Can a Methamphetamine-Word EEG Signal Distinguish Psychosis History? An Addiction Stroop Study of 68 People

Direct answer: In a study of 68 analyzable adults, an electroencephalography-derived response to methamphetamine-related words differed in one subgroup of abstinent methamphetamine users, even though their average reaction times and error rates did not differ significantly from those of controls. Participants without a history of methamphetamine-associated psychosis exhibited a more negative left-anterior N200 event-related potential than healthy controls during an addiction Stroop task. The post-hoc comparison was reported at p = 0.009. Other electrophysiological components and several behavioral comparisons were not statistically significant. This is a group-level research observation, not a validated EEG test for psychosis, current methamphetamine intoxication, or impairment in a particular person.

The primary study was Huang and colleagues, “Neural basis of the attention bias during addiction stroop task in methamphetamine-dependent patients with and without a history of psychosis: an ERP study,” published in Frontiers in Psychology in 2023. Read the original primary article (DOI: 10.3389/fpsyg.2023.1173711). Its most important forensic boundary is unusually clear: the researchers recorded scalp electrical signals and computerized responses but did not quantify methamphetamine in a contemporaneously collected whole-blood, plasma, serum, or other forensic specimen.

What scientific question did the investigators actually test?

The investigators compared three groups: healthy controls, methamphetamine-exposed individuals without a history of psychosis, and methamphetamine-exposed individuals with such a history. Their question was whether a brief brain electrical response to addiction-related words could reveal differences not apparent in standard computer-task accuracy and speed. They were studying the relationship among historical methamphetamine dependence, a psychiatric-history grouping, the way attention is allocated to drug words, and event-related potentials. The study was not designed to establish who had recently taken methamphetamine, whether a person was hallucinating during the experiment, or whether any subject could safely operate a vehicle.

That separation is important. A behavioral task may measure the speed or accuracy of selecting a key when a word appears. An event-related potential measures the average voltage pattern generated by neuronal activity time-locked to the event. A toxicology assay answers a different question: whether a specified chemical was identified or quantified in a validated physical specimen. The three observations cannot be treated as interchangeable forms of proof.

How many independent people were studied?

The abstract describes recruitment of 31 healthy controls, 14 methamphetamine users without psychosis history, and 24 with methamphetamine-associated psychosis history. One participant in the no-psychosis group was subsequently removed because too many EEG epochs failed quality requirements. The characteristics table and analytic results therefore use 31 controls, 13 without psychosis history, and 24 with psychosis history: 68 independent analyzed people. Reporting 69 as the final analytic sample would mix recruitment with analysis. The 256 word presentations per person, multiple electrode channels, and repeated electrical epochs do not turn 68 biological participants into thousands of independent humans.

All three groups were drawn from a specific research setting and were screened against neurological, major physical and other psychiatric exclusions. The participants identified as methamphetamine users supplied histories through an interview rather than a controlled dosing protocol. Some measures were available only for the two methamphetamine-history groups. Any application outside comparable people, tasks and recording procedures therefore requires new validation.

What did psychosis history mean in this study?

The psychosis-history group had a lifetime diagnosis of methamphetamine-associated psychosis, with eligibility tied to episodes of hallucinations or delusions. The comparison methamphetamine group had no current or past psychotic symptoms under the investigators’ criteria. This was not a randomized assignment to psychosis. People were grouped according to pre-existing clinical histories, making causal interpretations about the origin of any EEG difference inappropriate.

The investigators required abstinence from methamphetamine for at least 15 days. The characteristics table reported a much longer average recalled abstinence duration in the two patient groups—approximately 32 and 34 months, with broad ranges. Those historical statements do not constitute contemporaneous quantitative drug measurements. Nor do they establish an exact last-use time for anyone in a litigation matter. The group label “with psychosis history” must not be converted into proof that a participant was experiencing an active episode during the experiment.

What were the actual test materials and measurements?

The researchers used an addiction version of the Stroop task: eight methamphetamine-related words and eight matched neutral words, each displayed in four colors. Participants had to respond to the ink color instead of the meaning of the word, using four assigned keyboard keys. Four randomized blocks contained 64 trials each, for a total of 256 presentations after practice. The behavioral endpoints were reaction time for correctly answered trials and percentage of response errors. Those are task-derived measures, not chemical concentrations.

Scalp EEG was recorded with 32 electrodes at a 1,000 Hz sampling rate while the participants performed the task. The study filtered and processed the signal, constructed word-locked epochs from 200 milliseconds before to 1,000 milliseconds after stimulus onset, rejected specified artifacts, and assessed three event-related-potential components. Importantly, the detailed ERP group comparisons focused on methamphetamine-related word trials. They were not direct within-person electrical comparisons demonstrating how an individual responded differently to every neutral-word trial.

What the participant characteristics do—and do not—show

Group summaries matter when considering whether the observed EEG effect could arise from the study design or from measured background differences. The patient groups were similar in self-reported daily methamphetamine history and on the Barratt impulsiveness subscales in the table, and the three groups had no statistically significant reported difference in age or education. A nonsignificant baseline comparison is not proof that every potentially important characteristic was identical or that selection bias was absent. Numbers below are descriptive; the reported years of methamphetamine use and quantities are histories, not laboratory-confirmed dosages.

What is the N200, and what does “more negative” mean?

An event-related potential is created by averaging voltage fluctuations aligned to a defined stimulus. “N200” refers to a negative-going electrical component commonly studied in the roughly 200–300 millisecond window following an event. The paper averaged activity over left-anterior electrodes F3, F7 and FC5. Voltage amplitude is expressed in microvolts, not milligrams, nanograms per milliliter, or the numerical units of any validated drug-toxicology result.

More negative is a description of electrical polarity and relative voltage. It is not a measure of the “severity” of psychosis, proof of neuronal injury, or a universal scale of how much drug a person has ingested. EEG components depend on task parameters, reference choice, artifact processing, participant state and averaging rules. A technically sound interpretation has to stay with the exact task, component window and electrode region that produced the contrast.

What was the primary reported N200 difference?

During methamphetamine-word trials, the no-psychosis-history group showed a more negative left-anterior N200 mean amplitude than healthy controls. The authors reported a post-hoc comparison of p = 0.009. They also reported a group-level N200 analysis at p = 0.028. The graphical display shows considerable variation among individual participants and overlapping observations. This does not support classifying any one person’s diagnosis from a single amplitude reading.

The paper did not establish that the N200 difference cleanly separated the two methamphetamine-exposed groups from one another. Nor did it provide a sensitivity, specificity, receiver-operating-characteristic curve, prespecified diagnostic cutoff or external validation cohort. The observation is best characterized as a potentially interesting electrophysiological association in this particular sample—one that would require independent replication and prospective classifier testing before use as a diagnostic instrument.

Did participants perform worse on the addiction Stroop task?

Not on the study’s principal mean reaction-time or error-rate comparisons. The authors reported a nonsignificant group effect for correct-trial reaction time (p = 0.284), a nonsignificant word-condition effect (p = 0.930), and a nonsignificant group-by-word-condition interaction (p = 0.993). For error rate the corresponding tests were also nonsignificant: group p = 0.733, condition p = 0.930, and interaction p = 0.970. These are actual reported null statistical comparisons, not an inference from an absent table.

The result limits the strongest possible interpretation of the EEG finding. A significant scalp-voltage contrast did not come with independently established average behavioral slowing or excess task errors in the same study. An effect in one type of research outcome does not automatically create an effect in another. Likewise, failure to reject a null hypothesis is not an equivalence trial: these results do not establish that every participant, condition or outcome would be identical in a larger or more demanding experiment.

What did the P300, N450 and component-latency analyses show?

Other ERP components did not demonstrate the same clear mean-amplitude group effect. For P300 amplitude the paper reported p = 0.196, and for N450 amplitude p = 0.636. Their time windows were approximately 250–450 milliseconds and 400–650 milliseconds, respectively. The later N450 is often discussed in the context of cognitive conflict, but this study’s measured N450 group amplitude did not supply supporting statistical evidence for a group distinction.

The authors also tested response timing for each component. Group differences in N200, P300 and N450 latencies were not statistically significant, with reported p values of 0.365, 0.838 and 0.962. A claim that every EEG component was delayed in methamphetamine-associated psychosis would therefore be contrary to these findings. The original observation was comparatively narrow: selected left-anterior N200 amplitude in relation to one patient subgroup and controls, not a general collapse of electrical attention timing.

Which significant and nonsignificant results should appear together?

Separating positive outcomes from negative ones prevents selective citation. All entries below concern the same analyzed sample or the small no-psychosis-history subset; the correlation analysis is not an independent trial and should not be presented as replication. In particular, the absence of reaction-time and error-rate effects is as relevant to interpretation as the one reported post-hoc N200 difference.

How did impulsiveness scores relate to the EEG finding?

In the small group of 13 analyzed participants without a psychosis history, N200 amplitude had a positive correlation with the Barratt attention-impulsiveness score (r = 0.614, p = 0.026) and non-planning score (r = 0.656, p = 0.015). By contrast, the Barratt total association (r = 0.502, p = 0.080) and motor association (r = 0.083, p = 0.788) did not reach statistical significance. These results neither measure blood-drug exposure nor establish how strongly a particular person is impulsive outside the study.

Correlations based on thirteen people can be sensitive to unusual points and sample selection. Multiple correlations and component comparisons create a risk of chance findings even when several nominal p values are below 0.05. The authors’ scatterplots are useful for seeing the individual spread, but a correlation coefficient is not a tested decision threshold, causal treatment mechanism or prospective psychosis predictor. Statements about subjective impulsiveness should remain distinct from objective errors, legal intent, and observed behavior.

Why does a history of psychosis not make N200 an individual diagnostic test?

A diagnostic test needs clearly specified eligibility, a reproducible acquisition protocol, independent validation against an accepted reference diagnosis, a prespecified cutoff, and quantitative measures of false positives and false negatives. The present study provided none of those classifier-validation outcomes. It was an exploratory group comparison using a modest cohort, unequal subgroup sizes, and participants with differing abstinence histories. A clinician may consider EEG observations as research context, but the study does not show that a reading identifies active hallucinations, distinguishes methamphetamine-associated psychosis from schizophrenia, or predicts a future psychiatric episode.

Even a perfectly reproduced group mean difference would allow substantial overlap between individuals. Observing a particular N200 amplitude in someone would not tell an examiner whether that person has the studied history, how long ago drugs were used, or why an event occurred. The authors themselves describe the finding as an initial investigation requiring future research. That modest evidentiary status should remain visible in any litigation-oriented summary.

Does this paper contain a methamphetamine toxicology result?

No quantitative drug result was produced for a forensic biological specimen. The researchers did not report an analytically measured concentration of parent methamphetamine, amphetamine metabolite or another stimulant in venous whole blood, separated plasma or serum, urine, oral fluid, hair, vitreous humor, breath, milk or tissue. Their historical daily gram figures came from participant interviews. Their scalp microvolt signals came from electrophysiology; their reaction times were measured in milliseconds; the Barratt scores came from a questionnaire.

Those measurement systems answer different scientific questions. A valid measured drug concentration requires a specified analyte, sample collection, calibration, matrix, method, quantification limits, quality controls and reporting criteria. For broader context, the site’s published guide to biological matrices explains why compartments are not interchangeable, while its confirmatory testing guide identifies the analytical issues that an actual drug-testing result requires. Neither principle turns EEG voltage into a blood-drug concentration.

Can the findings establish present intoxication, driving impairment or event causation?

They cannot. The analyzed patient groups were abstinent rather than experimentally given a controlled dose of methamphetamine. No contemporary quantitative pharmacokinetic profile was developed, no on-road or simulator driving test was performed, and no crash, workplace injury, witness observation or criminal event was studied. There is consequently no measured dose–blood concentration–behavior chain linking this N200 result to an individual. Any claim that a particular person was intoxicated, drove unsafely or caused a specific event would require independent case evidence.

Researchers studying attention and clinicians evaluating psychosis may both find the result informative within its narrow scope. Forensic inference nevertheless depends on the question: evidence that a group differs on a brain signal is not the same as evidence that an individual had a chemical substance in a specific specimen at a specific time. The latter would need validated drug testing; impairment assessment would need appropriate behavioral and clinical evidence beyond a historical EEG research contrast.

How should an expert evaluate apparent conflicts within the source?

The published text contains differences between some recruitment descriptions and analyzed denominators, and certain reported degrees of freedom and omnibus statistics require careful verification before anyone recalculates or broadens an effect. A disciplined scientific review uses the final analyzed sample reported with the results, quotes published statistics as reported without silently repairing them, and avoids claiming a precise diagnostic accuracy never tested. Graphical distributions, complete analysis tables and methods matter more than an isolated conclusion sentence.

It is also essential to preserve the asymmetry between the findings. The apparent N200 effect was narrower than the list of task outcomes studied; P300 and N450 did not follow it; response latencies did not follow it; and average error rates and reaction times did not follow it. That pattern does not mean the EEG result is meaningless, but it places the result in a preliminary context. It argues against converting nominal statistical significance into certainty about neurological disease, substance exposure or legal responsibility.

What questions should a forensic reviewer ask before citing this study?

First establish whether the matter concerns historical dependence, psychiatric diagnosis, current intoxication, drug identity, impairment, or specific event causation. Those are different propositions. Then verify the actual cohort and exclusions, the periods of abstinence, clinical diagnosis criteria, the type of word stimuli, the EEG montage and N200 electrode sites, the precise post-stimulus windows, the multiple endpoints tested, and the negative comparisons. A reviewer should obtain the complete paper rather than relying on a headline suggesting “brain-based detection.”

For an actual biological drug result, separate inquiry must address chain of custody, specimen matrix, validated analytical method and uncertainty. The published therapeutic drug monitoring guide is useful background on timed concentrations, but this EEG experiment itself supplied none. Distinguish what was physically sampled from what was calculated or inferred and ask whether the proposed legal conclusion was among the measured study endpoints.

Frequently asked questions about methamphetamine-word EEG responses

1. Did the study prove EEG can diagnose methamphetamine-induced psychosis?

No. It reported a group-level N200 amplitude comparison in a small sample grouped by psychosis history. No diagnostic threshold, external validation set, false-positive rate or clinical decision rule was established. An individual clinical evaluation and appropriate differential diagnosis remain separate tasks.

2. Were all 69 recruited participants included in the EEG analysis?

No. Although the original recruitment account includes 31 controls, 14 methamphetamine users without psychosis history and 24 with psychosis history, one no-psychosis participant failed EEG quality criteria. The analyzed sample was 31, 13 and 24, totaling 68 independent people.

3. Did addiction-related words make the patient groups respond more slowly?

The reported mean reaction-time analyses did not find significant group, word-condition or interaction effects. In particular, the group-by-word-condition test yielded p = 0.993. The result should not be rewritten as proven slowed responses in all methamphetamine-exposed participants.

4. What EEG component showed the clearest reported group contrast?

The left-anterior N200 amplitude measured roughly 200–300 milliseconds after methamphetamine-word presentation. The no-psychosis-history subgroup differed from controls in a post-hoc analysis at p = 0.009. This does not supply a person-specific cutoff or prove that the psychosis-history subgroup had a separate statistically different distribution.

5. Did P300, N450 and reaction-time measurements confirm the N200 pattern?

No. Mean-amplitude group tests for P300 and N450 were not significant, and the N200/P300/N450 latency tests were all nonsignificant. Behavioral reaction-time and error-rate differences likewise failed to achieve statistical significance in the principal analysis.

6. Was methamphetamine measured in blood or urine?

No contemporaneous quantitative human biological drug measurement was reported. Historic intake and abstinence came from clinical interviews; the principal experiment recorded scalp EEG and computer responses. Neither N200 microvolts nor a reported daily drug-use history equals a confirmed blood concentration.

7. Could an expert use this article alone to explain one driver’s impairment or one witness’s accuracy?

No. The participants did not complete an actual driving or witness-identification experiment, and the study established no individual concentration–effect threshold or event-specific causal model. The source can inform a carefully bounded discussion of research on attention and psychosis history, but it cannot replace toxicological casework and directly relevant behavioral evidence.

Bottom line for forensic toxicology and expert testimony

Huang and colleagues’ primary research illustrates how an EEG-derived N200 contrast may appear when the simple task measures do not differ significantly. That contrast was observed in a small, abstinent methamphetamine-exposed subgroup without psychosis history, while group reaction times, errors, other ERP amplitudes, component latencies and several impulsiveness correlations yielded genuine null findings. The source reports useful preliminary neuroscience, not a blood-drug assay or a validated impairment detector.

A scientifically defensible report should preserve the actual analyzed denominator, the exact left-anterior stimulus-specific measurement, the null comparisons, the source’s limitations and the complete absence of forensic specimen quantitation. Original study: Huang et al., Frontiers in Psychology (2023), DOI 10.3389/fpsyg.2023.1173711.