How Would Expressivism Change Value Uncertainty?
1. Separate four questions, not one
When evaluating expressivism, this page distinguishes at least four questions.
- Semantics: what, literally, is asserted by sentences such as “X is wrong” or “Y ought to be done”?
- Speaker psychology: what beliefs, attitudes, plans, emotions, or commitments does a subject have when making such a judgment or utterance?
- Pragmatics and social function: how is the utterance used for blame, coordination, attitude regulation, shared planning, reputation, or other social purposes?
- Ontology: do reasons, values, or normativity exist independently of present human moral practice?
These questions interact, but they are not identical. In particular, the causal origin or social function of a linguistic practice need not directly determine the semantic content of its sentences. Nor does a semantics for human moral sentences immediately determine the ontology of normativity as a whole. The argument below centers on these two separations.
2. What expressivism captures remains important
Expressivism strongly captures the practical character of moral language. In actual use, “X is wrong” is often connected with opposition to X, blame, plans to avoid X, and coordination of norms with others. Sophisticated non-cognitivist theories after Blackburn and Gibbard have tried to preserve this practical character while explaining moral talk of truth, facts, logical inference, and disagreement.
This page therefore does not deny that moral language has expressive, planning, and coordination functions. Those functions may also be important in evolutionary, historical, and social explanations of how moral practices arise and persist.
But an additional argument is needed to move from that point to the stronger claim that the semantic content of moral sentences itself consists in attitudes or plans.
3. Function is not automatically meaning
An utterance can regulate action without its semantic content being an instruction. “It will rain tomorrow” can cause people to carry umbrellas or change their plans. That practical function does not make the sentence mean “carry an umbrella.”
Likewise, even if moral language historically developed for blame, coordination, attitude regulation, or commitment signaling, and even if it still performs those functions, it does not follow merely from those facts that the semantic content of “X is wrong” is identical to “disapprove of X” or “adopt a plan not to do X.”
Causal genealogy, speaker psychology, pragmatic function, and semantic content should be evaluated separately. Explaining the deep motives or social functions of moral practice is not yet the same task as explaining what ordinary moral sentences mean.
4. “Expression” is not a transparent readout of inner attitudes
At the same time, it would be too strong to object that expressivism requires every moral utterance to transparently reveal the speaker's actual current attitude. People can lie, speak under social pressure, misidentify their own attitudes, or use conventional expressions insincerely. A linguistic form can be conventionally associated with a type of attitude or commitment without every token utterance faithfully displaying the speaker's psychology.
Therefore, the mere possibility of deception, performance, self-deception, or unintended expression is not a direct counterexample to sophisticated expressivism.
The stronger question is whether a fully sincere and competent subject could genuinely judge that “X is wrong” while lacking the non-cognitive attitude that a particular expressivist theory treats as essential to that judgment.
5. Sincere moral judgment can lack matching motivation
Consider a subject who sincerely judges that “X is wrong” while also saying “I will do X anyway,” “I do not want to stop doing X,” and “I have no intention of preventing others from doing X.” This differs from mere lying or performance. The subject may have no intention of retracting the normative judgment and yet no motivation to comply with it.
This possibility connects to the classical moral amoralist and debates about moral motivation internalism and externalism. This project does not, at least conceptually, identify normative judgment with motivation. The same separation is developed independently in the thesis on normative judgment and goal revision.
A contemporary expressivist can respond that the relevant attitude need not be a simple desire or immediate motivation; it may instead be norm acceptance, a planning state, or a higher-order commitment. The mere possibility of an amoralist therefore does not refute expressivism by itself.
But that reply raises a further question: what makes such “acceptance” or “commitment” non-cognitive? If acceptance includes storing the content “X is wrong” in one's model of the world as true or correct, the analysis becomes highly belief-like. If it instead requires a substantive connection to plans, desires, or endorsement, the problem of a sincere but unmotivated subject returns.
6. Moral utterances can arise from an internalized worldview
A further point is that moral utterances need not arise directly from a subject's active value choice or present will. Through education, religion, culture, institutions, and repeated discourse, people can internalize a normative worldview according to which certain things are right, wrong, obligatory, forbidden, or sinful.
A subject may have no memory of choosing such a norm and no desire to comply with it, yet sincerely judge that “X is sinful” or “Y is unjust.” A cognitivist model can describe this in the same basic way as other world knowledge: the subject retrieves a proposition from an internalized world-model.
The causal history of that belief may involve coordination, sanctions, imitation, emotion, and evolved dispositions. But the social genealogy of a belief is not the same thing as the semantic content of that belief.
This project therefore allows that moral language may have deep expressive and social functions while remaining skeptical of moving those functions directly into the semantics of ordinary moral utterances.
7. Realist-seeming discourse and quasi-realism
Ordinary moral discourse often behaves more strongly than a report such as “I dislike X.” It is natural to say that “even if I and my whole society had approved X, X would still have been wrong,” or that another person is not merely differently disposed but mistaken.
Yet this does not straightforwardly refute expressivism. Quasi-realism is precisely the project of asking how far talk of truth, objectivity, error, logical inference, and apparent mind-independence can be recovered from a non-cognitivist foundation.
The relevant distinction is between being able to reconstruct those linguistic behaviors and providing the simplest and best explanation of them. As more theoretical machinery is used to recover realist-seeming assertions, inference, sincere amoralists, and internalized normative world-models from a non-cognitive basis, expressivism must separately show what explanatory advantage it retains over the cognitivist alternative of treating normative judgments as belief-like from the start.
This is not the claim that quasi-realism is self-contradictory. It is the more modest point that reconstructability and best explanation are different standards, and the explanatory costs of competing theories should be compared.
8. Do not treat folk morality as uniformly realist
None of this requires claiming that ordinary people uniformly understand morality as universal objective fact. Experimental work on folk metaethics finds substantial objectivist responses in some cases but also more relativist responses depending on the issue, cultural distance, and how disagreement is framed.
A more defensible claim is: human moral practice clearly contains uses that demand correctness beyond mere personal attitude and sometimes beyond the speaker's community or era. But that commitment is not necessarily uniform across all people and all moral domains.
Accordingly, judging a cognitivist analysis to be more natural does not by itself commit this project to universal folk moral realism.
9. Human moral semantics does not directly determine value ontology
Suppose every semantic objection above fails and the present meanings of humanity's vocabulary of “good,” “bad,” and “ought” are in fact best explained in expressivist terms. It still would not immediately follow that no normative facts exist independently of human practice.
There is no guarantee that the activities humans presently classify as “morality” completely track every kind of normativity that might exist. In an extreme possibility, present human moral discourse could function almost entirely as a mechanism for coordinating attitudes while there nevertheless exist practice-independent reasons or values that humanity has not yet recognized.
This is not an argument for realism. It is an anti-closure claim: semantic success for expressivism alone cannot close the possibility of unknown objective normativity.
10. Ethics might exist outside what we currently treat as ethical life
Even if objective normativity exists, there is no guarantee that it maps closely onto human guilt, fairness intuitions, coordination habits, present concepts of suffering, or existing moral vocabulary. Future cognitive subjects might discover normative structure in physical, phenomenal, or relational features that we do not presently classify as ethical questions.
Accordingly, fully explaining current moral discourse would not be the same achievement as fully explaining the ontology of normativity itself. This does not claim that unknown ethics must exist. It is a methodological refusal to treat the boundaries of present moral practice as the boundaries of the space of possible value.
11. Implications for the Core and the stopping condition
An earlier version treated strong expressivism as shifting value inquiry relatively directly from discovery of external truth toward construction and reconstruction. That was too strong.
On the present view, first, this project provisionally judges that a cognitivist analysis treating ordinary normative sentences primarily as truth-apt judgments is currently the more natural account. This does not presuppose moral realism. Error theory, on which the normative propositions are systematically false, remains a live logical possibility.
Second, even if expressivism is correct about present human moral language, the study of existing morality is not identical to value inquiry as a whole. A large part of existing morality may concern the construction and revision of attitudes, plans, institutions, and reflective endorsements while inquiry into practice-independent normativity remains a distinct layer.
Third, expressivism about human moral language alone does not defeat the stopping condition for foundational normativity. Ruling that out would require the additional ontological thesis that no normativity exists independently of human attitudes, practices, or constructions.
Conversely, if a future cognitive subject were to discover, independently of present moral vocabulary, normativity that itself supplied reasons either at the foundational layer or as an overwhelmingly strong best explanation, that discovery could still substantially update the Core.
12. Provisional position
- Accept that moral language has expressive, planning, and coordination functions, but do not identify those functions with semantic content.
- Separate causal genealogy, speaker psychology, pragmatics, semantics, and ontology.
- Do not treat expression as a transparent readout of the speaker's current psychology.
- Include sincere amoralists and utterances generated from internalized normative world-models among the phenomena a semantics must explain.
- Recognize that quasi-realism can recover realist-seeming discourse, while distinguishing reconstructability from best explanation.
- For now, adopt cognitivism—ordinary normative judgment as truth-apt and belief-like—as the more parsimonious working hypothesis.
- Do not infer moral realism from cognitivism. Anti-realist ontologies, including error theory, remain open.
- Do not assume folk morality is uniformly objectivist.
- Separate semantic expressivism from global ontological anti-realism.
- Do not infer the nonexistence of unknown practice-independent normativity from the constructed character of present human morality.
Separate why moral language arises and what it is used for from what its sentences mean. Then separate what human moral sentences mean from what normativity might be in the ontology of the world.
Sources / notes
For overviews of non-cognitivism, expressivism, and quasi-realism, see the Stanford Encyclopedia of Philosophy entries “Moral Cognitivism vs. Non-Cognitivism” and “Moral Anti-Realism”. On moral judgment and motivation, amoralists, and internalism / externalism, see SEP, “Moral Motivation”. On hybrid or ecumenical expressivism, see work by Michael Ridge. On variation in folk metaethics, see experimental work including Geoffrey P. Goodwin and John M. Darley, “Why Are Some Moral Beliefs Perceived to Be More Objective Than Others?” For the possibility of practice-independent moral realism, see SEP, “Moral Realism”. This page does not claim a comprehensive refutation of expressivism. It first examines the move from practical and social function to semantics, and then identifies the additional premises required when moving from a semantic account of moral practice to a global ontology of normativity.