Many everyday tasks require that we select from the environment stimuli relevant to our behavioral goals. Normally, this process of goal-directed selection may depend on a match taking place between goal-related information in working memory (WM) and incoming stimulus information (Desimone & Duncan, 1995;Duncan, 1998). Evidence supporting the role of top-down guidance of search to targets comes from experiments showing that search b benefits by providing participants with foreknowledge of targets (e.g., Anderson, Heinke, & Humphreys, in press;Wolfe, 2005), and that indeed some targets only "pop out" when foreknowledge is given (Hodsoll & Humphreys, 2001). Recent work suggests that these top-down effects are not confined to holding knowledge of the target, since there can also be effects of irrelevant information in WM. Downing (2000) had participants hold one stimulus in WM, then search for another target. The irrelevant cue in WM could reappear alongside the target or another item. Response times (RTs) were faster when the WM stimulus reappeared at the target's location (on valid trials) than when it fell elsewhere (on invalid trials). Importantly, this effect did not occur when the cue was presented but did not have to be maintained in memory. Soto, Heinke, Hump phreys, and Blanco (2005) further showed that irrelevant items in WM influenced the fastest RTs and the first saccades in search, and the effect occurred even when the WM cue was always irrelevant. This last result suggests that an irrelevant WM stimulus can capture attention automatically. Other studies have demonstrated effects, even with pop-out targets (Soto, Humphreys, & Heinke, 2006).Interestingly, in many of these studies, the WM item h constituted more of a "global" stimulus than the search target. For example, in Soto et al. (2005), the memory item d was an outline shape and the search target was an oriented line that appeared within the outline shapes presented in the search display. It could be that the WM stimulus captures attention at least partly because there is a bias toward the global properties of a display, as suggested by the "global precedence" hypothesis (Navon, 1977), which f holds that attention is biased toward global aspects of stimuli. When the WM cue matches early-emerging global properties of the search display, attention may be drawn to the global item matching the item in WM. It may also be that representing a stimulus in WM itself biases selection to the global level of a display, perhaps because the WM load reduces the resources available to process more local elements of a display. The biasing effects of WM on the selection of local and global properties of search displays were examined for the first time here. We had participants y g y carry out a global-local task in which they had to detect a Previous research has shown that stimuli held in working memory (WM) can influence spatial attention. Using Navon stimuli, we explored whether and how items in WM affect the perception of visual targets at local and global lev...

Participants held the size of a cue in working memory and looked for a target letter across local and global levels of hierarchical, compound letters. There was no effect of whether the cue size matched the size of the local or global letter, even when the size memory task was made difficult. The null effect of cue size was also found under priming conditions, when size had to be identified but not held in memory. Although cue size did not affect subsequent selection in these experiments, cue identity did. There were effects of size, however, on the magnitude of cueing effects, irrespective of whether targets were at the local or global level. The data are discussed in terms of the effects of overlap between the contents of WM and the attentional set for targets, and the role of focused and distributed attention on selection. Critically, the data run counter to the idea that local and global stimuli are selected by opening an ''attentional window'' of a particular size in working memory and matching this to the target letter.