How people decide to consume (more) alcohol when feeling stressed

Background and Aims: The tension reduction hypothesis suggests that people consume alcohol to alleviate stress. While previous studies showed stress increases alcohol's absolute value, alcohol's value relative to alternatives should be more relevant for drinking decisions. We aimed to test whether acute stress causes individuals to choose alcohol over appealing non-alcoholic alternatives and to identify the cognitive mechanisms underlying this choice behavior.

Design: Laboratory-based randomized 2×2 experimental study.

Setting: Controlled laboratory environment including a simulated bar setting.

Participants: 160 adults (56% male; mean age=31 years; median AUDIT=8) who regularly consume alcohol.

Interventions: Participants first rated beverages and made repeated choices between alcoholic and non-alcoholic options. They then received either alcoholic beverages (target BrAC=.06%) or nonalcoholic beverages, followed by either a personalized stress induction using autobiographical emotional memories or a neutral control procedure.

Measurements: Primary outcomes were proportion of choices for alcoholic beverages and decision response times. Choice behavior was analyzed using drift diffusion modeling to decompose decisions into three mechanisms: decision carefulness (boundary parameter), sensitivity to prior preferences (drift rate), and bias toward alcohol regardless of preference (bias parameter).

Findings: Stress moderately increased choices for alcohol (95% HDI [0.01, 0.13]), but only in sober participants. Drift diffusion modeling revealed that stress primarily affected decisionmaking by inducing a bias toward alcohol during evidence accumulation (95% HDI [0.19, 0.76]), without impacting decision carefulness or evidence sensitivity. This computational bias was stronger than observed in raw choice behavior, indicating that while stress consistently biases evaluation toward alcohol, this bias only sometimes overcomes competing considerations (i.e., a person might reverse a preference from 'a little bit' to 'not really', but not from 'a little bit' to 'not at all').

Conclusions: Our results are relevant to adults who report risky drinking without alcohol treatment history or comorbidity and provide support for the tension reduction hypothesis by demonstrating that stress can occasionally lead individuals to choose alcohol even when they prefer the nonalcoholic alternative. However, this effect only appeared in sober participants who have not yet consumed any alcohol, suggesting the hypothesis primarily explains decisions about initiating rather than continuing drinking episodes.