Learning to move, moving to learn: A quarter century of insights into infant motor development

Over the past quarter century, the field of infant motor development has undergone a profound conceptual shift from viewing motor behavior as a biologically preprogrammed sequence to understanding it as a dynamic, emergent process shaped by interaction, feedback, and prediction. This review traces that evolution across three key eras: the rise of Dynamic Systems Theory (DST) in the 2000s, which emphasized real-time coordination across bodily and environmental systems, the developmental cascades framework of the 2010s, which demonstrated how early motor milestones shape broader developmental trajectories, and the emergence of predictive, mechanistic models in the 2020 s, inspired by advances in artificial intelligence and robotics. Building on this trajectory, we propose a unifying framework termed Reinforcement from Sensorimotor Predictability (RSP, which posits that infants repeat actions not because they are goal-directed, but because those actions produce consistent and expected feedback. We present preliminary findings from a gaze-contingent eye-tracking study, along with a large-scale longitudinal project that applies machine learning to track sensorimotor trajectories in early infancy. Together, these lines of work suggest that predictability itself may serve as an intrinsic reinforcer, thus laying the groundwork for learning, agency, and the emergence of intentional behavior.