Abstract
Purpose. The hollow-mask illusion, i.e., obtaining the percept of a regular convex face when viewing a hollow mask, is weaker when the mask is inverted [Hill & Bruce, Perception, 1993]. The difference in performance can be attributed to global, top-down influences, since bottom-up cues, such as disparity range, foreshortening, texture gradient, motion parallax and perspective, are essentially the same for upright and inverted stimuli. Can this inversion effect also be obtained with natural non-face stimuli?
Methods. To compare the role of top-down processes in face and non-face stimuli, we used a hollow mask and a Patrick Hughes reverspective portraying a scene, respectively. Reverspectives, like hollow masks, are perceived in reverse (illusory) depth [Wade & Hughes, Perception, 1999; Papathomas, VSS 2001]. Each of the two stimuli was presented upright and inverted, and viewed both monocularly and binocularly. The strength of the illusions was quantified by three measures: 1) The critical distance at which the percept changes from illusory to veridical; both “approach” and “retreat” conditions were used here. 2) The relative “residence time” of the illusory percept as a percentage of the total viewing time. 3) The initial percept obtained immediately after the observer views the stimulus.
Results. The pattern of results is similar for all three measures. Hysteresis was observed between the “approach” and “retreat” conditions [Hill & Bruce, Perception, 1993; Papathomas, VSS 2001]. The critical distance was smaller, hence the illusion was stronger, for the reverspective than the mask. The strength of both, facial and non-facial, illusions is reduced significantly when the stimuli are inverted. However, the reduction is larger in the case of the facial than the scene stimuli.
Conclusion. The results seem to indicate that top-down influences are stronger for facial than for non-facial stimuli. Further work is needed to characterize these top-down influences.