Abstract
Studies of singleton search have found that response times may be decreased by priming the defining dimensions (Muller, Heller & Ziegler, 1995) or defining features (Maljkovic & Nakayama, 1994) of singletons. This phenomenon has been viewed as the direct enhancement or impairment of the relevant dimensions or features. A factor not considered in this work has been whether the set of target features and distractor features overlap. In Experiment 1 we found that, when the target feature of one trial can become the distractor feature of another trial (and vice versa), there was little dimension-priming effect. Experiment 2 was similar to Experiment 1, but target feature could not become a distractor feature (and vice versa); here we found a significant dimension-priming effect. In Experiment 3, we found that when target feature never become distractor feature (and vice versa), there was little feature-priming effect. Experiment 4 was similar to Experiment 3, but here the target feature of one trial could become the distractor feature of another trial (and vice versa), yielding a significant feature-priming effect. These experiments argue against previous singleton-priming accounts that posit direct enhancement or impairment of particular dimensions or features. We propose, as an alternative to such accounts, that both dimension priming and feature priming reflect a cost associated with inconsistent assignment of target and distractor features (Schneider & Shiffrin, 1977).
The research was supported by grant MH45584 from the National Institute of Mental Health to Harold Pashler. We are grateful to Todd Horowitz whose comments on an earlier study helped us develop the idea for the present work.