Can Humor Reveal Dark Personalities? Machine-Learning Detection of the Dark Tetrad

Journal of Individual Differences. 2026. No. 47(3). P. 132–143.

Glinkina L. S., Doptan E., Kosonogov V.

This study investigates a novel method for assessing Dark Tetrad personality traits (Machiavellianism, Narcissism, Psychopathy, Everyday Sadism) through humor evaluation. Traditional self-report measures are limited by response biases, creating a need for indirect behavioral assessment. We hypothesized that individual differences in these traits would correlate with distinct patterns in evaluating visual humor (funniness, perceived kindness/cruelty, and target direction) and that machine learning (ML) could classify individuals based on these patterns.

A sample of 200 participants rated 71 funny pictures and completed the Short Dark Tetrad (SD4) questionnaire. Correlation analyses revealed that Psychopathy and Sadism were associated with higher funniness and lower kindness ratings. Narcissism correlated only with higher funniness, while Machiavellianism showed no significant correlations. ML models confirmed these patterns: Psychopathy was classified with good accuracy (AUC up to .819), followed by Sadism. Narcissism was moderately predictable, while Machiavellianism was not predicted above chance level.

The findings demonstrate that computational analysis of humor evaluation can serve as a valid behavioral predictor for specific Dark Tetrad traits, particularly Psychopathy and Sadism. This approach offers a promising indirect assessment method, though its efficacy varies across traits and requires careful ethical consideration for any potential application.

Keywords: machine learning, The Dark Tetrad, humor evaluation

Publication based on the results of: Mental health research: behavioral and neurophysiological foundations (2025)