Deficits in Emotion Processing Ability Predict Autistic Traits in Children: Eye Tracking Results from a Retrodictive Mindreading Task

Background:  Research presents conflicting results regarding whether autistic traits are associated with emotion processing abilities (Harms, Martin and Wallace, 2010), or reduced attention to the eyes of faces (Cassidy et al. 2014; Jones and Klin, 2013; Jones, Carr and Klin, 2008). Inconsistent findings highlight the need to revisit experimental procedures; including the use of stimuli with more naturalistic responses.

Objectives:  1) To establish whether retrodictive mindreading (RM) ability predicts autistic traits; and 2) To establish whether visual perusal patterns predict autistic traits.

Methods:  38 children without ASC (18 males, 20 females; mean age (8.99 years), SD=19.24, range 72 months – 153 months) took part in a RM emotion task. This involved participants observing short video clips of people reacting to one of four prompts (being asked a difficult maths question, being given monopoly money, being given negative feedback and being given positive feedback), and subsequently guessing which prompt caused the reaction, and how the person felt. Participants’ visual perusal patterns were recorded via the Tobii eye tracker, and participants’ parents completed the Autism Spectrum Quotient child version (AQ).

Results:  A one way ANOVA showed that certain prompt types were more accurately recognised than others (F(2.40, 79.26)= 28.202, p<0.001). Post hoc Bonferroni corrected t-tests showed that responses to the maths question and monopoly money were significantly more accurately recognised than the negative and positive feedback prompts (p<0.001). There were no significant differences in the recognition rates between maths question and monopoly money (p=1), or negative and positive feedback (p=1). Recognition rates did not significantly predict AQ scores for any of the prompts (p>0.05). A multiple regression analysis was conducted with AQ scores as the outcome variable and RM ability, eye-mouth ratio and face-body ratio as predictors. The model was a significant predictor of AQ scores (F(3,28)=3.562, p = 0.027, adj. R² = .20), with total correct RM (p=0.02) a significant predictor. Looking to the face of the body (p=0.058) and eyes of the face (p=0.058) were borderline significant predictors.

Conclusions:  There were significant differences in the recognition rates of the different expressions, but this did not independently predict AQ scores. Results of the multiple regressions suggest that RM ability is associated with autistic traits. More time spent looking at the body also approached significance as a predictor of autism traits. Contrary to some claims the more time spent viewing the eyes relative to the mouth was suggestive of higher autistic traits, however this result was borderline significant. These results suggest autistic traits are related to difficulties in RM and by extension emotion processing, and do present with altered viewing patterns. There are implications for future research considering the stimuli used and if this accurately replicates naturalistic social responses allowing valid recognition testing and visual perusal. These results need to be replicated in a sample with confirmed ASC diagnosis.