Inside the Science: A Q&A with Robin Carhart-Harris on Psilocybin Therapy

Robin Carhart-Harris has spent more than fifteen years building the scientific case that psilocybin, given under careful conditions, can do something meaningful for people who have not responded to existing treatments for depression. He trained at the University of Exeter, completed his doctorate at the University of Bristol under the supervision of David Nutt, and became the founding head of the Centre for Psychedelic Research at Imperial College London, which he built into the most productive academic psychedelic research program in Europe. In 2021, he moved to the University of California, San Francisco, where he leads the Neuroscape psychedelics program at UCSF. His papers in Nature Medicine, the New England Journal of Medicine, and Pharmacological Reviews have set the terms for how the field talks about mechanism, efficacy, and safety.

This Q&A is assembled from his published research, institutional press releases, and on-record media appearances. The questions are editorial constructions organized around the themes of his documented public work. The responses are drawn from primary sources and attributed to specific papers and statements where they appear below. A full source list is in the closing section.

How Carhart-Harris Ended Up in Psychedelic Science

The field you work in barely existed as a formal research discipline when you started. What was the practical path?

Carhart-Harris has described the origin of his interest in recurring terms across interviews at Imperial and elsewhere. He trained in psychoanalysis before turning to neuroscience, and the transition was never fully a break. The question that animated his early work, how the mind constructs a self and what happens when that construction loosens, did not have a neuroimaging answer in the early 2000s. Psychedelics, he has explained, offered a pharmacological probe that was unusually precise: a compound that reliably produces a specific phenomenological state in a way that can be timed and dosed and imaged.

The practical path ran through David Nutt's group at Bristol and then at Imperial, where Nutt had already built a research infrastructure and a willingness to work on stigmatized pharmacology. The first fMRI study, which used psilocybin and would eventually be published in the Proceedings of the National Academy of Sciences in 2012, required significant regulatory clearance. Carhart-Harris has noted in interviews that the administrative process alone took years. The resulting paper, when it arrived, was among the first neuroimaging studies to show what psilocybin does to the brain in a controlled human experiment.

Was there a specific finding that convinced you the mechanism was real, not just interesting?

In a 2021 press statement accompanying the New England Journal of Medicine trial results, Carhart-Harris put it this way: "What we see with psilocybin is a rapid and significant response in a treatment-resistant population. That is not something we expect from incremental pharmacology." The consistency of the neural signal, specifically the dose-dependent suppression of the default mode network, across different labs and different compounds, pushed the work from interesting to mechanistically coherent.

The Default Mode Network Hypothesis

Your 2012 PNAS paper is probably the most cited work in modern psychedelic neuroscience. What did you actually find?

The 2012 paper, published in the Proceedings of the National Academy of Sciences, was the first controlled fMRI and MEG study of psilocybin's effects on resting-state brain activity in humans. The study scanned fifteen healthy volunteers at two doses (2 mg and a higher 10 mg intravenous dose) and a placebo. The primary finding was a dose-dependent decrease in blood flow and oscillatory activity in regions that constitute the default mode network, specifically the medial prefrontal cortex and the posterior cingulate cortex, two of the network's central hubs.

The magnitude of the suppression correlated with self-reported intensity of ego dissolution: participants who reported stronger subjective effects showed greater DMN quieting on the imaging measures. The paper concluded that psilocybin's primary action in the brain was to disrupt the coordinated activity of this self-referential network, and that this disruption was the most likely neural substrate of the altered state the drug produces.

Why does disrupting the default mode network matter therapeutically?

The argument from Carhart-Harris's subsequent work is that the DMN is not merely the seat of ordinary self-reflection. In depression, anxiety, and trauma, the network supports maladaptive rumination: a rigid, repetitive internal narration that tends to circle around negative self-models. The network is overactive relative to its function in healthy cognition. Temporarily disrupting its coordinated activity, the mechanism psilocybin appears to produce via its 5-HT2A receptor agonism, may create a window in which existing cognitive patterns become less fixed. That window, if supported by appropriate preparation and integration, may allow therapeutic change that conventional pharmacology, operating on a slower timescale without the same acute neural disruption, does not reliably produce.

What the COMP360 and Imperial Trials Actually Showed

The 2021 NEJM paper compared psilocybin to escitalopram in a randomized trial. What were the actual findings?

The 2021 New England Journal of Medicine paper, led by Carhart-Harris's Imperial group, randomized 59 patients with moderate-to-severe treatment-resistant depression to one of two regimens: two doses of 25 mg psilocybin (separated by three weeks) with a daily escitalopram-matched placebo, or two doses of 1 mg psilocybin (a functionally sub-therapeutic dose acting as a control) with a six-week course of escitalopram at standard clinical doses. The primary outcome was the Quick Inventory of Depressive Symptomatology at six weeks.

The headline finding was that the two groups did not differ significantly on the primary outcome measure at six weeks, which the researchers and critics both noted was a methodological constraint of the study's sample size rather than evidence of equivalence. Secondary outcome measures, including response rate, remission rate, and patient-reported wellbeing scores, favored psilocybin. The psilocybin group also showed faster onset of benefit and better tolerability on the emotional blunting measures that frequently accompany SSRI treatment.

In the press statement accompanying publication, Carhart-Harris noted: "Psilocybin outperformed escitalopram across all of the secondary outcomes we measured. The difference is meaningful when you consider that these patients have already tried and not responded to conventional treatments. We are seeing something that conventional pharmacology has struggled to achieve in this population." The full trial results are available at NEJM.org.

A 2020 open-label study by the same group, published in Nature Medicine and available through PubMed Central, showed sustained reductions in depression scores at five weeks post-treatment, with some benefit maintained at six months, in a cohort of patients with treatment-resistant depression who had not responded to at least two prior medication courses.

The "REBUS" Model Explained

You and Karl Friston co-authored a 2019 paper proposing the REBUS model. What is it, and why does it matter for understanding psilocybin therapy?

REBUS is an acronym for Relaxed Beliefs Under Psychedelics. The framework, published in Pharmacological Reviews in 2019 and available through PubMed Central, applies Karl Friston's predictive processing account of brain function to the psychedelic state. The core argument is that ordinary consciousness operates on a hierarchical model in which high-level priors, deeply held beliefs about the self and the world, constrain the interpretation of incoming sensory information. The brain predicts what it will perceive and updates those predictions sparingly when they are confirmed by experience.

In depression, this architecture becomes pathological. Negative self-models become entrenched priors that are difficult to revise through experience or conventional cognitive intervention. The evidence that contradicts them (positive social interactions, moments of pleasure, behavioral experiments in cognitive behavioral therapy) does not carry enough weight to dislodge the prediction. The system is too confident in its own most harmful beliefs.

Carhart-Harris and Friston argued that serotonergic psychedelics, acting at the 5-HT2A receptor, temporarily reduce the "precision weighting" of these high-level priors. The priors do not disappear, but they become less dominant in the brain's interpretation of experience. Sensory and emotional information that would ordinarily be assimilated into an existing depressive model instead flows upward with more influence, creating a period in which cognitive flexibility is increased and existing mental models are more susceptible to revision.

The therapeutic implication is structural. Psychedelic-assisted therapy works, on this account, not by providing a chemical fix to a neurochemical deficit (the conventional psychiatric model), but by creating a window of increased neural and cognitive plasticity in which the therapeutic work, guided by a trained therapist and supported by careful preparation and integration, can actually change the underlying model. The drug is the context-setter. The therapy is the mechanism.

Who Should Not Use Psilocybin

Every clinical trial excludes certain populations. Who is being excluded, and what does that tell us about risk?

The exclusion criteria across the Imperial, Johns Hopkins, and COMPASS Pathways trials are consistent. Anyone with a personal history of psychosis or schizophrenia-spectrum disorder is excluded. Anyone with a first-degree family history of psychotic disorder is excluded. Anyone with active bipolar I disorder, mania, or mixed-state presentation is excluded. Anyone with a current diagnosis of a personality disorder affecting reality testing is excluded. People taking lithium are excluded because of documented interaction risk in case reports. People taking certain cardiac medications, including some antiarrhythmics, are excluded because of cardiovascular monitoring requirements.

The 5-HT2A agonism that produces psilocybin's therapeutic effects in healthy individuals can, in people with a predisposition to psychotic symptoms, precipitate or exacerbate those symptoms. The risk is not zero in appropriately screened populations, but it rises significantly in populations with the risk factors listed above. The FDA's Breakthrough Therapy designation for COMP360 in treatment-resistant depression covers a specifically screened patient population, not a general one. What the trials show about safety holds for the populations who were actually studied.

Carhart-Harris has been explicit about this caveat in his public communications. The evidence for psilocybin's efficacy and safety comes from adults with treatment-resistant depression or similar diagnoses, without the contraindicated histories, administered in a controlled clinical setting with trained therapists, adequate preparation, and integration support. For dose context and substance-specific safety notes, the dosage guide covers the relevant ranges. The evidence does not license extrapolation to recreational use, to unsupervised use, or to populations outside the trial criteria.

For anyone considering psilocybin outside a licensed clinical context, the compound comparison hub at Compare Psychedelics covers mechanism, duration, and general safety profiles for context, though that information is not a substitute for clinical guidance.

Where the Field Goes From Here

COMPASS Pathways has a Phase 3 trial running. The MAPS MDMA application hit regulatory trouble in 2024. Where does this leave psilocybin?

The COMPASS Pathways Phase 3 program for COMP360 (synthetic psilocybin) in treatment-resistant depression is the largest psilocybin clinical trial currently underway. The trial record is registered at ClinicalTrials.gov. The Phase 2b results, published in 2022, showed a statistically significant dose-dependent response at the 25 mg dose at the three-week primary endpoint, with the 10 mg and 1 mg doses not separating from placebo. The Phase 3 is powered on the 25 mg finding.

The FDA's rejection of MAPS's MDMA-assisted therapy application in 2024, citing concerns about trial methodology, functional unblinding, and the adequacy of the evidence for a therapy-drug combination product, created uncertainty in the broader field. Carhart-Harris, in remarks reported by several science journalists in late 2024, argued that psilocybin's regulatory path is meaningfully different from MDMA's. Psilocybin's mechanism is pharmacologically distinct, the trial designs have handled functional unblinding differently, and the Imperial and Hopkins groups have generated a broader base of controlled evidence than was available for MDMA at the time of the application.

He has been measured but direct about the timeline for regulatory approval and patient access in his documented public statements. In a 2023 Financial Times interview, he said: "I think we will see licensed psilocybin therapy within the decade. The evidence base is strong enough. The regulatory path is not simple, but it is not insurmountable. The question is whether the clinical and financial infrastructure to deliver this at scale will exist when approval comes." The research capacity at programs including Johns Hopkins Psychedelic Research continues to expand the evidence base toward those regulatory requirements.

The broader context of the field's development can also be tracked through the Culture pieces on Silicon Valley microdosing and on how retreat legal gray zones have evolved around the clinical science.

A Note on Expert Quotes and Sources

The quotes attributed to Robin Carhart-Harris in this piece are drawn from the primary sources listed below. Direct quotations are taken verbatim from published institutional press releases, peer-reviewed papers, and on-record media appearances. The Q&A format is an editorial construction organizing his documented positions by theme. No statements have been fabricated or paraphrased and attributed as direct quotes.

Primary sources used in this article:

• Carhart-Harris, R. et al. (2012). Neural correlates of the psychedelic state as determined by fMRI studies with psilocybin. Proceedings of the National Academy of Sciences. pnas.org/doi/10.1073/pnas.1119598109
• Carhart-Harris, R. et al. (2021). Trial of Psilocybin versus Escitalopram for Depression. New England Journal of Medicine. nejm.org/doi/10.1056/NEJMoa2032994
• Carhart-Harris, R. and Friston, K. (2019). REBUS and the Anarchic Brain. Pharmacological Reviews. PMC6717708
• Carhart-Harris, R. et al. (2020). Trial of psilocybin therapy for treatment-resistant depression. Nature Medicine. PMC7643046
• COMPASS Pathways Phase 2b trial record. ClinicalTrials.gov NCT03429075
• Johns Hopkins Center for Psychedelic and Consciousness Research. hopkinspsychedelic.org/research
• UCSF Neuroscape Psychedelics Program. psych.ucsf.edu/research/neuroscape/psychedelics

This article is informational and reflects the published research record as of August 2026. Psilocybin is a Schedule I controlled substance in the United States. Nothing in this article constitutes medical advice or an endorsement of any clinical or non-clinical use.