Adverse Effect-Related Language in Recreational Psychedelic and non-Psychedelic Substance Experience Reports

In recent decades, psychedelic substances like LSD and psilocybin have been increasingly researched, but relatively few studies examined their adverse effects in recreational settings. As online substance experience reports may offer relevant information, we conducted a psycholinguistic (i.e., inferring psychologically relevant outcomes from natural language) assessment of popular psychedelic and related substances' adverse effects.

We applied latent semantic analysis to 4226 online substance experience reports, calculating each report's respective semantic similarity with 77 empirical literature-based adverse effect-related reference texts. We compared seven groups including ayahuasca, (isolated) DMT, ketamine, LSD, MDMA, psilocybin mushroom, and recreational antidepressant experiences (as a non-psychedelic comparator). Using sentiment analysis, we also classified each report's contextual valence (positive vs. negative) to examine its potential influence on between-substance semantic similarity differences.

Fifteen outcomes featured large effect sizes; with semantic similarity with vomiting-, excessive drowsiness-, cardiac event-, sleep problem-, and dissociation- or depersonalization-related terms leading the list. Report groups standing out with high scores were ayahuasca on vomiting-, ketamine on dissociation-, and MDMA on excessive drowsiness- and cardiac event-related semantic similarity. Report groups standing out with low scores were MDMA on dissociation-, ayahuasca on sleep problems-, and DMT on sleep problems- and drowsiness-related semantic similarity. A literature-driven secondary examination showed that LSD and psilocybin mushroom reports scored higher than all other groups on anxiety-related outcomes.

Despite the limitations of the observational study context, these results provide valuable insights into adverse effect-related language in recreational psychedelic substance experience reports. They show that psycholinguistic analyses of experience reports enable relatively feasible, low-cost, and useful substance comparisons that might be difficult to realize using other research designs.