Natural language signatures of psilocybin microdosing.

Natural speech can reveal whether someone has taken a microdose of psilocybin. In a double-blind, placebo-controlled experiment, 34 healthy adults provided speech samples after consuming either 0.5 grams of psilocybin mushrooms or a placebo. Machine learning classifiers distinguished between the two conditions with high accuracy (AUC ~0.8), based on features such as verbosity and sentiment scores, though semantic variability did not differ significantly. This suggests that low doses of serotonergic psychedelics leave detectable signatures in unconstrained speech, offering a potential low-cost, non-invasive method for monitoring microdosing regimens.

Study at a glance

Abstract

Serotonergic psychedelics are being studied as novel treatments for mental health disorders and as facilitators of improved well-being, mental function, and creativity. Recent studies have found mixed results concerning the effects of low doses of psychedelics ("microdosing") on these domains. However, microdosing is generally investigated using instruments designed to assess larger doses of psychedelics, which might lack sensitivity and specificity for this purpose. Determine whether unconstrained speech contains signatures capable of identifying the acute effects of psilocybin microdoses. Natural speech under psilocybin microdoses (0.5 g of psilocybin mushrooms) was acquired from thirty-four healthy adult volunteers (11 females: 32.09 ± 3.53 years; 23 males: 30.87 ± 4.64 years) following a double-blind and placebo-controlled experimental design with two measurement weeks per participant. On Wednesdays and Fridays of each week, participants consumed either the active dose (psilocybin) or the placebo (edible mushrooms). Features of interest were defined based on variables known to be affected by higher doses: verbosity, semantic variability, and sentiment scores. Machine learning models were used to discriminate between conditions. Classifiers were trained and tested using stratified cross-validation to compute the AUC and p-values. Except for semantic variability, these metrics presented significant differences between a typical active microdose and the inactive placebo condition. Machine learning classifiers were capable of distinguishing between conditions with high accuracy (AUC [Formula: see text] 0.8). These results constitute first evidence that low doses of serotonergic psychedelics can be identified from unconstrained natural speech, with potential for widely applicable, affordable, and ecologically valid monitoring of microdosing schedules.