Keven & Akins (K&A) have contributed an important new perspective to a growing body of literature that has challenged longstanding assumptions regarding the existence of innate imitative ability in infants (Jones Reference Jones1996; Oostenbroek et al. Reference Oostenbroek, Suddendorf, Nielsen, Redshaw, Kennedy-Costantini, Davis, Clark and Slaughter2016; Ray & Heyes Reference Ray and Heyes2011). We agree with their assertion that infant orofacial “gestures” likely do not in fact arise as a consequence of imitation of others’ actions. Nonetheless, evidence of imitation in slightly older children is indisputable, and rejection of a nativist interpretation necessitates adequate alternative explanations of such proficiency. We believe there is now a pressing need for consideration and investigation of such alterative accounts. In this commentary, we would specifically like to advocate the importance of understanding how early experience is involved in the development of imitative ability.

Early work by Piaget (Reference Piaget, Attegno and Hodgson1962) suggested imitation developed in a step-wise manner over the first 2 years, a finding now supported by more systematic research (Jones Reference Jones2007). Some theoretical models have proposed that this developmental trend is scaffolded by sensorimotor experience; for example, when an infant observes their own actions, or when a caregiver's actions correspond to those of the infant. According to this view, this experience builds associations between sensory and motor representations of the same action that will later facilitate imitation (Brass & Heyes Reference Brass and Heyes2005; Ray & Heyes Reference Ray and Heyes2011). It follows from these assumptions that the stereotypic actions discussed by K&A are likely to play a significant role in the development of these associations. Indeed, evidence suggests that caregivers frequently imitate infant action (Flynn et al. Reference Flynn, Masur and Eichorst2004; Pawlby Reference Pawlby and Schaffer1977), and stereotypies may increase during caregiver interaction (Thelen Reference Thelen1981a). Opportunities for caregiver imitation may, therefore, largely consist of rhythmic stereotypies. These imitative interactions could provide the infant with rich sensorimotor experience critical to the formation of associations which eventually underpin their own imitative capacities. Therefore, orofacial stereotypies in infancy might more accurately be viewed as inputs into, rather than outputs of, early imitation.

Our own work on action imitation in older children has found that the likely extent of sensorimotor experience of synchronous activity appears to be related to the automaticity of imitative responses (O'Sullivan et al., under review). Although such effects have yet to be identified in infants, we would expect that nascent imitative abilities would be similarly affected by the aggregated experience of correspondences between sensory and proprioceptive feedback associated with particular actions. To conclusively demonstrate that sensorimotor experience is a key component of imitative development, researchers must document naturally occurring sensorimotor experience and its implications for imitation throughout the first year. The study of behavioural stereotypies offers valuable groundwork that could inform the approach taken within such an enterprise.

A systematic developmental approach to the study of infant stereotypies and their environmental triggers could also considerably strengthen K&A's argument, as well as augmenting our understanding of the development of imitation. We are sympathetic towards the notion that apparently imitative responses in neonates may in fact be an artefact of stereotypic behaviour triggered by arousing stimuli. Thelen's (Reference Thelen1979; Reference Thelen1981a) observational studies of rhythmic stereotypies in infancy (cited in the target article) have provided important groundwork in this respect, identifying a wide range of stereotypies and some apparent releasers. However, to further elucidate the reasons for apparently imitative correspondences during infancy, experimental approaches (similar to those used in neonatal imitation studies) could be used to systematically investigate the responses elicited by a comprehensive suite of actions modelled by a caregiver (including, importantly, those identified as common rhythmic stereotypies with different peak postnatal frequencies). Arousal theories propose that correspondences between the actions of model and infant are found in studies of neonatal imitation because the model's actions tend to be arousing, and the infant's actions are developmental stereotypies which would increase in response to any arousing stimulus. If this is the case, it should follow that any stimulus known to release a particular stereotypy during a specific postnatal phase of development should also trigger the production of other stereotypies during other phases of development (i.e., in accordance with their documented peak production frequencies). Such evidence would corroborate the conclusion that correspondences are coincidental, rather than imitative responses. Nonetheless, assuming a role for experience in the development of imitation, one might also expect that, over the course of development, the specific actions produced by the infant should show an increasing tendency to match those of the model across the full range of behaviours (and taking into account expected production frequencies).

Finally, we would also like to highlight the implications of the argument in the target article, along with an experiential account of imitative ability, for understanding the social role of imitation. A rich literature has explored how imitation in children and adults is associated with affiliation towards imitators (Chartrand & Bargh Reference Chartrand and Bargh1999; Over et al. Reference Over, Carpenter, Spears and Gattis2013). It has been suggested, likely because of assumptions about the innateness of imitative ability, that this is an adaptive response which has been subject to evolutionary selection pressure (Lakin & Jefferis Reference Lakin and Jefferis2003). However, K&A have highlighted how behavioural stereotypies are linked with arousal in infants (Jones Reference Jones1996; Nagy & Molnar Reference Nagy and Molnar2004), and as proposed in the foregoing, these stereotypies may also provide opportunities for imitative interactions. Therefore, such interactions may further associate the experience of imitative correspondences within a prosocial context of positive arousal. To our knowledge, no research has yet examined how imitation becomes such a pervasively social phenomenon in childhood. However, as strong nativist views of imitative ability continue to be questioned, we must explore how an infant's social environment might facilitate the development of links between imitation and other experiences, including physiological and psychological states.

We hope that our reflections will encourage others to take an interest in the role of learning in the formation of flexible adaptive social behaviours, and to further consider how such processes are influenced by universal developmental phenomena such as behavioural stereotypies.