Risk-Taking Rises Before Drinking Does, Year-Long Smartphone Study Finds

People with alcohol use disorder tend to drink more in the month after their decision-making shifts toward risk, according to a year-long study that used gamified smartphone tasks to test the same people again and again. The finding, published in Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, reverses the assumption that heavy drinking comes first and scrambled judgment follows.

Researchers at University Hospital Würzburg and Technische Universität Dresden followed adults with mostly mild to moderate alcohol use disorder for twelve months. Participants completed decision-making tasks every month on their phones and reported their drinking every two days, a schedule designed to catch changes as they happened rather than asking people to remember a year of it.

What the monthly tasks measured

The tasks were built as games, not questionnaires. They probed three things: how much risk a person accepted when a choice mixed potential gains with potential losses, how they behaved when every option was a loss, and how much information they gathered before committing to a choice. Participants also completed measures of what researchers call cognitive control — response inhibition and working memory — the capacities usually invoked to explain why someone keeps drinking past the point they intended to stop.

Cameron S. Carter, editor-in-chief of the journal and a psychiatrist at the University of California, Irvine School of Medicine, said in a statement that decision-making and cognitive control "cannot be accurately examined through interviews or self-reports and instead require experimental tasks." The Würzburg and Dresden team built phone versions specifically to track how those functions move in ordinary life.

The direction of the arrow

Three monthly shifts predicted heavier drinking in the following month. Participants drank more when their risk-taking had been higher in the mixed gambling task, when their risk-taking had been lower in the pure-loss task, and when their information sampling had become less thorough — the pattern of deciding before looking.

The reverse did not hold. Drinking more in one month did not predict a riskier or more impulsive profile in the next, even after the analysis accounted for how much each participant had been drinking before. "This temporal direction was specific and was not observed vice versa," said Lorenz Deserno, an associate professor of neuroscience in developmental psychiatry at Würzburg and co-lead investigator.

The results were also narrower than the field's usual framing. Fluctuations in response inhibition and working memory — the self-control measures — were not associated with drinking changes at all. Neither was reward-driven behavior in general. What moved with drinking was a specific slice of decision-making: how a person weighed risk, and how quickly they stopped gathering evidence.

Why the timing matters more than the mechanism

A month is not a clinical deadline, but it is a window. Hilmar Zech, a postdoctoral fellow at Würzburg and Dresden who co-led the study, framed the practical implication as detection rather than treatment: if a monthly mobile check can flag a period when someone's risk calculus has shifted, support could arrive before the drinking increase rather than after it.

That is the premise behind just-in-time adaptive interventions, a digital-health approach already used in smoking cessation and diabetes management, in which a system watches for a change in state and responds within hours or days. The study does not test such an intervention. It establishes that the signal exists and that it precedes the outcome, which is the prerequisite for building one.

Treatment for alcohol use disorder today still mostly begins after a crisis — a hospitalization, an arrest, a family ultimatum. Intensive outpatient programs and medication-assisted treatment work best when people reach them before the consequences compound, and the study's authors position early detection of a high-risk window as a way to shorten that gap. For people already in care, the finding suggests that a month of unusual risk-taking is worth a conversation rather than a wait-and-see. Substance use disorder care generally treats relapse as a discrete event; this study describes something more like a slow tilt that happens first.

The limits of a phone-based study

The research release did not state how many people enrolled, what they were paid, or where they were recruited, and those omissions matter for judging how far the results generalize. A cohort that agrees to monthly tasks and reports drinking every other day for a year is, by construction, more engaged than the population with the heaviest drinking and the least stability — precisely the group that tends to be missing from longitudinal research.

The participants also had mild to moderate disorder, a group whose drinking fluctuates rather than holds steady. Whether the same monthly signal appears in people with severe dependence, or in those whose drinking is shaped by withdrawal, is untested here.

Those caveats do not undo the temporal finding, which required no assumption beyond the order in which measurements were taken. They do mean the study is best read as a proof that the signal can be measured cheaply and repeatedly — a map of where to look next, not yet a tool to hand a clinician.