Does Intranasal Methamphetamine Improve a Feeling of Control Without Improving Every Task? A 10-Person Study

Short answer: In a four-session, within-person experiment involving 10 adults who were current methamphetamine users, the lowest supervised intranasal methamphetamine dose, 12 mg per 70 kg body weight, improved performance on a computerized target-catching task and sharpened participants’ judgments about when they controlled the computer mouse. The larger 25- and 50-mg doses did not demonstrate the same overall improvement. They increased one form of commission error when the computer deliberately distorted mouse control, while several other task contrasts remained nonsignificant. Crucially, the investigators did not chemically measure postdose methamphetamine in forensic venous whole blood or establish any threshold for real-world impairment, legal culpability, driving, or witness reliability.

The original experiment is Kirkpatrick MG, Metcalfe J, Greene MJ, Hart CL, Effects of intranasal methamphetamine on metacognition of agency, Psychopharmacology 197:137–144 (2008), DOI 10.1007/s00213-007-1018-2. The distinctions below follow the actual original human study, rather than extrapolating from general stimulant pharmacology, an abstract, or a different experiment using oral amphetamine.

What does “metacognition of agency” mean?

Agency is a person’s experience of controlling an action or its result. Metacognition of agency is narrower: it asks whether the person accurately recognizes how much control they actually had. Someone can move a mouse quickly yet be mistaken about whether the device responded predictably. Conversely, someone may accurately recognize that an unpredictable computer was interfering even while missing moving targets. Treating these two outcomes as synonyms erases the scientific question.

The investigators deliberately altered the responsiveness of the computer mouse and then asked participants to judge their feeling of control. This is a tightly specified operational task. It is not a test of free will, moral agency, the truthfulness of a statement, ability to control an automobile, or legal responsibility during an alleged incident. The participants’ judgments were numerical task responses in an artificial environment, not validated opinions about real-world behavior.

Who actually participated, and was the sample independent?

Ten adults completed all four inpatient sessions: eight men and two women, with a mean age of 30.9 years (standard deviation 6.7). All met the original study’s DSM-IV criteria for a current methamphetamine-use disorder and were not seeking treatment. They reported methamphetamine use on about four days per week and an average history of 6.4 years. The authors also documented prior or concurrent use of other substances in this group, including tobacco, cannabis, cocaine, and alcohol.

Two additional men entered the study but did not contribute a complete four-session dataset. The biological study N is 10, not 12 completed subjects, 40 separate people, or 960 separate people. Four exposures per person, repeated target presentations and multiple ratings cannot be treated as additional independent human volunteers. The small and selected inpatient group sharply limits generalization to a community sample, a prescribed-medicine patient, an occasional user, or anyone in an actual forensic case.

Was the administered compound methamphetamine, dextroamphetamine or lisdexamfetamine?

The investigators administered methamphetamine hydrochloride powder intranasally, supplied through a regulated research arrangement and prepared by the research pharmacist. This is not ordinary dextroamphetamine, a mixture of d- and l-amphetamine salts, methylphenidate, or the lisdexamfetamine prodrug. All have distinct molecular identities or formulations, so a different stimulant study should not be quietly substituted for this one.

The original methods do not report a chiral forensic assay resolving d- versus l-methamphetamine in the participants’ specimens. It is appropriate to identify the administered research preparation as methamphetamine hydrochloride; it is not appropriate to invent a chemically measured d-isomer percentage or a participant-specific d/l ratio. Nor does milligrams of administered drug establish the concentration that was in someone’s blood at the task time.

What doses were administered, and how was sequence controlled?

Each person received intranasal placebo and methamphetamine at nominal doses of 12, 25 and 50 mg per 70 kg on separate blinded visits; each condition occurred within the same participant, and dosing order was counterbalanced. The four sessions occurred over approximately two weeks with at least 72 hours separating drug administrations. Drug and lactose powder were prepared to a common powder weight for blinding. Study personnel asked participants to insufflate the material within about 30 seconds.

Those values express intended administered mass normalized to a reference body weight, not recovered dose at the nasal mucosa, absorbed bioavailability, peak plasma concentration, or analyzed whole-blood concentration. A maximum single dose of 60 mg was imposed for safety. The cognitive task began about 45 minutes after administration. The investigators did not publish a postdose specimen-by-specimen methamphetamine concentration series for this metacognition experiment, so no valid one-to-one dose-to-blood calculation follows from these details.

How did the researchers distinguish real control from perceived control?

Participants operated an Apple computer mouse to intercept falling X targets and avoid O distractors. The task manipulated three factors: turbulence of mouse responsiveness, speed of descending stimuli, and a “magic” feature that awarded target hits even when the mouse approached within 10 pixels without directly contacting the X. Manipulating turbulence provided a situation where the computer, not just the participant, altered actual movement control.

The task was a 2 × 2 × 2 factorial arrangement with three repetitions per cell, giving 24 trials per laboratory visit. On each trial the experimenters recorded target hits and errors of commission; afterward the participant rated how much control they felt on a visual analogue line. Crucially, a hit rate is a performance endpoint and a control rating is a subjective judgment. The two may be related, but one cannot substitute for the other in a toxicological or psychological inference.

What did the original descriptive control ratings show?

The original paper reported average ratings on a numerical zero-to-one-like scale for the no-turbulence and turbulence conditions, separately for each drug dose. Participants consistently reported more control when the computer followed their hand than when unpredictable noise was inserted. This within-condition discrimination was present even after placebo. The comparison in this table is not itself a test that 50 mg was superior to 25 mg, nor does it provide an individualized calibration curve.

The published means are useful precisely because they prevent overstatement. For example, the no-turbulence rating at 25 mg was 0.558, lower than the 12- and 50-mg values of 0.576, while the turbulence rating at 12 mg was 0.173. The experiment’s more specific significant drug effects concern slow-speed subsets in the inferential analyses below. Reading a descriptive across-condition mean as a universal clinical effect would be misleading.

Did the low dose sharpen the judgment of control?

For the slow-speed, no-turbulence condition, each active drug dose increased perceived control relative to placebo, with the authors reporting p < 0.01 for the planned comparisons. For the slow-speed, turbulent condition—the computer was deliberately less controllable—only the 12-mg methamphetamine dose significantly decreased the control rating relative to placebo (p < 0.01). Those opposite-direction changes satisfy the authors’ narrow operational definition of enhanced metacognitive discrimination.

The distinction is counterintuitive but central: a lower confidence rating when real control is poor can signify better calibration, not simply reduced confidence. Higher apparent confidence on the easy trial is not enough; the subject should also recognize that the device is erratic on hard-to-control trials. The researchers found a dose-by-turbulence-by-speed interaction, F(3,27) = 4.969, p < 0.01. This provides evidence for a context-dependent effect on judgments, not a generalized claim about the drug making people aware, unaware, truthful, reckless, or competent in every situation.

Did the 25- and 50-mg doses produce false feelings of control?

The original hypothesis anticipated that greater stimulant doses might increase perceived control even where control was objectively poor. The results did not support a universal false-agency or overconfidence conclusion. At the larger doses, ratings increased under some optimal low-speed task conditions, but did not show the low-dose pattern of improved discrimination in the deliberately turbulent condition. The authors explicitly reported that no active dose disrupted the agency judgment according to their operational definition.

Failure to find the predicted disruption is an important negative finding. It does not prove such a disruption is impossible, nor does it show all high doses are harmless. With only ten people, one artificial task, one session per condition and no direct whole-blood exposure measurement, the experiment has limited power and generality. The direction and statistical support of each reported interaction—not an intuitive assumption that stimulants must make people overconfident—should control the interpretation.

What actually happened to target-catching performance?

The task measured hit rate separately from judgments of control. Across the dose comparison, the authors reported an improvement in target hit rate only at the 12-mg dose relative to placebo (p < 0.05). Neither the 25-mg nor the 50-mg dose was shown to improve hit rate. The source further reported that no dose interaction with turbulence, speed, or magic was significant for the hit-rate endpoint.

This is not a simple rising dose-response relationship where more methamphetamine reliably means greater performance. A favorable comparison at one dose in a repeated laboratory test does not establish a therapeutic recommendation or an individual performance forecast. Successful movements on a computer screen say little directly about complex traffic scanning, emergency hazard response, divided attention under fatigue, or sustained real-world functioning. The test also does not distinguish improved perceptual sensitivity from every possible motor mechanism.

Where did larger doses worsen performance?

The original commission-error analysis found a significant dose × turbulence × speed interaction, F(3,27) = 5.894, p < 0.01. In the slow-speed condition with imposed turbulence, 25 mg increased the fraction of O distractors not avoided compared with placebo (p < 0.05), and 50 mg increased this error measure with stronger reported significance (p < 0.0001). These errors were conditional—not a documented global deterioration across every trial type.

A commission error is a specific operational event: failing to avoid a distractor in a visual motor task. It should not be renamed an actual criminal error, a clinical inability to inhibit every response, or a driving violation. The authors considered whether hand movement or inhibitory control might help explain the error pattern, but the present study did not isolate and directly prove a neurochemical or muscular cause. Importantly, larger-dose errors were not accompanied by a clearly demonstrated lower hit rate across the whole task.

Which tested effects were statistically nonsignificant?

The negative results matter as much as the positive ones. Under fast-speed conditions, control ratings did not significantly differ across doses according to the source; there was no significant dose-by-magic interaction for agency ratings. Hit rate showed no significant dose-by-task-factor interaction, and the two larger doses did not receive the 12-mg improvement conclusion. Beyond the one turbulent/slow-speed setting, the commission-error analysis disclosed no additional significant interactions.

The original publication did not supply an exact p-value for every null contrast. Those are therefore identified below as “not significant; exact p not reported” instead of fabricating numerical values. A nonsignificant test is not proof of mathematical equivalence, perfect safety, zero physiological effects, or a guarantee against an adverse outcome in another population. Limited N, repeated tests, multiple planned contrasts and unobserved drug concentrations must all temper interpretations.

Why are target hits and commission errors separate endpoints?

The game deliberately asked participants to perform two forms of behavior: approach or touch X targets and refrain from contacting O distractors. A participant might catch more Xs because of faster tracking yet also touch more Os when the mouse is erratic. Combining those into one crude performance label could hide a tradeoff between detection and restraint. The original study correctly treated hit rate, error rate and subjective agency judgment as separate dependent variables.

Each endpoint also operates at a different inferential level. Target catching is measured behavior in a programmed environment; a commission error is a failure on a specified distractor; the agency rating reports the participant’s felt control. The task did not assay executive function in every possible sense. Claims about cognition should therefore retain the name and context of the endpoint on which they depend rather than present one significant contrast as a universal increase in “function.”

Did the results establish a dopamine-based mechanism?

The authors discussed possible nonmonotonic or inverted-U interpretations involving monoamine signaling, particularly dopamine-related cognitive modulation. That discussion describes possible explanations, not a direct biochemical finding from the ten participants. The experiment did not measure dopamine release in brain tissue or cerebrospinal fluid, did not use receptor imaging, and did not experimentally manipulate a dopamine receptor independently of methamphetamine.

Likewise, a proposed excessive-stimulation explanation for the larger-dose error pattern was not validated by direct mechanistic testing. The authors pointed to companion-session subjective measures that did not establish significant increases in feeling anxious, jittery, or on edge, and noted no significant change in certain subjective cognitive complaint ratings. Those companion observations cannot be converted into a dose-specific forensic drug concentration or proof that the participant’s internal state was known with certainty.

Was methamphetamine measured in postdose whole blood or another analytical matrix?

No quantitative postdose specimen data were published in this agency experiment. Screening urine toxicology before enrollment showed that all ten completing volunteers tested positive for methamphetamine; five also screened positive for a cannabis metabolite. Those initial screens helped characterize the population. They were not a serial, enantiomer-specific, confirmatory mass-spectrometric quantitation of methamphetamine during each task visit.

The metacognition study did not provide matched, timed concentrations of parent methamphetamine or amphetamine in legally collected human venous whole blood, centrifuged plasma, separated serum, quantitative urine, oral fluid, sweat, hair, exhaled breath, or vitreous. The nominal amount insufflated is not a substitute for a specimen result. Assumptions about absorption, distribution, elapsed time, tolerance and sampling matrix cannot magically turn the dose label into an individual ng/mL blood measurement.

Could the four doses establish a blood-concentration impairment cutoff?

No. A concentration-based impairment threshold requires adequate chemical exposure measurement, a validated impairment outcome relevant to the intended application, a sufficiently informative population, defensible decision criteria, and uncertainty analysis. This study had four nominal route-specific administrations but no paired quantitative human blood results at the moments of performance. The outcome was computer-mouse agency and target selection, not a legal, medical or occupational determination of impairment.

Neither the p-values from task experiments nor the percentage of targets hit is a blood-methamphetamine cutoff. No method in this source can convert the 12-, 25- or 50-mg/70-kg nominal doses to a safe or unsafe legal blood limit. The appropriate forensic inquiry starts with what was actually measured and the exact question being asked; it does not infer a specimen result, cutoff, or individual psychomotor status from a group-level experimental manipulation.

What do the sample size and design allow us to infer?

The blinded, within-participant, counterbalanced design is a strength: the same people experienced each dose, and differences across conditions were examined without treating every new session as an entirely new group. At least 72 hours elapsed between sessions and participants were trained beforehand. Those features help reduce certain person-to-person and task-learning differences, but do not create a large or representative population.

Ten selected adults who used methamphetamine regularly are not representative of drug-naive people, individuals with medical conditions, prescribed amphetamine users, every sex or age group, or defendants selected because of a traffic stop. The authors themselves acknowledged relying on one experimental agency task, not formally comparing every subjective rating against an independent measurement of actual mouse control, and using isolated doses rather than naturalistic patterns of repeated use. Confidence intervals for many planned contrasts were not reported; do not substitute confidence in the prose for unreported interval precision.

Does an enhanced sense of control mean a person is less impaired?

No. A participant correctly distinguishing a reliable from an unreliable computer interface is demonstrating task-specific metacognitive discrimination. It says nothing necessarily about visual field, complex traffic scene interpretation, hazard anticipation, divided attention, judgment under provocation, fatigue, speed selection or legal capacity during a completely different event. Agency is a construct with different meanings in experimental psychology, philosophy, psychiatry and criminal law.

The lowest dose produced a favorable pattern on two task measurements while larger doses produced a negative pattern in one commission-error condition. That coexistence itself defeats a simplistic categorical claim that methamphetamine either always improves performance or always disrupts self-awareness. Case opinions must address the actual subject, relevant medical and analytical records, direct observed behavior, timing and setting. This small experiment alone cannot determine any person’s degree of impairment.

How should toxicologists and trial lawyers use this primary study?

The study is useful for evaluating an overbroad inference. An expert who asserts that the mere presence of methamphetamine uniformly destroys metacognitive awareness must reckon with original significant low-dose results and null larger-dose agency-disruption tests. Equally, someone who relies on improved target catching to declare that methamphetamine improves driving, makes witnesses reliable or creates a protective effect is crossing the same unsupported boundary in the opposite direction.

A sound expert explanation identifies the precise dose, actual route, exposed population, computerized endpoint, the published statistical test, and the missing chemical matrix and individual criterion. The study may inform a discussion of complex cognitive effects under controlled conditions. It should not be used as a replacement for a validated toxicological identification, specimen collection record, chiral assay, medical differential, contemporaneous observation, or incident-specific causation analysis.

What is the narrow conclusion supported by all the results together?

Across four intranasal administrations in the same ten methamphetamine-using adults, the authors found that the lowest active dose enhanced certain agency judgments and target hits, and that the larger doses increased distractor-contact errors when slow-moving targets were combined with intentionally unreliable mouse response. These effects depended on task condition and did not extend to every tested factor. Under placebo, the group already discriminated between the computer’s more and less controllable states.

The scientifically defensible takeaway is therefore context-dependent task performance and metacognitive judgment, not a universal or person-specific impairment rule. The paper does not quantify active parent drug in a forensic specimen during task performance and does not test actual driving, truthful recollection, the timing of last use, involuntary action, or responsibility for a particular event. Both the significant and the nonsignificant results are essential to that conclusion.

Frequently asked questions about the original methamphetamine agency study

1. Did 12 mg improve every cognitive measure?

No. The source documented improved target hit rate and task-condition-specific calibration of perceived control after the lowest dose. The study measured only this computerized task, not a comprehensive neuropsychological battery; benefits in one condition should not be asserted for every form of cognition.

2. Did 25 or 50 mg make people falsely believe they controlled the mouse?

Not according to the authors’ prespecified operational definition. Larger doses did not produce the predicted universal disruption of agency judgments. They did increase commission errors in one slow-speed, turbulent-mouse condition. Feeling in control, hitting targets, and avoiding distractors are distinct.

3. Were there 40 independent people because there were four sessions?

No. Ten independent volunteers completed four sessions each; each person contributed repeated conditions and multiple 24-trial blocks. Two others entered but did not complete. Repeated sessions improve within-person comparisons but do not increase biological N from ten to forty.

4. What was the drug route and actual administered amount?

The experiment used intranasal methamphetamine hydrochloride at placebo, 12, 25 and 50 mg per 70 kg, each on separate days at least 72 hours apart. A 60-mg maximum single administration applied. These are nominal experimental amounts, not assayed whole-blood concentrations or an instruction for use.

5. Did the researchers measure methamphetamine concentration in venous whole blood?

No postdose quantitative whole-blood methamphetamine result, nor a paired plasma, serum or oral-fluid concentration curve, was reported for this agency task. Enrollment urine screening showed drug exposure history but did not establish timed mass-spectrometric exposure during the task.

6. Were any specific comparisons nonsignificant?

Yes. Fast-speed agency differences across doses, dose-by-magic agency interaction, and hit-rate dose-by-task-factor interactions were not significant as reported. Larger doses did not show the low-dose hit-rate benefit. The source did not give exact p-values for all these null comparisons.

7. Can the findings tell a court whether a particular driver was impaired or an accident was caused by methamphetamine?

No. The outcome was an artificial mouse-control task in ten regular methamphetamine users, not a prospective validation of road driving, blood concentration, impairment classification, crash risk or event causation. Individual opinions require independent case evidence and validated methods.

Original research and verified same-site background

Primary investigation: Kirkpatrick MG, Metcalfe J, Greene MJ, Hart CL. Effects of intranasal methamphetamine on metacognition of agency. Psychopharmacology. 2008;197:137–144. Read the original empirical article, DOI 10.1007/s00213-007-1018-2. The analysis above distinguishes direct results in that publication from the companion Hart et al. physiological report that used an overlapping study context and is not a replacement for this metacognition experiment.

For related principles on interpreting analytical evidence, see these independently verified published guides on the same website: why blood, plasma and other biological specimen matrices cannot be interchanged; why specific drug identification is different from a preliminary screen; and why timing and purpose matter when interpreting drug concentrations.