We recruited 28 healthy participants (12 females) ≥18 years of age from a list of individuals that expressed interest in participating in a psilocybin neuroscience study. Individuals participated in at least one of three sub-projects (sub-project 1 (n = 4), sub-project 2 (n = 10) and sub-project 3 (n = 21)) for a total of 35 psilocybin administrations (i.e., seven participants took part in two sub-projects). The sample included in this study only included those who had completed the Qualitative Experience Report as described below, and received at least 12 mg of psilocybin.

All participants underwent screening for somatic illness, including a medical examination, an ECG, blood screening for somatic disease, and screening for psychiatric disorders using the Mini International Neuropsychiatric Interview, Danish translation version 6.0.0 (Sheehan et al., 1998). Exclusion criteria for all sub-projects were: (Passie et al., 2002) present or previous primary psychiatric disease (DSM axis 1 or WHO ICD-10 diagnostic classifications) or in first-degree relatives; (Carhart-Harris et al., 2018) present or previous neurological condition/disease, significant somatic condition/disease; (Carhart-Harris et al., 2021) intake of drugs suspected to influence test results; (Davis et al., 2020) non-fluent Danish language skills; (Moreno et al., 2006) vision or hearing impairment; (Bogenschutz et al., 2015) previous or present learning disability; (Garcia-Romeu et al., 2019) pregnancy; (Johnson et al., 2017) breastfeeding; (Griffiths et al., 2016) magnetic resonance imaging (MRI) contraindications; (Grob et al., 2011) alcohol or drug abuse; (Ross et al., 2016) allergy to test drugs; (Anderson et al., 2020) significant exposure to radiation within the past year (e.g., medical imaging investigations); (Barrett et al., 2020) intake of QT-prolonging medication or electrocardiogram (ECG) results indicative of heart disease (Griffiths et al., 2011), blood donation less than 3 months before project participation; (MacLean et al., 2011) bodyweight less than 50 kg; (Erritzoe et al., 2018) low plasma ferritin levels (<12 μg/L).

Written informed consent was obtained from all individuals before inclusion. The study was conducted in accordance with the Declaration of Helsinki. The study was approved by the ethics committee for the Capital Region of Copenhagen (journal identifier: H-16028698, amendments: 56,023, 56,967, 57,974, 59,673, 60,437, 62,255, and Danish Medicines Agency (EudraCT identifier: 2016-004000-61, amendments: 2017014166, 2017082837, and 2018023295). The study was also preregistered at ClinicalTrials.gov (identifier: NCT03289949).

The outcome measures used in this study consisted of qualitative and questionnaire data collected on the day of psilocybin intervention and at 3-month follow-up. At baseline, we also collected descriptive information about depressive symptoms, perceived stress and sleep quality with the Major Depression Inventory (MDI) (Bech et al., 2001), Cohen’s Perceived Stress Scale (PSS) (Cohen et al., 1983) and Pittsburgh Sleep Quality Index (PSQI) (Buysse et al., 1989), respectively.

At the end of the psilocybin intervention (∼6 to 8 hours after ingestion of psilocybin), participants completed the self-report revised 30-item Mystical Experience Questionnaire (MEQ) (Barrett et al., 2015). The MEQ is a validated questionnaire comprising 30 items that capture the essential elements of a mystical-type experience relating to a single, discrete event (e.g., one psilocybin administration). The MEQ was derived from the MEQ43 which in turn was developed from the States of Consciousness Questionnaire based on classical descriptive work on mystical experiences and the psychology of religion (Stace, 1960; Pahnke, 1969; Richards, 1975). Items in the MEQ are rated using a 6-point Likert scale from 0 (none, not at all) to 5 (extreme, more than any other time in my life and more than a rating of 4). It has four sub-scales: Mysticality, Positive Mood, Transcendence of Time and Space and Ineffability (e.g., MEQ items: “Experience of unity with ultimate reality”, “Experience of ecstasy”, “Loss of your usual sense of time” and “Sense that the experience cannot be described adequately in words”, respectively). The threshold for a CME is a score of >60% on all sub-scales (Barrett et al., 2015). This threshold was used to divide data into CME and non-CME. MEQ total score was calculated as the mean of all items.

Following completion of the MEQ, participants were asked to draw a mandala of their experience using coloured pencils. Mandala is the Sanskrit word for “circle”, and it is typically interpreted as symbolising a transcendent function in the psyche (Bailer and Leeming, 2020). Practically, participants were given a paper with a circle represented on it and asked to draw their psilocybin experience within the circle using coloured pencils.

In the evening on the day of psilocybin intervention, participants completed a qualitative report describing their experience in free writing using a secured online survey system (https://www.limesurvey.org/). There were no constrains and no minimum or maximum length to the experience report, only that it had to be completed before the participants went to sleep.

Approximately 3 months following the psilocybin sessions, participants completed the Persisting Effects Questionnaire (PEQ) (Griffiths et al., 2006). The PEQ is used to measure long-lasting positive and negative effects attributed directly to the psychedelic experience. The PEQ is a non-validated self-report questionnaire that has been used, among others, by Griffiths et al. (2006), Griffiths et al. (2008), Griffiths et al. (2011), Griffiths et al. (2016) to assess persisting effects of psychedelics as well as long-term positive effects attributable to “naturally occurring” mystical-type experiences (Griffiths et al., 2019). The PEQ comprises 145 questions that describe how the participants perceive changes in their life that they feel are “due to the experiences during your last session [of psilocybin] and your contemplation of those experiences”. Items are rated using a 6-point Likert scale from 0 (none, not at all) to 5 (extreme, more than any other time in my life and more than a rating of 4). It has 12 subscale-themes, namely a positive and a negative change in: Attitudes About Life, Attitudes About Self, Mood Changes, Social Effects, Behavioural Changes, and Spirituality (e.g., Attitudes about life: “You have more joy in your life” and “You have less joy in your life”, Social Effects: “You express more love toward others” and “You express more hatred toward others”). Subscale scores were converted to percentage of maximum score as performed in previous analyses (Griffiths et al., 2011; Griffiths et al., 2016). For the statistical analyses within study, we used the positive subscales only since very few persisting negative effects were rated. See Figure 2.

In sub-project 3 plasma samples were drawn before and approximately 40, 80, 110, 160, 190, and 340 min post-psilocybin administration, and plasma psilocin levels (PPL) were evaluated as described in (Madsen et al., 2019). Peak values were chosen as the MEQ specifies “at any time during that session you experienced the following phenomena”.

Differences at baseline between those who experienced a CME and did not experience a CME (non-CME) on the MDI, PSS, PSQI, age differences, bodyweight, dose administered, and naivety to psychedelic experiences were compared using a random intercept model to take into account that some participants participated twice. Sex differences in MEQ subscale and total score were also estimated using a similar model. Associations between the highest PPL value, i.e., representing peak effects of the drug, for each participant and MEQ total score or PPE_LV were evaluated using univariate linear models. There were no repeated observations in this subgroup.

The association between the MEQ and the PEQ, was investigated with linear latent variable models (LVMs) using the “lava” package for R (Holst and Budtz-Jørgensen, 2013). LVMs can be seen as an extension of linear mixed models for the analysis of multiple repeated measurements. While linear models parameterised with a random slope assume a constant correlation between measurements, LVMs relax this assumption allowing certain measurements to be more correlated than others (i.e., measurements from certain subscales can be more intercorrelated than others). LVMs were evaluated using the “modelsearch” function from the lava package and typically structured as shown in Figure 1 with some variation in covariance as described in section Latent-Variable Model Construction. A latent variable representing the six subscales of the PEQ was constructed and termed Positive Persisting Effects (PPE_LV). The association between the MEQ total and PPE_LV was evaluated within this model. Post hoc analyses of the associations between MEQ subscale scores and PPE_LV were also evaluated, each in a separate model. In order to calculate the potential direct and indirect mediation effects of the co-variates “Age”, “Sex”, “Dose”, “Project 2” and “Project 3”, both direct regressions of these onto PPE_LV and indirect effects of these via MEQ total or subscales were calculated in each model. “Sex-Female” and “Project 1” are reported as the intercept in the analysis of these co-variates. We report loadings (i.e., parameter for the association subscale-latent variable, denoted by “β” in Figure 1) and confidence intervals of each PEQ subscale onto PPE_LV with “life-positivity” arbitrarily chosen as a reference. To control for the small sample size and repeated measures, robust standard error was used (Ozenne et al., 2020). The MEQ total model was evaluated first as a primary hypothesis and the p-value for the relation between the MEQ total score and PPE_LV is reported. Subsequent tests investigating the subscales report the p-value for the same interaction in the model, corrected using the Bonferroni method for a family of four tests (Dunn, 1961).

All evaluations were performed on the untranslated Danish text. Instances of words and quotes presented in this paper were independently translated by two bilingual individuals and recombined by a native English speaker and a third bilingual person.

In order to pre-process the qualitative data for this analysis, words were lemmatised using the python package “lemmy” v 2.1.0 (https://github.com/sorenlind/lemmy). Lemmatisation is the process of converting a word back to its root (i.e., “running” “ran” and “runs” would each be lemmatised to “run”). Danish words are not easily lemmatised as any given word may represent different cases for several different roots. For example, “slappe” (relax) could be lemmatised to “slappe” (relax), “slap” (loose), or “slippe” (drop). In order to control for this, the first lemma for each word was selected and others discarded. Upon random manual checking, this was accurate in all instances.

In this paper we attempt to control for investigator-narrative bias by utilising term-frequency inverse document frequency (tf-idf) to identify distinguishing themes in the text prior to qualitative analysis (Ramos, 2003). The corpus of all reports was split into two documents containing reports from either CME or non-CME. Tf-idf is a statistic that reflects the importance of a term to a specific subset of documents within a wider corpus. Term frequency is the number of times the word appears in each document divided by the total number of words in that document. Inverse document frequency is the log of the total number of documents divided by the number of documents containing the word, and tf-idf is the product of these. Reports were pre-processed by lemmatisation as described above, but stop-words (e.g., “the”, “if”, “and” etc.) were not removed as the tf-idf process accounts for these. Tf-idf scores were computed for each term within CME and non-CME documents. Analyses were performed using the R package “tidytext” (Silge and Robinson, 2016). t f − i d f = i n c i d e n c e   o f   x   i n   d o c u m e n t t o t a l   l e n g t h   o f   d o c u m e n t × l o g 10 ( t o t a l   n u m b e r   o f   d o c u m e n t s n u m b e r   o f   d o c u m e n t s   c o n t a i n i n g   x )

To further investigate the terms with the highest tf-idf scores, we manually searched through all reports and extracted quotes that included each of the five highest tf-idf terms. We first checked that each term was not driven by a single report that used the word repeatedly. We then selected quotes that reflected the most representative thematic usages of these terms.

To further illustrate our qualitative findings, available mandala drawings were located that aligned with the qualitative reports presented. Those interpreted as aligning with the themes of the associated qualitative report were selected, and the five participants were contacted to ensure that perception of depicted themes was accurate. These are presented in the results alongside their respective quotes.

All six positive subscales of the PEQ loaded well onto the latent variable PPE_LV (Estimate range = 0.75:1.22, p < 1.1 × 10⁻¹⁰ for all subscales). All subscale loadings for each model are reported in Supplementary Table S4. No additional covariances were modelled in the main analysis evaluating MEQ total. The models for subscales Ineffability, Transcendence of Time and Space, and Mysticality contained covariance between the PEQ subscales “Positive Attitudes about Life” and “Positive Attitudes about Self” whereas the model investigating the Positive Mood subscale contains a covariance between the MEQ subscale Positive Mood and the PEQ subscale “Increased Spirituality”.

The LVM investigating the association between MEQ total score and PPE_LV showed a statistically significant positive association (ß = 14.8, 95%CI = 8.66:20.96, p = 3 × 10⁻⁵). The subsequent LVMs investigating effects of the MEQ subscales on PPE_LV showed significant positive association with Positive Mood (ß = 14.5, 95%CI = 7.90:21.11, p_corr = 4.1 × 10⁻⁴), as did the LVM evaluating Mysticality (ß = 10.8, 95%CI = 6.14:15.51, p_corr = 2.0 × 10⁻⁴). However, no significant associations were observed for Transcendence of Time and Space (ß = 9.4, 95% CI = −1.72:20.56, p_corr = 0.38) or Ineffability (ß = 11.2, 95% CI = −2.75:25.14, p_corr = 0.45). See Table 2 for a full summary of the LVM results and Figure 3 for a graphical representation. LVM model summaries are reported in Supplementary Table S4. Direct and indirect mediation effects of covariates are reported in Supplementary Table S3. No indirect effects of any covariate via MEQ were significant (all p > 0.1, uncorrected). Scatter plots of MEQ total and subscales by PPE_LV are reported in Supplementary Figure S1.

Of the 35 reports analysed, 21 met the criteria for a CME i.e., scoring at least 60% on all subscales of the MEQ. The words with the highest tf-idf score from CME Qualitative Experience Reports indicating the greatest distinction from the overall corpus, were “Universe”, “dad”, “MR” (Magnetic Resonance), “beautiful”, “simultaneous”, “infinite”, “purple”, “in relation to”, “ray”, “happy”, and “brother” (See Figure 4). The highest tf-idf scores from non-CME reports were “gloomy”, “cycle”, “evil”, “cold” and “need”. To expand on these findings, these words were identified in text and a selection of quotes are presented below. Notably, as there are only two documents within the corpus (CME and non-CME), any word which appears in both documents has a tf-idf score of zero. Thus, tf-idf scores distinguish words that only appear in one document set (CME or non-CME).

By manually searching through the document body, we confirmed that high tf-idf scores were not driven by single individuals utilising the identified word multiple times. The word with the highest tf-idf score from the CME reports was “Universe”. Viewed in context, these references to the Universe appear to relate to feelings of unity and connectedness, analogous to MEQ item 14 “Freedom from the limitations of your personal self and feeling a unity or bond with what was felt to be greater than your personal self”, as well as a wonderment around the complexity of conscious experience.

One report juxtaposes the profound noetic quality of the experience with the tasks that participants were asked to perform in the scanner environment. This aligns with item 9 of the MEQ, i.e., “Certainty of encounter with ultimate reality (in the sense of being able to “know” and “see” what is really real at some point during your experience).”

The word that next-best distinguished the CME was “dad”. The reports describe feelings of gratitude, love and respect for the participants’ fathers. There are no items of the MEQ that reflect this theme directly. Notably, references to the participants mothers appeared in both CME and non-CME reports and thus had a tf-idf score of zero.

In two cases, the participants describe a desire to share this experience with their fathers specifically.

The word with the third highest tf-idf score was “MR” referring to the MRI scanner. In several cases, participants found the intense noises of the scanner influenced their experience, in some cases inducing synaesthesia-like experiences. More broadly, this represents the profound impact of the environment on the experience.

The MR environment was the setting for 21 out of the 35 experiences collected (sub-project 3). Some felt that it played a positive role in the experience, as described in item 6 of the MEQ “Experience of oneness or unity with objects and/or persons perceived in your surroundings.”

The word with the fourth highest score was “beautiful”. This particular excerpt below combines the themes of universality, family and beauty that appear repeatedly in the CME reports. This report also aligns with item 8 in the MEQ “Feelings of tenderness and gentleness”.

Several reports refer to the beauty of the closed eye visual effects and the participants interactions with them. Perceptual effects are not measured by the MEQ.

Some of the reports of beauty refer to participants experiences of reflecting on the nature of their consciousness and emotions.

Finally, many participants who had CMEs describe the beauty of the world and Universe around them, both natural and interpersonal. This is reflected in MEQ item 27 “sense of awe or awesomeness”

The fifth highest tf-idf word was “simultaneous”. These quotes largely refer to a duality of experience (e.g., perceived disconnection of body and mind) wherein participants report feeling several, sometimes conflicting, emotions at the same time or reflect on their experience as it is ongoing. No specific items in the MEQ match these themes.

Upon inspection, many of the words that had high tf-idf values from non-CME reports were driven primarily by one or two reports that used the words repeatedly. Thus, we did not feel that they provided a sound reflection of the themes within the non-CME. The only exception was the word “gloomy” which was used six times in non-CME reports, never in CME reports, and was reported by several individuals. In context this referred largely to interpretation of music.