Abstract
Purpose. Filling-in regions have recently been shown to affect shape perception, but it is unclear how these regions gain their influence. Here, we consider the possibility that filling-in regions affect shape perception via lightness induction. Method. Subjects discriminated fat and thin noise-corrupted rectangles, the tops and bottoms of which were connected by illusory contours or luminance-defined contours. Inducers were either all black, all white, or both black and white (mixed). In each of these six conditions, the fat and thin shape alternatives were created by either rotating the inducer elements clockwise or counterclockwise by 6.4 deg (small rotation condition) or 12.8 deg (large rotation condition). A Quest staircase adjusted signal (inducer) contrast to yield 70% performance for each observer. A classification image (CI) technique revealed correlations between each pixel of noise and response. A region-of-interest (ROI) analysis was performed on the narrow vertical bands of CI pixels that exactly bisected the fat and thin contours on each side of the rectangles. These ROI regions never contained signal, and thus could only correspond to filled-in regions. Results. There were three main results. First, ROI pixels strongly affected shape responses in equal but opposite ways when inducers were all white or all black. Second, when inducers were of mixed polarity, ROI pixels did not affect performance. Finally, ROI pixels affected performance comparably in the small and large rotation conditions. Conclusions. These results suggest that a) filling-in regions influence shape perception, even when those regions are relatively far from relevant contours; b) the direction of this influence depends on lightness induction; and c) the magnitude of the influence does not drop off sharply with distance from shape boundary. These findings, taken together, indicate that lightness induction plays an important role in determining how filling-in regions alter the perception of shape.
This work was supported by a UCLA Graduate Research Fellowship awarded to BPK.