Does D-Amphetamine Raise Psychosis-Like Ratings Without Improving Memory? Two 20-Person Drug Trials

Short answer: In two separate, randomized, double-blind crossover experiments involving 40 healthy volunteers altogether, oral d-amphetamine raised ratings of psychosis-like experiences without producing a statistically significant improvement in the forward spatial-span or digit-span tasks tested. The d-amphetamine experiment enrolled 20 people and compared 0.45 mg/kg with placebo. A different group of 20 people received 200 mg caffeine and placebo; caffeine neither significantly improved these span outcomes nor raised the combined psychosis-like-experience score. D-amphetamine nevertheless increased heart rate and blood pressure, while caffeine increased blood pressure. Within the d-amphetamine group, the drug-associated change in psychosis-like ratings was negatively correlated with spatial-span change at an eight-second delay (r = −0.58; p = 0.006). This exploratory correlation cannot establish that the ratings caused any change in memory, much less diagnose psychosis or prove impairment in an individual.

The underlying original human investigation is Kassim and colleagues’ 2023 PLOS ONE paper, The effects of caffeine and d-amphetamine on spatial span task in healthy participants. Understanding its result requires keeping four things separate: the actual drug administered, the mental and physiological endpoints measured, the independent people included in each experiment, and the outcomes the researchers did not measure.

What forensic question does this experiment actually answer?

The narrow scientific question is whether a defined oral stimulant challenge changes forward spatial or digit memory span and short-term psychosis-like questionnaire ratings in healthy participants. It is not a clinical validation study of a forensic blood concentration threshold. Nor is it an experiment on impaired driving, eyewitness identification, workplace performance, violence, or criminal responsibility. The legitimate inference concerns group-level task responses and self-reported or interview-rated phenomena under carefully described laboratory conditions.

These boundaries matter because the terms stimulant effect, cognitive enhancement, psychosis-like experience, and impairment are not interchangeable. A faster pulse shows a measurable physiological response, but does not establish that spatial memory improved. An increased composite questionnaire score documents a research-scale change, not a new psychiatric diagnosis. A nonsignificant drug-versus-placebo test does not prove identical drug and placebo effects or rule out smaller effects in another population or task.

Were 40 people given both d-amphetamine and caffeine?

No. The authors conducted two separate experiments, each with 20 independent healthy adults. In the first experiment, each participant encountered d-amphetamine and its placebo on separate, counterbalanced, blinded occasions. In the second, a different set of 20 encountered caffeine and its placebo. Thus the biological total was 40 people, with two different within-person drug-placebo comparisons. There was no single 40-person, three-arm crossover comparing d-amphetamine directly against caffeine in the same volunteers.

Repeated exposures, measurements at several times, multiple span trials, and observations under different delays do not increase the number of independently enrolled humans. Counting 20 people twice would artificially inflate the apparent sample and could distort any claimed precision. The study is appropriately described as two small crossover pilots, not a 40-person randomized head-to-head drug comparison.

What stimulant doses were administered?

The amphetamine experiment gave 0.45 mg/kg orally of d-amphetamine, a specified enantiomer of amphetamine, with placebo capsules containing glucose. The source reports a mean participant weight of approximately 74.9 kg, corresponding to a mean calculated administered amount of about 33.7 mg. That figure describes the nominal dose swallowed; it is not a measured circulating drug concentration. This was d-amphetamine, not methamphetamine, and it should not be substituted for an illicit methamphetamine exposure or an unknown prescription formulation.

In the other experiment, the investigators gave an oral 200 mg caffeine capsule versus a glucose placebo in a different within-person sequence. The experimental dose was not determined from a directly assayed serum, plasma, or venous whole-blood caffeine concentration. Cross-study comparisons require caution: the two groups, treatment clocks, drug mechanisms, and potentially baseline characteristics were not matched through a shared head-to-head randomization.

Which times and memory-delay conditions were actually tested?

Following d-amphetamine administration, spatial-span testing occurred about 90 minutes after dosing, with digit-span testing about 210 minutes after dosing. In the caffeine experiment, those tasks occurred about 60 and 120 minutes after dosing, respectively. The original Methods specify memory delays of zero, four, six, and eight seconds. The abstract instead summarizes a different set of delays; the task Methods and reported analyses are the appropriate basis for detailing the administered protocols.

Those times are experimental observation schedules. They do not establish when an individual would become safe or unsafe to drive, when a forensic specimen would be positive, or when a drug effect would end. Likewise, an eight-second retention delay is a cognitive manipulation within a memory task, not a post-ingestion time or an elimination half-life.

What is a forward spatial-span task?

Participants watched an examiner indicate successive locations among ten squares, then attempted to reproduce the same sequence. The task became harder as the sequence length increased. The researchers scored performance both as a maximum span and as the total number of correctly completed trials. Under delay conditions, the volunteer waited before reconstructing the indicated positions. This assesses a narrow form of visuospatial temporary storage and reproduction, rather than the full range of executive function, navigation, planning, or operational situational awareness.

The digit-span task similarly required people to reproduce spoken numbers in their original order. A forward digit span largely stresses immediate verbal storage; it does not require the same manipulation as reversing or reordering numbers. The investigators deliberately examined both spatial and verbal content, recognizing that working memory is not a single uniform faculty. Neither task alone is a validated intoxication test or a measure of professional competence.

Did oral d-amphetamine improve either form of memory?

No statistically significant main effect was demonstrated for these particular forward span outcomes. For the 20-person amphetamine experiment, the reported spatial-span maximum comparison was F(1,19) = 0.78, p = 0.38, and spatial-span total was F(1,19) = 1.6, p = 0.22. The verbal digit-span maximum was F(1,19) = 0.27, p = 0.60, and digit-span total was F(1,19) = 0.05, p = 0.80. Corresponding drug-by-delay interactions were not significant in either task. These are measured null statistical findings, not absent measurement.

It would be inaccurate to rewrite these findings as either “amphetamine invariably impairs memory” or “amphetamine has no memory effects.” The study instead lacked sufficient evidence for improvement on these tested endpoints at these doses and schedules. Other cognitive tasks, populations, timing windows, drug formulations, sleep states, or statistical power could yield different findings. The current experiment cannot adjudicate those broader hypotheses.

Did caffeine improve spatial or verbal span?

The separate caffeine cohort also showed no statistically significant improvement on the designated span indices. Spatial-span maximum was F(1,19) = 1.0, p = 0.30 and spatial total F(1,19) = 1.1, p = 0.30; verbal digit-span maximum F(1,19) = 1.4, p = 0.24 and verbal total F(1,19) = 1.2, p = 0.28. The paper additionally reported no significant caffeine-by-delay interaction for these span tasks.

These results are consistent with an important task distinction. Caffeine can produce alerting or cardiovascular effects without necessarily changing the number of spatial positions or spoken digits reproduced correctly. A span score is not identical to simple reaction time, sustained vigilance, or subjective wakefulness. The experiment therefore does not establish that caffeine was physiologically inert. Its directly measured blood-pressure response points the other way.

First source-grounded evidence table: memory and physiological tests

The table deliberately separates significant physiological outcomes from nonsignificant memory tests. A pulse or cuff-pressure change cannot convert a negative task comparison into cognitive enhancement or impairment. Similarly, a task result does not supply the missing analytical chemistry necessary to identify drug concentration in a forensic specimen.

What were “psychosis-like experiences” in this experiment?

Researchers assessed a group of five psychological scales addressing unusual perceptions, beliefs, positive-symptom descriptions, or broader psychiatric ratings: the Brief Psychiatric Rating Scale, Perceptual Aberration Scale, Magical Ideation Scale, revised Launay–Slade hallucination scale, and Scale for the Assessment of Positive Symptoms. They converted individual scale results into standardized values and combined them into a summary psychosis-like-experience index. The composite captures questionnaire and interview variation within a research protocol.

The scales are not proof that any volunteer developed a psychotic disorder. They should not be interpreted as independently adjudicated hallucinations, delusions, legal incapacity, dangerousness, or impaired witness reliability. Even a real drug-associated shift in a group-average composite cannot identify the state of a different person in another setting. A research-screen score is not a psychiatric diagnosis.

Did d-amphetamine significantly raise psychosis-like ratings?

Yes, on the combined score. The d-amphetamine-versus-placebo comparison for the composite psychosis-like-experience index yielded a Wilcoxon statistic V = 199 and p approximately 0.00048 (rounded in the abstract to 0.0005). Among the component scales, the authors reported significant increases in Brief Psychiatric Rating Scale, Perceptual Aberration, and Magical Ideation scores. Not every component changed significantly: the Launay–Slade and positive-symptom scales did not show a statistically significant drug effect in the reported comparisons.

The source’s printed descriptive values provide useful scale context. For example, the Brief Psychiatric Rating Scale averaged 26.4 ± 2.7 under d-amphetamine and 23.7 ± 1.1 under placebo. The Magical Ideation Scale averaged 4.6 ± 4.0 versus 3.5 ± 2.8, and the Perceptual Aberration Scale 4.1 ± 4.0 versus 2.5 ± 2.8. These descriptive means and standard deviations are not patient-specific thresholds and cannot be mechanically translated into a clinical psychosis incidence rate.

Did caffeine raise psychosis-like ratings?

In the other 20-person cohort, the combined psychosis-like score did not differ significantly between 200 mg caffeine and placebo (V = 106, p = 0.98). The authors also reported nonsignificant caffeine comparisons on the component questionnaires. This is an actual tested null for the specified scale endpoints, unlike blood-drug concentrations or real driving, which were not measured at all.

The two negative and positive outcomes should not be overextended into claims about the comparative likelihood of psychosis after all possible doses. The caffeine cohort was small, administered one selected dose, and evaluated brief psychological phenomena in healthy volunteers. The d-amphetamine cohort did not prospectively determine psychotic-disorder incidence. Neither experiment was designed to estimate rare adverse-event probabilities in clinical or general populations.

Second source-grounded evidence table: psychological measurements

The table combines descriptive scales and statistics without presenting them as an individual risk model. The correlation coefficients arise from the amphetamine experiment’s within-person change scores and should not be extrapolated to a different drug, dose, or patient. The psychological constructs in these rows differ from overt delusions, observed disorder onset, or behavior during a legally relevant event.

What does the eight-second negative correlation establish?

Within the d-amphetamine group, volunteers with larger increases in psychosis-like ratings tended to have less favorable changes in spatial span at the longest eight-second retention interval. The authors reported r = −0.58 with p = 0.006 for that association. By contrast, the analogous change in verbal digit span at eight seconds was r = −0.01 with p = 0.96. The distinction matters: the significant finding was not a uniform relationship between the psychosis-like index and every form of memory.

Correlation is not an experimentally identified causal pathway. Both ratings and span changes followed stimulant-versus-placebo conditions, but the participants were not separately randomized to levels of psychosis-like experiences; no intervention selectively manipulated that intermediate variable. Baseline traits, measurement variability, expectancy, shared drug sensitivity, and the multiple analyses can affect associations. In an evidentiary setting, it is defensible to describe the correlation and its uncertainty, not to assert that transient unusual beliefs caused a measured memory deficit.

Why can a drug change blood pressure without changing memory span?

Different endpoint systems respond differently. Blood-pressure cuffs and pulse measurements capture peripheral physiology. A span task requires perception, short-term retention, attention to instructions, and reproduction of a sequence. A drug may produce a reliable autonomic effect while the selected cognitive endpoint shows no significant average difference. These outcomes can coexist without contradiction.

The amphetamine experiment found increases in systolic pressure, diastolic pressure, and heart rate. The caffeine experiment found increases in systolic and diastolic pressure but not a significant heart-rate effect. That pattern confirms measured physiological changes under the study conditions. It does not validate a blood-concentration estimate, because heart rate and blood pressure are not calibrated analytical assays of parent amphetamine, caffeine, or their metabolites.

What biological specimen and chemical assays were used?

None was reported as a quantitative drug-concentration specimen for this trial. The paper describes nominal oral doses, task performance, psychological questionnaires/interviews, and noninvasive physiological measurements. It does not present an analytically validated measurement of parent d-amphetamine, caffeine, or metabolites in venous whole blood, separated plasma or serum, urine, oral fluid, hair, vitreous, human exhaled breath, or tissue. There is therefore no assay instrument, detection limit, quantification limit, chemical specimen unit, or patient-level time-concentration series to report for administered drug.

A forensic venous whole-blood concentration cannot be derived reliably from a nominal mg/kg dose alone. Plasma or serum drug measurements, had they existed, would require their own identity and analytical validation; they should never silently be relabeled “whole blood.” Likewise, a published background statement about expected peak plasma timing is not a plasma measurement performed in these 40 volunteers. The original data cannot calibrate a roadside cutoff or answer what concentration a hypothetical individual had at a particular time.

For background on why physical specimen type and analytical identification determine what a test actually supports, the independently published biological matrices guide and confirmatory drug testing guide provide additional context. Neither guide changes the unmeasured status of biological drug concentrations in Kassim’s experiment.

How should the short observation window affect interpretation?

Neither session followed every potential stimulant effect over a complete day. The selected cognitive tasks occurred at distinct post-dose intervals, with physiological measurements made at several additional times. Consequently the findings apply to that protocol, not to all phases of absorption, distribution, metabolism, and elimination. The experiment also did not develop a full human pharmacokinetic curve or a parent/metabolite exposure-response model. Doses prescribed or taken in a separate case may involve different schedules, tolerance, formulation, coexposures, medical conditions, and sleep histories.

Forensic toxicology separates measured concentration from measured function. A clinical discussion may use laboratory concentrations alongside treatment context, as described in the published therapeutic drug monitoring overview, but this study itself did not obtain the needed chemistry. Those missing observations should be named as absent evidence rather than assigned statistical insignificance.

Can the results be generalized to driving, testimony, or criminal responsibility?

No case-level threshold or validated individual performance classification was generated. Spatial-span and digit-span tasks differ from maintaining lane position, responding to unexpected road hazards, performing a standardized field sobriety assessment, recalling a witnessed event, or applying legal standards of responsibility. Nor did the investigators directly observe any of those events. Claiming the study proves an individual was impaired, unimpaired, psychotic, truthful, violent, or responsible would exceed its design.

The defensible forensic use is narrower: the trial demonstrates that a stimulant can yield strong questionnaire and cardiovascular findings without significant improvements on selected forward-memory outcomes, and that drug-sensitive research measures are not interchangeable. An expert can explain the group design, the actual statistics, and why the absence of individual drug chemistry or validated case performance prevents a stronger attribution.

What limitations and alternative explanations matter most?

Small sample size is central: each drug comparison had 20 participants, and the study was exploratory. Although placebo comparison and blinding mitigate some biases, a nonsignificant score could reflect limited statistical precision, ceiling effects, or the particular forward-span metric. Broad cognition and genuine daily-life executive function extend beyond these two instruments. Baseline habits, prior stimulant exposure, sex composition, and participant expectations may influence generalizability despite the controlled protocol.

Multiple scale components, cognitive delays, outcome scores, and correlations were tested. An isolated significant association warrants caution even when its nominal p-value is low, especially without preregistered individual mediation analysis or external replication. The paper’s abstract and detailed Methods also do not state identical delay lists; the detailed procedures describe zero, four, six, and eight seconds. The appropriate response is to preserve the full source context rather than silently invent a harmonized study design.

The two cohorts are not a random direct comparison between d-amphetamine and caffeine. Any seeming difference across arms remains confounded by distinct subjects, drug doses, task clocks, and histories. Finally, the investigation contains no objective relapse, incident, or legal blood-threshold endpoints. These limits do not negate the genuine group-level findings; they define the scientifically supported perimeter around them.

Frequently asked forensic-science questions

1. Did all 40 participants take both stimulant drugs?

No. Twenty separate volunteers underwent d-amphetamine-versus-placebo crossover testing, while a different twenty underwent caffeine-versus-placebo crossover testing. The two cohorts were not randomized to one another’s active drug.

2. Did 0.45 mg/kg d-amphetamine significantly improve spatial memory?

Not on the source’s tested forward spatial-span maximum or total scores: p = 0.38 and p = 0.22. Those nonsignificant results cannot prove a zero effect for every task or individual.

3. Did 200 mg caffeine significantly improve digit-span performance?

No significant caffeine-related increase was demonstrated for the forward digit-span maximum (p = 0.24) or total correct score (p = 0.28) in this 20-person protocol.

4. Does the increased psychosis-like composite mean d-amphetamine caused psychosis?

No. It shows a group-level increase in research questionnaire/interview ratings. The trial did not establish new psychotic-disorder diagnoses, an individual propensity to violence, or legal incapacitation.

5. Did psychosis-like ratings cause poorer spatial memory at eight seconds?

The change scores were negatively correlated (r = −0.58, p = 0.006), but the investigators did not independently randomize or manipulate the putative intermediate psychological ratings. Causation or clinically validated mediation was not demonstrated.

6. Can the oral dose be converted into a forensic whole-blood concentration?

Not from this experiment. No quantitative parent-drug or metabolite assay in venous whole blood or a validated alternate human matrix was reported. Dose and measured drug concentration are different evidence types.

7. Can this study establish that a particular driver or witness was impaired?

No. No road or simulator driving, individual case performance, blood threshold, event reconstruction, or eyewitness-validity endpoint was measured. The paper supports bounded group-level physiological, questionnaire, and forward-span findings only.

Scientific bottom line

The two original crossover pilots show a useful dissociation: oral d-amphetamine increased peripheral physiology and psychosis-like questionnaire scores, yet did not significantly improve measured forward spatial- or digit-span memory; separately, oral caffeine increased blood pressure without demonstrating span enhancement or an increased psychosis-like composite. The significant amphetamine-associated eight-second spatial-span correlation invites further investigation but does not establish a diagnosis, causal mediation, or forensic impairment determination. These are empirical observations in 20-person groups, not direct quantitative human drug-concentration measurements. The original complete primary study and its actual definitions remain controlling: Kassim et al., PLOS ONE (2023), DOI 10.1371/journal.pone.0287538.