Cognitive and Subjective Effects of Psilocybin Microdosing: Results from Two Double-Blind Placebo-Controlled Longitudinal Trials

Earlier anecdotal and observational reports have portrayed psychedelic microdosing as a means to boost focus, creativity and mood, prompting interest in its potential therapeutic use for attentional and affective disorders. However, placebo-controlled evidence has been inconsistent: some open or poorly controlled studies reported improvements in well-being, mindfulness, and convergent thinking, but many effects vanished once expectancy and blinding were modelled. The field is also hampered by small samples, heterogeneous dosing regimens, diverse outcome measures and a lack of a unifying mechanistic framework, leaving it difficult to determine whether microdosing produces reliable cognitive or emotional changes. Prochazkova and colleagues designed two randomized, double-blind, placebo-controlled longitudinal trials to address these gaps by testing whether psilocybin microdosing biases cognitive control toward flexibility at the expense of persistence. Using the Metacontrol State Model (MSM) as the organising theory, the investigators selected a battery of behavioural tasks and questionnaires predicted to detect shifts along the persistence–flexibility axis (e.g., AX-CPT, Multi-Armed Bandit, Reference-Back, Attentional Blink, N-back, RMET) alongside measures of well-being, psychological flexibility and momentary mood. The goal was to evaluate acute and cumulative effects of two microdosing regimens in semi-naturalistic workshop settings while maintaining rigorous placebo control and blinding.

Two parallel experiments were conducted at Leiden University in collaboration with community partners. Both used randomized, double-blind, placebo-controlled, between-subjects longitudinal designs with baseline, two acute (one-hour post-dose) lab assessments and a post-acute online follow-up roughly two days after the final dose. Participants were recruited via the Microdosing Institute and the Psychedelic Society of the Netherlands; screening excluded people with personal or family histories of psychosis, schizophrenia, mania or borderline personality disorder. Randomisation and blinding were implemented during community workshops where half of participants’ prepared doses were covertly swapped for placebo by staff; participants were informed they could receive active or placebo capsules. Experiment 1 employed a lower dose regimen (0.65 g fresh Psilocybe galindoii truffles, roughly 1/15th of a recreational dose) administered every three days across approximately four weeks (six microdoses total). Experiment 2 raised the dose to 1.00 g and extended the active period so Acute 1 and Acute 2 corresponded to later microdoses (6th and 10th), with an overall trial duration of about eight weeks. Truffle alkaloid content was verified post hoc by chemical analysis. Tasks were presented on standardised computer equipment and questionnaires via Qualtrics; data preprocessing used SPSS and final analyses were performed in JASP. The cognitive battery in Experiment 1 included the AX-Continuous Performance Task (AX-CPT), Multi-Armed Bandit (MAB) task, Reference-Back task, Remember-Know memory test and the Reading the Mind in the Eyes Test (RMET). Experiment 2 used the Attentional Blink task, N-back (with multiple loads), the Trust Game and the Inclusion of Other in the Self (IOS) scale. Subjective measures across studies included the Warwick–Edinburgh Mental Well-Being Scale (WEMWBS), Multidimensional Psychological Flexibility Inventory (MPFI), Affect Grid, Cognitive Flexibility Inventory (CFI), Freiburg Mindfulness Inventory (FMI), Self-Compassion Scale (SCS), PANAS and a bespoke Subjective Microdosing Experience Questionnaire (SMEQ). Participants also provided open-text responses about their acute experiences, which were pooled across experiments for qualitative analysis. Sample sizes varied by task after exclusions for protocol deviations and data preprocessing. In Experiment 1, analysed Ns ranged roughly from mid-50s to high-60s across tasks (e.g., MAB N = 67; AX-CPT N = 58; RMET N = 59). Experiment 2 produced similar per-task Ns (e.g., Attentional Blink N = 56; N-back N = 56; Trust Game N = 54). Qualitative analyses combined 178 open-ended responses (85 active, 93 placebo). Statistical analyses employed mixed-design repeated-measures ANOVAs with session and task factors, χ² tests for categorical data and corrections for sphericity; non-parametric alternatives were used when assumptions were violated. Blinding integrity was assessed by participants’ guesses and by perceived microdosing strength ratings.

Across both experiments, randomisation succeeded: demographic and baseline variables did not differ significantly between placebo and active groups. Blinding appeared effective; participants’ guesses about group allocation did not differ from chance at Acute 1 or Acute 2 in either experiment (Experiment 1: χ²(2, 62) = 2.18, p = .337 and χ²(2, 62) = 4.88, p = .087; Experiment 2: χ²(2, 71) = 3.24, p = .198 and χ²(2, 66) = 1.92, p = .383). Perceived strength of effects declined over time in both groups (significant main effect of time in both experiments), suggesting tolerance or waning expectation rather than a drug-specific change. Behavioural performance showed consistent learning or practice effects across sessions but minimal evidence of differential drug effects. In Experiment 1, AX-CPT, MAB, Reference-Back and Remember–Know tasks all exhibited main effects of time or trial type without significant time × condition interactions, indicating no reliable microdosing effects on proactive/reactive control, exploration–exploitation balance, updating, or long-term memory. The Reference-Back produced only a small gate-closing error difference that was present at baseline. The MAB task showed learning across sessions but no group differences. The RMET in Experiment 1 was an exception: a three-way interaction (item difficulty × session × group) was significant (F(2, 114) = 3.92, p = .023, η² = .064). Follow-up tests indicated the active group improved on difficult RMET items from baseline to both Acute 1 (p = .04) and Acute 2 (p < .01); this effect was not accounted for by expected group assignment. The authors note this isolated cognitive effect could be a true, selective enhancement of intuitive social cognition or a Type I error given multiple tests. In Experiment 2, Attentional Blink and N-back performance showed expected main effects of lag/load and practice, but the critical interactions involving group were non-significant (e.g., Attentional Blink time × lag × group F = 0.43, p = .780). The Trust Game and IOS both showed increases over time but no session × group interactions, indicating changes were not specific to the active condition. Some questionnaire subscales exhibited baseline imbalances; where small post-intervention differences emerged (e.g., a CFI subscale interaction), the authors caution these likely reflect chance or baseline differences rather than robust treatment effects. Subjective measures converged on null or non-specific findings. WEMWBS, MPFI, Affect Grid, FMI, SCS and PANAS generally revealed no significant session × group interactions; decreases in psychological inflexibility and reductions in negative affect occurred across both groups. SMEQ responses produced one uncorrected effect for visual clarity in Experiment 1 (t(68) = 2.54, p = .013, d = 0.617) that did not survive correction for multiple comparisons. Qualitative analyses of 178 open-text responses showed that most participants reported some subjective change regardless of condition. Overall, 49% described purely positive symptoms, 30% mixed, and 14% purely negative. A chi-square test indicated a greater frequency of negative symptoms in the active group (χ²(1, 146) = 9.5, p = .023). Participants in the active condition reported more bodily awareness experiences (t(176) = 2.8, p = .006, d = 0.49), while pro-social descriptors were more common in the placebo group (t(176) = 2.2, p = .025, d = 0.32). Two participants reported strong adverse reactions (flashbacks, extreme insomnia, intense emotions), both of whom were in the placebo condition. In sum, microdosing did not produce consistent cognitive or affective improvements relative to placebo but was associated with modest increases in somatic awareness in the active group.

Prochazkova and colleagues interpret their findings as providing no compelling evidence that psilocybin microdosing reliably enhances cognitive performance or emotional well-being in healthy volunteers under controlled conditions. Both experiments produced largely null behavioural and questionnaire results after accounting for multiple comparisons; practice and time effects were common, and blinding was maintained. The lone robust behavioural effect—a selective improvement on difficult RMET items in Experiment 1—could indicate a subtle, domain-specific effect on intuitive social cognition, but the authors warn it may also be a false positive given the extensive testing battery. The investigators situate their results within the mixed placebo-controlled literature, noting convergence with several recent trials that found no reliable cognitive or mood benefits but occasional changes in neural measures or perceptual domains. They emphasise that subjective reports were predominantly positive across both conditions, highlighting the potential role of expectancy. At the same time, the active condition produced more reports of somatic or interoceptive discomfort, suggesting that microdosing can yield subtle physiological effects even when cognitive benefits are absent. Key limitations acknowledged include the possibility of insufficient power to detect very small or domain-limited effects despite samples larger than many prior microdosing studies; the non-clinical nature of the samples, which reduces potential for measurable improvement in mood or cognition; and the limited ecological validity of laboratory tasks that may miss complex, real-world outcomes. The authors also note baseline imbalances on some questionnaire subscales and the uniform dosing irrespective of body weight as constraints. Selection bias and participants’ prior psychedelic experience are cited as additional factors that may amplify expectancy and complicate interpretation. For future research, they recommend larger samples, studies in clinical or symptomatic populations where room for improvement is greater, more ecologically valid and multimodal outcome measures, and systematic investigation of individual differences in response. The authors conclude that, under the tested regimens and in healthy individuals, psilocybin microdosing did not produce reliable cognitive or emotional enhancement beyond placebo, though targeted or perceptual effects merit further study.