Searching for stimulus-driven shifts of attention

Demonstrating a stimulus-driven attention shift is a thorny problem. In the attention capture literature, stimulus-driven attentional shifts are inferred when the task provides no incentive to attend to the critical feature. Using this definition, most studies suggest that when the critical feature is unpredictive of target location in a visual search task, dynamic events (e.g. onset, looming) capture attention but static singletons (e.g. unique colors or shapes) do not. In practice, even if a feature is unrelated to the target, subjects might attend to it voluntarily. For example, because observers must anticipate the appearance of the search display itself, this top-down goal might bias subjects to attend to any dynamic event. Without this bias, dynamic events might not capture attention (Gibson & Kelsey, 1998). We rule out this explanation with a search task in which all display changes (except the critical dynamic event) are made during saccades. Even when the task does not require a top-down bias toward display changes, the abrupt onset of a new item still garners attentional priority. In another case, top-down strategies do appear to cause illusory capture effects. A recent claim counters the assumption that irrelevant static singletons do not attract attention by showing that in some cases uniquely colored and shaped items do receive priority (Turatto & Galfano, 2000, 2001). However, this priority might not reflect stimulus-driven capture. The amount of priority is equivalent to 'attentional misguidance', a slight voluntary prioritization which appears in a restricted range of conditions. More importantly, in our replications of these experiments, prioritization appears rarely and appears to depend heavily on task demands and instructional details. Together, these findings provide additional support for the claim that some dynamic events capture attention in a stimulus-driven fashion, but that static singletons do not.