Abstract

Summation thresholds were measured at 16° and 32° eccentricity in order to determine the spatial properties associated with the perception of blur in low-pass filtered images. Both real-world and sinusoidal images were processed by using one of a series of Gaussian integrating kernels whose variances differed. The images were either circular or annular and were centered at either 16° or 32°. Stimulus area was varied, although the annular stimuli were always 8° wide. The images were presented in pairs with one member of each pair being an unprocessed mirror image of the image under test. The data for both types of stimuli were qualitatively similar to each other and to those previously reported for foveally presented stimuli, in that blur threshold varied monotonically with spatial cut-off frequency. The data obtained for the annular stimuli were unlike those typically obtained with circular stimuli, however, in that they often showed a nonmonotonic relationship between threshold and stimulus area. A series of control studies eliminated edge effects and meridional anisotropies in form perception as explanations for the nonmonotonicities. It is concluded that the differences in the spatial summation of circular and annular stimuli reflect differences in the functional (and by implication, the cortical) representation of these two classes of stimuli.

© 1992 Optical Society of America