Using the Centroid Method to Study Feature Based Selective Attention
Feature-based selective attention is a very important ability people use to explore the world around them. Being able to suppress distractions and focus on one feature of items is instrumental in a person being able to find what they are looking for. Historically, this has been studied with a search task—one that asks participants to find a target item amongst distracters. This method, while useful, has some flaws that leave open gaps in our knowledge of how feature-based selective attention works and what features people can use it on. This dissertation explores a relatively new method of exploring this ability—the centroid method—that is capable of covering these gaps. Using this paradigm, Chapter 1 shows that a feature-dimension the search literature implied was useful—orientation—is not. Subjects cannot ignore one orientation and attend only to another. The search literature conflated orientation with local orientation contrast, and the centroid task was able to separate these features. Chapter 2 compares the centroid task to a numerosity estimation task in order to show that computing centroids does not rely on computing numerosity, eliminating some otherwise reasonable theories on how these computations are done. Chapter 3 uses the centroid method to explore a feature dimension that heretofore has not been extensively studied—internal angles. Subjects were able to use the internal angle of items as a useful feature in the centroid task, to a limited degree. As long as the angles are discriminable enough, which seems to be determined by the ratio between them, subjects could selectively attend to just a target angle (or range of angles) while ignoring distracters. Overall, the centroid method proved to be a useful tool for studying how feature-based selective attention worked in three different experiments, and when subjects could and could not use it.