Towards a Shared Action Code for Human-Robot Interaction

Recent experiments show that the observation of robots controlling arm and hand movements in order to reach and manipulate objects activates the human mirror system. This suggests the possibility of developing a basic action code which is shared between humans and robots. What are the central features of this code, and its functional roles in the complex processes of action generation and understanding? A computational framework for addressing these issues is presented and critically compared to extant computational models of mirror neurons.