Temporal structure of psychedelic EEG: statistical unit, dose, and receptor mechanism in multi-scale dynamics
Zenodo (CERN European Organization for Nuclear Research) September 7, 2026 DOI: 10.5281/zenodo.22585152 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Pre-registered secondary analysis of four public EEG datasets Double-blind Preregistered Peer reviewed |
|---|---|
| Sample size | 31 |
| Population | Healthy adults from public EEG datasets (psilocybin microdose: N=31 paired; ketamine: N=8; DMT: N=6; cross-dataset DMT sample: N=25) |
| Interventions | Psilocybin microdose Ketamine DMT |
| Dose | 0.5 g psilocybin (microdose) |
| Measures | Lempel-Ziv complexity, autocorrelation (0.5/1/2 s), detrended fluctuation exponent (alpha_DFA) |
| Key points | At the subject unit, psilocybin microdose showed no significant changes in Lempel-Ziv complexity (p=0.27, d_z=+0.23) or detrended fluctuation exponent (p=0.20, d_z=-0.23); previously reported increases were pseudoreplication artifacts. Full-dose ketamine and DMT both increased alpha_DFA (d_z=+1.74 and +2.30, respectively), indicating no receptor-specific dissociation on persistence. |
Abstract
Rationale: Increased neural complexity under psychedelics is replicated, but the temporal structure carrying it — and the design choices that decide what a "complexity increase" means — remain contested: window-level reports treat correlated windows as independent; cross-dataset contrasts confound pharmacology with protocol.
Objectives: To test, in public EEG datasets with pre-registered criteria, (i) whether multi-scale temporal structure changes under psilocybin microdose at the correct statistical unit; (ii) whether full-dose ketamine (NMDA antagonist) and DMT (5-HT2A agonist) diverge on the persistence axis; (iii) whether the microdose effect is moderated by blinding and trait.
Methods: Four pre-registered analyses on four public datasets: psilocybin microdose (0.5 g, double-blind crossover, N=31 paired), unblinding/trait stratification of the same subjects (N=23/8), ketamine (within-subject, N=8), and inhaled DMT (within-subject, N=6; cross-dataset descriptive sample, N=25). Metrics per subject: Lempel-Ziv complexity, autocorrelation (0.5/1/2 s), and the detrended fluctuation exponent.
Results: (i) At the subject unit, no microdose metric reached significance (LZc p=0.27, d_z=+0.23; alpha_DFA p=0.20, d_z=-0.23); the previously reported LZc increase (p=1.35e-3) was a pseudoreplication artifact of pooled windows. (ii) Ketamine and DMT both increased alpha_DFA in paired designs (d_z=+1.74, 8/8; d_z=+2.30, 6/6) — the apparent pharmacological dissociation (DMT alpha approximately 0.58) was a cross-dataset artifact. (iii) No moderator reached significance; the largest exploratory effect was in non-perceivers (N=8, d_z=0.91).
Conclusions: Microdose temporal dynamics are silent at realistic sample sizes (N approximately 200+ for d_z approximately 0.23); full-dose persistence rises under both receptor mechanisms, dissolving the 5-HT2A/NMDA dissociation on this axis. The design lessons — statistical unit, within-subject baselines, cross-dataset caution — generalize beyond psychedelics.