Abstract
Because reflectance is confounded with illuminance in the retinal image, illuminance must be somehow discounted. But the illumination level need not be known if the visual system can determine which surfaces share the same illumination level. Two surfaces that are coplanar, adjacent, and separated by a sharp boundary support a strong inference of common illumination. But the inference is weakened as the two surfaces are separated by a gap, even though coplanar, because an illumination boundary might fall within the gap. But if one surface completely surrounds the other, even if separated by a gap, their illumination will be the same except in the unlikely event that the illumination boundary just fits within the gap. We tested this idea using the perpendicular planes stimulus of Gilchrist (1977). First, each 4cm target was moved 2cm laterally within its plane so that it appeared to float in space. This destroyed most of the coplanar ratio effect. Next, the large square coplanar neighbor of each target was extended so that the target was surrounded by a 1.7 cm square border, with a 2 cm gap between target and border. This produced a coplanar ratio effect half as strong as that produced by adjacency, but only for the target in the highly-illuminated plane. A further experiment with articulated planes showed no difference whether the highest luminance in the plane was adjacent to the target or three or more patches removed from the target. Many studies have shown that target lightness depends on proximity to the highest luminance when these are separated by a gap. Our results and others show that proximity plays little or no role when the space between target and highest luminance is filled with other surfaces, presumably because any intervening illumination boundary would be revealed by these mediating surfaces.
NSF (BCS-0643827) NIH (5R25GM060826).