Attentional selection of superimposed surfaces cannot be explained by modulation of the gain of color channels.

When two differently colored, superimposed patterns of dots rotate in opposite directions, this yields the percept of two superimposed transparent surfaces. If observers are cued to attend to one set of dots, they are impaired in making judgments about the other set. Since the two sets of dots are overlapping, the cueing effect cannot be explained by spatial attention. This has led to the interpretation that the impairment reflects surface-based attentional selection. However, recent single-unit recording studies in monkeys have found that attention can modulate the gain of neurons tuned for features such as color. Thus, rather than reflecting the selection of a surface, the behavioral effects might simply reflect a reduction in the gain of color channels selective for the color of the uncued set of dots (feature-based attention), as if viewing the surfaces through a colored filter. If so, then the impairment should be eliminated when the two surfaces are made the same color. Instead, we find that the impairment persists with no reduction in strength. Our findings thus rule out the color gain explanation.