Does Alcohol Make Every Face Look More Hostile? An 84-Person Crossover Study

No. In an 84-person experiment, alcohol did not significantly increase hostile judgments across faces overall. More specific analyses found increased hostile judgments of happy faces and a pattern involving ambiguous expressions. Those conditional findings do not establish that alcohol makes everyone aggressive, or that a person’s demeanor reveals their alcohol concentration.

What the experiment measured

Eastwood and colleagues tested 84 regular, nondependent drinkers aged 18–40 in two sessions at least seven days apart. There were 42 men and 42 women. Each person received alcohol in one session and a placebo beverage in the other, with the order randomized and counterbalanced.

The alcohol dose was 0.4 grams per kilogram, with the dose calculation capped at 90 kilograms. Participants drank over ten minutes and waited another ten minutes before testing. A face-judgment task came first, followed by a joystick approach-and-avoidance task. Breath alcohol was measured after testing.

The reported post-session breath concentrations ranged from 8 to 32 micrograms per 100 milliliters of breath, with a mean of 16.62. These were breath measurements, not direct blood ethanol results. A single post-task measurement does not establish each person’s concentration throughout both tasks.

The overall result and the conditional results differ

In the face task, participants briefly saw male composite faces displaying emotions at different intensities and classified each as hostile or not hostile. Data for surprise expressions were excluded because of programming errors.

Across the analyzed emotions, mean hostile judgments were 31.61% after alcohol and 30.51% after placebo. The overall drink effect was not significant, p = .342. Reporting only that alcohol increased hostility would omit this central null result.

For happy faces specifically, hostile judgments averaged 8.09% after alcohol versus 5.07% after placebo, with a significant drink effect, p = .009. This is an increase of about three percentage points, rather than a three-percent relative increase.

An exploratory interaction between drink and expression intensity was also significant, p = .002: the alcohol-related difference was larger for low-intensity, ambiguous expressions and diminished as expressions became clearer. The intensity analyses were exploratory, and the study’s power calculation addressed the main drink effect rather than every interaction.

Faster joystick responses did not establish greater approach

The second task required pushing or pulling a joystick according to a border cue around an emotional face. It was not a decision to confront or avoid a person in real life. Three participants were excluded from this analysis for excessive errors.

Responses were generally faster after alcohol, but there was no significant drink-by-approach/avoidance interaction for any tested emotion, all p values greater than .260. General response speeding therefore did not demonstrate a selective tendency to approach threat.

The approach/avoidance measure also had poor split-half reliability, with a negative correlation of −.32. That weak consistency limits confidence in strong conclusions from its null results. Neither a claim of increased approach nor proof of no possible effect follows from this task alone.

Why the DUI relevance has limits

A driver’s interpretation of another person’s expression may matter during a roadside interaction or road conflict. This experiment offers a possible mechanism involving ambiguous social information. It did not measure driving, road rage, actual aggression or police encounters, however, and state anger did not show a significant alcohol or time effect.

The faces were brief, isolated images, not moving people in a real conversation. Most participants correctly identified the alcohol session, leaving expectancy incompletely controlled. The relatively young, selected sample and the single dose also restrict generalization.

It would be unjustified to work backward from hostility, politeness or an ambiguous facial interpretation to a person’s BAC, legal status or driving ability. Likewise, an absence of hostility cannot exclude deficits in attention or motor control. The site’s analysis of laboratory tests and driving prediction explains why a task-specific finding requires additional evidence before becoming a driving opinion.

Primary source

Eastwood APR, Penton-Voak IS, Munafò MR, Attwood AS. Effects of acute alcohol consumption on hostile attribution bias towards emotional facial expressions and implicit approach/avoidance tendencies. Journal of Psychopharmacology. 2026;40(9):1392–1404. https://doi.org/10.1177/02698811251408743.