Overimitation in Children with Autism Spectrum Conditions

Background:  Typically developing children frequently copy unnecessary actions with high fidelity, this is termed overimitation.  The degree to which children overimitate seems to be socially modulated (Over & Carpenter, 2012). Children with autism have documented social difficulties and may also have an imitation deficit. Therefore, it seems unlikely that children with autism will engage in overimitation. However, children with autism have previously been shown to faithfully imitate inefficient tool selection and use with novel objects (Nielsen & Hudry, 2010) but it is unclear whether this overimitation is due to social responsiveness in participants with autism or due to the causal reasoning demands which may confound the task.

Objectives: The present study tests whether children with autism spontaneously overimitate unnecessary actions when the causal reasoning demands of the overimitation task are reduced. In addition, the study also tests whether children with autism are able to explicitly differentiate necessary from unnecessary actions.

Methods:  27 children with autism, 27 chronologically age-matched (CA) and 27 verbal mental age-matched (VMA) controls took part. Each participant had the opportunity to overimitate on five trials. In each trial the participant watched a demonstrator complete a three step sequence of actions on a set of simple, familiar objects such as building a block tower or retrieving something from a plastic lunch box. Within each action sequence, two actions were necessary to achieve the goal and one was unnecessary. After viewing the demonstration, participants were instructed to complete the action goal as quickly as they could. Overimitation was recorded if the participant completed the unnecessary action. Following imitation, children were asked to rate how ‘sensible’ or how ‘silly’ each action in each sequence was.

Results:  All children were able to complete the action goal. However, children with autism overimitated unnecessary actions significantly less than both the CA- (p=0.001) and VMA- (p=0.02) matched controls. Furthermore, children with autism were less able to differentiate sensible and silly actions when asked explicitly (CA: p=0.0001, VMA: p=0.007).

Conclusions: These results show that children with autism accurately copy goal directed actions but do not overimitate in circumstances where typical children do.  This lack of overimitation means that children with autism miss out on a wealth of social learning opportunities that typical children exploit. Our data show an interesting dissociation between implicit and explicit levels of measurement on the task. On one hand, children with autism implicitly understand the rationality of the demonstrated actions as they selectively imitate only the necessary actions. However, they are unable to explicitly distinguish ‘sensible’ from ‘silly’ actions. The consequences of these results for autism and for models of overimitation will be discussed.