Imitation of Atypical Biological Kinematics in Autistic Adults Is Associated with Sensorimotor Integration

Background: Novel sensorimotor actions are learned during voluntary imitation. Over repeated imitation attempts, a sensorimotor action model is configured by processes that represent the executed observed biological kinematics. Although voluntary imitation is functional from early development, autistic individuals generally show low-fidelity imitation of novel actions (DeMyer et al., 1972; Hayes et al., 2016). Notably, in these studies the imitation stimuli (e.g. typical and atypical kinematics) were presented in a random trial order. This non-predictable trial order was suggested to underpin low-fidelity imitation by attenuating the integration of sensorimotor information between trial n, and trial n+1. To this end, we attempted to influence imitation of biological kinematics in Experiment 1 by presenting stimuli in a constant trial order that is known to facilitate sensorimotor integration between trials. To further examine sensorimotor integration in Experiment 2, a visuomotor-interference task was introduced between trial n, and trial n+1, to interfere with inter-trial sensorimotor processing.

Objectives: Examine the contribution of sensorimotor integration processing during voluntary imitation in autism.

Methods: Autistic, and control, adults observed single point-light models moving horizontally on a monitor and imitated using a stylus on a digital graphics tablet. A control model contained typical biological kinematics and displayed a bell-shaped (peak velocity occurs at 44% of the movement trajectory) velocity profile. An experimental model contained atypical biological kinematics and displayed peak velocity that occurred earlier (18%) in the movement trajectory. In Experiment 1 (autistic, n = 15; control, n = 15) and 2 (autistic, n = 15; control, n = 15), the typical (30 trials) and atypical (30 trials) models were presented in a constant trial order, counterbalanced across participants. In Experiment 2, participants performed an additional incongruent (continuous circle drawing) visuomotor task in the inter-trial period.

Results: The discrete (time to peak velocity) and continuous (root mean square error) kinematic data in Experiment 1 indicated autistic adults significantly (ps < 0.05) adapted imitation by reducing error across the imitation trials. We showed no adaptation effect in Experiment 2 when autistic participants performed a secondary visuomotor interference task in the inter-trial period.

Conclusions: These data indicate functional voluntary imitation in autistic adults, and importantly that the reported deficits (DeMyer et al., 1972; Hayes et al., 2016) in voluntary imitation in autism may have been associated with an imitation context that prevented sensorimotor integration and adaptation across trials. The finding from the secondary task supports the suggestion that atypical imitation and motor learning in autism (Haswell et al., 2009) is associated with atypical sensorimotor integration.