Abstract
Feature-based attention is the ability to select relevant information on the basis of visual features, such as a particular color or motion direction. While some research has proposed a fixed limit on attention to a single visual feature (e.g., Huang & Pashler, 2007), it is unknown whether feature-based attention can be tuned more narrowly or broadly across different feature values. Here we test the limits of feature-based selection by asking whether and to what extent different ranges of feature values can be efficiently attended.
Across four experiments, we demonstrate that feature-based attention can be flexibly adjusted in response to current visual input. Participants were instructed to attend to a set of colored dots amongst a set of distractor dots to detect a brief luminance decrease. To vary attentional focus, we manipulated the range of target dot colors across six conditions (uniform distributions spanning 10°, 20°, 40°, 60°, 90°, or 120° in a luminance-matched CIELab space), and found consistent but surprisingly moderate reductions in performance as the range of target colors increased. In a second experiment, we replicated this finding in a 2-dimensional CIELab space, controlling for influences of target-distractor similarity between conditions. Additional experiments demonstrated that this effect could not be explained by participants simply attending to a subset of the target dots as the color range increased (Experiment 3 & 4).
Our findings show that feature-based attention can be directed to different ranges of colors, suggesting a flexible focus of attention, analogous to findings in spatial attention. Whether these two modes reflect the same underlying processes (a “zoom-lens” for feature-based attention) remains to be determined. Exploration of the representational spaces that underlie attention to features and locations could further bridge the connection between these literatures and help sharpen our understanding of the common mechanisms that influence visual attention.