EVIDENCE LIBRARY
After a Four-Day Drinking Festival, Can Faster Attention Responses Hide More Errors? A 13-Person Pilot
A 13-person festival pilot found faster but less accurate Flankers responses at 0.00% breath alcohol. Examine sleep confounding, mixed tasks and forensic limits.
Quick answer: Yes, a small two-phase experiment found that people tested with a 0.00% breath-alcohol reading on the final morning of a four-day music festival made more errors on a selective-attention task than they had in a sober laboratory session, even though they responded faster. The study cannot establish that any individual was unfit to drive: only 13 festival participants were analyzed, no driving was observed, and drinking, sleep loss, fatigue, practice, and festival conditions were not experimentally separated. Its most useful forensic lesson is narrower: a negative displayed breath reading and a fast reaction time do not, by themselves, settle performance accuracy.
What scientific question did the festival study actually test?
Norman and colleagues published Assessing driving-relevant attentional impairment after a multiday drinking session: A two-phase pilot study in Alcoholism: Clinical and Experimental Research (2022). In the first phase, 52 young adults completed tablet-based attention tests at different points along a controlled alcohol-administration session. Thirteen of those participants furnished usable measurements on the final morning of a subsequent four-day Australian music festival. The researchers compared festival-morning performance with the same participants’ earlier laboratory performance.
The investigators were trying to build a field-testing method for driving-relevant attention after several days of drinking. They did not perform a randomized trial in which some people attended a festival while otherwise equivalent people remained at home. Nor did they conduct actual vehicle-road tests. The evidence is about a limited set of attention-task scores in a selected pilot sample, not verified crash risk.
Original primary study: Norman T, Monds LA, Dilevski N, et al. (2022). DOI: 10.1111/acer.14788.
What was measured directly, and what was not?
The alcohol-related instrument was an Andatech AlcoSense Prodigy S breathalyzer. The article reports instrument readings in a percentage convention, including 0.00%, approximately 0.05%, and 0.08% across the laboratory session. The measured sample was exhaled breath. The study did not draw, preserve, and analyze a venous whole-blood specimen by forensic headspace gas chromatography at those testing points. Accordingly, the reported percentages must not silently become independently measured forensic blood-alcohol concentrations.
The researchers also administered two direct behavioral tasks: an Arrow Flankers test of selective attention and a Rapid Visual Information Processing test of sustained attention. These are observed laboratory-style performances rather than chemical measures. They recorded self-reported sleep, drinking, hangover severity, and sleepiness. Those answers add context but are neither laboratory proof of consumption nor a validated individual explanation for a specific error.
For context on the distinction between an instrument’s breath reading and a chemical result from another biological specimen, see breath alcohol measurement and the broader pharmacokinetics interpretation guide.
How did the laboratory and festival phases differ?
During the controlled laboratory phase, each participant started with an instrument display of 0.00%, consumed a vodka-containing drink calculated to target a peak reading near 0.08%, and took the cognitive tests during the ascending and descending portions of the resulting alcohol curve. The mean readings were 0.056% at the nominal 0.05%-ascending assessment, 0.080% at the higher assessment, and approximately 0.050% during descent. The 0.08% label was a targeted assessment point: the protocol also allowed testing at an individual’s peak if that value was not attained.
The second phase was naturalistic. Fifteen laboratory participants attended a four-day music festival. One with a 0.07% festival-morning breath reading and another who reported illicit drug use were excluded, leaving 13 analyzed festival participants. On the final morning they completed the attention tests, a breath reading, and questionnaires in an onsite test area. Their usual festival drinking and sleeping behavior was not randomized or standardized.
On that final morning, the 13 retained participants had reported 0.00% on the breath instrument. This is a reported reading at that moment, not a demonstration of absolutely zero ethanol in every biological compartment at all prior times and not a certification of capacity to drive.
What are the key observed exposure and fatigue results?
| Measure | Original study result | What it represents |
|---|---|---|
| Laboratory cohort | 52 people | Controlled breath alcohol and tablet attention measurements |
| Festival cohort analyzed | 13 people, selected from 15 attendees | Paired final-morning field measurements after exclusions |
| Ascending laboratory assessment | Mean reported breath reading 0.056% (SD 0.006) | Nominal 0.05% ascending point, not sampled venous blood |
| Higher laboratory assessment | Mean reported breath reading 0.080% (SD 0.010) | Peak-target laboratory condition |
| Descending laboratory assessment | Mean reported breath reading 0.050% (SD 0.003) | Nominal 0.05% descending point |
| Final festival morning | 13 of 13 analyzed participants displayed 0.00% | Breathalyzer reading at the test, not a functional test |
| Reported drinks during festival’s 72 hours | Mean 23.5 (SD 9.1) standard drinks | Self-report; the study defined one standard drink as 10 g ethanol |
| Reported drinks in preceding 24 hours | Mean 10.6 (SD 5.5) standard drinks | Self-reported exposure, not verified intake or measured BAC |
| Sleep in prior 24 hours | 8.1 hours at laboratory visit; 6.5 hours at festival | Participant-reported sleep including naps |
| Karolinska sleepiness score | 3.0 laboratory; 5.6 festival | Subjective 1–9 sleepiness scale, higher means sleepier |
| Hangover severity score | 0.8 laboratory; 2.2 festival | Mean subjective rating using the study’s hangover scale |
The sleep, sleepiness, and hangover differences were statistically significant in the 13-person paired analysis: sleep p = .005, sleepiness p = .003, and hangover severity p = .001. They support a difference in the group setting and self-reported state. They do not partition how much of any attention error arose from residual alcohol effects, poor sleep, expectancy, fatigue, or practice.
Why did the Arrow Flankers task show a speed–accuracy trade-off?
In the Arrow Flankers task, participants identified the direction of a central arrow while surrounding symbols could support or conflict with the response. The 80 trials included congruent, incongruent, neutral, and response-suppression conditions. Researchers could inspect both reaction time and whether a response was correct. A faster mean response is not automatically a better response if more answers are wrong.
On the festival morning, the 13 participants made significantly more incorrect responses than during their earlier sober laboratory baseline (p = .013) and than at the laboratory’s 0.05%-ascending timepoint (p = .015). The comparisons with laboratory 0.08% and 0.05%-descending assessments were not statistically significant (p = .22 and .24). Meanwhile, reaction times for correct Flankers responses were faster during the festival test than at each laboratory assessment, including the sober baseline (p = .001), with p < .001 against the ascending and higher laboratory conditions and p = .043 against the descending condition.
These are within-participant statistical comparisons of tablet responses. They neither reconstruct a vehicle’s lane position nor calculate the probability of a crash. They show that an observer who looks only at response speed could overlook poorer accuracy on this particular test.
Did sustained attention also worsen after the festival?
Not in the same way. The Rapid Visual Information Processing task presented digits at short intervals and asked participants to identify target sequences. In the laboratory’s 52-person sample, the percentage of correct responses was worse at the 0.05%-ascending assessment than at sober baseline (p = .032), but not significantly different at the higher or descending assessments in that specific analysis.
In the 13-person festival comparison, sustained-attention correct-response performance was better on the festival morning than at the laboratory’s 0.05%-ascending (p = .022) and 0.08% (p = .030) assessments, with no significant difference from the laboratory sober baseline (p = .095). This mixed result matters: the study did not show generalized deterioration across every attention domain or a monotonic worsening with each higher breath reading.
The authors considered possible temporary compensatory effort during a short task. That is a hypothesis rather than something the experiment directly established. A brief 300-trial screen was not designed to quantify what happens across a prolonged drive home.
Which comparisons were statistically supported?
| Comparison or endpoint | Reported result | Appropriate inference |
|---|---|---|
| Flankers incorrect responses: festival vs sober laboratory | More errors at festival; p = .013; effect size d = −0.81 | Group difference in selected task accuracy |
| Flankers incorrect responses: festival vs laboratory 0.05% ascending | More errors at festival; p = .015; d = −0.789 | Same participants, different contexts and assessments |
| Flankers incorrect responses: festival vs laboratory 0.08% | p = .22; d = −0.36 | No statistically significant difference detected; not equivalence |
| Flankers correct-response speed: festival vs sober laboratory | Faster at festival; p = .001; d = 1.140 | Speed and accuracy can move in opposite directions |
| Flankers correct-response speed: festival vs laboratory 0.05% descending | Faster at festival; p = .043; d = 0.627 | Reaction-time contrast only; no road-driving endpoint |
| RVIP correct responses: festival vs laboratory 0.05% ascending | Better at festival; p = .022; d = 0.727 | Different attention endpoint from Flankers errors |
| RVIP correct responses: festival vs sober laboratory | p = .095; d = 0.502 | Difference did not meet the conventional .05 threshold |
| Laboratory RVIP accuracy: 0.05% ascending vs sober; n = 52 | Worse at ascending point; p = .032; d = 0.33 | Small task-specific acute-exposure association |
Values are selected from the study’s original Tables 1 and 2 and should be read within the authors’ repeated-measures design. Multiple outcomes and comparisons were explored in a pilot study. A small p-value is not the probability that alcohol caused a named person’s driving conduct; a nonsignificant value is not a certificate that performance was identical.
Does a 0.00% breath reading prove that someone was safe to drive?
No. A breathalyzer describes a chemical measurement under the instrument’s sampling, measurement, and reporting conditions. Fitness to drive involves attention, judgment, vigilance, psychomotor coordination, fatigue, sleep, medical factors, and real-world task demands, among other considerations. This study offers a concrete illustration of why the two questions must not be collapsed.
There is also a limitation in the opposite direction: the fact that some festival participants made more Flankers errors with a 0.00% displayed reading does not establish that all people with a negative breath test are impaired. The group was small, selected, and assessed in a particular naturalistic setting. The experiment did not define a forensic decision threshold for Flankers errors or establish its sensitivity and specificity for unsafe driving.
Further background on analytical-method scope and interpretation is available in the forensic toxicology evidence library.
Does the result show residual ethanol effects, sleep deprivation, or hangover caused the errors?
It cannot isolate any of those causes. Festival participants reported greater hangover symptoms, more sleepiness, and less sleep than during the earlier laboratory session. Alcohol consumption during the festival was self-directed and estimated by self-report. The sessions also differed in location, time, social activity, and repeated familiarity with the testing battery. The laboratory had no placebo beverage condition to separate pharmacologic effects from expectancy.
These exposures can plausibly matter, but the study did not independently manipulate or fully model them to identify one causal pathway. The investigators themselves noted that their sample was too small to reliably determine the association between subjective impairment and the attentional effects. Describing the result as “alcohol alone still impaired everyone after elimination” would exceed the research.
What makes this study relevant to a DUI, fatigue, or collision case?
It can help formulate the right questions. First, what was actually tested—a direct breath sample, preserved whole blood, an attention task, a clinical examination, or a driving event? Second, what was the instrument’s purpose and reporting threshold? Third, did any purported performance evaluation measure errors, speed, both, or actual vehicle control? Fourth, what were the person’s relevant sleep and activity circumstances at the time in issue?
For a lawyer or forensic expert, a fast response time without an error-rate review is a potential interpretation defect. Likewise, a “negative alcohol” label cannot be elevated automatically to a full neurobehavioral clearance. But a postfestival group effect also cannot replace case facts: verified measurement records, timing, observations, alternative exposures, vehicle data, witness accounts, fatigue evidence, and the precise legal question still control individual analysis.
The article does not establish a statutory cutoff, a permissible forensic adjustment to any breath or blood test, a quantified crash-risk multiplier, or an expert-certified conclusion about any identified driver’s state. Those would require materially different data and validation.
What study-design limitations most affect forensic generalization?
Small field denominator: only 13 people remained in the festival analysis. Detecting a difference in this group is not the same as deriving population-wide error rates or identifying individual impairment.
Naturalistic, uncontrolled exposure: the field period included unknown combinations of drinking behavior, sleep restriction, food, exertion, and possible fatigue. Retrospective drink counts and sleep were self-reported.
No placebo and nonrandomized festival attendance: there was no equivalent control group placed into otherwise comparable festival and nonfestival conditions. Practice and assessment-order effects are difficult to disentangle from residual consequences of the event.
Task sensitivity varied: the laboratory battery did not uniformly detect impairment across the rising and falling alcohol curve. The field Flankers finding and the field sustained-attention finding pointed in different directions.
Proxy endpoints: no vehicle was driven, no roadside stop or crash was reproduced, and no evidential whole-blood ethanol specimen was assayed. The results must remain attention-task observations rather than vehicle-operation or causation measurements.
Scope of the population: participants were young, screened for specific health, tobacco, and substance-use criteria, and the festival group was selected from a laboratory sample. Generalization to older drivers, medical comorbidity, other substances, or different environments is uncertain.
Frequently asked questions
Can a person make more attention errors after drinking even when a breathalyzer shows 0.00%?
Yes, this 13-person field pilot observed more errors on one selective-attention task after a four-day festival despite 0.00% reported breath readings. It did not establish that every participant, or every future driver, was impaired.
Did the study measure real driving, lane keeping, or crashes?
No. It used tablet-based Flankers and sustained-attention tests. Neither road driving nor simulator driving was measured in this experiment.
Was blood alcohol independently measured by a forensic laboratory?
No. Exposure levels were assessed through the Andatech breath instrument and reported in the study’s percentage convention. The article does not provide paired forensic venous whole-blood GC results for these comparisons.
Did all 52 laboratory participants attend the festival?
No. Fifteen attended the festival; two were excluded from field analysis, leaving 13. The controlled laboratory comparisons used 52 participants, while field comparisons used the selected subset.
Is a faster reaction time proof of better cognitive performance?
No. In this study, participants responded faster on the festival morning but also made more Flankers errors than during sober laboratory baseline testing. Speed and accuracy must be analyzed together.
Was sustained attention consistently worse on the morning after the festival?
No. RVIP correct-response performance was not significantly worse than sober laboratory baseline and exceeded some intoxicated laboratory comparison points. The findings differed across tasks.
Can this study establish why a particular collision occurred?
No. It provides preliminary group-level task findings and questions for evaluating breath, fatigue, and cognitive evidence. Individual crash causation requires separate evidence of the person, timeline, conduct, relevant alternatives, and event.
Bottom line: keep the analytical result and functional conclusion separate
Norman and colleagues’ two-phase pilot raises a useful caution: 0.00% on a breathalyzer is a reported chemical result, while selective-attention accuracy is a separate functional measurement. Thirteen participants showed more Flankers errors and faster responses after a multiday festival, but the sustained-attention findings were mixed, the field sample was very small, and the source of the differences could not be isolated. This paper supports careful testing and qualified testimony—not a blanket rule that negative breath alcohol guarantees safe driving or that a postfestival error proves alcohol-caused impairment.