Behavioral and Neural Basis of Anomalous Motor Learning in Autism

Background: Theory suggests that motor learning either underlies or parallels learning of social and communication skills, deficits of which are at the core of autism spectrum disorder (ASD).  Children with ASD demonstrate both hyper- and hypo-sensitivity to sensory stimuli, as well as generalized motor impairments.  Specifically, studies of motor learning have suggested an over-reliance on proprioception.  The cerebellum, which plays a key role in motor learning, is frequently implicated in ASD for abnormalities on both a cellular and gross level.  As such, cerebellar structure may be related to patterns of sensorimotor learning in children with ASD.

Objectives: Quantify how children with ASD learn from different sensory modalities of error and relate this to gross cerebellar anatomy.

Methods: Forty children, ages 8-12, participated in the task: 20 with ASD and 20 typically developing (TD) controls.  Groups were matched for gender (Fischer’s exact test, p=0.41), age (t(38)=2.0, p=0.18), intelligence (WISC-IV Perceptual Reasoning Index) (t(38)= -1.51, p=0.14), and Edinburgh Handedness score (t(38)=-0.64, p=0.52).  For the psychophysical study, children participated in a reaching task in which their movements were given either a visual or proprioceptive perturbation during the course of their reach.  Sensitivity to these errors was measured as the relative compensation that occurred on the subsequent trial.  Children also participated in an imaging study including a high resolution T1-weighted image using a 3T Philips Achieva.  The resulting image was parcellated using an automatic protocol to produce regional volumes, according to the atlas in Schmahmann (Schmahmann et. al. 2000).

Results:  We found that children with ASD showed an increased sensitivity to proprioceptive error (repeated measures ANOVA, significant effect of group (F(1,38)=5.0, p=0.032), field (F(1.1,40.4)=8.8,p=0.004) and group by field interaction (F(1.1,40.4)=4.3, p=0.043)) and a decreased sensitivity to visual error (repeated measures ANOVA, significant effect of group (F(1,38)=6.4,p=0.015), visual gain (F(1.1,42.0)=5.2,p=0.024), but no significant interaction (F(1.1,42.0)=1.7,p=0.202)) when compared to TD controls.  Further, the sensitivity to error in each modality was anti-correlated (R²=0.56, p < 0.001), such that as one expressed greater sensitivity to proprioceptive error, they showed less sensitivity to visual error, with the ASD group showing a cluster towards proprioceptive reliance.  Using a generalized linear model (GLM), we related this proprioceptive bias to the volume of the right anterior lobe and right lobule VI in the cerebellum.  The GLM had a significant omnibus result (likelihood ratio χ²(5)=13.6, p=0.018), compared to an intercept only model.  Further, we found a significant main effect of lobule VI volume (Wald χ²= 5.125, p=0.024), as well as a significant group by lobule VI volume interaction (Wald χ²= 6.495, p=0.011) on the proprioceptive bias.

Conclusions: Our results demonstrate that children with ASD show an abnormal pattern of motor learning with a bias towards proprioceptive rather than visual feedback, and that this pattern is related to the volume of specific regions of the cerebellum.  The nature of this abnormality in motor learning may underlie deficits in acquiring social and communication skills in autism and could provide opportunities for developing novel therapies for improving motor, social, and communicative skills in people with ASD.