Abstract
During visual search, it is typically thought that possible targets (candidates) are selected from a search set, then serially inspected until the target is found (e.g., Wolfe, 1994). The relationship between search set-size and the amount of time it takes to locate the target is traditionally thought to be linear: the more items present, the longer it takes to complete search. Wolfe, et. al, (2011) demonstrated that this linear relationship does not translate to search in real scenes. Despite very large set-sizes, search in real scenes is highly efficient. Recently, Lleras, Cronin & Buetti (submitted) proposed a model for visual search (Information Theory of Vision, ITV) in which the RT X Set-Size function is mostly logarithmic. Instead of beginning with a selection of candidates, ITV proposes that search begins with the sequential rejection(s) of items unlikely to be targets (lures): lures dissimilar to the target are rejected rapidly and those that are more similar to the target are rejected more slowly. Because the search function is logarithmic, large set-sizes do not necessarily elicit extremely long reaction times, better modeling real-world search times. However, real-world scenes contain many different lures varying in dissimilarity to the target and thus in their ease of rejection. Here, we investigated the impact of an array having multiple types of lures on the RT X Set-Size function. Experiment 1 employed sets of two different lures highly dissimilar to the target. Experiment 2 employed sets of two different lures highly similar to the target. Trials in each experiment contained four candidate items and a varying number of lures from one or both sets. The results of these experiments provide supporting evidence for ITV and a better understanding of search in the presence of multiple non-candidate objects, as is the case in real world scenes.
Meeting abstract presented at VSS 2014