Decision Under Normative Uncertainty

• In Event: Normative Formal Theory

Abstract

How should we evaluate options when we are uncertain about the correct standard of evaluation? Such normative uncertainty differs from ordinary empirical uncertainty and raises new challenges for economic policy making, decision theory and ethics. The debate, which has originated in contemporary meta-ethics, is of high interest for economics also since world economics leaders have seriously been keen on integrating moral concerns in their social science (Tirole, Bénabou, Fleurbaey) but also in their practice within international institutions such the World Bank, OECD and through commission (for instance Stiglitz-Sen-Fitoussi French Commission), the IPSP Panel. So far, within political economy and economic theory, researchers have dedicated effort to modeling and managing empirical uncertainty. We wish to bridge a gap between both disciplines, ethics, which has been interested exclusively in normative uncertainty, and political economy, which has been interested exclusively in empirical uncertainty.
Suppose we must evaluate some choice options. The criterion could take several forms, for instance personal well-being, utility in a suitable sense, moral value, social welfare, legal value, or artistic value. In this evaluation, one can face two fundamentally different types of uncertainty. Decision theory has so far focused on empirical uncertainty: uncertainty about empirical facts, states, or consequences. A doctor may be uncertain whether a particular treatment cures the disease. Philosophers have recently turned to so-called normative uncertainty: uncertainty about the correct standard of evaluation itself (e.g., Oddie 1994, Lockhart 2000, Jackson and Smith 2006, MacAskill 2014, Bradley and Drechsler 2014, Ross 2006, Sepielli 2006, 2009, Weatherson 2014, Grieves and Ord 2017, Lazar 2017, MacAskill and Ord forthcoming.) Even if the treatment is certain to cure the disease and known also in all its other empirical features, so that no empirical uncertainty whatsoever exists, the doctor can be uncertain about the moral value of the treatment, perhaps by wondering about the moral trade-offs between recovery and (fully known) side effects of the treatment, or by hesitating between utilitarian and deontological evaluations. Normative uncertainty can be more pressing and harder to resolve than empirical uncertainty which makes it more surprising that formal decision theory so far neglects normative uncertainty. The most widely discussed proposal is to form the expected value of options, relative to correctness probabilities of competing valuations. We show that the expected-value theory is just one of four natural expectation-based theories. These theories differ in their attitudes to normative risk and to empirical risk. The ordinary expected-value theory imposes neutrality to normative risk, whereas its attitude to empirical risk is impartial, i.e., determined by the risk attitudes of the competing valuations deemed possible. The three new theories are, respectively, neutral to both types of risk; impartial to both types of risk; or neutral to empirical but impartial to normative risk. We conditionally defend the theory which is impartial to all risk in the impartial value theory on the grounds that it respects risk-attitudinal beliefs rather than imposing an ad-hoc risk attitude. Meanwhile, our analysis shows how one can address empirical and normative uncertainty within a unified formal framework, and rigorously define risk attitudes of theories.

Authors

• Brian Jabarian

• Franz Dietrich, Paris School of Economics & CNRS