Bradley S. Gibson - US grants

When we gaze at a complex visual scene, what determines where we look? Current theory suggests that visual attention is controlled by two distinct mechanisms: "goal-directed" when visual attention is oriented in terms of the observer's goals, and "stimulus-driven" when visual attention is "captured" by objects and events in the external world irrespective of goals. The primary question that will be addressed in the present research is, "Can visual attention ever be captured in purely stimulus-driven fashion?" Over the past decade, substantial empirical evidence has supported the possiblity that at least one stimlus feature, abrupt onset, can drive attention. This evidence has been obtained using a visual search paradigm in which a single abrupt visual onset (which happens to be the target only by chance) appears among several"no-onset" elements (which are typically formed by removing camouflaging parts from preceding placeholder objects). In contrast, other visual features such as color, luminance, flicker, and motion do not appear capable of capturing attention. Such findings have led to the belief thatabrupt visual onsets are special. One popular interpretation is that they capture attention because they correspond to the appearance of new perceptual objects. However, the present research seeks evidence for an alternative account in which abrupt visual onsets may capture attention because they correspond to the earliest (and largest) form of change in the visual field. What actually happens is that perception of the no-onset elements is delayed. This hypothesis is controversial primarily because such a claim has been thought to be inconsistent with traditional accounts of temporal vision (involving a phenomenon known as temporal integration). Thus, the first phase of this research program is devoted to developing a new, temporal coding account of temporal integration that is based on human psychophysical data obtained from a novel change detection paradigm as well as mathematical modeling techniques. In addition, a dichoptic presentation paradigm will also be used in an attempt to localize the temporal coding process within the visual system (i.e., before V1 or after V1). The second phase of this research will then focus on how this temporal coding process is affected by other higher-level states of the visual system such as whether attention is focused on a single object or divided over several objects (as it typically is in the visual search paradigm). Together, the present series of experiments are important because they will extend our knowledge of temporal vision while also providing a more thorough understanding of why abrupt visual onsets appear unique in their ability to capture attention.