We examine how the RG cardinal mechanism adapts to a shift from one spatially uniform field to another. A probe-flash paradigm allowed the response to be estimated at an initial adaptation state, an end adaptation state, and four intermediate time conditions. The threshold curves were higher and flatter in the intermediate conditions than at the beginning and end of adaptation. The intermediate results cannot be explained by combinations of cone or opponent-level multiplicative and subtractive gain controls, and implicate higher-order adaptation processes. (C) 2000 John Wiley & Sons, Inc. Col Res Appl, 26, S43-S47, 2001.