Neural complexity is increased after low doses of LSD, but not moderate to high doses of oral THC or methamphetamine

Earlier research has linked neural complexity, often measured with EEG signal-diversity metrics, to levels of consciousness: complexity falls during sleep, anaesthesia and coma and rises during certain altered states. Psychedelics such as LSD have been reported to increase neural complexity at high doses, and the entropic brain hypothesis proposes that such increases may destabilise maladaptive patterns of thought and behaviour. At the same time, low-dose or "microdosing" practices have become popular and controlled studies have shown some acute improvements in well-being after low doses, but objective neural markers at these doses remain poorly characterised. Murray and colleagues set out to determine whether low doses of LSD (13 and 26 µg) increase neural complexity in the absence of overt psychedelic-like altered states, and to compare those effects with two other drugs that do alter consciousness and mood at the tested doses: oral THC (7.5 and 15 mg) and methamphetamine (MA; 10 and 20 mg). The investigators focused on resting-state Lempel–Ziv (Limpel-Ziv in the text) complexity as the primary neural outcome, alongside spectral power and self-reported subjective and mood measures, to clarify whether increased complexity is necessary or sufficient for altered states of consciousness.

Three independent, but procedurally aligned, within-subject studies were conducted in healthy volunteers between 2020 and 2022 at a single laboratory. Each participant attended three 5-hour sessions and received placebo and two active doses of one drug (LSD, THC or MA) under double-blind, randomised conditions, with sessions separated by at least 7 days. EEG and self-report measures were obtained during the anticipated peak drug effects. Participants were healthy adults aged 18–35 (N = 73 overall; 21 in the LSD study, 23 in the THC study, 29 in the MA study; 33 women). Screening included physical exam, ECG, a modified SCID for DSM-5 and drug-use history. Common inclusion criteria were BMI 18–32 kg/m2, English fluency and at least high-school education; exclusions included history of psychosis, severe PTSD or panic disorder, recent substance use disorder (past year, except nicotine), pregnancy and current medication (apart from birth control). The THC study additionally excluded frequent lifetime cannabis users and required abstinence from cannabis for 30 days prior to participation. Drug administration: LSD (13 or 26 µg as tartarate solution) was given sublingually in 0.5 mL; THC (dronabinol, 7.5 and 15 mg) and MA (Desoxyn, 10 and 20 mg) were administered orally in opaque capsules. Doses were chosen to reflect low or microdose ranges for LSD and clinically relevant or behaviourally active doses for THC and MA. Participants were told they might receive placebo, stimulant or sedative, and possibly a hallucinogenic or cannabinoid drug, to minimise expectancies. Subjective measures included the Drug Effects Questionnaire (DEQ) item "Do you feel a drug effect?" on a 0–100 mm visual analogue scale, and selected Profile of Mood States (POMS) subscales for anxiety and elation. At session end, the 5D-ASC questionnaire was completed in the LSD and THC studies to assess altered states of consciousness across five subscales. EEG acquisition used a 19-channel 10–20 montage (BioSemi ActiveTwo), recording 5 minutes of eyes-closed resting activity 90–120 minutes after dosing. Neural complexity was quantified using a Lempel–Ziv algorithm (LZ76 as described), where each channel's signal was binarised relative to its mean and normalised to an entropy rate yielding values between 0 and 1. Spectral power was computed for delta (1–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), beta (13–30 Hz) and gamma (30–80 Hz). Statistical analyses used repeated-measures ANOVA (dose as within-subject factor) with planned contrasts comparing each dose to placebo and tests for linear dose effects. EEG electrode-level follow-ups were corrected for multiple comparisons with an FDR of 0.05. Pearson correlations explored relationships between EEG metrics (from the highest dose condition) and self-report outcomes.

Sample characteristics: Mean participant age was in the mid-twenties; overall N = 73 with 33 women. The LSD group had about twice as many males as females. Lifetime cannabis use differed by design: the THC cohort reported a mean of about 12.6 lifetime uses, whereas the LSD and MA groups reported much higher lifetime use (~250), reflecting the THC study's exclusion criteria. Subjective effects: All three drugs produced dose-dependent increases in the DEQ item "Do you feel a drug effect?" (LSD dose×time F1,20 = 11.00, p = 0.003; THC F1,22 = 82.96, p < 0.001; MA F1,28 = 8.87, p = 0.006). However, the lower LSD dose (13 µg) did not differ significantly from placebo on that measure (F1,20 = 2.80, p = 0.110), consistent with a sub‑perceptual dose. Mood measures showed that LSD and THC increased anxiety (LSD: F1,20 = 4.97, p = 0.038; THC: F1,22 = 22.01, p < 0.001) and LSD and MA increased elation (LSD: F1,19 = 4.68, p = 0.043; MA: F1,28 = 13.29, p < 0.001). On the 5D-ASC, the low doses of LSD produced no change on any subscale, whereas THC increased all five subscales (for example, OBM F1,22 = 12.77, p = 0.002; VR F1,22 = 70.38, p < 0.001). Neural complexity and spectral power: Resting-state Lempel–Ziv complexity increased dose‑dependently after LSD (dose F1,20 = 19.49, p < 0.001), with a broadly distributed topography but relative absence of increases at midline sites including Pz. No significant effects on LZ complexity were observed after THC or MA. In spectral analyses, LSD reduced low-frequency delta and theta power and increased gamma power, without affecting alpha power across the 10–20 montage. By contrast, THC reduced alpha power (primarily frontal), while MA increased alpha power and also increased beta and gamma power, with frontal emphasis for the alpha effects. Associations between EEG and subjective measures: LZ complexity after the higher LSD dose (26 µg) did not correlate with any subjective ratings, including felt drug effect, anxiety, elation, or any 5D-ASC subscale. However, LSD-induced reductions in delta and theta power were associated with increases in elation; participants who reported greater elation after 26 µg also reported higher anxiety. After 15 mg THC, lower alpha power related to two 5D-ASC subscales (dread of ego dissolution and vigilance reduction). No EEG–subject relationships were found for the MA study. Age-related supplementary finding: In placebo data from the THC study, adult (30–35 years) relative to adolescent-aged (18–20 years) participants showed globally reduced low-frequency oscillations and increased neural complexity, mirroring the LSD-versus-placebo pattern.

Murray and colleagues interpret their primary finding as evidence that low doses of LSD (13 and 26 µg) increase resting-state Lempel–Ziv complexity without producing retrospective reports of altered states on the 5D-ASC. This dissociation leads the investigators to conclude that increased neural complexity is neither necessary nor sufficient for psychedelic-like altered states: THC produced altered-state ratings without increased complexity, and low-dose LSD increased complexity without altered-state reports. The authors situate these results relative to prior work showing increased complexity at high psychedelic doses and to one microdosing study that did not observe complexity changes, arguing that their findings nuance the view of complexity as a biomarker of the psychedelic state. The discussion links possible mechanisms to 5-HT2A receptor activity, which can increase neuronal sensitivity and excitatory transmission, potentially generating greater signal diversity even at low receptor engagement. The researchers note that complexity has been associated in previous work with better task performance, developmental maturation and clinical response prediction, and that complexity is generally reduced in several psychiatric conditions. They speculate that the lateralised pattern observed (with reduced complexity at midline Pz) might reflect network-level changes such as default mode network disintegration; complementary task-based data in the same participants reportedly showed increased reward responses and altered face‑processing accuracy after low-dose LSD. On spectral power, the authors emphasise a dissociation between complexity and oscillatory measures: changes in LZ complexity did not map onto oscillatory power changes. They highlight alpha desynchronisation as a more sensitive correlate of altered states of consciousness, given that THC and higher-dose psychedelics previously reported reduced alpha alongside subjective alterations, whereas low-dose LSD altered delta/theta and increased gamma. The discussion posits functional interpretations for alpha changes, suggesting THC-related alpha reductions may reflect active, content-rich altered states, while MA-related frontal alpha increases might index reduced content or different cognitive states. Limitations acknowledged by the investigators include pooling data from three separate studies (albeit with identical EEG procedures and within-subject designs), group differences in lifetime cannabis exposure due to enrolment constraints, and the absence of a dedicated measure of "richness of experience" that has been tied to Lempel–Ziv complexity in other studies. The authors recommend future work to determine whether LZ increases after low-dose LSD relate to cognitive, behavioural or therapeutic outcomes, and to further examine the role of alpha desynchronisation in mediating altered states of consciousness.