Abstract
Robilotto, Khang, & Zaidi (2001) showed that for physically different filters on identical backgrounds, perceived transparency is predicted by both perceived contrast of the overlaid region and filter transmittance (proportion of incident light passing through the back surface). We now extend this study to filters on dissimilar backgrounds. Two Achromatic backgrounds differing either in contrast or mean luminance were generated on a CRT as random overlapping ellipses. Two neutral density filters were simulated as moving in circular motions, one on each background. The filters were generated based on two independent physical properties, reflectivity (proportion of light reflected at a change in media), and inner transmittance (proportion of light passing from front to back surface within the filter). The Standard filter had both physical properties fixed. The Matching filter had one property fixed while observers adjusted the other. In Exp. 1, observers adjusted the variable parameter to make the two filters appear equally transparent. In Exp. 2, the overlaid regions were rotated to abolish transparency cues, and observers adjusted the variable parameter to make the two overlaid regions appear equal in contrast. In all background conditions, observers were able to equate the perceived transparency of physically different filters through a linear tradeoff between reflectivity and inner transmittance. Equated transmittance was the physical determinant of equated perceived transparency and equated perceived contrast of the overlaid region was the sensory determinant. In conditions where Standard filters were presented over backgrounds of lower contrast or lower mean luminance, observers made settings so that the transmittance of the Matching filter was lower than for identical backgrounds. When identical filters were placed on dissimilar backgrounds, perceived transparency was lower on backgrounds of lower contrast or luminance, as was perceived contrast.