To get the grasp: seven-month-olds encode and selectively reproduce goal-directed grasping

Infants need to analyze human behavior in terms of goal-directed actions in order to form expectations about agents' rationality. There is converging evidence for goal encoding during the first year of life from looking time as well as social learning paradigms using imitation procedures. However, conceptual interpretations of these abilities are challenged by low-level motor resonance accounts that propose task-specific lower level sensorimotor associations underlying looking time tasks rather than abstract conceptual knowledge. To test the differential predictions derived from the two accounts, we investigated within-child consistency of performance on different, but conceptually related, tasks requiring goal encoding. This study presented seven-month-old infants with a looking time task and an imitation task, both testing their ability to encode an action goal based on a reaching action, as well as a working memory task to control for the influence of general cognitive capacity. Results showed inter task convergence to be independent of working memory: infants who spent more time looking at goal change events in the looking time task were more likely to selectively reproduce the goal in the imitation task when the model had performed an intentional grasping action rather than a back-of-hand object contact. These findings support the view that low-level motor resonance mechanisms are not sufficient to explain the capacities of action understanding in infants.