Research Seminar in Decision and Action Theory: Ron Aboodi (LMU/MCMP)

Title:

Justifying and Deploying Ordinal Intertheoretic Value Comparisons

Abstract:

When we try to decide among multiple options in a choice-situation, we often try to evaluate them; asking ourselves something like which of these options are better, or more choiceworthy. And often we justifiably remain uncertain about the answer. For example, an educator could be justifiably uncertain about whether it is better to devote class time to cultivating sensitivity to animal suffering or to social injustice. I focus on epistemic states wherein the uncertainty is ultimately about a purely normative matter, such as whether animal suffering is as intrinsically bad as human suffering. Such purely normative uncertainty could sometimes be analyzed as a credal distribution among incompatible normative views, or theories, or families of theories (assigning to each a different value or rank that represents the relative likelihood of it being authoritative). For example, our educator’s credence could be divided between two (families of) consequentialist moral theories — one that claims that animal suffering is as bad as human suffering, and another that considers animal suffering to be negligible when compared to human suffering. Here the familiar problem of intertheoretic value comparisons might arise: Could the badness of animal suffering according to one normative theory be comparable—at least in ordinal terms such as greater or equal—to the badness of animal (or human) suffering according to a different normative theory? The question can arise for non-consequentialists as well: Could the severity of the wrongness of hurting animals according to one theory be comparable to that of hurting animals according to a different normative theory? More generally, can one justifiably compare values, or choiceworthiness levels, across normative theories? If so, how should one deploy such comparisons in decision making? I will refer to such comparisons as “Intertheoretic Value Comparisons”; hereafter “IVCs”. I will focus on the worry that IVCs can never be justified, and the worry that there can be no justifiable, systematic method to deploy IVCs in practice. Against these worries, I will argue that a certain kind of IVC is practically deployable and, in a certain type of epistemic state, justifiable. In epistemic states of this type, the normative uncertainty stems from indecisive normative intuitions, and the agent constructs each normative hypothesis on the basis of a different (internally consistent) subset of her intuitive normative judgments (or “seemings”). I illustrate how—in some such epistemic states—an agent could justifiably construct normative hypothesis “a” partly by reference to a value of an option in normative hypothesis “b”, so that a’s propositional content already includes some IVCs. My argument is based on (and perhaps illuminates) the perspectival nature of normative-uncertainty decision theory, which renders the justification of IVCs dependent on the agent’s evidence (normative intuitions included). Furthermore, I will argue that, in some—if not all—of the relevant epistemic states, one’s indecisive normative intuitions justify a credal distribution only among normative hypotheses that contain merely ordinal IVCs, without cardinal IVCs. Accordingly, I will focus on ordinal IVCs, unlike the vast majority of the relevant literature. Ordinal IVCs can be deployed in practice by eliminating stochastically dominated options, without maximizing expected intertheoretic value (a procedure which has been the target of most criticisms of IVC).