Can Evening Methamphetamine Disrupt Sleep and Next-Day Cognitive Flexibility? Evidence From Five Rhesus Monkeys

What question did the experiment actually answer?

The investigators asked whether evening methamphetamine treatment, at doses previously shown to interfere with sleep in rhesus monkeys, would be associated with poorer performance on a complex learning and cognitive-flexibility task during the following day. The question was not whether methamphetamine directly produces a particular blood concentration or whether every cognitive domain becomes impaired. It concerned a sequence: administer a drug in the evening, observe activity patterns during the subsequent dark period, and assess trained touchscreen behavior later.

That sequence is scientifically important because a next-day change after stimulant exposure can reflect several overlapping pathways. The drug may affect arousal, sleep, motivation, learning, or task performance. A sleep–performance correlation is compatible with a sleep-related pathway but cannot, by itself, prove that sleep disruption mediated all observed behavioral changes. Nor can an animal task score be automatically translated into a courtroom conclusion about a person.

How many independent subjects were studied?

Five adult rhesus macaques were studied: four males and one female, with reported weights between 9.5 and 11.7 kg. This is the biological sample size. It is not 15 monkeys merely because several dose conditions were considered, and it is not hundreds of monkeys because the animals completed many touchscreen trials. The same animals contributed repeated baseline, drug, sleep, and cognitive observations. The study design therefore exploits within-animal comparisons but offers a limited number of independent biological units.

All five animals were experimentally naïve when enrolled and learned the touchscreen procedures before experimental testing. They were individually housed under a 12-hour light/12-hour dark schedule, with lights on at 6 a.m. and darkness beginning twelve hours later. This controlled schedule makes the evening administration interval interpretable within the animal facility; it does not recreate the environmental, medication, sleep, or substance-use histories of any particular human.

What was administered, and how?

The tested compound was (+)-methamphetamine hydrochloride, obtained from the National Institute on Drug Abuse, dissolved in physiological saline, and administered intramuscularly at nominal doses of 0.3 and 0.56 mg/kg. Injections occurred at zeitgeber time 9, three hours before lights off at zeitgeber time 12. The order of the two methamphetamine dose conditions was randomized across monkeys. This was neither oral dextroamphetamine nor lisdexamfetamine, and the injected salt identity and route should not be replaced with those other stimulants when summarizing the experiment.

The paper describes repeated evening injections during the relevant testing phases, rather than one universal single-dose exposure followed by a one-time cognitive examination. Each drug condition had its own nearby baseline testing period. The investigators cited earlier rhesus-monkey pharmacokinetic work when reasoning that appreciable direct drug action would be unlikely by the next cognitive session; that reasoning is not a reported chemical assay of the five animals’ blood the next morning.

How were baseline, dose order, and potential practice effects handled?

An initial task session served to familiarize the monkeys with the test and was not included as an analyzed baseline. Two later baseline test sessions were each followed, after a separation interval, by a methamphetamine condition. The investigators left approximately two months between successive testing periods in an effort to reduce carryover of learned task solutions. The order of the two drug doses was randomized, while the baseline-before-treatment structure was retained.

This arrangement is useful but not equivalent to an independently randomized placebo-versus-drug parallel trial. Repeated exposure to the task can still change how an animal solves it. The paper reported that comparisons between the two baseline sessions were not statistically significant for its listed task outcomes. A nonsignificant baseline comparison in five animals does not prove an absence of learning, but it is relevant evidence against simply attributing every later difference to gross baseline drift.

What was measured overnight: sleep or physical movement?

Collar-mounted MotionWatch-R actigraphy sensors registered activity during the 12-hour dark period. Software converted movement records into estimates of sleep latency, sleep efficiency, percent wake, and sleep fragmentation. For this experiment, sleep efficiency is the percentage of the dark period classified as sleeping or inactive; sleep latency is the interval between lights off and the first immobile bout. Percent wake and the fragmentation index are similarly derived from movement-defined activity or immobility.

Those outcomes are valuable for within-animal comparisons but remain actigraphy-based sleep proxies. The article did not report polysomnography, electroencephalographic sleep stages, or a chemical assay of methamphetamine as part of this overnight protocol. In particular, a movement-defined inactive period is not automatically interchangeable with an electrophysiologically scored sleep stage. A collar monitor also cannot determine what a human defendant subjectively experienced on a different night.

Which sleep changes were significant?

The clearest result was dose-dependent. At 0.56 mg/kg, all four group-average actigraphy endpoints differed significantly from baseline in the paper’s Bonferroni post-hoc comparisons. The 0.3 mg/kg averages often moved in the same direction, but those lower-dose comparisons did not meet the specified significance criterion against baseline. That distinction prevents the inaccurate shorthand that either administered dose established statistically significant sleep disruption on every measure.

The following original evidence table selects the study’s central actigraphy values and reorganizes them by the inferential question, rather than reproducing the publisher’s table layout. Entries are mean ± standard error of the mean across the five monkeys; they are not individual sleep records or confidence intervals.

The reported repeated-measures analysis detected a treatment effect for latency (F[2,8] = 4.925), efficiency (F[2,8] = 7.885), wake percentage (reported F[4,16] = 10.23), and fragmentation (F[2,8] = 8.888). The unusual reported degrees of freedom for percentage wake should be preserved as printed, not silently harmonized with the other three. A separate private scientific review documents this reporting irregularity and the small-sample implications.

How was next-day cognitive flexibility examined?

The monkeys completed a touchscreen intradimensional/extradimensional (ID/ED) attentional set-shifting task adapted from the Cambridge Neuropsychological Test Automated Battery. Different stages required discovering which visual object predicted a food reward, reversing a previously successful choice, ignoring irrelevant visual features, and eventually shifting attention to a new stimulus dimension. Correct choices were rewarded with small food pellets, while incorrect or missing responses carried different timeouts.

Eight named stages were assessed: simple discrimination, its reversal, compound discrimination, its reversal, intradimensional discrimination and reversal, and extradimensional discrimination and reversal. The broad label “executive function” conceals these distinct task demands. Earlier stages depend heavily on acquiring or reversing stimulus–reward contingencies; later stages place additional demands on changing an attentional set. It matters where errors occur, not just whether a total score moves.

What constituted an error, and what does task accuracy mean?

The investigators assessed trials required to complete the task, response omissions, percent accuracy, total errors, and perseverative errors. Perseveration is especially relevant during reversal learning: an animal continues selecting a response that previously earned a reward after the rule changes. More perseverative responses indicate difficulty adapting to the changed contingency in this experimental setting, not a psychiatric diagnosis or a validated measure of dangerous real-world decisions.

The main dose-to-dose cognitive table normalized each drug condition to its corresponding baseline, so many reported values are percentages of that condition’s baseline performance. A value such as 201.53 in a normalized trial row does not mean the animal literally completed 201.53 trials in that stage. Likewise, large percentages arising from small baseline counts can look dramatic while carrying substantial variability. Preserving that denominator distinction is essential for an accurate quantitative summary.

Did the higher evening dose worsen overall next-day performance?

Compared with its own baseline, 0.56 mg/kg was associated with significantly greater total trials required, total errors, and total perseverative errors. In the 0.3 mg/kg condition, total perseverative errors also differed significantly from its baseline. When the two drug conditions were compared directly after baseline normalization, the higher dose produced worse overall trials, average accuracy, total errors, and perseverative errors at the paper’s reported significance threshold.

The finding is not “all functions impaired.” The overall omission comparison between 0.3 and 0.56 mg/kg was nonsignificant (reported P = .71), even though other outcomes moved. The investigators also reported no significant change in the number of days needed to finish the whole ID/ED protocol relative to the respective baselines. Those outcomes belong alongside positive findings so that the result is not presented as a uniform cognitive collapse.

Which specific task stages drove the results?

The strongest dose-related differences appeared early, during simple-discrimination reversal and compound discrimination. Reversal errors are behaviorally different from simple response slowness: they may reflect difficulty abandoning a previously reinforced choice after contingencies change. Compound discrimination adds irrelevant visual features and tests whether subjects focus on the rewarded ones. Results in those stages support a narrower description than saying that every kind of attentional set-shifting failed.

The dose comparison was often not statistically significant during the later extradimensional stages. In a small repeated-measures animal experiment, a nonsignificant comparison can reflect limited power, individual variation, or an effect smaller than the available evidence can resolve. It does not prove equal performance. The table below uses selected original normalized group summaries with their reported paired comparisons, mixing positive and negative rows to make that balance explicit.

For some stage outcomes the paper reports P values at or near .05, and multiple stage-level comparisons were examined. The displayed significance statements follow the paper’s specified tests; they should not be recast as independent confirmatory replications or precise effect sizes. Table 2 values represent the same five animals under different experimental conditions, not different groups of five.

Was reaction speed reduced as an alternative explanation?

The authors reported no significant changes in the response-latency analysis for 0.56 mg/kg relative to baseline across the task stages, and they did not notice overt motor impairment while staff handled and chaired the monkeys. This is useful negative evidence against a simplistic explanation that all task deficits reflected inability to touch the screen quickly. It is not proof that subtle movement changes, fatigue, attention, or motivational factors were completely absent.

Task accuracy, perseveration, and latency have different meanings. A subject can respond at roughly the same speed while choosing the wrong stimulus more often. That distinction is central both to the behavioral interpretation and to forensic restraint: stable touchscreen response latency under the tested conditions is not a measured normal driving reaction time, and an altered error rate is not a traffic-collision probability.

Did sleep–cognition correlations prove mediation?

Across baseline and drug conditions, higher movement-derived wake percentage correlated with more trials, more total errors, more perseverative errors, and poorer accuracy. Lower estimated sleep efficiency correlated with poorer accuracy and more perseverative errors. These directions fit the idea that disturbed sleep and altered next-day performance are linked in this laboratory regimen. They are an internally coherent pattern, not a formal demonstration of the mechanism.

The correlations pooled observations from the same five monkeys across repeated conditions. Repeated observations can be statistically dependent, and no independent randomized sleep-restoration intervention isolated sleep from all other consequences of methamphetamine exposure. Therefore, the correlations should not be described as causal proof that methamphetamine had no residual direct or other indirect effects on cognition. Pooling conditions can also make a relationship partly reflect the common influence of treatment.

Which sleep–cognition relationships were not significant?

The source separately identified many null correlation tests. For example, total errors versus sleep fragmentation had r = .09424 with P = .7099; total errors versus sleep efficiency had r = −.3371 with P = .1713; trials to completion versus sleep efficiency had r = −.3206 with P = .1946. Mean accuracy versus sleep latency had r = −.1200 and P = .6352. These negative tests matter: the association was selective across the variables examined, not universal for every measure of actigraphy and every behavioral output.

“Not significant” does not mean an effect was proved to be zero. With five biological subjects and multiple related tests, statistical estimates can be unstable and more uncertain than a single P value conveys. The article’s correlation table also pairs an apparent r value with an R-squared number that is not consistently its mathematical square for every displayed entry, making source-level transcription and interpretation a matter for private methodological quality control rather than public overconfidence.

Were both baseline cognitive tests identical?

No two behavioral sessions will be literally identical. The paper compared its second and third baseline periods and reported that none of the listed comparisons reached the conventional significance threshold. For example, total trials across stages averaged 602.50 ± 198.35 in one baseline and 332.00 ± 80.39 in the next (P = .12). Mean accuracy averaged 60.74 ± 2.80% and 65.75 ± 2.66% (P = .13). These numbers show a possible direction of practice improvement even without formal significance.

Because the authors deliberately separated testing periods by approximately two months and normalized drug-condition measures against the corresponding nearby baseline, they attempted to reduce this problem. Those design safeguards are meaningful; they do not convert a repeated task into an entirely fresh, independent biological experiment on each occasion. In any legal or scientific report, the baseline-control architecture should appear beside the claimed dose effect.

Could the results be generalized to humans receiving methamphetamine?

Only cautiously, as a hypothesis-generating animal model. The subjects were five healthy, laboratory-housed rhesus monkeys, not people with heterogeneous sleep schedules, substance-use disorders, medications, sleep apnea, mental-health diagnoses, or other clinical conditions. Methamphetamine was given by a controlled intramuscular route at specified mg/kg doses. Human exposure commonly occurs through other routes, and species differences affect absorption, disposition, and behavior.

The study makes a limited inference about next-day performance on one specialized touchscreen test in this animal setting. It did not enroll human drivers, operators, injured persons, witnesses, police interviewees, or patients. It contains no clinical psychosis diagnoses and no on-road or simulator outcome. Consequently, a human case requires its own properly collected drug evidence, sleep history, medications, timing, behavioral observations, and other material facts; one cannot derive those missing facts from an animal group mean.

Was methamphetamine chemically measured in a forensic blood specimen?

No. The administered exposure was an intramuscular quantity of methamphetamine hydrochloride expressed in mg/kg body weight. The paper did not report a quantitative assay of parent methamphetamine or amphetamine metabolite in these animals’ blood, and it did not analyze legally collected human venous whole blood, separated plasma or serum, urine, oral fluid, hair, sweat, vitreous humor, or exhaled breath for the drug. There is no study-specific LC–MS/MS or GC–MS concentration, lower limit of detection, lower limit of quantification, or validated conversion from a biological matrix into a whole-blood value.

The actigraph measured collar acceleration, and the touchscreen measured choices and response behavior. Neither is a chemical toxicology instrument. A body-weight-adjusted injected dose is not interchangeable with a circulating concentration, and a literature-derived pharmacokinetic half-life used for scheduling is not an observed quantitative analyte result from an individual test subject. Readers should preserve the distinction between administered dose, unmeasured internal exposure, and observed behavioral outcome.

What would a human forensic toxicology comparison require?

A defensible human comparison begins with the actual specimen and analytic method. A properly identified legal venous whole-blood result is different from a centrifuged plasma or serum result; even validated measurements in one matrix cannot simply be relabeled as values in another. Urine, oral fluid, hair, sweat, vitreous, and breath answer different exposure or detection questions. Chain of custody, sample handling, confirmation criteria, quantitative range, uncertainty, and time relative to the alleged conduct also matter.

The separately published guide to biological matrices in forensic toxicology explains why specimens cannot be treated interchangeably. The live guide to confirmatory drug testing distinguishes analytic identification and quantitation from broader inferences about effect. Neither guide turns this animal study into a blood-drug reference table; they provide methodological context for deciding what an actual human laboratory record does and does not establish.

What are the most important limitations?

The biological sample contained five monkeys and only one female. The experiment measured activity-based sleep proxies rather than electrophysiologically verified sleep stages. Drug and baseline periods occurred in a repeated sequence over months; practice, time, habituation, and individual variability could influence performance. The study assessed many sleep, overall cognition, task-stage, and correlation endpoints, creating a need to interpret its collection of P values together rather than selecting only the most striking result.

Also, the experimental schedule makes a sleep-related mechanism plausible but does not separately manipulate sleep to establish causal mediation. The high-dose drug exposure was repeated during the relevant task phases. The source’s reported wake-percentage ANOVA degrees of freedom differ from those of the other sleep parameters; some correlation-table R-squared values should be checked against the printed r values. These are appropriate private-review flags, not reasons to invent corrected calculations or discard all data. Finally, no individual forensic chemical concentration or legally relevant human outcome was measured.

What conclusions survive a conservative reading?

A restricted but meaningful conclusion survives: in five trained rhesus monkeys, a higher evening intramuscular methamphetamine dose produced statistically significant movement-based sleep disruption and was associated with poorer performance on selected elements of a next-day touchscreen set-shifting task. The lower dose did not significantly change the principal group-average sleep endpoints, and multiple cognitive stages, omissions, response latencies, and sleep–behavior correlations were not significant. This is evidence of a dose- and task-dependent pattern in an animal laboratory, not a universal demonstration of cognitive impairment at any methamphetamine exposure.

It remains possible that disrupted sleep contributed to task deficits, but the data do not isolate sleep as their only cause. They cannot specify a legal whole-blood threshold, diagnose an individual patient, identify the time of a particular ingestion, show whether a person safely drove, or determine responsibility for a specific event. Evidence should be characterized at the level actually measured and no further.

Frequently asked scientific and forensic questions

1. Did the investigators study five monkeys or many independent animals across conditions?

Five independent adult rhesus monkeys—four males and one female—participated. They contributed repeated measurements under baseline and two drug-dose conditions. Extra nights, cognitive stages, and touchscreen trials do not create extra independent monkeys; the main paired comparisons therefore have a biological N of five.

2. Was the injected drug dextroamphetamine or methamphetamine?

It was (+)-methamphetamine hydrochloride, administered intramuscularly at 0.3 or 0.56 mg/kg. It should not be rewritten as oral dextroamphetamine, mixed amphetamine salts, methylphenidate, or lisdexamfetamine. Doses describe the administered salt and route; no blood concentration was measured.

3. Did both dose levels significantly worsen overnight sleep?

No. The 0.56 mg/kg condition significantly differed from baseline on the four specified movement-derived measures after the authors’ stated post-hoc procedure. The 0.3 mg/kg condition had descriptive changes but did not significantly differ from baseline on those primary sleep measures.

4. Did every part of the next-day cognitive task become impaired?

No. Several significant differences concentrated in early discrimination and reversal stages, along with some overall error and accuracy measures. Multiple other task-stage contrasts, overall omissions, and measured response-latency comparisons were nonsignificant. The study supports task-dependent effects, not blanket deterioration across every cognitive domain.

5. Can a correlation between wake time and errors prove sleep caused all errors?

No. Such a correlation is consistent with a sleep-related contribution but is not a complete causal mediation test. The same five animals appeared across repeated conditions, and methamphetamine could affect performance through additional pathways. Some sleep–behavior correlations were also nonsignificant.

6. Does 0.56 mg/kg correspond to a known human venous whole-blood methamphetamine concentration or DUI cutoff?

No. This was a nominal intramuscular animal dose, not an analyzed human legal whole-blood concentration. The study supplied neither measured parent-drug levels nor a validated matrix conversion or human forensic impairment threshold. No human drivers were tested.

7. Can this paper establish the cause of a person’s collision, unreliable statement, or injury?

No. The experiment did not observe such an event or study the person concerned. Its appropriate contribution is limited to animal-model evidence about sleep-related activity estimates and next-day touchscreen behavior. Individual case causation requires independent case-specific records and scientifically justified analysis.

Original research citation and scope

Borgatti DA, Rowlett JK, Berro LF. Effects of methamphetamine-induced sleep disruption on intradimensional/extradimensional attentional set-shifting in adult rhesus monkeys. Journal of Pharmacology and Experimental Therapeutics. 2025;392(10):103693. Original primary DOI: 10.1016/j.jpet.2025.103693. This independent interpretive article is based on the full primary research and does not reproduce the publisher’s figures, tables, or extended text. Its findings remain specific to the experimental model and endpoints actually studied.