Amphetamine reduced slowing after errors in a working-memory experiment without improving overall accuracy. The key post-error analysis included 55 participants, not the full 110-person accuracy sample.
Wardle and colleagues used placebo and 5-, 10- and 20-mg d-amphetamine sessions under double-blind counterbalanced conditions. Participants performed an N-back task. After excluding responses inconsistent with the instructions, 110 participants contributed to accuracy analyses. Post-error slowing required enough correct responses after both errors and correct trials; those additional requirements and missing-data handling produced a final sample of 55 for that endpoint.
The 20-mg dose reduced reaction times after errors, without significantly changing post-correct reaction times. Overall 3-back accuracy did not significantly change. Drug-induced elation did not account for the post-error effect, and personality and baseline performance did not significantly moderate the drug effect. Thus, the result was not simply a general acceleration of all responses or demonstrated improvement in remembering the sequence.
Slowing after an error can reflect several processes, including adjustment and orienting. Less slowing is therefore not automatically better control or evidence that a person failed to recognize an error. The endpoint-specific sample and imputation assumptions matter to precision and generalization. This controlled acute-dose study in healthy volunteers should also remain separate from observational studies in abstinent methamphetamine users. It supplies no individual error-awareness diagnosis, driving conclusion or legal-capacity rule. Accurate reporting names the actual analysis population alongside the dose and task result.
See post-stop adaptation in methamphetamine dependence.