Increasing Facial Recognition Skills in Children with Autism

Background: Children with autism consistently demonstrate deficits in facial processing compared to their typically developing peers.  They also demonstrate difficulties with global processing of both facial and non-facial stimuli.  Research has shown that children with autism spectrum disorders tend to focus more on the mouth area, rather than the eyes, when looking at faces.  Facial recognition skills have been shown to be highly positively correlated with improved social skills, which are correlated to such desirable outcomes as better emotion recognition abilities and increased prosocial behaviors (Denham et al., 2003, Eisenberg et al., 1996).
Thus far, studies investigating attempts to increase facial recognition skills in populations of children with autism have not found significant improvements (Faja et al., 2008). However, recent research utilizing avatars in the computer-based social skills intervention FaceSay™ has shown it to significantly increase facial recognition skills in a population of adolescents with autism spectrum disorders (Hopkins, 2007) and socially at-risk children in Head Start (Perez, 2008).

Objectives: The current study sought to replicate previous findings of studies using the FaceSay™ intervention in a population of three- to six-year-old children with autism spectrum disorders and typically developing children.

Methods: The games in FaceSay™ utilize avatars, which are virtual people that are capable of interacting with humans, and each game in the intervention is targeted at training children pay attention to all the components of the face, promoting appropriate facial recognition skills.    The children played the game for 15-30 minutes, depending on their ability to attend to the games, twice per week.  The intervention lasted 12 weeks, and the children in this study were tested on the Benton Facial Recognition Task prior to and after participating in the intervention.
Results: A factorial analysis of covariance (ANCOVA) was completed to assess the effect of game assignment (Face Say™, Tux Paint) and diagnosis (with or without an Autism Spectrum Disorder), as well as the interaction between the two factors on post-intervention facial recognition skills.  The pre-intervention scores on the Benton Facial Recognition test and age served as the covariates. After adjustment by the covariate, there were no significant main effects or interaction effects.  However, when diagnosis was removed from the model and FaceSay™ was compared to Tux Paint, a significant difference was found, F (1,22) = 4.715, p = .041, η² = 0.18. 
Conclusions: After looking at the means of the groups before and after taking part in the intervention, it became apparent that the children with autism were performing at the same level as their typically developing peers on the Benton Facial Recognition Task after the intervention.  The lack of significant overall findings in the first analysis was likely due to issues related to power and the fact that the typically developing children were likely performing at near ceiling levels prior to the intervention.  Our findings suggest that this is a useful intervention to increase the facial recognition abilities of young children with autism.