Validation of the revised Mystical Experience Questionnaire in experimental sessions with psilocybin

Reports of mystical-type experiences describe common dimensions such as a felt unity with all that exists, sacredness, noetic quality (a sense of encountering fundamental truth), intense positive mood, transcendence of time and space, and ineffability. Earlier versions of the Mystical Experience Questionnaire (MEQ), most notably the 43-item MEQ43, were developed to quantify these dimensions in single, discrete experiences occasioned by classic hallucinogens and have been used in prospective studies of psilocybin; however, psychometric investigation of the MEQ43 has been limited. A 30-item revised instrument (MEQ30) was previously derived from retrospective online reports of psilocybin-related experiences and modelled into four factors (mystical, positive mood, transcendence of time and space, and ineffability), but that work relied on uncontrolled, retrospective data collected on average eight years after the original experiences. This paper aims to validate the MEQ30 in prospectively collected experimental data. Barrett and colleagues pooled data from five laboratory studies in which participants (n=184) received a moderate to high oral dose of psilocybin (at least 20 mg/70 kg) and used confirmatory factor analysis and structural equation modelling to test the MEQ30 factor structure, examine relationships between MEQ30 latent scores and persisting outcomes attributed to the session while controlling for overall drug intensity, and adapt a dichotomous "complete mystical experience" scoring method from the MEQ43 to the MEQ30. The study therefore evaluates internal, external and convergent validity of the MEQ30 in controlled-dose psilocybin sessions.

Data were pooled across five Johns Hopkins laboratory experiments in which each participant's single highest-dose psilocybin session was included. The five studies contributed the following observations: Study 1 (double-blind pharmacology comparison) contributed 30 psilocybin-session completions, Study 2 (dose-sequence study) contributed 18 high-dose sessions (30 mg/70 kg), Study 3 (open-label smoking cessation pilot) contributed 15 highest-dose sessions, Study 4 (spiritual practices trial, in preparation) contributed 71 high-dose sessions (mostly 30 mg/70 kg), and Study 5 (cancer-related anxiety/depression, in preparation) contributed 50 moderately high-dose sessions. One observation per participant was used and all studies had institutional review board approval and informed consent. Measures collected at the end of each session included the 100-item States of Consciousness Questionnaire (SOCQ), which contains the 43 MEQ43 items; the MEQ30 scores were derived from the relevant 30 items within that battery. Items were rated on ordered categorical scales (MEQ items used a 6-point scale from 0="none" to 5="extreme"). Persisting outcomes were assessed approximately three to eight weeks after sessions using items that rated personal meaningfulness (8-point ordinal), spiritual significance (6-point), change in well-being/life satisfaction (7-point range −3 to +3), and positive behaviour change (6-point). The Hallucinogen Rating Scale (HRS) intensity subscale and the single HRS intensity item (item 99) were used as measures of overall subjective drug intensity to test whether persisting outcomes were better explained by mystical-type experience than by global drug intensity. Analyses comprised confirmatory factor analyses (CFAs) comparing three competing structural models for the MEQ30 items: the previously proposed four-factor MEQ30 model, and two alternate three-factor models denoted "Hood" and "Stace." An extended second-order model (MEQ30-total) was also tested, with the four first-order factors loading onto a single higher-order latent variable. External and convergent validity were evaluated using structural equation models (SEMs) that regressed each of the four persisting-effect items on latent MEQ30 factor scores (or on the MEQ30-total latent variable) while simultaneously controlling for HRS-derived intensity (either the factor or item 99). Model fitting used maximum likelihood estimation in the lavaan package in R; fit was assessed using comparative fit index (CFI) and the standardized root mean square residual (SRMR), with the Satorra-Bentler scaled chi-square reported due to sample size. RMSEA was not considered because of known small-sample properties. Finally, the authors adapted the MEQ43 dichotomous "complete mystical experience" rule (score ≥60% of maximum on each subscale, discarding the lower of internal or external unity for MEQ43) to the MEQ30 by classifying observations with ≥60% on each of the four MEQ30 subscales as "complete." Classification performance of MEQ30-derived complete scores was evaluated against MEQ43-derived classifications using hit rate, false alarm rate, proportion correct and sensitivity index (d').

Internal validity: Confirmatory factor analyses replicated prior findings that the four-factor MEQ30 model fits MEQ30 item data better than the two three-factor alternatives. The four-factor model met conventional fit thresholds (CFI >0.90, SRMR <0.09) whereas the Stace and Hood three-factor models did not (CFI <0.90). Cronbach's alpha indicated excellent internal consistency for each MEQ30 factor: mystical α=0.97, positive mood α=0.92, transcendence of time/space α=0.86, and ineffability α=0.90. Standardised factor loadings were generally high and similar to the original MEQ30 development sample, with the exception of one item (#11) whose loading was lower in the present sample (0.46 vs 0.70 previously); the authors report this value as acceptable and made no questionnaire modifications. A second-order MEQ30-total model, in which the four factors load onto a single higher-order latent variable, also demonstrated acceptable fit, supporting use of an overall MEQ30-total score. External validity and convergent prediction: Structural equation models that regressed persisting-effect ratings on MEQ30 latent variables and HRS-derived intensity demonstrated acceptable fit (structural MEQ30 model CFI=0.905, SRMR=0.058; structural MEQ30-total model CFI=0.900, SRMR=0.060). The single HRS intensity item did not predict the persisting outcomes in either model, suggesting overall subjective intensity did not account for the longer-term effects. Latent scores on the mystical factor positively predicted retrospectively rated spiritual significance and also predicted positive changes in behaviour attributed to the session; the positive mood factor predicted increased personal well-being attributed to the session. The extracted text does not clearly report the exact regression coefficients for these effects (values shown as B=0.. in the extraction), so precise effect sizes and standard errors cannot be stated here. No first-order latent factor significantly predicted rated meaningfulness of the experience. By contrast, the MEQ30-total latent variable significantly and positively predicted all four persisting-effect outcome variables in the models reported. Scoring of "complete mystical experiences": Applying a ≥60% per-factor cutoff on the MEQ30 yielded a classification that compared well with the MEQ43-based classification. Using the MEQ43 classification as the reference standard, MEQ30-derived complete mystical scoring produced a hit rate of 0.95, a false alarm rate of 0.14, an overall proportion correct of 90.2%, and a sensitivity index d' = 3.464. The 14 false-alarm cases (MEQ30 positive but MEQ43 negative) were each attributable to a single MEQ43 subscale falling just below the 60% threshold; the positive mood MEQ43 subscale was the most common such sub-threshold contributor (n=5). The mean relative sub-threshold score for the discrepant MEQ43 scale was 0.51 (SD=0.046, range 0.43–0.58), indicating the misses were close to the cutoff. The authors interpret these results as consistent with somewhat greater sensitivity of the MEQ30 classification, possibly owing to removal of weaker items from the MEQ43 during MEQ30 construction.

Barrett and colleagues interpret their findings as supporting both the internal and external validity of the MEQ30 in prospectively collected laboratory psilocybin sessions. Factor analyses confirmed a stable four-factor structure and high internal consistency for each MEQ30 factor, and a second-order MEQ30-total score is psychometrically defensible. Structural models showed that MEQ30 latent scores—particularly the mystical and positive mood factors, and most robustly the MEQ30-total score—predict persisting subjective effects attributed to psilocybin sessions even when controlling for rated drug intensity, indicating that mystical-type dimensions convey information about longer-term outcomes beyond simple drug strength. The authors situate these results within the theoretical literature that conceptualises mystical experience as a unitary latent phenomenon expressed across multiple dimensions; finding that the MEQ30-total score predicted all persisting outcomes is taken as consistent with that theory. They also note that while some individual first-order factors showed selective predictive relations (mystical → spiritual significance and positive behaviour change; positive mood → well-being), none predicted the rated meaningfulness item, and the MEQ30-total generally provided the best predictive validity. Several practical implications are discussed. The MEQ30 appears resilient to whether its items are presented within the full SOCQ or as stand-alone items; a stand-alone MEQ30 is offered for future use and is recommended over the MEQ43 and SOCQ for experimental applications. The adapted dichotomous rule for identifying a "complete mystical experience" on the MEQ30 performs comparably to the MEQ43-based rule, albeit with a modestly higher false-alarm rate that the authors attribute to greater sensitivity rather than reduced specificity. The authors acknowledge that biological mechanisms underlying mystical experiences remain unidentified and that the MEQ30, while validated here in psilocybin sessions, should be further validated in studies of mystical experiences produced by other means. They also note that some extracted numerical details (e.g. specific regression coefficients) are not fully reported in the provided text, which limits precise quantification of effect sizes within this summary.

The study demonstrates that the MEQ30 is psychometrically sound when applied to prospectively obtained experimental data from controlled-dose psilocybin sessions. MEQ30-total scores have greater predictive validity for persisting personal and spiritual outcomes than a simple measure of overall drug intensity, and the MEQ30 can be used to classify "complete mystical experiences" in a manner comparable to the MEQ43. Barrett and colleagues recommend the MEQ30, and specifically the MEQ30-total score, for future investigations of individual mystical episodes, and suggest further validation across other induction methods.