Building a Bridge between Mirror Neurons and Theory of Embodied Cognition

摘要

Mirror neurons are specialized neurons recently discovered in the brains of primates. In experiments mirror neurons showed activity both when a subject performed an action and when it observed the same action performed by self or another (possibly conspecific) subject. We formulate and study possible computational consequences of the hypothesis in which the experimentally observed properties of mirror neurons are generalized to other perceptive modalities and the underlying mechanism for coupling sensory and motor information is extended by an associative mechanism serving for completion of cross-modal information composed of perception and motor information. Depending on of what kind of information is completed, the hypothesis opens the door for understanding the mechanisms for sensorimotor coordination, imitation learning, and thinking and is inspiring for the design of such mechanisms in the case of artificial agents. Our results justify the hopes generally related to the discovery of mirror neurons.