State Spaces, Local Logics, and Non-monotonicity

This paper uses the information-theoretic machinery developed in IF 2] to relate the way error is handled in the sciences to the problems of non-monotonic reasoning. In Information Flow in Distributed Systems 2] (hereinafter IF) Jerry Seligman and I develop a mathematical theory of information. One problem tackled there that of accounting for the sorts of exceptions to law-like regularities associated with information ow that are also traditionally associated with non-monotonic reasoning. Another problem addressed is that of reconciling everyday common sense theories with scientiic theories. Again, non-monotonicity becomes an issue one has to face since if human reasoning is essentially non-monotonic but scientiic reasoning is classical (hence monotonic), there would seem to be a serious obstacle to reconciling the two. In this paper I want to use the tools from IF to further explore these issues.