What Matters and How It Matters: A Choice-Theoretic Representation of Moral Theories

Abstract

We present a new "reason-based" approach to the formal representation of moral theories, drawing on recent decision-theoretic work. We show that any moral theory within a very large class can be represented in terms of two parameters: (i) a specification of which properties of the objects of moral choice matter in any given context, and (ii) a specification of how these properties matter. Reasonbased representations provide a very general taxonomy of moral theories, as di↵er-ences among theories can be attributed to di↵erences in their two key parameters. We can thus formalize several distinctions, such as between consequentialist and non-consequentialist theories, between universalist and relativist theories, between agent-neutral and agent-relative theories, between monistic and pluralistic theories, between atomistic and holistic theories, and between theories with a teleological structure and those without. Reason-based representations also shed light on an important but under-appreciated phenomenon: the "underdetermination of moral theory by deontic content".

Towards a Framework Combining Machine Ethics and Machine Explainability

“…First there is the question of how to model the content and ordering of principles in a more sophisticated way and how to quantify these orderings -and if one even needs to do so. After all, in light of our results, one could be inclined to switch to a framework genuinely relying on reasons if one finds a way to decouple reasons from principles ( [17] might offer a useful approach here). Second, the principle order might be context dependent.…”

Responsibility Voids and Cooperation

“…This suggests that if deontological theories are consequentializable, then deontological moral reasoning can be incorporated by amending the preferences accordingly. Second, Dietrich and List (2017 , 456) provide a reason-based decision-theoretic framework that can represent virtually any moral theory by way of a so-called “reasons structure,” “which encodes the theory’s answer to the two-part question of which properties matter and how they do so.” A moral theory’s associated reasons structure can readily be plugged into the individual-reasoning schema. 10 Instead of assessing the available options in terms of their consequences (as in (I2) and (I3)), they would be assessed in terms of their properties.…”

Dagstuhl Manifesto

“…The discussion group concluded -in agreement with a number of philosophers -that consequentialization has its formal limits. We can consequentialize some conventionally non-consequentialist theories only at the cost of stretching or redefining the notion of "consequence" (for discussions, see, e.g., Brown 2011, Dietrich andList 2016). If we are willing to build various contextual features into the notion of "consequence", then "consequentialization" becomes vacuously possible, but it will no longer be very useful for the purpose of encoding moral theories in a machineimplementable way.…”

“…This approach is an attempt to represent a large class of moral theories in a canonical format, without "consequentializing" them in a potentially trivialising manner. Specifically, Dietrich and List (2016) offer a "reason-based" formalisation of moral theories. They encode the action-guiding content of a moral theory in terms of a choice function (here they share the starting point of the standard decision-theoretic approach), which they interpret as a "rightness" function.…”

“…The discussion group recognised that the same action-guiding recommendations can often be systematised by competing moral theories (as argued, e.g., by Broome 2004, ch. 3, andDietrich andList 2016). This is a consequence of the fact that moral theories specify not only how we ought to act, but also why we ought to act in that way.…”