Control of Dynamic Attention

Brigham And Women'S Hospital, Boston MA

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): The broad goal of the proposed research program is to understand the deployment of attention in a changing visual scene. Attending to multiple moving objects is critical to everyday visually guided action, such as driving in traffic or walking through crowds. Humans are able to perform in such situations with high accuracy, despite cognitive demands from other tasks. For professional drivers, pilots, or air traffic controllers, the smallest loss of performance on this task may result in terrible consequences. However, we currently know very little about this core visual skill. Much of what we do know comes from multi-element visual tracking experiments, in which subjects are able to track up to five moving targets selected from a larger set of otherwise identical moving items. Our preliminary data show that tracking is surprisingly flexible and robust. Tracking performance is unimpaired even when stimuli are invisible for up to 300 ms. The predictive ability required to perform this task may be the ability that allows humans to switch between various tasks requiring visual attention while still keeping track of important moving objects in a dynamic world. Current theories of tracking are inadequate to account for this ability. We propose to draw on existing models of scene parsing and path extraction in static images to help understand path construction in dynamic scenes. In particular, we hypothesize that prediction can be modeled using spatiotemporal association fields. These are a natural extension of the spatial association fields that have proven useful in our current understand of contour integration in static images. Using multielement tracking and visual search paradigms, we propose to (1) test the adequacy of existing accounts of the ability to track multiple objects over time; (2) test the hypothesis that spatiotemporal association fields contribute to tracking of multiple objects over time; (3) test the role of spatiotemporal grouping in successful tracking; and (4) investigate the interaction of cues in the tracking of multiple objects over time. These experiments that will allow us to understand the mechanisms of multielement tracking and to use this task as a tool to study attention in a dynamic world.