Microdosing with psilocybin mushrooms: a double-blind placebo-controlled study

Interest in the practice of taking very low, sub-perceptual doses of psychedelics (“microdosing”) has expanded rapidly among the public and researchers. Anecdotal reports and observational studies claim improvements in mood, cognition, creativity and wellbeing, but these data are vulnerable to selection bias, expectancy effects and the lack of placebo controls. Neurobiological mechanisms remain underexplored for microdoses, although studies with full psychedelic doses show alterations in brain oscillations, connectivity and subjective experience that could in principle scale with dose. Cavanna and colleagues set out to evaluate the acute and short-term effects of a representative psilocybin mushroom microdose on subjective experience, behaviour, creativity (convergent and divergent thinking), perception, cognition and brain activity measured with EEG. To limit artificial motivation and to better reflect real-world practice, the investigators recruited people who were planning to start microdosing with their own Psilocybe cubensis material and implemented a randomized, double-blind, placebo-controlled within-subjects protocol using 0.5 g dried mushroom per dosing day versus an inert mushroom placebo.

Thirty-four healthy adult volunteers (11 female; mean age 31.26 ± 4.41 years; mean weight 74 ± 17 kg) were recruited between December 2019 and August 2020. Participants reported prior psychedelic use (mean 11 ± 14.9 lifetime experiences) and only six had substantial prior microdosing experience. All provided written informed consent and completed the protocol. The study was registered (ClinicalTrials.gov NCT05160220) and approved by a local ethics committee. A within-subjects randomised double-blind design allocated each participant to two measurement weeks separated by one week without measurements: one week corresponding to an active condition (0.5 g ground dried Psilocybe cubensis in a gel capsule) and one to placebo (0.5 g edible mushroom prepared identically). Order was randomised by a third party so that both participants and experimenters were blind to condition during data collection. Participants completed two dosing days per measurement week (Wednesday and Friday) and underwent repeated assessments across those days. Primary and secondary measures spanned subjective and objective domains. Acute subjective effects were assessed with a 21-item visual analogue scale (VAS) adapted from prior psychedelic research. Trait and state questionnaires included the Big Five Inventory, State-Trait Anxiety Inventory, Short Suggestibility Scale, PANAS, Mind Wandering Scale, Perceived Stress Scale, Tellegen Absorption Scale, Psychological Well-being Scale, Flow State Scale, Creative Personality Scale, Cognitive-Affective Empathy Test and Cognitive Flexibility Scale. Creativity was probed with convergent and divergent tasks: the Remote Associates Test (RAT), the Alternative Uses Task (AUT) and the Wallach-Kogan (WK) test. Perceptual and cognitive tasks comprised backward masking, binocular rivalry, attentional blink, Go/No-Go, Stroop and the Trail Making Test (TMT). Daily physical activity was monitored with a Fitbit Charge 4 (steps and distance). Resting-state EEG (5 min eyes open, 5 min eyes closed) and an auditory Local-Global paradigm were recorded with a 24-channel mobile system. EEG preprocessing included bandpass filtering (1–90 Hz), notch filtering, channel rejection/interpolation, epoching, ICA-based artifact removal and power spectral density estimation in standard bands (delta, theta, alpha, beta, gamma). Broadband signal complexity was estimated using a Lempel–Ziv compression metric on binarised envelopes. Event-related potentials (ERPs) for local and global auditory deviants were analysed from −200 to +1300 ms relative to stimulus onset. Chemical analysis of mushroom samples (three independent sources) was performed by LC–MS on 150 mg aliquots, yielding concentrations of psilocybin (640.2 μg/g), psilocin (950.7 μg/g), baeocystin (50.4 μg/g) and norbaeocystin (12.5 μg/g). The authors report an estimated effective psilocybin-equivalent dose of ≈0.9 mg for the preparation used, but note possible variability across sources and potential degradation during storage. Statistical comparisons between active and placebo conditions used non-parametric paired Wilcoxon signed-rank tests (or Mann–Whitney U for unpaired data), with Bonferroni correction applied when indicated. Chi-squared tests evaluated breaking of the blind. Bayesian analyses computed BF10 values (Jeffrey–Zellner–Siow prior) to quantify evidence for alternative versus null hypotheses. Results are reported both uncorrected and with correction for multiple comparisons where applicable.

Blinding: Participants correctly identified the experimental condition in 49 of 64 measurement weeks (75%). For the active weeks, 25 of 34 (73.5%) were correctly unblinded; the placebo weeks showed a similar rate (70%). Chi-squared tests indicated that participants did not break blind in the first measurement week but did so in the second week. Acute subjective effects: Total VAS scores (sum of 21 items) were significantly higher for the active dose compared to placebo on dosing days. Crucially, this overall effect was driven by participants who correctly identified their condition: the “unblinded” subset showed significantly higher VAS totals for active versus placebo (Bayes factors BF10 > 30, very strong evidence), whereas the “blinded” subset did not show robust differences (BF10 values between 1/3 and 3, interpreted as inconclusive, with only a trend on the first dosing day). Individual VAS items showed differences on measures related to imagination, dreamlike quality, spatial distortions and mind-wandering, but these did not survive Bonferroni correction for multiple comparisons; most BF10 values for sub-items were inconclusive. Questionnaires and trait scales: No significant differences emerged between active and placebo conditions on the battery of self-report scales at p < 0.05 after correction. One uncorrected finding (conscientiousness, p = 0.01) did not remain significant after Bonferroni correction. Most BF10 values were <1/3, indicating moderate evidence favouring the null hypothesis. Analyses restricted to the blinded subset generally produced inconclusive Bayes factors (between 1/3 and 3), consistent with limited sample size. Creativity tests: Convergent (RAT) and divergent (WK, AUT) creativity measures showed no significant differences between active and placebo after correction. Two metrics (AUT fluency and WK elaboration) reached p < 0.05 uncorrected but did not survive Bonferroni correction; BF10 values predominantly supported the null hypothesis (most <1/3). Perception and cognition: Across the perceptual and cognitive battery, a few uncorrected differences were observed. Visibility of the second target in the attentional blink task was reduced at 300 ms lag under psilocybin relative to placebo (p < 0.05 uncorrected). Reaction times in the Stroop task were longer with psilocybin (p < 0.05 uncorrected). Part A completion time of the TMT was increased in the active condition. These effects did not consistently survive correction for multiple comparisons and are reported as trends toward impaired performance rather than robust enhancements. EEG and ERPs: Resting-state spectral analysis revealed decreased power in the theta band (4–8 Hz) during eyes-closed recordings under the active microdose compared with placebo; this reduction survived Bonferroni correction across frequency-band comparisons (n = 4). No reliable differences were found in alpha or beta bands. Global Lempel–Ziv complexity (a measure of signal diversity/entropy) did not differ between conditions for eyes-open or eyes-closed recordings. ERP analysis of the Local-Global auditory paradigm reproduced expected larger late amplitudes for global deviants versus local deviants at central-frontal sites (e.g. AFz) but revealed no significant differences between active and placebo conditions. Physical activity: Step counts, distance travelled and activity/resting times measured by Fitbit did not differ between active and placebo weeks (uncorrected comparisons). The authors note lower activity on Wednesdays, likely attributable to protocol-related seated measurements carried out that day. Chemical characterisation: LC–MS of sample aliquots gave psilocybin 640.2 μg/g and psilocin 950.7 μg/g. The authors estimated an effective psilocybin-equivalent per dose of ≈0.9 mg but emphasised inter-source variability and the potential for potency loss during storage.

Cavanna and colleagues interpret their findings as showing that a single microdose regimen of 0.5 g dried Psilocybe cubensis (per dosing day) produced noticeable subjective effects for participants who correctly identified receiving the active material, but did not yield reliable improvements in creativity, cognition, physical activity or self-reported wellbeing. Rather than consistent benefit, some cognitive tasks showed trends toward impaired performance under the active condition. The EEG finding of reduced theta power during eyes-closed resting state aligns with previous reports of broadband power reductions at higher psychedelic doses, while the absence of increased Lempel–Ziv complexity suggests that signal entropy elevation may be a feature of full psychedelic states rather than low-dose exposure. The investigators emphasise the role of expectation and unblinding: subjective acute effects were largely limited to participants who correctly guessed the condition, and unblinding rates were high (≈75% of measurement weeks). They note that prior open-label and survey-based reports of microdosing are vulnerable to confirmation and placebo biases, and that several other double-blind studies have yielded minimal or mixed effects. Bayesian analyses suggested insufficient evidence to resolve effects in the blinded subgroup, highlighting sample size constraints for that comparison. Key limitations acknowledged by the authors include variability in alkaloid content across mushroom sources and potential loss of potency during storage, absence of dose adjustment by participant body weight, and the short-term, two-dose-per-week design that precludes assessment of cumulative or schedule-dependent effects. The sample comprised healthy volunteers, so ceiling effects could have masked benefits that might appear in clinical populations. The study also prioritised acute laboratory assessments (e.g. computer tasks, EEG) that were impractical to repeat over extended dosing periods. For future research, the authors recommend studies that test a variety of dosing schedules and durations, include patient populations with baseline impairments, measure longer-term outcomes and safety-relevant physiological endpoints (for example, effects of chronic 5-HT2B receptor stimulation on cardiovascular health). They conclude that, based on their controlled data, small amounts of dried Psilocybe cubensis can induce subjective effects and alter EEG rhythms, but do not provide conclusive evidence of enhanced wellbeing, creativity or cognitive function; expectation effects likely account for at least some of the anecdotal benefits attributed to microdosing.