Are social robots more interesting than humans? Quantitative assessment of Joint Attention in autistic and typically developing children

Introduction: Autism Spectrum Disorder (ASD) is characterized by deficits in social interaction and communication, including joint attention (JA) impairment. This observational case-control study explores quantitative JA metrics in ASD preschool-aged children and typically-developing (TD) toddlers, interacting with both a human agent and a social robot. Material and methods: Ten ASD and sixteen Typically Developing (TD) children, matched for developmental age, participated completing two JA tasks of the “Échelle par la Communication Sociale Precoce”(ECSP)-Gaze Following and Object Spectacle-administered by both a human (ECSP-H) and a social robot NAO (ECSP-R). A Kinect Azure Camera was used to track children's gaze and five performance indicators were defined to quantify JA-related behaviors: Number of Responses to JA, Time Latency to response and Fixation Time, both towards robot and human agent. Results: Significant group differences emerged in responsiveness to social cues, particularly in the Gaze Following task. TD children responded more consistently to human prompts (p = 0.01) and showed greater visual interest in the human agent in both ECSP-H (p < 0.01) and ECSP-R (p = 0.02) conditions. In contrast, percentage of fixations on the robot was comparable across groups. Discussion: This study introduces unobtrusive, gaze-based quantitative measures to assess joint attention in ASD. Significant differences between ASD and TD groups suggest that children with ASD exhibited greater attentional preference to the robot agent, giving potential for these metrics in distinguishing attentional patterns in these two populations. This feature may nonetheless be clinically relevant, as it may be leveraged to facilitate attention, promote participation, and support robot-assisted therapeutic interventions.