Abstract
We define ‘weaves’ to be a class of stimuli that contain intertwined vertical and horizontal bars. There are two subgroups of weaves: luminance-defined weaves consist of patterns in which the vertical and horizontal bars have different luminance levels; equiluminant weaves consist of patterns in which the bars have the same luminance levels. The classic Hermann grid display (i.e., a white grid superimposed on a black background, or vice versa) is a special case of an equiluminant weave. As with the Herman grid, both luminance-defined and equiluminant weaves produce smudges when the background is brighter or darker than the bars. We demonstrate that 1) luminance-defined weaves produce smudges at every other intersection and equiluminant weaves produce smudges at every intersection; 2) for luminance-defined weaves, the smudges are present in the high-spatial frequency components of the pattern but not in low spatial frequency components; 3) unlike the smudges for equiluminant weaves, the smudges for luminance-defined weaves are not disrupted by jaggy bars, wavy bars, thick bars, or orientation changes; and 4) unlike the smudges for equiluminant weaves, the smudges for luminance-defined weaves occur foveally and can be created with contrast variation (contrast-contrast weaves). One possible framework for considering these effects is that, as suggested in the theory of Schiller and Carvey (2005), the smudges arise out of the co-activation of ON and OFF simple cells with colour-selective simple cells—except that the colour selective cells represent both hue and achromatic sensations. We also suggest another possibility in which the smudges arise from a bi-product of lightness interpolation mechanisms that builds a 1st-order representation from 2nd-order information.
This research was supported by the German Science Foundation (Graduate Program ‘NeuroAct’ 885/1). We thank Karl R. Gegenfurtner for valuable comments on this topic.