Theory-based Models of Attention in Visual Workspaces

Abstract

ABSTRACTTo minimize the risk of operator distraction or overload, system designers must tailor the arrangement of display channels in a visual workspace to the operator’s attentional abilities. Toward this end, computational models that embody the theoretical principles of attention control can inform the designer’s choices and narrow the space of potential display layouts for consideration. Wickens’ SEEV model incorporates four mechanisms of attentional guidance—stimulus salience, effort, expectancy, and information value—along with a probabilistic choice mechanism to predict the steady-state distribution of attention over multiple information channels. An extension of the model also predicts the speed and probability with which participants notice critical signals in a dynamic workspace. Empirical tests in single- and multi-task supervisory monitoring environments confirm that the models predict performance well, capturing the influence of various top-down and bottom-up attentional control mechanisms. Results suggest that the models can serve as theory-motivated tools to guide and assess the design of multi-channel visual workspaces.