Infants distinguish shadows from markings

Purpose: To investigate the ability of seven-month-old infants to decompose the retinal image into a set of "intrinsic images" which capture scene characteristics such as reflectance, illumination and orientation. We investigated the development of the ability to distinguish between a dark region, caused by a cast shadow, and an equally dark region, created by a surface that reflects little light. Method: In order to measure this ability, we habituated infants to a truncated cube mounted in the center of a dark screen presented at the infant's eye level. The cube was illuminated by a slide projector such that it could be presented either bisected by shadow or entirely in bright light. Two cubes were created that varied in reflectance. One cube was half white and half dark gray while the other was all white. During habituation, the gray and while cube was presented with a shadow bisecting it. After the habituation criterion was reached, test trials were presented. Test trials consisted of the gray and white cube with the shadow removed and the cube with the shadow unchanged but with the gray surface removed. Additional subjects were run in a control study to provide baseline measures of the attention-holding properties of the test displays. Results: In an initial study, adults tended not to notice a change in the shadow, but they did notice the change in the reflectance of half of the cube. Infants, 28-32 weeks of age, showed evidence that they also ignored the shadow in that they looked longer at the display that changed in reflectance than at the display in which the shadow was removed. Conclusion: Infants discriminate between shadows and low reflectance surfaces and tend to ignore shadowed regions relative to regions that vary in reflectance.