Expressivism in Ethics

Core Idea

Expressivism holds that moral utterances do not express beliefs or state facts, but rather express non-cognitive mental states such as emotions, attitudes, or prescriptions. Early emotivism (Ayer, Stevenson) said 'murder is wrong' means roughly 'boo to murder.' More sophisticated versions, such as Blackburn's quasi-realism and Gibbard's norm-expressivism, try to explain how moral discourse can mimic truth-apt language without positing moral facts. Expressivism faces the Frege–Geach problem: if moral terms in assertions are non-descriptive, what do they contribute when embedded in conditionals ('if lying is wrong, then getting children to lie is wrong')?

Explainer

From your introduction to metaethics, you know the central divide: moral realism holds that moral sentences express beliefs about mind-independent moral facts, while non-cognitivism holds that moral sentences do something other than state facts. Expressivism is the most developed non-cognitivist position, and its central claim is deceptively simple: when you say "cruelty is wrong," you are not describing the world — you are expressing an attitude of disapproval toward cruelty. On this view, moral language is more like saying "boo to cruelty!" than like saying "cruelty causes suffering." No moral facts are needed, no moral perception faculty is required, and no difficult metaphysics about the nature of goodness is necessary.

Early emotivism (A.J. Ayer in *Language, Truth and Logic*, 1936; C.L. Stevenson in *Ethics and Language*, 1944) made this claim in its bluntest form: moral utterances are expressions of emotion and attempts to influence others' emotions, with no descriptive content at all. This position has obvious appeal for anyone attracted to a scientifically austere worldview — it eliminates mysterious moral facts while explaining why moral disagreement feels so persistent and heated (we are expressing and contesting attitudes, not discovering facts). But it faces an immediate objection: we do not *talk* as though we are merely venting emotions. We argue, reason, change our minds, call each other right and wrong, and use moral terms inside complex logical structures. If "lying is wrong" just means "boo, lying!", what does it mean embedded in a conditional?

This is the Frege-Geach problem, named for philosopher Peter Geach (building on Frege's logic). Consider the valid argument: (1) Lying is wrong. (2) Getting your children to lie is getting them to do something wrong. (3) Therefore, getting your children to lie is wrong. This looks logically valid. But if "lying is wrong" in premise 1 just expresses an attitude (rather than asserting a proposition), it cannot serve as a premise in a logical argument — the same expression in premise 2 appears inside a conditional, where no emotion is being expressed. The logical structure requires that "lying is wrong" mean the *same thing* in both positions, but emotivism has no account of what that constant meaning could be.

Simon Blackburn's quasi-realism and Allan Gibbard's norm-expressivism are sophisticated responses to this problem. Blackburn argues that we can "earn the right" to talk as if there are moral facts — including using truth-apt language, making inferences, and claiming objectivity — without actually positing moral facts. We project our attitudes onto the world and then regulate them through social norms of rational coherence, producing discourse that mimics factual language. Gibbard analyzes moral judgments as expressions of *norms* for governing action and reactive attitudes: to say an act is wrong is to express acceptance of norms that forbid it. Both approaches try to show that the logical and epistemic features of moral discourse (inference, revision, testimony, disagreement) can be explained in purely expressivist terms, without invoking moral facts. Whether they fully succeed — whether quasi-realism "earns the right" to moral truth-talk without quietly smuggling in realism — remains the central contested question in contemporary metaethics.