Poster Presentation

Poster F in Poster Session F: Thursday, August 6, 1:45 – 3:30 pm, Kimmel Center, Shorin & Rosenthal Rooms

Computational Natural Language Analysis Reveals Psilocybin-Induced Changes in Meaning-Making

Caitlin Rooney¹, Krisztina Jedlovszky¹, Joanna Kuc¹, Jeremy Skipper¹; ¹University College London

Presenter: Caitlin Rooney

Emerging evidence indicates that psilocybin produces sustained improvements in wellbeing, thought to be mediated by changes in how individuals construct meaning from internal and external experience. However, the mechanistic basis of this process is poorly understood. We draw on Linguistic Active Inference Theory (Yao, 2025) and the REBUS model (Carhart-Harris & Friston, 2019), to argue that naturalistic speech provides a tractable window into psychedelic-induced perturbations of precision weighting. Within this framework, psilocybin is hypothesised to reduce the precision of high-level prior beliefs, increasing sensitivity to incoming prediction errors and promoting more flexible inference. We tested this account in a sample of 40 healthy participants undergoing a single high-dose psilocybin intervention. Meaning-making dynamics were assessed via naturalistic thought sampling across 10 days both pre- and post-intervention, with speech data analysed using bag-of-words, probabilistic topic modelling and graph-based semantic network analyses. Psilocybin was associated with an increased frequency of self-referential and meaning-making terms. Graph-based analyses revealed reduced modularity and increased global integration within semantic networks — a pattern we interpret as reflecting reduced top-down prior precision and more flexible integration of information. These computational signatures were associated with a shift in inner speech valence towards more positive and regulatory forms, and increased wellbeing, supporting a mechanistic account linking psychedelic-induced precision modulation to psychological outcomes.

Topic Area: Methods, Tools, Theory & Neural Coding