CamPoS: Cambridge Philosophy of Science

Philosophy of Probability and Statistics

Cambridge has made a contribution to the philosophy of probability and statistics that is perhaps unrivalled. Some of the most important thinkers in probability and statistics have worked here, including John Venn, who developed the frequency theory of probability as well as Venn diagrams; John Maynard Keynes, who worked on the foundations of probability theory, among many other things; Frank Ramsey, who developed the notion of subjective probability; Sir Ronald Fisher, the founder of much of modern statistics, whose work has been of great significance for genetics and evolutionary biology; and Sir Harold Jeffreys and Dennis Lindley, who made important contributions to Bayesian probability and statistics.

More recently, Richard Braithwaite, Ian Hacking, and Hugh Mellor have done fundamental work on the subject and the University remains a centre for important research in the foundations of probability. This research deals with ‘subjective’ questions about credence, evidence, and confirmation: Can probabilities measure degrees of belief? Do the rules of probability have some normative application to changes in belief? Do they have rational bearing on practical reasoning (decision theory)? Other, ‘objective’ questions concern chance and frequency: Is there an empirically meaningful notion of chance? Does it place any constraints on rational belief-formation or on rational choice? And does the answer have implications for the interpretation of quantum mechanics?

Across the University, Cambridge continues to be a lively place for asking important questions about the foundations of statistics too. In the Statistical Laboratory, for example, David Spiegelhalter has examined the potential role of a subjectivist Bayesian approach within everyday statistical practice. Meanwhile, Philip Dawid has investigated the logical bases of, and the relationships between, several current approaches in statistics, as well as evidential and causal reasoning from statistical data. 

Arif Ahmed
October 2012