Abstract
In the aperture capture illusion (ACI), a vertical bar moves horizontally with its top and bottom appearing sequentially through two vertically separated and horizontally offset apertures (Palmer & Kellman, VSS, 2001). Observers incorrectly perceive the bar as misaligned in the direction of the apertures, and the magnitude of the illusion grows with velocity. Palmer et al. (VSS, 2002) suggested that the visual system continues to represent object parts while occluded but that the velocity of occluded parts is underestimated. Faces (and other objects of expertise) are thought to be processed configurally, and subjects have difficulty selectively attending to half of a face composite, especially when it is aligned the other half (Young et al., 1987). Here, we ask whether such configural and holistic biases for upright faces would reduce the magnitude of the ACI. We compared the ACI for a moving upright face to each of a number of control stimuli matched in size, including a grey square, a textured square, a scrambled face and an inverted face. The stimuli moved behind a black occluder and appeared sequentially through offset apertures that were each 80% of the width of the stimuli. Observers compared the relative magnitude of the ACI for pairs of stimuli moving at the same velocity, in sequentially presented displays. The ACI was stronger for an upright face than for the control stimuli, for most observers tested. Because the ACI appears stronger with inverted than upright faces, it is possible that configural processing is responsible for the reduction in the illusion and thus, survives occlusion. Future work will explore whether the inversion effect on the ACI occurs for any object with a strong canonical orientation or is systematically related to the face inversion effect typically measured on static unoccluded displays.