The EPIMP project aims to pin down the accuracy-centered properties of imprecise probabilities, describe reasonable ways to measure these properties, and use the resulting accuracy measures to justify and extend imprecise probability theory.
Over the past years, Arthur has been investigating the foundations of choice functions---a very general imprecise-probabilistic framework. He has done pioneering work on irrelevance, exchangeability, indifference, and many other important notions for choice functions.
Kevin has been exploring questions about rationality, learning, and decision-making: how should we respond to evidence, and how can we make effective use of the information we have? He has developed a new rule - Continuous Probability Kinematics - that specifies how to respond to a heretofore unexplored type of learning experience.