Resumen

CASTILLO ARCEO, David; ESCOBAR, Esaú; HERMOSILLO, Jorge  y  LARA, Bruno. Model of a Mirror Neuron System in an Artificial Autonomous Agent. Nova scientia [online]. 2013, vol.5, n.10, pp.51-72. ISSN 2007-0705.

The research presented here is based on the on going multi-disciplinary work in the cognitive sciences addressing topics such as Mirror Neurons, Behavior Recognition, Imitation and Cognitive Robotics. The work involves the design of a system, implemented on an Artificial Autonomous Agent, addressing a perspective grounded in Simulation Theory and based in computational models of Mirror Neuron Systems. As a base and first step, the agent learns associations between its movements and the sensory consequences these have on the world; once this knowledge forms part of its baggage, the agent is capable of imitating the movements of a second agent. The design of the proposed system is based on two assumptions: (1) The mirror neuron system seen as the coupling of Inverse and Forward Internal Models - being the latter, and its function as predictor, the hypothesized function of mirror neurons, (2) The basis for the recognition of other's behaviors is the ability of living things to link their own behaviors with behaviors performed by others by means of a common language developed during their experience on interacting with their environment. We present an experiment where an agent imitates a second one to proof whether the recognition of the other's behavior is possible from the adopted perspective. We believe that our experiment is a proof of concept and presents a very solid ground for further research.

Palabras llave : Mirror Neurons; Imitation; Internal Models; Sensorimotor Learning; Artificial Intelligence (AI).

        · resumen en Español     · texto en Español     · Español ( pdf )