The reviewer is right, this has been corrected in the new draft. Nevertheless, Koenderink's representation is still not reified in 3-D, but is a 2-D abstraction of 3-D surfaces, i.e. a 3-dimensional percept is supposedly encoded as a 2-dimensional Gaussian curvature map, which limits this representation to a single point in depth for every (x,y) map location. As such, it cannot in principle represent multiple transparent surfaces, for example, nor can it represent the perception of empty space between the viewer and the object, which is evident in subjective experience. Nor can it represent amodal completion of objects occluded by nearer objects, or the occluded far surface of objects.