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. a “horizontal effect”) and that this effect remains even with natural scene stimuli containing a natural bias in content at a given orientation. We suggested that this effect evolved to discount the dominant horizontal content typically present in natural scenes in order to ‘emphasize’ content at other orientations. Here we investigate this idea with natural scene imagery containing different amounts of horizontal content bias to determine if sensitivity for detecting amplitude increments at off-horizontal orientations depends on the relative predominance of horizontal content in a given image. Sets of natural scene imagery were compiled; each set contained images with a different predominance of horizontal content relative to the other image sets. Additional control sets contained images biased at other orientations. All images were made to have identical isotropic amplitude spectra, differing only in content carried by each image's phase spectrum. Test stimuli were generated by making increments at one of four orientations (0, 45, 90, or 135deg.). Ability to detect the oriented increments was measured with a single-interval Yes/No task. Consistent with our previous work, performance for detecting increments was poor for horizontal relative to the other orientations despite the content bias in the image. A significant positive relationship between the amount of horizontal content and sensitivity to increments at the other orientations was obtained. The horizontal effect may have evolved in accordance with the typical bias in oriented content in natural scenes, acting to emphasize less-dominant content as the current results show sensitivity for detecting off-horizontal amplitude increments is related to the relative amount of horizontal content present in the natural scene stimuli.