Action Prediction in Children with Autism Spectrum Disorder

Background:  Predicting others’ action goals is a basic social skill. Predictive eye-movements in action observation have been linked to the Mirror Neuron System (MNS). Very few studies have investigated predictive eye movements in ASD. Whether the MNS is matching observed actions to motor plans as efficiently in Autism Spectrum Disorder (ASD) as in neurotypical individuals is a matter of debate. In particular, it has been argued that children with ASD fail to activate a representation of the final state of an action sequence during its initial acts.

Objectives:  The main aim of the study was to investigate whether prediction of an initial act is influenced by the ambiguity of the final goal of the sequence.  It was hypothesized that this would be the case in neurotypical, but not in ASD. In addition, we wanted to illuminate the relationship between predictive gaze performance and the level autistic symptoms and adaptive functioning.

Methods:  Eye-tracking was used to measure gaze as children (6-year-olds) with Autistic Disorder (AD, n = 40), Pervasive Developmental Disorder (PDD, n = 25) and typical development (TD, n = 21) looked at an action sequence presented on a computer monitor. The sequence consisted of a two-step action: pick up object and move it to a final location. While keeping movement parameters and initial goal constant, we changed the final goal of the sequence. In one condition, the final goal was ambiguous, and in another it was unambiguous. In a third control condition, the final goal was occluded.

Results:  For the reach-to-grasp actions toward the object, there was no group difference and no effect of condition. During the latter phase, when the object was moved to the final goal, there was an effect of condition, but not of group. This effect could be explained by reactive gaze in the occluded goal condition. The groups predicted the ambiguous and unambiguous actions to a similar degree, but substantial individual differences were observed in the AD group. Children with AD who arrived with their gaze at the goal after the arrival of the hand (reactive performance) had lower levels of adaptive functioning than their ‘predictive’ peers, but these groups were equal in terms of the level of autistic symptoms.

Conclusions:  In line with a previous small sample study, the present data suggest that on a group level, children with ASD use similar eye-movements in action observation as neurotypical individuals. That ambiguous and unambiguous actions elicited predictive eye movements to a similar existent suggests that prediction in this context reflects a bottom-up process activated by the sight of a hand approaching a goal object, irrespective of the congruency between the initial act and the final goal. Importantly, the results suggest that prediction is more related to the level of adaptive functioning than to autism specific symptomatology.