Notions of rationality and uncertainty should – ideally – be underpinned by appropriate logics, not just widespread beliefs. Some key works are:
- Kant: on the limits of knowledge, the seductions of subject-object languages and the dangers of a priori probability estimates.
- Mill: on the limits of induction, e.g. as would be needed to justify numeric assessments of probabilities.
- Boole: on the limits of logic, the tendency for ‘almost certain’ beliefs to result in falsehoods when used in long chains of logic, and the limitations of probabilities.
- Russell builds on Keynes’ Treatise on Probability (which followed Boole), discussing the link between conventional probability and ‘normal science’, and their assumptions. (Russell and Keynes had studied under Whitehead.)
- Good: on technical developments needed to further the approach of Keynes.