EVIDENCE LIBRARY
Can Passing One Cognitive Test Exclude Alcohol Impairment? A Six-Test Dose Study
A 56-person dose-response study found alcohol affected cognitive tests differently at ~0.048% breath-reported BAC. Learn what normal scores cannot prove.
No. Passing one attention, memory, or psychomotor test does not establish that alcohol has left every cognitive function unaffected. In a controlled 56-person study comparing six cognitive tasks, different abilities responded differently as alcohol exposure increased. At the investigators’ lowest alcohol testing point, approximately 0.048% reported BAC, working-memory performance, visual inspection time, and errors of response inhibition showed significant alcohol–control differences; several other measures did not. The absence of a significant difference on one test did not erase differences on the others.
This is the central forensic lesson from Dry and colleagues’ 2012 primary experiment. It is evidence of group-level, task-dependent cognitive effects, not a validated method of declaring any particular defendant impaired or unimpaired from a single test score. The paper’s alcohol readings came from a breath instrument and were expressed as BAC values; it was not an evidential headspace gas-chromatography study of separately collected venous whole blood. Read the original study (DOI: 10.1371/journal.pone.0050977).
What question did the six-test alcohol study actually answer?
The investigators asked whether the effects of alcohol rose uniformly across distinct cognitive processes and whether one convenient test might serve as a sensitive index over several progressively higher alcohol exposure points. The answer to the first question was no: task sensitivity and the shape of dose-related changes differed. The investigators favored a brief inspection-time test as a promising research measure because it registered a graded change without relying on fast hand or finger movements. They did not validate it as an independent diagnostic test for intoxication, legal impairment, or crash causation.
The distinction matters in testimony. A failed laboratory task is not synonymous with unsafe real-world performance in every setting; conversely, a passed task does not guarantee that every other safety-critical ability is preserved. Questions about a specific driver require case-specific observations, testing methods, timing, and uncertainty.
How were the participants assigned and exposed to alcohol?
The study enrolled 56 adults aged 18–45: 32 men and 24 women. Participants were assigned to an alcohol group or a control group, 28 per group. Eligibility required recent experience drinking at least five alcoholic beverages on an occasion and excluded certain disorders, visual conditions, drug dependence, pregnancy, and medications expected to stimulate or sedate. The sample therefore does not represent all drivers, older adults, patients taking medications, or people with every drinking pattern.
There was a familiarization day with several practice administrations of the cognitive battery. On the test day, the alcohol group received three incrementally administered portions of vodka mixed with sugar-free orange juice. The control group received juice alone. Each set of tasks began around baseline and then at approximately 60, 120, and 180 minutes. The reported mean/target alcohol points were approximately 0.00%, 0.048%, 0.082%, and 0.100%. Individual readings varied around those group values.
Doses were calculated using a Widmark-based formula to reach successive targets. A Lion Alcometer 500 breath device measured alcohol before individual tasks. Thus the reported “BAC” values are breath-instrument-derived study readings rather than contemporaneous separate measurements of whole-blood ethanol by laboratory headspace gas chromatography. The control participants were told their assigned condition and received juice, not a deceptive alcohol placebo; expectancy and blinding limitations follow. The rising-dose schedule also means time/order and accumulating alcohol exposure changed together.
Which cognitive tests were compared?
The investigators deliberately avoided treating cognition as one interchangeable quantity. Six named tests supplied eight scored endpoints across perceptual, memory, attentional, inhibitory, strategic, and psychomotor processes.
| Task | What it probes | What the study found at the approximately 0.048% point |
|---|---|---|
| Inspection Time (IT) | Speed of visual discrimination without requiring a speeded motor response | Significant alcohol–control difference; response continued to change across higher exposure points. |
| Self-Ordered Pointing Task (SOPT) | Visual working memory and monitoring already selected locations | Significant difference; the authors described a graded dose relation. |
| Sustained Attention to Response Task — commission errors | Failure to withhold a habitual response when required | Significant difference; most of the increase appeared by the first alcohol point, with limited further increase. |
| Useful Field of View (UFOV) | Divided visual attention and processing under a search-like task | No statistically significant alcohol–control difference established at that first point; differences occurred at higher exposure. |
| Trail-Making Test A and B (TMT) | Speeded visual sequencing, switching and psychomotor flexibility | No significant first-point difference established; study-level alcohol effects emerged across the experiment. |
| Traveling Salesperson Problem (TSP) | Strategic route optimization and multistep problem solving | No significant overall alcohol main effect in this experiment. |
| Sustained Attention to Response Task — omission errors | Failures to respond to frequent go stimuli | No significant overall alcohol main effect in this experiment. |
Interpretation note: A nonsignificant result is not proof that the process was unaffected or that a participant is sober. The authors used baseline adjustment and corrections for repeated comparisons. These statements describe the study’s statistical results, not an individual diagnostic cutoff.
Was cognition already affected at a reported 0.048% alcohol level?
In this sample and protocol, three of the eight scored endpoints showed significant alcohol–control differences at the first post-drink testing point, approximately 0.048%: inspection time, self-ordered pointing, and SART commission errors. The article reports effect-size estimates at that point of approximately d = 0.65 for inspection time, 0.76 for working memory, and 1.06 for response-inhibition errors. These are standardized group contrasts; they are not probabilities that a particular individual will fail, and they cannot be translated directly into a forensic BAC margin or an individual level of disability.
The findings also do not mean everyone at 0.048% is identically impaired. Nor do they demonstrate a universal threshold at which impairment begins. The experiment evaluated a limited range of exposure levels in experienced drinkers and was not designed to establish person-specific impairment or the legal significance of a particular reading.
Why did some tasks change steadily while others did not?
The researchers highlighted two distinct patterns. Inspection time and working memory worsened in a more graded fashion as the alcohol readings increased. By contrast, SART commission errors rose early but showed relatively little additional worsening as the study moved toward approximately 0.10%. The authors discussed a possible task-related ceiling limitation: when relatively few response-withholding trials are available, an error measure may stop discriminating well among higher exposure points.
This is why one cognitive score should not be substituted for another without validating what both measure. Inspection time depends on a brief visual judgment; trail making adds manual speed and switching; strategic problem solving has a different time scale and error structure; commission and omission errors in the same sustained-attention task index different events. A normal result on an insensitive or ceiling-limited task does not falsify a deficit detected by a different method.
What did the principal study comparisons show?
| Evidence element | Direct study observation | Boundary of the inference |
|---|---|---|
| Participants and groups | 56 adults; 28 allocated to alcohol and 28 to control; repeated performance evaluations after practice. | Not an individual impairment reference population; control assignment was not concealed with an alcohol-tasting placebo. |
| Alcohol exposure | Successive target/attained average readings near 0.048%, 0.082%, 0.100%; breath-instrument checks during testing. | Not independently determined venous whole-blood BAC at each task; exposure increments and elapsed time were linked. |
| Lowest alcohol point | Working memory, inspection time and SART commission errors differentiated groups. | Significance/effect size describes groups; not a threshold for a particular driver’s competence. |
| Higher levels | Most other endpoints showed alcohol-related changes, but TSP and SART omission errors lacked significant overall main effects. | Failure to reject a null hypothesis is not equivalence or immunity to ethanol. |
| Study end point | Laboratory task scores in a controlled rising-dose experiment. | No actual driving, collision rate, field sobriety validation, individually proven BAC at an earlier event, or cause-of-injury measure. |
Was inspection time validated as a stand-alone alcohol impairment test?
No. The investigators argued that inspection time was promising because it differentiated the groups at the lowest studied post-drink point, changed with dose, and can be administered without speeded manual responding. They also noted its comparative convenience, potentially requiring about five minutes rather than the longer working-memory protocol used here. That is a research recommendation, not prospective validation for determining the cause of a crash or for classifying an individual as legally impaired.
A forensic validation claim would require additional evidence about individual test–retest reliability at relevant exposure levels, diagnostic operating characteristics, demographic and clinical variation, alternative causes of poor performance, comparison with appropriate reference outcomes and the actual circumstances in which the test is administered. This study did not supply such a case-classification rule.
Why does the difference between breath readings and blood BAC matter here?
The authors used the term “BAC” and described target percentages, but reported using a Lion Alcometer 500 to obtain readings during the cognitive battery. Do not conflate an instrument’s breath-derived reported BAC scale with independently analyzed venous whole-blood ethanol, serum or plasma ethanol, or an assumed blood–breath partition. Likewise, the Widmark calculation was a dosing estimate, not a direct laboratory measurement. A study’s group average exposure cannot independently establish the concentration or change at a defendant’s earlier event.
For specimen-method context, see the site’s breath-alcohol measurement guide and pharmacokinetics overview. These links provide background, not additional experimental support for this specific six-test result.
Could the study show impairment on the descending side of the alcohol curve?
No. The experiments evaluated increasing readings after repeated alcohol portions. They did not compare cognition at equal readings on the ascending and descending limbs or determine the size of acute functional tolerance. An expert therefore cannot use these data to assert that an individual would score identically during alcohol absorption and elimination.
Repeated task practice also matters. Participants trained on the battery before testing, yet learning, fatigue, expectancy, unequal group awareness and test-order effects may still influence outcomes. The investigators applied a baseline-regression adjustment and checks for individual recorded alcohol levels, but those statistical choices do not convert the design into a randomized, blinded equal-BAC ascending-versus-descending crossover.
What are the most defensible courtroom uses—and misuses?
A defensible use is to explain that alcohol’s group-level effects are multidimensional: different standardized research tasks can show differences when another task does not, and responsiveness changes over an exposure range. That scientific principle is relevant when a witness claims that one normal cognitive or roadside observation proves a complete absence of ethanol-related functional effects.
An indefensible use would be to announce that an accused person was impaired simply because their reported breath reading matched a study group mean, to convert standardized group effect sizes into individual impairment probabilities, or to claim that a single omission-free or strategic-problem-solving score exonerates all other functions. Nor can this paper assign responsibility for any particular collision, infer the drinking timeline of a stranger, or establish a jurisdiction’s legal DUI threshold.
The case-specific assessment remains grounded in the actual analytically supported specimen, collection time and assay; documented behavior and observations; relevant competing causes of performance changes; and the limits of every inferential bridge. For additional research navigation, see the site’s forensic toxicology evidence library.
Frequently asked questions about alcohol and cognitive testing
Can a person pass one cognitive test while alcohol affects a different ability?
Yes. In Dry and colleagues’ controlled sample, sensitivity differed among six tasks. A nonsignificant result on one endpoint could coexist with a significant group difference on another. Neither outcome establishes a particular individual’s impairment.
Which tests showed group effects around 0.048%?
Inspection time, the self-ordered pointing working-memory task, and SART commission errors showed significant alcohol–control differences at the first post-drink point. The percentage represents the study’s breath-instrument-reported exposure scale, not a separate forensic whole-blood assay.
Did all eight measured endpoints worsen significantly?
No. The strategic Traveling Salesperson Problem and SART omission-error outcomes did not show significant overall alcohol main effects. That is an evidentiary limit of this experiment, not proof those functions are biologically unaffected by ethanol.
Does a group effect size tell how impaired a particular driver was?
No. Standardized group effect sizes characterize between-group score differences in this protocol; they cannot identify one person’s BAC, safe-driving capacity, or accident causation without additional evidence.
Was this a test of real driving?
No. The outcome was performance on laboratory cognitive tasks. The experiment contained no real-road drive, validated collision-risk endpoint or crash investigation.
Does the paper support the same conclusions during falling BAC?
No. The researchers tested increasing alcohol readings after successive drinks. Their results do not establish equal effects during elimination, nor quantify a person’s acute tolerance.
Was inspection time approved as a forensic sobriety test?
No. The investigators identified inspection time as a promising research measure, not a stand-alone legally validated impairment classifier or substitute for an evidential ethanol analysis.
Primary scientific source
Dry MJ, Burns NR, Nettelbeck T, Farquharson AL, White JM. Dose-Related Effects of Alcohol on Cognitive Functioning. PLOS ONE. 2012;7(11):e50977. doi:10.1371/journal.pone.0050977. This analysis draws on that complete original study, not on a secondary review. It explains the experiment’s group statistics and limits; it does not offer an individual medical, toxicological, or legal determination.