Does Dextroamphetamine Improve Cognitive Control—or Just Change EEG Signals? A 23-Person Masked Attention Trial

Short answer: In a randomized, double-blind crossover study, 23 healthy adults completed a difficult visual attention task after placebo, 10 mg oral dextroamphetamine, and 20 mg oral dextroamphetamine. The drug improved the signal-detection index known as d-prime and reduced target-response time in the overall analysis. It also increased a frontal EEG response called P3a during correctly withheld responses. But drug treatment did not significantly improve every behavioral measure: separate hit and false-alarm outcomes, response bias, and mean target P3b amplitude lacked significant drug effects. A difficulty-dependent theta pattern further showed that an EEG change cannot simply be labeled “better control” at every workload.

This is evidence about performance on a specific laboratory task and its electrical correlates, not an individualized blood-drug threshold, a roadside sobriety validation, or proof that a particular person was fit or unfit to drive. The empirical source is Bhakta and colleagues’ 2022 original study, EEG reveals that dextroamphetamine improves cognitive control through multiple processes in healthy participants, published in Neuropsychopharmacology. The original primary study and DOI identify the experiment examined here.

What question did the researchers actually ask?

The investigators wanted to determine whether an oral stimulant changed performance and brain electrical responses on a demanding, deliberately cross-species-compatible cognitive-control task. Instead of treating “attention” as one indivisible ability, they examined detecting an occasional target, withholding a response to a non-target, choosing promptly, and processing visual stimuli that were either easy to view or quickly masked. Simultaneously recorded EEG provided a separate family of measurements: scalp voltage components and oscillatory power. One endpoint could improve while another failed to change or moved in an unexpected direction.

This distinction matters in forensic interpretation because everyday words such as attention, inhibition, alertness, and performance often conceal different operational definitions. Faster button or joystick movement during a research paradigm is not necessarily a lower accident risk; a larger P3a waveform is not a quantification of parent drug in blood. The task must be identified, the contrast must be specified, and the actual null results must remain visible alongside statistically significant ones.

Who was studied, and how many independent people contributed?

Investigators enrolled 36 apparently healthy adults but reported outcomes for 23 who completed all three test days. Their demographic table described 12 men and 11 women; average age was 22 years with a standard deviation of 4.82. Participants underwent clinical screening, including medical and psychiatric assessment, electrocardiography, urine toxicology screening, and, where applicable, pregnancy screening. The protocol was designed for healthy research volunteers, not patients with suspected intoxication, defendants in a driving case, or people undergoing a clinical performance assessment.

Each completed participant encountered three dosing visits. Therefore, there were 23 independent human participants, not 69 independent humans. Likewise, 216 trials per visit do not create 216 separate volunteers, and thousands of EEG data points are not independent biological replicates. Repeated observations can improve estimation within a person, but conclusions about generality must still respect the modest participant denominator, the completion/attrition history, and the narrow eligibility criteria.

How were dextroamphetamine and placebo administered?

The study used a randomized, placebo-controlled, double-blind, counterbalanced within-participant design. On separate visits roughly a week apart, participants took placebo, 10 mg dextroamphetamine, or 20 mg dextroamphetamine orally. The cognitive task with EEG took place approximately 120 minutes after the pill. The publication refers to these as active doses expected to be biologically active based on earlier work. It did not report a matched series of quantitative human blood or plasma amphetamine concentrations from which the reader could establish the actual exposure in each individual at the moment of performance.

Dextroamphetamine is d-amphetamine, not methamphetamine. Neither the administered milligram dose nor the nominal time after dosing is itself a directly measured forensic drug concentration. A separate question would require a suitable collected specimen, analytically validated drug identification and quantitation, a known collection time, and careful interpretation of the particular matrix. Those ingredients are not supplied by an EEG performance experiment merely because a medication was administered under controlled research conditions.

What made the five-choice continuous performance test difficult?

The researchers used a modified five-choice continuous performance task, abbreviated 5C-CPT. In target trials, a participant moved a joystick toward the one illuminated circle. In non-target trials, all five circles appeared illuminated and the participant was supposed to withhold a response. Target trials therefore test detection and action selection; non-target trials provide a direct requirement to suppress an inappropriate movement. These outcomes should not be casually combined into one global label of “impairment.”

Each full session contained 216 task trials: 108 easy/unmasked and 108 difficult/masked, with 90 target and 18 non-target trials in each difficulty condition. Easy targets appeared for 100 milliseconds. In the hard condition the image appeared for only 10 milliseconds before a visual mask covered it for the following 90 milliseconds. A response window of one second was provided, with varied delays between trials. The masking manipulation changed visibility while holding the total presentation interval comparable, making it possible to evaluate whether drug effects depended on perceptual challenge.

This artificial task is useful precisely because it isolates responses under controlled conditions. It does not reproduce lane keeping, divided attention in live traffic, situational judgment, risk selection, fatigue, tolerance, simultaneous sedatives, or sudden hazards. An effect demonstrated under a 10-millisecond visual display cannot be generalized into an on-road safety classification without independent validation.

What was measured, and what was not?

Computerized behavior included response speed, successful target hits, false alarms when a non-target should have been ignored, a signal-detection measure called d-prime, and an index of response bias. Physiological recording included event-related EEG voltages and frequency-specific theta power. Screening urine was part of eligibility procedures; it was not a published quantitative postdose urine drug-excretion study. The article also described changes in pulse and blood pressure, but it did not transform those clinical measurements into a validated classifier of individual impairment.

Did the medication actually speed responses?

The mixed-model analysis reported a significant overall drug effect on target-hit reaction time, F(1, 52.36) = 11.06, p = .002, along with a significant effect of task difficulty, F(1, 46.44) = 16.98, p < .001. Masked trials were generally harder. A planned comparison identified faster hit reaction time after 10 mg than after placebo for difficult targets (p = .007 in the Results text). However, the study did not find a significant drug-by-difficulty interaction for this endpoint. That means the data do not support a sweeping claim that the stimulant helped only hard stimuli for reaction speed.

The 20-mg-versus-placebo post hoc reaction-time contrast was not significant in the reported analysis, and the two active doses were not significantly different from one another. Thus, “20 mg is twice as effective as 10 mg” would be an invention. Nor is a shorter response time proof of better choices: speed and accuracy may trade off, and real-life activities impose constraints that are not replicated by a laboratory joystick movement.

What happened to the signal-detection index?

D-prime quantifies the separation between detecting targets and incorrectly responding to non-targets under signal-detection assumptions. Its overall drug-model association was significant, F(1, 70.46) = 4.23, p = .044. The original Results text reports a 10-mg-versus-placebo post hoc comparison of p = .03 and a 20-mg-versus-placebo comparison of p = .057. The latter is not below the conventional .05 threshold; one graphical caption rounds that contrast differently. It should not be presented without qualification as a clean significant 20-mg benefit.

Separate models for hit frequency, false alarms, and response bias did not show significant main or interaction effects. This is not a logical contradiction: a composite signal-detection index can show a model difference while its contributing individual proportions do not each meet a significance criterion. But it is a major limitation on an overly broad assertion that dextroamphetamine “fixed response inhibition.” The appropriate conclusion is narrower: the task’s composite sensitivity shifted, with no demonstrated significant change in several component behavioral measures.

Why is frontal P3a important—and why is it not a drug concentration?

The frontal P3a is a voltage component observed shortly after a relevant stimulus and commonly investigated in orienting, attentional allocation, and response-selection processes. Here the investigators examined P3a at a fronto-central scalp location during correct rejections of non-targets. They reported a significant main drug effect, F(1, 59.64) = 16.89, p < .001, with higher frontal P3a amplitude under active treatment. A secondary way of scoring P3a relative to an earlier N2 trough also yielded a significant treatment association, F(1, 64.70) = 15.43, p < .001.

That finding supports the study’s mechanistic investigation, but P3a amplitude has no one-to-one conversion to ng/mL dextroamphetamine, an exact dose, a time since ingestion, or an impairment percentage. A larger waveform does not prove that the same individual made fewer false alarms, especially when the separate false-alarm model showed no significant drug effect. In case analysis, identifying the level of measurement is indispensable: stimulus-locked microvolt behavior, task outcome, and chemically quantified specimen are separate forms of evidence.

Did parietal P3b amplitude increase at the group level?

No significant group-level drug effect on mean target P3b amplitude was shown. The source reports a nonsignificant effect of task difficulty on P3b, F(1, 42.05) = 2.20, p = .15, with drug and drug-by-difficulty effects below an F statistic of one. This genuine null finding matters because a reader could otherwise see the positive correlation analysis and mistakenly conclude that active treatment universally raised the P3b waveform.

In secondary within-participant analyses limited to correctly detected difficult targets, changes in target-hit rate were correlated with changes in P3b amplitude. Reported p-values for three dose contrasts were .04, .01, and .05. These are associations between two change measures, not evidence of a significant average P3b treatment effect. Some are near a conventional decision threshold; multiple exploratory tests and restricted trial counts further constrain predictive claims. The original researchers discussed this physiological-behavioral association as a potential lead for future research, not as a clinically validated response test.

What did the theta result reveal about task difficulty?

Frontal theta is an oscillatory EEG measure commonly considered in studies of performance monitoring and cognitive control. The investigators differentiated brain activity anchored to target presentation from activity anchored to the participant’s response. They found no significant drug influence on stimulus-locked theta power in the analysis reported. The more informative result came from response-locked theta during target responses: a significant treatment-by-difficulty interaction, F(1, 72.33) = 5.95, p = .02.

When the task conditions were considered separately, response-locked theta decreased with active treatment on easy targets, F(1, 34.26) = 8.34, p = .007. On difficult targets, the direction was described as increased power but the relevant model did not reach statistical significance, F(1, 29.59) = 1.51, p = .23. A claim that dextroamphetamine significantly increased theta on difficult trials would therefore be inaccurate. The reliable pattern was the interaction and the easy-condition reduction, not significant improvements in both halves.

The source also examined correlations between drug-related theta changes and changes in hit rate on easy targets. Two contrast p-values were .01 and .02; the third was .08, a nonsignificant result. Correlation direction and interpretation should be understood in the context of the study’s EEG scoring, not simplified into the idea that any reduced theta value means an impaired or intoxicated individual. Electrical activity is sensitive to the epoch, reference, frequency band, trial definition, and analytic model.

Which statistically significant and nonsignificant contrasts belong together?

The table intentionally places the positive findings next to negative ones. It avoids numerical bar-height estimates that cannot safely be extracted from the printed graphs and avoids treating individual dose visits as separate independent participants. The reported F tests came from mixed models with non-integer denominator degrees of freedom, not simple two-sample tests based on 69 unrelated people. A nominal significance statement should also be distinguished from clinical significance, a validated classifier, and a causal explanation of one person’s behavior.

Were heart rate, blood pressure, and subjective experience aligned?

The Results text states that dextroamphetamine increased heart rate and blood pressure, whereas the subjective symptom scale did not show significant changes in fear, drowsiness, or happiness. Detailed physiological and symptom analyses were directed to supplementary materials rather than supplied as complete numeric tables in the present original PDF. The primary article also notes a participant with elevated heart rate and blood pressure lasting more than eight hours after pill administration; the condition resolved, but the participant was excluded out of caution.

This does not justify an incidence estimate from a small selected sample or establish that all medically eligible people respond similarly. It does illustrate why psychomotor or EEG task improvement cannot be equated with an absence of pharmacological or cardiovascular effects. Likewise, a nonsignificant subjective drowsiness change does not establish an absence of stimulant action, because symptom reports and computerized performance are distinct endpoints with their own variability and sensitivity.

Why does the absence of quantitative drug levels matter?

The study protocol included a urine toxicology screen during eligibility and visit preparation. That is different from reporting validated quantitative concentrations of parent dextroamphetamine in a collected human forensic specimen during the EEG experiment. The original article did not publish corresponding numerical concentrations of dextroamphetamine in venous whole blood, separated plasma or serum, urine, oral fluid, hair, or breath. It also provided no method validation establishing the analytical limit, calibration, extraction recovery, or matrix-specific uncertainty for such an assay. A dosing assignment is not a reported laboratory result.

These distinctions prevent an expert from substituting “10 mg” or “20 mg” for “10 ng/mL” or “20 ng/mL,” and from assuming that one person’s reported dose fixes that person’s blood level at the moment of an incident. Whole blood includes cellular elements whereas plasma and serum are separated fractions; concentration units and matrix-specific pharmacokinetics cannot be interchanged by a casual ratio. EEG theta power and P3 voltage are physiological responses, not substitutes for direct chemical identification of parent drug and metabolites.

Can the study establish an individual person’s intoxication or impairment?

No. It was a small healthy-volunteer experiment under controlled dosing and a specific computerized task. It did not enroll known impaired drivers, test actual crash risk, observe collisions, validate a diagnostic impairment threshold, or establish sensitivity and specificity for detecting intoxication in individual people. It did not compare measured legal whole-blood concentrations with contemporaneous roadside findings. The paper’s evidence supports limited inference about experimental contrasts under its stated conditions; it does not supply case-specific facts about medication adherence, combined substances, tolerance, impairment, witness credibility, medical injury, or event causation.

Forensic case evaluation requires the actual event timeline, independently documented clinical and behavioral observations, a validated specimen collection and laboratory process when chemical findings are used, relevant medication history, and any competing explanations. Neither an EEG component nor a population mean task difference can replace those inputs. Favorable performance on an artificial test cannot prove that an individual was safe, and an unusual electrical response cannot by itself prove culpable impairment.

How should the study be presented in expert testimony or scientific writing?

Begin with what was randomized: dose condition within the same volunteer. Identify the true 23-person completion sample and the laboratory task. Report dextroamphetamine as the administered molecule, not methamphetamine or an undifferentiated “stimulant.” State the nominal oral dose and testing interval without inventing pharmacokinetic concentrations. For each result, identify the exact contrast: the overall mixed-model drug term is not the same thing as a 20-mg post hoc contrast, and a significant association between within-person changes is not a significant average treatment effect.

A scientifically balanced summary must also mention the genuine nulls: no demonstrated separate hit or false-alarm drug effect, no significant response-bias difference, no group mean P3b effect, no hard-condition theta effect, and no stimulus-locked theta effect. The original publication’s contrast-description inconsistencies should not be silently repaired with more favorable numbers. Independent confirmation from additional samples and studies designed for clinically meaningful decisions would be needed before claiming personalized predictive utility.

What are the primary limitations?

The sample was small and selected for health. Only 23 of 36 enrolled participants completed all three test days. Repeated-measures analyses improve efficiency but do not overcome limited population coverage. Participants were young adults tested roughly two hours after standardized oral dosing. The study did not establish how effects change during chronic treatment, withdrawal, long-term adaptation, older age, relevant psychiatric illness, co-ingested alcohol or other drugs, or the timing of an external incident.

The task involves only 18 non-targets in each difficulty condition, leaving comparatively sparse opportunities to measure false alarms. The manuscript itself discusses a likely response-performance ceiling and limited power for complex multivariable analyses. Its within-person ERP correlations are interesting physiological observations, but their role as predictive biomarkers is prospective, not validated. Supplementary numerical source material was cited in the original paper; where absent from the user-supplied PDF it should not be replaced by estimates or conjecture.

Frequently asked questions

1. Did both 10 mg and 20 mg dextroamphetamine significantly improve d-prime versus placebo?

Not under a strict reading of the Results text. The overall drug effect was significant at p = .044; the 10-mg-versus-placebo contrast was p = .03, while the 20-mg-versus-placebo contrast was p = .057. A figure caption rounds the latter differently. The stronger conclusion is an overall modeled association and a significant 10-mg contrast, not a clean demonstrated benefit of every dose.

2. Did dextroamphetamine reduce false alarms on the no-go trials?

The original investigators did not find a statistically significant drug effect in the separate false-alarm analysis. P3a voltage during correctly withheld responses increased, but a voltage change and a reduced false-alarm frequency are not identical findings.

3. Does higher P3a mean higher dextroamphetamine concentration in whole blood?

No. P3a is an event-related scalp EEG voltage component. The study did not publish corresponding quantitative legal venous whole-blood parent-drug concentrations or validate P3a as a concentration assay.

4. Was parietal P3b significantly increased by the drug?

No significant average drug effect on P3b amplitude was reported. The significant observations involved correlations between within-person changes in P3b and difficult-target hit rate. Those findings must not be recast as a significant mean treatment effect on P3b.

5. Did the stimulant increase frontal theta activity when the visual task was hard?

The hard-target change was not statistically significant (p = .23). The significant results were the overall drug-by-difficulty interaction (p = .02) and a reduction in response-locked theta on easier targets (p = .007).

6. Could this experiment diagnose acute amphetamine-related driving impairment?

No. It did not validate roadside thresholds, a person-level sensitivity or specificity, actual driving outcomes, or a mapping from a chemical specimen result to individual impairment. It is a laboratory experiment about a masked attention task and EEG signals.

7. What primary research should a reader examine before citing this result?

Read the complete Bhakta et al. original research publication, including its original figures and available supplemental analyses. Check the specific reported contrasts and sample exclusions rather than relying on an abstract, a figure caption, or a generalized statement that stimulants improve cognition.