Abstract
The role of memory in visual search is under debate. Wolfe et al. reported that RTs to confirm the presence of a target increased with the number of display items even when search displays were repeated multiple times. They took these results to indicate that subjects continued to search through repeated displays as if they had no memory for the display items. In contrast, Chun et al. showed that with repeated presentation, subjects moved their attention to the target location more quickly. These “contextual cueing” effects suggested that memory plays a role in visual search. On the assumption that memory for the target location takes the form of an attention vector, we hypothesized that competing attention vectors might produce the appearance of memory-free search in the Wolfe et al. paradigm. Suppose subjects move their attention to the target location before they affirm its presence; a display with multiple repeated targets will have multiple associated attention vectors. To move attention to the location of one target, competition among attention vectors must be resolved. If resolution time increases with the number of competing attention vectors, these implicit memories may produce Wolfe et al.'s pattern of results. By this account search time should vary with the number of probed targets in repeated displays rather than with the number of display items, factors that were confounded in the original design. Like Wolfe et al. we repeated search displays containing 2, 3, 5, or 8 items multiple times by block. In each block, every display item was probed an equal number of times. We also included a condition in which 8 items were always displayed, but only 2, 3, or 5 items were probed by block (there was also a set of target absent trials in each condition). When the number of probed targets varied from 2 to 5, search slopes were equivalent (34 ms) regardless of the number of items in the display, consistent with the competing attention vectors account.