The ability to recognize animate agents based on their motion has been investigated in humans and animals alike. When the movements of multiple objects are interdependent, humans perceive the presence of social interactions and goal-directed behaviours. Here, we investigated how visually naive domestic chicks respond to agents whose motion was reciprocally contingent in space and time (i.e. the time and direction of motion of one object can be predicted from the time and direction of motion of another object). We presented a ‘social aggregation’ stimulus, in which three smaller discs repeatedly converged towards a bigger disc, moving in a manner resembling a mother hen and chicks (versus a control stimulus lacking such interactions). Remarkably, chicks preferred stimuli in which the timing of the motion of one object could not be predicted by that of other objects. This is the first demonstration of a sensitivity to the temporal relationships between the motion of different objects in naive animals, a trait that could be at the basis of the development of the perception of social interaction and goal-directed behaviours.