A comment on "assimilation of achromatic color cannot explain the brightness effects in the achromatic neon effect" by Marc K Albert.

in the achromatic neon effect'' by Marc K Albertô Albert (1998) describes a number of visual displays, which he uses to argue against current models of neon color spreading. He particularly emphasizes models which I and my colleagues have developed. He erroneously concludes that our models cannot explain the percepts that are generated by these displays. I am writing to set the record straight. His figure 1a describes a well-known example of achromatic neon color spreading in Redies ^ Spillmann displays. This figure depicts a small light-gray cross on a darker-gray background. The cross has horizontal and vertical sides. The small cross is embedded within a larger cross whose sides, beyond the light-gray parts, are white. The image as a whole leads to a percept of neon color in which a circular gray disk intersects the four white ^ gray edges of the cross. The gray disk, which is formed by neon color spreading, seems to be darker than the light-gray cross. Albert claims that such percepts `̀ provide strong evidence against a number of visual theories and models that offer explanations of the neon effect based on mechanisms that predict that assimilation will determine the direction of the brightness illusions seen in neon stimuli'' (page 844). In particular, he claims that our ``theory also predicts that the target region [the neon disk] ... should appear brighter than the background (ie brightness assimilation), since the cross appears brighter than the surround'' (page 840). Albert correctly notes that our model predicts how a circular illusory contour forms between the four white ^ gray edges of the cross, and that the disk that is enclosed by this circular contour fills-in brightness values that are computed at positions within the boundaries of the light-gray cross. In this sense, our model is the type of assimilation model that Albert is criticizing. On the other hand, Albert does not describe the crucial property of how our model computes the brightness signal that gets filled in. Instead, he simply observes the percept of a brighter cross than neon disk, rather than describing how the model mechanistically explains the darker-gray that is actually filled in within the disk. He hereby confounds the percept with its cause. Later in the article, Albert acknowledges that our model has more structure than his critique based on his figure 1a might suggest. In particular, while discussing the model's presumed explanation of a percept involving the Kanizsa square (see his figure 2), Albert claims that the model predicts `̀ enhanced brightness signals (owing to edge contrast)'' (page 844). Thus, our `̀ assimilation'' model includes `̀ contrast'' mechanisms, which are the opposite of assimilation mechanisms. This is not a contradiction, because our model, which has come to be called FACADE theory, describes several types of processing that occur in multiple processing stages. We pioneered models (eg Cohen and Grossberg 1984; Grossberg 1987a, 1987b; Grossberg andMingolla1985a,1985b,1987) inwhich perceptual boundaries and surfaces are processed in parallel by the visual cortex. These boundary and surface systems are often called the `boundary contour system' and the `feature contour system'. FACADE theory attempts to explain how these boundary and surface systems interact to generate conscious percepts. Many data have since supported the hypothesis that separate boundary and surface systems exist; see Grossberg (1994) for a review. Two of the most recent sets of experiments are those of Elder and Zucker (1998) and Rogers-Ramachandran and Ramachandran (1998). Discussion Perception, 1999, volume 28, pages 1291 ^ 1302