It’s All Good: Contextual Emotional Incongruity in Spontaneous Facial Expressions of Children with ASD

Background: Facial expressions of children with ASD are often perceived as awkward (Grossman, 2015) or ambiguous (Yirmiya, Kasari, Sigman, & Mundy, 1989), and are more difficult to recognize (Brewer et al. 2016). Young children with ASD show emotional facial expressions that aren’t always congruent with social context (McGee, Feldman, & Chernin 1991), particularly when positive expressions are paired with negative behaviors (Costa, Steffgen, & Samson 2017). Most previous data into expression incongruity rely on posed or mimicked facial expressions, which may not be representative of spontaneous expressiveness. On the other hand, it can be difficult to determine the contextual valence of social interactions.

Objectives: Analyze facial expressions produced spontaneously by children with and without ASD in response to brief video clips whose emotional content was well defined.

Methods: We presented a pseudo-randomized sequence of funny and disgusting YouTube clips to neurotypical (NT) and autistic children. Participants’ mean age was 13:8 (ASD) and 13:1 (NT). Groups were not significantly different in age, sex, IQ, and language ability. Participants with ASD had significant higher scores (p < .0001) on the Social Communication Questionnaire Lifetime form (SCQ), indicating greater social communication impairment.

We videotaped participants’ facial responses to the videos and coded the durations (in ms) of positive/negative/flat expressive valence during the most evocative six seconds of each video. All coders were blind to diagnosis and all videos were independently double coded. When coders disagreed, a third coder was involved and final disagreements were resolved via discussion (Ventresca et al. 2017).

We calculated the percentage of positive, negative, and flat expressive duration relative to the duration of each 6,000ms clip. We then grouped responses to four happy video clips (funny kitten videos and laughing babies) and four disgusting video clips (people eating grubs, brains, and spiders, disgusting things being pulled out of a nose) and calculated the percentage of positive, negative, and flat responses to those composites.

Results: Neurotypical (NT) children were significantly more likely to have flat responses (Disgusting videos: F (1, 34) = 6.59, p = .015, Happy videos: F (1, 34) = 6.86, p = .013), while ASD children were significantly more likely to show positive responses (Disgusting videos: F (1, 34) = 7.03, p = .012, Happy videos: F (1, 34) = 7.42, p = .01). There were no group differences for negative expressions. There was a significant positive correlation between the SCQ and positive expressions in response to videos (disgusting videos: r = .47, p = .007; funny videos: r = .4, p = .028).

Conclusions: Children with ASD are more expressive overall in response to video stimuli and make more positive expressions regardless of stimulus valence. For disgusting stimuli this results in a mismatch between the emotional quality of the content they are watching and the facial expression they produce. This bias toward positive expressions and resulting expressive incongruity for stimuli with negative content seems to be related to greater social communication impairment and could lead to misunderstandings in social context.