EVIDENCE LIBRARY
Can Alcohol Impair the Decision to Stop Without Slowing a Stop Signal?
A controlled alcohol study found less voluntary stopping without general slowing on a stop-signal task. What can forensic experts actually infer?
Quick answer: Yes, moderate alcohol exposure can affect the decision to stop an ongoing action without producing the same pattern on a laboratory task that requires stopping after an external signal. In a randomized, placebo-controlled experiment, adults who consumed alcohol were less likely to voluntarily disengage from a computer tracking task, while alcohol did not produce a general slowing of externally cued stopping. The distinction matters: a breath-alcohol reading, a modeled laboratory stopping measure, and a person’s real-world capacity to perform a particular act are different evidence.
The primary study is Liu and colleagues’ original article, Moderate acute alcohol use impairs intentional inhibition rather than stimulus-driven inhibition, published online in 2020 and in the 2021 volume of Psychological Research (DOI: 10.1007/s00426-020-01353-w). Its findings are relevant to scientific interpretation of alcohol-associated behavior, but do not provide a field test, an individualized BAC calculation, or proof of why a particular person acted or failed to act.
Can someone respond to a stop signal yet be less willing to stop voluntarily?
That is the narrow finding this study helps explain. The researchers distinguished stimulus-driven inhibition—stopping a response when the environment presents a stop cue—from intentional inhibition—deciding without an external command to terminate an ongoing action. Both involve behavior control, but the study measured them using different computer tasks and different statistical endpoints.
In the externally cued task, an on-screen change told a participant to interrupt a keypress response. In the intentional task, participants followed a moving bottle with a cursor and chose whether, and when, to stop tracking in a short decision window. Tracking and stopping were associated with different point and lottery incentives. Neither task asked a participant to drive, handle a firearm, undergo a sobriety examination, or make a legal decision.
What was the experiment’s actual design?
The investigators randomly assigned 111 mostly student-age adults to an alcohol condition, a placebo condition, or a water control. Five were excluded from the reported main sample: three placebo participants did not believe they had consumed alcohol, one had a nonzero initial breath reading, and one had a positive THC screening result. The resulting study sample comprised 106 participants. Five additional participants received inconsistent instructions and were excluded from the analysis of the intentional-stopping task; that task therefore did not use the identical participant set as the overall study.
In the alcohol condition, men received a target dose of 0.55 g/kg ethanol and women 0.45 g/kg, mixed as vodka and orange juice. The placebo used tonic water with alcohol-associated sensory cues, including a vodka-soaked lemon and vodka applied to the glass rim. It should not be characterized as an absolutely ethanol-free preparation at the molecular level. The control received water. The design compared pharmacological alcohol exposure, alcohol expectancy, and a no-alcohol expectation, but was not a fully balanced placebo design.
The investigators monitored breath alcohol with an Alcoscan ALC-1 and collected subjective alcohol-effect ratings. They did not report a paired venous whole-blood ethanol time series or use a blood concentration to validate individual psychomotor deficits. The alcohol condition’s mean reported breath readings were 0.39‰ during each of the two tasks as expressed in the article; readings changed across the session, with mean values of 0.34‰ at one time point and 0.44‰ at a later one. Those values are reported breath measures in the paper’s notation, not separately assayed whole-blood BAC values.
What did the intentional-stopping task find?
For the chasing-bottles task, alcohol exposure was associated with a lower modeled rate of disengaging from tracking within the decision window. The study reported a hazard ratio of 0.68 (95% confidence interval 0.61–0.76) for the alcohol exposure contrast. Interpreted in the model’s context, that is approximately a 32% lower instantaneous hazard of stopping—not proof that exactly 32% fewer people stopped, and not a numerical prediction for a real person’s behavior.
The type of bottle cue also mattered. Within the alcohol condition, stopping was more frequent when the tracked picture represented an alcoholic beverage than when it represented a soft drink (reported hazard ratio 1.14, 95% CI 1.01–1.28, p = .03). The interaction of alcohol consumption and bottle category was statistically significant (p = .03). That pattern was not the simple claim that an alcohol cue always worsens behavior control. In fact, the study’s most pronounced group contrast concerned soft-drink stimuli.
The outcome was a statistical rate of voluntary tracking disengagement in a reward-structured task. It is neither an observed measure of drinking restraint outside the laboratory nor a diagnostic measurement of impulse-control capacity in an individual. The task’s design, reward structure, and 20-second stopping window all influence its interpretation.
Did alcohol slow the external stop-signal response?
Not systematically in this experiment. The stop-signal task generated an estimated stop-signal reaction time (SSRT) from keypress reaction times and an adaptive stop-signal delay. SSRT is a derived behavioral measure, not a direct recording of an isolated neural impulse or an ethanol concentration.
The main finding was a sex-by-alcohol interaction rather than a uniform slowing. In the article’s post hoc comparison, women receiving alcohol had a shorter mean SSRT than women in the no-alcohol conditions (216 versus 242 milliseconds; p = .02). There was no significant male contrast (reported p = .47). The investigators cautioned that the female difference may have reflected unusually prolonged stopping in the placebo condition; it should not be recast as proof that alcohol improves female driving reactions or increases public safety.
The study also did not find a general alcohol-related difference in ordinary go reaction time or stop rate. A nonsignificant test is not proof of identical performance across conditions, and a positive result in one task does not validate the other as a forensic impairment classifier.
Measured findings versus inferred interpretations
| Study quantity | What was actually measured or calculated? | What it does not establish |
|---|---|---|
| Administered dose | Weight-based oral ethanol target: men 0.55 g/kg, women 0.45 g/kg | An individual’s absorbed dose or venous BAC at a separate event |
| Breath alcohol | Instrument readings during the experiment; mean 0.39‰ across the two named task periods in the article’s notation | A directly analyzed whole-blood specimen, or an identical concentration at every task moment |
| Externally signaled stopping | Estimated SSRT calculated from behavioral response data and adaptive stop delay | Roadside coordination, driving safety, or a direct brain measurement |
| Voluntary task disengagement | Observed track-or-stop decisions represented by a mixed-effects survival model; alcohol exposure HR 0.68 (95% CI 0.61–0.76) | That 32% of all individuals failed to stop, or that a particular act was alcohol-caused |
| Alcohol expectancy | Experimental beverage-belief manipulation and ratings | A guarantee that placebo participants had no microscopic ethanol contact, or a field-validated deception test |
Selected results and their appropriate interpretations
| Comparison | Published finding | Scientific interpretation |
|---|---|---|
| Alcohol exposure and voluntary stopping | HR 0.68; 95% CI 0.61–0.76; p < .01 | Lower model-estimated rate of task disengagement during the 20-second choice window |
| Women: externally cued stopping | Mean SSRT 216 ms with alcohol versus 242 ms without alcohol; p = .02 | A subgroup task effect, not general proof of improved real-world reactions |
| Men: externally cued stopping | Reported p = .47 | No significant contrast detected in this task/sample; not proof of no possible effect |
| Alcohol group: alcoholic versus soft-drink bottle cues | HR 1.14; 95% CI 1.01–1.28; p = .03 | More voluntary stopping with alcohol-related images within the alcohol condition, contrary to a simple cue-impairment account |
| Breath measurements in the two task periods | Mean 0.39‰ for each; task-period difference p = .96 | Group averages do not mean every individual’s alcohol level or time course was constant |
Why the distinction matters in forensic testimony
When a witness says that alcohol impairs inhibition, the first question is which kind of inhibition, under what measurement procedure? A delayed response to a stop cue, a choice not to discontinue a rewarded activity, and a self-report of difficulty resisting a temptation are not interchangeable. The Liu experiment is a useful example because the two tasks yielded different patterns under the same overall experimental framework.
A scientifically bounded expert explanation may state that moderate oral alcohol exposure changed group-level intentional-stopping behavior in this reward-driven laboratory paradigm. It should not state, without independent evidence, that the study proves a defendant could not stop an action, that an intoxicated person necessarily lacked intent, or that a measured BAC mechanically determines behavior. A scientific association in a group is not a factual finding about a separate event.
Before translating task research into a case, identify the person’s specimen type and method, collection time, exposure history, other substances, the actual behavior alleged, competing explanations, and the time relation among those facts. The site’s separate guides on breath alcohol measurement and pharmacokinetic interpretation address why a concentration alone cannot supply all of that information.
What limitations require particular caution?
A young, selected sample and modest group sizes
Participants were mainly 18- to 25-year-old students without alcohol dependence and within the study’s screening criteria. The recruited N = 111, main N = 106, and additional five exclusions from the voluntary-stopping analysis are different denominators. This was not a clinical sample of people with severe intoxication, complex injuries, or polysubstance exposure.
Incentives, instructions, and task design
The chasing-bottles task rewarded continuing with immediate points and stopping with other point/lottery opportunities. The investigators acknowledged that incentive learning, design choices, and the arbitrary 20-second window could contribute to observed behavior. The model cannot fully separate failure to trigger a stop decision from difficulty executing one after it is initiated.
Alcohol exposure was not clamped
Breath results rose across the testing period, and task order required statistical adjustment. Sex-specific oral doses were intended to align exposure but did not eliminate biological variability. The external stop signal also used a specific visual format; task modality may matter. Only breath, not independently paired venous ethanol, was available for linking measured exposure to the task results in the published protocol.
No validated individual threshold or field endpoint
Neither a particular breath reading nor an SSRT or hazard ratio was validated in this paper as a cutoff for driving impairment, intent, incapacity, or causation. No road-driving outcomes, collision outcomes, field-sobriety assessments, or legal case determinations were measured. The work should be used to clarify mechanisms and competing inferences, not to substitute for actual case evidence.
Frequently asked forensic questions
Does alcohol always slow reaction time?
No. This experiment did not find general slowing on its externally signaled stopping measure. Effects depend on the task, sample, exposure, and statistical comparison. It does not follow that all reaction-time tasks are unaffected by alcohol.
What does a hazard ratio of 0.68 mean?
It represents a roughly 32% lower modeled instantaneous disengagement rate in the defined computer task, conditional on the model. It is not the percentage of intoxicated participants who committed an error, and it is not an individual risk estimate.
Did the study prove people became unable to stop?
No. Participants could and sometimes did disengage. The finding concerns how frequently and when they elected to stop while performing a rewarded task; it did not test absolute incapacity.
Did the study measure blood alcohol concentration directly?
No paired venous whole-blood ethanol assay is reported as the relevant concentration series. The investigators administered oral alcohol and monitored instrument breath values. Specimen and conversion distinctions matter; see the verified forensic toxicology evidence library for additional method-focused guides.
Was the placebo completely free of ethanol?
It contained tonic water as the beverage base but used alcohol-associated presentation cues, including vodka applied to the glass rim and a vodka-soaked lemon. It is therefore inaccurate to assert that absolutely no ethanol could contact the placebo preparation.
Does this research show that alcohol caused a specific criminal act?
No. The study has no individual criminal-case facts, no reconstruction of a particular act, and no validated legal intent classifier. The finding does not resolve volition, responsibility, or medical incapacity in an individual case.
Can a negative stop-signal test rule out other alcohol-related performance changes?
No. Voluntary decision-making, externally triggered stopping, coordination, attention, and other functions involve different tasks and evidentiary questions. A nonsignificant result in one assay does not establish intact functioning in every domain.
Primary evidence and bottom line
Primary original research: Liu Y, Grasman RPPP, Wiers RW, Ridderinkhof KR, van den Wildenberg WPM. Moderate acute alcohol use impairs intentional inhibition rather than stimulus-driven inhibition. Psychological Research. Published online May 19, 2020; 2021;85:1449–1461. https://doi.org/10.1007/s00426-020-01353-w.
Bottom line: The original experiment supports a distinction between group-level voluntary stopping and stopping in response to an external signal. A breath measurement establishes an instrument-reported concentration in the studied testing conditions; a derived SSRT or hazard ratio describes a particular model of task performance. None of them, separately or together, proves a particular person’s real-world impairment, subjective state, legal intent, or the cause of an event without additional independently evaluated evidence.