Evidence for Specificity of Anomalous Motor Learning in Autism

Background: Differences in procedural (skill-based) learning appears highly relevant to autism, with several investigators positing that impaired development of social and communicative skills may be linked to anomalous formation of action models beginning early in infancy.  Consistent with this idea, there is increasing evidence that autism is associated with anomalous motor development, including impaired imitation and execution of goal-directed actions (“dyspraxia”).  One of the crucial steps in motor learning is for the brain to form internal models: a mapping between motor commands and the expected sensory feedback.  We recently reported (Haswell et al., 2009) that children with autism spectrum disorder (ASD) show an anomalous pattern of motor learning as compared to typically developing (TD) children, with excessive reliance on proprioceptive, rather than visual, feedback; furthermore, for children with ASD, this bias towards proprioceptive-based formation of action models was strongly correlated with motor, as well as social, skill impairment.  Critical questions remain, including whether this anomalous pattern of motor learning is specific to ASD.

Objectives: To explore the specificity of anomalous motor learning in ASD, by comparing formation and generalization of action models in children with ASD to that of another population with developmental motor impairments – children with Attention Deficit/Hyperactivity Disorder (ADHD).

Methods: We quantified the representation of internal models by measuring patterns of generalization in age, gender, and IQ-matched groups of 14 children with ASD (age 10.5±1.7), 10 with ADHD (age 11.0±1.7 years), and 13 TD children (age 10.4±1.8).  Subjects were trained to reach to the forward direction in left workspace while holding a robotic arm; the robotic arm produced a curl force field so that subjects had to learn to adapt their movements to hit the target (“Target 1”).    In this task, the brain builds an association between self-generated motor commands and the sensory consequences (visual and proprioceptive).  The strength of each association can be inferred by how the brain generalizes the errors from the trained movements to novel movements.  As such, we quantified generalization in the right workspace in the intrinsic (proprioceptive) coordinates of the arm (Target 3, identical joint rotations as compared to Target 1) and in the extrinsic (visual) coordinates of the task (Target 2, identical hand motion as compared to Target 1).

Results: Superficially, learning appeared similar across the groups, with 3-group ANOVA revealing no effect of diagnosis on adaption to the force when reaching to Target 1.  In contrast, 3-group ANOVA revealed a significant effect of diagnosis on generalization pattern (F(2,34)=8.89, p=0.0008) with follow-up two-group contrasts revealing that children with ASD showed significantly greater generalization in intrinsic versus extrinsic coordinates compared with ADHD (F(1,22)=9.99, p=0.0045) as well as TD (F(1,25)=55.7, p<0.0001 – as reported in Haswell et al., 2009), but that the pattern in TD and ADHD children did not significantly differ (F(1,21)=0.19, p=0.67).

Conclusions: Children with ASD show an anomalous pattern of generalization compared to children with ADHD, as well as TD children.  The findings provide evidence suggesting that the pattern of excessive reliance on proprioceptive, rather than visual, feedback during motor learning is specific to autism.