Shared Mechanism for Emotion Processing in Adolescents with and without Autism

Thursday, May 11, 2017: 12:00 PM-1:40 PM
Golden Gate Ballroom (Marriott Marquis Hotel)
C. Ioannou1, M. El Zein1, V. Wyart1, I. Scheid2, F. Amsellem3, R. Delorme3, C. Chevallier1 and J. Grèzes1, (1)Ecole Normale Supérieure, Paris, France, (2)Robert Debre University Hospital, Paris, France, (3)Institut Pasteur, Paris, France

Autism spectrum disorders (ASD) are characterised by significant deficits in social interaction and communication skills. Atypicalities in the affective domain are central to ASD and research suggests that individuals with ASD react to social cues differently from typically developing (TD) individuals. The roots of such difficulties are still debated but it has been suggested that difficulties in processing emotional cues play an important role in these social deficits. This hypothesis has led to a considerable amount of work pointing to emotion recognition difficulties, yet, a number of studies have found that people with ASD do recognise emotions accurately.


Here, we shift the focus to ask whether the mechanismsbehind emotion processing are the same in ASD, irrespective of potential differences in accuracy between ASD and TD groups. To do so, we focus on the contextual impact of gaze on emotion recognition for two reasons: Firstly, we know that neurotypical adults better decode and rate as more intense anger paired with direct gaze and fear when coupled with averted gaze (labelled here Threat+), as compared to the other emotion-gaze combinations (threat-), and secondly, because it is theoretically possible to distinguish different mechanisms that may affect the integration of these social cues, using model-based analyses of participants’ accuracy.


20 adolescents (12-17 yrs) with ASD and 20 age-, gender-, IQ- matched TD controls participated in a two-choice emotion categorisation task (fear or anger). In each trial, participants were presented with a facial expression of anger or fear of varying intensity (7 levels of emotion strength), paired with direct or averted gaze, for 250ms, and had to categorise the expressed emotion (4 second response window).

To identify the computational mechanisms accounting for the influence of gaze direction on emotion categorization, we compared two models: model 1, where gaze direction would bias responses towards Threat+ combinations and model 2, where gaze direction would enhance the perceptual sensitivity to Threat+ combinations. We used Bayesian model selection and calculated the Bayesian information criterion (BIC) to determine which of the two models was more likely to explain the observed data.


Behavioural data revealed that the ASD group displayed lower overall performance (TD:86%; ASD:82%). Yet, gaze direction had a similar impact on emotion categorisation in both groups, i.e. improved accuracy for Threat+ combinations (figure 1a). Critically, computational modelling of participants’ behaviour reveals that the same mechanism, i.e. increased perceptual sensitivity, underlies the contextual impact of gaze direction in both groups (figure 1b).


Adolescents with ASD, while being overall less accurate in emotion recognition, are able to integrate gaze direction with facial expressions of emotion similarly to TD adolescents. Importantly, the mechanisms behind such integration are the same as in both TD adolescents and neurotypical adults (El Zein et al., 2015). We discuss the specific experimental conditions that may favour emotion processing and the automatic integration of contextual information in ASD.