A single 25 mg dose of psilocybin produces what 94% of subjects rate as the most unusual state of consciousness of their entire life. The question that has dogged psychedelic neuroscience for a decade is whether anything measurable survives that experience once the molecule clears. A new Nature Communications paper from Robin Carhart-Harris's group is the cleanest answer yet — and the finding worth slowing down for is structural.
Why this paper, why now
The entropic-brain hypothesis, proposed by this same lab, holds that psychedelics work by temporarily increasing the complexity of spontaneous brain activity — letting the cortex escape its predictive grooves and explore a wider state space. The molecular story underneath it has filled in over the years: psilocybin metabolizes to psilocin, psilocin is a 5-HT2A agonist, 5-HT2A activation triggers BDNF transcription in the hippocampus and prefrontal cortex, BDNF drives synaptic plasticity. In animal models the downstream effects are concrete and durable — dendritic spines grow, oligodendrocyte populations shift, white-matter integrity changes.
The human evidence has been thinner, mostly acute. What this paper sets out to do is look for structural and functional change a full month after one supervised dose, in people who have never used a psychedelic before, with a real placebo arm.
The design
Twenty-eight healthy adults, average age 41 (range 29–59), gender and education balanced, every one of them entirely psychedelic-naive. Each participant received two doses, in fixed order, one month apart:
- A sub-perceptual 1 mg dose, framed to participants as "a variable dose up to 25 mg." This served as an active placebo and an order/expectancy control.
- A full 25 mg dose one month later — a high clinical dose, capable of producing a full psychedelic experience.
EEG was recorded at baseline and at 1, 2, and 4.5 hours into each session. fMRI and diffusion tensor imaging were collected before and one month after each dose. Behavioral measures — cognitive flexibility, psychological insight, well-being, subjective intensity — ran across the same timeline.
A true randomized crossover wasn't possible. 25 mg of psilocybin can produce carry-over effects that would compromise a subsequent session. So the 1 mg dose is doing the work a placebo arm would normally do — and because it produced no measurable neural or psychological change, it does that work well. Every effect described below appeared only after 25 mg.
What changed acutely: entropy and alpha
During the 25 mg session, two well-established psychedelic signatures appeared together. Signal entropy — measured via Lempel-Ziv complexity, an information-theoretic index of how unpredictable the EEG trace is — rose sharply. Alpha power dropped. Both effects peaked at the two-hour mark, exactly when participants reported peak subjective intensity. Decreased alpha is generally read as cortical disinhibition. Increased entropy means the cortex is visiting a wider repertoire of states than its baseline allows. The interaction between dose and time on subjective intensity was overwhelming: F(8,197) = 57.73, p < 0.0001.
This part of the picture is not news. Acute entropy effects under psychedelics have been replicated many times. What this paper does is use that acute signal as a predictor of what shows up downstream.
The structural finding: white-matter compacting in prefrontal tracts
One month after the 25 mg dose, diffusion imaging revealed decreased axial diffusivity bilaterally in two white-matter tracts: one connecting the prefrontal cortex to the striatum (PFC–STR; t(24) = −3.72, p = 0.006), one connecting the prefrontal cortex to the thalamus (PFC–THA; t(24) = −3.85, p = 0.005). The tracts overlap substantially in space — the authors interpret this as a common prefrontal origin, not two independent findings. With Bonferroni correction the left-hemisphere effect held. The result survived free-water correction. The 1 mg control produced no equivalent change.
What does decreased axial diffusivity actually mean? In plain terms, water is moving less freely along the long axis of these fibers — consistent with a compacting or thinning of the tracts. The authors put two non-exclusive interpretations on the table:
- Pruning of weak or redundant connections — the brain shedding fibers it doesn't need.
- New neurogenesis with under-myelinated axons — new fibers laid down that haven't yet been wrapped in their insulating sheath. This is structural growth, the kind of architectural change the BDNF and NGF pathways drive.
Both are plasticity. Both are consistent with the high density of 5-HT2A receptors in the prefrontal cortex and the predominance of cortical 5-HT2A terminals over subcortical ones, which points the finger at PFC 5-HT2A as the locus of action.
The authors then immediately advise caution. They write that the DTI finding should be interpreted conservatively until it replicates with multi-shell diffusion sequences — work they say is already underway. That kind of restraint is rare and it strengthens the finding rather than weakening it. The cleanest reading of the result is: psilocybin appears to leave a structural footprint in prefrontal-subcortical wiring one month after a single dose, the direction of the change is consistent with either pruning or new growth, and the next round of imaging will tell us which.
The functional finding: modularity and well-being
Brain network modularity measures the degree to which functional connectivity divides into distinct, semi-independent subnetworks. High modularity is segregated; low modularity is more globally integrated. Earlier psilocybin-for-depression work from this lab found that decreased modularity tracked with symptom improvement.
In the present study the group-average modularity decrease wasn't significant against the 1 mg control (p = 0.21). But against pre-dose baseline a significant decrease appeared (t(24) = −2.95, p = 0.007). And more interestingly, the size of each person's modularity decrease correlated with the size of their well-being increase (r = −0.40, p = 0.04). The brains that became more integrated felt better.
The honest counterweight: most resting-state functional measures returned to baseline. The authors put it directly in the abstract — enduring functional brain changes are largely absent. The persistent changes are in the wiring, not in the activity patterns those wires support. The activity can re-form on the new architecture.
The behavioral findings and the mediation chain
At one month post-25 mg, participants improved on three measures: cognitive flexibility (extra-dimensional shift errors on an IDED task), psychological insight (the validated Psychological Insight Scale; dose × time F(2,44) = 14.05, p < 0.0001), and well-being. None moved after the 1 mg dose.
The analysis that ties everything together is the mediation model. A cross-validated predictive model showed that acute brain entropy at 1, 2, and 4.5 hours predicted improved well-being a month later (r = 0.66, p = 0.006). The same entropy signal at posterior electrodes at 2 hours predicted psychological insight measured the following day (r = 0.59, p = 0.006). A formal three-time-point mediation analysis confirmed the chain: acute entropy → next-day insight → one-month well-being. Adding insight as a mediator weakened the direct entropy-to-well-being link, satisfying standard mediation criteria.
The molecule sets the conditions. The acute neural state predicts what is understood the next day. What is understood the next day predicts what is felt a month later. The brain effect doesn't cause the outcome — it has to pass through insight to land.
This is the part of the picture too often skipped over in the popular psychedelic discourse. A high-entropy state on its own doesn't reliably produce a lasting benefit. The state has to be metabolized into understanding, and the understanding has to do work, before the outcome appears at one month. The molecule cannot be the whole intervention. The integration is part of the intervention.
What the authors don't claim
It is an unusually candid limitations section. The authors flag, among other things:
- The DTI structural findings need replication with multi-shell sequences. Until then, the structural interpretation is tentative.
- The group-average modularity decrease wasn't statistically significant against the 1 mg condition. The individual well-being correlation held; the headline group effect did not.
- The study was exploratory and hypothesis-generating, not powered to test specific predictions or examine demographic differences.
- The sample was healthy, mostly British, and entirely psychedelic-naive. Generalization to clinical populations or experienced users is a separate question.
Read those caveats alongside the result rather than around them. The findings get more interesting, not less, when the authors are this disciplined about what they show.
The pathway this belongs to
The same 5-HT2A → BDNF axis we've written about in the dual-pathway article runs straight through this paper. The compacting of prefrontal-subcortical tracts and the modularity shift sit in the same conceptual neighborhood as the maintenance and construction signals — BDNF and NGF — that drive durable neurogenesis. Whether the compacting reflects pruning or new growth, the underlying machinery is the plasticity machinery the entire myko Library is organized around. This paper does not change anything about how myko thinks about formulation. It strengthens the through-line: the durable wins in cognition come from architecture, not arousal. From wiring, not stimulant noise.
It also reinforces a quieter point. Mechanism is not promise. The Carhart-Harris group flags its own headline structural finding as preliminary pending replication. Saying that out loud in an abstract is exactly the discipline this field has needed.
PATHWAYS, NOT PROMISES.
Further reading
BDNF + NGF: the dual-pathway story · Pathways, not promises · Integration
Psilocybin remains a controlled substance in most jurisdictions outside approved clinical and research settings. Nothing here is medical advice.
Source: Lyons T, Spriggs M, Kerkelä L, Rosas FE, Roseman L, Mediano PAM, Timmermann C, Oestreich L, Pagni BA, Zeifman RJ, Hampshire A, Trender W, Douglass HM, Girn M, Godfrey K, Kettner H, Sharif F, Espasiano L, Gazzaley A, Wall MB, Erritzoe D, Nutt DJ, Carhart-Harris RL. Human brain changes after first psilocybin use. Nature Communications (2026). DOI: 10.1038/s41467-026-71962-3.
