What is an Appropriate Theory of Imitation for a Robot Learner?

When faced with learning new skills or trying to solve problems humans sometimes have a choice, either learn the skill oneself (individual learning) or alternatively learn the skill from someone who already has that knowledge (social learning). Learning from another has the advantage of bypassing a long and costly search process and can therefore speed up learning by exploiting the knowledge of the other. It may also be the case that new and original solutions to problems may never be found when searching for them individually whereas readily available solutions may be found by exploiting the knowledge of another. Imitation is one mechanism which allows this knowledge transfer to take place. In this paper comparisons of selected theories underlying imitation and how these may be useful from a robotics viewpoint and especially a robot learner are presented. Many of these theories underlie existing robotic implementations and thus to understand the efficacy of these robot learners an understanding of the underlying concepts may prove useful and necessary.