Longitudinal neural and psychological changes following psilocybin under compassion imagery priming

Psilocybin has been associated with lasting changes in wellbeing, personality, cognitive flexibility and self-related beliefs, but the size and quality of these effects depend on the acute experience and on context or “set and setting”. The authors note that compassion-based practices engage affiliative, care-related systems in the brain and may be especially relevant as preparatory states for psychedelic experiences. They frame this within the REBUS model, which proposes that psychedelics temporarily loosen rigid high-level beliefs and increase psychological flexibility, with the possibility that compassionate priming may steer these changes towards social safeness and self-compassion. Pallavicini and colleagues set out to test whether psilocybin effects on absorption, decentering and self-compassion would vary when psilocybin was preceded by compassion imagery rather than attention to breathing. They also aimed to examine whether these psychological effects were accompanied by changes in large-scale functional connectivity measured with fMRI. The study was designed as a naturalistic field experiment, with a high-dose and low-dose psilocybin comparison and a compassion imagery versus attention-to-breathing priming comparison, making the work exploratory and hypothesis-generating rather than confirmatory.

The study used a naturalistic, longitudinal design in which participants were recruited through word of mouth and social media and screened by mental health professionals. A total of 107 participants were enrolled, 105 completed assessments up to the intake day, 100 completed the post-intake session, and 93 provided follow-up data. Participants were adults aged 21 to 65 years who had already arranged to take psilocybin in a group ceremony and agreed to be blinded to dose, undergo contemplative priming, questionnaires, EEG, fMRI, blood and saliva sampling, and wear activity trackers. Exclusion criteria included major psychotic, bipolar, depressive and several anxiety or eating disorders, recent substance dependence, neurological disorders, pregnancy, psychiatric medication, high state anxiety, and prior severe adverse psychedelic experiences. Participants were not prospectively randomised in a fully balanced way; they selected ceremony dates based on availability, and groups were formed around those dates. The researchers therefore compared four groups: high dose with compassion imagery priming (N=30), high dose with attention-to-breathing priming (N=20), low dose with compassion imagery priming (N=25), and low dose with attention-to-breathing priming (N=25). High dose was 3 g of dried Psilocybe mushrooms and low dose was 1 g of the same material, administered in fruit juice. The ceremony setting was naturalistic and could be indoors or outdoors; facilitators provided music, aromatics and support as needed. The priming lasted about 20 minutes immediately before ingestion. Psychological outcomes were assessed with online questionnaires one week before dosing, on the intake day, and at 7, 30, 90 and 180 days afterwards. Measures included Tellegen Absorption Scale, Experiences Questionnaire decentering, Compassion Engagement and Action Scales, anxiety, personality and other variables; some subscales that were unbalanced at baseline were excluded from analysis. For neuroimaging, participants completed 5-minute resting-state and 5-minute compassionate imagery fMRI scans before dosing and again 2 to 4 days after dosing. fMRI preprocessing included smoothing, band-pass filtering, normalisation, nuisance regression and motion scrubbing; analyses were based on 90 Automated Anatomical Labeling regions and on eight canonical resting-state networks. The authors used random forest classifiers with permutation testing to distinguish resting state from compassionate imagery, and to test whether pre-post changes in compassionate imagery connectivity distinguished compassion imagery from attention-to-breathing within each dose group. The main brain-behaviour correlations were limited to functional connectivity changes in the high-dose condition, because only that comparison showed significant classification. Questionnaire analyses used t-tests, within-group longitudinal comparisons and false discovery rate correction. Correlations between brain and questionnaire change scores were considered significant at p<0.05 with at least medium effect size, and were restricted to measures that had shown longitudinal effects.

The acute subjective effects of psilocybin were broadly similar across conditions. In the high-dose versus low-dose comparison, only a few uncorrected differences emerged, including transcendence, audiovisual synaesthesia, anxiety and complex imagery, with small-to-moderate effect sizes. No significant differences were found between the high-dose compassion imagery and high-dose attention-to-breathing groups, suggesting that the priming condition did not markedly alter acute subjective reports. The authors note, however, that these data do not directly establish successful blinding because participants’ beliefs about dose were not measured. Across follow-up questionnaires, the most robust longitudinal finding was a sustained increase in absorption. In the high-dose group, absorption rose significantly from baseline at 30, 90 and 180 days after dosing, with effect sizes around d=0.48 to 0.57 after correction. The low-dose group showed only non-significant trends in the same direction. In the high-dose priming comparison, absorption was significantly higher in the compassion imagery group than in the attention-to-breathing group at 30, 90 and 180 days, with effect sizes around d=0.72 to 0.82. Within-group analyses also showed clear increases in the high-dose compassion imagery group, whereas the high-dose attention-to-breathing group did not show significant longitudinal change. Decentering also increased over time. Both high-dose and low-dose groups showed significant false discovery rate-corrected increases on the Experiences Questionnaire decentering subscale at multiple follow-up points, with effect sizes roughly between d=0.48 and d=0.83. The high-dose compassion imagery group showed a trend towards greater decentering than baseline by 180 days, whereas the high-dose attention-to-breathing group did not. Self-compassion measures showed more mixed results. For the engagement component, the high-dose group increased significantly at 90 and 180 days, and the high-dose compassion imagery subgroup showed a significant increase at 180 days; the corresponding attention-to-breathing subgroup showed only a non-significant tendency. For the action component, both dose groups tended to increase over time, but only the low-dose group at 180 days survived correction. These findings were described as within-group temporal changes rather than reliable between-group differences. The fMRI classifiers successfully distinguished resting state from compassionate imagery at baseline, with an AUC of 0.716 ± 0.008 and p<0.001. This distinction was mainly driven by connectivity among the left and right dorsal attention networks and their connections with executive control, default mode and auditory networks. When the authors examined changes in compassionate imagery network connectivity before versus after psilocybin, the high-dose group showed significant classification between compassion imagery and attention-to-breathing priming, with an AUC of 0.696 ± 0.008 and p<0.05. In that contrast, stronger coupling involved the two dorsal attention networks and connections involving the auditory and sensorimotor networks. The same analysis in the low-dose group was not significant and performed at chance level. Brain-behaviour correlations were then examined only in the high-dose groups. Absorption showed no significant correlations with connectivity change. By contrast, decentering and self-compassion action were positively correlated with visual cortex–sensorimotor connectivity, suggesting that larger increases in these psychological measures were associated with stronger integration between those networks. Self-compassion engagement showed predominantly negative correlations, including reduced coupling between dorsal attention and other networks and a negative association between executive control and sensorimotor connectivity. These associations were reported mainly in the high-dose compassion imagery group, with few or none in the high-dose attention-to-breathing group.

The authors interpret the findings as suggesting that psilocybin, particularly when combined with compassion imagery priming, is associated with sustained increases in absorption and with trends towards enhanced decentering and self-compassion. They argue that absorption may reflect a greater capacity to become deeply immersed in inner experience, decentering may indicate a more observational stance towards thoughts and emotions, and self-compassion may reflect a shift towards caring and affiliative self-related processing. They present these outcomes as broadly consistent with the REBUS framework, which predicts reduced rigidity of high-level beliefs and greater psychological flexibility under psychedelics. Relative to earlier research, the authors position their work as extending separate literatures on psychedelics and compassion-based practices by testing them together in a single protocol. They suggest that compassion imagery may have amplified some of the longer-term effects of high-dose psilocybin, especially for absorption, and that the fMRI findings indicate distinct network-level signatures for compassionate imagery and for changes induced after psilocybin. They interpret the involvement of dorsal attention, executive control, auditory and sensorimotor networks as compatible with attentional control, embodiment and integration of internal experience during compassion imagery under psilocybin. The authors acknowledge several important limitations. The study was naturalistic, so dose was not standardised chemically and participants were not fully randomised or prospectively balanced across groups. Blinding effectiveness was unknown because participants’ beliefs were not measured. Baseline differences existed for some measures, some acute effects may have been hard to judge because there was no reference point, and attrition across follow-up may have introduced bias. The brief 20-minute compassion imagery training may also have been insufficient. In addition, the fMRI analyses were performed at the network level, limiting spatial specificity, and the study was not preregistered, increasing the possibility that some findings are exploratory or selective. For implications, the authors suggest that compassion-focused imagery may be worth studying further as a preparatory or post-experience practice in psychedelic contexts. They propose that Compassion-Focused Therapy and Compassionate Mind Training could be explored as ways to support emotional safeness, integration and longer-term therapeutic change, but they present this as a future research direction rather than a settled clinical recommendation.