Social Monitoring of Conversation Partners in Young Children with ASD

Background: Attenuated attention to ostensive social bids (e.g., gaze cues, speech) appears to be a prodromal and early syndromal feature of ASD (Charwarska et al., 2012; Shic et al., 2014). These early differences in social monitoring may compromise the uptake of foundational social communicative inputs and contribute to difficulties in language and social competency. Recent work suggests that simplified humanoid figures not only facilitates attention to faces (Tsang et al., 2018), but may also support learning when used in ASD interventions (Scassellati et al., 2018). However, it is unclear whether simplified humanoid figures (e.g., puppets) are regarded as equivalent social agents to people, and whether children with ASD are receptive to their social communicative bids. This will critically inform the application of simplified social figures as tools to increase social learning opportunities in ASD.

Objectives: (1) Examine how children with ASD monitor a conversation between a puppet and person, and (2) its association with social communicative competence. Simpler humanoid figures (e.g., puppets) may facilitate attention to relevant social information for children with ASD, relative to their typically developing (TD) and developmentally delayed (DD) peers, which may provide supported social learning opportunities.

Methods: 81 participants (ASD: N=29, Mean=46.25(SD=14.76) months; DD: N=25, Mean=51.29(SD=17.39) months; TD: N=27, Mean=39.83(SD=10.94) months) were eye-tracked while they freely viewed a video depicting a puppet and person engaged in an animated conversation (Figure 1). Participants were characterized using the ADOS-2 and a cognitive test (e.g., Mullen or DAS). Eye-tracking data were analyzed under two conditions: 1) Puppet-Speaking/Person-Listening and 2) Person-Speaking/Puppet-Listening. For each condition, we measured overall percent looking at the scene (%Valid), percent of valid looking directed at faces (%Face_Speaker, %Face_Listener), and a composite ratio of relative attention to the speaker versus listener within the dyad (i.e., time spent looking at the speaker divided by looking to either speaker or listener). Attention to faces was analyzed using a 3 (Diagnosis) x 2 (Condition) x 2 (Social Agent: puppet vs. person) repeated-measure ANCOVA, covarying for age. Associations between the composite ratio metric and clinical features were evaluated.

Results: The ASD group had lower %Valid in Puppet-Speaking/Person-Listening (F(2,80)=3.984,p=.022) and Person-Speaking/Puppet-Listening conditions (F(2,80)=7.033,p=.002) than TD and DD groups. There was a significant 3-way interaction (F(2,77)=5.34, p<.001, partial eta2=.12)—children with ASD had lower %Face_Speaker-Person, but not %Face_Listener-Puppet, in the Person-Speaking condition; all diagnostic groups showed similar patterns of %Face_Speaker-Puppet and %Face_Listener-Person in the Puppet-Speaking condition (Figure 2). In the ASD group, a higher puppet-speaker to person-listener ratio was moderately associated with higher ADOS scores (r=.47,p=.06). Conversely, a higher person-speaker to puppet-listener ratio was associated with lower symptom severity (r=-.41,p=.03), and higher verbal (r=.46,p=.01) and nonverbal (r=.59,p=.001) cognitive skills.

Conclusions: Overall, children attend to the speaker in a conversational dyad. While children with ASD attended less to the social scene than their TD and DD peers, the presence of the puppet effectively engaged children with ASD to socially monitor conversational partners. Perceptually salient, humanoid forms may serve as alternative routes for social learning especially in lower functioning and cognitively/verbally challenged children with ASD.