Lasting effects of a single psilocybin dose on resting-state functional connectivity in healthy individuals

Psilocybin is a serotonergic psychedelic whose acute effects include profound alterations in consciousness lasting about six hours and characterised by sensory and affective changes. Previous clinical and healthy-volunteer studies have reported persistent improvements in mood, well-being and personality traits (for example, increased openness and mindfulness) after one or a few moderate-to-high doses. Although acute effects on resting-state functional connectivity (RSFC) have been documented — including transient disruption of the default mode network (DMN) and increased between-network coupling during the drug experience — there is limited and mixed evidence regarding persistent changes in brain connectivity after the acute effects subside. Few studies have imaged the brain beyond the immediate post-session period, and none, according to the authors, have robustly linked longer-term RSFC changes to concomitant psychological changes or PET measures of 5-HT2A receptor binding. Drummond and colleagues therefore set out to evaluate lasting changes in RSFC following a single psilocybin dose in a small cohort of healthy, psychedelic-naïve volunteers. The study measured within-network and between-network RSFC at baseline, one-week and three-months post-administration, attempted to replicate a prior region-to-region analysis, and explored whether RSFC changes related to acute phenomenology, persisting psychological measures (for example, personality and mindfulness), and neocortex 5-HT2A receptor binding assessed with [11C]Cimbi-36 PET. The aim was to identify candidate brain-system changes that might mediate lasting behavioural effects of psilocybin.

This was an open-label imaging study of 10 healthy, psychedelic-naïve volunteers (6 male, 4 female; mean age 28.3 ± 3.4 years). Participants were screened for somatic and psychiatric illness and gave informed consent. Psilocybin was administered in a controlled setting with two supporting psychologists present; dosing was 0.2 mg/kg (n = 4) or 0.3 mg/kg (n = 6) given orally in capsule form. Subjective drug intensity was sampled verbally every 20 minutes during the session. Participants completed questionnaires at baseline and at three-month follow-up, including the NEO Personality Inventory-Revised (NEO PI-R), the Mindfulness Attention and Awareness Scale (MAAS) and the Persisting Effects Questionnaire (PEQ). Acute subjective measures included the 11-dimension Altered States of Consciousness questionnaire (11D-ASC), the MEQ30 mystical experience questionnaire and the Ego-Dissolution Inventory (EDI). Neuroimaging comprised three identical MRI sessions (baseline, one-week and three-months) on a 3T scanner and two PET sessions (baseline and one-week) using the agonist radioligand [11C]Cimbi-36 to estimate neocortex 5-HT2A receptor binding (BPND). Resting-state BOLD data consisted of 10 minutes with eyes closed; participants were instructed to let their minds wander but not fall asleep. fMRI preprocessing used SPM12 and CONN: slice timing correction, realignment/unwarping, coregistration and segmentation, normalisation to MNI space and spatial smoothing (8 mm FWHM). Denoising included bandpass filtering (0.008–0.09 Hz), aCompCor physiological component regression (first five components for white matter and CSF), motion parameter regression with derivatives, and censoring of outlier volumes using ART (global variance and composite motion thresholds). Connectivity analyses used an a priori atlas defining 36 regions across seven canonical resting-state networks: DMN, dorsal attention (DAN), executive control (ECN), salience (SN), sensorimotor (SMN), visual (VN) and auditory (AN). Within-network connectivity was defined as the mean Fisher z-transformed correlation between all pairs of ROIs in a network; between-network connectivity was the mean across ROIs in two different networks. A secondary analysis used a 268-region Shen atlas to attempt replication of a prior region-to-region approach. Statistical testing was performed in R using paired t-tests comparing baseline to one-week and baseline to three-months separately. The authors adjusted p-values across the 28 within- and between-network comparisons with the Bonferroni–Holm method and reported Cohen's d effect sizes. Exploratory post-hoc analyses were performed for any network effect that passed the primary threshold, but the authors caution that these are not inferential due to limited power. For the Shen268 analysis, they followed a two-stage approach similar to the prior study but applied Bonferroni–Holm correction across the tested edges to control family-wise error. Correlations between RSFC change and behavioural or PET measures were reported descriptively and via latent-variable modelling for PEQ subscales.

All 10 participants completed the imaging sessions; psilocybin was well tolerated and no serious adverse events occurred. Mean time intervals were reported: baseline to intervention ~15 days, intervention to one-week rescan ~6.5 days, and intervention to three-month rescan ~101.5 days. Motion and number of censored volumes were low and did not differ significantly between time points. The principal finding was a statistically significant decrease in within-network RSFC of the executive control network (ECN) at one-week post-psilocybin. The reported statistics were punc = 0.00039, pFWE = 0.010 and Cohen's d = -1.73, with nine of ten participants showing reduced ECN RSFC at one-week. At three-months ECN RSFC remained numerically decreased relative to baseline but the effect was not statistically significant (punc = 0.23, pFWE = 1; Cohen's d = 0.4 as reported). No other within- or between-network connectivity estimates survived correction for multiple comparisons at either one-week or three-months. Some medium-sized uncorrected effects were noted: at one-week SMN-SMN, ECN-VN and DAN-AN connectivity decreased (Cohen's d ≈ -0.6) while DMN-ECN connectivity increased (Cohen's d ≈ 0.5). At three-months ECN-VN and DAN-AN connectivity remained decreased (Cohen's d ≈ -0.6 and -0.5 respectively). In the Shen268 region-to-region replication framework, 405 edges showed evidence for significant connectivity at baseline under the initial filter step; applying paired tests, 25 edges changed at one-week (19 increases, six decreases) and 18 edges changed at three-months (12 increases, six decreases) at an uncorrected p < 0.05. However, none of these edges remained statistically significant after Bonferroni–Holm correction across the 405 tests. Exploratory association analyses linked the one-week decrease in ECN RSFC to psychological and PET measures. The positive subscales of the PEQ loaded onto a single latent construct (p < 10^-6) and change in ECN RSFC at one-week was negatively associated with that latent positive PEQ factor. The investigators also report that greater ECN disintegration at one-week (and lesser disintegration at three-months) correlated with increased mindfulness at three-months and with decreases in neocortex 5-HT2A binding at one-week. Change in ECN RSFC was not associated with measures of the acute psychedelic experience, and trait personality changes such as openness, neuroticism and conscientiousness were not correlated with ECN RSFC change in these analyses.

Drummond and colleagues interpret their findings as indicating that a single psilocybin administration produces a transient decrease in ECN resting-state connectivity detectable at one-week but not at three-months, suggesting this may be an afterglow-period phenomenon. They propose that ECN disintegration could be linked to improvements in traits such as mindful awareness and to positive persisting effects reported three months later, and they note associations with reductions in neocortex 5-HT2A binding that could mechanistically relate receptor-level change to network-level alterations and behaviour. The authors highlight that ECN is implicated in executive functions (for example, cognitive flexibility and attentional control) and that alterations in ECN connectivity have been reported in conditions such as depression, obsessive–compulsive disorder and addiction; accordingly, ECN modulation is a plausible candidate mechanism for some therapeutic effects of psilocybin. The discussion situates the null or small effects on other networks in context: despite large subjective and reported personality changes after psilocybin, the authors did not find widespread, lasting RSFC reorganisation beyond the one-week ECN effect. They caution that persistent alterations in the DMN are unlikely to be the dominant long-term mechanism, given the small DMN effect observed and convergence with prior studies that did not find sustained DMN changes. The attempted replication of a prior region-to-region finding reproduced the scale of candidate edges but, after correction for multiple comparisons, yielded no robust effects; this highlights the power limitations of exploratory edgewise analyses in small samples. Key limitations are emphasised: the small sample size (n = 10) limits statistical power and generalisability, and resting-state atlas definitions vary across studies which complicates comparisons. The investigators did not measure plasma psilocin levels, so individual pharmacokinetic variability could not be related to connectivity changes. They suggest alternative analytic approaches (for example, dynamic connectivity or entropy measures), task-based fMRI, and integrated PET–fMRI in larger clinical and healthy cohorts to further map the neurobiological pathways by which psilocybin may produce lasting psychological effects. Overall, the authors present ECN modulation at one-week as a candidate neural correlate worthy of follow-up in adequately powered studies.