The ‘Reading the Mind in the Eyes' Test: Complete Absence of Typical Sex Differences in Performance in ~400 Men and Women with Autism

Background:  

The ‘Reading the Mind in the Eyes’ test (Eyes test) is an advanced test of theory of mind. Typical sex differences have been reported (female advantage). Individuals with autism show more difficulty than typically developing individuals, yet it remains unclear how this is modulated by sex, as previous studies have been relatively small scale and females with autism have been under-represented. 

Objectives:  

We recruited a large, sex-balanced sample to test for the effects of sex, diagnosis, and any interaction of these, on the performance of the Eyes test. We also tested if the patterns of group differences fit predictions from the ‘extreme male brain’ (EMB) theory of autism. 

Methods:  

The Eyes test was administered online to 395 adults with autism (178 males, 217 females) and 320 typical adults (152 males, 168 females). Performance was examined in terms of total correct score, using a 2 x 2 factorial design. Correlation between Eyes test correct score and self-reported empathy (measured using the Empathy Quotient, EQ) and autistic traits (measured using the Autism Spectrum Quotient, AQ) were examined separately for the four groups. Support Vector Machine (SVM) classification, stratified by sex, was also performed. 

Results:  

Two-way ANOVA showed a significant sex-by-diagnosis interaction (F(1,711) = 5.090, p = 0.024, η_p² = 0.007) arising from a significant sex difference between typical males and typical females, and a complete abolition of sex difference in males and females with autism (p = 0.907, d = 0.01). Case-control differences in males and females were observed. The pattern of group differences was in line with the EMB theory predictions (typical females > typical males > males with autism ≈ females with autism). Eyes-EQ and Eyes-AQ correlations were significant only in females with autism (r = 0.35, r = -0.32, respectively), but not in males with autism, typical males, or typical females. SVM on Eyes items classified autism diagnosis at higher accuracies for females than males (males, 63%; females, 72%). Furthermore, an SVM model trained within one sex generalizes equally well when applied to the other sex. 

Conclusions:

In a large adult sample we confirmed that performance on the Eyes test is a sex-common phenotypic marker of autism, reflecting core social-communication difficulties in autism. Given the existence of typical sex differences, performance of females with autism differs from same-sex controls more than that of males with autism. The findings also provide support for the EMB theory predictions in both sexes in terms of overall performance.