1650 



HANDBI )i IK c ll I'MYSM ll 1 K . , 



NEt'ROl'HYSHiI ci( ,Y III 



are probablv based on diverse physiologic mecha- 

 nisms. This has been shown, for instance, in para- 

 metric studies by Leibowitz et al. (514) on effects of 

 tachistoscopic presentation upon constancy of shape, 

 size and brightness. The same variable, viz. briei 

 exposure, had differential effects on these three 

 measures. Short exposure diminished shape con- 

 stancy, had little effect on size constancy and en- 

 hanced brightness constancy. Similarly, tests of 

 shape and size constancy have different outcomes 

 when they are performed on photographs of the test 

 objects rather than on the test objects themselves; in 

 the photographs, shape constancy is reduced, while 

 size constancy is virtually abolished. The experienced 

 photographer takes these reductions into account 

 when planning an exposure, just as he attempts 

 (with varying success) to overcome his natural 

 tendency towards brightness constancy which leads 

 him, if unchecked, to overexpose his films at noon 

 and to underexpose them at dusk. 



RECENT WORK ON CONSTANCY OF COLOR AND BRIGHT- 

 NESS. The approach of Hcring (but without recourse 

 to memory colors) has been revived in the studies by 

 Helson (196), just as the tradition of von Helmholtz 

 continues in Brunswik's probabilistic approach (73, 

 74, 76) to constancy effects. Helson endeavors to show- 

 that adaptation and contrast suffice in explaining 

 chromatic and achromatic constancies. In any pat- 

 terned held, an "adaptation level' is established which 

 is the "weighted geometric mean of the reflectance 

 of all parts in the visual scene." Under colored il- 

 lumination, for instance in green light (as under the 

 foliage of trees), the (greenish) background will 

 contribute disproportionately to this adaptation 

 level. All surfaces that have reflectances above- 

 adaptation reflectance take the hue of the (comple- 

 mentary) afterimage. Surfaces near the adaptation 

 level are either seen as colorless or with very low 

 saturation. In this formulation, chromatic 'con- 

 stancy' is a consequence of adaptation (to the chro- 

 matic level) and of contrast which is, in turn, a 

 consequence ol gradients from the level. The adapta- 

 tion processes postulated here, however, need to be 



fastei than those ordinarily considered. 



Helson's work is in close contact with the modern 

 views on color vision represented bv Hurvich & 

 Jameson (231). In older view--, -black' is usuallv re- 

 garded .i- the one color which is due whollv to the 

 eve |n fact, however, .inv color, in am saturation, 



can lie evoked from positive or negative chroma ticity 

 gradients. In spectrall) homogeneous yellow il- 



lumination, "'samples of high reflectance are yellow, 

 while those of low reflectance are reddish-blue"; or 

 in monochromatic blue illumination, "samples of 

 high reflectance are blue and those of low reflectance 

 reddish-yellow. . . . The colors arising from either 

 positive or negative gradients are equally 'good,' the 

 latter appearing more saturated in strongly chromatic 

 illumination than the former" (iob). 29 



Attempts at reducing (achromatic) brightness con- 

 stancv to simultaneous contrast or, more generally, to 

 gradients of illumination in the visual field have 

 likewise been made [see Wallach (518) and Leibowitz 

 et al. (315)]. A complication in this area is introduced 

 by the fact that brightness contrast appears to be 

 primarily unidirectional; brighter objects depress the 

 brightness of less bright objects, but the dimmer 

 object seems to have little comparable effect on those 

 of higher luminance. In experiments with a gray 

 test object viewed over an illuminance range of 1 

 million to 1 against either "black," "white' or 'gray' 

 backgrounds (315), the major portion of the con- 

 stancy effect could be predicted from independently 

 derived contrast relationships. 



Doubts have been expressed, however, whether all 

 phenomena of brightness constancy can be thus ex- 

 plained (337). Several classic experiments reveal this 

 difficulty. In Gelb's experiment (145) a black disk is 

 shown in a dark room and illuminated bv a spotlight. 

 It appears white until a small piece of white paper is 

 brought near it; at that moment the disk turns 

 abruptlv black. In the converse experiment by 

 Kardos (248), a white disk is shadowed so that its 

 surround remains brilliantly lit. Such a disk looks 

 black until the shadow-caster is shifted so that a 

 penumbra becomes visible, in that instant, the disk 

 begins to look white, though shaded. This demonstra- 

 tion is essentially the same as the older ringed-shadow 

 experiment introduced bv Hering: .1 small object 

 i asts a shadow on a faintly illuminated white surface. 

 The shaded area looks as white as the rest of the 

 surface, though shaded. Draw a heavv black outline 

 around the shadow and the shaded region turns dark 

 gray. On obliteration of the penumbra, the shadow is 

 changed into .1 slain ( JOO, 202). 



I'hese lime experiments 1 bv Hering, (ielb and 

 Kardos) are often invoked as instances detracting 

 from an interpretation of brightness constancy in 



: ' I lirsr t.H ts lui in tli*- lusis dt the [[null |>ul ilicized dem- 

 onstrations mi color vision by Land (998) which illustrate 

 the principles implicit in 1 [ering*s experiments on colored 

 shadows (200, 202) .mil in tin- systematic quantitative work of 

 I lin \ ich & Jameson ' -■ ; 1 



