Epistemic Risk : Issues on the normative basis of risk analysis
Abstract: The articles included in this thesis centre on what can be called the normative basis of risk analysis. Since risk analyses are decision procedures, they should adhere to norms of rational decision-making, and this they do. Most risk analysis schemes are built on a notion of the decision theoretic maxim that the goal of every decision is to maximize expected utility. However, for some thirty years now this maxim has been called into question. The idea is not that it is unreasonable in itself, but that the theoretical construct of which it is a consequence is in some way inappropriate. Idealizations of the theories are criticized for being too excessive. They do not properly cover cases where the decision maker has poor or, as it were, imprecise knowledge. These are situations where risks are not only about the consequences of the decision but also there are risks with the knowledge itself, epistemic risks. If it is important for a normative decision theory to be not only ideal as a theoretical construct, but to have, in addition, a prescriptive purpose, that theory should also be applicable to actual decision makers and to somehow deal with epistemic risks. The first two papers are about what kinds of decision theory there are, and how best to construe them. The first describes the differences between normative, descriptive and prescriptive decision theories. The second asks whether there is anything intrinsic to the theory that makes it normative or descriptive. Paper 3 asks on what grounds risk assessment is required to be separated from risk management in the risk analysis process. It is argued that the separation is there to ensure that epistemic values are held apart from non-epistemic values. The paper ends by arguing that, in efforts to improve risk management decision-making, it is more important to be explicit about scientific uncertainty than it is to separate sets of values. Paper 4 argues for a more complete decision-theoretic basis for risk analysis. The contrast here is with the bifurcated process of maximizing expected utility when sufficient knowledge is available and using the precautionary principle in cases where knowledge is more limited. Paper 5 investigates the notion of epistemic risk in detail.
This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.