Evidence for tolerance in psychedelic microdosing from the self-blinding microdose trial

Interest in therapeutic applications of classic psychedelics has grown sharply, with the best‑established model being psychedelic‑assisted therapy that pairs one or two supervised, moderate-to-high dose drug sessions with psychotherapy. A distinct and less-regulated practice, microdosing, has also become popular; although definitions vary, typical regimens involve very low doses taken 2–4 times per week (for example 5–20 µg LSD or 0.1–0.3 g dried psilocybin mushroom). Anecdotal reports and some observational studies claim benefits for mood, creativity and cognition, but placebo‑controlled trials have generally failed to find robust support for such effects, raising the possibility that reported benefits are driven by placebo or expectancy effects. Baumann and colleagues used data from their self‑blinding microdose trial to test whether tolerance develops during microdosing. The self‑blinding design is a citizen‑science, at‑home placebo‑controlled approach that allowed a large real‑world sample. The study specifically examined whether the probability of correctly guessing a microdose (used here as a proxy for noticeable subjective drug effects) declined as participants accumulated previous microdoses during the trial, and whether any tolerance effect differed between LSD/LSD‑analogues and psilocybin microdosing.

The analysis used data from the self‑blinding microdose trial, a citizen‑science trial in which participants who planned to microdose prepared and used their own capsules and implemented a randomised, blinded schedule at home. Participants were randomly allocated to one of three groups in a 1:1:1 ratio: placebo (four placebo capsules/week for four weeks), half‑half (microdose schedule for two weeks, placebo schedule for two weeks) and microdose (four capsules/week for four weeks, of which two capsules/week were active microdoses and two were placebos). At the end of every dosing day participants completed a forced binary choice indicating whether they thought that day's capsule was active or placebo. The primary dependent variable for the present analysis was a binary "correct microdose guess" indicator defined on dosing days when a microdose was taken: True if the participant guessed "microdose" on a microdose day, False if they guessed "placebo" on a microdose day. Placebo dosing days were removed from the dataset. Mixed linear regression models were fitted with participant ID as a random effect to account for repeated measures; the independent variable of interest was the number of previous microdoses taken earlier in the trial. Models were adjusted for baseline covariates selected by backward elimination (variables were removed iteratively until all remaining covariates were significant at α≤.05). In a secondary, post‑hoc analysis the same model formulation was applied separately to LSD/LSD‑analogue and psilocybin microdose sessions. Dose conversion was applied to place psilocybin mushroom doses on an approximate LSD scale (0.1 g dried psilocybin mushroom ≈ 4.6 µg LSD); LSD‑analogue doses (1p‑LSD, 1cp‑LSD) were left unaltered. The analysed dataset comprised 993 microdose dosing days with guess data (757 LSD/LSD‑analogue sessions, 236 psilocybin sessions). The authors note that these counts refer to dosing sessions, not numbers of participants. Code and data to reproduce the analyses are reported as available in the manuscript repository.

When all microdose sessions were analysed together (n=993 dosing days with guess data), the number of previous microdoses was associated with a reduced probability of correctly identifying a microdose: estimated coefficient mean±se = -0.017±0.007, p = .009. In other words, each additional prior microdose was associated with approximately a 1.7% decrease in the probability of a correct microdose guess. Dose level was positively associated with correct guessing (mean±se = 0.009±0.004, p = .016), meaning higher reported dose increased the chance of correctly detecting an active microdose. When analyses were stratified by substance type, results diverged. For LSD/LSD‑analogue sessions (n=757), the number of previous microdoses remained a significant negative predictor of correct guess probability (mean±se = -0.026±0.007, p < .001), an effect corresponding to roughly a 2.6% decrease in correct guess probability per prior microdose. By contrast, the psilocybin subsample (n=236) showed a non‑significant positive estimate (mean±se = 0.013±0.014, p = .36). The authors interpret these patterns as evidence that tolerance to noticeable subjective effects accumulated with repeated LSD microdosing but was not detectable for psilocybin in their sample.

Baumann and colleagues interpret the observed decline in correct microdose guess probability with increasing prior microdoses as evidence consistent with tolerance developing during microdosing, at least for LSD and its analogues. They note that the magnitude of the estimated effect is modest per dose (approximately 2–3% decrease in correct guess probability per prior microdose), but cumulative exposure—given typical regimens of multiple microdoses per week—could produce a substantial reduction in recognisable effects over time. The authors situate the LSD finding within previous literature showing rapid tolerance to repeated, higher‑dose LSD in humans and animal data implicating pharmacodynamic mechanisms. They propose a pharmacological hypothesis for the substance difference: LSD experiences last longer (~8–10 hours) than psilocybin (~4–6 hours), so prolonged agonist exposure at the 5‑HT2A receptor might induce receptor desensitisation and tolerance more readily with LSD. They acknowledge that human receptor occupancy data exist for low‑dose psilocybin but not for LSD, and that higher receptor occupancy at typical LSD microdoses could plausibly trigger desensitisation processes. An alternative explanation offered by the authors is statistical power: the LSD session count was substantially larger than the psilocybin session count, so the study may have been underpowered to detect a tolerance effect for psilocybin. Key limitations are emphasised: this analysis was post‑hoc and exploratory rather than pre‑planned; tolerance was inferred via a behavioural proxy (correct‑guess rate) rather than direct pharmacological measures; participants self‑prepared microdoses, introducing dose uncertainty; and declining task engagement or questionnaire fatigue over time could also reduce correct guessing independently of pharmacological tolerance. The authors conclude that clinical trials with verified dosing and direct pharmacological measures are needed to confirm whether tolerance differs between substances and to clarify mechanisms. They suggest practical implications consistent with their findings: microdosers might benefit from periodic suspensions of dosing to avoid tolerance, and psilocybin could be preferable for long‑term microdosing if the lack of tolerance is confirmed.

The paper concludes that, based on the self‑blinding microdose trial data and the proxy measure used, repeated LSD/LSD‑analogue microdosing shows evidence of tolerance (a declining probability of detecting a microdose with accumulated prior doses), whereas psilocybin microdosing did not show such an effect in this dataset. The authors suggest that microdosers may consider periodic breaks to mitigate tolerance, and that psilocybin may be better suited than LSD for sustained microdosing regimens, while calling for controlled clinical studies with verified dosing to confirm these implications.