A Local Sign for Depth

WHEN a human observer views a tridimensionally ordered scene, each of his eyes receives a slightly different perspective view of the scene. Differences in tridimensional location are coded on to the retinae as differences in bidimensional position, that is to say, as retinal disparities. As is well known1, retinal disparity alone produces compelling perceived spatial separation. The way in which the nervous system decodes retinal disparity into perceived spatial separation is somewhat mysterious, since retinal disparity alone does not specify whether an object is nearer or farther than a fixated object, but merely that they are not in the same plane. (Disparities are geometrically ambiguous. For them to become unambiguous, additional information is required. Different authors have involved different sources of information, none of which is wholly satisfactory: see, for example, ref. 2).) The nervous system obviously can distinguish these two situations but the way in which it does this is not clear. Hering3 solved the problem by giving each point on the two retinae a specific depth-value. On this model all points on the nasal hemiretinae of each eye have positive depth-values, so that any point imaged there will be seen as further than the fixation point, while points on the temporal hemiretinae have negative depth-values, so that points imaged there will be seen as nearer than the fixation point. This simple and elegant solution has never gained wide acceptance, largely because Hering never specified any mechanism which could add depth-values to inputs. The specification of a mechanism to do that is as great a problem as the specification of a mechanism to decode retinal disparity into perceived spatial separation. It is no wonder that the majority of investigators felt that Hering had solved one mystery by another mystery.