Mixed-Initiative Transactions: A Cognitive Engineering Approach to Interface Agent Modeling

The fundamental goal of mixed initiative systems is to enhance joint cognitive work among humans and computer agents. The designs of these systems have leveraged knowledge and principles form the areas of linguistics, cognitive science, and computer science. In this article we suggest that knowledge from the field of cognitive engineering is also relevant to the design of mixed initiative systems. We present design principles for mixed initiative agent modeling that are motivated from this perspective. Here we discuss these principles and present a computational architecture based, in part, on them.