Infants’ sensitivity to the predictability of exchanged actions in socially contingent exchanges: the contingency cube

Previous studies on social contingency have revealed that young infants are sensitive to the consistent temporal pairing of contingent responses and can draw various, evolutionarily relevant inferences from it. For example, when infants identify that it is highly probable for an unfamiliar entity to produce temporally contingent responses to another agent's actions, they can infer that the reactive entity is an intentional agent, even if it shows no other cues of agency. However, the consistent temporal pairing of contingently exchanged actions is not sufficient to account for the full range of inferences induced by contingent social reactivity. Based on recent findings, I argue that recognizing socially contingent interactions relies on the sensitivity to the predictability of subsequent actions along three dimensions. I propose that young infants monitor 'what', 'when' and 'how' response actions are produced. By tracking these three factors, infants can represent the relatedness of subsequent actions exhibited by interacting social partners, guiding their contingency-based inferences, including those that have traditionally been viewed as independent of social contingency. I conjecture that the cognitive mechanism to monitor the predictability of exchanged actions may have become further specialized during human evolution, enabling social partners to engage in complex cooperative and communicative interactions. This article is part of the theme issue 'Mechanisms of learning from social interaction'.