Abstract
When two targets of equal luminance are seen against adjacent dark and light backgrounds, as in simultaneous contrast displays, does their lightness depend on the physical luminance of the two backgrounds, or on background lightness? We tested this question using the well-known Mach folded-card illusion. A folded gray card was placed in the convex (roof) position on a large sheet of glass and illuminated from one side, producing luminances of 3.2 and 0.5 ftL. For one group of observers who saw the card in the roof (convex) position, the two sides appeared differently illuminated but roughly the same in lightness. By the use of an artificial shadow cast onto a white surface located below the glass sheet and illusory linear perspective cut into the card, we caused a second group of observers to see the card in the book (concave) position and the two sides appeared as black and white. Equi-luminant target squares were added to each face of the folded card by cutting an aperture in the center of each face. Although from the observer viewpoint the apertures actually revealed sections of a common distant surface, they appeared as opaque chips lying on the two faces. These were matched for lightness using a Munsell chart. If background lightness were crucial, we would expect a larger contrast effect in the book condition. If background luminance were crucial then we would expect no difference between our conditions. In fact neither of these outcomes occurred. In the convex condition the targets appeared as Munsell 2.7 and 6.7, while in the concave condition they appeared as Munsell 5.7 and 5.9. In a separate study, observers voluntarily reversed the Mach card, matching the targets in each perceived orientation. The results were qualitatively the same, but with a reduced difference between conditions. Thus, target lightness depends, not on background lightness or background luminance, but on background illumination level.
NSF: BCS-0236701 NIH: BM 60826-02