Cognitive Models in Computing: Results and Directions

Eric Horvitz, Microsoft Research

I will present research on harnessing statistical methods to model properties of human attention and memory–and on harnessing such cognitive models in computing applications. I will focus first on the construction of predictive models of workload and interruptability. I will review user studies of disruptions in computing settings, and discuss the Priorities, BusyBody, and Notification Platform projects. Then, I will turn to efforts to build probabilistic models of memory landmarks, reviewing work on the MemoryLens project. I will present several prototype applications to demonstrate how such models can be applied. Finally, I will discuss longer-term directions with the use of probabilistic methods to model multiple aspects of human cognition.

Speaker Biography

Eric Horvitz is a Senior Researcher at Microsoft Research, where he manages the Adaptive Systems and Interaction group. His interests include principles of sensing, learning, and reasoning under uncertainty, and applications of probability and utility in problem solving, communications, and human-computer interaction. He is an Associate Editor of the Journal of the ACM, Chairman of the Association for Uncertainty in Artificial Intelligence, and serves on the Naval Research Advisory Committee (NRAC). He has been elected a Councilor and Fellow of the American Association for Artificial Intelligence. He received PhD and MD degrees from Stanford University.