The preview benefit: Visual marking, feature-based inhibition, temporal segregation, or onset capture?

Abstract

The preview effect demonstrates that if observers in a visual search task are allowed a preview of a subset of elements before another subset of elements is added to the display, the first subset of elements no longer competes for attentional selection in the search process. Watson and Humphreys (1997) explained this effect by proposing that the locations of previewed elements are top down inhibited during the preview by a process they refer to as visual marking. The results of recent studies cannot easily be explained by the original visual marking account. As a consequence, three alternative views have emerged. According to one notion, a feature-based inhibition account, the preview benefit is mediated by inhibition applied at the level of feature maps in addition to location-based inhibition. A second view, the temporal segregation hypothesis, assumes that prioritized selection of new elements results from observers being able to selectively attend to one group of elements that can be perceptually segregated from another group on the basis of temporal asynchrony. A third view assumes that the preview benefit is caused by onset capture mediated by the appearance of the new elements. The present paper reviews the key findings concerning the preview benefit with the aim to resolve some of the controversies about how observers prioritize selection of new over old elements.

Acknowledgments

I thank Chris Olivers, and an anonymous reviewer for comments on an earlier version of this paper.

Notes

¹In addition to colour-based inhibition, Braithwaite et al. (2003) as well as Braithwaite and Humphreys (2003) also assume that the prioritization of new elements can be further supported by the possibility to use an anticipatory set for a known target colour.

²In Experiment 4, Jiang et al. (2002b) had participants to indicate whether the rotation of a target “T” was up, down, left, or right. The target was always presented among the old elements containing multiple L-shaped objects. New elements consisting of multiple rotated “T”s were added to the display after 150 ms. The results indicated that participants were able to correctly report the identity of the target in about 65% of the trials. Jiang et al. inferred on the basis of these results that observers can prioritize the selection of old over new elements. However, it should be noted that if prioritized selection for old elements had been perfect, performance should have been close to 100%.

³It is important to note that the results of several other studies (Humphreys, Watson, & Jolicoeur, 2002; Olivers & Humphreys, 2002; Watson & Humphreys, 1997 2000) suggest that prioritized selection is based on top-down processing. These studies have, however, generally utilized a preview task in which old and new elements not only differed in their temporal onset, but also in colour. If there is, however, no colour difference, as in Atchley et al. (2003) and Donk and Theeuwes (2003), prioritized selection seems to be completely bottom-up driven.

⁴Jiang et al. (2002b) remark that the interval between the presentation of the old and new elements should be “long enough for attention to be deployed to one group and not the other” (p. 719). They suggested that the required length of this interval should be at least 200 ms.