| This thesis examines the influence of shadows on the recognition of natural objects. Shadows introduce edges that can be confused with object contours, but they can also provide information about the shape and illumination of objects. The experiments in Chapter 2 examined the influence of shadows on recognition of digitized images of fruits and vegetables. Observers identified the images presented under a variety of "degraded" viewing conditions, including grayscale, blur, two-tone images, and brief exposure times. These manipulations were expected to make shadows more difficult to identify, and thus impair performance (accuracy and reaction time). However, the results revealed little if any effect of shadows. Chapter 3 examined the effects of shadows and illumination on face recognition. In one experiment, observers determined whether two sequentially-presented faces were the same or different people. The direction of illumination was either the same or different for the two faces, and the faces either did or did not contain shadows. Performance (sensitivity and reaction time) declined when illumination direction was changed, and was also worse when cast shadows were present. These results indicate that face representations maintain information present in the original image. A second experiment explored whether the same findings extend to longer-term representations. Observers learned eight faces by name under a single illumination and shadow condition, and later identified the faces under new illumination and shadow conditions. Performance declined when observers were tested under novel illumination conditions, but shadows had no differential effect. |
