Baseline power of theta oscillations predicts mystical-type experiences induced by DMT in a natural setting

This was an observational, naturalistic study in which participants consumed DMT in the contexts they normally use the drug. Thirty-five volunteers (7 female; mean age 33.1 ± 6 years) were recruited between May and December 2019. Inclusion required at least two prior experiences with ayahuasca or DMT and willingness to use DMT in the presence of research team members; participants were asked to abstain from other psychoactive drugs for at least 24 hours. Pregnant individuals were excluded, and screening included a non-diagnostic psychiatric interview (SCID-CT) and exclusion for past difficult psychedelic experiences. The extracted text reports that six participants were excluded from EEG analyses due to excessive artefacts, leaving 29 subjects for electrophysiological analyses. DMT was self-administered or administered with a facilitator using freebase DMT recrystallised on non-psychoactive plant leaves and inhaled as smoke or vapour after combustion; facilitators typically withdrew the pipe when the participant stopped inhaling. The mean load of the pipes was estimated by participants or facilitators at 40 mg freebase DMT, and presence of DMT in samples was verified by high-performance liquid chromatography–mass spectrometry. The researchers did not supply the drug. EEG was recorded with a 24-channel mobile amplifier using standard 10-20 electrode positions. Baseline recordings included both eyes-open and eyes-closed periods (reported as five minutes each), and acute DMT recordings began at exhalation and continued until subjects reported return to baseline (mean duration 6 ± 1.4 minutes). Preprocessing included division into 2 s epochs, bandpass (1–90 Hz) and notch (47.5–52.5 Hz) filtering, automated and manual channel inspection and interpolation, rejection of artefactual epochs, and independent component analysis (Infomax) to remove artifactual components. On average 30% of channels were rejected (maximum eight channels), about 21.3 ± 13.7 epochs were removed per subject, and 2.7 ± 1.1 ICA components were removed. For source estimation the investigators used Brainstorm: preprocessed epochs were mapped to a standard MRI grid (4 mm), forward solutions were computed with a 3-shell sphere model, and sources were estimated with minimum norm imaging and dynamical Statistical Parametric Mapping (dSPM). Band-specific power spectral density was obtained via Welch's method and normalised; spectral power was summarised in standard bands (delta 1–4 Hz, theta 4–8 Hz, alpha 8–12 Hz, beta 12–30 Hz, low gamma 30–40 Hz, high gamma 40–70 Hz) and averaged across the 90 regions of the Automated Anatomical Labeling (AAL) atlas. Statistical comparisons of DMT versus baseline used Wilcoxon signed-rank tests. Predictive analyses of baseline activity used Spearman rank correlations between eyes-closed regional spectral power and questionnaire subscales (5D-ASC, MEQ30, NDE), with false discovery rate (FDR) control via the Benjamini–Hochberg procedure at q = 0.05 pooling p-values across bands and subscales. In addition, a machine learning regression analysis used a random forest regressor (1000 trees) trained on 50% of samples to predict three subscales (5D-ASC Unity, MEQ30 Mystical, MEQ30 Transcendence) from the 90-region theta power vector; this train/test split was repeated 1000 times with and without shuffling of targets to generate a permutation-derived p-value. The extracted text does not clearly report some procedural details such as exact cross-validation splits beyond the 50/50 split or hyperparameter tuning.

Contrasting source-localised spectral power during the acute DMT condition with the eyes-closed baseline, the investigators observed posterior increases in delta power and less marked increases in theta. Alpha power decreased in occipito-temporal regions under DMT, while beta power showed a mixed pattern — increases in posterior regions and decreases in prefrontal areas. In addition, gamma power increased in occipital, parietal and temporal regions. The authors note that these source-level effects largely replicate prior scalp-level reports and align with independent EEG studies of intravenous DMT, while also revealing additional significant changes in theta and beta bands. When correlating baseline (eyes-closed) regional spectral power with post-DMT questionnaire scores, FDR-corrected analyses identified significant associations only for the theta band. Specifically, lower baseline theta power in frontal and temporal regions was inversely associated with higher scores on three subscales: 5D-ASC Experience of Unity ("Unity"), MEQ30 Mystical experience ("Mystical"), and MEQ30 Transcendence of time and space ("Transcendence"). The extracted text emphasises these inverse relationships but does not provide exact correlation coefficients or confidence intervals for the regional correlations. In parallel analyses using baseline psychometric measures, trait absorption showed significant positive correlations with the mystical-type outcomes (R = 0.43 for 5D-ASC Unity and R = 0.51 for MEQ30 Mystical), consistent with previous survey work. The text indicates that a random forest regression using theta power across the 90 AAL regions was applied to predict Unity, Mystical and Transcendence scores and that a permutation procedure was used to assess significance; however, the extracted Results section does not clearly report numerical performance metrics (for example mean squared error, explained variance, or permutation p-values) for the machine learning models, so model-level results and their statistical significance are not ascertainable from the provided text.

Tagliazucchi and colleagues interpret their findings as evidence that baseline neural oscillations — particularly theta power in frontal and temporal sources — are related to the propensity to experience mystical-type phenomena under DMT. They argue that naturalistic recordings, while lacking the control of double-blind placebo-controlled trials, afford broad heterogeneity in set and setting that can be an advantage when searching for contextual predictors of psychedelic response. The discussion situates the DMT versus baseline source results within existing literature: posterior increases in delta and decreases in alpha replicate prior analyses, and occipito-temporal gamma increases are concordant with earlier reports on DMT and ayahuasca. Novel significant changes in theta and beta bands were noted at the source level and the authors suggest that mixed scalp sources could obscure these effects in non-source analyses. Regarding the theta–mystical link, the investigators propose mechanistic interpretations rooted in prior findings: theta oscillations have been associated with mind-wandering, anxious rumination, working memory and autobiographical memory retrieval — processes that might interfere with a mental set conducive to mystical-type experiences. Conversely, other literature links enhanced theta to mindfulness and sustained attention, which the authors acknowledge as an apparent contradiction; they suggest that different cortical generators of theta may underlie distinct cognitive roles and that the same frequency band can reflect different functions depending on source and context. The authors note they did not measure mind-wandering directly prior to DMT, so subjective or behavioural validation of the proposed disruptive role of theta is lacking. The investigators also report that baseline psychometric measures predicted some outcomes: trait absorption correlated with unity and mystical scores, whereas other experiential domains (perceptual, affective, cognitive) were not predicted by baseline EEG in this sample. They speculate that unmeasured individual factors, such as regional 5-HT2A receptor density, might explain variability in those other dimensions, and recommend that future prediction efforts incorporate richer neuroimaging modalities (functional, anatomical and neurochemical data). Key limitations acknowledged by the authors include the absence of a placebo condition and limited precision about administered dose — constraints inherent to naturalistic designs. They recommend replication under randomised double-blind placebo-controlled conditions while also noting that some contextual heterogeneity may be desirable when studying how set and setting influence psychedelic experiences. Finally, the discussion frames the study as an initial step toward predicting acute psychedelic effects from baseline neurophysiology and suggests that identifying and modulating predictors of mystical-type experiences could ultimately help to optimise therapeutic outcomes, a goal requiring further systematic research and larger datasets.