There is an ongoing debate as to whether prioritizing new objects over old objects (the so-called preview benefit) is the result of top-down inhibition of old objects (i.e., visual marking; Watson & Humphreys, 1997) or attentional allocation to new objects, presented with a luminance transient (Donk & Theeuwes, 2001). In the two experiments reported here, we tested whether prioritization by luminance transients alone can produce a subset-selective search similar to the preview effect. Subjects viewed multiobject displays while a subset of objects was briefly flashed. The subjects prioritized up to 14 flashed objects over at least 14 nonflashed objects. Since prioritization by luminance transients can produce a subset-selective search on its own, it may well play an important role in the preview benefit.
The visual system utilizes a variety of attentional mechanisms to prioritize the selection of relevant information in the environment. Recently, Watson and Humphreys (1997) showed that visual search can be significantly improved when the presentation of search objects is separated in time. If a subset of distractors is previewed for at least 400 msec before the search display, the previewed objects can be excluded from search. This phenomenon, termed the preview effect, appears to be fairly robust across a variety of conditions; it occurs with previewed subsets of up to 15 objects (Theeuwes, Kramer, & Atchley, 1998) and is not affected if search objects change color (Watson & Humphreys, 1997,2002). Watson, Humphreys, and colleagues (Watson & Humphreys, 1997, 2002) argued that the preview effect is a result of top-down inhibition of the previewed (old) objects, a mechanism that they termed visual marking.
Even though top-down inhibition represents the modal view of the mechanism underlying the preview benefit, recently Donk and Theeuwes (2001) proposed an alternative account. They suggested that the preview effect occurs as a result of the prioritization of new objects (Yantis & Jones, 1991). They showed that the preview effect critically depends on whether the new objects are accompanied by luminance increments. When the new objects were presented as isoluminant with the background, the preview effect was abolished.
Even though Donk and Theeuwes's (2001) account seems viable, several arguments in favor of the top-down inhibitory account of visual marking have been put forward. First, the preview effect seems to have a relatively long time course; old objects have to be previewed for at least 400 msec in order for the preview effect to emerge, and it reaches asymptote at about 1,000 msec. Second, the preview effect is reduced if subjects are engaged in a demanding secondary task during the preview (Olivers & Humphreys, 2002; Watson & Humphreys, 1997). This implies that inhibition is an active top-down process that can be disrupted by a secondary task. Third, probe detection studies have shown that the detection of a probe dot was the worst at the locations of old objects, suggesting active inhibition (Watson & Humphreys, 2000). Finally, it is the accepted view in the literature that only a limited number of luminance transients, about four, can be prioritized for search (Burkell & Pylyshyn, 1997; Yantis & Jones, 1991). In visual marking studies, it has been demonstrated that at least up to 15 new locations can be prioritized for selection (Donk & Theeuwes, 2001; Theeuwes et al, 1998).
Given the literature reviewed above, the fundamental question as to whether the preview benefit is the result of top-down inhibition of old objects or of attentional allocation to new objects, accompanied by luminance transients, is still undecided. The experiments we report here were designed to test whether prioritization by luminance transients alone could produce a subset-selective search similar to the one found in the preview paradigms. We designed a paradigm in which a search subset was defined solely on the basis of luminance transients and there was no preview of a subset of objects. …