Abstract
According to ‘biased-competition model’ (Desimone & Duncan, 1995), the contents of working memory (WM) guide the allocation of selective attention. Recent work by Downing (2000) with human observers supported this model using probe discrimination task. In contrast, it has been failed to demonstrate this guidance effect during visual search (Downing, 2000; Houtkamp, Spekreijse, & Roelfsema, 2002). However, in the typical visual search task such as this, it is possible that a target template stored in WM is behaviorally most relevant, so that the other contents in WM do not guide selective attention. To investigate this possibility, we used two different types of visual search task. In one task, the search target was defined as an object whose shape was symmetric around the vertical axis (Chun & Jiang, 1999). In the other task, the search target was defined as a specific shape. The guidance effect of WM was measured by the difference in search response times between the matched condition (in which the WM item was presented in the search array as a distractor), and the non-matched condition (in which the WM item was not presented in the search array). The significant guidance effect of WM was found when the target was defined as vertically symmetric but not when the target was defined as a specific shape. These findings suggest that WM items can guide attention in visual search in the absence of behaviorally more relevant items in WM-such as target templates in the context of visual search.