Purchase this article with an account.
Xin Chen, Gregory J. Zelinsky; Monitoring the use of target memory during visual search. Journal of Vision 2004;4(8):683. doi: https://doi.org/10.1167/4.8.683.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Visual search requires comparing a target representation to patterns appearing in a search display, but few studies have asked how working memory (WM) limitations constrain our ability to represent and maintain this target code during a search task. We report two experiments that tracked target memory use during search on a moment-by-moment basis. The stimuli were Chinese characters and the subjects were either Chinese-readers (CR) or non-Chinese-readers (NCR). In Exp 1, 9 search characters were arranged in a circle enclosing a central target character. The task was to indicate whether the central target appeared in the circle array. Eye gaze was monitored throughout search, with memory defined in terms of the number of refixations back to the central target. In Exp 2 we introduce the “flipping paradigm”, a manual counterpart to the eye movement paradigm used in Exp 1. The stimuli were again Chinese characters, but the central target and the search array now appeared on separate displays. Subjects were free to flip back and forth between these target and search displays (by pressing a button) if they felt a need to review the search target. We found that (1) both target refixations (Exp 1) and flips back to the target (Exp 2) were twice as frequent in NCR subjects compared to CR subjects and (2) that these refixations and flips occurred disproportionately near the end of a trial immediately before the search judgment. We interpret the first pattern as indicating a greater WM load for the NCR subjects. Because these subjects had no semantic label for the target, they had to hold more purely visual information in memory. However, the many refixations and flips near the end of search suggests that these behaviors were serving a checking function and were not caused by target information fading from visual WM. We discuss these data in terms of a tradeoff between the completeness of the target code and the duration that this representation can be held in WM.
NIMH grant R01 MH63748 and Army Research Office grant DAAD19-03-1-0039
This PDF is available to Subscribers Only