Predictive Motor Abilities in Children with Autism Spectrum Disorder: Evidence from Kinematics and Muscle Activity

Background: Motor clumsiness has often been reported in individuals with autism spectrum disorder (ASD). However, as findings highly depend on the testbeds and inclusion/exclusion criteria, no common denominators or causal underpinnings of motor impairments in ASD have been identified. A recent theoretical framework from our group suggests that the seemingly distinct manifestations of ASD in multiple domains may share a common core: an impaired ability to make predictions. Difficulties in language, cognition, social interactions, repetitive behaviors, and motor coordination may all result from the same underlying impairment, albeit at different time scales. This study thus examined motor impairments through the lens of predictive control.

Objectives: We tested the hypothesis that individuals with ASD show reduced motor coordination when interacting with moving objects, specifically when performing actions that involve prediction. We examined whether impairments depended on the degree of prediction challenges in ASD.

Methods: Sixteen children with ASD (aged 7-12 years) and 26 age- and IQ-matched neurotypical (NT) children participated in a series of motor tasks involving catching in both naturalistic and virtual environments. While performing naturalistic ball catching, 3D joint kinematics and muscular activities in participants’ arm, leg, and trunk were recorded to quantify predictive motor behaviors. A set of control tasks assessed postural balance, reaction time and movement speeds when reaching to a static ball. The same individuals performed virtual interception tasks that simplified the coordination challenges and also afforded manipulating the time window for prediction. Catching accuracy was evaluated by temporal and spatial errors. A control virtual task was designed to be analogous to the experimental tasks when assessing reactive action.

Results: In naturalistic ball catching, ASDs showed a lower percentage of successful catches compared to NTs (p=0.01). While NTs approached the ball with a relatively stable profile of hand velocity, ASDs exhibited a collision-like motion at ball contact, suggesting insufficient prediction. Reduced predictive control in ASD was also evident in the muscles; ASDs showed peak muscular activity in trunk muscles at catching moment while NTs showed preparatory muscle activation prior to catching (p=0.02). Further, ASDs exhibited more pronounced co-activation of biceps and triceps, consistent with abrupt ball interception (p=0.01). In contrast, postural sway during quiet standing and reaction and movement time during the control tasks did not show difference between ASDs and NTs (p>0.3). In the virtual interception tasks, ASDs’ accuracy in catching indicated significantly worse performance, especially when occlusion of the ball trajectory enhanced the predictive challenge (p=0.02). In a control test of reaction time, ASDs did not differ from NTs, neither in reaction nor movement time (p>0.5).

Conclusions: Results from a suite of prediction-based tasks revealed that ASD children manifest reduced predictive motor control. Further, the impairment depended on the degree of predictive challenge. These findings suggest that the ability to predict may be a unified mechanism explaining phenotypic variations across ASD individuals. With this theoretical context, the results have implications beyond motor skills towards a more encompassing understanding of autism.