There is renewed interest in the study of illusory contours, so much so that an international conference devoted exclusively to this phenomenon was recently held (see Petry & Meyer, 1986). Much of the mystery surrounding illusory contours concerns how they can be perceived in the absence of physical discontinuities in luminance. Indeed, a major theme of research on illusory contours is the specification of the relevant stimulus parameters in the luminance domain required for their formation. A good example of this approach is the recent work of Prazdny (1985Prazdny ( , 1986, who has argued that (1) luminance differences must be present in order for illusory contours to be perceived, and (2) differences in temporal correlation between the background and the inducing elements elicit the perception of illusory contours in random-dot cinematograms.In this paper, we report several observations relevant to this issue that cast doubt on the assertions that luminance-domain information or spatiotemporal differences are essential variables for the induction of illusory contours. In short, we have found that observers can readily perceive illusory contours induced solely by global stereoscopic forms under stimulus conditions in which neither monocular luminance differences nor motion discontinuities are present.To produce these forms, we used a hard-wired electronic system that generates, on modified color television receivers, dynamic matrices of randomly ordered red and green dots. All elements in the matrices are replaced randomly at a field rate of 60 Hz, resulting in apparent motion of the elements that resembles Brownian motion. Such motion, however, does not impair the perception of global stereoscopic forms, which can be presented for controlled durations and moved about quickly to virtually any position in stereoscopic space. A unique feature of the system is an image digitizer that converts any twodimensional achromatic shape into a global stereoscopic form of equivalent configuration. (For a more complete description of this system, see Lehmkuhle &Fox, 1980, andShetty, Brodersen, &Fox, 1979.) For our purposes, we presented to the digitizer inducing elements (Pac Man) arranged to produce simple Kanizsa-type illusory squares, rectangles, and triangles Support for this investigationwas provided by Grant EYOO590-22 from the National Institutes of Health. We would like to thank Susan Petry for her comments on a previous version of this manuscript. Requests for reprints should be addressed to Robert Fox, Department of Psychology, Vanderbilt University, Nashville, TN 37240. of varying sizes (i.e., 5.5 0 , 6 . 8°, 8 . 2°) . The stereoscopic displays were completely devoid of any monocular or brightness cues, and no motion or texture discontinuities were present at any time between the illusory figures and the inducing elements; the inducing elements were specified entirely by horizontal binocular disparity.We had 15 observers, 8 of whom were inexperienced with random element stereograms, view the resulting stereoscopic...