The effectiveness of the visual system is permitted by a complex processing network, which relies on cortical, sub-cortical and peripheral structures. The purpose of this research is to improve the knowledge of the process sustaining the visual perception of motion, and to produce a computational model able to reproduce the features of human visual tracking of a moving object.This work includes an extensive bibliographic research, and a series of experiments. The thesis consists of two parts:The first part pertains to the determination of performance in the "backward" inference of a partially visible movement. It consist of defining the involvement of exogenous information (retinal signals) and endogenous information (internal models of observed action) in the kinematic reconstruction of a partially hidden trajectory of a moving target. Our results support the hypothesis that the CNS adopts a mechanism based on the use of internal models in the reconstruction of past biological kinematics.The second part complements the first one, and aims to identify the structure and the functional characteristics of the tracking system; it also aims to understand the origin of systematic errors present in the location of a target, in humans.We developed a computational model in Matlab, based on the extrapolation mechanism of movement, which is capable of reproducing the experimental data for the localization task
