Abstract
Shapiro et al (2007; 2008) demonstrated that most brightness illusions can be accounted for by removing low spatial frequency content from the image. Here we examine the ability of this approach to account for a class of phenomena we refer to as Long-Range Argyles (LoRA). LoRA elements are diamonds divided into four quadrants: two gray, one black, and one white (black and white quadrants are on opposite sides of the diamond). The diamonds are arranged in columns (or rings), with black quadrants facing each other and white quadrants facing each other. When the columns are close together on a gray background, the display is the same as Adelson's Argyle illusion (the gray area between the white quadrants appears dark; the gray area between the black quadrants appears bright). In LoRAs, however, the columns can be moved independently of each other. We demonstrate that the brightness effects 1) occur over large spatial distances (i.e., when columns are separated by more than 15 deg); 2) are fast, relative to standard induction (i.e., the effect can be seen when white and black quadrants modulate at 8 Hz); 3) do not depend on sharp edges (i.e., the effects in LoRAs occur when one or more columns are blurred; this is important because under such conditions, the columns appear to be in different depth planes); 4) can occur simultaneously at different spatial scales; and 5) appear in the presence of other long-range spatial interactions (i.e., a column that swings horizontally appears to shift vertically when flanking columns are present). A high-pass filter model with a single cutoff spatial frequency can account for some aspects of LoRA phenomena, but not at all spatial scales simultaneously, and not at spatial ranges much larger than the diamonds. We discuss the role of a separate contrast pathway (Shapiro, 2008).
Supported by NEI (EY017491).