evidence prior to decision in the beads task compared with healthy volunteers in both the behavioural and computational modelling analysis. There were no group differences in the IST or delay discounting task. Greater impulsivity as indexed by lower evidence accumulation in the beads task was associated with smaller dorsolateral prefrontal cortex and inferior parietal volumes. In contrast, greater impulsivity as indexed by lower evidence accumulation in the IST was associated with greater dorsal cingulate and precuneus volumes. Binge drinking is characterized by impaired reflection impulsivity suggesting a deficit in deciding on the basis of future outcomes that are more difficult to represent. These findings emphasize the role of possible therapeutic interventions targeting decision-making deficits.

Keywords: Binge drinking; decision making; reflection impulsivity; voxel-based morphometry.

Figure 1

Beads task and relationship with brain volumes. Top panel: Beads task. Subjects viewed two jars with opposite ratios of red and blue beads (Jar 1: P = 0.80 red; P = 0.20 blue/Jar 2: P = 0.80 blue; P = 0.20 red). Beads selected from a single jar were sequentially shown to the participants. The goal was to infer from which jar the beads were being selected. After each bead was drawn, participants either chose to draw another bead or to make a decision. The drawn beads remained on display at the top of the screen. Bottom panel: Left dorsolateral prefrontal (DLPFC) and left inferior parietal cortices volumes positively correlated with the number of beads drawn in beads task. The data are shown for healthy volunteers (HV) and binge drinkers (BD)

Figure 2

Information Sampling Task (IST) and relationship with brain volumes. Top panel: IST. A 5 × 5 matrix of grey boxes were presented on a touch‐screen monitor. Upon being touched, each box opened to reveal one of two colours, shown as panels below the matrix. The goal was to decide which colour was predominant in the matrix. The ‘fixed win’ condition is shown in which the number of points decreased by 10 points for every box touched. Bottom panel: Left dorsal cingulate cortex (left) and right precuneus (right) volumes negatively correlated with the number of boxes opened in the IST. The data are shown for healthy volunteers (HV) and binge drinkers (BD)

Figure 3

Behavioural results. (a) Group comparisons of binge drinkers (BD) and healthy volunteers (HV) for the number of beads drawn on the beads task (BT) and the number of boxes opened on Information Sampling Task (IST) and K‐value of the Delay Discounting Task (DDT). For illustration purposes, the raw data are shown rather than the transformed data, **P < 0.005. (b) Total points won for the IST comparing BD with HV, *P < 0.05. Group and group‐by‐cost interaction. (c) Correlation analyses between samples drawn in the BT and the Alcohol Use Disorders Identification Test (AUDIT) scores for each group (BD and HV)

Figure 4

Computational modelling results. Binge drinkers (BD) showed higher ‘cognitive noise’ than healthy volunteers (HV) and greater variability within the group. The plots show medians, interquartiles (solid), full ranges (whiskers) and outliers (circles). The notches are non‐parametric confidence intervals