Moral and epistemic evaluations: A unified treatment

Abstract

A rich tradition in metaethics seeks to explain the meaning of moral language in terms of desire-like attitudes. This approach can be implemented in di erent ways. On a contextualist implementation, moral discourse describes the desire-like attitudes of some agent(s), for example, the speaker or the speaker’s community. On an expressivist implementation, moral discourse does not describe desire-like attitudes; it expresses them. On a relativist implementation, the truth-value of a moral assertion at a context of assessment depends on the desire-like attitudes of the assessor—that is, someone assessing the utterance for truth or falsity. Despite these di erences, all such “attitudinal metaethics” are bound by a common thread: they analyze moral discourse in terms of conative states. Attitudinal metaethicists sometimes propose extending their approach to other varieties of normative discourse, including epistemic discourse.1 A generalized attitudinal semantics along these lines carries obvious attractions. But it also faces important challenges. An initial challenge concerns how to even spell out a generalized attitudinal semantics. While much ink has been spilled in pursuit of a precise attitudinal semantics for moral discourse, the extension to other fragments of normative language has not received a comparable degree of attention. Second, and more worrisome, some philosophers have argued that there are principled obstacles to the very idea of a generalized attitudinal semantics. For example, Boult and Köhler 2020 argue that a generalized attitudinal semantics is under-motivated, since the primary arguments for an attitudinal metaethics do not carry over to the epistemic domain. And Wodak 2017 argues that a generalized attitudinal semantics over-predicts disagreements across normative domains. These obstacles can be used to frame a dilemma for any attempt to generalize an attitudinal semantics. On the one hand, generalizers need to show that there are su cient commonalities between di erent normative domains to warrant a uni ed treatment. On the other hand, generalizers had better not erase the obvious di erences between di erent avors of normative judgment.