Quantifying Imitative Behavior Deficits in Children with Autism Spectrum Disorder

Background: Individuals with Autism Spectrum Disorder (ASD) experience coordination and gait difficulties, slower speed in timed movements, and reduced balance and postural stability compared to healthy controls. However, the timing of this divergence from the typical developmental trajectory of motor abilities remains unclear. Given the role of imitation in early motor learning, failure to appropriately interpret and mimic another’s movements may produce downstream consequences in coordination, particularly with regard to gestures. Children with ASD have difficulty initiating and engaging in imitative behavior, and clinical measures that evaluate imitation rely on observation and categorical data of “yes” or “no” rather than quantifiable assessments of the degree to which movements are atypical.

Objectives: We aimed to assess the development of motor function and quantify deficits in imitative behavior in children with ASD.

Methods: A cross-sectional longitudinal design was used to investigate development of motor function in 10 children with ASD (ages 2–12). Evaluations of motor functions were conducted quarterly for one year and compared with those of 10 age- and gender-matched typically developing children. Children performed dynamic daily tasks such as pointing, reaching, balance, and walking as they interacted with virtual environments. They also imitated a humanoid robot as he performed gestures such as “waving hello/goodbye” and “fist bump”. We used kinematic data and a Dynamic Time Warping algorithm to quantify the spatial and temporal accuracy of imitative behaviors.

Results: Children with ASD had significantly higher variability in Center of Pressure compared to controls at all ages. Children with ASD consistently have significantly lower Froude numbers than age-matched controls. The Froude number (Fr) is directly proportional to the ratio between the kinetic energy and the gravitational potential energy needed during movement and is given by Fr = V²/gL, where V is the average speed of locomotion, g the acceleration of gravity, and L the leg length. These results suggest that development of an adult-like walking pattern characterized by an optimal energy transfer from cycle to cycle takes places at a slower rate in children with ASD. Children with ASD use significantly longer time to successfully point and reach a target resulting in longer “time per target” values (p<0.5). This suggests that the coordination of eye-arm movement in order to reach and point to target is delayed in children with ASD. The results of Dynamic Time Warping showed that children with ASD have poorer imitation behavior, as evidenced by higher discrepancy values of imitation based on weighted joint angle contributions, during the dynamic task compared to the control group.

Conclusions: Although improvements in motor function are evident in both groups with increasing age, the trajectories for development of balance, walking, and reaching have a different slope (i.e., they develop at a slower rate) compared to controls. During early childhood, specific motor and imitation impairments can serve as markers for screening/diagnosis of ASD. Given the potential influence of visual information processing on successful imitation, we are conducting a follow-up study using mobile eye tracking to assess visuomotor integration in ASD.