Abstract
Traditional visual search studies assume that a visually-distinct, target item will be immediately attended and identified. These paradigms employ odd-item displays with one target present amongst distractors and homogeneous displays of only distractors, allowing subjects to use the strategy of an odd-item detection, without necessarily identifying the target item. Using a new paradigm (Seiffert, Dupuis & Caramazza, in prep), we enforce target recognition, by adding trials that have homogeneous displays of only targets and odd-item displays with one different distractor amongst other distractors. Observers detected a typical “pop-out” target, such as vertical targets amongst slanted distractors. We compared these results to search which enforced odd-item detection without specifying its identity (the odd-man-out task). Reaction times for odd-item detection did not differ from recognition of a single specified target amongst distractors, though both were from 50-100 msec slower than recognition of multiple targets. Therefore, whether detection was based on local contrast with distractors or on recognition of a specified item, emergence of a single target item requires more processing relative to multiple targets. To assess the role of grouping and segregation, we measured responses to a group of targets compared to scattered targets. These manipulations had negligible effects. To more closely assess the role of identification, we repeated the first experiment with targets defined at a higher level of abstraction — digits among letters. Heterogeneous targets were identified as quickly as a homogeneous targets. We conclude that sensory properties of the display and item identity do not weigh heavily into the analysis, but rather decisions are reached through accumulating evidence of target presence from all items in the display.