Does Methamphetamine Dependence Weaken Reactive More Than Proactive Control? A 178-Person AX-CPT Study

Answer: In a 2025 observational study, men with a history of methamphetamine dependence performed worse than a comparison group on the reactive-control condition of a modified AX Continuous Performance Test (AX-CPT), but the reported proactive-control and baseline conditions did not show statistically significant differences after the authors’ adjustment for age. The authors also reported a lower Proactive Behavioral Index in the methamphetamine-dependence group. A separate working-memory experiment found lower mean span scores in both methamphetamine-dependence and social-media-overuse groups compared with their own controls. These are group-level computer-task associations, not evidence that methamphetamine was present in anyone’s blood at testing, that a particular individual was impaired, or that the drug caused a specific event.

The primary source is Zhang, Zhang, Xu, Zhou, and Shen, “Differential Impact of Methamphetamine Dependence and Social Media Overuse on Cognitive Control: Based on the Dual Mechanisms of Control Theory,” published in Behavioral Sciences in 2025 (original primary study, DOI 10.3390/bs15081086). This analysis examines the measured tasks, denominators, positive and null findings, and limitations relevant to interpreting executive-function evidence in a forensic setting.

1. What was the actual scientific question?

The investigators asked whether people recruited because of methamphetamine dependence and people selected for unusually heavy social-media use had different cognitive-control profiles from their respective comparison groups. They did not administer methamphetamine, determine an acute dose-response curve, or observe participants driving. They instead compared preexisting groups in two different experiments that measured related—but not interchangeable—components of cognitive control.

The distinction matters because “cognitive control” is not a single laboratory analyte or a standalone clinical diagnosis. It encompasses keeping instructions active, deciding when to prepare a response, and revising that response when the stimulus changes. A task may isolate one part of this process without establishing how a person would act while driving, performing a job, or encountering a sudden hazard. In this study, the first experiment used a calculation-and-recall span procedure; the second used a modified letter-cue response task and a derived index comparing its trial conditions.

2. Who participated, and how many independent people were studied?

Experiment 1 included 199 analyzed adults in four groups: 43 methamphetamine-dependent men, 43 male comparison participants, 40 people selected for social-media overuse, and 73 social-media comparison participants. Experiment 2 used newly recruited participants and reported 178 analyzed adults: 43 methamphetamine-dependent men, 38 male controls, 42 social-media-overuse participants, and 55 comparison participants. These are two separate experimental samples. The hundreds of stimulus trials within the AX-CPT do not add hundreds of independent people to its sample size.

The methamphetamine-dependence participants were recruited from a compulsory rehabilitation center in Zhejiang Province following physiological detoxification. Their comparison participants were recruited separately and were also male. The social-media comparison was conducted with college students identified using adapted screening questionnaires and reported frequency-of-use thresholds. Accordingly, the results are not a randomized experiment assigning otherwise comparable people to years of methamphetamine use versus no use, and “social-media overuse” should not automatically be treated as the same clinical disorder as substance dependence.

The source describes 181 initial participants and 178 retained in the second experiment after reaction-time exclusions, while its tabulated group sizes sum to 178. The appropriate denominator for the reported final condition comparisons is therefore 178, not 181, and the final group breakdown is the one displayed in the study’s Table 2. A separate count of repeated trials must never be substituted for that denominator.

3. How did the researchers measure working-memory span?

In Experiment 1, participants solved sequential single-digit arithmetic problems while remembering the second digit from each operation. After a sequence, they recalled those digits in order. The published span metric was the last level completed correctly before the participant’s first error; higher scores indicated better performance on that particular procedure. It is a behavioral test score, not a chemical biomarker of methamphetamine.

The authors adjusted the methamphetamine-group comparison statistically for age and the social-media comparison for gender. The first comparison was especially challenging because the methamphetamine group’s mean age was 35.30 years, versus 24.12 in its comparator. Statistical adjustment is useful but does not, by itself, make groups with large demographic differences interchangeable. In a forensic review, the actual recruitment and covariate overlap matter as much as the p-value.

The group means and reported adjusted tests support a narrower statement than “methamphetamine causes permanent memory loss.” The measured phenomenon was lower group-level span performance under one task. Participants’ earlier education, treatment setting, years of drug use, other exposures, sleep, psychiatric characteristics, and differences in recruitment may also matter. The study did not randomly assign prior dependence and cannot quantify the causal contribution of a specific substance to any participant’s score.

4. What is the AX Continuous Performance Test?

The second experiment used a modified AX-CPT. A cue letter appeared, followed after an interval by a probe letter. The meaning of the second letter depended in part on the first. Frequent “A-X” combinations encouraged a dominant response expectation. In “A-Y” trials, the participant had to suppress the response prepared after the expected A-X sequence when the probe turned out not to be X. In “B-X” trials, the initial B cue could help the participant anticipate that the otherwise familiar X probe should not trigger the dominant response. “B-Y” trials supplied a lower-conflict baseline; a B-Z supplement was intended to discourage a shortcut strategy.

According to the paper’s formal design, 70% of 280 trials were A-X; the other four listed classes each represented 7.5%. The authors excluded the common A-X training-dominant trials and the B-Z supplementary class from the main task-condition analysis. Thus the published AY, BX, and BY figures refer to selected subsets and derived measures, not overall correctness across every presented letter.

The researchers interpreted AY as reactive-control demand and BX as proactive-control demand. That mapping is a theoretical operationalization: it specifies why the authors expected one condition to engage a particular process. It does not provide a direct observation of which neurons activated or establish an independent biological diagnosis of proactive or reactive control.

5. What did the authors actually measure in Experiment 2?

The key reported metric was the paper’s “Inverse Efficiency Score,” abbreviated IES, for BY, AY, and BX conditions. The published description characterized higher reported scores as better task performance. These are constructed performance indices derived from accuracy and response speed, not milliseconds of reaction time, blood concentrations, or percentages of correct drivers. An interpretive problem remains: the paper describes the index using accuracy divided by reaction time, although “inverse efficiency” is often defined differently in methodological literature. For fidelity, the results below preserve the source’s reported index values and directions without silently converting them into a different metric.

Its separate Proactive Behavioral Index (PBI) was calculated using the relative difference between BX and AY indices, divided by their sum. A positive result favors the BX measure relative to AY; a negative result goes the other direction. This index is dimensionless and comparative. It should not be represented as the proportion of someone’s brain relying on one network, as a measured neurotransmitter signal, or as a clinical score with a validated legal cutoff.

6. What were the actual reactive-control findings?

After age adjustment, the methamphetamine-dependent group had a mean AY index of 1.44, compared with 1.68 in its male comparison group. The authors reported p = 0.017 and partial η² = 0.071. The result is a statistically significant difference under the study’s conventional p < 0.05 criterion and is consistent with lower performance on this selected reactive-control condition among the participants recruited for methamphetamine dependence.

The social-media-overuse group had an AY index of 1.56 versus 1.67 in its comparison group, with gender-adjusted p = 0.056. That comparison did not cross p < 0.05. It can be described as an observed numerical difference with unresolved statistical uncertainty, not as a second proven impairment. Calling both results significant would change the experiment’s reported outcomes.

Neither comparison used a randomized drug challenge. Moreover, the methamphetamine group was substantially older than its comparator in Experiment 2, with means of 34.10 and 19.45 years. An ANCOVA may reduce confounding under model assumptions, but such a large age separation leaves questions about extrapolation, model fit, and unmeasured correlates. The word “reactive” in a task label does not eliminate those questions.

7. Did the proactive-control condition also differ significantly?

No statistically significant difference was reported for the BX proactive-control index in either comparison. The methamphetamine-dependent group scored 1.35 and its comparator 1.50, age-adjusted p = 0.201. The social-media-overuse group scored 1.35 versus 1.45, gender-adjusted p = 0.120. Both p-values are above 0.05.

This does not prove the groups were exactly equivalent or that every participant possessed intact anticipatory control. It means the reported analyses did not resolve a difference in those BX scores at the selected significance threshold. Sample size, measurement precision, participant heterogeneity, covariate modeling, and the extent to which the adapted AX-CPT captures a real-world skill all affect interpretation. An expert should not turn “not statistically significant” into “zero effect under every condition.”

8. What happened in the baseline and relative-control indices?

The lower-conflict BY baseline condition was also nonsignificant: methamphetamine dependence 1.71 versus control 1.85, p = 0.191, and social-media overuse 1.74 versus control 1.83, p = 0.081. The authors’ PBI comparison was different. Men in the methamphetamine-dependent group had mean PBI −0.035 versus +0.067 in the male control group, p = 0.009. The social-media PBI comparison was 0.079 versus 0.074, p = 0.446, a nonsignificant finding.

The PBI’s negative mean is sometimes summarized as relatively greater reliance on reactive control. That summary needs qualification: the very group described as relying on reactive control also performed worse on the AY condition used as its reactive proxy. A relative score cannot establish that reactive control is working well, that reactive control dominates actual neural activity, or that individuals consciously select an identified strategy. It captures the relationship between two derived task-performance measures.

The methamphetamine comparison uses age adjustment; the social-media comparison uses gender adjustment. The four groups in this experiment were 43 methamphetamine-dependent men, 38 male controls, 42 social-media-overuse participants, and 55 social-media controls. Reporting each metric with its own p-value avoids a misleading headline that treats every component of cognitive control as similarly affected.

9. Does the study show acute methamphetamine intoxication?

No. The methamphetamine-dependent participants were recruited during a rehabilitation phase after physiological detoxification. The investigators did not give methamphetamine during testing, identify an administered dose, or report a validated quantitative postdose specimen series. The study’s group label concerns prior substance-use history and recruitment criteria. It is not itself a measured blood concentration or evidence of drug presence at the time of the tasks.

Forensic toxicology requires discipline about the actual physical specimen. A confirmed parent-methamphetamine result in properly collected venous whole blood is a different kind of evidence from a dose documented in a study protocol or from a behavioral score in a person with substance-use history. Separated plasma and serum also have different specimen definitions and cannot be silently substituted for whole blood. Urine or oral-fluid screening may address recent exposure under their own assay limitations, but this study does not provide a validated quantitative series in any of those matrices. The site’s guide to biological matrices in forensic toxicology explains why the physical sample must remain tied to the analytical claim.

Without measured blood concentration, collection time relative to use, laboratory methods, and specimen validity, it would be erroneous to convert an AX-CPT score into nanograms per milliliter of methamphetamine. Equally, a history of dependence does not establish that an acute stimulant effect was present on a particular date. Prior exposure, current exposure, withdrawal, psychiatric status, and learned task performance are different evidentiary questions.

10. Why is the age difference a serious interpretive limitation?

In Experiment 1, the methamphetamine-dependent men were on average roughly 11 years older than their control men. In Experiment 2, the separation was roughly 15 years. Age can correlate with cognitive performance, educational opportunity, clinical history, occupational experience, and patterns of digital testing familiarity. The investigators addressed age with ANCOVA, which is preferable to pretending the groups were age-identical, but covariate adjustment depends on overlap and model assumptions.

When participant groups occupy markedly different age ranges, an adjusted comparison may rely partly on a statistical prediction about participants scarcely represented in one group. The source does not supply a randomized design, common-age matching sufficient to eliminate this concern, or an experimental manipulation of years of dependence. Therefore the results can support a reported association after specified adjustment; they cannot isolate the causal effect of years of methamphetamine exposure or a specific lifetime dose.

Other differences also matter. The methamphetamine cohort came from a rehabilitation institution, while controls were recruited separately; the social-media cohort came from a college environment. Educational attainment differed between methamphetamine and control groups. A courtroom presentation must not collapse these recruitment contrasts into a single clean exposure variable.

11. Are the two experiments a longitudinal recovery study?

No. Experiment 2 used newly recruited participants rather than repeated testing of the 199 people from Experiment 1. The 178 AX-CPT participants were therefore a separate group, not a follow-up outcome for the initial 199. The study reports no randomized baseline treatment assignment, no sustained longitudinal monitoring of abstinence and cognition, and no within-person demonstration that changing methamphetamine exposure changed these indices.

It follows that the observed span difference in one sample cannot be matched participant-by-participant to an AY or PBI difference in the other sample. There is no published correlation establishing that an individual’s arithmetic span predicts that same individual’s reactive-control score across the two experiments. Similarly, the report does not quantify a recovery trajectory following a measured abstinence interval or assess whether a future rehabilitation intervention will restore a particular person’s executive-function performance.

12. How should positive, nonsignificant, and unmeasured findings be separated?

There are three analytically distinct categories. A positive statistical result here is an adjusted difference the authors reported below their selected significance threshold, such as AY p = 0.017 or PBI p = 0.009 in the methamphetamine comparison. A nonsignificant result is an analysis that was performed but did not clear that threshold, such as BX p = 0.201 or BY p = 0.191. An unmeasured outcome is something the investigators never tested—for example, actual vehicle collisions, whole-blood methamphetamine concentrations, or a legally validated single-person impairment classification.

It is misleading to label an unmeasured question “p > 0.05,” because there is no relevant p-value. It is likewise misleading to treat p = 0.056 as a positive result merely because it is close to 0.05. Statistical nonsignificance does not establish equivalence, and nominal significance does not necessarily imply a large or clinically decisive effect.

The paper examined several outcomes: working-memory span, multiple AX-CPT conditions, the comparative PBI, and two separate types of recruitment comparison. A reader should consider the family of analyses and potential multiple-testing issues. The manuscript reports selected adjusted tests, but that does not create a validated decision threshold for diagnosing any particular individual from a single score.

13. What does a task-derived index say about real-world behavior?

A test score represents performance under a defined set of stimuli, instructions, deadlines, response keys, and scoring rules. The AX-CPT’s repeated expected A-X cue structures are intentionally artificial; their purpose is to separate response-preparation and correction demands under laboratory control. The study did not observe participants navigating traffic, reacting to an unexpected pedestrian, operating machinery, or documenting an actual incident.

Task performance may be relevant when forming hypotheses about executive functioning. It is not automatically an individual safety certification or an accident reconstruction. Real-world performance also depends on fatigue, distraction, practice, visual ability, situational workload, risk-taking, road conditions, competing tasks, and many other variables not validated here. The gap between a group mean and an individual prediction is especially important when the study provides no externally validated classification sensitivity, specificity, decision threshold, or legally relevant error-rate estimate.

Nor was this an electroencephalography, fMRI, neuronal-recording, or gene-expression study. The researchers used keyboard responses on computer tasks. Statements that a particular prefrontal brain region was damaged or that neural activity was directly measured in these participants would exceed the actual data.

14. Can a history of methamphetamine dependence establish a blood level or use time?

No. The experiment includes information about prior dependence and, in its tables, reported histories such as years and monthly consumption. Those are background descriptors, not measurements of methamphetamine in a collected blood, plasma, serum, urine, hair, or oral-fluid specimen from the testing session. They also do not establish when the person last used a particular dose.

A case requiring chemical identification or quantitative interpretation should be evaluated using the case’s real test records, specimen matrix, timing, chain of custody, validation, and confirmatory procedures. The separate guide on confirmatory drug testing addresses the evidentiary distinction between an instrument-supported chemical finding and broader interpretations. Even properly confirmed methamphetamine presence would be a different proposition from proving a quantified individual cognitive deficit or incident cause.

Because this paper did not perform serial methamphetamine concentration measurement in the relevant participants, it cannot establish a pharmacokinetic elimination slope, a dose-specific impairment threshold, a concentration at a past event, or how abstinence changes a person’s measured AX-CPT performance. Importing such a number from another population without an appropriate basis would not repair this gap.

15. What evidence would be needed for a person-specific forensic opinion?

An opinion about a real person first requires a precisely defined question: chemical detection, time since administration, clinical diagnosis, functional impairment, or causal contribution to an incident. Evidence useful for one question may be irrelevant to another. For drug-presence opinions, obtain the complete original laboratory report and validation, actual specimen identity, collection time, analytes, measurement uncertainty, possible medication exposures, and chromatographic confirmation where applicable.

For a functional assessment, consider documented contemporaneous behavior, the specific event demands, competing medical and environmental explanations, standardized methods appropriate to the individual, and evidence about the test’s operating characteristics. The group AX-CPT differences reported here do not supply an individual cutoff or substitute for that record. It would be especially inappropriate to infer that all persons with a prior diagnosis must make a particular type of response error during a specific incident.

If the issue concerns prescribed stimulants rather than historical methamphetamine dependence, do not interchange drug identity, route, therapeutic dose, matrix, or dosing schedule. Therapeutic drug monitoring and medication adherence have their own clinical purposes and timing requirements. The site’s therapeutic drug monitoring overview is relevant background, but neither that guide nor this observational paper provides a case-specific impairment equation.

16. What limitations should be disclosed in litigation or expert reporting?

Several limitations are central rather than peripheral: recruitment by preexisting group membership; a male-only methamphetamine cohort; markedly different average ages between methamphetamine and control participants; unequal education and recruitment settings; the absence of random assignment to drug-exposure histories; no directly assayed active methamphetamine concentration at testing; separate participants in Experiments 1 and 2; and no driving, occupational accident, or person-specific classification validation.

The control-label terminology also needs care. “Reactive-control impairment” is the authors’ interpretation of one selected behavioral task contrast. The PBI is a calculated ratio from task indices. Neither establishes a unique brain mechanism for a given subject. The source’s IES nomenclature and formula deserve methodological scrutiny before making a numerical comparison to another AX-CPT study that may compute efficiency differently. Relative differences among conditions should not be stated as universal clinical deterioration or proof of the permanence of any deficit.

Finally, this study did not directly compare people who had received the same methamphetamine dose under controlled experimental conditions. It compared groups with different histories, environments, and characteristics. A responsible opinion should state that the data support an association under this design, rather than extending the result into an unsupported causal or predictive conclusion.

17. Bottom line for interpreting this experiment

The original research provides a useful but narrow contrast. In Experiment 1, methamphetamine-dependent men scored lower than their male controls on a working-memory span task, while a separately recruited social-media-overuse group also scored lower than its comparison group. In Experiment 2, methamphetamine dependence was associated with a lower AY reactive-control score and lower PBI, whereas the analyzed BY baseline and BX proactive-control comparisons were nonsignificant. The social-media AX-CPT and PBI contrasts did not meet the reported significance threshold.

Those details are scientifically more defensible than the oversimplified claim that dependence “switches off” cognitive control or that a negative PBI proves one particular person’s intoxication. The study does not quantify legal blood methamphetamine, demonstrate acute drug causation, test actual road driving, or assign an individual accident probability. See the full original 2025 primary article and DOI record for the underlying trial design, figures, adjusted comparisons, and the authors’ own limitations.

18. Frequently asked forensic questions

1. Did 178 people receive methamphetamine during the AX-CPT experiment?

No. The 178 analyzed adults belonged to preexisting recruitment groups. The methamphetamine group consisted of men in rehabilitation with a reported dependence history. The study did not dose the participants with methamphetamine or report a controlled acute drug concentration at the time of the cognitive testing.

2. Was reactive control worse but proactive control normal?

The methamphetamine group had a statistically lower AY reactive-task score, while its BX proactive comparison was nonsignificant. “Normal” is too strong: a nonsignificant group contrast is not a demonstration of equivalence or universally normal individual function.

3. Does a negative Proactive Behavioral Index establish brain injury?

No. PBI is a relative calculation from selected AX-CPT behavioral scores. It is not an imaging result, a tissue diagnosis, a measurement of brain damage, or a validated person-specific legal impairment test.

4. Were the social-media-overuse results identical to the methamphetamine results?

No. Both comparisons had lower span scores in the first experiment, but in the second experiment social-media AY p = 0.056 and PBI p = 0.446 were nonsignificant. The cohorts also differed in recruitment and demographics, so this was not a controlled ranking of the causal toxicity of two exposures.

5. Can these results reveal a driver’s methamphetamine blood concentration?

No. No serial, chemically quantified methamphetamine whole-blood or plasma results were obtained for these task sessions. No conversion factor relates the published IES or PBI values to a driver’s nanograms-per-milliliter concentration.

6. Does the separate memory-span experiment track cognitive recovery?

No. It used a different set of people from the AX-CPT experiment and was not a longitudinal treatment or abstinence study. Between-group span scores do not show how one person’s performance changes after stopping or restarting any drug.

7. Can an expert use this article to prove methamphetamine caused an accident?

Not by itself. It describes laboratory task associations in groups, without validated individual prediction, event observation, measured active drug concentration, or causal reconstruction. Any opinion about a particular incident requires independent event-specific and laboratory evidence.