Are moral judgements merely expressions of non-cognitive attitudes?
Expressivism, in the broadest sense, argues that moral judgements’ primary function is to express non-cognitive attitudes. According to expressivism, the usage of moral terms describes and reports no new information. Rather, the usage of moral terms expresses one’s attitude toward the target of the moral terms. For example, saying “it is bad that George was robbed” only reports the fact that George is robbed, the usage of “bad” only serves to express the speaker’s disapproval toward this turn of events. Expressivism is traditionally thought of as a non-cognitivist theory, though some variants of expressivism has moved toward closer and close to cognitivism. Non-cognitivism makes two distinct claims. Firstly, non-cognitivism states that the content of moral judgements have no truth conditions in the robust sense (as opposed to the minimalistic sense). To deny that moral judgements have truth conditions in the robust sense is to deny that the contents of moral judgements purport to describe features of the world. Secondly, non-cognitivism states that moral judgements are not beliefs. Rather, moral judgements serve to express emotions, prescribe actions, endorse norms, and so on. In this essay, we will first discuss the advantages of expressivism. Then, we will move on the embedding problem. Lastly, we will consider and evaluate Blackburn’s response to the embedding problem.
Expressivism can appear unintuitive at first. After all, we clearly say things like “It is true that murder is wrong”, and “he believes that murder is wrong”, both of which appear to be incompatible with expressivism. Expressivists claim that there are three important concerns that motivate expressivism and are worth the prima facie unintuitiveness. Firstly, expressivism has a simpler ontology than non-naturalistic realism. Unlike realism, expressivism claims that there are no moral facts. This stands in contrast with non-naturalistic realism where a sort of sui genesis moral facts is proposed to exist. The existence of such sui genesis moral facts can be seen as supernatural and difficult to accommodate by a scientific view of the world. In addition, having a simpler ontology itself is seen as a theoretical virtue, other things being equal, regardless of whether the theory in question is compatible with science or not.
Avoiding moral facts also grants expressivism an epistemological advantage. In contrast with non-naturalistic moral realism, expressivism need not explain how we can come to know facts that are not part of the natural world. In contrast with naturalistic moral realism, expressivism doesn’t need to explain how we can know about the bridging laws between natural facts and moral facts. Moore’s open question argument shows that we can’t identify moral properties with natural properties a priori. Since natural properties and moral properties don’t have the same meanings, moral naturalists need to explain how we can discover that they in fact denote the same properties. Expressivism avoids these difficult tasks and simply acknowledges that there are no facts or bridge laws for us to discover.
Positing that moral judgements are just expressions of non-cognitive attitudes also grants expressivists the ability to explain the motivational aspect of moral judgements. Moral theorists commonly grant that there is some close connection between motivation and moral judgements, even though the motivation might be defeasible, or the connection might only be present for rational people. Yet a close connection between motivation and moral judgement is hard to explain if moral judgements were simply beliefs. Beliefs are seen as motivationally inert according to motivational Humeanism. Beliefs influence motivations only in conjunction with other attitudes such as desires. Expressivism, on the other hand, has a much more straightforward explanation for the motivational character of moral judgements. According to expressivism, sincere moral judgements are expressions of existing non-cognitive attitudes. It seems natural that non-cognitive attitudes such as the approval or disapproval of an action have motivational power. If one constantly acts in a way that one disapproves of, then it is reasonable to question the sincerity of the disapproval. Expressivism, therefore, has an easier time explaining why moral judgements are motivating.
Note that none of the motivations give a decisive advantage to expressivism, as moral realism has responses to each of the challenges. Still, it is difficult to deny that expressivism can more naturally deal with these issues compared to moral realism. However, error theory agrees with moral realism in denying non-cognitivism. Nevertheless, it agrees with the expressivists that there are no moral facts. As a result, error theory also avoids the ontological and epistemological problems realists face. Deciding between error theory and expressivism just comes down to whether expressivism’s version of non-cognitivism better accounts for the moral language usage than moral cognitivism.
One argument that aims to show the inadequacy of expressivism in accommodating common moral practices is called the embedding problem. The problem is that our moral claims can be used in many contexts. We can assert a moral claim or embed it in complex sentences such as conditionals. Moral reasoning often require that the same meaning is assigned to the moral claims regardless of whether it is used in an assertive context or in complex sentences. For example, if I argue that:
1. Cats are bad.
2. If cats are bad, you shouldn’t get a cat.
Therefore, you shouldn’t get a cat.
It appears to be a valid argument. But the validity of the argument depends on the clause “cats are bad” having a common meaning across moral judgement 1 and 2. It is straightforward for cognitivists to explain why this is the case. “Cats are bad” is a proposition that attribute a property of badness to the cat. In both sentences, “cats are bad” means the same proposition. For expressivists, however, asserting that cats are bad is to express an attitude of disapproval toward cats. Yet, in judgement 2, the clause can’t be expressing an attitude toward cats. The clause appears in the antecedent of a conditional where no factual attitude is plausibly being expressed. That an attitude need not be present for proposition 2 to be accepted is evident from the fact that people who don’t hold a disapproving attitude toward cats can still accept the second proposition. Attitudes also change depending on the context under which it is used. If these moral claims were just expressing attitudes, the attitudes being expressed would consequently be different.
Moreover, standards theories of logic give an account of how the truth conditions of propositions relate to each other and constrain the possible truth values of other propositions. It seems clear that standard theories of logic wouldn’t apply to moral judgements, if they were just expressions of attitudes with no truth conditions. Since we clearly do reason about moral judgements and are convinced by logical arguments in ethics, expressivists need to find a way to preserve and explain the logical relations between moral judgements and other judgements that embed them.
Blackburn’s response to the embedding problem is to identify attitudes as the common element between contexts. In an assertive context, I am avowing an attitude. In other contexts, the same attitude is entertained or discussed. To use the cat example, I’m saying that if I were to disapprove of cats, I would also disapprove of you getting a cat. To address the logical relations between moral judgements, Blackburn sees complex sentences as higher order attitudes toward combinations of attitudes. “If cats are bad, you shouldn’t get a cat” can be translated as an approving attitude toward the disapproval of cats when it is followed by a disapproval of you getting a cat as well as a disapproving attitude toward violations of the combination, such as when the disapproval of cats is followed by a lack of disapproval toward people getting cats. If Blackburn is right, then expressivists can construct a normative space where some combinations of attitudes are normative, and some combinations are not, and they can construct the normative space in such a way that mirrors propositional logic.
Unfortunately for the expressivist, Blackburn’s solution isn’t completely satisfactory. Suppose a person agrees with the moral judgements 1 and 2 but denies the conclusion. According to common sense, I suggest that there is something objectively non-normative about the inconsistency displayed by this individual. However, the expressivist fails to provide a satisfactory account of why such inconsistencies are non-normative. Blackburn is committed to the claim that if you disapprove of a combination of attitudes, then it is non-normative for you to hold such a combination of attitudes. The problem is that it is not entirely clear what it means to be non-normative in this context. If we were to think of moral judgements as beliefs, then what is normative is just what is truth-preserving. But this clearly doesn’t apply to attitudes. Blackburn cannot appeal to some kind of irreducible normativity, because that would go against the ontology of expressivism. The only solution open to expressivist seem to be to appeal to some other attitude, whether it is an even higher-order attitude, or an attitude toward some goal that would be maximally achieved by being consistent. If normative here just means approved by some other attitudes, then it is not clear if these attitudes can serve as a satisfactory basis for normativity. One way to think about normativity is that what is normative should be justified. Yet, if normativity just traces back to some ultimate attitude for the expressivist, as it must do to avoid regression, the ultimate attitude would seem to be arbitrary and doesn’t seem to be justified itself (particularly because there are no facts that can support it according to expressivists). Since the ultimate attitude is not justified, the entire system that is supposedly justified by it also seems unjustified. If the relationship between higher and lower order attitudes is not that of justification, the alternative seems to be a causal relationship. Yet, if the relationship is merely causal, then higher order attitudes have very little to do with normativity at all. In moral practices, people clearly do think of some moral judgements as justified. If expressivists are committed to say that what people actually mean is either that their moral judgements are grounded in some set of ultimate attitudes that they possess or that their moral judgements are caused by some set of attitudes, then their position clearly don’t seem very attractive. Even if Blackburn does work out a way to base normativity entirely on attitudes of a person, it only explains why holding 1 and 2 but denying the conclusion is non-normative for people who happen to share certain higher-order attitudes. Yet, common sense requires that everyone ought to find holding 1 and 2 but denying the conclusion non-normative.
In conclusion, we discussed the motivations behind expressivism and my view that whether expressivism can accommodate common moral language practices is the key factor that decides between it and error theory. We discussed embedding problem and why it prima facie appear to be problematic for expressivism. Then, we discussed Blackburn’s solution to expressivism and why it appears to be unable to account for the normativity of some moral inferences but not the others.
Blackburn, S. Spreading the Word. Chapter 5 & 6.
Blackburn, S. Ruling Passions (OUP 1998). Chapters 3
Shafer-Landau Moral Realism ch. 1, 5.
Van Roojen, “Moral Cognitivism and non-Cognitivism” The Stanford Encyclopedia of Philosophy. http://plato.stanford.edu/entries/moral-cognitivism/