Neural correlates of the psychedelic state as determined by fMRI studies with psilocybin

Psychedelic drugs such as psilocybin have a long history of ritual and therapeutic use, yet their neural mechanisms remain poorly understood. Earlier work has characterised the phenomenology of psilocybin and implicated serotonergic (particularly 5-HT2A) receptors, but direct, high-resolution imaging of the transition from normal waking consciousness into the psychedelic state has been lacking. Carhart-Harris and colleagues set out to characterise the neural correlates of the acute psychedelic state using task-free functional MRI. They combined arterial spin labelling (ASL) perfusion and blood-oxygen-level-dependent (BOLD) fMRI to map cerebral blood flow (CBF) and BOLD changes before and after intravenous infusions of placebo and psilocybin, aiming to capture short‑latency, phasic effects produced by a 2 mg i.v. dose that begins acting within seconds and peaks within minutes. The study focused on resting-state activity and connectivity, with particular attention to high-level association regions and connector hubs such as the medial prefrontal cortex (mPFC), anterior cingulate cortex (ACC), posterior cingulate cortex (PCC) and thalamus.

Two separate samples of healthy, hallucinogen-experienced volunteers were studied. For ASL perfusion imaging, 15 subjects (five female; mean age 34.1, SD 8.2) underwent two 18-minute task-free scans with a fixation cross; saline placebo was infused in the first scan and psilocybin (2 mg in 10 mL saline) in the second. Subjective intensity was rated on a 0–10 visual analogue scale at multiple time points (including 5 and 12 minutes post-infusion). For BOLD fMRI, a separate group of 15 subjects (two female; mean age 32.0, SD 8.9) underwent eyes-closed, task-free 12-minute BOLD scans on two visits about 14 days apart, with placebo and psilocybin infusions given in balanced order. Infusions in both protocols were administered manually over 60 s, beginning 6 minutes after scan start. Imaging: ASL perfusion used a single-shot pulsed PICORE-QUIPSSII sequence with 16 axial slices; 240 tag-control pairs were acquired. BOLD data were acquired with gradient-echo EPI (TR/TE 3000/35 ms, 53 slices, 3 × 3 × 3 mm voxels, 240 volumes). First-level models incorporated the tag-control difference (ASL) and a pharmacodynamically informed square function (BOLD) to estimate within-scan modulation by psilocybin; higher-level mixed-effects analyses contrasted post-infusion versus pre-infusion and psilocybin versus placebo. Statistical maps were cluster-thresholded (z > 2.3) and whole-brain corrected at P < 0.05. Functional connectivity/PPI: A ventromedial prefrontal (vmPFC) region of interest was defined from the CBF/BOLD decreases and used for psycho‑physiological interaction (PPI) analyses testing for changes in linear coupling associated with psilocybin infusion. Gray matter, white matter and CSF time series were entered as nuisance regressors. Physiological confounds were addressed by recording respiration, cardiac pulse and end‑tidal CO2; RETROICOR and GLM-based regressors were used to remove cardiac/respiratory harmonics and CO2-related variance. Cerebrovascular reactivity (CVR) was assessed with a breath‑hold task (paced breathing and 10-s breath-holds) to test whether psilocybin altered vascular responsiveness; BOLD responses to breath-hold were normalised to CO2 changes and compared voxel-wise between conditions. Ethical screening and exclusion criteria were applied (age ≥21, no personal/family psychiatric disorder, no substance dependence, no cardiovascular disease, no recent hallucinogen use within 6 weeks, etc.).

Subjective effects: Psilocybin produced rapid, robust subjective effects. In the ASL group mean intensity ratings were 6.7 (±1.9) at 5 minutes post-infusion and 5.2 (±2.3) at 12 minutes; in the pooled data all ten subjective items scored significantly higher after psilocybin than after placebo (P < 0.01 as reported). ASL perfusion results: Group-level ASL analyses (n = 15) revealed significant decreases in CBF after psilocybin in multiple subcortical and cortical regions. Subcortical decreases included bilateral thalamus, putamen and hypothalamus. Cortical decreases were prominent in high‑level association and hub regions: PCC, retrosplenial cortex, precuneus, bilateral angular and supramarginal gyri, rostral and dorsal ACC, paracingulate gyrus, mPFC, frontoinsular cortex, lateral orbitofrontal cortex, frontal operculum, precentral gyrus and superior/middle/inferior frontal gyri. Temporal analyses of ROIs (thalamus, ACC, PCC) showed steep CBF reductions after infusion that were sustained for the duration of the scans. A correlation analysis indicated that larger CBF decreases in ROIs were associated with greater subjective intensity; the ACC CBF change correlated negatively with intensity (Pearson r = -0.55, P = 0.017, one-tailed). The extracted text reports trend‑level associations for thalamus and PCC but is ambiguous about exact statistics. BOLD fMRI results: The BOLD sample (n = 15) showed regional decreases in BOLD signal after psilocybin that broadly mirrored the ASL CBF decreases, including the mPFC, ventral PCC, putamen and subthalamic nuclei. Additional BOLD decreases were observed in higher‑order visual areas that were not seen in ASL. Importantly, no regional increases in CBF or BOLD signal were reported in either modality. Functional connectivity/PPI: Using a vmPFC seed, the PPI analysis identified a significant decrease in positive coupling between the mPFC/vmPFC and the PCC after psilocybin. The authors describe this as a reduced positive coupling rather than an emergence of genuine negative connectivity. Physiological correction: Removing physiological variance (heart rate, respiration rate and depth, end‑tidal CO2) from the BOLD data did not materially alter the significant maps. Breath-hold CVR testing found no difference in the BOLD response to hypercapnia between psilocybin and placebo, arguing against a direct vascular action of the drug on CVR.

Carhart-Harris and colleagues interpret their findings as evidence that acute psilocybin decreases neural activity and functional coupling in key cortical and subcortical connector hubs, most notably the mPFC/ACC, PCC and thalamus, and that these decreases are related to the intensity of subjective psychedelic experience. They note that these results were unexpected because psychedelics are often assumed to increase neural activity; however, the authors point to supporting animal data showing broadband decreases in resting local field potential power after psilocybin and propose that decreased activity observed with fMRI is plausible. The authors contrast their fMRI results with prior PET work reporting global increases in glucose metabolism after oral psilocybin, suggesting a timescale difference: the PET tracer integrates activity over much longer periods and might therefore reflect rebound or later effects not captured by the phasic fMRI measures. Mechanistically, the team proposes that stimulation of 5-HT2A receptors, known to be central to psychedelic action, could lead indirectly to net inhibition of pyramidal cell activity via excitation of GABAergic interneurons, consistent with the observed deactivations in cortical regions rich in 5-HT2A receptors. They acknowledge that receptor-level mechanisms cannot be definitively established from these data. Functionally, the authors highlight that the most strongly affected regions are high‑baseline activity hubs within the default-mode network (DMN) and have unusually high cortico‑cortical connectivity. They argue that deactivation and decoupling of these hubs may reduce top‑down constraint on cognition, enabling an "unconstrained" or less‑constrained style of cognition seen subjectively during the psychedelic state. The observed decrease in vmPFC–PCC coupling is interpreted as a perturbation of reciprocal hierarchical interactions between prefrontal and parietal association areas, which could reflect reduced top‑down influence or a relative increase in bottom‑up influence. Clinical relevance and caveats: The discussion connects the findings to hypotheses about depression, noting that mPFC hyperactivity and increased mPFC connectivity have been implicated in depressive rumination; thus, acute mPFC deactivation by psilocybin may provide a putative biological mechanism for observed improvements in wellbeing and depressive symptoms in prior clinical reports, though the authors emphasise that further work is required to test this therapeutic hypothesis. They also mention decreased hypothalamic CBF as potentially relevant to anecdotal reports of relief in cluster headache. Limitations and uncertainties acknowledged by the authors include the need for more direct measures of neural activity to validate the fMRI inferences, incomplete resolution of receptor‑specific mechanisms, and the temporal differences between imaging modalities. They also report attempts to exclude non‑neural confounds: physiological regression and breath‑hold CVR testing argued against simple vascular explanations for the observed fMRI effects. In conclusion, the investigators propose that their multimodal fMRI approach provides a detailed account of the brain correlates of the psychedelic state, characterised principally by decreased activity and connectivity in connector hubs, which they suggest permits the altered, unconstrained cognition characteristic of high‑dose psilocybin experiences.