EVIDENCE LIBRARY
Can Moderate Alcohol Reduce Visual Memory Before Working Memory? A 154-Student Study
A 154-person study found different visual and working-memory results across alcoholmeter bands. Why breath readings and test centiles do not prove witness accuracy.
Short answer: In a laboratory study of 154 university students aged 18–19, immediate visual-memory scores were lower in adolescents with a history of binge drinking who had breath-instrument alcohol readings reported as 0.3–0.5 g/L, compared with alcohol-refraining controls. In contrast, working-memory scores did not show the same significant lower-dose group difference; significantly poorer working-memory performance was reported in the higher 0.54–1.1 g/L group. The study compared different participants and used an alcoholmeter, not a forensic analysis of drawn whole blood. It does not establish a universal concentration threshold for memory impairment, a witness’s reliability, or the cause of any particular event. The underlying original study is Vinader-Caerols and colleagues (2017), DOI 10.3389/fpsyg.2017.01720.
What did researchers actually investigate?
The central question was whether two different kinds of memory performed similarly as experimentally observed alcohol readings rose in 18–19-year-olds with a prior binge-drinking pattern. The first domain, immediate visual memory (IVM), involved recognizing faces and remembering scenes. The second, working memory (WM), involved ordering letter–number sets and reproducing visual–spatial sequences. These are not interchangeable cognitive functions. One test emphasizes recognition or recall of recently displayed visual material; the other requires actively maintaining and manipulating information over a short interval.
The authors were interested both in acute alcohol-associated test differences and in whether a prior binge-drinking history appeared to affect performance when no alcohol was detected by their instrument. They also examined sex-related group effects. Those are distinct research questions: a historical drinking pattern is a participant characteristic; an alcoholmeter reading is a contemporaneous instrumental observation; an assigned memory-test score is an observed test outcome.
The study’s most defensible public conclusion is task-dependent, group-level differences. It is not evidence of a validated impairment detector for all adolescents, drivers, witnesses, or people with an alcohol-use history.
Who participated, and how were the groups formed?
The final analyzed sample consisted of 154 University of Valencia students, including 67 male and 87 female participants, all aged 18–19. The paper reports that refrainers had never previously consumed alcohol. Participants classified as binge drinkers reported a prior-year pattern of consuming at least six Spanish 10-g standard drink units consecutively for males or five for females, at least three times monthly over the preceding 12 months. Consequently, the comparison groups were not simply random samples of the general population.
Some participants drank a flavored nonalcoholic refreshment; binge-drinking participants assigned alcohol drank vodka mixed with refreshment. The paper describes fixed-dose and weight-based dosing approaches, followed by measurement and sorting into groups according to the achieved alcoholmeter reading. The publication does not establish a fully randomized, blinded, placebo-controlled comparison in which every participant, including refrainers, was equally eligible for alcohol. That design distinction matters when drawing causal inferences.
Four result categories were analyzed for each sex: refrainers with a zero display (BAC0-R), prior binge drinkers with a zero display (BAC0-BD), prior binge drinkers in the reported 0.3–0.5 g/L band (BAC1), and prior binge drinkers in the 0.54–1.1 g/L band (BAC2). The original authors called these BAC groups. The description below preserves the published numerical bands while distinguishing the actual measuring device from drawn blood.
Were the reported alcohol concentrations directly measured in blood?
No venous whole-blood ethanol assay is reported in the paper’s Methods. The investigators used an Alcoquant 6020 alcoholmeter and designated its displayed values as “BAC.” They measured alcohol before and after beverage consumption, repeatedly during the waiting interval, and during the memory-test sequence. The group bands are therefore author-reported blood-alcohol-style values obtained using an alcoholmeter, not independent headspace gas-chromatography results from blood drawn and preserved for a forensic laboratory.
A reading of 0.3–0.5 g/L is a published grouping label. It should not silently be converted into proof of the participant’s independently measured forensic whole-blood concentration or into a laboratory’s own reporting unit. The testing protocol, instrument calibration and uncertainty, breath sample quality, elapsed time, and any device-specific blood-equivalence assumption are material to a case-specific interpretation. For the general measurement distinction, see breath-alcohol measurement principles and blood alcohol testing by headspace gas chromatography.
How did the study measure immediate visual memory and working memory?
Both outcomes were assessed with subscales of the Spanish-adapted Wechsler Memory Scale, third edition (WMS–III). For IVM, participants had to recognize faces and remember previously presented scenes. For WM, participants completed letter–number sequencing and visual–spatial sequence reproduction. The investigators then transformed scores to age-referenced centiles. The figures in their Table 2 are summarized group centile scores, not percentages of correct answers or memory-loss percentages.
Researchers also collected questionnaire information on consumption patterns and anxiety. The memory tasks were performed after the achieved alcohol-reading category was established, with further instrumental readings around the tests. Therefore, a single peak reading used for allocation should not be interpreted as a continuously maintained or precisely contemporaneous ethanol concentration at every cognitive observation.
This matters in litigation because there are several layers of measurement: the administered amount of vodka; the achieved alcoholmeter display; the task administration and scoring; and a subsequent retrospective inference about an individual’s memory for a real-world event. Only the first three were elements of the experiment, and the last was not validated.
What were the main observed results?
In the authors’ factorial analyses, the alcohol-category factor was statistically significant for IVM (F(3,146) = 9.354, p < .001) and WM (F(3,146) = 10.353, p < .001). The important scientific question is what the pairwise contrasts showed, not merely whether an omnibus statistical test was positive.
Immediate visual memory: the BAC1 group scored lower than refrainers with a zero reading (p < .05), and BAC2 scored lower than refrainers (p < .001). BAC2 also scored below the zero-reading binge-drinker group (p < .005). Refrainers and the zero-reading binge-drinker group did not have a statistically significant IVM difference in this sample. The authors did not report BAC1 as significantly below the zero-reading binge-drinker group in their primary pairwise account.
Working memory: the lower BAC1 category did not show significant differences from either zero-reading comparison group. The higher BAC2 group performed worse than both zero-reading groups (each p < .001) and below BAC1 (p < .05). Thus “IVM differed at the lower reported category whereas WM did not” is a description of these particular group contrasts—not proof that every person loses visual memory at that concentration while retaining working memory.
What do the actual group sizes and centile scores show?
| Published alcoholmeter category | Men (n) | Women (n) | Men: IVM mean ± SEM | Women: IVM mean ± SEM | Men: WM mean ± SEM | Women: WM mean ± SEM |
|---|---|---|---|---|---|---|
| Zero; never-drinking refrainers (BAC0-R) | 18 | 24 | 39.24 ± 6.57 | 55.62 ± 5.49 | 56.34 ± 5.74 | 40.41 ± 4.05 |
| Zero; binge-drinking history (BAC0-BD) | 23 | 27 | 44.73 ± 5.65 | 40.51 ± 5.43 | 51.37 ± 4.99 | 36.05 ± 3.26 |
| 0.3–0.5 g/L; binge-drinking history (BAC1) | 12 | 12 | 23.89 ± 7.26 | 31.14 ± 10.17 | 43.00 ± 6.59 | 35.46 ± 6.58 |
| 0.54–1.1 g/L; binge-drinking history (BAC2) | 14 | 24 | 10.89 ± 4.49 | 26.68 ± 4.60 | 23.80 ± 4.40 | 23.55 ± 3.47 |
Reproduced as an independently formatted factual summary from the study’s original Table 2; values are mean age-based centiles ± standard error of the mean (SEM), not independent BAC measurements, impairment probabilities, percentages of failures, or raw event-memory scores. Totals: 67 men + 87 women = 154 participants. Source: original DOI.
These descriptive numbers are useful for transparency but should not be used to invent exact sex-specific pairwise significance. The study’s relevant inferential tests operated on groups and factors; the sex-by-category interactions were not statistically significant. Nor does the table establish a natural monotonic dose-response curve within each individual: different students populate the alcohol categories, and the intervals themselves are broad.
Did prior binge drinking cause permanent memory deficits?
That conclusion is not supported by this study. The paper did not identify statistically significant differences between refrainers and prior binge drinkers when both groups had a zero alcoholmeter display. Its own Discussion cautions that the comparisons had low statistical power for isolating a long-term drinking-history effect. Absence of a significant difference is not proof of identical memory ability or absence of chronic neurological effects.
The design cannot separate all historical and preexisting differences between those who had never drunk and those who reported repeated episodes. Refrainers were not assigned the same alcohol treatment, and subjects were not followed longitudinally from before the onset of drinking. The study therefore does not determine whether an individual’s reduced test score preceded alcohol exposure, resulted from acute exposure, followed years of consumption, or reflected other characteristics.
The original authors discuss the possibility of compensatory neural recruitment during working-memory tasks, but this experiment did not directly image the brain, quantify recruited neural networks, or demonstrate a biological compensation mechanism. That explanation should remain a hypothesis, not a measurement or a claimed neurophysiological finding.
Did the experiment establish different alcohol effects in women and men?
Not for alcohol-category-related changes in the two memory domains. The authors reported a nonsignificant sex factor for IVM (F(1,146) = 3.847) and a nonsignificant IVM sex-by-category interaction (F(3,146) = 1.476). For WM, they reported a sex main effect (F(1,146) = 7.970, p < .01), but the WM sex-by-category interaction was not significant (F(3,146) = 1.254).
A statistically significant overall sex main effect on one test is not equivalent to showing that ethanol changes that test more strongly for one sex. And an interaction that is not statistically significant is not proof of exact equivalence. The sampled population, group sizes, body-dose protocol and individual variation all restrict extrapolation to a particular male or female person.
Forensic opinions should avoid substituting an assumption based solely on sex for the person’s actual measured performance, sampling circumstances, and relevant medical evidence. In addition, no sex-based finding here validates memory reliability, the accuracy of a witness’s recollection, or legal capacity.
Could these findings establish that a witness misremembered what happened?
No. Recognizing faces and scenes during a standardized task is not the same as encoding an unexpected altercation, retaining the details of an accident, accurately describing a perpetrator, or giving an eyewitness identification after later conversations. Test performance is relevant to understanding possible cognitive effects, but no person-specific witness was tested for real-event accuracy.
The paper reports means across categories, inferential group comparisons, and centile scores. It supplies no validated cutoff for classifying an individual’s memory as accurate versus inaccurate and no method for translating a particular alcoholmeter reading into a probability that a statement is false. A source that demonstrates a group difference also cannot decide whether a particular account is deliberate deception, honest memory error, impaired perception, or accurate recollection.
A scientifically bounded evaluation of witness memory would examine actual event chronology, timing of ethanol exposure, documented measurements and sampling matrix, the opportunity to perceive details, intervening information, contemporaneous descriptions, and known alternative explanations. This study informs one possible dimension—acute memory-test performance—without adjudicating any of those factual questions.
Why can a nonsignificant WM result at BAC1 not prove intact function?
A nonsignificant contrast says that the analysis did not establish the specified group difference under its data, test, and error criteria. It does not prove the effect was zero. The BAC1 group consisted of 12 men and 12 women, while the two zero-reading comparison groups were larger. The paper itself gives limited power for some contrasts. For WM, the estimated statistical power associated with the BAC1 comparison against refrainers in Table 2 is 0.386, whereas its higher BAC2 comparison was reported as 0.999. Those are study-specific power estimates rather than probabilities that a single participant was impaired.
Conversely, IVM’s lower-dose group finding is not a threshold measurement obtained by titrating each person’s blood alcohol upward until the first failure. There was no such subject-by-subject threshold procedure. Different subject groups, varied alcohol readings inside broad categories, and transformed scores limit what a concentration cutoff means.
The defensible interpretation is that the selected WMS–III IVM task appeared sensitive to lower alcoholmeter categories in these comparisons than the selected WM tasks. Whether this ordering holds in older persons, non-binge drinkers, different tasks, real-world memories, or clinical populations requires independent evidence.
How do exposure, assay, test score, impairment and causation differ?
| Evidence layer | What the study actually supplies | What cannot be concluded from that layer alone |
|---|---|---|
| Alcohol administration | Vodka-containing drink offered to selected prior binge drinkers, with described fixed/weight-based methods | Exact absorbed dose or historical concentration in another person |
| Alcohol analytical observation | Alcoquant alcoholmeter readings, reported as BAC-style g/L categories | Independently measured forensic whole-blood GC ethanol or confirmed identical blood:breathe conversion |
| Cognitive performance | WMS–III immediate visual and working-memory centiles after drinking/control | Identical effect on all memory systems, driving or field-sobriety tasks |
| Statistical result | Group-level omnibus and selected corrected pairwise differences; sex main effect in WM without a significant interaction | A universally valid personal threshold, diagnostic sensitivity/specificity or individual probability of error |
| Real-world event | No witnessed crime, collision, interview accuracy or longitudinal brain-injury endpoint was validated | Personal truthfulness, fault, legal incapacity or specific causal memory failure |
For understanding how concentration, distribution, and elapsed time interact, the site’s published pharmacokinetics overview provides context. That educational background does not replace the paper’s original measurements or clinical/legal case evidence.
What limitations should scientific and legal readers preserve?
Age and setting: all participants were 18–19-year-old students at one university, with specific health and drinking-history selection criteria. This was neither a representative national sample nor a real-world event study.
Selection and comparison: abstainers were not assigned ethanol, and the groups were partly defined by history and achieved instrument readings. A between-person group contrast cannot isolate every preexisting difference, unlike a complete randomized within-subject dose sequence.
Analytical matrix: the study’s alcohol variable came from an alcoholmeter. Direct drawn whole-blood ethanol was not described, so breath-to-blood equivalence and legal reporting cannot be assumed.
Timing: peak instrument reading defined categories, but memory tasks occupied a later interval with additional measurements. A category is not an exact continuous concentration at every response.
Endpoints: WMS–III centiles describe standardized tasks, not rates of failure to remember a collision or criminal act. There was no real-event truth criterion, clinical neurological diagnosis or measured brain activation.
Uncertainty: some cell sizes were small; the publication reported nonsignificant interactions and acknowledged limitations in detecting historical-drinking effects. Statistical significance and the scientific scope of a conclusion are separate matters.
Frequently asked questions
Does this study show visual memory is impaired before working memory?
It shows that lower visual-memory scores were detected in the lower BAC1 category versus refrainers, whereas corresponding working-memory contrasts at that category were not statistically significant. That is a task- and study-specific group finding, not a fixed biological sequence for every person.
How many participants were tested?
There were 154 analyzed late-adolescent students, including 67 men and 87 women. Subgroup totals are displayed in the evidence table; no single group consisted of all 154 subjects.
Was the alcohol concentration measured in a blood tube?
No direct venous whole-blood assay is described. The authors used an Alcoquant 6020 alcoholmeter and reported BAC-style concentration bands from its readings.
Did the lower category prove working memory was unimpaired?
No. No statistically significant lower-category working-memory contrast was found, but that does not establish zero effect or equivalence. Small subgroup sizes and power affect interpretation.
Did binge-drinking history alone cause a lower score?
The zero-reading binge-drinker and refrainer groups did not differ significantly in IVM or WM. The design was not sufficiently powered or controlled to prove that a drinking history had no long-term effect or caused one.
Can an expert use these numbers to decide whether an eyewitness was accurate?
No. The experiment did not validate a case-specific witness-memory accuracy test, a probability of false recollection, or a conversion from alcoholmeter result to truthfulness.
Did the study prove women and men react identically?
No. It reported no statistically significant alcohol-category-by-sex interactions in the memory endpoints, which does not prove biological identity or exact equivalence. It also found a working-memory sex main effect unrelated to a demonstrated differential alcohol interaction.
Primary research and interpretive bottom line
Vinader-Caerols C, Duque A, Montañés A, Monleón S. Blood Alcohol Concentration-Related Lower Performance in Immediate Visual Memory and Working Memory in Adolescent Binge Drinkers. Frontiers in Psychology. 2017;8:1720. doi:10.3389/fpsyg.2017.01720. This original primary study corresponds to the user’s complete Ethanol Papers Batch 05, item 1. The numerical tables and statistical contrast descriptions above are specific to that study; they are not independently validated forensic blood-alcohol or case-causation calculations.
Bottom line: in this selected adolescent sample, immediate visual memory appeared to differ across a lower alcoholmeter band than working memory, but there was no direct forensic blood assay, no within-person impairment threshold, no test of real-world eyewitness accuracy, and no demonstration that the reported group contrast determines an individual’s impairment or any legal event.