Characterizing Temporal-Contextual Effects on Social and Object-Directed Attention in ASD Via High-Density Video Coding

Oral Presentation
Friday, May 11, 2018: 2:21 PM
Grote Zaal (de Doelen ICC Rotterdam)
A. Rozga1, A. Southerland1, M. McCall1, E. A. Stubbs1, M. R. Silverman2, E. Ajodan2, K. Chanda1, E. Chong1, J. Rehg1 and R. M. Jones3, (1)Georgia Institute of Technology, Atlanta, GA, (2)Sackler Institute for Developmental Psychobiology, New York, NY, (3)Weill Cornell Medicine, New York, NY
Background: Atypical patterns of gaze in social interactions remain among the earliest behavioral red flags for autism and are commonly targeted in treatment. Observational studies typically examine global frequencies of behaviors such as eye contact and joint attention. Eye-tracking technologies have enabled more detailed characterization of gaze dynamics in laboratory “screen tasks,” but such high-density measurements have rarely been applied to more ecologically valid “live” social interactions.

Objectives: We used high-density video coding to characterize patterns of social and object-directed gaze, along with temporal-contextual factors, during live play interactions in children with autism (ASD) and typically developing (TD) children.

Methods: 56 ASD and 58 TD children were administered the Early Social Communication Scales, videotaped with a camcorder as well as a camera embedded in a pair of glasses worn by the examiner that captured the child’s looks to the examiner’s eyes. A subset of the sessions – the six ESCS toy-spectacle tasks for 21 ASD (mean age 45.1mo, SD=10.6; mean VIQ=74, SD=33; 71% male) and 38 TD (mean age 27.4mo, SD=5.3; 63% male) participants – were coded at the frame level for the onset/offset of each instance of (1) child looking to the examiner’s eyes; (2) child looking at the toy; (3) examiner activating the toy, (4) the toy inactive and out of the child’s reach, and (5) the toy in the examiner’s vs. child’s possession.

Results: Dependent variables were not normally distributed; hence, we report the results of nonparametric tests. Compared to TD children, the ASD children spent less time making eye contact with the examiner when the toy was inactive (p=.001) and in their possession (p=.01), with no group differences during toy active periods (p=.06). The ASD group produced fewer shifts of attention from the toy to the examiner’s eyes across toy active (p=.01), toy inactive (p<.001), and child possession (p=.005) periods. Within-group analyses indicated that during periods of examiner toy possession, both groups spent more time making eye contact with the examiner and less time looking at the toy when it was inactive compared to when it was active (all p’s<.0001). They also engaged in more attention shifts when the toy was inactive compared to when it was active (ASD p=.-001; TD p<.0001). Children in both groups spent more time looking at the toy when it was in their possession vs. the examiner’s possession (p’s <.0001), with no effect of possession on amount of eye contact and rate of gaze shifts (p’s>.1).

Conclusions: These results confirm previous findings of decreased eye contact and lower rates of joint attention gaze shifts in children with ASD, but indicate that temporal context matters for the pattern of group differences observed. Importantly, our analyses revealed that children with ASD modulated their gaze across toy active/inactive and child/examiner possession periods similarly to TD children, indicating they were sensitive to the temporal context of the interaction. These results have implications for measurement of social and object attention in research studies aiming to characterize early behavioral markers of ASD, and for treatment outcome measures.