Specificity of Action Model Formation Deficits in Autism and Their Relationship to Social and Motor Impairments

Background: Internal action models, or sensorimotor programs that form the brain basis for a range of skilled behavior and for understanding others' actions, are compromised in autism and may impact social development in affected individuals.   

Objectives: Determine the specificity of deficits in action model formation in autism relative to ADHD, and examine this deficit as a putative mechanism underlying social dysfunction in autism.  

Methods: Motor adaptation was examined in 23 children with autism and two control groups: 20 TD children and 17 children with ADHD.  Participants learned to control a robotic arm while making arm movements in which the robot perturbed their movements by producing a velocity-dependent force field perpendicular to the direction of motion. In this task, the typically developing 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 learning from the trained movements to novel movements. The training took place in the left workspace (Target 1), and we quantified generalization to novel movements in the right workspace matching the intrinsic coordinates of the arm (Target 3; identical joint rotations as Target 1), and extrinsic coordinates of the task (Target 2; identical hand motion as Target 1). Movements to Targets 2 and 3 were always made in 'error-clamp' trials, in which the robot artificially eliminated movement errors, but allowed for measurement of force output from the hand.    

Results: The adaptive learning for Target 1 was indistinguishable across the three groups of subjects (P = 0.44). However, generalization patterns were markedly different (p<0.001). Children with autism generalized joint rotations to a greater degree than TD children (p<0.001), whereas this generalization was not distinguishable between TD and ADHD (p=0.29). The difference between autism and ADHD was close to significant (p=0.06); the autism group showed greater generalization of the joint rotation than the ADHD group. These results suggest that children with autism built a motor memory that more strongly relied on proprioceptive coordinates than did TD children (and, to some degree, than did ADHD children).  Over-reliance on proprioceptive coordinates in autism was related to impairments in social function on the ADOS-G (p<.02) and motor imitation (p<0.05) and basic motor control as measured using the PANESS (p<0.005).   

Conclusions: Findings suggest a specific pattern of altered motor learning associated with autism.  When learning a novel movement, children with autism show a bias toward reliance on proprioceptive, rather than visual feedback that was not observed in a clinical control group of children with ADHD.   Furthermore, this anomalous pattern of motor learning was found to robustly correlate with measures of social and motor dysfunction, suggesting that compromised action model formation may contribute to impaired development of social (as well as motor) capacity in autism. This line of study can lead to important advances in understanding the neural basis of autism and, critically, can be used to guide effective therapies targeted at improving social, communicative, and motor function.