Illusory contour clarity decreases with retinal eccentricity. This effect could be the result of a loss of phase information in the periphery. To test this, we measured the relative clarity of edges with differing retinal eccentricities using an adaptive staircase that adjusted inducing element alignment, and compared that to subject’s misalignment detection threshold. In Experiment 1, a pair of illusory rectangles with circular inducing elements was shown along with a fixation point. While fixating, subjects judged the relatively clarity of the illusory rectangles' edges nearest fixation. The fixation point appeared in one of three locations, the center, 1.7 deg to the left, and 3.2 deg to the left (just outside the left rectangle). The vertical alignment of the inducing elements of the left illusory rectangle varied; misalignment increased if the subject indicated the left illusory edge was clearer than the right, and decreased otherwise. In Experiment 2, the same subjects viewed similar displays with fixations points in identical locations. The horizontal alignment of the inducing elements on the right illusory figure was varied, and subjects made 2AFC aligned-misaligned judgments. Generally, the nearer the illusory edge was to fixation, the more misalignment was needed to match the clarity of the two illusory edges. The misalignment needed to equate illusory contour clarity (mean PSE for most eccentric condition=1.2 deg) was considerably larger than the corresponding alignment threshold (mean for most eccentric condition = 0.7 deg). The tolerance for misalignment of inducing edges reported here is considerably greater than previous estimates (15 arc min) suggesting different limits apply for peripherally viewed and free-viewed illusory contours. These results confirm previous findings of decreasing illusory contour strength with increasing retinal eccentricity, and exclude a particular explanation based on loss of alignment information as eccentricity increases.

Funded By NEI Grant RO1EY13518 to PJK and TFS.