Risky choices come before heavier drinking, a one-year study finds

Every four weeks for a year, participants with alcohol use disorder, the clinical term for problematic drinking, opened a smartphone and played short games. Four tasks measured response inhibition (holding back an automatic reflex), working memory, risk-taking and information sampling (how much evidence someone gathers before deciding). Between sessions, they answered questionnaires about their drinking every two days. Dr. Hilmar Zech, of the University Hospital Würzburg and Technische Universität Dresden, and Dr. Lorenz Deserno, of the Würzburg lab, followed 603 people, mostly with mild to moderate disorder, and tracked which changes came first.

Until now, most evidence came from snapshot studies that measure people at a single moment. They had linked alcohol use disorder to weaker cognitive control and riskier decisions, but a snapshot cannot show whether the risky choices or the drinking came first. That gap matters, because the disorder contributes heavily to disability and death worldwide.

Monthly swings in decision-making predicted drinking in the following month. When participants took more risk in a mixed gambling context, and less in a loss context, they drank more the next month. When information sampling biases decreased, they drank more in the subsequent month. The arrow ran one way: the same pattern did not appear in reverse, even after accounting for how much participants had already been drinking.

Other traits did not follow that path. Fluctuations in self-control, meaning response inhibition and working memory, and in reward-driven risk-taking were not associated with changes in drinking. The journal's editor-in-chief, Cameron S. Carter, notes that decision-making cannot be reliably captured through interviews or self-reports alone, which is why the team built smartphone versions of experimental tasks. The study offers no treatment yet.

For people living with alcohol use disorder, and the clinicians who support them, the possible gain is timing. If a phone can flag stretches when risk-taking is climbing, a warning or a nudge could arrive before a high-risk period, rather than after it. Carter describes the goal as "mechanism-based, real-time interventions." The study does not say how soon such tools could reach clinics, and it tested prediction, not any intervention.