Abstract
Previous study had reported that collinearly placed short line segments which had border-ownership attribute (i.e., figural side) could easily be integrated into a long contour when they had the same ownership polarity, but they could hardly be integrated when their ownership polarities spatially alternate (Kikuchi & Oguni, ECVP2005). This study performed psychophysical experiments investigating the perception of the closed contours represented by such small line segments with border-ownership attribute. The experimental method was based on path-paradigm (Field, Hayes, & Hess, Vision Research, 1993). Instead of Gabor patches, small equilateral triangles were used as stimulus elements, as used by the previous study by Kikuchi & Oguni. A path was composed of twenty triangles, embedded in background (randomly placed triangles in 16×16 grids). Each path-triangle participated in the global contour so that one of the three edges in the triangle was on a global smooth curve. All paths had circular shape. Unlike the preceding study (Kovacs & Julesz, PNAS, 1993), two types of closed contours can be defined in our experiment according to the ownership polarities. Type I: global ownership polarity is inward of the circle, but local ownership polarity is outward, corresponding to “hole”. Type II: both global and local ownership polarities are inward of the circle. Each trial in our experiment had two successive stimulus screens, and only one of them had the path. Subjects' task was to answer which screen included the path. We compared the correct rates of path-detection for above two types of contours, and obtained the result that type II contours were salient but type I were not, for all three subjects (p[[lt]]0.01). This result indicates that there seems interaction between local and global ownership signals on global closed contours, and that salient perception of the contours needs the coincidence of global and local ownership polarities.
Supported in part by a Grant-in-Aid #19700648 from MEXT Japan.