This thesis proposes a methodology for the design of man-machine interfaces by combining top-down and bottom-up processes in vision. From a computational perspective, we propose that the scientific-cognitive question of combining top-down and bottom-up knowledge is similar to the engineering…