BackgroundAutistic people are disproportionately vulnerable to anorexia nervosa and other eating disorders (ED), and within the general population, autistic traits correlate with ED psychopathology. A putative mechanism which may underpin this heightened risk is alexithymia, a difficulty identifying and describing emotional states which is observed in both autism and ED. In two experiments with independent non-clinical samples, we explored whether alexithymia might mediate the heightened risk of eating psychopathology in individuals high in autistic traits.MethodsOur first experiment used the PROCESS macro for SPSS to examine relationships between alexithymia (measured by the Toronto Alexithymia Scale (TAS-20)), autistic traits (autism quotient (AQ)), and eating psychopathology (Eating Attitudes Test (EAT-26)) in 121 participants. Our second experiment (n = 300) replicated and furthered this analysis by examining moderating effects of sex and controlling for anxiety and depression as covariates. We also included an additional performance-based measure of alexithymia, the Levels of Emotional Awareness Scale (LEAS).ResultsStudy 1 suggested that TAS-20 scores mediated the relationship between heightened autistic traits and eating psychopathology. Replication and further scrutiny of this finding, in study 2, revealed that this mediation effect was partial and specific to the female participants in this sample. The mediation effect appeared to be carried by the difficulty identifying feelings subscale of the TAS-20, even when depression and anxiety were controlled for. LEAS scores, however, were not significantly related to autistic traits or eating psychopathology.LimitationsCross-sectional data prevents any conclusions around the direction and causality of relationships between alexithymia, autistic traits, and eating psychopathology (alongside depression and anxiety), necessitating longitudinal research. Our non-clinical sample was predominantly Caucasian undergraduate students, so it remains to be seen if these results would extrapolate to clinical and/or autistic samples. Divergence between the TAS-20 and LEAS raises crucial questions regarding the construct validity of these measures.ConclusionsOur findings with respect to autistic traits suggest that alexithymia could partially explain the prevalence of ED in autistic people and may as such be an important consideration in the pathogenesis and treatment of ED in autistic and non-autistic people alike. Further research with clinical samples is critical to explore these ideas. Differences between men and women, furthermore, emphasize the importance of looking for sex-specific as well as generic risk factors in autistic and non-autistic men and women.