How does feature-based attention affect visual processing?

Five experiments are reported from which it is concluded that attending on the basis of a stimulus feature (e.g., red) does not directly affect the sensory quality of stimuli that possess that feature. Feature-based attention was manipulated in a visual search task by providing information about the probability that the target would possess a given feature (e.g., "The target has a 1.0 probability of being red when present."). Feature-based attention failed to aid performance under "data-limited" conditions (i.e., those under which performance was primarily affected by the quality of the stimulus) but did affect performance under conditions that were not data limited (Experiments 1-3). If attending to a feature had affected the sensory quality of stimuli, performance should have been aided under all conditions. Experiments 4 and 5 provided converging support for this conclusion.