Can Eye Tracking Identify Methamphetamine Cue Bias Without Measuring Intoxication? A 66-Man Study

Short answer: In a 2022 observational experiment, 30 men recruited from methamphetamine rehabilitation displayed longer initial fixations and greater gaze dwell on methamphetamine-related photographs than on paired neutral photographs. Thirty-six men without a reported drug-use history showed a different visual preference. The participants with a methamphetamine-use history also reported more craving after the image task, but the researchers did not find a statistically significant relationship between their baseline craving scores and the attention-bias measures. These findings concern responses to drug-associated images, not the concentration of methamphetamine in blood, acute intoxication, driving impairment, or a validated individual diagnostic test.

The underlying primary research is Mirtorabi, Saleki, Rahmanian, Hadizadeh, Rostami, and Yoonessi, Direct and Indirect Measures of Attention Indicating a Bias Toward Cues in Methamphetamine Users (2022), DOI: 10.32598/bcn.12.6.1589.1. This article examines what was actually observed, how direct gaze recording differs from a button-press proxy, and where the study’s forensic inference ends.

What question did the experiment answer?

The investigators asked whether men with a history of methamphetamine use preferentially directed visual attention to methamphetamine-related pictures, compared with neutral pictures and with a control group. They combined two distinct measures: a 60-Hz eye tracker that sampled the direction and duration of gaze, and a dot-probe task that recorded how quickly someone pressed a key after a probe appeared on one side of a screen. Participants with a history of methamphetamine use also scored their momentary subjective craving before and after the task.

That is a narrower experimental question than whether eye tracking can determine who is intoxicated. The exposure of interest was a brief visual cue. No methamphetamine dose was administered under a controlled experimental protocol. The laboratory did not establish a measured postdose methamphetamine concentration curve or test real-world driving. Its comparison was principally between groups distinguished by use history, after eligibility screening, during a single laboratory picture-viewing session.

Who was tested, and how many independent people were represented?

The final analyzed sample comprised 66 different men: 30 rehabilitation-center methamphetamine users and 36 controls described as having no prior drug use. The investigators initially recruited 40 men from rehabilitation. Four were excluded for psychological disorders identified during evaluation and six for positive urine screening, leaving 30. A psychiatric interview and urine analysis were used as eligibility procedures, not as a published quantitative toxicokinetic dataset.

The two groups were similar in reported mean age: 33.6 ± 6.6 years for the methamphetamine-history group and 33.7 ± 6.0 years for controls. The users reported that their most recent methamphetamine consumption had occurred between eight and 29 days before testing; the reported average interval was approximately 19.96 days. This retrospective time history was not independently established by a measured serial blood concentration profile.

A subject could view many picture pairs, make repeated eye movements, and produce numerous button responses. Those events were repeated observations within a person. They do not increase the sample to thousands of independent humans. In the main mixed analyses, the group factor was the person-level methamphetamine-use history, while image category was varied within participants.

What did participants actually look at?

The investigators paired 15 images related to methamphetamine with visually and semantically matched neutral pictures. Each drug–neutral pair appeared four times, balancing the left and right placement of the cue. Twenty pairs of neutral natural-scene images were also shown to examine side preferences. The principal task therefore contained 60 cue–neutral presentations and 20 neutral–neutral presentations per person, each image pair displayed for about one second.

Before each pair, a central fixation cross was displayed for approximately two seconds. A probe appeared after the images disappeared; participants identified its location with a left or right key. The study used an EyeTribe ET1000 device reported at 60 Hz. A stable gaze location for at least 100 milliseconds was classified as a fixation. These procedural definitions matter: the measured “first fixation” is a rule-based screen event, not a neurological recording of reward, a drug assay, or evidence that the person was unable to disengage from a roadway hazard.

Investigators removed incorrect dot-probe responses, response times below 200 milliseconds, and specified outlying response times. They summarized each individual’s data for analysis. The design was intended to compare where attention went and how quickly the probe was detected, not to recreate driving, employment duties, treatment decisions, or legal interrogation.

What did the first-fixation results show?

In the methamphetamine-history group, the reported average first fixation on a drug cue lasted 192 ± 37.4 milliseconds, compared with 178 ± 24.4 milliseconds on a neutral image. The within-group comparison was reported as t(29) = 3.80, p < .001. In the control group, the pattern went in the opposite direction: neutral-picture first fixations averaged 190 ± 22.6 milliseconds, versus 183.5 ± 29.7 milliseconds for cue pictures, p = .004.

The mixed-design image-type-by-group interaction for first-fixation duration was also significant, F(1,64) = 25.4, p < .001. This interaction supports a different response pattern across the two groups in this task. It does not establish that every user showed a drug-cue preference or that a fixed millisecond cutoff can identify methamphetamine use in an unfamiliar individual.

The relevant denominator for the comparisons is the number of people, not the number of recorded gaze samples. Nor should the 14-millisecond difference between the two averages in the user group be described as an individually diagnostic test. Group means, overlapping distributions, and an unvalidated screening threshold answer different statistical questions.

What is dwell time, and why is it distinct from first fixation?

First fixation captures an early viewing event. Dwell time concerns the portion of a picture-presentation interval spent looking at a particular image. A person may initially look at one picture and subsequently shift repeatedly; both metrics can therefore point in similar directions without measuring the same behavior.

In the methamphetamine-history group, the authors reported mean dwell percentages of 31 ± 3.9% for cue pictures and 26 ± 3.3% for neutral pictures, p < .001. In controls, cue dwell averaged 31 ± 4.5% and neutral dwell 36 ± 5.2%, with the within-group comparison reported as p = .002. The interaction between image type and participant group was F(1,64) = 31.89, p < .001.

Thus, the directly recorded gaze measure showed a group-by-image difference, while the absolute cue-picture dwell mean was reported as 31% in both groups. The informative contrast is how cue and neutral viewing differed within each group and in the interaction—not the inaccurate claim that the user group’s absolute cue-picture mean was categorically higher than the control group’s. The stimulus pairing and baseline visual preferences are central to interpretation.

What did the proportion of first looks at drug cues establish?

The researchers also calculated whether a person’s first fixation landed on a cue picture rather than the neutral picture in the paired trials. Among users, the mean cue-first proportion was 54.58% ± 6.2%, reported as statistically greater than 50% chance. The control-group mean was approximately 50.6%, and its chance comparison was not statistically significant.

Being modestly above chance on average does not turn the measure into a definitive yes–no classification for a single person. A 54.58% average represents a small preference in a specially designed 1-second picture task; it is not a rate of drug consumption, a probability of relapse, or a percentage of impairment. The study did not present an externally validated cutoff, prospective individual diagnostic sensitivity and specificity, or a confusion matrix for use in forensic testing.

Primary evidence table: direct eye-tracking results

The table intentionally keeps first-fixation duration, dwell proportion, and cue-first probability separate. They share the same group of participants and overlap psychologically, so they should not be counted as three independent replications. Likewise, an interaction result does not imply that all subjects had the same direction or magnitude of bias.

Did the keyboard dot-probe task agree with eye tracking?

Broadly, it pointed toward cue-related attention in the methamphetamine-history group, but it is an indirect observation. In a dot-probe design, a probe replaces the location previously occupied by a cue or neutral picture. Faster responses at the cue location may reflect attention being allocated to that location. They may also be influenced by motor response variability, eye movements, practice, stimulus layout, and response-cleaning rules.

The article reports a within-user comparison between probe-location conditions with p = .033, while the analogous comparison among controls was p = .376 and therefore not significant under a conventional .05 criterion. The analysis used a nonparametric test after assessing response-time distribution. The paper displays the response comparison graphically but does not supply a sufficiently clear tabulation of absolute milliseconds for every probe condition to justify invented precise cell means here.

It is tempting to describe a significant result in one group and a nonsignificant result in another as proof that the groups differ. Such a conclusion ordinarily requires an appropriate between-group interaction or direct contrast. The published separate dot-probe tests should therefore be described as separate within-group findings, rather than as a complete, independently validated diagnostic model.

Did viewing methamphetamine images increase self-reported craving?

Among the 30 users, subjective craving was rated on a single-item 0–10 scale before and immediately after picture viewing. The mean rose from 2.80 ± 1.99 to 4.23 ± 2.029, a mean difference of 1.43 scale points. The reported paired comparison was t(29) = 5.78, p < .001. That result is evidence of a short-term change in self-reported craving during this task, not a laboratory measurement of dopamine release, ingested methamphetamine, or a subsequent relapse.

The study did not report the same pre/post craving protocol as a quantitative comparison in the 36 controls. It did not randomize users to a separate neutral-only image intervention. Accordingly, the pre/post increase can be described as occurring after the task, but the size of the increase cannot be exclusively attributed to drug-related images in a controlled causal comparison that the study did not perform.

Momentary craving can change with context, expectancy, embarrassment, or the way an interview or questionnaire is administered. The single-item scale captures a reported experience, not the actual probability that a person will obtain or use methamphetamine. Clinical follow-up would be required to establish a reliable prediction of real subsequent behavior.

Were craving and eye movements correlated?

This is one of the most important qualifications. Although users had cue-related gaze patterns and a statistically significant pre/post craving increase, the investigators reported no statistically significant relationship between the initial craving score and their studied attentional-bias measures, including gaze percentage, first fixation, and reaction time. Two outcomes changing in a group do not require those outcomes to track one another across individuals.

The paper also described a small negative correlation between reported craving and duration of methamphetamine-use history, r = −0.30, p = .034, with approximately 10% of variation described by the squared correlation. By contrast, duration of use was not significantly correlated with dot-probe reaction time, p = .282. The correlational results do not establish that a shorter or longer history caused craving, and they provide no individualized estimate of relapse risk.

These distinctions matter where parties selectively cite only the significant eye-tracking and craving differences. The same source expressly reports null relationships. A defensible synthesis must account for both, rather than implying that a gaze pattern objectively quantifies self-reported craving.

Primary evidence table: task, craving, and negative findings

Multiple endpoints and correlations are examined in a relatively small study. Readers should ask which analyses were prespecified, what multiplicity safeguards were used, and whether later independent samples reproduced the result. The original contrasts are legitimate observations; treating each nominal p value as a separate, conclusive legal test is not.

Was any methamphetamine concentration measured?

No quantitative contemporaneous methamphetamine concentration was reported for the attention task. The paper describes urine screening as an eligibility measure and excludes six rehabilitation recruits who screened positive. It does not provide an analytically validated set of individual quantitative urine concentrations, whole-blood concentrations, plasma pharmacokinetics, oral-fluid results, or a postdose chemical assay matched to each eye-tracking observation.

A screening result, a person’s reported most recent use, and a directly quantified drug concentration are different types of evidence. Even a negative screening result does not independently establish the exact time of last use without method-specific detection windows, cutoffs, collection time, and confirmatory data. A history of methamphetamine use is not equivalent to current methamphetamine in venous whole blood.

The general distinctions between specimens and what a method can support are explained in the site’s biological matrices guide and its confirmatory drug testing guide. Neither resource transforms this behavioral experiment into a drug-concentration study. The source provides no formula linking a 192-millisecond first fixation to a specific ng/mL reading.

Why is a visual cue response not proof of acute impairment?

Acute drug effect requires its own relevant exposure and performance evidence. Here, the evaluated men reported a period since last use, and the task consisted of a display of photographs and a keyboard response. It did not involve administered acute methamphetamine, real-world driving, simulated lane keeping, standardized occupational performance, or an independent clinical intoxication assessment as an outcome.

Looking longer at a familiar or salient image is not synonymous with inability to attend to traffic. A probe response difference in milliseconds is not interchangeable with reaction to an unexpected roadway event, especially when the experiment used repeated trials in a quiet room. The study does not establish a sensitivity, specificity, likelihood ratio, decision threshold, or error distribution suitable for deciding whether an unknown individual was under the influence.

Other studies may directly measure drug concentrations or simulated driving, but a distinct experiment cannot retroactively supply absent measurements to this one. Conflating different participant cohorts, substance administrations, biological matrices, and tasks would compromise scientific traceability. Proper interpretation names the endpoint, specimen, design, and limits of generalization each time.

Can these results predict relapse or treatment adherence?

The investigators discussed potential future usefulness for understanding attention and craving. Their reported data, however, were cross-sectional laboratory observations plus a same-session craving rating. No prospective abstinence follow-up, verified later relapse outcome, intervention allocation, or external prediction cohort was provided as the validation endpoint of this specific analysis.

A prediction rule requires more than a statistically different group average. It needs a stated target population, an operational outcome, adequate follow-up, prespecified decision thresholds, discrimination and calibration measures, and validation in an independent sample. Without those elements, the findings can motivate research hypotheses but should not be used to label a particular patient noncompliant, dishonest, or likely to relapse.

This limitation applies even if the descriptive eye-tracking result is reproducible. A reproducible difference in mean cue viewing can be scientifically interesting without yielding actionable diagnostic performance. Scientific description and individual prediction are separate evidentiary stages.

What are the main design limitations?

All participants were men, and the rehabilitation group was selected from a treatment setting after excluding some recruits. The investigators sought approximate age matching, but a single cohort cannot represent women, adolescents, all patterns of methamphetamine use, people with comorbid psychiatric conditions, or individuals facing acute high-dose exposure. Self-reported use timing and the lack of quantitative paired toxicology limit exposure interpretation.

The task was a controlled 1-second photograph-pair paradigm. Familiarity with particular cues, prior experiences, image salience, motor performance, expectations, eye-tracker calibration, and sample exclusions may contribute to observed differences. Correcting or retaining reaction-time trials according to specified rules can affect indirect results. A different stimulus set, testing environment, or eye-tracking system might produce a different magnitude or classification reliability.

The authors’ inferential statistics describe several significant results and several null correlations, without demonstrating the accuracy of an individual forensic classifier. The study is observational with respect to use history. It cannot disentangle preexisting attention tendencies from effects of years of substance use, experiences in rehabilitation, or other correlated factors. Group differences do not establish that methamphetamine exposure caused the specific gaze pattern in each participant.

How should a forensic expert use this study?

First, identify the proposition. It is reasonable to say that in one selected male sample, prior methamphetamine users as a group looked relatively longer at methamphetamine cues than neutral cues, and reported higher average craving after exposure to the image task. It is also necessary to state that baseline craving and measured attention bias were not significantly correlated and that the controls had their own viewing preference pattern.

Second, match the source to the disputed factual issue. A toxicology report about a measured blood concentration requires validated analytical data from the actual specimen and case. An allegation about impaired driving requires evidence suited to driving and the legally relevant time, not a cue-image laboratory task. A question about future relapse requires longitudinal prediction studies. For the difference between a clinical drug monitoring question and a forensic one, see the site’s therapeutic drug monitoring overview.

Third, keep denominators and conclusions honest. Thirty rehabilitation participants produced many repeated eye observations, but still represent 30 biological people. A p value concerns a hypothesis about sampled patterns; it does not establish an error-free verdict about an individual. The proper evidentiary statement is about an observed association in a defined design, not about what an unknown person must have used, intended, remembered, or done.

Frequently asked questions

1. Did eye tracking directly detect methamphetamine?

No. The device recorded direction and duration of eye movements while participants looked at photographs. It did not chemically identify methamphetamine or quantify a drug concentration in blood, plasma, serum, urine, or oral fluid.

2. How many independent people completed the experiment?

Sixty-six men were analyzed: 30 with a methamphetamine-use history and 36 controls. Repeated image trials and 60-Hz gaze samples were not additional independent participants.

3. Was the study a controlled methamphetamine administration trial?

No. Drug-use history defined the groups. Users reported last use eight to 29 days before testing, and the experiment administered picture cues, not methamphetamine. Urine testing was described as part of eligibility screening.

4. Does 54.58% first fixation on a cue diagnose substance use?

No. It is the average proportion in a selected group under this particular visual task, modestly exceeding chance. The researchers did not validate a person-level diagnostic cutoff or report the prospective sensitivity and specificity needed for such use.

5. Did faster dot-probe responses occur in everyone?

The published statistical analysis reports a within-group probe-location difference for users, p = .033, and a nonsignificant comparison for controls, p = .376. Those results describe group distributions, not that every individual reacted faster in a particular condition.

6. If craving increased, was it correlated with the eye-tracking scores?

Not significantly in the reported analysis of baseline craving and attention-bias variables. A rise in mean self-reported craving after picture viewing does not establish that gaze duration quantifies an individual’s craving intensity.

7. Can this study prove a person’s impairment, relapse, or collision causation?

No. It did not measure contemporaneous quantitative whole-blood drug levels, real driving, subsequent verified relapse, or any individual collision. Each of those claims requires distinct validated evidence and a case-specific scientific analysis.

Bottom line

The scientific contribution is the convergence of direct gaze measures and an indirect dot-probe measure around drug-cue attention in a selected methamphetamine-history group, with a separately observed increase in self-reported craving. Equally important are the nonsignificant craving–attention relationships, absence of a quantitative drug-concentration endpoint, lack of individual classifier validation, and absence of driving or follow-up outcomes. Treat this as a laboratory study of cue-directed attention—not a chemical test or standalone determination of impaired conduct.

Primary original study: Mirtorabi SD, Saleki S, Rahmanian MS, Hadizadeh H, Rostami R, Yoonessi A. Direct and Indirect Measures of Attention Indicating a Bias Toward Cues in Methamphetamine Users. Basic and Clinical Neuroscience. 2022;13(4):455–463. https://doi.org/10.32598/bcn.12.6.1589.1.