Skip to content

Galen Ballentine

SUNY Downstate Health Sciences University

3 papers in the library · 12 citations · publishing 2021-2023

Papers

Trips and Neurotransmitters: Discovering Principled Patterns across 6,850 Hallucinogenic Experiences

bioRxiv (Cold Spring Harbor Laboratory) July 14, 2021 Galen Ballentine, Sam Friedman, Danilo Bzdok 6 citations preprint

Psychedelic drugs alter consciousness by disrupting how the brain's higher association cortex processes incoming sensory signals. Analyzing 6,850 free-form testimonials about 27 drugs and linking them to 40 neurotransmitter receptor subtypes via gene transcription maps, a pattern-learning approach revealed that specific changes in awareness—such as dissolving self-world boundaries or fractal visual distortions—correspond to distinct distributions of receptor densities across the cortex. Ego-dissolution-like experiences were tied to 5-HT2A, D2, KOR, and NMDA receptors in both deep hierarchical (associative higher-order cortex) and shallow hierarchical (visual cortex) brain regions. Emotional effects involved 5-HT2A and Imidazoline1 receptors, while auditory and visual sensations involved SERT, 5-HT1A, and 5-HT2A receptors. Each receptor-experience factor spanned between higher-level association and sensory input poles, potentially relating to a collapse of hierarchical order among large-scale brain networks.

Trajectories of sentiment in 11,816 psychoactive narratives

Human Psychopharmacology Clinical and Experimental December 20, 2023 Sam Friedman, Galen Ballentine 5 citations

Machine learning models trained on 11,816 publicly available testimonials about 52 different drugs can predict and quantify subjective experiences. The models identified 11 statistically significant latent factors linking drug receptor affinities to word usage, which mapped to brain regions. A pervasive distinction emerged between universal psychedelic experiences of heightened feeling and the grim, mundane experiences of addiction and mental illness. MDMA was linked to “Love”, DMT and 5-MeO-DMT to “Mystical Experiences” and “Entities and Beings”, and other tryptamines to “Surprise”, “Curiosity”, and “Realization”. The methods show potential for characterizing psychoactivity through data-driven sentiment analysis.

Language Models Learn Sentiment and Substance from 11,000 Psychoactive Experiences

Research Square August 17, 2022 Sam Friedman, Galen Ballentine 1 citation

Machine learning applied to 11,816 publicly available drug testimonials reveals distinct subjective experiences linked to specific substances. Using BERT, a transformer model, the study predicted 28 dimensions of sentiment across narratives, validated by psychiatrist annotations. Canonical correlation analysis connected 52 drugs' receptor affinities with word usage, uncovering 11 latent receptor-experience factors mapped to a 3D cortical atlas. Results distinguished lucid from mundane phenomena: MDMA was linked to "Love," DMT and 5-MeO-DMT to "Mystical Experiences," and other tryptamines to "Surprise," "Curiosity," and "Realization." These models could potentially guide real-time biofeedback in therapeutic sessions.