Abstract
When viewing a single image in the periphery, such as the number 2, accurate identification is fairly easy, but becomes more difficult with increasing eccentricity. However, if one takes that same image and flanks it to the left and right with other images, (e.g., the number 2 flanked by 5′s (525)), accurate identification becomes much more difficult (Bouma, 1970). It is said that the flanking images “crowd” the target image thus making identification more difficult. This phenomenon, known as crowding, has been documented since the early 1900s (Wagner, 1918), but the biological basis of this phenomenon has been difficult to identify. In order to explore the neural basis of crowding, we used shape-from-shading gradients. These stimuli were chosen because they give a strong illusion of shading and will pre-attentively ‘pop-out’ of a scene when oriented horizontally. However when the same stimulus is rotated by 90 degrees, the illusion of shading is drastically reduced, and the images will no longer ‘pop-out’ of a scene. It has been proposed that crowding may occur late in the visual system and actually be a form of limited attentional resolution. If this were true, then one would expect that the horizontally oriented shape-from-shading gradients would be less affected by crowded than the vertically oriented shape-from-shading gradient. This is because the horizontally oriented gradients can ‘pop-out’ of a scene pre-attentively, and thus require little or no attention in the later stages of vision, thus bypassing the attentional resolution limits of crowding in the later stages of vision. Our results agreed with this hypothesis and showed a strong interaction between shading orientation and crowding such that the horizontally oriented shape from shading gradients were much less affected by crowding than the vertically oriented shape-from-shading gradients, thus supporting the limited attentional resolution theory.
Edward Hubbard