Do Methamphetamine Cues Disrupt Conflict Adaptation? A 44-Man Simon Task Study

Short answer: In a controlled, non-dosing laboratory comparison of 22 men diagnosed with methamphetamine use disorder and 22 male healthy controls, methamphetamine-related pictures altered a specific measure of response conflict. Within the clinical group, the mean Simon interference effect was 56 milliseconds for drug-related pictures and 22 milliseconds for neutral pictures. The larger effect came principally from faster responses on drug-cue trials when the correct response used the hand on the same side as the picture. More importantly, the usual reduction in conflict after a previously conflicting trial was observed for neutral pictures but not clearly for drug pictures. This is evidence about a cue-dependent laboratory response pattern, not proof that anyone had recently taken methamphetamine or was impaired during a real-world event.

The primary experiment was reported by Yining Xu, Yingzhi Lu and Lihui Wang in 2026 in Psychological Research, DOI: 10.1007/s00426-026-02277-7. The question it addresses is narrower than whether methamphetamine simply slows reaction time: can a drug-associated image preferentially activate a spatially compatible button response and modify trial-to-trial adaptation when that automatic tendency conflicts with a task rule?

What actually happened in the experiment?

The researchers enrolled 44 independent adult men. Twenty-two were patients with a DSM-5 methamphetamine use disorder recruited from a male rehabilitation center; 22 were healthy men recruited from the community. Each person performed one modified Simon-task experiment. The clinical participants’ reported mean duration of abstinence was 19.82 months (standard deviation 9.38), while their average reported history of substance use was 10.91 years (standard deviation 6.70). These historical summaries were not blood-concentration measurements and did not establish when any participant last consumed methamphetamine.

Each trial displayed either a drug-related picture or a neutral picture in the left or right visual field. A red or blue frame surrounded the picture. The subject’s instruction was to ignore the picture’s meaning and position, identify the frame color, and press a fixed key with the corresponding hand: left for red, right for blue. A picture appearing on the same side as the required responding hand produced a congruent trial; a picture on the other side produced an incongruent trial. There were 24 trials in each of the four cue-by-congruency conditions, 96 formal trials per participant, not 96 independent people.

What does the Simon effect mean here?

In the Simon paradigm, an irrelevant stimulus location can prime a motor response even when the participant knows to follow a different rule. The Simon effect in this study was the individual’s mean reaction time for incongruent trials minus the mean for congruent trials. A positive difference indicates that following the correct frame-color rule took longer when the pictured cue sat on the opposite side from the correct hand. The measure is an interference contrast measured in milliseconds; it is not a concentration, an impairment percentage, or the rate at which drugs are metabolized.

The crucial comparison is within the same group and person: drug-related versus neutral pictures. The study asks whether those pictures produce different interference effects in the diagnosed group, not whether an unspecified response threshold classifies an impaired driver. The response-key mapping and probability of frame colors were deliberately held constant. The pictures themselves, however, had naturally occurring visual differences that were not strictly matched. That distinction matters when interpreting a cue-specific effect as learned drug association rather than a completely isolated physical-stimulus property.

How did the 22 men with methamphetamine use disorder perform?

The clinical group’s mean correct-response reaction times were 552 milliseconds for drug-cue congruent trials and 608 milliseconds for drug-cue incongruent trials. Corresponding neutral-picture means were 583 and 605 milliseconds. Drug cues therefore yielded a 56-millisecond mean Simon effect (608 − 552), compared with a 22-millisecond effect for neutral cues (605 − 583). In the within-patient analysis, cue type by congruency interacted, F(1,21) = 7.84, p = .011; the source reports a between-cue Simon-effect contrast with Cohen’s d = 0.597 and a confidence interval for the within-group difference of 8.84 to 59.90 milliseconds.

The result should not be misdescribed as a general worsening of every response when drug material appeared. In fact, congruent responses were faster to drug cues than to neutral cues by an estimated 31 milliseconds, with a within-group test t(21) = 3.18, p = .005. The study did not show a clear corresponding drug-cue response difference in the incongruent condition. A cue could therefore widen the interference contrast by facilitating its compatible response rather than simply causing every response to slow down.

How did the 22 healthy controls compare?

Among the healthy controls, the mean drug-cue reaction times were 551 milliseconds congruent and 591 milliseconds incongruent, while the neutral-cue means were 559 and 609 milliseconds. The corresponding mean interference differences were 40 and 50 milliseconds. Spatial congruency was significant within the control group, F(1,21) = 43.76, p < .001, but the control-group cue-by-congruency interaction was not significant, F(1,21) = 1.16, p = .293. Nor was the cue-type main effect significant in controls, p = .095.

The full mixed analysis reported a significant group-by-cue-by-congruency interaction, F(1,42) = 8.49, p = .006. Yet two additional direct between-group comparisons, one for drug-minus-neutral responding in congruent trials and the other for incongruent trials, had p = .074 and p = .062 respectively: neither met the study’s conventional .05 threshold. A statistically significant higher-order interaction and nonsignificant decomposed tests can coexist. An expert should identify the exact estimand and contrast rather than select whichever isolated result sounds strongest.

Evidence table: mean response time and error rate

In the full data, mean reaction time had a significant effect of cue type, F(1,42) = 8.46, p = .006, and of congruency, F(1,42) = 78.56, p < .001. The overall group main effect was not significant. Because error proportions did not satisfy normality assumptions in most tested conditions, the researchers used nonparametric error-rate comparisons rather than automatically applying the reaction-time model to the error outcomes. The patient-group Friedman test across the four error conditions reached p = .008, whereas the healthy-control test was p = .168. These error findings require their own interpretation; reaction-time evidence should not be treated as proof of a universal accuracy deficit.

What is conflict adaptation, and why is the previous trial relevant?

A response-conflict task can show sequential adaptation: after completing an incongruent trial, people may become better prepared to resolve conflict on the next trial. The researchers separated current-trial Simon effects depending on whether the previous trial had been congruent or incongruent. The expectation was a smaller interference effect after a prior incongruent trial. This is a difference of differences across sequential task conditions; it should not be interpreted as the recovery trajectory of an individual after stopping methamphetamine.

To calculate sequential measures, the previous trial needed a correct response; eligible prior-correct trials accounted for 88.6% of clinical-group trials. Thus, the sequential contrasts are conditional on an analytic selection rule. The main hypothesis concerned whether current drug cues preserved interference despite the preceding conflict. It did not evaluate a biological change between visits, a pharmacokinetic process, or the effects of an acute administered dose.

What happened after a previously incongruent trial?

For neutral cues in the clinical group, the average Simon effect shifted from 37 milliseconds after a congruent previous trial to −7 milliseconds after an incongruent previous trial. That is the source’s evidence of trial-to-trial conflict adaptation, with a reported paired-test t(21) = 3.95, p < .001. For drug cues, the corresponding effects were 59 and 49 milliseconds. The study reported no significant change between those two drug-cue conditions, and a Bayes-factor analysis favored the no-difference hypothesis over the alternative by a factor of 3.33. This is moderate model-dependent evidence for a null difference in that comparison, not proof that every patient’s capacity to adapt was absent.

The cue-by-previous-congruency interaction in patients was reported as F(1,21) = 6.29, p = .020. In controls, by contrast, the main effect of prior congruency was significant, F(1,21) = 16.71, p < .001, while the cue-type main effect, p = .320, and cue interaction were not significant. The patterns therefore support a particular laboratory contrast: control participants showed sequential adaptation across both picture categories, whereas clinical participants showed the clearest adaptation for neutral pictures.

Evidence table: sequential Simon effects

This table captures the distinctive finding more directly than any single mean reaction time. It also shows why one must not collapse different levels of the design: the healthy group’s two cue categories exhibited adaptation, whereas the clinical group’s drug-cue contrast remained positive after both kinds of prior trial. The numbers describe group averages and standard errors from repeated computer responses. They cannot be used to calculate the probability that an untested individual recently used methamphetamine or would fail an unrelated real-world inhibition task.

Did the study show poorer response accuracy?

Not as a simple across-the-board statement. In the clinical group, drug-cue congruent trials had mean error rates of 2.5%, compared with 6.6% on drug-cue incongruent trials. The Friedman test across the four error conditions was significant, and selected corrected pairwise comparisons reached significance, including drug-congruent versus drug-incongruent (p = .042). But the healthy group had no significant overall four-condition error test, p = .168. In analyses of sequential error-rate interference, pairwise tests after correction did not reach significance in the clinical group (all reported p > .065).

Error proportions and reaction times capture different aspects of performance and have different distributions. The authors explicitly checked normality and variance assumptions before choosing tests. A longer time can sometimes reflect a deliberate accuracy-preserving choice, while fast responses can sometimes generate mistakes. The reported pattern does not justify imposing an individual binary “impaired/not impaired” classification on a single error percentage or on a selected 56-millisecond average contrast.

Did the researchers administer or measure methamphetamine?

No acute methamphetamine dose was administered in this experiment. The drug stimuli were pictures, not a pharmacological challenge. Methamphetamine-use-disorder status came from clinical diagnostic eligibility and history; the reported years of substance use and months of abstinence were background characteristics. The experimental endpoints were keyboard response times and error percentages, with a computer display controlling visual stimulation. The report contains no calibrated quantitative measurement of parent methamphetamine or amphetamine metabolite in the participants’ post-task venous whole blood, separated plasma or serum, urine, oral fluid, hair, vitreous fluid or exhaled breath.

Even a correctly measured drug concentration in one matrix would not automatically be equivalent to a concentration in another, and an administered milligram amount would still be distinct from a measured nanogram-per-milliliter specimen concentration. Here, there was no administered amount or measured drug concentration at all. A pharmacological mechanism may be discussed in the theoretical background, but this observational group comparison cannot attribute the measured Simon effects to a known contemporaneous dose, individual elimination curve or dose-response relationship.

What can the 19.82-month mean abstinence tell us?

It characterizes the recruited clinical sample; it does not establish continuous verified abstinence for each participant. The reported duration also varied, with standard deviation 9.38 months. In a small exploratory association within the clinical sample, the reduction in sequential conflict adaptation was correlated with reported abstinence duration, r = .430, p = .046. An analogous correlation with total reported substance-use years was r = .400, p = .065 and did not meet a .05 threshold. Neither correlation provides a longitudinal, within-person recovery or deterioration estimate.

The researchers did not measure craving during this experiment, although they considered craving one possible explanation. Smoking and alcohol-drinking histories differed between groups: mean smoking years were 20.04 versus 8.63 and mean alcohol-drinking years 16.55 versus 8.14. The authors conducted some smaller post hoc matched-subgroup comparisons, but those exploratory procedures cannot eliminate all selection bias, life-history differences or residual confounding. Inferring that longer abstinence itself caused deterioration would reverse the logic of a cross-sectional correlation without longitudinal intervention evidence.

Is this the same thing as an attentional-bias or dot-probe test?

No. A standard attentional-bias measure often compares where attention goes, how long the gaze remains on a drug picture, or whether a probe is detected faster at a cued location. Xu and colleagues instead tested a response-conflict task in which the relevant dimension was frame color and the irrelevant dimension was picture position. The central observation was an interaction between cue category and spatial response congruency, and another involving the preceding trial. That permits a more precise statement about interference and adaptation than a generic claim that someone “noticed drug pictures more.”

The study did not record calibrated eye tracking or a direct measure of gaze dwell time as its primary outcomes. Nor did it perform a cue-approach training intervention, assess substance relapse prospectively, or administer methamphetamine experimentally. Related literatures can motivate hypotheses, but results from different tasks cannot be pooled as though the same endpoint, population and time scale were measured. The task distinction is especially important when scientific material may later be used in a legal setting requiring a particular functional inference.

Why do the null findings matter in a forensic interpretation?

Several results specifically limit overgeneralization. Healthy controls did not show a significant cue-by-congruency interaction (p = .293), but that does not mean the controls lacked a standard Simon effect; their overall congruency effect was strong. The direct between-group simple comparisons of drug-related responding had p = .074 and .062, so those selected comparisons alone did not establish group separation. Patients showed neutral-cue conflict adaptation but no detected adaptation for drug cues; a Bayes factor of 3.33 is not an immutable physiological law or a person-level validation statistic.

Other outcomes were not merely “nonsignificant”; they were not measured. There were no contemporaneous chemical drug measurements, no driving simulator, no actual road test, no crash outcome, no observed relapse and no eyewitness accuracy task. Those missing endpoints cannot be assigned imaginary p-values or described as negative trials. Keeping true null statistics distinct from unobserved outcomes prevents inappropriate extrapolation from a group cognitive-control result to a litigated person’s behavior.

What are the most important design limitations?

The clinical and control groups each comprised just 22 men, restricting precision and generalization to women, older clinical populations or people without the study’s eligibility characteristics. The clinical group came from a rehabilitation center, so findings do not necessarily extend to currently intoxicated individuals, other treatment settings, or people with coexisting psychiatric diagnoses excluded by the study. The pictures were naturalistic and not strictly equated for every visual attribute; learned association is a plausible interpretation, but alternative contributions from picture properties are not removed merely by randomizing the response rule.

The groups also differed in smoking and drinking history. Smaller post hoc subgroup comparisons are informative but not a substitute for randomizing methamphetamine exposure or accurately measuring and balancing all confounders. Sequential analyses used only trials with a correct preceding response, which can change the available trial set. Finally, several statistical contrasts and correlations were tested within a relatively small sample. The overall experimental pattern merits attention, but no prospective individual sensitivity, specificity, impairment cutoff, likelihood ratio or forensic validation against independent real-world cases was reported.

How should a forensic toxicologist use this study?

It is suitable as a carefully bounded example of how a methamphetamine-associated stimulus can influence laboratory stimulus-response competition in people with a history of methamphetamine use disorder. It can support a discussion of cue-dependent automatic response tendencies and the difference between immediate spatial interference and adaptation after a preceding conflict. It cannot establish that a specified person has methamphetamine in blood, is impaired at a particular time, is untruthful, will relapse, caused a crash or could not inhibit a specific real-world action.

Any case-specific opinion would need evidence appropriate to the actual proposition: authenticated specimen collection and confirmatory chemical analysis for a drug-concentration question; a defined, validated behavioral method for a performance claim; independent event records and causal alternatives for an incident; and consideration of the limits of transferring group averages to an individual. The correct forensic formulation here is an observed group-specific computer-task response pattern with meaningful null contrasts and unmeasured chemical and driving endpoints.

Frequently asked questions

Did this experiment give 44 men methamphetamine?

No. Twenty-two men had a clinical history of methamphetamine use disorder and 22 were healthy controls. The experiment presented drug-themed and neutral pictures and recorded keyboard decisions; no study methamphetamine was administered.

What is the main reaction-time difference?

Within the clinical group, the drug-picture Simon effect was 56 milliseconds versus 22 milliseconds for neutral pictures. These are incongruent-minus-congruent group mean contrasts, not delays on every trial or an individual performance threshold.

Did drug pictures always make the clinical participants slower?

No. The key within-group finding included a 31-millisecond faster drug-picture response in congruent trials. The data do not show a uniform slowing effect across all conditions.

Was adaptation completely absent in methamphetamine-use-disorder participants?

No. Sequential adaptation was observed for neutral cues. The drug-cue comparison showed little detected adaptation; the source reported a Bayes factor favoring a null difference by 3.33, which is moderate evidence concerning that contrast, not a universal human incapacity.

Were both groups distinguishable by every simple comparison?

No. Two follow-up direct between-group comparisons had p = .074 and .062. A significant three-way interaction does not turn each decomposed contrast into a significant difference or a validated classifier.

Can the task indicate a recent methamphetamine dose or blood concentration?

No. No recent dose was experimentally given, and no quantitative methamphetamine concentration in whole blood, plasma, serum or other forensic biological matrix was obtained during the experiment.

Can the result prove driving impairment, relapse or fault in an incident?

No. The paper did not measure real or simulated driving, subsequent relapse, incident causation or person-specific impairment. It can inform a bounded scientific explanation of cue-dependent response conflict, not replace direct evidence addressing those different questions.

Primary research and final interpretation

Source: Xu Y, Lu Y and Wang L. “Drug cue enhances response conflict and reduces conflict adaptation in methamphetamine addiction.” Psychological Research 90, article 57 (2026). Read the original primary study (DOI 10.1007/s00426-026-02277-7). The study’s tables and task design should be consulted when assessing its findings; the numeric summaries above concern laboratory group contrasts, not tested forensic drug impairment thresholds.