Psychedelic drug use and schizotypy in young adults

Earlier research into classical serotonergic psychedelics (for example LSD, psilocybin, DMT, mescaline) has re-emerged over the past decades with growing interest in their therapeutic potential. At the same time, investigation of possible harms has lagged behind, and the long-standing question of whether psychedelic use is linked to psychosis-spectrum outcomes remains unresolved. Some smaller studies reported associations between psychedelic use and psychosis-like symptoms, whereas larger population studies have not confirmed such links; differences in assessment strategies and the ability to detect subclinical, dimensional traits (for example schizotypy, cognitive biases such as Bias Against Disconfirmatory Evidence, and learning abnormalities) may underlie these discrepant findings. Lebedev and colleagues set out to examine whether past psychedelic use in otherwise healthy young adults is associated with schizotypal traits and cognitive biases typically observed in the schizophrenia spectrum. The study aimed to address limitations of prior cross-sectional work by combining a large online survey (n = 1,032) with an experimental subsample (n = 39) that completed behavioural tasks probing evidence integration (BADE task) and instructed aversive reversal learning, and by collecting follow-up data on frequency and recency of drug use to derive an exposure metric. The pre-registered hypothesis was that psychedelic use and greater total exposure would be associated with higher schizotypy and with impairments in evidence integration and reversal aversive learning.

The investigators used a cross-sectional design with an online survey and a laboratory-based experimental arm. Recruitment targeted Swedish adults via social media and research-recruitment platforms, including forums likely to include recreational and scientific users of psychedelics. The survey screened 1,032 participants; a subset of 701 participants aged 18–35 with no neurological, psychiatric or serious medical illness and no history of head trauma constituted the "young healthy" sample for focused analyses. An experimental subsample of 39 subjects completed behavioural testing. The study was approved by the regional ethics board, preregistered on OSF, and analysis scripts and modelling code were made available by the authors. The primary survey outcome was a composite z-score measure of schizotypy formed by merging scores from the Peters Delusion Inventory (PDI) and O-LIFE. A follow-up survey collected extended drug-use histories to derive a composite exposure measure merging frequency and temporal proximity of use. The authors explicitly asked respondents not to attribute questionnaire items to acute drug effects. Representativeness of subsamples was assessed with analysis of variance; demographic and psychopathology features (for example ADHD and autism-spectrum symptoms) were compared and reported as representative of the screened sample. Two behavioural tasks were used in the experimental arm. The Bias Against Disconfirmatory Evidence (BADE) task required participants to rate plausibility of four interpretations of scenarios across three sequential hints; the Evidence Integration Impairment (EII) score was computed from plausibility ratings following established approaches (summing Absurd ratings after each hint, plus late Lure ratings, minus the final True rating). The instructed aversive fear reversal task used individually titrated mild electric shocks as the unconditioned stimulus and two angry faces as conditioned stimuli; subjects were explicitly informed about contingencies and three rule reversals occurred during the task. Physiological responses were measured with pupillometry and skin conductance response (SCR); peak change within an eight-second window after CS offset (relative to a 500 ms pre-stimulus baseline) was the metric. Behavioural data were modelled using a modified Rescorla–Wagner framework implemented in Stan, estimating parameters such as a ρ-parameter that indexed sensitivity to instructed information. Statistical analysis employed general linear modelling in R, adjusting for demographics, psychiatric diagnoses and concomitant drug use, with post-hoc group comparisons where appropriate. Normality and outliers were checked (Shapiro–Wilk, 1.5*IQR), and supplementary analyses examined schizotypy subdomains. The authors report use of intention-to-analyse principles for the available data, and the experimental analyses excluded participants who failed to complete tasks or lacked usable physiological signals as described in the Results.

Survey results: The full survey sample comprised 1,032 individuals; 701 met criteria for the young healthy subsample (mean age 27.99 ± 6.26 years). In multiple linear regression models that adjusted for concomitant drug use, psychedelic use was not significantly associated with the composite schizotypy score in the full sample (n = 1,032, β(SE) = 0.003(0.02), t(1021) = 0.191, p = 0.85) nor in the young healthy sample (n = 701, β(SE) = 0.0004(0.02), t(691) = 0.024, p = 0.98). A subsample with extended exposure data (n = 197) showed no significant relationship between total psychedelic exposure (frequency and recency) and elevated schizotypy. By contrast, stimulant use (cocaine, amphetamines, ephedrine) consistently predicted higher schizotypy scores in young healthy participants (β(SE) = 0.057(0.02), t(691) = 2.8, p = 0.005) and in the full screened sample (β(SE) = 0.065(0.02), t(1021) = 3.87, p = 0.0001). Alcohol use was associated with lower schizotypy in the young healthy sample (β(SE) = -0.04(0.01), t(691) = -2.78, p = 0.005) but not significantly so in the full sample. The model's pseudo-R2 (Cragg–Uhler) was 0.09 (AIC reported). When psychiatric diagnosis was entered into the model for the total sample, diagnosis was significantly associated with schizotypy (β(SE) = 0.075(0.01), t(1028) = 6.82, p < 0.001) while the diagnosis-by-psychedelic-use interaction was non-significant, indicating that psychiatric comorbidity explained a substantial part of the association with schizotypy. Direct group comparisons (unadjusted) showed that psychedelic users (n = 323) had higher mean schizotypy than non-users (n = 709) with a small effect size (Cohen's d = 0.16, 95% CI [0.025, 0.29], one-tailed t = 2.38, p = 0.009); the difference in the young healthy subgroup was marginal (Cohen's d = 0.13, 95% CI [0.035, 0.30], one-tailed t = 1.58, p = 0.06). Analyses of schizotypy subdomains produced results consistent with the primary model: no significant association between psychedelic use and subdomains after adjustment, and stimulants showing the strongest positive associations, particularly with unusual experiences. Experimental arm — BADE: Of 39 participants, 37 completed the BADE task and were included in analyses (22 psychedelic users and 17 age/sex-matched non-users). Mean EII scores were lower (indicating better evidence integration) among psychedelic users (mean = -0.19, SD = 1.04) than non-users (mean = 0.26, SD = 0.91), but this group difference did not reach significance at the group-comparison level (p = 0.17). A multiple regression analysis that used the composite exposure metric (frequency and recency) found that greater psychedelic exposure significantly predicted lower EII; the extracted text reports a significant effect but does not clearly present the full regression coefficient and test statistic in the available excerpt. Experimental arm — fear reversal learning: One subject discontinued and another lacked usable SCR data, yielding 38 participants with pupillometry (n users = 22, n non-users = 16) and 37 with SCR (n users = 21, n non-users = 16). Model fits were stronger for pupillometry-derived expected value dynamics (marginal/conditional R2 = 0.21/0.22) than for SCR-derived fits (marginal/conditional R2 = 0.005/0.22). Between-group t-tests showed that psychedelic users were more influenced by instructed rule reversals: for the ρ-parameter estimated from pupillometry, Cohen's d = 0.82 (95% CI [0.11, 1.52], t = 2.20, p = 0.04), and for the SCR-derived ρ-parameter the between-group difference was larger (Cohen's d = 4.1, 95% CI [2.5, 5.68], t = 11.9, p < 0.001). Linear regression showed that overall exposure to psychedelics (temporal proximity and frequency) was associated with greater instruction sensitivity in pupillometry-derived ρ (β(SE) = 0.04(0.02), t(36) = 2.37, p = 0.02; Cohen's d = 1.05, 95% CI [0.56, 1.54]). These associations remained when controlling for demographics and concomitant drug use in multiple regression models.

Lebedev and colleagues interpret their findings as showing only a weak relationship between psychedelic use and schizotypal traits in the studied population. Although unadjusted group comparisons indicated slightly higher schizotypy among psychedelic users, the effect size was small and the association disappeared after excluding participants with psychiatric diagnoses and after adjustment for concomitant drug use; psychiatric diagnosis and stimulant use were stronger predictors of schizotypy than psychedelic use. The authors note that this pattern aligns with large population studies that found no robust link between psychedelics and psychosis-related outcomes, while contrasting with older studies that did not adjust for confounders such as stimulant use. Contrary to their initial hypothesis, the experimental data suggested that greater psychedelic exposure was associated with improved evidence integration (lower EII) and with increased sensitivity of fear responses to instructed knowledge in a reversal aversive learning paradigm. The researchers propose that these effects—greater readiness to revise beliefs in the face of disconfirmatory evidence and increased top-down influence on aversive learning—could provide a mechanistic rationale for the therapeutic potential of psychedelics in disorders characterised by inflexible cognitive styles (for example depression or certain anxiety-related conditions). They also report that stimulant exposure showed the opposite pattern, being associated with worse evidence integration and higher schizotypy, which accords with prior work. The investigators acknowledge several limitations. Drug-exposure measures were retrospective and subject to recall bias; the study did not collect family psychiatric history; schizotypy was operationalised as a composite score combining different subdomains (a choice made to preserve power), which limits domain-specific inference despite supplementary subdomain analyses. The experimental arm was modest in size, limiting power and the ability to generalise physiological findings. The authors further note that they did not examine co-occurring trait domains such as autism-spectrum or ADHD symptoms in depth in the present analysis. Finally, they emphasise that the absence of a strong link between psychedelic use and psychosis-associated symptoms in healthy young adults does not exclude the possibility that psychedelics could be harmful for individuals at elevated risk for psychotic disorders, and they call for further experimental and longitudinal studies to clarify causal relationships and underlying neural mechanisms.