PERCEPTION 



1651 



fig. 32. A distorted room (con- 

 structed by Ames) which induces 

 apparent distortions of size. The 

 window on the left is actually 

 much more distant than the 

 one on the right, i.e. the floor 

 plan of the room is not rec- 

 tangular. As noted in the text, 

 it can be argued that the illusion 

 merely shows failure of size con- 

 stancy under conditions of 're- 

 duction.' [From Lawrence (31 1).] 



simple terms of contrast; the areas involved in these 

 experiments differ too much. However, one can 

 consider these demonstrations as proof for the crucial 

 role of boundaries (or contours, as discussed above on 

 p. 1605), and as evidence of the trigger action of the 

 introduction of boundaries between areas of high ami 

 low luminance (in analogy u> the demonstration in 

 fig. 4, above). Undoubtedly, main additional factors 

 need lo be considered for these and other constancy 

 effects. Size and shape constancy, in particular, have 

 been shown to vary considerably with the instructions 

 given to the observer or with the 'set' he adopts in 

 looking at the scene. 



ROLE OF INSTRUCTION AND EXPERIMENTAL SETTINC. 



The relative success of explicit instruction or de- 

 liberately adopted attitudes (159) in diminishing 

 constancies probably differs for different constancy 

 effects: loudness judgments for speech (at varying 

 distances) show nearly complete constancy, in spite of 

 instructions to judge the loudness 'at the ear,' and 

 not 'at the source' (350); for pure tones or noise, 

 loudness constancy is considerably less. Even for the 

 latter stimuli, constancy increases if the intervals 

 between standard and comparison sounds are pro- 

 longed, just as visual size constancy seems favored 

 when the angular separation of standard and variable 

 object is increased (243). 



effects of 'reduction.' The simplest, yet most 

 effective way of diminishing constancy effects is the 

 use of the 'reduction screen' (249, 251) or of equiv- 



alent procedures. The use of reduction can be shown, 

 for instance, in the experiments on brightness con- 

 stancy introduced by Kalz (249); two color mixers 

 (wheels with adjustable black and white sectors) are 

 placed side by side, separated by a shadow-caster. 

 II an observer is asked to adjust the black-white 

 mixture of the illuminated color wheel so that it 

 matches the brightness ol the shaded wheel, he will 

 add some black to obtain a match but not as much 

 l.\ far as would be required by the (lower) photo- 

 metric luminance of the shaded wheel If he now 

 looks at each of the wheels through a "reduction 

 screen' (an opaque screen with a hole), the match 

 breaks down and the gra) ol the shaded wheel looks 

 much darker than that of the illuminated one. Reduc- 

 tion can be accomplished similarly by removing 

 accessor) objects hum the field of view, bymonocular 

 observation, or, as we have seen, by photography. 

 The famous size illusion in Ames' distorted room 

 (8, 236; see fig 32) is in fact a special instance of 

 destruction of size constancy under conditions of 

 extreme 'reduction'; the actual room is not rectangu- 

 lar but shaped so that the far wall slants away to the 

 left. As a result, the window at the left is at a greater 

 distance than that on the right. To obtain the illusion 

 of a discrepancy in the size of the heads appearing 

 in the two windows, the observer must see no micro- 

 structure (which would yield a gradient on the far 

 wall), must use only one eye and must avoid eye or 

 head movements. The same "reduction' is of course 

 accomplished by photographing the room; hence 

 the fact that the illusion is obtained in the photograph 



