Silence is golden: Documenting the speech production of participants and support providers in psilocybin administration sessions for the treatment of post-traumatic stress disorder

There is increasing interest in psilocybin for psychiatric disorders because clinical trials have reported rapid and sometimes sustained symptom improvements after a single high-dose administration. In the Compass trial context, participants receive structured preparation, continuous supervision during dosing, and follow-up sessions, but the nature of the support provider’s role during administration has remained debated. The paper notes that this model is often misunderstood as psychotherapy in the conventional sense, even though high-dose classic psychedelics can produce intense inwardly focused states in which active dialogue may be difficult or unnecessary. The authors also contrast psilocybin trials with MDMA-assisted psychotherapy, where verbal therapeutic interaction is more central, arguing that the two approaches are sometimes conflated inappropriately. The study aimed to clarify what actually happens during psilocybin administration sessions in Compass clinical trials by quantifying speech and examining participant experiences of support. Dougherty and colleagues sought to measure words-per-minute (WPM) for participants and support providers across preparation, administration, and follow-up sessions, and to analyse post-session interviews to understand how support was perceived. The broader purpose was to reduce confusion about the Compass monitoring-and-support model and to provide a more concrete description that could aid replication and interpretation of future studies.

The researchers analysed data from a 12-week, Phase II, multicentre, non-randomised open-label clinical trial of a single 25 mg dose of COMP360 psilocybin for post-traumatic stress disorder. The trial took place in the United Kingdom and the United States, received ethics approval in both countries, and was conducted under Good Clinical Practice and Declaration of Helsinki principles. Twenty-two English-speaking adults with PTSD were enrolled after screening 39 individuals, and none withdrew. Participants were outpatients with at least moderate PTSD severity on CAPS-5 at baseline. The extracted text lists several exclusion criteria, including complex PTSD, psychotic disorder, bipolar disorder, personality disorder, obsessive compulsive disorder, recent alcohol or substance use disorder, recent major depressive disorder, recent trauma exposure, significant childhood abuse, and significant suicidal risk. Participants attended an initial screening visit and then a run-in period with one to six further screening-period visits. They met a primary and assisting support provider three times before dosing: twice during screening and once the day before administration. These were the preparation sessions. During the administration session, all participants received 25 mg COMP360 psilocybin in a dimly lit room, wore eyeshades, listened to a standardised music playlist, and were continuously attended by two trained support providers for approximately 6–8 hours. The follow-up phase included three support-provider sessions after dosing: the day after administration and at weeks 1 and 2. All sessions were audio recorded. For the quantitative analysis, recordings were professionally transcribed and checked with automated speech recognition when needed. The researchers calculated WPM for each speaker in each session by dividing total words spoken by session duration. They also created 5-minute time bins to estimate speech across the session and calculated silence as the proportion of session time without speech. Because the data were nested and repeated across sessions and participants, they used a linear mixed-effects model with session and speaker as fixed effects, their interaction, and random intercepts for participants nested within support providers. Planned contrasts were used to examine differences within and between speakers across visits. For exploratory correlations, the researchers used 5D-ASC scores completed after the administration session to represent acute psychedelic experience. They examined associations between 5D-ASC dimensions and administration-session WPM using Spearman’s rank correlation. No p-values were calculated for these correlations because the analysis was post hoc and the sample size was small. The qualitative component used Braun and Clarke’s six-phase thematic analysis framework. It focused on transcripts from post-administration interviews conducted the day after dosing, selected to capture immediate reflections on the administration session. Coding was iterative and reflexive, led by NLM, with discussion and refinement involving EW and GMG. The authors state that quotations were used to ground the analysis rather than as formal evidentiary proof.

The dataset included audio recordings from 64 preparation sessions, 18 administration sessions, and 66 follow-up sessions; two preparation and four administration recordings were missing. Descriptively, speech was markedly reduced during administration compared with the other session types. On average, 78% of the administration session was spent in silence, whereas silence accounted for about 25% to 30% of the time in preparation and follow-up sessions. The authors also note that speech during administration was less variable across support providers than across participants. The mixed-effects model showed significant effects of session type, speaker role, and their interaction. The session-by-speaker interaction improved model fit substantially (likelihood ratio test χ² = 105.85, p < 0.001), indicating that the effect of visit on WPM differed depending on whether the speaker was a participant or a support provider. Within-session contrasts suggested that speech rates were balanced during preparation, similar between speakers during administration, and much higher for participants than support providers during follow-up. Between-session contrasts showed that both participants and support providers spoke significantly less during administration than during any other session type. Support providers also spoke significantly less in follow-up than in preparation. The time-series analysis suggested that most speech that did occur during administration was concentrated at the beginning and end of the session. The authors interpret this as consistent with brief opening orientation and later reorientation, comfort, and post-session guidance, rather than sustained dialogue. In the exploratory correlations, lower speech rates during administration tended to align with more intense psychedelic experiences, especially for Oceanic Boundlessness and Anxious Ego Dissolution. The reported correlations for participant and support-provider WPM were negative for these dimensions (for example, OB at -0.311 and -0.325, and AED at -0.311 and -0.261), while other 5D-ASC dimensions showed weaker associations. Because no p-values were calculated, these findings are presented as exploratory only. The thematic analysis identified three themes. First, support was described as minimally enacted but still experientially important: participants often paid little conscious attention to support providers, yet valued their presence as stabilising and reassuring. Second, support was seen as promoting autonomy through a non-directive approach that allowed participants to remain with their own inner experience without interruption, reinterpretation, or intrusive guidance. Third, brief reassurance and validation were experienced as the main forms of active support during intense moments, sometimes through simple verbal orientation or consensual hand-holding. Selected participant accounts indicated that being reminded of safety and location, or having distress acknowledged, was helpful.

The authors argue that the results demonstrate that psilocybin administration sessions in this Compass trial were dominated by silence: the participants and support providers spent most of the dosing period without speech, and speech was concentrated at the start and end of the session. They present this as evidence that the support model is distinct from conventional trauma-focused psychotherapies, which rely on structured verbal engagement, interpretation, and cognitive work. In their view, the data support the idea that support during psilocybin administration is not primarily a psychotherapeutic dialogue but a background framework that enables safety and tolerability. The qualitative findings are used to deepen this interpretation. The authors emphasise that the moments of speech were remembered less for their content than for their timing and tone, and that participants did not describe support providers as actively directing reflection or providing interpretations. Instead, participants valued presence, availability, validation, and reassurance. The discussion suggests that in the altered state induced by psilocybin, extended conversation may be unwanted or disruptive for many people, while non-directive support may help preserve inward focus and autonomy. The paper positions these findings against earlier debates about whether psychedelic trials involve psychotherapy or something closer to structured support. The authors contrast their findings with trauma-focused therapies and note that the Compass model differs from MDMA-assisted psychotherapy, where active psychotherapeutic engagement is more central. They argue that their data help clarify misconceptions created by the broad term “psychedelic-assisted psychotherapy” when applied to psilocybin. The authors acknowledge several limitations. The study was open-label and modest in size, and WPM is described as a reductive measure compared with more sophisticated language-analysis methods. The correlational analyses were exploratory and underpowered, so they are not treated as confirmatory. They also note that the report focuses only on administration sessions, even though broader support interactions may matter. For future work, they suggest larger samples and more advanced approaches such as natural language processing and large-language-model-assisted analysis to examine the communicative features of support more deeply. They also note that speech analytics may eventually help examine whether the support model mediates clinical response.

The authors conclude that the Compass monitoring-and-support model during psilocybin administration is unlike conventional psychotherapy. In their account, the support provider functions mainly as a steady, attentive, non-directive presence that maintains safety, autonomy, and reassurance while the participant undergoes a self-directed psychedelic experience. They end by stating that, in this context, silence is golden.