It is now that Wedgwood posits his favored semantic theory for normative judgments, drawing inspiration from proponents of conceptual role semantics to argue that the nature of a concept is its role in thought and reasoning, and from this that the essential role of normative concepts is a regulative role in practical reasoning. This semantic theory is well suited to accommodate the internalism Wedgwood argued for earlier, as on this view, what makes normative concepts what they are is the function they play in deciding what to do. Pushing further, Wedgwood posits that the essential role of “ought” as a concept must be that of committing the thinker to an intention, in that if a speaker notes “I ought to x”, the proper function of ‘ought’ in the mind of the speaker will yield the intention to x, such that one is rationally committed to producing the intention if one is using the concept correctly in reasoning.
Wedgwood quickly disarms an objection that flies up according to which concepts only play a representational role, where they represent states of affairs, such that they cannot play a regulative role in practical reasoning. His response is that concepts are only representational if they are in propositions, which in turn depend on the semantic values of those concepts, their interrelations, and the correct use of logical operators that allow those interrelations. But from this it only follows that the representational role of a concept is essential to that concept insofar as the essential features of a concept must determine the concept’s reference or semantic value. In other words, the essential representational role of a concept is just the features of that concept that determine it’s semantic value, so Wedgwood can avoid this objection by showing that it is the role a normative concept plays in reasoning that determines the semantic value of that concept, such as that the concept points to the appropriate ought-relation.
The more specific version of conceptual role semantics that Wedgwood favors is one in which the “nature of a concept is given by the basic rules of rationality governing its use” (82), with the result that there are rules by which it would be rational or irrational to use a concept, and this theory explains how the concept functions in the thought of the agent, as well as fixes the specific referent of the concept, the two conditions Wedgwood posited were necessary for a semantics of normative judgments in chapter 3. The non-normative example that Wedgwood provides is a disjunction, such that rules of rationality require that if one accepts p or one accepts q, then one, if not to be irrational, must accept (p or q). The rules of rationality also then require that one not deny (p or q). This view is a functionalist semantics, where a concept is the role that concept plays in yielding the correct thought output given the thought that was inputted. So in this light, the essential nature of a disjunction is the role it plays in the inference from the input thought p to the output thought q, and for Wedgwood, if acceptance of p is correct, then in virtue of the truth-preserving nature of the rule-bound role of the concept that is used for such an inference, acceptance of p or q as its conclusion is uniquely correct, in the sense that the correct output state could not be ~ (p or q).
This is still not enough to fix the semantic value of a disjunctive, though we saw how the rules governing the use of the disjunctive in reasoning came close to fixing the semantic value of the concept. The problem is that the basic rules of use for a concept (and they must be basic in the sense that only the basic rules of inference for a concept are relevant, as we want to say that though logically equivalent, ‘p or q’ and ‘not both not-p and not-q’ have different semantic content, as one could believe one and not the other) cannot be incomplete in such a way that a thinker mistakenly rejects an output belief state, because surely the semantic value of a concept entails the possibility of a set of rules that is complete in its fixing that semantic value as the referent. We should not settle for anything but this complete set of rules, then, in order to set the semantic value of a concept, and a set of rules is “complete just in case it captures every possible condition in which rejecting p or q would be a mistake” (87). So, the semantic value of a disjunctive is its correct binary operation on semantic values of smaller scope, that is to say, its semantic value is the binary role it plays on the semantic value of thoughts, where this role is completely defined to yield only uniquely correct output states.
Wedgwood’s next step is to show that ‘ought’ and other normative terms are propositional operators like ‘or’, ‘not’ ‘and’ and ‘if’, in such a way that normative terms can play a role in practical reasoning. It might be tempting to understand ‘ought’ as a relational predicate, where a course of action is indexed to a certain agent at a certain time, but Wedgwood thinks that this is mistaken, as seeing ‘ought’ as a relational predicate can lead to ambiguities of meaning in a statement, some of which we can clearly see the speaker did not mean, such that the relational-predicate approach introduces too much confusion into the discussion when it is easier to take a more unified approach of positing that ‘ought’ is a propositional operator (Wedgwood attributes this argument to Bernard Williams). In this light, “‘ought’ is an operator that attaches to a proposition” (93), and this proposition can be implicitly indexed to a person and time without introducing the problems brought in by the relational predicate approach. While there might be a grammatical barrier to translating English sentences of this type into logical form, there is no logical barrier, such that, to Wedgwood, this does not suggest a failure of his view, but rather offers the reward of straightforward truth conditions in virtue of that proposition.
‘Ought’ and Conceptual Role Semantics
Wedgwood has thus shown the outlines of the version of conceptual role semantics for propositional operators that he favors, and that ‘ought’ is a propositional operator, such that he is justified in presenting his theory of conceptual role semantics for ‘ought’. Crucially, Wedgwood notes that his reason for being drawn to conceptual role semantics for ‘ought’ is its ability to account for internalism, but because the version of internalism that he favors is narrowed to judgments of the appropriate sort, the conceptual role of ‘ought’ must be broad enough that there can be judgments of the appropriate sort, and thus judgments of an inappropriate sort. That is to say, the conceptual role of ‘ought’ must be broad enough that when judgments from this concept are of the appropriate sort, one has an intention to act in accord with that judgment, but still allow for the use of this concept ‘ought’ in ways that do not lead to this intention to act, when the judgment is not of the appropriate sort.
One of the conditions for being a judgment of an appropriate sort was that the course of action be dependent on intention, meaning that the action judged to be brought about would not obtain without the judger’s realizing it. To allow the conceptual role of ‘ought’ to be broad enough, then, it must be possible to use it when the course of action is independent of intention, and this Wedgwood does by shifting from intentions to plans. A plan is just a proposition of how you might behave at event t, and a way things might be if you behaved that way. One can plan on many things that are outside of your direct actions, such that judgments about what one ought to do can involve plans that are independent of ones intentions.
The other condition for being a judgment of an appropriate sort is that the judgment is made with no relevant uncertainty, as given uncertainty we might judge that we ought to do either A or B, but not knowing which to do, form no intentions to do either A or B. Wedgwood broadens the conceptual role of ‘ought’ by positing that we have ideal plans and actual plans, such that in this case the judgment ‘I ought to do A or B’ features in our ideal plans, even if the judgment does not feature in our actual plan, which is to do C, a second-best option. One’s ideal plan at event t, is the proposition you would hold if your practical reasoning was not affected by relevant certainty, such that when one experiences no relevant uncertainty, the ideal plan and the actual plan of the agent coincide. As such, when there is relevant uncertainty, your ideal plan separates from your ideal plan, such that your actual plan might be to C, even though your ideal plan is A or B, and though you judge you ought to A or B, you are not practically irrational to do C.
With the distinction between intentions and plans, and ideal plans and actual plans in place, we can understand the essential conceptual role of ‘ought’ in a speaker’s acceptance of “I ought to x” as being such that it commits the speaker to making x part of their ideal plans about what to do at time t. This commitment is such that if one rationally accepts this ought judgment, then one is irrational to not make x part of their ideal plan at t. Moreover, given the cognitivist argument presented in response to expressivism in chapter two, acceptance of “I ought to x” is a mental state that is just an ordinary belief, such that internalism and cognitivism are seen to be highly compatible in Wedgwood’s theory.
Having shown the conceptual role in thought of ‘ought’, Wedgwood, in order to meet the representationalist objection posited at the beginning of the chapter, shows how the conceptual role fixes the conditions of ‘ought’ in a way that sufficiently determines the referent of ‘ought’. For Wedgwood, the ‘ought’ concept must surely fix its referent as a relational property of propositions as they appear to an agent at a given time, because the conceptual role of the concept provides conditions for when the concept is correctly applied in inferences, in such a way that the concept functions to guarantee when the rules of the concept are used validly. The valid use of the concept follows rules of that concept, which must be the complete set of rules that would rule out false conclusions from reasoning from that concept. The validity of a rule can be determined by nothing short of the fact that the correctness of the input guarantees the correctness of the output, in the manner outlined before for the concept of a disjunction where the output state is uniquely correct because the opposite output state would be incorrect. Wedgwood thinks that such a negated proposition does feature in the ideal plan of the speaker, in that in judging that “I ought to x”, rationally commits the speaker to making x part of their ideal plans, and rationally commits the speaker to not making ~x (not x) part of their ideal plans. Thus, the semantic value of ‘ought’ is “the weakest property of a proposition p that makes it the case that it is correct for A to make the proposition p part of her ideal plan about what to do at t, and incorrect for A to make the negation of p a part of her ideal plan” (102).
Wedgwood notes that something further must be said about his use of the words ‘correct’ and ‘incorrect’, and his general point is that in theoretical reasoning there are rules of rationality and justification that make beliefs correct not just in the sense that they meet these standards, but also in achieving these conditions, a further goal of just “getting things right” will be achieved, wherein the goal of beliefs correctly fitting the world will be met. Wedgwood thinks that the same point is true for practical reasoning, where the end goal is not having consistent plans just for consistency sake, but because consistent plans are more likely to be correct in mapping out a plan of acting in a way that is “genuinely choiceworthy”, and this is the goal of practical reasoning.