Philosophers’ volume 17, no. 7 Abstract Imprint april 2017 The idea that there is a fundamental difference in value between persons and things, and that respecting this difference is an impor- tant moral requirement, has strong intuitive appeal. Kantian ethics is unique in placing this requirement at the center of a moral system KANTIAN DECISION and in explicating the conditions for complying with it. Unlike chal- lenges to Kantian ethics that focus on tragic cases that pit respect for one person against respect for another, this paper focuses on the ques- tion of how we can respect the value distinction between persons and MAKING UNDER things under conditions of uncertainty. After exploring why decision making under uncertainty is a neglected topic among Kantians and demonstrating how uncertainty challenges our ability to comply with UNCERTAINTY: this norm, we propose a notion of morally insignificant risk within a framework that allows agents to navigate real-world decisions involv- ing material benefit and some risk to dignity without violating the DIGNITY, PRICE, Kantian’s core commitments. We conclude by exploring some of the challenges facing this approach. 1. Introduction AND CONSISTENCY There is something compelling in the idea that there is a fundamen- tal moral difference between persons and things, and that a significant portion of ethics is concerned with ensuring that our treatment of our- selves and others reflects this distinction. Kantian ethics places this Adam Bjorndahl, Alex John London, intuitively appealing idea at the center of its moral system. A common challenge to Kantian frameworks concerns how they deal with situa- Kevin J.S. Zollman tions in which all the choices open to an agent fail to respect the value of some individual.1 There is, however, a different fundamental prob- Carnegie Mellon University 1. For instance, Marcus says, “But Kantian ethics is notably deficient in cop- © 2017, Bjorndahl, London, and Zollman ing with dilemmas. Kant seems to claim that they don’t really arise, and we This work is licensed under a Creative Commons are provided with no moral grounds for their resolution” (1980, pp. 125–126). In cases where each of the actions open to an agent will imperil the lives of Attribution-NonCommercial-NoDerivatives 3.0 License different groups of people, Kantians face the challenge of how to decide which <www.philosophersimprint.org/017007/> action to take without aggregating lives in the way a consequentialist would. Trolley problems present a similar challenge; the literature on doing/allowing, intending/foreseeing, killing/letting die, etc., represents strategies Kantians bjorndahl, london, and zollman Kantian Decision Making Under Uncertainty lem facing Kantian ethics that arises in far more mundane situations: Kantians have failed to appreciate how uncertainty applies to their how can the Kantian make decisions that do not equate the value of own moral standards. Kantian decision makers, we argue, routinely persons and things in circumstances of uncertainty? have to make choices under uncertainty. But intuitive ways of extend- Uncertainty is a ubiquitous feature of life. Decisions that are easy ing Kantian decision making to cases of uncertainty are too extreme: to make under certainty can become much more difficult and morally they trivialize decision making if they prohibit all acts that might vi- fraught under uncertainty. It is surprising, therefore, that the Kantian olate Kantian norms, or they trivialize the Kantian’s moral commit- moral tradition has so little to say about decision making under uncer- ments if they permit any act that does not violate Kantian norms for tainty. This stands in sharp contrast to consequentialism, which has a certain. Nor can familiar techniques from expected utility theory be long and robust tradition of theories encompassing uncertainty (von applied to resolve the problem; they violate the Kantian’s moral com- Neumann and Morgenstern, 1953; Savage, 1954). mitments by forcing a comparison between the value of persons and This lack of development might be explained by two distinct but things. interrelated assumptions. First, the most detailed and mathematically The fundamental challenge of decision making under uncertainty sophisticated treatments of uncertainty have been developed under for Kantian theory, therefore, is to develop a framework that allows very different assumptions than those used in Kantian moral theories. agents to meaningfully evaluate acts that risk violating dignity with- Uncertainty is usually modeled as attaching to consequences.2 Since out making invidious comparisons between the value of persons and Kantian moral theory does not evaluate actions in terms of their conse- the value of things. We undertake the first steps in this process by quences, uncertainty might seem like an orthogonal concern. Second, articulating a notion of morally insignificant risk within a framework the categorical nature of moral imperatives in the Kantian framework that allows agents to navigate real-world decisions involving material appears to offer a straightforward response to practical uncertainties. benefit and some risk to dignity without violating the Kantian’s core Because Kantian theories enjoin the agent to never perform certain commitments. types of acts, one might suppose this injunction extends easily to acts We begin in Section 2 by identifying key features of Kantian moral that involve uncertainty. If an act that has property x is morally prohib- theory, and then in Section 3 demonstrating the relevance and perva- ited, then it seems natural to extend this prohibition to those that might siveness of decision making under uncertainty within this theory. In have property x. We contend, however, that both of these assumptions Section 4 we illustrate that there is not a simple, existing solution to are misguided.3 the problem of uncertainty about an act’s moral status. In Section 5 we develop our own model for Kantian decision making under uncer- have deployed in attempting to address such issues (Steinbock, 1980; Fischer tainty, and in Section 6 we defend this model. In Section 7 we explore and Ravizza, 1992). a generalization of this model. Section 8 investigates the relationship 2. For a discussion of the consequentialist commitments of standard decision theory, see (Colyvan et al., 2010). between these theories and so-called “consistency” principles from de- 3. Jackson and Smith( 2006) have identified a similar problem for absolutist cision theory. Section 9 concludes. theories, which they take Kant to instantiate. They focus on absolute prohibi- tions, such as against killing innocent people, and argue that uncertainty can arise when the status of a person as innocent or not is unclear. They argue that absolutist theories cannot deal with uncertainty without compromising which certain trade-offs in values may represent a moral breach. Additionally, their absolutism. The model we present here is more general in that it applies although we identify a similar problem, our positive approach represents a not only to absolutist theories, but to a wider range of pluralist theories in unique attempt to incorporate uncertainty into a Kantian framework. philosophers’ imprint - 2 - vol. 17, no. 7 (april 2017) bjorndahl, london, and zollman Kantian Decision Making Under Uncertainty Although formal models of decision making have been almost the nity. What has a price can be replaced by something else as its exclusive province of consequentialist moral theory, this paper demon- equivalent; what on the other hand is raised above all price and strates that the project of formalizing a Kantian decision theory can therefore admits of no equivalent has a dignity (4:434–435). contribute significantly to this moral tradition and to mathematical A thing x might be replaceable in at least two senses. It is “token- models in decision theory. We hope the framework outlined here will replaceable” just in case there is another token of the same type that provide a foundation for further exploration in this area. could be substituted for x without loss of value. For example, any 2. Dignity, Price, and “Kantian” Theories new pair of Levi’s 501 jeans could be replaced by any other new pair of Levi’s 501s in the same size without a loss of value: they perform For our present purposes there are three definitive features of “Kan- the same function or have the same use and fill this role in the same tian” moral theories: (1) a claim about the nature and source of the manner. A thing x is “equivalent-replaceable” if there is an amount of value or status of persons that provides the operational content to (2) some other thing that can take the place of x without a loss of value. a fundamental moral norm and (3) a procedure for applying the norm The jeans are equivalent-replaceable in the U.S. by some number of directly to acts. Our claim is not that these features capture the com- U.S. dollars. Although the dollars do not serve the same function as prehensive structure of Kant’s moral theory. Instead, these are core fea- jeans, they have the same relative value or worth. For Kant, things can tures that some recognizably “Kantian” theories include. Explicating be both token-replaceable and equivalent-replaceable, and this is the these features helps to illustrate why these theories have not seriously foundation upon which markets and the use of currency are based. engaged with the impact of uncertainty on decision making. In contrast to things, Kant says that beings with dignity “admit The first defining feature of Kantian moral theory is a fundamental of no equivalent”. Presumably, this means one cannot simply replace distinction between the value of “things” and “persons”. We regard one person with another without loss of value, even if the two people this as the most important insight that Kantians should want to pre- have the same skills and abilities. In other words, people are not token- serve.