Deep Structure and Smart Mechanisms: Designing Perspicacious Systems

A fundamental challenge in the design of any cognitive system is to support productive thinking and efficient control. Research shows that human problem solving can be greatly enhanced using representations that reflect the deep structure of problems. Further, research on human action shows that selectively constraining degrees of freedom can improve both speed and accuracy of performance. This talk will discuss how these two insights from the basic research literature can be incorporated into work analysis and interface design to enhance performance of cognitive systems. The goal is to design interfaces so that the deep structure of the problem is well mapped to the opportunities for action. A major challenge is to operationalize the basic constructs of deep structure and smart mechanism in terms of specific work domains. Examples from the medical and aviation domains will be used to illustrate how this challenge is being met.