Effects of a Targeted Face-Processing Intervention On Visual Attention to Naturalistic Social Scenes

Background: Children with autism spectrum disorders (ASD), as compared to typically-developing peers, are selectively impaired in their ability to recognize facial identity. Eye-tracking studies have shown that viewing patterns in children with autism are driven by the physical contingencies of a stimulus rather than its social context. A critical question is thus whether viewing patterns can be enhanced through a direct training intervention.

Objectives: The aims of the current study were (1) to assess the effect of a face-processing intervention on visual attention to dynamic social scenes and (2) to examine the relationship between performance on targeted face-processing tasks and patterns of visual attention to faces within more naturalistic settings.

Methods: In a randomized clinical trial, school-age children diagnosed with ASD were pre-screened with the Let’s Face It! skills battery that measures both face and object processing abilities. Participants who were significantly impaired in their face processing abilities were assigned to either a treatment (N=42) or waitlist group (N=37). Children in the treatment group received 20 hours of face training with Let’s Face It! (LFI!) computer-based intervention. The LFI! program was composed of seven interactive computer games that targeted specific face processing impairments associated with ASD, including the recognition of identity across image changes in expression, viewpoint, and features, analytic and holistic face processing strategies, and attention to information in the region of the eyes. Children were re-assessed using the LFI! skills battery post-treatment.

In a subset of participants (13 of the children in the treatment group and 15 in the waitlist group), eye-tracking data were collected at both time points while the children viewed videos of children and adults engaged in naturalistic, age-appropriate social interaction. Percentage of visual fixation time on eyes, mouth, body, and object/background regions was calculated across all scenes for each viewing. Time-varying visual scanning patterns were also measured using kernel density estimation. Viewing patterns at Time 1 were compared to those of Time 2, as well to normative viewing patterns from a sample of 36 age- and IQ-matched typically-developing peers.

Results: Preliminary analyses on viewing patterns in a subset of the sample (n=5) suggested an interaction between time point and treatment condition (η_p²=0.12). Following LFI! training, children looked more at faces (eyes and mouth regions combined) and less at object/background regions than during pre-training (d=0.79). The viewing patterns of children in the waitlist condition changed less across time (d=0.25). Ongoing analyses will examine the effect of LFI! training in the full sample, where the greater power will enable evaluation of the statistical significance of these findings. These analyses also will include more detailed study of time-varying visual scanning patterns as well as an investigation of how changes in performance on the LFI! face-processing skills battery may mediate changes in visual attention to the naturalistic stimuli.

Conclusions: The present study suggests that direct face-processing training using the LFI! program alters how children with ASD engage with their natural social visual environment.