From imitation to emulation: A developmental framework for continual robot learning

Chaolan Lin

Keywords: Robot learning, developmental robotics, imitation, emulation

Abstract: Imitation and emulation are foundational mechanisms through which humans, particularly children, acquire knowledge and adapt to complex social environments. Developmental research shows that children initially engage in high-fidelity imitation, including the reproduction of causally irrelevant actions, but gradually shift toward emulation, selectively reproducing goal-relevant aspects of behavior. This trajectory, from surface-level copying to intentional, goal-directed understanding, offers a powerful model for continual robot learning. In this position paper, we propose that robots can benefit from a similarly staged learning process, beginning with broad imitation and advancing toward flexible emulation guided by sustained human interaction and feedback. Drawing on theoretical and empirical insights from developmental psychology, we define key learning stages and introduce benchmark tasks that assess fidelity, causal reasoning, generalization, and social alignment. By aligning robot learning with human cognitive trajectories, we aim to develop systems that are not only robust and adaptive but also interpretable and capable of evolving within human environments.

Submission Number: 16