The largest psychedelic brain-imaging study yet says 'desynchronized' was the wrong word

Since the earliest fMRI studies of psilocybin, the standard description of the psychedelic brain has been disorder: desynchronized, entropic, noisier than normal. A study led by Devon Stoliker and Adeel Razi at Monash University, published in Nature alongside a commentary from Stefan Borgwardt and Mihai Avram, argues that description captures only half the picture, and the half it misses may matter more for understanding why psychedelic therapy works.

How the study was designed

The team built what they describe as the largest single-site psychedelic neuroimaging dataset assembled to date: 65 psychedelic-naive adults, 62 analyzed in the final sample, each scanned with both fMRI and EEG twice, once at baseline and once roughly 80 minutes after a fixed 19 milligram oral dose of psilocybin. The same four conditions repeated in both visits: resting with eyes closed, guided meditation, listening to a curated music program, and watching a wordless film of moving clouds. That design is the point. Most prior psychedelic imaging studies scan people at rest and infer everything else; this one deliberately varied what participants were doing and watched how the brain’s organization moved with them.

What they found, and why it isn’t simply “more noise”

Standard, time-averaged measures did show the pattern the field already expected: functional connectivity within brain networks dropped, and modularity, the degree to which the brain organizes into distinct, separable networks, decreased significantly across every condition tested. Read on its own, that looks exactly like the desynchronization story already told. But the team also used machine learning to track each person’s brain activity as a continuous trajectory through time, rather than averaging it into a single snapshot, and a different picture emerged entirely. Under psilocybin, a classifier could identify which of the four conditions a person was in from their brain activity alone at well above chance level for every participant tested, a result that held across the full sample. If psilocybin were simply injecting randomness into brain activity, that classification accuracy should have gotten worse, not held up across the group. Instead, the brain’s moment-to-moment activity became more tightly organized around whatever context the person was in.

Why this directly revises the field’s own recent framing

This finding sits in direct tension with a widely cited 2024 Nature paper describing psilocybin as producing a broadly desynchronized brain state, and the Monash team addresses that tension head-on rather than ignoring it. Their argument is that time-averaged connectivity measures, the kind used in most prior studies including that one, do show reduced within-network connectivity and reduced modularity. What those measures miss is a structure that exists specifically in the moment-to-moment, temporal dynamics of brain activity, a structure that disappears the instant you average it away. Both descriptions are accurate to what they measure. The disagreement is about which measurement captures what matters for the psychedelic experience and its aftereffects.

The part that gives “set and setting” a measurable basis

Clinical guidelines for psychedelic-assisted therapy have treated the physical and psychological environment as decisive for years, largely as accumulated clinical wisdom rather than something directly measured in the brain. This study’s strongest evidence for a neural version of that idea comes from a network perturbation analysis run on a six-participant subgroup: baseline classification accuracy of roughly 62 percent rose to roughly 98 percent under psilocybin in that subgroup, and when the researchers computationally removed the psilocybin-specific activity from the default mode network and the visual network separately, replacing each with baseline activity, that gain dropped substantially, roughly six to eight percentage points for each. Those two networks, the study notes, sit at opposite poles of the brain’s principal gradient from internally directed to externally directed processing. Their joint alteration is what the team argues enables context-tracking, meaning the specific setting a person is in during a session may be doing more neurological work than a passive backdrop.

Embeddedness, and why the paper insists it isn’t the same as connectedness

The team’s proposed name for the subjective state that tracked this reorganization is embeddedness, participants reporting a felt sense of being continuous with their surroundings rather than separate from them, distinct, the authors are careful to specify, from mere connectedness, which implies links between things that remain separate entities. The strength of this brain-level context alignment scaled directly with how strongly participants reported self-dissolving and boundary-dissolving effects, and separately with how much their mindset had shifted one day after the session. Sensory and hallucinogenic effects showed weaker associations, and negative effects, anxiety and impaired cognition, showed essentially none, a pattern that argues against reading this as a generic drug-intensity effect and toward reading it as specifically tied to the positively experienced, boundary-dissolving quality of the trip.

The limitations the study states plainly

This was an open-label study with no blinding at any stage, including during data analysis, a real methodological limitation this desk has flagged repeatedly across other psychedelic trials this year, though here it applies to a mechanistic imaging study rather than a clinical efficacy trial, a meaningfully lower-stakes context for the concern. The four conditions were always presented in the same fixed order, rest before meditation before music before film, for participant safety and comfort reasons the authors state directly, which means condition order and psilocybin’s continued absorption over the session cannot be fully disentangled from the specific pattern of results, though the authors note strong within-condition clustering was already present during the first condition, before order effects could plausibly accumulate. The sample was healthy adults with no psychiatric diagnosis and no prior psychedelic experience, not a clinical population, and a companion arm testing eight weeks of mindfulness training beforehand found no measurable difference from participants who received no such training, a real null result worth noting rather than omitting.

The frame

This desk covered a related, complementary finding a few weeks ago: a separate Michigan-led study showing psychedelic states and sedation produce literal mirror-image patterns of brain network integration, evidence bearing on what a psychedelic state is at a structural level. This study extends that same basic question in a different direction, not what distinguishes a psychedelic state from a sedated one, but what distinguishes a psychedelic state that tracks its environment from one reduced to a single, context-free description of disorder. Neither finding settles the underlying debate about psilocybin’s therapeutic mechanism. Together, they represent a real shift in how rigorously that question is now being asked, with large, multimodal datasets and analysis methods built specifically to detect structure that a simpler, averaged measurement would erase.