Replication and extension of a model predicting response to psilocybin

Psychedelic compounds such as psilocybin have shown promise as treatments for conditions including depression, death-related anxiety, substance use and obsessive–compulsive disorder, and as enhancers of well‑being in healthy volunteers. However, even in controlled settings responses vary: some participants report highly positive, meaningful or mystical-type experiences while others have challenging or adverse sessions. Earlier research has identified dose, certain personality traits (for example absorption and openness), and contextual factors of the setting as contributors to outcome, but much variation in individual responses remains unexplained. A prior crowdsourced study found that transient mental states at ingestion—specifically a state of surrender versus a state of preoccupation—explained substantial variance in mystical and adverse psilocybin experiences, beyond trait predictors. Russ and colleagues set out to replicate and extend that prior model in a new, independent sample. The study aimed to (1) conceptually replicate regression models predicting mystical and adverse response using different dependent measures, (2) test additional hypothesised predictors (including motivational, prior-state and setting variables), and (3) examine whether mystical-type experiences relate to longer‑term positive change. Data were retrospective and crowdsourced from people who had self‑administered psilocybin within the previous 12 months rather than being collected under controlled drug administration.

Participants were 183 US residents recruited via Amazon Mechanical Turk who reported ingesting psilocybin in the prior 12 months. The screened sample comprised 97 males (53.0%) and 85 females (46.4%), mean age 31.9 years (SD = 9.43, range 18–70). Education, prior psychedelic experience, age at first psychedelic exposure, and time since the indexed psilocybin session (1–12 months) were recorded; detailed information on setting, dose and co‑ingested substances was reported in an appendix but not fully reproduced in the extracted text. Data were collected with an extensive online survey (286 items) that combined open‑ended memory prompts with standardised scales. Sections covered background/traits, prior life situation, proximal state at ingestion, in‑session experience, and current state/long‑term change. Key psychometric measures included selected items from the Tellegen Absorption Scale (18 items, α = .91 in this sample), newly developed trait scales (Barriers, Deservingness), a Confusion state measure (α = .91), a 14‑item motivations inventory (spiritual, norm, pleasure, betterment factors), and novel set measures: State of Surrender (10 items, α = .92) and State of Preoccupation (4 items, α = .757). In‑session experiences were assessed via 5D‑ASC subscales: an Oceanic Boundlessness‑based mystical index (Mystical‑OBN; α = .93) and a Dread of Ego Dissolution adverse index (Adverse‑DED; α = .91). Ego dissolution was measured with the Ego Dissolution Inventory (EDI; ego dissolution α = .87). Long‑term change used composite positive and negative change scales (α = .929 and α = .883 respectively) and a flourishing measure. Recruitment sought 230 participants; 624 individuals entered screening, 318 passed screening and consented, and after validation checks and completion requirements 183 remained. Compensation for completing the survey was USD 2.00. Analyses used hierarchical linear regression models with blockwise entry of predictor clusters (demographics, trait variables, prior state, motivations, setting, proximal state/set). Adverse‑DED scores were log‑transformed to reduce skew; Mystical‑OBN showed only slight skew and was analysed without transformation. Outliers were examined and retained where not removable. Subgroup regressions compared sex and levels of spiritual motivation, and long‑term outcomes were compared by presence versus absence of a “complete” mystical experience using independent t tests, with a >70% threshold on the OBN scale used to define a complete experience.

Sample characteristics: 183 respondents, mean age 31.9 (SD 9.43), 53.0% male. Time since the indexed psilocybin session ranged across 1–12 months. Internal reliabilities for key measures were strong (for example surrender α = .92; Mystical‑OBN α = .93; Adverse‑DED α = .91). Replication of the original model for mystical experience: Models from the prior sample (S1) generalised to the current sample (S2). The best predictors from S1 that explained 65.7% of variance in mystical‑MEQ in S1 explained 52.9% of variance in mystical‑OBN in S2 (R2 = .529, F(3,179) = 40.98, p < .001). In both samples the state of surrender was the strongest predictor (β = .607 in S1; β = .477 in S2 reported earlier in the text; later S2 estimates are β = .525 or .453 depending on model specification), followed by the trait absorption (β ≈ .308 in S1; ≈ .359 or .247 in different S2 models). Exposure to spoken words during the session predicted mystical experience negatively in both samples (words β ≈ -.140 in S2). Sex‑specific models explained more variance for males than females in S2 (males R2 = .593; females R2 = .423 in one reported comparison), with surrender remaining the dominant predictor for both sexes. Revised, expanded model for mystical‑OBN in S2: Blockwise regression showed that demographics added little (block 1 R2 = .010), traits increased explained variance to R2 ≈ .312, prior state factors to R2 ≈ .333, motivations to R2 ≈ .493, and proximal set factors (surrender, apprehension, preoccupation) raised R2 to ≈ .63. A parsimonious backward‑eliminated model with five predictors accounted for 61.0% of variance in mystical‑OBN (R2 = .610, F(5,177) = 55.372, p < .001). Significant predictors in that model were state of surrender (β = .453, p < .001), spiritual motivations (β = .289, p < .001), absorption (β = .247, p < .001), deservingness (β = -.162, p < .01), and apprehension (β = -.117, p < .05). The authors note a suppressor effect involving deservingness and surrender, such that deservingness becomes an inverse predictor in the presence of surrender. Prediction of ego dissolution: Using a similar variable-selection approach, the final model explained R2 = .487 of variance in the EDI (F(4,178) = 42.29, p < .001). Significant predictors were surrender (β = .543, p < .001), deservingness (β = -.177, p < .001), absorption (β = .170, p < .05), and spiritual motivations (β = .155, p < .05). Adverse experience results: The S1 model that had explained 56.2% of variance in adverse‑SES explained less variance in S2 adverse‑DED (R2 ≈ .315 for initial blockwise model; best‑fitting model R2 = .344, F(4,178) = 23.305, p < .001). Across samples, preoccupation was the strongest positive predictor (β ≈ .440 in S1; β = .359 in S2). In S2 the best‑fitting adverse model included preoccupation (β = .359, p < .001), apprehension (β = .159, p = .013), confusion (β = .220, p = .003), and openness (β = -.129, p = .040), together explaining about 34.4% of the variance. Sex‑stratified analyses produced similar explanatory power for males (R2 ≈ .348) and females (R2 ≈ .392), with some sex differences in which additional variables contributed significantly (for example age inversely predicted adverse experience for females in one analysis). Long‑term outcomes and mystical experience: Independent t tests compared participants who met a threshold for a complete mystical experience (>70% on the OBN subscale) with those who did not. Those with a complete mystical experience reported higher positive change (M = 4.00/5) than those without (M = 3.39; t(178) = 4.909, p < .001). Negative change scores were lower among the complete mystical group (M = 2.22) than the non‑complete group (M = 2.74; t(178) = -3.569, p < .01). Participants with a complete mystical experience rated the session as more important (M = 75.66) than others (M = 42.75; t(178) = 9.261, p < .001), and flourishing scores were higher for the complete mystical group (M = 4.78) versus others (M = 4.39; t(178) = 2.170, p < .05).

Russ and colleagues interpret the findings as a robust replication of their earlier model: transient mental states at the time of psilocybin ingestion—principally a state of surrender and, conversely, a state of preoccupation—were consistent and strong predictors of mystical‑type and adverse experiences, respectively, across two independent samples and using different outcome measures. The authors highlight that surrender and preoccupation demonstrated good internal consistency and retained the largest beta weights in both samples, supporting their conceptual relevance for predicting session quality. They also note that lower exposure to spoken words during the session predicted greater mystical experience, aligning with therapeutic practices that control auditory stimuli in clinical settings. The discussion places these results in the context of earlier work linking mystical experience to favourable long‑term outcomes. Data from this study supported that association: meeting the authors' threshold for a complete mystical experience was associated with greater reported positive change, lower negative change, higher subjective importance of the session, and higher flourishing. The authors suggest that state factors such as surrender, in combination with certain dispositional traits (for example absorption), contribute more substantially to mystical‑type outcomes than motivational factors alone, although spiritual motivation did add predictive power in some models and correlated with absorption and surrender. Limitations acknowledged include the retrospective, self‑report and crowdsourced study design which precludes causal inferences and reliable dose information; potential sample bias favouring individuals with positive experiences due to survey length; and specificity to psilocybin limiting generalisability to other psychedelics. The authors recommend future prospective research to obtain reliable dosing and contextual data, formal validation of the surrender and preoccupation scales, exploration of physiological correlates of surrender, and investigation of whether surrender states can be induced or supported in clinical settings to improve therapeutic outcomes. They conclude that despite limitations, the replication across two samples lends credibility to the importance of assessing and attending to psychological set (notably surrender and preoccupation) when seeking to optimise the therapeutic potential of psilocybin.