1 626 



HANDBOOK OF PHYSIOLOGY 



NEUROPHYSIOLOGY III 



the enclosed object is seen as carrier of motion, 

 rather than its surround). 



kinetic: depth effects. A striking instance of the 

 close connection between depth and motion are the 

 kinetic depth effects (346, 519); a two-dimensional 

 shadow on a translucent screen may give the im- 

 pression of three-dimensionality if the shadow-casting 

 object behind the screen (e.g. a wire triangle) is 

 slowly rotated. To Wallach and his colleagues (520) 

 this effect is indicative of the influence of memory of 

 past impressions on present experience — a revival of 

 the classical distinction of sensation and perception. 

 In fact every instantaneous impression of the shadow 

 does look fl.u, but their succession provides, in the 

 terminology of the Gibsons (157), a continuous series 

 of perspective transformations which can form an 

 adequate stimulus correlate of both depth and 

 motion. They point out, further, that the face of a 

 solid object is usually given "not as a form, but as a 

 unique family of transformations" (157). 



These kinetic depth effects thus furnish particularly 

 convincing instances of the general rule that sequential 

 as well as instantaneous patterns are the important 

 bases of perception. It is no accident that constancy 

 of shape is very much enhanced when objects are 

 slowly rotated, as in the experiments by Langdon 

 (299). Perceptions of shape as well as depth and 

 motion are normally dependent on sets of trans- 

 formations, rather than instantaneous impressions of 

 static patterns presented to an immobile observer. 



binocular parallax. If all this is true, then the 

 most famous of all single 'cues' of depth — the one 

 resulting from binocular parallax- -is likewise de- 

 moted from its primary to an auxiliary role. Effects 

 of disparity, in binocular vision, may be merely a 

 special form of the gradients which Gibson has 

 described (1 54). Binocular fixation v ields two slightly 

 differing views of a single object; the difference 

 increases as the object approaches the eyes. To tin 

 earlier investigators of depth perception, it was the 

 trianguladon thus produced that seemed to be, if 

 not the only, then the most powerful source of per- 

 ceived depth (38, "ii, 154). This belief was but- 

 tressed by Wheatstone's discovery in 1833 (540) of 

 the stereoscope which proved that horizontal disparity 

 of two flat patterns presented one to each eve could 

 act as a single determinant of depth But reliance on 

 the stereoscope .is paradigm of depth makes one miss 



hall ol the problem. It might seem that all we need to 

 understand is how the nervous system combines the 



two retinal images' into a single view. Yet for a real 

 tridimensional scene, geometry of binocular vision 

 requires that the singleness of vision be restricted to 

 the points that arc fixated; all objects lying closer or 

 farther than the fixation point should appear double. 



Since the time of Aguilonius (3), the term 'horopter' 

 has been in use to describe the set of all points that 

 ought to be perceived as single in the monocular 

 Held of vision. The corollary of the horopter notion is 

 binocular diplopia for all points lying outside this 

 horopter surface. The normal binocular observer, 

 however, has consistently ignored these geometric 

 constructions; he sees a tridimensional scene in which 

 objects are fused throughout. Does he do this by 

 'suppressing' his double images? Some investigators 

 in the nineteenth century thought he did, but Hering 

 pointed out (199 1 that by suppressing his double 

 images, the perceiver would lose valuable information 

 about depth. Again for geometric reasons, all objects 

 King closer than the fixated one should appear in 

 double images that are crossed, while objects beyond 

 the fixated ones should appear as uncrossed double 

 images. But this observation raises only a further 

 paradox; since the perceiver cannot distinguish which 

 eye is obtaining which view, how can he tell near 

 from far, i.e. crossed from uncrossed disparity? 



It seems to us that all these paradoxes arise only 

 in an 'image theory' of perception. They disappear if 

 one considers instead the serial patterns of stimulation 

 with which binocular depth perception is correlated. 

 There is a gradient of increasing disparity which 

 stretches from any fixated object toward the horizon, 

 and there is a converse gradient which decreases 

 from maximal disparity near the eves towards the 

 fixated object where it becomes minimal [see Gibson 

 (154, pp. 100-108)]. A stepwise change in such 

 disparity gradients v ields an edge, just as it does on 

 monocular viewing ol gradients of texture (see above, 

 fig. 21). These discontinuities furnish the "empty 

 spaces' that are seen in such compelling fashion 

 between overlapping objects appearing at different 

 depths in a stereoscope. 



Consider the stereograms in the upper part of 

 figure 23. When combined into a single view in a 

 Stereoscope, the arrangement on top gives the im- 

 pression of a surface like a wall which slants to the 

 right, if (he horizontal gradient of disparitv increases 

 to the right. Interchanging right and left images 

 v ields a corresponding slant to the left A gradient of 

 disparitv that runs vertically up or down (as the one 

 obtained on combining the two lower stereograms in 

 the bottom half of fig 23 I provides a slant SUCh as that 



