Abstract
Search can be accomplished by categorizing each item into target and distracter categories and applying the single-object categorization procedure multiple times. To what extent is human visual search consistent with this notion? To answer this question, we applied two different categorical mappings on a single set of stimuli, and compared performance under these two mappings during search or categorization tasks. The stimuli were diamond-shaped objects missing one corner segment or a portion of one side segment. One categorical mapping (the component mapping), placed “corner-missing” objects into one category and “side-missing” objects into another category, while the other mapping (the orientation mapping) placed objects with either the top corner or a segment of the two top lines into the “top” category and objects with the bottom corner missing or a segment of the bottom lines missing into the “bottom” category. When searching for a target among multiple distracters, subjects responded most quickly when cued by the exact target, less quickly when told only which component category the target belonged to, and least quickly when told which orientation category the target belonged to. However, when categorizing single stimuli in the absence of distracters, subjects were significantly faster categorizing the stimuli according to the orientation mapping than the component mapping. Thus, the component cue appears to be more helpful than the orientation cue for guiding visual search, while the converse was true when a single stimulus was categorized. We conclude that human search performance is not consistent with the notion that search is accomplished by employing single-object categorization multiple times.
Supported by NIH 071788 and NSF 0345525.