Using Facial Expression Analysis Software to Examine the Relationship Between Abnormal Facial Expressions and Alexithymia in Children with and without Autism

Background:  Buck and Powers (2013) argue that the ability to nonverbally express one’s emotions is an important developmental skill, because the ways in which caregivers respond to those emotions can help children understand what emotion they are experiencing, why they are experiencing it, and may subsequently offer strategies on how to regulate that specific emotion. If children are non-expressive or express their emotions in confusing ways, learning opportunities are lost and alexithymia (i.e., difficulties identifying and describing one’s feelings) could develop over time. Buck and Powers’ hypothesis may be especially relevant for children with Autism Spectrum Disorder (ASD), as approximately 50% of this population has high rates of alexithymia compared to 10% in the neurotypical population (Berthoz & Hill, 2005), and because ASD is characterized in part by blunted and/or confusing nonverbal emotional expression (APA, 2013). While Buck and Powers’ hypothesis has been supported by several studies using neurotypical adults, to date, no studies have examined the relationship between nonverbal emotional expression and alexithymia in child participants or in the ASD population.

Objectives:  The objective of this study was to examine associations between nonverbal expression and alexithymia in children with and without ASD. 

Methods:  Participants (see Table 1) viewed images and videos extracted from YouTube and Google Images designed to evoke various emotional responses, while being covertly recorded with a webcam. The stimuli were collated into a 15-minute .mp4 video file. Participants were instructed to sit still and watch carefully for the entire video. Subsequently, webcam recordings were analyzed using iMotion’s Emotient software (ImotionsGlobal, 2015), which estimates which emotions are being expressed based on Ekman et al.’s (2002) facial action coding system (FACS). In cases where the entire 15-minute recording was not usable (e.g., due to participant distraction), “inattention time” was partialed out. Concurrently, parents completed the parent-report Children’s Alexithymia Measure (CAM; Way et al., 2010), with higher scores on the CAM reflecting stronger alexithymic traits. 

Results:  A significant negative correlation was observed between CAM scores and expression of contempt, r(32) = -.351, p = .021, as well as an aggregate of all negative emotions (including sadness, contempt, disgust, anger and fear), r(31) = -.336, p = .028). In addition, higher CAM scores were associated with higher rates of neutral expression r(32), = .341, p = .024. Thus, children with stronger alexithymic traits produced weaker negative expressions and were more expressively neutral than children with weaker alexithymic traits.

Conclusions:  This study is the first to show that less salient facial expressions are associated with higher levels of alexithymia in children with and without ASD, consistent with Buck and Power’s (2013) hypothesis. These results support the possibility that blunted or confusing emotional expression may contribute to the high rates of alexithymia in the ASD population, although causation cannot be inferred from this study’s design. Future research should use more ecologically valid stimuli, longitudinal designs, and examine facial expressions during interactions between children and caregivers to gain a clearer understanding of how abnormal facial expressions may contribute to the development of alexithymia.