Development of a Data-Driven Taxonomy of Non-Ordinary States of Consciousness through Artificial Intelligence

Overview

This study will use artificial intelligence to analyze people’s descriptions of non-ordinary states of consciousness (NOS) in order to elevate an expert taxonomy and develop a clear, experience-based classification system. Through this research, investigators aim to integrate AI-driven insights with existing expert frameworks and create validated online tools to advance scientific understanding and communication about NOS.

Abstract

Non-ordinary (altered) states of consciousness (NOS) are psychophysiological states that differ from the ordinary baseline state of consciousness. Although for some the ordinary waking state of consciousness is the only one that provides access to “reality,” or at least is not delusional, there is compelling evidence that it entails important limitations including egocentricity, biased perception, and cognitive rigidity, which can impact negatively the mental health and well-being of young people. Conversely, spontaneous (e.g., some transcendent experiences) or induced (by psychedelics, meditation, hypnosis, etc.) NOS can reduce some cognitive limitations and biases, have therapeutic effects, and promote personal and social wellbeing. Nonetheless, the study of NOS has lacked clear terminology, an agreed-to taxonomy, and has often and incorrectly equated the trigger or an induction of an NOS (e.g., through a hypnotic induction) with the state itself (“the hypnotic state”). An expert, multidisciplinary (psychology, neurosciences, anthropology, etc.) taxonomy centered on NOS phenomenology proposed eight types at the basic or “species” level: proto and transitional, delirium, minimal to no-awareness, experiential detachment, enhanced physicality, altered identity, imaginary/fantasy/visionary, and unity/mystical. This proposal has not been validated empirically. Large Language Models (LLMs) use AI to analyze words within their natural contexts, making them useful to establish the characteristics of states of mind. They can generate high-resolution, multidimensional categorizations, which could be later integrated with the expert-consensus taxonomy. This project will: 1) develop a taxonomy of NOS through AI (e.g., large language modeling) analyses of first person-reports in the literature, datasets, and social networks reports; 2) compare it with the consensus taxonomy to modify/expand it, in order to develop online questionnaires in different languages; 3) and, after statistical analyses of the questionnaire answer, establish a taxonomy of NOS based on top-down and bottom-up approaches, which can improve communication, research, and theory among different research areas.

Broader Impact

This project will: 1) develop a taxonomy of NOS through AI (e.g., large language modeling) analyses of first person-reports in the literature, datasets, and social networks reports; 2) compare it with the consensus taxonomy to modify/expand it, in order to develop online questionnaires in different languages; 3) and, after statistical analyses of the questionnaire answer, establish a taxonomy of NOS based on top-down and bottom-up approaches, which can improve communication, research, and theory among different research areas.