Abstract
How do humans search for multiple targets from more than one category? In contrast to animal research, such foraging has been largely neglected in human visual search, typically involving single-category, single-target trials that terminate with the first response. Here, we introduce a new iPad foraging task where observers cancel a series of targets among distractors by tapping them until all are gone. The number of possible target types and distractor types can vary independently. We asked how foraging changed as a function of target complexity. During feature-based foraging 16 naïve observers cancelled 40 green and red disks among yellow and blue distractors (or vice versa). During conjunction-foraging they cancelled red disks and green squares among green disks and red squares (or vice versa). Our main dependent variable was the number of "runs" per trial, which could vary between 40, if a target switch occurred after each response, and 2, if sequential subset searches were carried out. Random switching would result in mean run length of 20. For feature-based foraging, behavior was characterized by many short runs (M = 14), suggesting that observers could simultaneously maintain two or more color templates in mind and were happy to switch between them. For conjunction-based foraging, the pattern was dramatically different: observers focused on one target-type, finishing most or all of those before switching to the other target type (M = 5). In both conditions, more runs resulted in less overall movement, suggesting that switching might be the more "optimal" strategy. For conjunction foraging, this was mostly driven by 4 observers who were able to switch frequently despite the increase in target complexity. Our main, novel finding -- the striking difference between conjunction and feature-based behavior -- suggests that attention imposes a very sharp constraint on human foraging.
Meeting abstract presented at VSS 2014