The theory of nomic probability (Q1187227)

This readable paper provides a welcome summary of the author's views on probability and a corresponding form of defeasible inference. The basic building block is ' nomic probability' --- that is, probability as it appears in physical laws connecting projectible properties. The theory of nomic probability comprises a bit more than the classical probability calculus. Given the theory of nomic probability, Pollock provides a formulation (in terms of defeasible arguments) of the statistical syllogism. This leads to a characterization of definite probabilities --- probabilities that apply to individuals. It is important that these definite probabilities are based on indefinite objective probabilities. We thus arrive at the ``mixed physical/epistemic probabilities that are needed for decision theory.'' The next problem is the source of our knowledge of these probabilities, and the answer is `induction'. The author offers a new view of statistical inference which makes it an example of defeasible reasoning based on probabilities of just the sort he develops in the beginning of the article. --- The material here is developed more fully in [the author, Nomic probability and the foundations of induction (Oxford University Press, 1990)].