Unreliable probabilities, paradoxes, and epistemic risks

Abstract

This paper explores the pros and cons of classical theories of rational decisionmaking and so-called generalized theories of decision-making. It argues that even if they teach us a great deal about rationality and sound decision-making, and are very useful tools, theories of this kind have serious limitations. In particular they breakdown in the presence of epistemic risk and where there is value uncertainty. It is argued that in these situations we need to take a Socratic approach to risk analysis and risk management.