Natural language analysis of the structure of altered states of consciousness

Abstract Background and aims Altered states of consciousness (ASC) represent acute and marked deviations from normal waking consciousness. Investigations into ASC are significant to problems in medicine, science, and philosophy, including the structure of conscious experience. Here, we conducted a preliminary investigation into the structure of ASC while addressing the role of psychedelics, which purportedly manifest features of mind. Methods We performed quantitative and qualitative analyses of 300 narrative reports across 12 ASC induction methods: meditation, float tank, psilocybin, lysergic acid diethylamide (LSD), N,N-dimethyltryptamine (DMT), 5-methoxy-N,N-DMT (5-MeO-DMT), ketamine, salvia, 3,4-methyl​enedioxy​methamphetamine (MDMA), cannabis, datura, and diphenhydramine (DPH). We hypothesized that reports from the psychedelics (serotonin 5-HT 2A receptor agonists) would contain similar content with non-pharmacological induction methods, alongside greater positive sentiment and reported authenticity relative to reports from other substances. Results In quantitative analysis, most psychedelics, except LSD, as well as salvia and ketamine, shared similar content with non-pharmacological methods. In qualitative analysis, most psychedelics, except LSD, were deemed both positive and authentic, with authenticity predicting positive sentiment across the 12 ASC induction methods ( R = 0.68; p = 0.015). We uncovered latent themes charting a trajectory of ASC from baseline to metaphysical experience, incorporating text-to-image generative artificial intelligence to illustrate underlying phenomenological structure. Conclusions Our findings suggest that reproducible structural observations may be externally validated across methods to support a “mind-manifesting” characterization for some ASC induction methods, such as salvia, ketamine, or 5-MeO-DMT, but not for others, such as LSD, datura, or DPH, together informing future studies of psychedelics, ASC, and structuralism.