Speaker: Giovanni Tessitore - VINELab - Università degli Studi di Napoli "Federico II"
Date: 06/10/2010, h. 10.00
Location: DISI - room 322 "sala conferenze"
Various computational models have been proposed to account for the behaviour of mirror neurons in the broader context of mirror system functionalities. Mirror neurons behaviour is usually modelled there as the outcome of computational processes unidirectionally flowing from sensory (and usually visual) input to mirror neuron activity. A different approach to the computational modelling of mirror neurons, inspired to the direct matching hypothesis, is presented in this talk. A central functional role is assigned to mirror mechanisms in visual perception, predicting that motor information coded by mirror neurons simplifies the processing of sensory inputs and improves the results of action recognition tasks. The following questions are specifically addressed: What is the information coded by mirror neuron activity? How does the information coded by mirror neurons bear on sensory processing? A computational account of the interaction between motor information and perceptual (visual) processes is provided within a probabilistic framework. Model behaviours observed in simulation experiments that are carried out in the context of reach-to-grasp actions are described and analyzed. These results show that visual sensory processing benefits in various experimental settings from the interaction with motor information.