Reduced Anticipatory Responses during Dynamic Object Interactions in Autism

Background: The dynamic world presents challenges to those with autism. Previous studies indicate difficulties in navigating traffic and driving (Autism Wandering and Elopement Initiative, awaare.org; Feeley, 2010). First person accounts highlight difficulties with dynamic playground games (Robison, 2006). These reports are puzzling given that moving objects are not inherently aversive to those with autism. Indeed, many individuals with autism are especially drawn to moving objects, enjoy video games (Mazurek et al., 2012), repeatedly set objects in motion, and engage in visual stimming involving motion. What then underlies autistic individuals’ difficulties with dynamic objects?

Several studies have shown that basic motion detection and direction perception thresholds are largely unimpaired in autism (Bertone et al., 2003; Milne et al., 2005; Mottron et al., 2006). However, interacting with dynamic objects involves a crucial step beyond detection: anticipating where the moving object will be and adjusting motor movements to intercept/avoid the object. Computational systems for dynamic object tracking rely on predictive techniques such as Kalman filtering (Welch and Bishop, 2001; Gao et al., 2005). They demonstrate that a key consequence of impaired prediction is errors in online position estimation. In daily life, these errors would render difficult seemingly straightforward tasks like catching an object or avoiding an oncoming car, even when basic motor skills are seemingly adequately developed.

Objectives: This study tested whether the difficulties that autistic individuals face in interacting with dynamic objects may arise from impairments in prediction. The task of ball-catching, combined with high speed video recording and computational analysis, were employed to reveal differences in predictive abilities with dynamic objects.

Methods: High-speed video recordings were captured of neurotypical and autistic children performing ball-catching. Throws varied systematically in distance from the participants, sometimes eliciting reaching, thereby testing participants abilities in anticipating the ball’s trajectory. Three observers reviewed the videos and assigned a rating to the participant’s predictive abilities. Additionally, computer vision techniques were applied to extract flow fields from each video, providing a detailed, quantifiable trajectory for both the ball and the child’s hands.

Results: Initial individual data suggest that autistic participants exhibit reduced accuracy of spatio-temporal relations between the hands and ball, compared to age-matched controls. Anticipation scores for ASD participants were lower on average than those of neurotypicals. The patterns seen in a small number of detailed computer vision tracings of the hand and ball quantify increased accuracy in prediction in NT participants relative to ASD participants (Fig 1).

Conclusions: This study is consistent with impaired spatial-motor anticipation in ASD. Results mesh with a broader hypothesis of impaired prediction in autism, and may help to explain broader impairment in interactions with dynamic objects in ASD. Future experiments will investigate variations on the features of dynamic stimuli, such as speed, size, and task demands. Future work will evaluate differences in other dynamic object tasks and correlate performance with features of the autism phenotype, as well as examine specificity to autism. Results could translate into development of effective safety skills and other dynamic tasks that autistic individuals encounter in everyday life.