In meta-ethics, expressivism is a theory about the meaning of moral language. According to expressivism, sentences that employ moral terms – for example, "It is wrong to torture an innocent human being" – are not descriptive or fact-stating; moral terms such as "wrong", "good", or "just" do not refer to real, in-the-world properties. The primary function of moral sentences, according to expressivism, is not to assert any matter of fact but rather to express an evaluative attitude toward an object of evaluation. Because the function of moral language is non-descriptive, moral sentences do not have any truth conditions. Hence, expressivists either do not allow that moral sentences to have truth value, or rely on a notion of truth that does not appeal to any descriptive truth conditions being met for moral sentences.
Expressivism is a form of moral anti-realism or nonfactualism: the view that there are no moral facts that moral sentences describe or represent and no moral properties or relations to which moral terms refer. Expressivists deny constructivist accounts of moral truths – e.g., Kantianism – and realist accounts – e.g., ethical intuitionism.
Because expressivism claims that the function of moral language is not descriptive, it allows the realist to avoid an error theory: the view that ordinary ethical thought and discourse are committed to deep and pervasive error and that all moral statements make false ontological claims.
Expressivism does not hold that moral sentences, as used in ordinary discourse, describe the speaker's moral attitudes. Expressivists are united in rejecting ethical subjectivism: the descriptivist view that utterances of the type "X is good/bad" mean "I approve/disapprove of X". Subjectivism is a descriptivist theory, not an expressivist one because it maintains that moral sentences are used to represent facts – namely, facts about the subject's psychological states.
Some early versions of expressivism arose during the early twentieth century in association with logical positivism.