Kennedy and Bai (2000 Perception 29 399–408) argued incorrect border polarity blocked perception of faces in shape-from-shadow ‘Mooney faces’ with dark lines at the contour, a display inspired by Hering. Their hypothesis was tested with several displays, notably binocular gratings made of lines of dots. The stereo-induced depth involved a shadow falling on two surface planes. Most of a dark-dotted (shadow) region appeared to be on one surface, but a strip of dark dots at the shadow's border appeared to be on another—to the fore or rear. Control conditions involved ‘negative’ images (white dots). Subjects saw the shadowed object as easily in dark-dotted images with stereo depth as in an image with uniform depth for all the dots, and more readily than in the negatives. Our results favour the border-polarity hypothesis.