Automated Measurement of Head Movement Coordination in Infant-Parent Dyads and Later ASD Outcomes

Friday, May 12, 2017: 10:00 AM-1:40 PM
Golden Gate Ballroom (Marriott Marquis Hotel)
K. B. Martin1, D. S. Messinger2, Z. Hammal3 and J. F. Cohn4, (1)Psychology, University of Miami, Coral Gables, FL, (2)University of Miami, Coral Gables, FL, (3)Carnegie Mellon University, Robotics Institute, Pittsburgh, PA, (4)Psychology, University of Pittsburgh, Pittsburgh, PA

The Face-to-Face/Still-face (FFSF) is an early index of social-emotional functioning, which assesses infant and parent abilities to coordinate emotion and arousal states. Early atypicalities during social interactions, such as unusual head movements, may contribute to the social impairments that characterize children with autism spectrum disorder (ASD). Symptoms of ASD may alter the coordination of head movements between infant and parent during interactions, disrupting successful communication and social interactions. The quantity and speed of head movement—pitch (nods), yaw (turns), and roll (lateral inclinations)—between infants and parents may be more or less coordinated over early interactions in ASD. Compared to human coding, automated measurement can objectively quantify movement behaviors in infants with and without ASD risk. Our group recently found that older children with and without ASD systematically differ in their head movements in response to social stimuli.


Examine whether coordination of the quantity of infant and parent head movement in the FFSF differs by ASD risk and outcome.


Infant-parent dyads (N=64) completed the FFSF at 6 months. Dyads included infants with (high-risk) and without (low-risk) an older sibling with ASD. Three degrees of rigid head movement—pitch, yaw, roll— were tracked from the video-recordings using an automated, computer-vision approach (Zface; Jeni, Cohn, & Kanade, 2015). Head angles were quantified as angular displacement and angular velocity. Root mean square values were calculated to measure the magnitude of angular displacement (quantity) and angular velocity (speed) of head movement. At 36 months, infants were assessed for ASD by a clinician, informed by the Autism Diagnostic Observation Schedule. We assessed differences in head movement coordination between infants and parents in low-risk children (Low-Risk/No-ASD, n=22), high-risk children without ASD (High-Risk/No-ASD, n=31), and high-risk children with ASD (High-Risk/ASD, n=10). Analyses were within-group, zero-order correlations of the total quantity of infant and parent displacement and velocity of head movement in the FFSF. R-to-Z transformations were used to compare correlations between groups.


Infant-parent correlations of angular displacement tended to be greater for the High-Risk/ASD group than the High-Risk/No-ASD (ps<.09) and the Low-Risk/ASD groups (ps<.09) for pitch, yaw, and roll. Infant-parent correlations of angular velocity were greater for the High-Risk/ASD group than the High-Risk/No-ASD group for pitch and roll (ps<.01) and tended to be greater for yaw (p=.09). Infant-parent correlations of angular velocity were greater for the High-Risk/ASD group than the Low-Risk/ASD group (ps<.05; Table 1, Figure 1).


Here, automated measures of early interaction revealed that infants later diagnosed with ASD had the highest levels of coordination with their parents in the quantity and speed of head movement. For angular displacement and angular velocity, High-Risk/ASD infants and parents exhibited the highest associations of head movements. Low-Risk/No-ASD infants and parents exhibited the lowest associations of head movements. High-Risk/No-ASD infants and parents typically exhibited associations of head movements between the other two groups. While it is unlikely that infants match their parents’ head movements, parents of High-Risk/ASD infants may match those of their infants more vigilantly than parents of infants in other groups (Beebe, 2011).