Abstract
The phenomena and theories of brightness illusions are a fascinating field of vision research. Neurophysiology has left little footprint in this field besides the lateral inhibition concept that did not fare so well. The purpose of this contribution is to point out a possible common neurophysiological explanation for a diverse sample of brightness and color illusions and emphasize the quest for a comprehensive theory. Single cell recordings have shown that neurons in the visual cortex emphasize luminance and color borders and generally signal their orientation. Many of these neurons (over 50% in area V2) are also selective for border ownership (Williford & von der Heydt: Scholarpedia 8(10):30040, 2013). Recent studies suggest that border ownership selectivity reflects grouping circuits in which neurons responding to the contours of an object activate grouping cells at higher levels which then, by feedback, enhance the responses of the same contour neurons. I argue that the activation of grouping circuits corresponds to the formation of proto-objects in cognitive theory. This process is parallel, fast and automatic. The activated grouping cells point to salient objects and provide handles for top-down attention and other object-based cognitive operations. Specifically, I propose here that lightness and color of objects are computed from the border signals in the corresponding proto-object representations. I will discuss how this theory can explain the phenomenon of filling-in (Krauskopf: JOSA 53, 741, 1963), the influence of image segmentation on color (Nakayama et al.: Perception 18, 55, 1989) and lightness perception (Anderson & Winawer: Nature 434, 79, 2005) and color after-images (van Lier etal. Curr. Biol. 19, R323, 2009), White’s illusion (White: Perception 8, 413, 1979) and attention-induced brightness changes (Tse: Vision Res. 45, 1095, 2005). Some of these explanations are backed up by neurophysiological or psychophysical studies, but many questions await experimental answers.
Meeting abstract presented at VSS 2014