Quantifying the effects of feature similarity on attentional selection using psychophysical scaling

Abstract

The similarity between target and distractor features is a main factor determining the efficiency of attentional selection. Prior studies investigating the impact of target-distractor similarity on performance have mostly selected similarity levels in a categorical and qualitative way (e.g., by using discrete color or shape categories), or have used distance in physical stimulus space to quantify similarity. However, findings from psychophysics have shown that the physical stimulus space does not map linearly to psychological similarity. Thus, we here assessed psychological similarity using psychophysical scaling methods and tested how these similarity functions relate to attentional selection. We measured psychological similarity of colors chosen from CIELab space as a test case. Then, using a standard visual search task, we measured search times using a wide range of target-distractor similarity levels (Exp. 1) and also manipulated the number of distractors (2 or 7) to measure search efficiency (Exp. 2-3). In another sustained attention task (Exp. 4), participants continuously monitored a set of target dots among distractor dots to detect brief changes in luminance, and the color similarity between target and distractor dots was manipulated across trials, as in the visual search tasks. We found that performance in both tasks was comparably affected by target-distractor similarity, with search RTs and luminance discriminability plateauing at approximately 40°-50° around the CIELab color wheel. In contrast, the number of visual search distractors only affected performance when similarity was very high (below 20°). Interestingly, performance in each task was non-linearly related to both the physical and psychological similarity between targets and distractors. Overall, these findings suggest that there is no simple linear relationship between feature similarity and attentional selection, and point to the importance of understanding how the underlying perceptual organization of a given feature space interacts with mechanisms of attentional selection.