The primary aim of this project is to develop a formal framework for the interaction between defeasible assumptions, norms, and the evaluation of actions. Its secondary aim is to compare this framework to existing quantitative approaches to uncertainty in the context of normative reasoning.
In the framework, assumptions will be represented by object-level expressions, rather than by abstracta such as probability measures or preferences over models or scenarios. This emphasis on explicit representation is motivated by the view that such abstracta are often not available or even nonsensical in realistic cases, and hence that we should develop an alternative to constructions that use them. If we want to distinguish between different types of such assumptions as they bear on our evaluations, working with object level assumptions is also necessary.