Psilocybin enhances insightfulness in meditation: a perspective on the global topology of brain imaging during meditation

Berit and colleagues situate their work at the intersection of two rapidly expanding literatures: meditation neuroscience and psychedelic research. Earlier studies have documented overlapping phenomenology and some shared neurophysiological signatures between meditation and classical serotonergic psychedelics (e.g. changes in self-consciousness, reduced default mode network activity, altered functional connectivity and increased markers of plasticity). However, most prior work has used conventional measures such as functional connectivity, resting‑state networks, signal variability and entropy. The authors identify a gap: more holistic, geometry‑aware approaches that capture the global spatiotemporal organisation of whole‑brain activity during different meditation styles and under psychedelic influence have not been widely applied to these states. This study applies a topological data analysis (TDA) method based on the Mapper algorithm to fMRI collected before and after a five‑day psilocybin‑assisted meditation retreat in experienced meditators. The authors aim to characterise the ‘‘topological landscape’’ of meditative brain states (focused attention, open monitoring, resting state), test whether psilocybin modulates that landscape, and relate topological descriptors (for example optimal transport distances and graph centrality measures) to subjective reports from an altered‑states questionnaire. They hypothesise that psilocybin will alter the organisation of meditative brain states (consistent with increased informational richness) and that topological measures will relate to experiential dimensions such as meta‑awareness, insightfulness and positive derealization.

Design and participants: This was a double‑blind, randomised, placebo‑controlled parallel study of a five‑day silent meditation retreat (sesshin) in experienced meditators. Thirty‑eight participants were enrolled and two were excluded from analysis, yielding 36 subjects (18 per group: psilocybin and placebo). Participants were matched on age, sex, lifetime meditation experience and dispositional mindfulness; mean age reported was 51.66 ± 8.32 years. Psilocybin was administered on the fourth day at a dose of 315 µg/kg body weight (mean absolute dose 21.82 ± 3.7 mg); placebo was lactose. The Five‑Dimensional Altered States of Consciousness (5D‑ASC) questionnaire and 11 lower‑order scales (11‑ASC) were collected 360 minutes after ingestion as retrospective measures of subjective effects. Imaging and experimental tasks: Each participant underwent two fMRI sessions: one day before and one day after the retreat. Each session comprised three fixed, eyes‑closed 7‑minute blocks (resting state, focused attention (FA), then open monitoring (OM)), producing 21 minutes (600 frames after slicing) per session after concatenation and exclusion of transition frames. FA instructions focused on breath, OM on open observation without attachment; during resting state subjects were instructed to rest but not meditate. MRI acquisition used a 3T Philips Achieva scanner with standard structural and T2*-weighted functional sequences (TR = 2000 ms, TE = 35 ms). Preprocessing followed a standard pipeline (C‑PAC) including motion correction, skull stripping, ANTs registration to MNI152 template, aCompCor nuisance regression, temporal bandpass (0.009–0.08 Hz), spatial smoothing (FWHM 4 mm) and z‑scoring. Topological pipeline and graph measures: The Mapper algorithm (using UMAP for the filter function) was applied to subject‑specific concatenated fMRI time series (middle 100 frames of each labeled block). The projected space was covered with overlapping bins (30 bins, 50% overlap), DBSCAN clustering of slices produced nodes representing clusters of similar fMRI frames, and nodes were connected when clusters shared frames, yielding a Mapper shape graph per subject. Nodes were annotated by meditation state labels. From each graph the investigators computed several topological descriptors per state: degree centrality (weighted mean degree of nodes for a label), closeness centrality (weighted mean reciprocal shortest path length), diameter of the set of nodes for a label, and pairwise optimal transport (OT) distances between label distributions on the graph (1‑Wasserstein / Earth Mover’s distance). The OT distance quantifies the minimum transport cost to transform one label’s node distribution into another given graph geodesic distances. Statistics and associations with phenomenology: Primary inferential tests were two‑way repeated‑measures ANOVAs with group (psilocybin vs placebo) as between‑subject factor and session (pre vs post retreat) as within‑subject factor, applied separately to each graph measure and state (or tuple of states). For OT distances comparing a state pre vs post (d(*1,*2)) only paired t tests were used to test drug effects. Correlations between graph measures and psychometric scales (35 measures) used Kendall or Pearson correlations as appropriate, with multiple comparisons correction: Benjamini‑Yekutieli for t‑tests and an eigenvalue‑based estimate of 10 effective tests for psychometric correlations. Significant correlations informed mixed linear models (session and psychometric measure as fixed effects, subject as random effect) and linear regression models examining insightfulness with predictors such as positive derealization and d(OM1,OM2).

Sample and scans: Thirty‑six experienced meditators (18 psilocybin, 18 placebo) completed fMRI one day before and one day after the five‑day retreat. Psilocybin dosing averaged 21.82 ± 3.7 mg (315 µg/kg). Each session included three 7‑minute blocks (RS, FA, OM) and graphs were constructed from concatenated middle frames to reduce transition noise. Main topological findings: The global effect of the retreat was a shift in resting‑state fMRI activity across all three states, evidenced by nontrivial OT distances d(OM1,OM2), d(RS1,RS2) and d(FA1,FA2). This retreat‑induced shift tended to be larger than differences between meditation styles measured within the same day. Specific psilocybin effects were concentrated on OT distances involving open monitoring and resting state. Optimal transport distances and subjective experience: The OT distance d(OM2,RS2) (open monitoring vs resting state after the retreat) showed a significant group difference, with greater mean and variance in the psilocybin group (independent t test t20.6 = -2.7, p = 0.010). Across subjects, d(OM2,RS2) correlated positively with the ASC subscale positive derealization (Kendall r = 0.379, p = 0.00118; corrected p = 0.0118). Mixed linear models for outcome d(OM*,RS*) with positive derealization and session as predictors showed a significant main effect of retreat (t = -2.624, p = 0.0129) and a significant interaction between retreat and positive derealization (t = 2.315, p = 0.0268); analogous models run in the psilocybin group yielded comparable significant retreat and interaction terms (retreat t = -2.293, p = 0.0358; interaction t = 2.446, p = 0.0264). A median split showed that for participants with low positive derealization, d(OM*,RS*) decreased due to the retreat (t = 2.3348, p = 0.03207), whereas no retreat effect appeared in the high derealization subgroup. Pre/post changes and drug modulation of OM and RS: Psilocybin increased the OT distances comparing the same state pre vs post: d(OM1,OM2) was larger in the psilocybin group (t25.7 = -2.2, p = 0.037), and d(RS1,RS2) showed a similar group difference (t26.6 = -2.2, p = 0.04). Additionally, the OT distance measures d(OM1,OM2), d(RS1,RS2) and d(OM2,RS2) were strongly intercorrelated and exhibited significant drug effects. Relations with insightfulness and ASC: Insightfulness correlated strongly with positive derealization (Pearson r = 0.82, p = 6.56e-10). Linear models predicting insightfulness with positive derealization and d(OM1,OM2) were significant and explained substantial variance: for the whole sample the model F(3,31) = 25.1, p = 1.96e-08 with multiple and adjusted R‑squared 0.7084 and 0.6802; in the psilocybin subgroup F(3,13) = 7.061, p = 0.00464 with multiple and adjusted R‑squared 0.6197 and 0.5319. Interaction terms between positive derealization and d(OM1,OM2) trended toward significance (p ≈ 0.075–0.081). Changes in graph centrality and diameter: Across both groups, retreat produced significant increases in measures for the OM node: degree centrality (ANOVA F1,34 = 24.8, p corr = 5.5e-5; paired t t35 = -5.04, p corr = 4.2e-5), closeness centrality (ANOVA F1,35 = 20.8, p corr ≈ 9e-5; paired t t35 = 4.6, p corr = 1.5e-4), and diameter (ANOVA F1,35 = 8.7, p corr = 0.017; paired t t35 = 2.97, p corr = 0.016). The OT distance between FA and OM decreased after the retreat across subjects (session effect F1,34 = 9.7, p = 0.0038; paired t t35 = 3.14, p = 0.0034). No significant effects were found for FA–RS OT distances. An exploratory group difference showed greater FA degree centrality post‑retreat in the psilocybin group, but this effect was near nonsignificant and appeared present pre‑retreat as well. Timing and acute effects: Importantly, post‑retreat fMRI was acquired approximately 48 hours after drug administration, so the observed psilocybin‑related topological differences are described as persisting beyond acute intoxication in this setting.

The investigators interpret their findings as demonstrating that TDA, operationalised via the Mapper algorithm and OT distances, provides a useful, complementary lens on the organisation of whole‑brain activity across meditation states and how this organisation is modulated by a psilocybin‑assisted meditation retreat. They argue that OT distance quantifies similarity or overlap of whole‑brain activation patterns between labelled states and can be read conceptually as a measure of information flow or geometric dissimilarity between state distributions on the Mapper graph. Authors emphasise that the retreat produced a consistent shift in resting‑state activity across FA, OM and RS that exceeded within‑day differences between meditation styles, indicating that the Mapper approach is sensitive to broad retreat effects. Superimposed on this stable topology were specific psilocybin‑related alterations, notably larger OT distances between OM and RS post‑retreat that correlated with positive derealization. The authors link these observations to the experiential construct of insightfulness: psilocybin increased positive derealization, which associated with larger OM–RS dissimilarity, and insightfulness was best explained by positive derealization together with the magnitude of OM pre/post change (d(OM1,OM2)). From this pattern they propose a mechanistic interpretation in which meditation‑trained increases in meta‑awareness and ‘‘aperture’’ (represented by increases in OM centrality and diameter) combined with psilocybin‑induced enrichment of informational content support greater insightfulness. In placing results relative to prior work, the authors note convergences with studies reporting increased brain entropy with psychedelics and changes in default‑mode and task‑positive network dynamics with both meditation and psychedelics. They suggest the combined state may be ‘‘divergent yet synergistic’’: psilocybin increases informational richness while meditation provides a stabilising meta‑awareness that preserves the capacity for reflection. The paper proposes that topological graph measures could serve as candidate brain markers for beneficial psychological processes in psychedelic‑assisted interventions. The authors acknowledge several limitations reported in the extracted text. These include the novelty and evolving nature of the TDA pipeline (choice of UMAP as filter may affect results), comparatively short fMRI block durations (7 minutes per condition), concatenation of blocks and exclusion of transitions, and the challenge of defining and controlling ‘‘task performance’’ in meditation settings. Additional limitations noted are the lack of a non‑meditator control group and absence of a psilocybin‑without‑retreat arm, which limit inferences about the necessity of the meditation context. The authors recommend future work to annotate Mapper graphs with intrinsic network labels and entropic measures, and to combine topological analysis with richer first‑person phenomenological methods (neurophenomenology) to deepen understanding of how changes in entropy and topology relate to conscious experience.