Can alcohol slow a person’s stopping response even when visual-cue processing is not measurably disrupted?
Yes, in a particular laboratory task. In a randomized, double-blind experiment involving 39 analyzable social drinkers, participants assigned 0.4 or 0.8 grams of ethanol per kilogram of body weight had longer estimated stop-signal reaction times than placebo participants during testing. A companion visual control task and functional MRI comparisons did not show the same pattern of alcohol effects in brain regions identified as processing infrequent visual cues. The distinction is important: seeing or processing a signal and suppressing an action already underway are different functions. This experiment does not establish what an individual driver saw, when that person braked, or whether alcohol caused a particular collision.
The primary study is Nikolaou, Critchley and Duka (2013), Alcohol Affects Neuronal Substrates of Response Inhibition but Not of Perceptual Processing of Stimuli Signalling a Stop Response, PLOS ONE 8(9):e76649, DOI: 10.1371/journal.pone.0076649. This analysis relies on that original article’s methods, numerical results, tables and discussion, rather than on its abstract alone.
Why does the distinction between detecting and obeying a stop cue matter?
Responding to a changing situation involves several processes. A person must register the relevant stimulus, select what it means under the current instructions, modify an ongoing motor plan and successfully execute or suppress a response. A delay in the last steps is not equivalent to failure to perceive the stimulus. Conversely, evidence that early visual processing remains relatively intact does not demonstrate safe stopping, timely attention shifts or normal behavior under realistic conditions.
Consider an illuminated signal presented during a repetitive hand-response task. A participant may process its sudden change yet take longer to cancel the response that was already being prepared. The trial by Nikolaou and colleagues was designed to separate aspects of these operations. Its result should not be translated into a purported laboratory reconstruction of a particular driver’s brake-light perception or braking distance. Those quantities were never measured.
This mechanistic distinction also helps avoid a common evidentiary shortcut: assuming that a significant mean behavioral difference identifies precisely which person was impaired, or that a nonsignificant sensory contrast proves that perception was unaffected in every participant.
How were the 39 participants selected and assigned?
The investigators initially recruited 42 right-handed, English-speaking social drinkers from a university participant pool. Three datasets were excluded from analysis: one because acquisition failed technically and two because the participants never withheld a response during the scanning stop trials. The final analysis comprised 39 participants, 19 men and 20 women, divided equally into placebo, 0.4-g/kg ethanol and 0.8-g/kg ethanol groups (13 per group).
Allocation was randomized under double-blind conditions. All participants first completed a baseline and familiarization visit without the experimental alcohol challenge, followed by a separate beverage-administration and MRI visit. The assigned ethanol was mixed with tonic water into a 500-mL beverage; placebo contained tonic water with flavoring intended to approximate the drinks. The experimental visit was therefore a parallel-group comparison, not a design in which every participant personally received all three drinking conditions.
The modest group sizes and selected healthy-drinker population constrain extrapolation. The sample excluded several health conditions and current medications. Investigators also assessed drinking history, impulsivity and cognitive characteristics, but collecting baseline covariates does not make this a population-level validation of legal impairment.
Was blood alcohol directly measured?
No forensic venous whole-blood ethanol measurement was reported for the exposure comparisons. The investigators used a Lion Alcolmeter SD-400 to measure alcohol in exhaled breath, then converted those breath values into quantities expressed as blood alcohol concentration in grams per liter. The paper labels the converted values “BAC”; analytically, however, the instrument sampled breath, not a venous blood specimen. A breath-to-blood conversion is an estimate dependent on the chosen relationship between matrices.
Approximately 10 minutes after the drinking period, the authors reported group-average converted values of 0 g/L for placebo, 0.59 ± 0.13 g/L for the 0.4-g/kg group, and 1.16 ± 0.34 g/L for the 0.8-g/kg group. Five minutes after the scan, the corresponding values were 0, 0.35 ± 0.09 g/L and 0.99 ± 0.11 g/L. These are different measurement occasions and cannot be silently substituted for an incident-time venous whole-blood BAC.
For legal or analytical interpretation, distinguish a dose expressed in grams per kilogram, a directly measured breath concentration, its reported blood-equivalent conversion, and an actual laboratory whole-blood ethanol assay. Related questions of sample timing, elimination and reconstruction require additional observations and assumptions; see the site’s separate discussion of alcohol elimination from two blood samples.
What did participants do in the visual stop-signal task?
Participants repeatedly saw a green arrow and pressed a key corresponding to its direction. On a minority of trials, the green arrow changed to red, indicating that the participant should refrain from making the response. The principal scanning version contained 120 Go trials and 40 Stop trials. After a successful stopping attempt, the researchers lengthened the interval before the red signal by 50 milliseconds, making the next stop harder; after a failure, they shortened it by 50 milliseconds, making the next stop easier.
That adaptive staircase was designed to maintain stopping success near 50% while estimating a less-obvious quantity: the speed of the stopping process. The resulting stop-signal reaction time (SSRT) was computed by subtracting the mean signal delay from mean correct Go-response latency. SSRT was a model-derived behavioral task parameter; it was not the elapsed time between a real hazard and a driver’s actual foot movement.
Why not simply compare stopping accuracy? Because the algorithm intentionally adjusts task difficulty to make success rates converge. Similar proportions of successful stops therefore need not mean that inhibitory capacity was equal. A person whose stop process becomes slower can be given an earlier signal to reach approximately the same stopping accuracy.
What were the actual stop-signal reaction-time results?
The experiment reported a significant alcohol-group-by-session interaction for SSRT (F(2,36) = 3.34, p < .05). Baseline SSRT did not significantly differ across the three groups. During the beverage/scanning session, each alcohol group had longer average SSRT than placebo in the reported post hoc comparisons (t(24) > 2.47, p < .025). Although the high-dose group’s mean exceeded the low-dose group’s mean, the direct low-versus-high difference was not statistically significant (t(24) = 1.3). The descriptive means do not justify a precisely calibrated individual dose-response law.
| Assignment (n = 13 each) | Baseline SSRT, mean ± SD (ms) | Scanning-session SSRT, mean ± SD (ms) | Breath-derived converted value 10 min after drinking, mean ± SD (g/L) |
|---|---|---|---|
| Placebo | 257.21 ± 25.31 | 241.84 ± 36.82 | 0 |
| 0.4 g/kg ethanol | 273.66 ± 40.99 | 279.95 ± 41.44 | 0.59 ± 0.13 |
| 0.8 g/kg ethanol | 278.32 ± 48.55 | 300.86 ± 40.45 | 1.16 ± 0.34 |
Source: Nikolaou et al. (2013), Tables 1 and 3. SSRT is a computed task parameter. The last column is the study’s reported breath-derived blood-equivalent value, not directly assayed whole blood. These are group means, not individual predictions.
The scanning-session mean SSRT difference was approximately 38 milliseconds for low dose versus placebo and 59 milliseconds for high dose versus placebo. These subtraction results describe the group means in that experimental setting; they are not demonstrated additions to stopping time for a vehicle, and the experiment did not measure road speed, perception-to-braking time or stopping distance.
Why did the researchers also run a visual control task?
In the control variant, called VSST_C, participants again saw the green arrow and the same kind of occasional red cue, but were instructed to continue responding even when the color changed. The control included 60 Go trials and 20 cue-change trials. This was intended to test processing of a distinctive, infrequent visual signal without requiring cancellation of the motor response. The classic stop task always preceded this control task, a fixed sequence that can create order or practice effects.
Accuracy and response latency in that control variant did not show statistically significant group-by-trial interactions or group differences as reported. This supports a narrower interpretation: the selective behavioral deficit in the main stop task was not mirrored in the authors’ chosen overt control measures. It does not establish that every component of visual attention, color perception or perceptual decision making was invariant under alcohol. A null comparison in 13-person groups is not an equivalence test.
Nor was the control an eye test or a test of attention to hazards in a moving visual field. Both tasks used simplified, repetitive colored arrows under defined instructions. Those design choices are strengths for isolating candidate mechanisms, but limits for eyewitness, driving and collision reconstruction.
What did the functional MRI findings add?
The authors acquired functional magnetic resonance imaging using blood-oxygenation-level-dependent (BOLD) contrast. BOLD signals reflect coupled vascular and neural activity; they are not direct chemical measurements of ethanol in a brain region, proof of a damaged neuron, or a standalone clinical diagnosis. The investigators compared activation patterns during successful inhibition, unsuccessful inhibition and ordinary responses, and contrasted those regions with activation during the perceptual control task.
At the lower assigned ethanol dose, the principal stop-related analyses showed changes distributed across prefrontal, temporal, occipital and motor-related cortices. At the higher dose, reported changes extended to subcortical areas including the thalamus and globus pallidus. These are group-level imaging contrasts within this experimental paradigm, not maps that locate or quantify a given defendant’s neurological impairment.
The authors also analyzed the intersection of activation associated with a rare visual cue in the stop task and that cue in the continue-responding control task. In those conjunction-defined regions, the study did not find the same alcohol effect pattern. This is the evidentiary basis for the paper’s distinction between stop-network modulation and early processing of a salient cue. It is not a claim that all visual or attentional brain activity was unchanged: the authors discussed alcohol-associated activity differences in portions of the separate control-task analysis and possible attentional effects.
The results are consistent with a mechanism involving motor-response selection or inhibition rather than solely failure to register the infrequent visual change. As with all task-based BOLD contrasts, the analysis depends on acquisition settings, spatial statistical thresholds, subtraction contrasts and the sampled participants. Imaging findings are mechanistic context, not individualized forensic classification.
What is directly observed, calculated or merely inferred?
| Observation or claim | What was actually obtained | Appropriate forensic boundary |
|---|---|---|
| Ethanol exposure | Assigned oral ethanol dose of 0.4 or 0.8 g/kg; instrument sampled breath | Not a separately verified individual venous whole-blood alcohol result |
| Reported “BAC” | Breath values translated into g/L blood-equivalent numbers in the research report | A matrix conversion, not proof of a legal whole-blood concentration at another time |
| Stopping performance | Key presses, adaptive red-cue timing and derived SSRT | Laboratory hand-response inhibition, not a driver’s foot-brake latency |
| Visual cue processing | Control-task responses and fMRI cue-related contrasts | Narrow task results; nonsignificance is not proof that real-world perception is perfect |
| Brain activity | Group BOLD activation contrasts across experimental conditions | Indirect hemodynamic proxy, not cellular toxicity or a participant-specific diagnosis |
| Individual impairment or collision cause | Not directly measured | Requires incident-specific observations, scientifically relevant evidence and uncertainty assessment |
Separating these layers of evidence matters in deposition and at trial. One can accurately say that a randomized study measured a slower estimated stopping process under its alcohol conditions without claiming that a particular person failed to see a hazard or that a specific numerical vehicle delay follows from the group mean.
Can the experiment establish a driver saw a signal but failed to respond?
No. The study shows that, among these participants, alcohol assignment was associated with slower model-derived motor inhibition in a defined visual task while selected perceptual-control outcomes and conjunction-defined fMRI regions did not show comparable effects. The investigators did not put participants in cars, expose them to actual roadway hazards, examine traffic-camera footage, or quantify perception-response intervals in collision settings.
Translating a group SSRT contrast of tens of milliseconds into vehicle stopping distance would require assumptions about detection, expectation, motor coordination, pedal movement, road speed, friction and task generalization. None was empirically estimated in this experiment. A court-facing opinion should explicitly identify any external evidence used to bridge that gap, rather than present this study as direct validation for collision reconstruction.
There is also an evidentiary distinction between “the average group effect is statistically significant” and “this person was impaired at this event.” Within-group variation was substantial, the dose groups each contained only 13 people, and the researchers did not validate a person-level sensitivity, specificity or decision threshold.
What limitations should a forensic reviewer identify?
First, the sample was small, composed of selected healthy social drinkers, and excluded many medications and health conditions. Second, although beverage allocation was randomized and double blind, participants’ subjective awareness of intoxication could have limited effective blinding. Third, the classic stop task always preceded the continue-responding control, so task order was not independently counterbalanced. Fourth, testing occurred at a particular interval after the experimental drinks, with breath-derived exposure values changing between reported time points.
Fifth, the adaptive staircase intentionally targeted similar stop-success rates, making interpretation of stop-accuracy alone inappropriate. Sixth, SSRT is computed from observed reaction times and cue delays; its reliability depends on the assumptions of that computational task. Seventh, the BOLD analysis supplies indirect signals and involves choices about multiple spatial comparisons. Finally, overt perceptual-test null findings do not constitute a formal equivalence demonstration of intact vision or attention across all contexts.
The paper’s behavioral results also warrant careful reading of the original tables alongside the inferential text rather than isolating one group mean. Its account of nonsignificant Go-accuracy effects and the tabulated scanning accuracy percentages requires source-level checking before anyone draws a detailed accuracy-specific conclusion. None of these cautions erases the principal reported SSRT finding; they define its proper scope.
How should the research be used in an expert report or deposition?
The source is best employed to establish a mechanistic possibility backed by randomized laboratory observations: alcohol can interfere with the control process that cancels an action, and that interference may be distinguishable from some early sensory responses to the stop cue. It is not suitable as a substitute for the facts of a particular case, nor as a universal impairment test tied to a converted breath reading.
For a case-specific analysis, preserve the identity of the actual biological matrix tested, the method and uncertainty of any alcohol measurement, the relevant timeline, the person’s observed conduct, and alternative explanations. Distinguish any calculated values from measurements. Present opposing or null findings alongside significant outcomes. For research sourcing and citation methodology, consult the site’s forensic research resources and editorial and research source policy.
Frequently asked questions
1. Does a 50% stop-success rate show that alcohol had no inhibitory effect?
No. The protocol adjusted stop-cue timing by 50-millisecond steps to keep success near 50%. SSRT, rather than raw stop-success percentage alone, was the study’s key index of stopping speed.
2. Were the reported 0.59 and 1.16 g/L values measured directly in whole blood?
No. The research instrument measured exhaled breath and the authors reported values converted into blood-equivalent units. A direct forensic whole-blood sample assay was not described for these exposure measures.
3. Did the higher ethanol dose significantly worsen SSRT compared with the lower dose?
The high-dose mean was descriptively longer, but the direct low-versus-high SSRT comparison was not statistically significant. Do not represent the means as a validated quantitative dose-response curve.
4. Did alcohol leave every aspect of vision unchanged?
No such universal conclusion follows. Selected behavioral control measures and conjunction-defined cue-related brain regions did not show a corresponding alcohol effect. That does not exclude effects on other visual or attentional systems, and the authors discussed control-task regional changes.
5. Is fMRI evidence equivalent to measuring alcohol in brain tissue?
No. BOLD fMRI records a blood-oxygenation-related signal used to infer task-related brain activity. It is not an ethanol assay, proof of neural injury or an individualized diagnosis.
6. Can the study establish a minimum safe driving concentration?
No. It tested laboratory response cancellation, not actual driving or crashes, and did not derive person-specific legal thresholds. Its breath-derived exposure values are not direct legal whole-blood determinations.
7. Does a longer SSRT prove that an individual involved in an accident failed to brake because of alcohol?
No. That conclusion requires reliable event-specific information and a justified causal link. Group-average SSRT and MRI observations alone cannot identify an individual’s perceptions, motor response or accident cause.
Primary research and scope of this summary
Nikolaou K, Critchley H, Duka T. Alcohol Affects Neuronal Substrates of Response Inhibition but Not of Perceptual Processing of Stimuli Signalling a Stop Response. PLOS ONE. 2013;8(9):e76649. https://doi.org/10.1371/journal.pone.0076649.
This article is an evidence-based explanation of a published primary experiment. It does not offer a medical diagnosis, determine any specific person’s intoxication or provide an expert opinion about an identified legal proceeding. The paper’s exact measurements, procedures and statistical qualifications should be checked directly before use in expert testimony.