Drugs like entactogens and psychedelics appear poised to enter clinical psychiatry, however we lack a unified framework for quantifying the changes in conscious awareness optimal for treatment. We address this need using transformers (i.e. BERT) and 11,821 publicly-available, natural language testimonials from Erowid. First, we predicted 28 dimensions of sentiment across each narrative, validated with clinical psychiatrist annotations. Secondly, another model was trained to predict biochemical (pharmacological and chemical class, molecule name, receptor affinity) as well as demographic (sex, age) information from the testimonials. Thirdly, canonical correlation analysis (CCA) linked the 52 drugs' affinities for 61 receptor subtypes with word usage across the testimonials, revealing 11 latent receptor-experience factors each mapped to a 3D cortical atlas of receptor gene-expression. Together, these 3 machine learning methods elucidate a neurobiologically-informed, temporally-sensitive portrait of drug-induced subjective experiences. These models are mutually confirmatory, pointing to an underlying structure of psychoactive experience dominated by the distinction between the lucid and the mundane, but also sensitive to effects unique to individual drugs. For example, MDMA was singularly linked to mid-experience swelling of "Love", potent psychedelics like DMT, and 5-MeO-DMT were associated with "Mystical Experiences", while other tryptamines were associated with an emotional constellation of "Surprise", "Curiosity" and "Realization". Applying these models to real-time biofeedback (e.g. EEG, or MRI) with zero-shot learning that tunes the sentimental trajectory of the experience through changes in audiovisual outputs, practitioners could guide the course of therapeutic sessions, maximizing benefit and minimizing harm for patients.