Abstract
When looking for things in our daily environments, we often rely critically on the contextual associations between objects, learned through prior experience. For example, when searching for a parked bicycle on a busy street, we typically scour locations where bicycles tend to occur, such as near bike racks and light poles, and we avoid searching unlikely locations, such as the tops of buildings or cars. How might these learned associations be acquired? We investigated this question using a series of visual search tasks. Specifically, we examined whether people can learn, over time, that target items are likely to occur within a particular category of objects (e.g., animals or dessert food). Participants searched through displays containing four categories of items, each localized to a particular quadrant of the display. One category was predictive; it always contained the target (e.g., the target always appeared near animals). Targets were randomized on each trial, and were unrelated to the category of images in which they were associated. Several combinations of categories were presented randomly across trials, each with a different predictive grouping. In order to verify that participants were learning to associate categories with targets and not simply relying on spatial cueing, we also included a condition wherein each category's quadrant location was randomized. In both conditions, participants found targets more quickly when a specific category always contained the target, compared to a control condition where no category was predictive of the target location. These results indicate that (1) people learn arbitrary associations between item identities relatively quickly, and (2) this category learning is independent of learning repeated spatial locations, and can be used to help guide the eyes to relevant items in the environment.
Meeting abstract presented at VSS 2014