Processing of self-related thoughts in experienced users of classic psychedelics and non-users: a source localisation EEG study

The study analysed two independently collected datasets from different laboratories to evaluate replicability. Dataset I (reported N = 70) was collected at University SWPS and Dataset II (reported N = 38) at Jagiellonian University. Participants were recruited via online surveys distributed through Polish harm-reduction and psychedelic-related channels. Two subgroups were defined: ‘‘users’’ with at least 15 lifetime experiences of classic psychedelics and ‘‘non-users’’ who had never used psychedelics but expressed willingness to try them in future. The extraction contains some inconsistencies in reported counts (for example, the text states 36 users and 36 non-users were tested in Warsaw but also reports N = 70); the exact final sample after exclusions is therefore not unequivocally clear from the extracted text. Exclusion criteria included substantial lifetime use of other psychoactive drugs, current psychoactive medication, AUDIT-C scores ≥ 4 on any item, and psychiatric or neurological diagnoses. Participants were instructed to abstain from psychedelics for at least 30 days prior to EEG. Behavioural measures administered before EEG included the Ego Dissolution Inventory (recalled past experiences), the Rumination–Reflection Questionnaire (RRQ), Beck Depression Inventory II (BDI-II), and State–Trait Anxiety Inventory (STAI). The experimental manipulation was a 20-minute self-related thoughts (SRT) induction adapted from Nolen-Hoeksema’s rumination protocol: blocks of 30-second trials asked participants to focus on self-related statements or neutral/distraction statements, followed by ratings of task focus, self-focus and sadness on 1–10 scales. Participants kept their eyes closed during focus periods. EEG was recorded with high-density systems (Dataset II: 64-channel BioSemi ActiveTwo; Dataset I recording specifics are described elsewhere in the extraction). Electrode positions were digitised for most participants (captrak available for 84% of Dataset I). Preprocessing used MNE-Python: 1 Hz high-pass filtering, visual rejection of gross artifacts, ICA to remove stereotyped artifacts (eye blinks, movement, muscle), interpolation of bad channels and re-referencing to common average. Source reconstruction used DICS beamforming (Dynamic Imaging of Coherent Sources) applied to cross-spectral density (CSD) matrices computed with Morlet wavelets (seven cycles) across 4–30 Hz. Power spectral densities (PSDs) were estimated at source vertices. Statistical analyses were conducted in source space and focused on alpha (7–12 Hz) and beta (14–30 Hz) bands. The analysis pipeline comprised: (a) within-group whole-brain comparisons of SRT versus distraction using cluster-based permutation t-tests (cluster size limited to 100 vertices to improve spatial specificity); (b) ROI-based between-group tests using clusters derived from the within-group contrasts (clusters retained for ROI analysis only if >50 vertices); and (c) exploratory whole-brain between-group tests on condition differences (DIFF = PSD[SRT] – PSD[distraction]). ROI comparisons employed Welch t-tests and Bayesian linear regression to evaluate the added predictive value of group after adjusting for lifetime cannabis use and lifetime meditation hours. Bayesian regression compared a null model with confounders only against models including group. Shepherd correlation, a robust measure for EEG–behaviour relationships, was used to test associations between PSDs and behavioural/trait measures. Behavioural statistics used Mann–Whitney U tests and mixed ANOVAs for condition and group effects. Corrections for multiple comparisons (FDR/Holm) were applied to correlation results.

Demographics and questionnaires: Groups were matched on age, sex and education. Across datasets, users reported more meditation hours and greater lifetime use of dissociatives and empathogens; Dataset I users also reported higher cannabis use (caveat: cannabinoid data was collected only for subsets due to a programming error). In Dataset I, users scored significantly lower on state and trait anxiety (STAI I/II), depression (BDI-II) and rumination (RRQ) than non-users. In Dataset II, no group differences emerged on anxiety, depression or rumination, but users reported higher reflection scores on the RRQ. Behavioural response to SRT induction: Mixed ANOVAs showed that both datasets produced the anticipated within-subject manipulation: participants reported higher self-directed attention and higher sadness during the SRT condition than during distraction (large effects: η2 up to 0.89 for self-focus). Dataset I showed a main effect of group on attention to the statement (users reported greater attention overall) and an interaction indicating users reported less sadness during SRT than non-users. Dataset II did not replicate these group or interaction effects; it showed only a condition effect for attention to the statement. EEG within-group comparisons: In both datasets alpha power was higher during SRT than distraction in medio-posterior and ventral-temporal regions (Dataset I: two alpha clusters ≥ 50 vertices, p = 0.014 and p = 0.016; Dataset II: cluster p < 0.001). Beta power increases during SRT were observed but reached statistical significance only in Dataset II (Dataset I beta clusters were present but cluster p-values did not reach conventional significance thresholds). Between-group ROI and whole-brain contrasts (Dataset I): In Dataset I, the DIFF (SRT minus distraction) in alpha band averaged over ROI cluster A (temporo‑occipital/parietal region) was significantly smaller for users than for non-users (Welch t = 2.51, p = 0.015; permuted p = 0.02). Bayes Factor (BF10) for this contrast was 3.24, interpreted as substantial evidence favouring the group difference. Cluster B in alpha did not reach significance (t = 1.724, p = 0.09; BF10 = 0.868). In the beta band, several ROIs (labelled C, E, F) showed smaller DIFF for users versus non-users (text indicates significance but precise statistics for each ROI are not fully presented in the extracted fragment). Whole-brain cluster-based tests in Dataset I also identified occipital, medial and lateral clusters with smaller DIFFs in users for alpha (clusters 1a, 1b p = 0.022 and p = 0.044) and for beta (clusters with p ~0.026 and p = 0.05), with spatial coverage in occipital and ventral‑temporal regions. Between-group analyses (Dataset II): Predefined ROI comparisons found no significant group differences in alpha or beta bands. Whole-brain tests identified no clusters for between-group DIFF in Dataset II; the investigators therefore did not pursue further exploratory analyses on that dataset. Bayesian regression and confounder control: Bayesian model comparisons incorporating lifetime meditation hours and lifetime cannabis use often favoured null models (confounders only) at the ROI level, although evidence was generally weak (BF10 range 0.25–1.27). For whole-brain alpha clusters in Dataset I, models including group had stronger evidence than null models (BF10 = 3.99 and BF10 = 5.843). In the beta band whole-brain clusters, evidence modestly favoured null models (BF10 ~0.6–0.7), indicating that confounders attenuated group effects in some analyses. Correlations: Shepherd correlations showed initial positive relationships between cluster-averaged PSDs and lifetime psychedelic use, and between PSDs and BDI-II scores in several clusters (pi between 0.30 and 0.54). However, after correction for multiple comparisons (FDR/Holm), none of these correlations remained significant. Lifetime psychedelic use did not correlate with BDI-II (pi = -0.03, p = 0.873).

Ruban and colleagues interpret their results as partially consistent with the hypothesis that experienced psychedelics users exhibit attenuated electrophysiological responses during induced self-related thought. Within both datasets the SRT manipulation reliably increased alpha power (and beta power in Dataset II) in medio‑posterior and ventral‑temporal regions implicated in self-related processing. In Dataset I, users showed smaller increases in alpha and beta power during SRT relative to distraction, concentrated in temporo‑occipital, medio‑posterior and insular regions; this pattern is congruent with prior findings linking posterior alpha/beta dynamics and DMN activity to self-related cognition, and with acute psychedelic effects that reduce posterior alpha/beta power. The authors caution that the between-group electrophysiological effects did not replicate in Dataset II. They note several plausible explanations: Dataset II had fewer participants (lower statistical power), lacked individual MRI scans and in some cases electrode digitisation (reducing source localisation precision), and thus may have been less sensitive. Bayesian analyses further suggested that lifetime meditation and cannabis use reduced the apparent group effect in multiple ROIs, indicating these behaviours may confound or share variance with long-term psychedelic exposure. Exploratory correlations between PSDs and behavioural measures produced nominal associations with lifetime psychedelic use and depression scores but none survived correction for multiple comparisons, limiting interpretability. Key limitations acknowledged by the investigators include the cross-sectional, naturalistic design that precludes causal inference and precise characterisation of dose, context and quality of psychedelic experiences. Source localisation was constrained by incomplete MRI/captrak coverage, and cluster-based permutation methods limit claims about precise anatomical or frequency localisation. Sample selection biases are also discussed: recruitment via harm-reduction and psychedelic-interest channels may have produced a self-selected, potentially unrepresentative sample whose prior experiences, expectations and preparedness differ from broader psychedelic-using populations. The authors recommend future longitudinal and better-powered studies with comprehensive control for confounders, richer characterisation of subjective experience and improved neuroimaging localisation to clarify whether and how naturalistic psychedelic use relates to long-term alterations in self-related information processing. Overall, the paper presents preliminary evidence from one dataset suggesting attenuated alpha/beta increases during self-related thought in experienced psychedelic users, but emphasises non-replication in an independent dataset and the need for cautious interpretation and further research.

The study represents an initial source-localised EEG investigation of self-related thought processing in naturalistic users of classic psychedelics versus non-users. Dataset I showed that users reported lower rumination, depression and anxiety and exhibited smaller SRT-related increases in alpha and beta power in regions implicated in self-processing. These electrophysiological effects were not replicated in Dataset II. Given methodological limitations, potential confounders and lack of replication, the authors conclude that their findings are tentative and that further rigorous, better‑powered and longitudinal research is required to clarify long-term effects of psychedelic use on self-related neural processing.