This paper seeks to foster a discussion on whether experiments with robots can inform theory in infant motor development and specifically (1) how the interactions among the parts of a system, including the nervous and musculoskeletal systems and the forces acting on the body, induce organizational changes in the whole, and (2) how exploratory behaviour and selective informational signals at the timescale of skill learning may allow behaviour to become stabilized at the longer timescale of development. The paper describes how three generative principles, inspired from developmental biology and shown to underlie the dynamics of infants learning to bounce in a Jolly Jumper, were broken into a set of mechanisms suitable for controlling a robotic system and resulted in a similar developmental profile. A comparison of infant and robot data leads to a set of criteria for improving the usefulness of robotic studies. (Contains 4 figures and 1 table.)