Abstract
This study examined the associations between psychopathy dimensions (triarchic phenotypes and classical factors), empathy domains (cognitive and affective), and interoception (interoceptive attention and accuracy) while accounting for the putative role of alexithymia. A community sample (n = 515) completed an online survey encompassing: Triarchic Psychopathy Measure (boldness, meanness, disinhibition); Levenson Self-Report Psychopathy Scale (primary and secondary psychopathy); Body Perception Questionnaire (interoceptive attention); Interoceptive Accuracy Scale; Toronto Alexithymia Scale. Hierarchical linear regression models were implemented for hypothesis-driven analyses examining the associations between psychopathy, empathy, and interoception while controlling for sex, age, and alexithymia. Exploratory path models were employed to investigate alexithymia and/or cognitive empathy as mediators between interoception and psychopathy. Our results largely confirmed the postulated empathy profiles across psychopathy dimensions, as meanness and primary psychopathy displayed a broad empathy impairment, while disinhibition and secondary psychopathy were only associated with diminished cognitive empathy. Importantly, boldness displayed a unique pattern (enhanced cognitive empathy and reduced affective empathy), further reinforcing its importance within the constellation of psychopathy traits. Contrary to our hypotheses, self-perceived interoceptive attention and accuracy were not associated with either psychopathy dimension after controlling for alexithymia. However, interoceptive accuracy and alexithymia were associated with cognitive empathy, while alexithymia was also positively related to all psychopathy dimensions (as expected), despite the unexpected strong and negative association with boldness. Exploratory analyses suggested significant indirect effects (mediation) between interoceptive accuracy and psychopathy via alexithymia and/or cognitive empathy. These mediating effects must be interpreted with caution and future studies should be designed to formally test this model.