Abstract
Purpose. The luminance ratios of adjacent regions often stay constant across changes in illumination (Wallach, 1948). Chien (ARVO 2002) found that 4-month-old infants showed lightness constancy when a consistent local luminance ratio cue was present across changes in illumination. By manipulating the reflectance of the surround, we tested whether infant lightness constancy breaks down when the local luminance ratio cue was inconsistent across changes in illumination. Methods. The forced-choice novelty preference method (Chien, Palmer & Teller, in press) was used. Two incandescent illuminations differing by about a factor of 3 were used. The stimuli were light-grey (refl. 60%) and dark-grey (20%) papers, patterned as smiley faces, and were surrounded by either a white (90%) or a mid-grey (30%) surround. Three conditions were tested. In the “control” condition, both the surround and the illumination were unchanged between the study and the test phase. In the “change of surround” condition, the surround was changed from white to mid-grey but the illumination remained unchanged. In the “change of surround and illumination” condition, a change of illumination from low to high was further added. In the study phase of each trial, an infant was presented two identical faces with one surround and one illumination. In the test phase, the infant was presented one face that had the same reflectance and another face with a novel reflectance with a possibly different surround and illumination. The key was that, with a different test surround, the reflectance-matched face yielded an inconsistent (novel) luminance ratio, and this could cause a failure of lightness constancy. Results. In the control condition, infants showed preferences to the face with a novel reflectance. Neither the change of surround nor the change of illumination and surround conditions showed the same results. This suggests that constancy in infants depends at least in part on a local luminance ratio cue.