Quick answer: Yes. In a double-blind, counterbalanced crossover experiment involving 20 young adults, alcohol was associated with more eye fixations and saccades, a longer delay before the first fixation, and slower completion of a computer-screen visual maze than a placebo beverage. The researchers measured eye movement with a 300-Hz tracker and estimated alcohol exposure from a breath-testing device. This demonstrates differences in a defined laboratory visual-search task; it does not validate eye tracking as a stand-alone test of an individual’s blood alcohol concentration, ability to drive, eyewitness reliability, or cause of an accident.
The primary source is Silva and colleagues’ original 2017 PLOS ONE research article, “Effects of acute alcohol ingestion on eye movements and cognition: A double-blind, placebo-controlled study” (DOI: 10.1371/journal.pone.0186061). The findings below are taken from its methods, results, and original Table 2, rather than an abstract-only description.
What did the experiment actually establish?
When the same 20 participants completed a visual maze after alcohol rather than placebo, the investigators recorded an average of 40.60 versus 26.55 fixations and 40.15 versus 25.45 saccades. Average maze-completion time was 11,355 versus 8,526 milliseconds, respectively. The paired comparisons reported statistically significant differences for fixation count, saccade count, and task time. These are group outcomes under a particular stimulus and protocol, not cutoffs for determining whether one person was intoxicated.
The study adds useful evidence about how alcohol can influence the sequence of looking and searching. But the instrument did not measure whether participants could operate a vehicle safely, accurately perceive an unfolding collision, or recall an actual event. No diagnostic sensitivity, specificity, false-positive rate, or individual legal-impairment threshold was validated.
How was the placebo-controlled study designed?
Silva and coauthors enrolled 20 volunteers, comprising 11 men and nine women, ages 18–32. They reported normal or corrected vision and were screened for medical, neurological, psychiatric, medication, and substance-related exclusions. The study used a within-person design: each participant attended an alcohol session and a placebo session, with the order counterbalanced. Sessions were separated by a reported average interval of 20 days, with considerable variation.
In the alcohol session, volunteers drank vodka diluted with orange juice. The amount was calculated using anthropometric inputs to target the authors’ nominal 0.08% blood-alcohol-equivalent condition. The comparison drink was designed to appear and smell alcoholic, including a small quantity of vodka for masking, but the breath instrument displayed no detectable alcohol in the placebo condition. The researchers describe a two-researcher procedure: one prepared the beverages and collected alcohol readings, while another conducted the maze assessment.
Each participant started a maze at a central point, followed visual pathways, encountered two defined areas of interest where a direction was chosen, and sought an exit. The displayed maze was a specialized laboratory visual-search and problem-solving stimulus, not a roadside nystagmus test or actual driving task. The eye tracker recorded patterns of gaze; it did not read intentions or thoughts.
What was measured directly—and what was estimated?
| Evidence | How obtained | What it supports | What it does not establish |
|---|---|---|---|
| Eye position and movements | Binocular Tobii TX300 eye tracker, 300-Hz recording, with fixation classification software | Gaze-derived fixation and saccade metrics during a visual maze | Direct measurement of cognitive capacity, brain ethanol, or roadside fitness |
| Maze completion | Duration and scanning behavior while locating an on-screen exit | Performance in this specific laboratory visual-search task | Driving performance, legal impairment, or accident causation |
| Ethanol exposure | Digital BFD-50 breathalyzer at serial times, plus calculated oral dose | Breath-device readings expressed by the authors as BAC-equivalent percentages | Forensic headspace-GC measurement of venous whole-blood ethanol |
| Decision-point behavior | Fixations in two predefined maze areas of interest | Gaze distribution at the task’s route-choice locations | Whether a participant made a correct real-world decision or told the truth |
The paper repeatedly calls the breath-derived values “BAC,” but its methods identify a breathalyzer rather than a blood specimen or blood assay. A prescribed beverage dose and a breath-based estimate are also different things: the target was not an independently verified, fixed concentration in every participant during each instant of tracking. If a forensic dispute turns on a person’s actual blood ethanol, the collection and analytical records for that person’s specimen must be assessed separately.
What were the original eye-tracking results?
Original Table 2 gives the following descriptive means for all 20 participants. The task-time and gaze measures are presented in the authors’ reported units; rounding below does not turn these group statistics into personal thresholds.
| Outcome | Alcohol mean | Placebo mean | Authors’ paired-test finding |
|---|---|---|---|
| First fixation latency | 62.00 ms | 32.25 ms | t(19) = 2.806; p = .011 |
| Number of fixations | 40.60 | 26.55 | t(19) = 4.434; p = .001 |
| Mean fixation duration | 272.6 ms | 246.5 ms | t(19) = 3.245; p = .004 |
| Total fixation duration | 10,488.2 ms | 7,758.7 ms | t(19) = 2.418; p = .026 |
| Number of saccades | 40.15 | 25.45 | t(19) = 4.633; p = .001 |
| Mean saccade duration | 37.75 ms | 31.61 ms | Not statistically significant: p = .413 |
| Total saccade duration | 1,410.25 ms | 838.70 ms | t(19) = 2.374; p = .028 |
| Maze execution time | 11,355.05 ms | 8,525.60 ms | t(19) = 2.695; p = .014 |
A particularly important negative result is often overlooked: although the numerical mean duration of an individual saccade was larger after drinking, the mean-saccade-duration comparison was not statistically significant in this sample. It would be inaccurate to say that all eye-movement parameters demonstrated a confirmed alcohol effect. The authors also found more fixations in each of two route-choice areas of interest, but those measurements are still gaze counts, not an externally validated decision-quality score.
What do fixation and saccade measurements mean?
A fixation is an observed pattern of relatively stable gaze
A fixation allows information at a visual location to be sampled. Longer or more numerous fixations can reflect different search demands, stimulus complexity, strategy, fatigue, motivation, or experimental conditions. In this paired study, the alcohol condition yielded more fixations and longer average fixation duration on the same maze. That supports a task-specific alteration in visual exploration, while leaving open exactly which cognitive component produced it.
A saccade is a rapid change in gaze direction
Saccades move the eyes between locations of interest. More saccades need not mean an individual lacks control over their eyes; it can mean that more visual sampling moves were made while trying to solve the task. The study counted and timed these moves but did not validate a saccade-count cutoff for drunkenness. Saccade metrics in a maze are not interchangeable with horizontal gaze nystagmus observations.
Areas of interest describe where participants looked
The two marked decision locations were defined by the task’s designers. More fixations in those locations under alcohol may be compatible with less efficient route selection, but fixation count is an indirect proxy for processing demands. It does not independently identify the participant’s actual reasoning, accuracy on an unrelated task, ability to judge a real traffic event, or honesty as a witness.
Was the alcohol concentration actually 0.08% throughout the maze?
No fixed, directly assayed whole-blood concentration is demonstrated. The authors calculated beverage doses intended to produce approximately 0.08% and began tracking around the expected time. Their serial breathalyzer-derived group means were reported as about 0.09% at 20 minutes, 0.06% at 40 minutes, 0.03% at 60 minutes, and 0.014% at 90 minutes after administration. Those changing values alone rule out treating the entire experiment as a constant verified 0.08% venous-blood exposure.
A reviewer should also preserve the distinction between the timing of a device reading and the timing of the gaze recording. A nominal target or mean at one time point cannot supply the exact chemical concentration in the brain, venous blood, or breath of a named participant at every later movement. The paper is not an individual pharmacokinetic reconstruction.
Does more eye movement mean less reliable eyewitness testimony?
Not on this evidence alone. The experiment did not show a staged crime, ask subjects to describe people or objects from memory, score identification accuracy, compare accurate with inaccurate witnesses, or validate a truthfulness assessment. It measured how young adults moved their eyes while navigating a simple on-screen maze. Whether the same pattern predicts what someone actually saw at a collision scene would require separate witness-perception research and case facts.
Forensic analysis should separate exposure (the participant received alcohol), physiological or behavioral observation (gaze counts and task time changed), task performance (this visual maze took longer), and individual case conclusion (what a particular witness perceived or what caused a collision). The study informs the first three under its laboratory conditions but does not resolve the fourth.
Can this study validate an alcohol eye-tracking impairment test?
No. To validate a stand-alone diagnostic or forensic decision rule, researchers would need an explicitly defined intended use, reference standard, population, measurement protocol, error rates, thresholds, replication, and testing across relevant alternatives. This study’s paired statistical differences were not designed to estimate sensitivity or specificity for identifying alcohol exposure, a legal BAC threshold, or driving impairment in previously unseen individuals.
Nor did it test standard field sobriety procedures. The visual-maze tracker generated continuous gaze measurements under controlled screen-viewing conditions; those values cannot be applied numerically to another device, different stimulus, roadside lighting, head-position constraints, disease state, or medication exposure without separate evidence. A single person’s longer maze time might be compatible with many explanations other than alcohol.
What are the main methodological strengths and limitations?
The crossover design is a strength because each volunteer provides both an alcohol and a comparison observation, reducing some between-person variation. Counterbalancing the session order helps address systematic practice effects. A masked evaluator and objective gaze-recording equipment also strengthen the link between the administered condition and measured scanning behavior. Several related outcomes point toward more prolonged visual exploration under alcohol.
There are limits. Only 20 relatively young, screened adults participated. The study tested one specially designed maze under laboratory viewing conditions rather than multiple traffic scenarios. A breathalyzer estimated alcohol exposure instead of analyzing paired forensic whole-blood samples. There was no independently validated individual impairment classifier or driving outcome, and the published work reports numerous paired measurements and exploratory correlations without turning these into reliable individual predictions. The placebo included sensory masking, but the study does not establish how completely expectancy was eliminated.
Some printed statistics in the source require careful reading: for example, its narrative labels one sex comparison significant while printing a p value incompatible with that label. The numerical inference in this article therefore relies on the clearly reported paired eye-tracking outcomes and original table, not on that contradictory sex comparison. Replication, broader ages, independently assayed exposure, and a task with a validated forensic outcome would be necessary before extending these findings further.
What should a lawyer or forensic reviewer ask for?
Start with the proposition being offered. If someone cites this study to support that alcohol affects visual attention, its paired maze results are relevant as bounded laboratory evidence. If the argument instead claims that eye movements prove a driver’s legal BAC, unreliable testimony, or fault in a specific collision, request the missing bridge: a validated comparison population, task-to-outcome relationship, individualized measurement records, timing, and competing explanations.
Request the source’s original Table 2, figure definitions, the eye-tracker filter and calibration details, the actual gaze records if case-specific comparisons are contemplated, and a description of how the proposed expert inference would be tested. An analytical finding is different from a performance measure. For the separate question of reconstructing blood ethanol over time, see this site’s discussion of why two blood samples do not establish a universal elimination rate. The research library organizes additional measurement and interpretation questions, and the editorial source policy explains the status of literature-based appraisals.
Frequently asked questions
Did all 20 participants take both an alcohol and placebo test?
Yes. The reported within-subject design had 20 volunteers complete both sessions, with the order counterbalanced. It is not a comparison of an alcohol-only group against 20 different untreated people.
Did the researchers draw and analyze venous blood?
The methods describe using a BFD-50 digital breathalyzer for serial alcohol readings and expressing its results as BAC-equivalent percentages. No laboratory headspace-GC analysis of a drawn venous whole-blood specimen is documented as the exposure measure for these maze comparisons.
Did the study show longer mean saccade duration after alcohol?
The alcohol condition had a numerically longer group mean, 37.75 versus 31.61 milliseconds, but its reported paired comparison was not statistically significant (p = .413). That is not proof the effect is exactly zero in all populations; it is also not a demonstrated positive effect in this sample.
Were more fixations recorded at the maze’s choice points?
Yes. The authors reported more fixations in both predefined areas of interest after alcohol. These are counts of gaze events during a particular visual maze, not direct measures of the quality of every decision a person makes.
Can eye-tracking changes prove that a witness was mistaken?
No. Witness memory, identification accuracy, and truthfulness were not measured. At most, this study helps explain a possible task-specific effect of alcohol on visual scanning in young adults.
Does this article establish a roadside impairment cutoff?
No. The study supplied no validated classification threshold, sensitivity/specificity estimates, or comparison with roadside impairment or actual driving. Its group results should not be converted to a forensic pass/fail rule.
What is the most defensible conclusion?
Under a small, counterbalanced alcohol-placebo crossover protocol, alcohol changed several eye-tracking measures and lengthened maze-solving time. The findings support a limited causal inference about the experimental condition and the tested task, not a diagnosis or legal conclusion for an untested individual.
Primary source and reading boundary
Silva JBS, Cristino ED, Almeida NL, Medeiros PCB, Santos NA. Effects of acute alcohol ingestion on eye movements and cognition: A double-blind, placebo-controlled study. PLOS ONE. 2017;12(10):e0186061. DOI: 10.1371/journal.pone.0186061. This source provides experimental group evidence, not a case-specific toxicology report or an assessment of any particular person’s behavior.