Abstract
Previous results from our lab on color search conducted in controlled laboratory settings showed that search speed (the inverse of the search slope) for a target color increases linearly as a function of the target-distractor distance in CIELab color space. In other words, the larger the distance between target and distractor color is, the faster participants will find the target. Importantly, because this relationship between search speed and target-distractor distance is strictly linear, it challenges theories of visual search and visual attention that assume feature-based attention facilitates performance in feature search by “boosting” the processing of the target color because such boosting would result in a non-linear relationship between search speed and target-distractor distance. Here, we wanted to achieve two goals. In Experiment 1, we aimed to replicate the findings linking search speed to target-distractor separation in an online experiment. This is challenging because of the uncontrolled variability with respect to monitors and illumination conditions across participants, which adds substantial noise and/or distortions to the color stimuli. The results of this online color experiment were indeed noisier than the results measured in the lab, but a larger sample size was able to mostly counteract these concerns. Importantly, the results replicated the linear relationship between search speed and target-distractor distance. Experiment 2 tested a variation of the boosting hypothesis, the so-called Optimal Tuning Theory. According to this theory, feature-based attention boosts a nearby color that amplifies the representational difference between the target and distractor colors. As a result, when all distractor colors are on the same side of the target, search should become more efficient, particularly at small target-distractor separations (compared to Experiment 1 where no boosting was observed). Results are discussed in the context of Target-Contrast Signal theory of attention, which does not rely on feature boosting of any kind.