Abstract
Attention to a specific feature (e.g., the color red) enhances processing of that feature over other, unattended features. However, whether and how focusing attention to a specific color may alter feature representations outside the focus of attention is largely unknown. In two experiments, we show that feature-based attention changes the appearance of colors, and that this perceptual warping extends through large parts of the feature space. Participants performed two tasks: First, they continuously attended to a set of colored target dots among distractor dots (30° away along a CIELab color wheel) to detect a brief decrease in luminance of the targets. Simultaneously, they judged which of two briefly, unpredictably flashed probe colors was most similar to the attended target color. Probe colors were sampled from around the color wheel, both on the same side as the distractor color and on the opposite side. We modeled the effects of attention on color similarity judgments using maximum-likelihood difference scaling (Maloney & Yang, 2003). In Experiment 1, participants (N=20) more accurately discriminated colors on the same side of the color wheel as the distractor compared to equally-distant colors on the opposite side, leading to a reliable separation of the similarity functions on each side of the color wheel. In Experiment 2 (N=30), we replicated this finding when probe colors appeared without the target and distractor dots on the screen, eliminating the possibility of color contrast. These findings suggest that feature-based attention can alter color representations, such that colors closer to targets and distractors are perceived with higher fidelity relative than colors more distant in feature space. This demonstrates that attention does not act uniformly on a given feature space, solely enhancing target processing, but can warp feature representations across the feature space, effectively decreasing the similarity between targets and potential distractors.