Mapping the simulated output onto actual percepts is a good point, and so is the distinction between modal illusory contours as in the Kanizsa figure, and amodal grouping percepts. I address these issues in my MLRF paper.