Dept. of Biomedical Engineering, McGill University; Mila Quebec AI Institute
SUNY Downstate Medical Center
Data Sciences Platform, Broad Institute
Recent research indicates that certain hallucinogens can be used to treat a wide variety of mental illnesses and require only a few doses to produce durable effects. Different compounds have distinct but overlapping biological mechanisms, for example the serotonin 2A receptor agonism shown by classical psychedelics. Subjective reports of these drug experiences describe overlapping phenomenological profiles, often including vivid imagery, mystical experiences, and imaginary creatures. Elucidating the mechanisms that underpin the extreme variability of these experiences is vital for developing and refining these drugs for use in the clinic. While most studies on psychedelics include dozens of participants and one molecule, our study integrates 6,850 real-world testimonials and receptor affinity fingerprints of 27 hallucinogenic drugs. With Canonical Correlation Analysis (CCA) we derive the underlying experiential factors from user-generated data in a manner that is intrinsically linked to drug receptor affinity. These distilled receptor-semantic factors are mapped to 3D voxels of the human cortex via RNA expression patterns of different brain regions given by the Allen Brain Atlas. The underlying components show semantic landscapes that span the sensorial (eg audio vs visual hallucinogens), emotional (eg terror vs bliss) and mystical. Each component is simultaneously described in neurochemical terms across the serotonin, opioid, and dopamine systems. Given testimonial data and molecular receptor affinities, this framework extends to new drugs to help identify relationships across domains of experience, molecules, and neuroanatomy.