Integration of Visual and Proprioceptive Perception of Movement Kinematics in Children with and without Autism Spectrum Disorder

Background: Children with autism spectrum disorder (ASD) often exhibit deficits in sensory information processing, yet the nature and consequence of these deficits is unclear. An important goal of sensory information processing is the integration of information from multiple sensory modalities for perception of body and limb movements in space. Vision and proprioception are usually the two most informative modalities serving this purpose. According to current models of sensorimotor integration, if uncertainty increases in one modality (e.g. vision), the contribution of the other (proprioception) should increase.

Objectives: We used robotic technology and psychometric modeling to test the hypotheses that children with ASD exhibit deficits in multimodal sensory integration for motion perception relative to typically developing (TD) children (i.e. that children with ASD favor proprioception over vision to a greater extent than TD controls).

Methods: Ten high-functioning children with ASD and ten TD children performed a two-interval, forced choice, hand path curvature discrimination task. Seated participants grasped the handle of a robotic device that moved in the horizontal plane and / or watched a cursor representing the hand move on the screen (Figure 1a). Task instructions were presented on the screen (Figure 1b). We characterized the ability to discriminate between hand and / or visual cursor paths (both called endpoint paths) of differing curvature (Figure 1c) and to characterize the relative importance of vision vs. proprioception in the perception of endpoint motion. Children performed the task under single-modality (visual or proprioceptive stimuli) or bimodal conditions (simultaneous visual and proprioceptive stimuli). We additionally investigated how uncertainty in the visual stimuli might influence perception by varying the width of the cursor’s Gaussian spatial distribution on different trials (Figure 1d). We estimated two psychometric variables from standard cumulative Gaussian functions fitted to the task response data: the standard deviation of the underlying psychometric model (also called discrimination threshold) and the point of subjective equality (PSE). We used these variables to estimate the contributions of vision and proprioception to the perception of endpoint motion. The discrimination threshold is inversely related to the precision in which participants discriminate between curvatures.

Results: Children with ASD were more variable in their discrimination (i.e. they exhibited increased discrimination thresholds) during the bimodal condition as compared to TD children, and as compared to discrimination during the single-modality experiments (Figure 2a). By contrast, both groups of children exhibited systematic re-distribution of sensory weights such that visual dominance over proprioception decreased as the amount of visual uncertainty increased (Figure 2b).

Conclusions: We identified deficits in the precision in which children with ASD discriminate hand path curvatures when provided simultaneous visual and proprioceptive information. Contrary to our original hypothesis, we found no evidence for ASD-related deficits in the re-weighting of visual and proprioceptive information for perception of hand-path kinematics. Our findings suggest that the neural mechanisms underlying multisensory re-weighting remain intact in high-functioning children with ASD, whereas mechanisms underlying multisensory perception of movement kinematics are compromised.