Facial Emotion Recognition of Dynamical Morphing Facial Expressions in Children with Autism Spectrum Disorder

• In Event: Poster Session 06
  In Poster Session: PS 06 Topic - Developmental Disabilities

Abstract

Faces are the source of important social information revealing emotional signals of intention from others. Facial emotional recognition (FER) skills improve with age with differential developmental trajectories as a function of emotion polarity (Thomas, De Bellis, Graham, La Bar, 2007). Autism spectrum disorder (ASD) is a developmental disorder that involves social and emotional impairments (Lombardo, Chakrabarti, Bullmore, & Baron-Cohen, 2011) yet the research regarding FER ability in children with ASD has yielded mixed results, Some researchers report significant impairments in FER accuracy in ASD samples (Loth, et al., 2018) while others have reported little to no differences in performance between ASD and typically developing (TD) groups (Capps, Yirmiya, & Sigman, 1992). Additional research suggests that impaired FER in individuals with ASD is only identified when the emotional expressions are more subtle (Wong et al., 2012) or in timed conditions (Behrmann et al. 2006). Further, while some researchers have postulated that deficits in FER abilities are more noticeable for negative emotions (Shanok, Jones & Lucas, 2019) others have found evidence for impairment in identifying happy expressions in ASD samples (Sato et al., 2017). Given that the ability to discern emotional expressions has adaptive value and is essential for successful social interaction (Tay, 2015) even in those with ASD, it is imperative to reconcile these differences. The overarching aim of this project was to gain understanding of the emotion processing abilities of children with ASD by examining their ability to accurately recognize facial emotions from computer-generated, dynamically morphed displays. The first goal was to investigate if accuracy and FER time varied as a function of the morph type (e.g. neutral to negative versus neutral to positive) across groups. It was predicted that children would have more success with the positive condition. Second, we compared group differences in the recognition ability of emotional expressions for familiar (mother) vs unknown faces. Both accuracy and reaction time were considered, and it was predicted that the typically-developing (TD) group would perform better than the ASD group but, in line with previous research, those in the ASD group would have more success with familiar faces. Ten children between the ages of 4-8 years were recruited from local schools and autism centers. Using Morph Man (Stoik, 2016), facial expressions from the participants’ mothers and unknown faces from the NimStim FER Stimuli (Tottenham et al., 2009) were morphed from a neutral expression to each of four emotional expressions (happy, sad, anger, fear). Written emotion descriptors accompanied by emoticons were displayed under the facial morphing videos. Participants touched the word that correctly identified the viewed expression as soon as they recognized the emotion allowing for tracking of FER processing in real time. As predicted, participants with ASD had slower FER processing speed than their typically developing peers and required higher emotional intensity before recognition. These impairments were mediated by face familiarity (Figures 1 and 2). Ultimately, this task featuring dynamic morphing of faces may provide a more comprehensive assessment of socio-emotional abilities in ASD children.

Authors

• Krystal D. Mize, Florida Gulf Coast University
  Presenting Author

• Katy Brooks, Florida Atlantic University
  Non-Presenting Author

• Nancy Aaron Jones, Florida Atlantic University
  Non-Presenting Author