EVIDENCE LIBRARY
Can Brain MRI Show How Drunk Someone Feels? What an Alcohol-Challenge Study Can—and Cannot—Establish
A study of 89 social drinkers linked MRI gray-matter density with self-rated alcohol effects. Why those findings cannot determine BAC, driving impairment or causation.
Can a brain MRI show whether someone felt intoxicated after drinking alcohol? Not from an MRI alone. In a controlled study of 89 healthy social drinkers, regional gray-matter-density estimates from structural MRI were statistically associated with how intoxicated participants said they felt on the rising limb of an alcohol challenge. Different regional associations appeared for reports of calmness or mellowness later in the session. These were group-level associations between an imaging-derived structural measure and subjective ratings—not a validated MRI test for an individual’s blood alcohol concentration (BAC), psychomotor impairment, driving fitness, or the cause of an incident.
The key source is the complete primary research paper by Stennett-Blackmon, Sevel, and Boissoneault (2023), Scientific Reports, DOI 10.1038/s41598-023-34546-5. Its actual protocol, time-point measurements, voxel-based morphometry findings, placebo comparisons, and limitations matter more than a headline about a “brain scan for drunkenness.”
What did the alcohol-and-MRI researchers actually investigate?
The researchers asked whether differences in gray matter density (GMD) were associated with participants’ own reports of subjective intoxication and subjective alcohol response. Their sample comprised 89 healthy social drinkers, including 55 women, with a mean age of 26.2 years. Participants attended counterbalanced, double-blind alcohol and placebo sessions separated by at least 48 hours. The alcohol beverage was designed to approach a reported target concentration of 0.08 g/dL; the placebo was soda presented with alcohol cues, including a small amount of alcohol on the glass rim and drink surface.
Alcohol exposure was monitored through breath-alcohol measurements before and after the MRI scan and salivary alcohol measurements before, during, and after scanning. The researchers did not report paired venous whole-blood ethanol measurements for this analysis. They also obtained structural T1-weighted brain images with a 3-T scanner and analyzed the images using voxel-based morphometry. That process generates statistical estimates of tissue characteristics across image voxels; it is not a microscopic assay of brain tissue.
Participants rated their own degree of intoxication on a 100-mm visual analog scale and rated different feelings using the Subjective Effects of Alcohol Scale (SEAS). SEAS separates high-positive, high-negative, low-positive, and low-negative subjective experiences. For example, feeling calm or mellow is a low-positive response. Those ratings are not interchangeable with observed balance, reaction time, divided attention, simulated driving, or a clinical examination.
What did the study directly measure, and what was inferred?
| Evidence component | What the study obtained | What it cannot establish by itself |
|---|---|---|
| Breath and saliva alcohol | Instrument readings from breath and saliva at protocol-specific times. | A directly analyzed venous whole-blood ethanol value, a universal breath-to-blood conversion, or alcohol-related functional impairment. |
| Subjective intoxication | Each participant’s self-rating on a 100-mm scale. | An objective measure of coordination, attention, driving ability, or actual legal impairment. |
| Subjective response | SEAS ratings of positive/negative and stimulating/sedating feelings. | That the participant performed well or poorly in a motor or cognitive task. |
| Structural MRI | T1 images analyzed to estimate regional gray matter density across groups. | Direct cell counts, acute tissue damage from the session, or an individual intoxication diagnosis. |
| Statistical regression | Associations between image-derived density and subjective ratings, with specified covariates and corrections. | That brain anatomy caused a particular report, that alcohol altered the anatomy during the session, or a validated individual prediction. |
This separation matters in forensic review. A breath instrument reading is not a histological result; a self-report is not a laboratory BAC; a voxel-based regression coefficient is not a measured dose; and none of these is an independently validated test of real-world driving performance. For more on what a breath instrument does and does not measure, see the site’s breath-alcohol measurement guide.
Did people feel less intoxicated later in the drinking session?
On the study’s designated early (ascending) and later (descending) assessment occasions, average subjective intoxication ratings were lower later. During the alcohol session, the mean early rating was 37.96 (SD 20.19) and the mean later rating was 29.76 (SD 18.75); the within-person difference was reported as p = 0.0005. Placebo ratings were also lower later, from 9.01 (SD 9.87) to 5.23 (SD 7.90), p = 0.0008. These results concern perceived intoxication on the specific rating scale, not demonstrably improved task performance.
The underlying concentration measurements deserve special care. The investigators reported mean breath-alcohol readings of 0.053 g/dL at 10 minutes after administration and 0.060 g/dL at 20 minutes, while the post-scan mean was 0.062 g/dL. The post-scan sampling occurred on average approximately 87 minutes after drinking; saliva measurements taken throughout scanning helped characterize the time course. Accordingly, readers should not describe the group comparison as if identical contemporaneous breath BAC values had been documented at the two subjective-rating occasions. The authors labeled the occasions as limbs of the concentration curve, but the group means are not a matched-concentration experiment for each individual.
That qualification matters when invoking acute tolerance. A lower self-rating later in a session can be consistent with changing subjective response, but it does not show that coordination, divided attention, or driving risk fell proportionately. Nor does it alone establish a specific BAC at an unobserved time. More generally, see the discussion of concentration-time relationships in the site’s pharmacokinetics guide.
Which MRI regions were associated with subjective alcohol effects?
Whole-brain analyses found that greater self-reported intoxication on the designated ascending limb was associated with greater estimated gray matter density in a cerebellar cluster. No corresponding statistically significant whole-brain association of subjective intoxication with gray matter density was reported for the descending occasion. Separately, greater low-positive SEAS ratings on the descending occasion were associated with regional gray matter density in the cerebellum and in a cluster spanning precentral/postcentral and related cortical territory. Those low-positive ratings capture sensations such as calmness and mellowness, not measured motor precision.
The voxel analyses adjusted for age, sex, and total intracranial volume; cluster-level family-wise-error correction was part of the specified imaging analysis. Additional hierarchical regressions used placebo scores and model covariates to examine whether alcohol-specific subjective ratings accounted for incremental variance in the image-derived regional measure. An incremental variance statistic is a statistical model property, not the percentage of gray matter created or destroyed by alcohol.
| Reported finding | Selected study result | Forensic interpretation boundary |
|---|---|---|
| Subjective intoxication, early versus later alcohol session | 37.96 (SD 20.19) versus 29.76 (SD 18.75); p = 0.0005. | A difference in self-reported feeling; not a demonstrated difference in actual driving or motor impairment. |
| Ascending-occasion intoxication and MRI gray matter | Positive association in a cerebellar cluster; adding alcohol-specific subjective ratings increased explained variance by ΔR² = 0.109 in follow-up regression. | Association conditional on selected model and sample; not a diagnostic prediction for a person. |
| Descending-occasion low-positive response and MRI gray matter | Associated cerebellar and pre/postcentral regions; alcohol-specific scores produced ΔR² = 0.052 and 0.090 in the respective follow-up models. | Calm/mellow subjective-response ratings, not proof of preserved coordination or task performance. |
| Measures without the highlighted imaging associations | The reported significant voxel-wise patterns were limited to specified subjective-rating dimensions and occasion. | Do not generalize the identified clusters to every subjective response, all brain regions, or an individual impairment classifier. |
The location of an association does not transform the experiment into proof that a particular region generated a reported feeling. It also does not show an acute change in brain structure from that day’s alcohol dose. Imaging-based gray matter density varies with preprocessing, anatomy, and statistical choices; it is not a direct observation of active alcohol molecules or neural firing.
Does a cerebellar association prove alcohol-related balance or coordination impairment?
No. The cerebellum participates in coordination and many other functions, making its involvement scientifically relevant. But in this research, the key outcome linked to cerebellar gray matter density was a reported feeling of intoxication or positive low-arousal response. Researchers did not validate an MRI threshold against horizontal gaze nystagmus, postural sway, field sobriety tests, instrumented gait, divided-attention tasks, or crash outcomes in this experiment.
It would therefore be erroneous to substitute a group association of cerebellar density with subjective intoxication for a demonstrated participant-level motor deficit. It would be equally erroneous to treat the absence of a significant association in another region as proof of unimpaired behavior. Different measures can answer different questions. A case involving suspected impairment requires its own contemporaneous behavioral, analytical, contextual, and methodological evidence.
Could lower reported intoxication mean the person was safe to drive?
Again, no. Perceived intoxication is not the same construct as driving safety. The alcohol challenge collected self-reported ratings, breath and saliva readings, and structural MRI measures. It did not test actual or simulated driving, reaction-time-based fitness, crash probability, or the accuracy of a screening rule that would classify a single driver as impaired. The investigators discussed prior work on the possible mismatch between willingness to drive and actual impairment, but those external findings are not newly tested by this MRI experiment.
A forensic expert can accurately say that subjective ratings changed and that region-specific density associations were observed in a selected research sample. The expert cannot infer from this study that a person who felt more sober after an interval had recovered all functions or was not impaired at a specified time.
What design and generalizability limitations matter?
The sample was 89 young, relatively low-risk social drinkers; participants with substance-use disorder histories, problematic AUDIT scores, relevant medical or neurological conditions, and other specified characteristics were excluded. These entry criteria limit extrapolation to persons with alcohol-use disorder, chronic high intake, major medical conditions, or different age distributions. The authors also identified possible effects of daily medications that were permitted but not recorded on the scan day.
The placebo had alcohol-related sensory cues, and almost half the sample believed the placebo drink contained alcohol. This makes the placebo comparison useful for studying expectation, but it would be too strong to call it a setting entirely free of exposure cues or trace ethanol on the beverage surface.
Structural MRI measurements and subjective ratings were linked observationally within an intervention protocol. Randomizing alcohol versus placebo improves a challenge comparison, but it does not randomize a person’s preexisting regional brain anatomy. Consequently, an anatomical association cannot by itself distinguish cause from correlated background factors. Follow-up imaging and longitudinal data would be needed before making claims about whether a particular drinking history changed regional gray matter.
The whole-brain analyses also rely on image segmentation, smoothing, voxel-wise models, and multiple-comparison controls. Their association statistics have sampling uncertainty and are not field-validated diagnostic operating characteristics. Sensitivity, specificity, predictive value, individual classification error, and legal decision thresholds for using MRI to infer intoxication were not established.
What should an attorney or forensic reviewer request before citing this research?
First identify the proposition being offered. Is the question whether a laboratory detected ethanol in a particular specimen, whether a witness said they felt intoxicated, whether a standardized psychomotor task changed, or whether an image metric correlates with a subjective score? Each needs a different evidentiary foundation.
For an alcohol-test dispute, examine the actual instrument, specimen matrix, collection and sampling times, calibration, raw records, limitations, and the intended reported quantity. For a disputed impairment claim, examine contemporaneous behavior and task evidence rather than recasting a structural MRI association as an impairment test. For a neuroscience claim, inspect the original voxel analysis, correction method, sample eligibility, model covariates, placebo comparisons, and whether the result has ever been externally validated for individual diagnosis. The site’s forensic toxicology evidence library provides related method and specimen guides.
The most defensible use of this study is a restrained one: subjective experiences of acute alcohol exposure varied by occasion and were associated with certain structural MRI measurements in a research sample. It does not establish an individual’s BAC, actual driving capability, neurological injury caused by alcohol, or causal responsibility for an event.
Frequently asked questions
Can an MRI determine a person’s BAC?
No. This study measured breath and salivary alcohol at specified times and estimated gray matter density from MRI. It did not validate brain MRI as a quantitative blood-ethanol assay or develop an individual BAC estimator.
Did alcohol cause gray matter to shrink during the experiment?
The study did not demonstrate that. Its findings concerned associations between structural density estimates and subjective ratings, not measured before-and-after tissue loss caused by a single drinking episode.
Why did people report less intoxication later?
The average self-reported intoxication score was lower at the later measurement. The study discusses changing subjective response and acute tolerance, but its early and late breath measurements were not identical matched BAC exposures for each person. The finding cannot identify one exclusive mechanism.
Does feeling less drunk prove less psychomotor impairment?
No. Subjective feelings and task performance are distinct endpoints. This research did not test a driving or validated forensic psychomotor classification rule.
What does gray matter density mean here?
It is a statistically estimated MRI feature derived through image processing and voxel-based morphometry. It is not a direct tissue biopsy, measurement of ethanol concentration in the brain, or a standalone clinical diagnosis.
Can this study establish that a particular drinker had a cerebellar abnormality?
No. It identified group-level regional associations after specified image and regression analyses. An individual’s image would require appropriate clinical acquisition, interpretation, differential diagnosis, and independent validity evidence for any particular proposed conclusion.
Could an attorney use this paper to prove a crash was caused by intoxication?
Not from this paper alone. Crash causation requires case-specific facts and evidence. The experiment reported no collisions, driving outcomes, or validated method for attributing an individual’s conduct to alcohol on the basis of these MRI findings.
Primary scientific source
Stennett-Blackmon BA, Sevel L, Boissoneault J. Association of cerebellar and pre-motor cortex gray matter density with subjective intoxication and subjective response following acute alcohol intake. Scientific Reports. 2023;13:7340. doi:10.1038/s41598-023-34546-5. The interpretations above are confined to this complete primary study’s protocol, results, and stated limitations.