Abstract
Recently we have shown that sensitivity for detecting oriented content in visual noise or natural scenes is worst at horizontal and best at the obliques (i.e., the “horizontal effect”). We suggested that a bias in the number of striate units tuned to different orientations (horizontal>vertical>obliques) leads to anisotropic contrast gain control and the anisotropy observed with broadband patterns. Since typical natural scenes possess more oriented content at cardinal orientations relative to the obliques, one might presume that as we view such environments, the visual system is in an anisotropic state of adaptation, causing less sensitivity to horizontal and, to a lesser extent, vertical content. In order to determine if the horizontal effect results from such an adapted state, we measured sensitivity to increments of broadband oriented content under various states of adaptation. Stimuli consisted of visual noise patterns (random phase spectra) and had amplitude spectra with the characteristic1/f fall-off of natural scenes. The main adapting stimulus was a uniform luminance field. Other adapting stimuli were generated by making broadband amplitude increments at one of four orientations (0, 45, 90, or at 135deg) or all four orientations combined. Ability to detect the increments was measured using the method of constant stimuli in a 2AFC paradigm. Results from the uniform adapting field condition showed a clear horizontal effect, suggesting that the main source of the horizontal effect is an inherent bias arising from anisotropic contrast gain control pooling. The magnitude of single-orientation adaptation was similar for each of the four orientations, indicating an additional source contributing to the horizontal effect. Multiple-orientation adaptation yielded less suppression for horizontal orientations (compared to the single orientation condition) suggesting the existence of anisotropic ‘cross-orientation’ suppression with broadband visual stimuli.
This work was supported by a grant from the Office of Naval Research (grant # N00014-03-1-0224) and from the Kentucky Space Grant Consortium (KSGC — NASA EPSCOR).