Reconciling Expressivism with Normative Inference: A Brandomian Account — Epoche B2
Expressivism and the Validity of Moral Arguments: How Much Can Inferentialism Do? Expressivists say that a moral judgement expresses an attitude rather than reporting a fact. The standard objection is that attitudes do not stand in logical relations, so an expressivist cannot explain why moral arguments are valid. Robert Brandom's inferentialism [1] , which makes the meaning of any expression consist in its role in inference, looks like an off-the-shelf answer: if meaning just is inferential role, then moral terms have stable meanings and moral arguments are valid for the same reason as any others. This essay argues that the off-the-shelf answer is not available. Inferentialism does not solve the expressivist's problem; it dissolves the setting in which the problem is posed, at a price the expressivist may not wish to pay. What it genuinely offers is narrower and still worth having: an account of what it is, at the level of public practice, to be committed to an inference. The semantic work — explaining when two attitudes clash — remains for the expressivist's own resources. The problem, precisely stated Fix the terms. A non-cognitive attitude is a mental state that is not a belief: desiring, approving, intending, accepting a norm. Expressivism holds that a moral judgement expresses such a state. A. J. Ayer's Language [2] , Truth and Logic (1936 [3] ) is the crude ancestor; R. M. Hare's The Language of Morals (1952) is the first serious version, treating moral judgements as universalisable prescriptions. Peter Geach's 'Assertion' (1965 [4] ) set out the difficulty that has organised the field since. He used an argument of this shape: If doing a thing is bad, then getting your little brother to do it is bad. Stealing is bad. Therefore getting your little brother to steal is bad. The inference is valid. Its validity requires that 'stealing is bad' means the same in the second premise, where it is asserted, and in the first, where it is not. Someone can sincerely assert the conditional while having no view at all about stealing, taking it as a claim about corrupting siblings. So the conditional expresses no disapproval of stealing. If expressing disapproval is what the sentence means, its meaning has changed between the premises, and the argument equivocates. The demand is thus not for a defence of expressivism in general. It is for a compositional semantics: an account of what a moral sentence contributes to a larger sentence containing it, given in terms of attitudes rather than facts. Hare's own attempt, an analogous logic for imperatives, is generally judged to have failed, because the entailment relations he needed among imperatives were stipulated rather than explained. The expressivist attempts Two serious answers exist, and both work by explaining logical relations as relations among mental states. Simon Blackburn's approach [5] , in Spreading the Word (1984), treats the conditional as expressing a higher-order attitude: disapproval of a certain combination of first-order attitudes — here, the combination of disapproving of stealing while not disapproving of inducing a sibling to steal. Someone who accepts both premises and rejects the conclusion is not contradicting themselves in the logician's sense; they are in a state that, by their own lights, they disapprove of. Inconsistency becomes a kind of practical clash. Allan Gibbard's Wise Choices [6] , Apt Feelings (1990) is more systematic. A normative judgement expresses acceptance of a system of norms, and its content is given by which combinations of factual and normative possibilities it rules out; two judgements are inconsistent when, between them, they rule out every combination that remains. This delivers the formal behaviour of negation, conjunction and the conditional without appeal to truth or facts, by a definite construction rather than an analogy. In Thinking How to Live (2003 [7] ) Gibbard recasts the machinery in terms of plans. These are the accounts on the table. Any claim that a further framework solves the problem must say what it adds to them. What Brandom actually says Robert Brandom's Making It Explicit (1994) is not a contribution to metaethics. It is a general theory of meaning, and its ambitions run in the opposite direction from expressivism's. Brandom inverts the usual order of explanation. Instead of starting with representation — meaning as a relation between words and world — and deriving inference from it, he starts with inference. To be a language user is to take part in the game of giving and asking for reasons. Asserting is undertaking a commitment : one becomes answerable for the claim, licensed to draw certain further conclusions and precluded from others. Participants keep score on each other's commitments and entitlements , and the content of an expression is its role in that scorekeeping. Logical vocabulary, on his account, is expressive: the conditional exists so that a commitment to an inference, previously implicit in practice, can be stated and thereby argued about. Why this does not solve the problem The attraction is obvious. If content is inferential role, then 'stealing is bad' has the same content wherever it occurs, since its inferential role does not vary with assertion; and the conditional in Geach's first premise is precisely the device for making explicit a commitment to an inference. The equivocation charge lapses. The trouble is what this costs. Brandom's inversion applies to all discourse. On his account 'the cat is on the mat' is also not fundamentally representational; its content is also given by commitments and entitlements. So inferentialism does not vindicate a contrast between moral language and descriptive language. It abolishes the contrast, by treating both in the way expressivists wanted to treat only the first. Expressivism is not the thesis that moral language is non-representational full stop. It is the thesis that moral language differs from descriptive language i