Sensorimotor Adaptation Underpins Imitation Learning of Biological Motion Kinematics in Autism Spectrum Disorders

Background:  Internal action models underpinning social interaction are developed by imitating biological motion. Although processes subserving automatic imitation of biological motion are functional in autism spectrum (autism) disorders (Sowden et al., in press), they are suggested to be compromised during voluntary imitation (Williams et al., 2004). This is said to be based on processing difficulties associated with integrating sensorimotor information across repeated trials of imitating biological motion. To this end, we examined sensorimotor integration and adaptation using a novel methodology that displayed atypical and typical biological motion in a random (control) and fixed (experimental) trial order. 

Objectives:  (1) examine imitation of biological kinematics; (2) examine whether imitation of biological kinematics is facilitated by promoting sensorimotor integration

Methods:  Twenty one (neurotypical) adults, and twenty one adults with autism, diagnosed by a clinical assessment and ADOS, participated in the study, which was approved by the local ethics committee. Participants imitated atypical and typical biological motion. During pre-test and post-test, atypical and typical stimuli were presented in a random trial order. During acquisition, the stimuli were presented in a fixed predictable trial order, which was counterbalanced between participants. Orthogonal planned comparisons were used to answer a number of apriori questions. Alpha was set at p < 0.05.

Results:  The neurotypical control group imitated biological motion accurately in all phases of the experiment. The autism group imitated atypical biological motion more accurately during acquisition when stimuli were presented in a fixed order, compared to the pre-test when the trial order was random (p < 0.05). Moreover, they imitated atypical motion more accurately across acquisition blocks (p < 0.05), thus demonstrating sensorimotor adaptation. Learning was confirmed with the autism group imitating atypical motion more accurately in the post-test compared to the pre-test (p < 0.05). 

Conclusions:  Consistent with previous work (Hayes et al., 2015), imitation of atypical biological motion was impaired in autism when stimuli were presented in a random order in the pre-test. When processing was facilitated using a fixed trial order, adults with autism demonstrated sensorimotor adaption resulting in high fidelity imitation of atypical biological motion. The ability to imitate atypical biological motion remained when stimuli were subsequently presented in a random order in the post-test. This persistence demonstrated sensorimotor learning. We suggest the fixed trial order removed the requirement to upregulate and downregulate alternative action models relating to the different kinematic properties of the observed atypical and typical stimuli. Thus, sensorimotor integration and consolidation of the representation of atypical kinematics was likely facilitated through repeated trial-to-trial sensorimotor transformations of the inverse (motor plan) and feedforward (efferent copy) models. This process would have yielded further opportunity for trial and error learning by facilitating trial prediction, such that sensory visual input from the next atypical model would have be processed to further consolidate the developing representation. These adaptation and learning effects offer some of the most promising evidence against a core deficit in voluntary imitation of biological motion in autism.