One of S. Siegel, L. G. Allen, and T. Eissenberg's (1992) recent arguments in support of associative-learning explanations of colored aftereffects (CAEs) is that the contingencies underlying these effects are not constrained by simple stimulus dimensions, such as contour orientation. Specifically, the authors claim to have generated CAEs contingent on sets of spatiotopic relationships between orientation components of a pattern (as opposed to orientation components per se). The present article illustrates how Siegel et al.'s claims are compromised by their failure to adequately address the role of fixation and eye movements during CAE induction.