Noncognitivism PDF

Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 4 Noncognitivism...

Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 4 Noncognitivism 4.1 The Nature of Noncognitivism Cognitivists take substantive, first-order moral claims (like “lying is never permissible”) at face value—as genuine assertions attempting to say something true about the moral facts. To be sure, nihilists think that such claims are, in fact, always false (since there are no moral facts to be accurately described); but for all that, as cognitivists even they take such claims to be genuine (if mistaken) assertions. But noncognitivists think it is a mistake to interpret moral claims in this way. Strictly speaking (the noncognitivists say), they are not genuine claims at all, since they are not so much as even attempting to describe some putative domain of moral facts. According to noncognitivists, when we utter sentences like this, we are not actually using them to make assertions; we are doing something quite different. (Accordingly, the noncognitivist thinks that if we were being careful it would be more accurate to use scare quotes in discussing these matters. Instead of talking about moral claims and assertions, we should instead talk about moral “claims” and moral “assertions”—that is, sentences that look like claims or assertions but are not actually either. Nonetheless, since a profusion of quotation marks quickly grows tiring, I will continue to refer to the relevant sentences in the intuitive, familiar way.) If noncognitivism is right, of course, then moral realism is mistaken, since realists hold that such claims are indeed genuine assertions (and what’s more, some of them are true). Thus, noncognitivism will strike most of us as a form of moral skepticism. Admittedly, a noncognitivist need not take this view of their own position. They might claim that they are not skeptics about morality at all; rather, they only want to free us from some common but mistaken beliefs about what we are up to when we engage in moral discourse. Nonetheless, precisely be- cause noncognitivists deny that there are moral facts (why believe in some- thing that we never so much as assert the existence of?) and deny that there 80 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 are moral truths (how can there be any, if our moral utterances don’t really assert anything at all, and so aren’t even truth apt?), most of us will indeed view this position as a skeptical one. So noncognitivism is a view that we’ll want to examine with some care. At first glance, if nothing more, noncognitivism seems like a difficult position to maintain. After all, our substantive moral claims certainly look like they are making assertions; they certainly appear to be statements to the effect that the moral facts are one way rather than another. Obviously, any particular such claim might well be mistaken. But why believe that such claims are not really making assertions at all? They look like assertions. Why think otherwise? Presumably, the noncognitivist is not going to deny that the appearances are as I have just described them. Substantive moral utterances do indeed look like genuine claims. But they will insist that the appearances are deceptive. We are misled by the surface grammar. Such sentences look like ordinary assertions. They have the grammatical form of ordinary assertions. But if we dig below the surface, the noncognitivist assures us, we will discover reason to believe that surface appearances are here leading us astray. We should, I think, be prepared to concede that at least in principle something like this could be the case. For sometimes it really does seem plausible to suggest that a sentence that looks like it is making an assertion (it has the grammatical form of a statement) is actually—below the surface— doing something rather different. Consider, for example, the sentence “I order you to shut the door.” At first glance, this seems to be making a factual claim, to wit, that I am engaged in a certain activity, the activity of ordering you. But it would be silly to suggest this is what the sentence is really up to. A sentence like this isn’t really in the fact-stating business at all. On the contrary, it seems plausible to suggest that it is actually an imperative. We use sentences like this to give orders, not to state facts. So despite having the grammatical form of a statement (a form that is normally used to make assertions), in this particular instance what is really going on is that this sentence and sentences like it are used to give commands. Here’s another example. Suppose you point to someone attractive across the room and say to me, “I need to know what his name is.” Here too, at first glance this looks like a factual claim; you are telling me that you need something—information about someone’s name. But once again, it seems somewhat strained to suggest that this sentence is being used to state facts.  4.1 The Nature of Noncognitivism 81 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 Rather, it is plausible to suggest that below the surface this sentence is actually being used to ask a question (“what’s his name?”) or perhaps to give a command (“find out what his name is!”). We needn’t quibble about which of these it is (questions may themselves simply be disguised imperatives). The important point is simply that although this sentence also has the grammat- ical form of a statement, it is actually being used to do something other than to state facts. So we shouldn’t dismiss out of hand the noncognitivist’s claim that something similar is going on when it comes to substantive moral utterances. Perhaps here too we should ultimately conclude that although these sentences look like they are being used to state facts, that appearance is simply an illusion brought about by the surface grammar. Perhaps if we dig below the surface we will find that such sentences are actually being used to do something else altogether. That, at any rate, is the central claim of the noncognitivist. Note, however, that so far all we have from the noncognitivist is a negative thesis—that substantive, first-order moral sentences are not used to state facts. We don’t yet have a positive proposal concerning just what it is (according to the noncognitivist) that these sentences are being used to do. If sentences that look like they are stating facts are actually doing something else, what is the something else that they are doing? In principle, I suppose, a noncognitivist could make any alternative pos- itive proposal that they feel like. But as it happens, there is an answer that is extremely common among noncognitivists: the sentences in question are really being used to express and reveal certain attitudes. Thus, for example, according to this view my saying that lying is wrong is (roughly speaking) a way of expressing (revealing, showing, displaying) my attitude of disapproval toward acts of lying; my saying that one is required to keep one’s promises is a way of expressing my approval of acts of promise keeping (and disapproval of acts of promise breaking); saying that equality is intrinsically good is a way of expressing my favorable attitude toward states of affairs with less inequality; and saying that honesty is a virtue is a way of expressing my approval of people who are honest. And so on. (A view like this is sometimes called expressivism, to capture the thought that the relevant moral sentences are used to express attitudes, rather than to state facts. Sometimes the view is called emotivism, to capture the thought that what is expressed is a kind of emotional attitude, a strong liking or disliking of the acts (or states of affairs, or character traits) in question.) 82 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 As it happens, working out the details of this general proposal consumes a fair bit of time among contemporary noncognitivists. But for our purposes the details mostly won’t matter. What is important is that if something like this is right, then we now have a positive thesis in place concerning just what it is that substantive moral claims are being used for when we utter them. It supplements the negative thesis that such claims are not being used to make assertions. I hope it is clear that in terms of posing a threat to the moral realist, what is crucial is the noncognitivist’s negative thesis. For on the face of it, at least, there is no obvious reason why a moral realist couldn’t agree with the noncognitivist’s positive thesis, that substantive moral claims are used to express attitudes. It does seem plausible, after all, to suggest that expressing attitudes is at least one of the things that we are doing when we make such claims. That is to say, even the realist might and perhaps should agree that if, for example, I say that stealing is wrong, part of what I am doing in saying this is revealing my disapproval of stealing! It is hard to see why the realist should resist acknowledging such a plausible thought. The realist only wants to insist that this is not the only thing we are doing when we say things like this— that part of what we are doing, and indeed the central part of what we are doing, is asserting something that we take to be a fact, namely, that stealing is wrong. As the realist might put it, it is precisely because I believe it to be true that stealing is wrong—precisely because I take this to be a moral fact—that I disapprove of it. So while it is certainly true that when I say that stealing is wrong, I am indeed expressing my disapproval of it, acknowledging this fact (accepting the noncognitivist’s positive thesis) doesn’t give us any reason to accept the noncognitivist’s negative thesis, that I am not actually making any kind of assertion about the moral domain. For the realist, then, what is most centrally going on when I make a sub- stantive moral claim is that I am attempting to accurately describe a set of moral facts. To be sure, I may also be revealing or expressing my emotional attitudes when I do this, but it is the describing that is primary. But for the noncognitivist there is no such describing going on at all. There is only the expression of approval and disapproval. There is no attempt to state facts. For this reason it is important not to confuse noncognitivism with a view that holds that when I say to you, for example, that stealing is wrong, what is really going on is that I am telling you that I disapprove of stealing—that I am stating a fact about my attitudes, and saying, in particular, that I have a disap- proving attitude toward theft. On this alternative view, when I make a moral  4.1 The Nature of Noncognitivism 83 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 claim I am indeed making a factual claim after all: I am making a claim about my attitudes. I am telling you what I approve and disapprove of. We can call a view like this subjectivism. But subjectivism isn’t a form of noncognitivism at all. On the contrary, it is actually a form of cognitivism! For according to the subjectivist, when I make a moral claim I am indeed making assertions about facts (facts about my attitudes, to be sure, but facts nonetheless). And so, if I have described my attitudes accurately (if, say, I really do disapprove of theft), then what I have said is true! And if I have misdescribed my attitudes, then what I have said is false. Far from being a form of noncognitivism, subjectivism is actually a realist view, albeit one that most realists would never find acceptable. (It is, in particular, a constructivist view, since according to the subjectivist, moral facts boil down to facts about the reactions or attitudes of the relevant minds—in this instance, the mind of the speaker.) In contrast, the noncognitivist insists that when I make a moral claim I am not describing my attitudes at all. I am not telling you what I approve or disapprove of. I am, rather, showing you. I am revealing my preferences, not describing them. To help make this distinction clear, suppose that I sit down in front of you and eat a gallon of chocolate ice cream. In doing this, I obviously am not saying that I like ice cream. Indeed, I am not saying anything at all. I am however showing you that I like ice cream. I am revealing this fact to you through my behavior. I am not reporting anything about my tastes, but I am nonetheless showing you what they are. Turning next to a slightly more complicated example, suppose that I say to you, “It is raining outside.” Here I am indeed asserting something: I am asserting something about the weather. Note however that something fur- ther is going on here as well. When I (sincerely) tell you that it is raining outside, I am revealing something about my beliefs. I am revealing to you that I believe it is raining! But of course, for all that, I am not saying that I be- lieve it is raining. I am not describing anything about my psychology. I am not talking about my psychology at all, I am talking about the weather! And yet, for all that, I am revealing to you something about my beliefs. Although I am not reporting anything about my beliefs, I am showing you what they are. Similarly, then, according to the noncognitivist, when I say that stealing is wrong, I am displaying or revealing my disapproval of theft. But I am not saying anything about my disapproval; I am only showing it to you. My utter- ance is an expression of my disapproval, but it is not a statement to the effect 84 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 that I disapprove of theft. I am not reporting anything about my attitudes; I am only showing them to you. So far, the three examples are similar. But according to the noncognitivist this last case—the moral case—differs in an important way from the first two examples. In the case of eating ice cream or the case of telling you about the weather, the fact that I am revealing something about my tastes or beliefs may be largely incidental to what I am doing. My primary aim is to eat ice cream or report on the weather. But when I tell you that stealing is wrong, the fact that I am revealing my attitudes by expressing them in this way isn’t at all incidental to my fundamental aim. On the contrary, says the noncognitivist, this is my very purpose in using this language. I say these things (stealing is wrong, inequality is bad, and so on) precisely so as to show you what my attitudes are. I am deliberately revealing them to you. That’s the whole point of using moral language. There certainly do seem to be places where language functions in just this way, precisely to reveal (rather than report) one’s attitudes. Suppose you offer me some strawberry ice cream, and I reply, “Strawberry ice cream, yuck!” Here I express my dislike of strawberry ice cream, and I do it verbally. (I could of course do it nonverbally instead, by throwing it out!) What’s more, the act of expressing my attitude isn’t incidental to some other purpose, it is my very goal in using this language in the first place. I reveal my attitudes without reporting them; and the very point of having language like this (“yuck”) is to make this sort of expression possible. Similarly then, says the noncognitivist, when I say, “Lying is wrong,” the entire point of having and using sentences like this is simply to express my disapproval of lying. Moral language exists precisely to show you my attitudes by expressing them in this way. So as a very rough first approximation, we can translate “lying is wrong” as something like “boo, lying!” or “lying, yuck!” And (again, roughly) we can translate “you are required to keep your promises” as something like “promise keeping, hooray!” or “yay, promise keeping!” Of course, putting the idea this way runs the risk of making it seem as though the noncognitivist thinks that making a moral claim is doing some- thing silly and unimportant. But that’s not at all the case. The kind of approval and disapproval involved in moral cases is far from silly or trivial. For when we use our moral vocabulary, we are talking about matters incomparably  4.1 The Nature of Noncognitivism 85 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 more important to us than whether someone eats strawberry ice cream or not. We are revealing our attitudes toward acts of murder, or theft, or promise keeping; we are responding to the possibility of a society’s being just or fair, or of people being honest, or compassionate, as opposed to being cruel or disloyal; we are contemplating the thought of widespread inequality, or of people being happy rather than suffering. We care incredibly deeply about whether people perform these sorts of acts, whether outcomes have these sorts of features, whether people have these character traits and so on. For these matters are utterly central to how we live with one another and to what sorts of people we are going to be. Unsurprisingly, then (say the noncognitivists), we have a special vocab- ulary that we use only in cases that involve this deep form of approval and disapproval. It is the vocabulary of right and wrong, good and bad, forbidden, permitted, and required. It is our moral vocabulary. So that is one main thing that noncognitivists typically think we are doing when we make moral claims: we are using a special vocabulary to express the deeply felt attitudes of approval and disapproval that come into play when we are concerned with these central questions about how we are to live. There is, in fact, a second thing, closely related, that noncognitivists typi- cally think we are doing as well—beyond expressing our attitudes of approval and disapproval—when we make moral claims. We are uttering imperatives, giving commands. We are telling people to behave in certain ways, to bring about certain kinds of outcomes, to be a certain type of person. Thus, for example, part of what I am doing when I say that stealing is wrong is issuing an imperative: “don’t steal!” When I say that promise keeping is obligatory, I am saying “keep your promises!” When I say that inequality is bad, I am ordering people to “minimize inequality!” I am giving commands—and not just to others, but to myself as well. I am telling everyone (myself included) to avoid theft, keep promises, and reduce inequality. Here too, we can think of our moral vocabulary as something we reserve for commands that concern these central questions about how we want all of us to behave, and what we want all of us to bring about. If it is cold outside, I may utter the everyday imperative “shut the door!” But I won’t say anything “stronger” than this, because normally this isn’t a matter of life and death. Yet if we imagine that somehow it were—if, say, someone’s life depended on 86 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 it—then I might well appropriately insist that it is wrong to leave the door open. I would still be issuing a command, but I would be doing it using the specialized vocabulary that we save for imperatives that have this more pressing significance. So what is the function of substantive moral claims? According to the noncognitivist it is a combination of these two things: expressing (in this special way) my attitudes of approval and disapproval, and issuing commands (to myself and to others) to act accordingly. This is the positive account of our moral utterances that is put forward by the noncognitivist. There are, of course, in-house debates among noncognitivists. Some emphasize the expressive aspect of moral discourse; others, the imperatival aspect. For our purposes, I think, there is no need to choose between them. And there are any number of technical details that need to be worked out as well, once you try to take either of these basic ideas and refine them so as to make them precise. This is very much a live research program among contemporary metaethicists. But here too, I think these details won’t much matter for our purposes. (For the curious, though, here’s a quick example. How does the noncognitivist propose to translate the claim that killing in self-defense is morally permissible but not morally required? This certainly isn’t the giving of a command to kill in self-defense (since I just said that doing this isn’t required); nor is it a matter of expressing an attitude roughly equivalent to “yay, killing in self-defense!” (for we used that locution to explicate moral requirements). So what’s going on here? It would seem that the noncognitivist will need to introduce the idea of an expression whose function is to mark the explicit absence of a command or the absence of certain attitudes. But as I say, we needn’t worry about these sorts of details.) What is important for our purposes is acknowledging that the two basic roles for moral vocabulary that we have identified on behalf of the noncognitivist—revealing attitudes and issuing commands—are important ones to fill. We care deeply about how people act, and so it is crucially important to have a way to communicate our attitudes of approval and disapproval and to command one another (and ourselves) to act accordingly. So the noncognitivist certainly thinks that moral utterances play a significant, indeed central, role in our lives. The jobs they perform are important ones. But they do not include the attempt to state or describe any purported moral facts.  4.2 The Case for Noncognitivism 87 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 4.2 The Case for Noncognitivism I have suggested that it is helpful to think of noncognitivism as putting forward two theses. The negative thesis says that substantive moral claims are not, in fact, genuine assertions at all; they are not used to describe purported moral facts. The positive thesis says that such claims are, however, used to express attitudes of approval and disapproval and to issue commands. Now as I have also already suggested, on the face of it there is no obvious reason why a moral realist should need to reject the positive thesis, if we understand that thesis as indeed restricted to making positive claims about the use of moral utterances. For there is no obvious reason why a realist should resist the suggestion that among the things moral language is used for are expressing attitudes and giving commands. All that the realist needs to insist upon is that these are not the only things that moral language is used to do. In particular, one further use—indeed the central use—of moral claims is to make assertions, to attempt to say what the moral facts are. So on the face of it, at any rate, it certainly looks like there is no reason for the realist to resist accepting the noncognitivist’s positive thesis. It is only the noncognitivist’s negative thesis which must be rejected. Indeed, it seems a virtue of realism that it can take this sort of pluralist attitude, that it can acknowledge that we use moral language not just for describing, but also for expressing and commanding. For just as it seems natural to take moral claims at face value as attempts to make genuine assertions, it also seems obviously correct to say that we also use them to reveal our attitudes of approval and disapproval and to give commands. On the one hand, when I (sincerely) say, for example, that murder is wrong, surely part of what I am doing is revealing to you my disapproval of murder. So it does seem undeniable that part of what moral claims are used for is to express attitudes. And, on the other hand, when we think of how easily we move between moral claims (“it is wrong to lie”) and moral rules (“don’t lie!”), it also seems undeniable that part of what moral claims are used for is to issue commands. So the positive thesis seems highly attractive in its own right, and it is an advantage of realism that it can so readily embrace it. Accordingly, it looks like we can focus our attention on the negative thesis, the claim that substantive moral claims are not used to describe. And here, I think, it really does seem as though the burden of proof will fall squarely on the noncognitivist. After all, as I have several times suggested, the sort of moral sentences we have in mind (“inequality is bad,” “white lies are 88 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 wrong,” and so on) certainly look like they are attempts to make claims about purported moral facts. If the noncognitivist thinks that this isn’t really the case—if they hold that the claims aren’t merely false (which is what the ni- hilist believes), but not genuine claims at all—then surely it falls upon the noncognitivist to explain to us why we should believe that this appearance is indeed a mere illusion. Moral claims look like claims. Why think otherwise? It might seem as though the noncognitivist should here avail herself of the same skeptical arguments that are routinely used to challenge moral realism. We haven’t yet examined them, but can’t the noncognitivist simply insist that if one or another of these arguments works, realism is defeated and we should, accordingly, embrace noncognitivism? But any suggestion like this would be too quick. For even if some argument convinced us that there are no moral facts and so realism must be rejected (perhaps because, say, any such facts would be metaphysically dubious), that would apparently still leave untouched the cognitivist claim that moral claims really do attempt to describe such facts. Even if they inevitably fail to say anything true (precisely because there are no such facts), that wouldn’t yet give us reason to think that moral claims aren’t really claims at all. An argument to the effect that moral claims are all false isn’t at all the same thing as an argument to the effect that they aren’t really assertions. In short, even if there are no moral facts, it seems as though ordinary skeptical arguments are actually going to support nihilism, rather than noncognitivism. To be sure, if we do become nihilists, we will then face the question of whether to abandon moral language. And at that point we might well decide that even if there are no moral facts there is still value in using moral language to express our attitudes of approval and disapproval and to tell people how to live. So we might well become revisionist nihilists, suggesting that we stop using moral utterances to make claims and use them instead only to express attitudes and issue commands. That would certainly be an intriguing proposal, but still, for all that, it would be an articulation of a cognitivist position (since even revisionist nihilism is a form of cognitivism), not at all a defense of noncognitivism. So we need to ask again, just how does the noncognitivist propose to argue that despite the surface grammar, despite the fact that moral claims look like genuine claims, in point of fact these sentences are not genuine claims at all? Interestingly enough, the most promising line of argument appeals to the thought that despite appearances, the various proposed uses of language are not in fact all compatible, that if it really is the case (as surely it is) that moral  4.2 The Case for Noncognitivism 89 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 claims are used to express attitudes and issue commands, then it just can’t be the case that they are also used to describe (purported) facts. In effect, the noncognitivist argues that the positive thesis entails the negative one, that the positive thesis suffices, all by itself, to rule out the cognitivist’s claim that moral claims are used to make assertions as well. Let’s look at this strategy at work. (Warning: the arguments that we’ll be considering until the end of this section may be a bit hard to follow the first time through.) One version of the argument focuses on the fact that we use moral claims to express our approval and disapproval. The noncognitivist then asks how we are to reconcile this with the cognitivist’s insistence that such utterances also assert the existence of various moral facts. For suppose that substantive moral claims really were assertions. It would seem, then, that I could in prin- ciple believe the given assertion (agree that the facts are just as described) and share this thought with you, without yet telling you anything at all about what I approve or disapprove of. After all, if I really thought there were moral facts, how could my merely acknowledging them tell you anything about what I favor? Suppose, for example, that the sentence “murder is wrong” really does assert something, namely, that acts of murder have the property of being wrong. It would seem, then, that I might sincerely believe that the facts are just as described. Accordingly, I might communicate that belief to you by uttering the sentence “murder is wrong.” Surely I could do that—state my belief about those facts—without indicating in any way whether I disapprove of murder. So imagine someone giving the following speech: “Murder is wrong. That is a simple fact. Murder really does have the property in question, wrongness. But what of it? I just don’t care about the fact that murder has that particular property! Indeed, I favor murder under certain circumstances. It is wrong, but I don’t disapprove of it at all!” It seems that if cognitivism were right, this would be a coherent speech to give. Sincerely asserting murder’s wrongness would not necessarily commit you to any form of disapproval at all. But—the noncognitivist continues—that gets things completely backwards. It really does seem obvious that (sincerely) saying that murder is wrong necessarily expresses disapproval of murder. So cognitivism must be mistaken. For if cognitivism were correct, one could think something wrong without disapproving of it. Since one can’t, cognitivism has to be rejected. 90 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 (The same basic argument could have been presented in terms of approval rather than disapproval: if cognitivism were correct one could think some- thing obligatory without approving of it; but one can’t, so cognitivism has to be rejected.) This argument against cognitivism is certainly suggestive. But we might still wonder: what is it about disapproval (or approval, for that matter) that necessarily takes us beyond merely acknowledging a given set of facts? What more does moral disapproval involve? A natural suggestion for the noncognitivist to make at this point is that there is a necessary connection between disapproving (in the morally relevant sense) and being motivated to act. To disapprove in this sense requires at least some minimal motivation to avoid the relevant acts or act type. (Thus, for example, if I truly disapprove of lying then I am at least some- what motivated to avoid lies.) So perhaps the noncognitivist’s argument is really this: There is a necessary connection between thinking something wrong and disapproving of it, and another necessary connection between disapproving of something and being at least somewhat motivated to avoid doing it. But the cognitivist cannot acknowledge these connections, since mere beliefs cannot motivate. So cognitivism is false. If something like this captures what the noncognitivist has in mind, we can simplify the argument a bit by dropping the references to disapproval and appealing instead directly to the implied connection between moral utterances and motivation. The resulting argument would then look like this: There is a necessary connection between thinking something wrong and being at least somewhat motivated to avoid doing it. But the cognitivist cannot explain that connection. So cognitivism is false. This revised argument involves an appeal to motive internalism (introduced in 1.2), one version of which holds that moral beliefs necessarily motivate, at least somewhat. The noncognitivist is thus claiming that motive internalism is incompatible with cognitivism. Therefore, given the plausibility of motive internalism, cognitivism should be rejected. Here’s the argument spelled out a bit more fully. If an agent sincerely believes that an act is wrong, this entails at least some motivation to avoid doing the act. But if cognitivism is correct, then the belief that an act is wrong is indeed a genuine belief, namely, the belief that a certain fact holds (to wit, that the given act has the property of wrongness). But a belief cannot, all by itself, guarantee any motivation to do anything. So cognitivism is mistaken.  4.2 The Case for Noncognitivism 91 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 (Moral “beliefs” are not genuine beliefs at all, and moral “claims” are not genuine assertions.) In effect, the noncognitivist is arguing that cognitivism cannot accommodate the truth of motive internalism. (How could a mere belief guarantee motivation?) In contrast, it might be added, noncognitivism easily accommodates motive internalism. For if moral claims are not genuine assertions at all, one cannot literally believe them and so we needn’t worry that one might have the relevant belief and nothing more, and thus not have the corresponding motivation at all. On the contrary, if noncognitivism is correct then when I sincerely say that murder is wrong, all I am doing is expressing my attitude of disapproval (and, perhaps, issuing a corresponding command). So if an agent sincerely “believes” that murder is wrong, they will necessarily be motivated (at least somewhat) to avoid killing others, since they really do disapprove of it. How might the cognitivist reply to this line of thought? The issues turn out to be quite complicated, and so I am going to postpone detailed examination of them until a later chapter (chapter 10). But let me quickly note two basic possibilities. One possible response, of course, would be to reject motive internalism. After all, if moral beliefs needn’t entail anything about motivation, then it won’t be any sort of problem for the cognitivist to concede that on their view one might sincerely believe that murder is wrong and yet not be motivated to avoid killing. Is the denial of motive internalism plausible? Some—the motive externalists—certainly think so. They claim that we can easily imagine someone who is completely unmoved by his moral beliefs. For example, a hit man for the Mob might say, “Of course murder is wrong, but so what? I just don’t care about morality!” If that is indeed a genuine possibility, then it seems as though motive internalism is mistaken, and so the noncognitivist’s argument (which presupposes it) fails. In effect, a position like this defends cognitivism by rejecting part of the noncognitivist’s positive thesis (or, alternatively, by modifying our under- standing of it). Instead of agreeing that substantive moral claims are always used to express approval and disapproval, the cognitivist would claim, rather, that they are only usually used to do this. So it won’t threaten cognitivism to note that mere beliefs per se won’t entail anything about motivation; it can still be the case that moral claims are indeed genuine assertions. (The 92 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 cognitivist might say: moral beliefs are typically accompanied by the relevant motivation; but not always.) But not everyone believes that a thoroughgoing amoralist (like our imaginary hit man) is a genuine possibility. (Does the hit man really think it wrong to kill?) And in any event, not all cognitivists are willing to embrace motive externalism. Some will think that anything genuinely worthy of the name morality must inevitably move us, at least a little. So it is worth noting as well a second possible defense of cognitivism, one that embraces motive internalism. According to this second possible reply, it isn’t really true that a mere belief can never motivate. No doubt, most beliefs are motivationally “inert.” (The mere belief that snow is white surely doesn’t move me to do anything at all.) But perhaps some beliefs really are such that they do guarantee the presence of at least some motivation. What sort of beliefs might be like that? Unsurprisingly, the cognitivist’s answer might be: moral beliefs! Here the cognitivist would be agreeing that if someone genuinely thinks that murder is wrong, then they must be moved by this thought (at least a little) to avoid killing. Thus the cognitivist would be accepting the unqual- ified version of the positive thesis, where that thesis is construed to mean that moral claims are always used to express approval and disapproval. And the cognitivist would be agreeing as well that the relevant notion of approval and disapproval necessarily involves motivation. But they would simply be insisting that moral claims can be genuine assertions for all of that, since it is the nature of moral beliefs that they do necessarily motivate (at least somewhat), unlike more ordinary beliefs. Here too, then, if this is indeed a genuine possibility—if certain beliefs (and in particular, moral beliefs) can motivate all by themselves—then the noncognitivist’s argument (which presupposes that this is impossible) fails. In principle, then, there are two different ways that the cognitivist might try to respond to the noncognitivist’s argument. So if the noncognitivist is to shore up her attack on cognitivism she must rule out both of them. She must argue both that genuine beliefs alone cannot motivate (so as to rebuff the second cognitivist reply), and that moral “beliefs” necessarily can (so as to rebuff the first). At this point, I suspect that most of us will agree that it isn’t obvious who has the better of this dispute. Adjudicating it will require a much more thor- ough examination of the various connections between motivation, on the  4.2 The Case for Noncognitivism 93 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 one hand, and morality and belief, on the other, than we have yet undertaken. Indeed, the topic is sufficiently complicated that we will later devote an entire chapter to it. For the time being, then, I only hope to have made it clear just how much turns on the outcome of that discussion. Instead of considering this argument further now, I want to note a second argument that is similar to it—except that instead of focusing on our use of moral claims to express attitudes, it focuses on our use of moral claims to issue commands. Since we certainly do use moral claims in this way, it concludes that these claims cannot be genuine assertions. Despite the shift in focus, the general strategy of this new argument is similar to that of the first one. It starts by pointing out that if cognitivism is true, substantive moral claims make genuine assertions about purported moral facts. But if that were the case, then it seems that one could believe that the facts are, indeed, just as the claim described. And if so, one could then appropriately communicate that belief by uttering the claim in question. But—the argument continues—surely one could assert the existence of such facts without thereby offering any commands at all about how to behave. The mere act of acknowledging facts cannot commit one to issuing imperatives. So cognitivism has to allow for the possibility of making moral claims without issuing commands. Since that possibility is absurd, cognitivism must be rejected. After all, suppose for the moment that the sentence “murder is wrong” really did assert that acts of murder have a certain property: wrongness. It seems as though I could then use that sentence to make that very asser- tion, and do so without yet issuing any kind of command at all. For saying or thinking that the world is a certain way (that the facts are a certain way) does not, in and of itself, commit me one way or the other to telling people how to behave. In particular then, I could believe murder is wrong without commanding anyone (myself or others) to avoid killing. But this, the noncognitivist argues, is unacceptable. Surely if I believe that murder is wrong this is equivalent to accepting some principle along the lines of “murder is morally forbidden.” And there is no real difference between accepting such a principle and accepting the corresponding moral rule “don’t kill!” But rules just are imperatives; so it is obvious that believing that murder is wrong necessarily involves the acceptance of certain commands, and saying it necessarily involves the issuance of such commands. Since cognitivism has to allow for the possibility of making moral claims without issuing commands, it has to be rejected. 94 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 Just as the strategy of this second argument is very similar to that of the first, the replies open to the cognitivist are similar as well. Once again, there are two of them, and here too I am only going to note them quickly. First, then, the cognitivist might try to reject the assumption—crucial to the argument—that accepting a moral claim necessarily involves the giving (or accepting) of a command. Perhaps this is only usually the case, not always. Perhaps one really can think an act wrong without issuing the corresponding imperative. Consider our hit man once again. Imagine that he says to his assistant, “I want you to kill Frank! Admittedly, killing is wrong, but what of it? I just don’t care about morality.” If that’s a coherent possibility, then moral claims can indeed be genuine assertions after all, affirming the existence of relevant facts. It won’t be a problem for the cognitivist to admit that the mere recognition of such facts need not result in the issuing of imperatives. (Taking this position would of course require rejecting or at least somewhat modifying the positive thesis, if that thesis is construed as asserting that moral claims are always used to issue imperatives.) Once again, not everyone will accept the idea that this kind of amoralist (the hit man) is a genuine possibility. And in any event, many cognitivists will agree with the noncognitivist that there is an extremely tight connection between making moral claims and issuing commands. So a second possible cognitivist response would be to insist that it isn’t really true that mere assertions never bring commands in their wake. No doubt most assertions are “imperativally inert” (if I tell you that snow is white, no particular commands follow, one way or the other). But there may be particular assertions which nonetheless have a uniquely close connection to commands. What sorts of assertions might be like that? Unsurprisingly, the cognitivist’s answer might be: moral beliefs! Here’s the idea. We all have an intuitive distinction between valid and invalid commands. If a random stranger standing next to you turns and says, “Give me $100!” this is going to strike you as a command in name only, nothing you need pay attention to. Why? Because there is no good reason to do what he said. In contrast, most of us think that the imperative “don’t tell gratuitous lies!” is a valid command, one backed by compelling reasons. (Recall the related discussion in 1.2.) Perhaps then the cognitivist should insist that there is a necessary con- nection between the truth of a given moral claim (what it would take for the claim to be true) and the existence of compelling reasons. This is a ver- sion of reasons internalism (introduced in 1.2.) Suppose then that I sincerely  4.2 The Case for Noncognitivism 95 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 claim that murder is wrong. Given reasons internalism, I have in effect com- mitted myself to the existence of compelling reasons not to kill. This would not in any way undermine the thought that I have made a genuine claim. Indeed, part of what I would have been saying or implying was that such reasons exist. So in saying that murder is wrong, I would have committed myself to a further claim, namely, that the imperative “don’t kill!” is a valid one, a command backed by good reasons. Thus, although the claim that murder is wrong would still be a genuine assertion, in sincerely making that claim I would be committing myself to the validity of the corresponding imperative. Perhaps, strictly speaking, making a moral claim would not automatically involve issuing a command, but it would come very close to doing that, since I would be implying that there were good grounds for issuing such a command. And under ordinary circumstances, at least, saying that there were such grounds might constitute one familiar way of (implicitly) issuing that very command. Thus there are at least two ways the cognitivist might try to resist the second version of the noncognitivist’s argument. She might deny that there is any necessary connection between making a moral claim and issuing commands, or she might insist that—thanks to the truth of reasons internalism—whenever you sincerely make a moral claim, you implicitly commit yourself to the validity of the corresponding imperative. (Note, in passing, that the second response also offers a possible reply to a different argument that the noncognitivist might propose, the argument from synonymy. This argument starts with the idea that uttering a moral principle like “it is wrong to tell lies” is basically the same thing as uttering the command “don’t lie!” But the latter of these two sentences is undeniably an imperative, while it is at least a possibility that the former only appears to be a genuine assertion. So if they are truly synonymous, it is more plau- sible to conclude that a claim that killing is wrong is actually an imperative as well, not a genuine assertion. Against this argument the cognitivist can now reply that although the two sentences (principle and imperative) are closely related, they are not truly synonymous. Rather, the principle implies the existence of good reasons for obeying the imperative, and the imperative presupposes the existence of such reasons. Thus anyone who affirms the principle is likely to be ready to issue the command; and anyone who issues the command is likely to be ready to affirm the principle. But for all that, the two are not synonymous: one is an imperative, while the other is indeed an assertion.) 96 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 To be clear, I don’t take these quick responses to the noncognitivist’s arguments to be decisive. We still need to take a closer look at the possibility of amoralists like the hit man, and we need to think further about the plau- sibility of both reasons internalism and motive internalism. These are issues to which we will eventually return. But I do think these quick replies suffice to show that providing a successful defense of noncognitivism (if such can be given) won’t be a trivial undertaking. So for the time being, at least, it seems reasonable to tentatively retain our natural inclination to take moral claims at face value—as genuine assertions about purported moral facts. 4.3 Objections to Noncognitivism So far, we have imagined the cognitivist on the defensive, trying to respond in one way or another to the noncognitivist’s attack. But it is worth pointing out that the cognitivist can also go on the offensive, noting various difficulties with noncognitivism. Beyond the most obvious objection—the simple fact that moral claims certainly look like they are attempts to describe a moral reality—there are additional problems that may further reduce noncognitivism’s plausibility. To the extent that noncognitivism has a difficult time responding to these objections, this will reinforce our sense that it is indeed cognitivism that is the more plausible of the two views. However, given that our main interest lies in evaluating skeptical challenges to moral realism, a survey of noncognitivism’s difficulties is less important to us than the attack against cognitivism that I sketched in the previous section. Accordingly, I won’t dwell on any of the objections that I am about to describe. Still, it is worth saying enough to show that noncognitivism really does face some significant problems of its own. First, then, consider the noncognitivist’s claim (part of their positive thesis) that when I make a moral assertion I am (often, or maybe even always) expressing an attitude of approval or disapproval. That does seem a plausible enough thing to suggest. But it is important to bear in mind that not all approval or disapproval is moral approval or disapproval. If, for example, I tell you, “That’s an ugly blouse!” then although I am clearly indicating my disapproval of your choice of clothing, this isn’t an instance of moral disapproval. Similarly, if I say, “What a delicious cookie!” then although I may be indicating my approval of your culinary skills, it isn’t moral approval that is being displayed.  4.3 Objections to Noncognitivism 97 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 So what is the particular sort of approval or disapproval that is involved when I make a moral claim? What differentiates moral approval or disap- proval from other forms of approval and disapproval? One natural suggestion is that the relevant forms of approval and disapproval are those that are based on judgments about the act’s being right or wrong. I morally disapprove of an act if my disapproval is due to my thinking that the act is morally forbidden. I morally approve of an act if my approval is due to thinking the act permissible or required. (Similarly, my disapproval of an outcome is moral disapproval if it is due to my thinking that the outcome is morally inferior. And so on.) Unfortunately for the noncognitivist, however, any proposal along these lines presupposes that I have a logically prior belief about the moral status of the act (or outcome) in question. My approval or disapproval can be based on my moral judgments only if I have moral judgments—only if I take the relevant moral claims to be true. And while a cognitivist can certainly allow for this possibility (since cognitivists take moral claims to be truth apt, and thus the sort of thing that can be believed or rejected, providing the basis for one’s moral attitudes), the noncognitivist cannot. According to the noncognitivist, moral claims are not genuine claims at all, so there is nothing to believe; there are no moral beliefs to provide the basis for one’s approval or disapproval. Thus, the noncognitivist needs some alternative way to pick out what it is that distinguishes moral approval and disapproval from other forms of approval and disapproval, a way that doesn’t covertly smuggle in an appeal to genuine moral assertions. The noncognitivist needs a way to characterize moral attitudes without presupposing the truth of cognitivism! And while there are various proposals that might be made, it is far from clear which, if any, are satisfactory. A second challenge facing noncognitivism grows out of a criticism which is, I think, fairly easily answered. The criticism is this. It certainly seems as though people disagree about moral matters. One person thinks white lies permissible, while another thinks them forbidden. Some people think all pleasures have intrinsic value, others do not. Disagreement is a familiar fact of our moral discourse. But how is the noncognitivist to explain what is going on when people disagree? The cognitivist has a straightforward account of the nature of moral disagreement. One side of a given debate is asserting—genuinely asserting— that the moral facts are such and such, and the other side is denying this. Each side is making a truth apt claim, but they are disagreeing about which of the 98 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 two claims is the true one, which claim describes the facts correctly. That’s the nature of moral disagreement: one side asserts what the other denies. But this natural account isn’t available to the noncognitivist, since the noncognitivist denies that moral claims are genuine assertions. Instead, according to the noncognitivist, all that is happening here is that each side is revealing their preferences. But if neither side is actually asserting anything, what could it possibly mean to say that they are disagreeing? If I reveal that I like chocolate ice cream, while you reveal that you prefer strawberry, there is no incompatibility here between what you are doing and what I am doing! So where’s the disagreement? As I say, this initial criticism is readily answered. Unlike the case of ice cream, where it is fine with me if you prefer strawberry and it is fine with you if I prefer chocolate, in the kinds of cases where moral attitudes are involved it is not in fact a matter of indifference to any of us how other people choose. We may not have differences of opinion with regard to the facts (says the noncognitivist) but we can still have disagreements concerning how we want ourselves and others to behave. This sort of practical disagreement occurs even in nonmoral contexts. If you and I are deciding what to do tonight, and you prefer the movies, while I prefer to go to a concert, there is no disagreement in our beliefs—neither of us is making an assertion that the other denies—but for all that we have a genuine disagreement, a disagreement about what to do! All the more so, then (says the noncognitivist), when we are disputing a moral question. If you think abortion is morally wrong, for example, while I think it is permis- sible, we will both care deeply about whether abortions are available. This won’t be a disagreement of belief, but it will be a disagreement about what sorts of behavior we are going to disapprove of. This is a perfectly genuine— and important—form of disagreement. So far, so good. The noncognitivist can meet the initial criticism. But now a new challenge arises in its wake. For what we might want to know, next, is this: given our moral disagreement, what is it, on the noncognitivist’s account, for us to engage in moral argument? Here too it seems as though the cognitivist has an easier time of it. For normally when we engage in a rational dispute with one another we try to sway those on the other side by offering reasons to believe that our own view is true. And if cognitivism is correct, then this is what we are trying to do in moral disputes as well: we are trying to persuade the other side by providing compelling reasons to believe that the particular claims we are putting forward are the true ones.  4.3 Objections to Noncognitivism 99 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 But what kind of account of moral argument can the noncognitivist pro- vide? If moral claims are not genuine assertions, then I am not trying to get you to change your beliefs but only your attitudes; I am simply trying to get you to change your preferences. Can this be anything other than an attempt on my part to influence your tastes by “pushing” your psychic buttons? I may succeed in tugging at your heartstrings. But is that argument, or just manipulation? In some cases, no doubt, it may be possible to provide reasons for changing one’s preferences. Perhaps you deeply want something, X, and don’t realize that Y will help bring X about. So you don’t currently want Y; you may even be opposed to Y. But in the course of arguing with you, I may be able to persuade you that Y really is an effective means of getting X. I have given you a reason to want Y, and in light of it, you may change your attitude toward Y. In cases like this, cases involving “derivative” preferences, even the noncognitivist can point to something that we will recognize as an instance of offering a rational argument for changing your attitudes. But even if the noncognitivist can do this for derivative preferences, it seems difficult to see how a similar account could possibly work when it comes to more basic or “ultimate” preferences. Here, it seems, noncognitivists have to dig in their heels and insist that no further reasons can be given for having one (fundamental) attitude rather than another. Anything that looks like rational argument can be nothing more than manipulating your psy- chology in such a way as to cause you to change your preferences without actually providing you with reasons to do so. (Why must the noncognitivists concede that on their view there cannot be good reasons to embrace one ultimate preference rather than another? Because if even ultimate preferences were subject to rational evaluation, then a moral realist could simply insist that right and wrong is ultimately a matter of which preferences are justified all the way down; and then there would be moral facts after all! For example, they might hold that a claim to the effect that a given act is wrong is true if and only if there is good reason (all the way down) to disapprove of it. And that’s a form of cognitivism, not noncognitivism.) So if we accept noncognitivism, won’t we need to hold that at a certain point in any moral “argument,” what passes for the giving of good reasons will inevitably be nothing more than verbal posturing and psychological manipulation? That rational argument about fundamental moral questions must actually be impossible? That seems a rather disappointing conclusion 100 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 to reach. (In contrast, the cognitivist can insist that even for basic moral issues the question is always whether a given moral claim is true, and there is no reason to assume that such questions are beyond our rational grasp.) Perhaps there is a way for the noncognitivist to avoid this unhappy conclusion. At the very least, perhaps noncognitivists can find a satisfying way to distinguish between acceptable ways to influence another’s basic preferences and sheer propaganda and brainwashing. That might at least somewhat reduce the unattractiveness of the view. Or perhaps some noncognitivists are prepared to bite the bullet here and concede the point. Once again, I won’t consider the matter further. My goal remains the more modest one of pointing out some of the different ways in which noncognitivism faces its own challenges. Here’s a third objection. According to noncognitivism’s positive thesis we use moral claims to express attitudes and issue commands. So if I say, “White lies are wrong,” I am expressing my disapproval of white lies and issuing a command (to myself and others) not to tell them. That noncognitivist interpretation of what I am doing when I use moral language is a tolerably clear proposal when we limit our attention to simple (“atomic”) sentences like the one I just mentioned. But we often use moral language in more complex sentences than these, and it is not yet at all clear what the noncognitivist thinks we are up to when we do. Suppose, for example, that I say, “If white lies are wrong, then it is wrong to tell one to your roommate.” What am I doing if I say something like that? As usual, the cognitivist has a relatively straightforward answer. I am asserting that if one set of facts holds (that is, if it is indeed a fact that white lies are wrong), then a second set of facts holds as well (that is, then it is a fact that telling such a lie to your roommate is wrong). In effect, I am asserting that there is a relationship between the truth of the first clause of the sentence and the truth of the second clause, such that if the former is true then the second is true as well. Note, however, that in saying this, I am not actually committing myself to the truth of the antecedent. I am not taking a stand on whether it really is true that telling white lies is wrong. I am merely claiming that if such lies are indeed wrong, then something follows from that. Details aside, all of this is more or less straightforward if we are cognitivists. But what should we say if we are noncognitivists? It won’t be acceptable for the noncognitivist to say that I use this more complex sentence to express my disapproval of white lies—for as we just noted, in saying this sentence I am not yet committing myself to the claim that telling white lies is wrong.  4.3 Objections to Noncognitivism 101 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 Nor am I commanding others not to tell them. What I am doing, of course, is saying that if white lies are wrong, so is telling one to your roommate. But how is the noncognitivist to capture this idea armed only with the notions of expressing attitudes and issuing commands? (Even if we agree that in uttering the sentence I am expressing my disapproval of your telling a white lie to your roommate if such lies are wrong, or I am ordering you not to tell such a lie if they are wrong—how is the noncognitivist to make sense of the conditional nature of such commands and attitudes? The most natural way to interpret them might be to say that if a certain moral claim is true—if telling white lies is wrong—then I do disap- prove and command accordingly. But of course, the noncognitivist cannot help herself to such talk of moral truths. Alternatively, is there some logically complex attitude or command at play? Am I somehow trying to express my disapproval of your telling your roommate a white lie in those circumstances where I (first?) disapprove of white lies more generally? But that doesn’t seem to adequately capture the content either. It would be more accurate to say that I disapprove of your telling the lie to your roommate if white lies are worthy of disapproval. But how is the noncognitivist to capture that last bit?) Closely related problems show up everywhere, once we start looking for them. Sometimes in moral deliberation we work our way from prem- ises to conclusions about right and wrong. I might, for example, argue as follows: (1) if white lies are wrong, then telling one to my roommate is wrong; but (2) white lies are wrong; so (3) telling a white lie to my roommate is wrong. That certainly seems to be a logically valid argument: the conclusion, (3), really does seem to follow logically from the premises. That is to say, if the premises are true, the conclusion is true as well. That, at least, is how we would ordinarily describe the logical force of an argument like this. We may be uncertain about one or more of the prem- ises of this argument (in particular, we might be uncertain about the second, which says, of course, that white lies really are wrong). But we ordinarily wouldn’t hesitate to say that the truth of the conclusion really would follow if the premises were true. Yet none of this talk—about how the truth of the conclusion would follow from the truth of the premises—can be taken literally if we are noncognitivists, since noncognitivists don’t think that sentences like (2) and (3) are so much as truth apt! What could it possibly mean to talk about how the truth of the conclusion would follow from the truth of the premises, if the conclusion and at least one of the premises are not the sort of thing which can 102 Noncognitivism Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 even have a truth value? (And of course, as we have already noted, it remains obscure what the noncognitivist wants to say about (1) in the first place.) Abstractly stated, the problem is this. When we evaluate arguments we ask whether the conclusion follows from the premises. But this very notion of “following” seems to be one that we most naturally understand by means of the notion of truth. Yet if noncognitivism is right and simple moral claims are not truth apt, what can this notion of one thing following from another even mean? (How is this second version of the problem, which concerns the idea of logically valid arguments, related to the first, which concerns complex moral utterances? Both involve contexts in which something more complicated than simple assertion is going on, and it is difficult to see how to make sense of the phenomena without appealing to the idea that moral claims can be truth apt.) As usual, none of this is problematic for the cognitivist. For the cognitivist thinks that all of the relevant claims in the argument are indeed truth apt, so it makes perfect sense to ask whether the truth of the premises really would guarantee the truth of the conclusion. But the noncognitivist needs some alternative account of what it means for the conclusion of a moral argument to “follow” from its premises, and it certainly isn’t clear what this alternative account would look like. Let me close with one more example of what is probably the same under- lying problem at work. Imagine that I am trying to decide whether or not it is permissible to tell a white lie. Perhaps I am planning a surprise party for my wife, and she has asked me what our plans are for the evening. Imagine that I ask myself, “Are white lies permissible?” What is going on when I raise this question? (Note that this is yet another area where I am using moral language, but not making a simple assertion.) Here too, the cognitivist has an easy time of it. According to the cogni- tivist, the claim that white lies are permissible is truth apt—it is either true or false. In asking my question, I am wondering which it is. (Even the nihilist, who thinks there are no moral facts, can recognize that I am here assuming there are such things, and wondering what the relevant facts are.) But once again, the noncognitivist’s account of what I am up to in asking my question is difficult to make out. Obviously enough, it can’t be that I am expressing approval or disapproval of white lies, since in my current state of uncertainty I don’t yet approve or disapprove. I am trying to figure out which to do! But if we rephrase my question as “what should I do with regard to white lies,  4.3 Objections to Noncognitivism 103 Downloaded from https://academic.oup.com/book/55119/chapter/423936314 by UNIV OF TORONTO SCARBOROUGH LIBRARY user on 30 September 2024 approve or disapprove of them?” no real progress has been made. For the obvious thought here is that I should disapprove of them just in case they are indeed wrong—if it really is true that white lies are wrong. And this is a thought that the noncognitivist cannot appeal to. (Similarly, I clearly am not issuing a command not to tell white lies. At best, I am trying to decide what commands are appropriate here. But how is the noncognitivist to make sense of that notion?) So here we have yet another case—an everyday, familiar case—of moral language being used in a way that the noncognitivist cannot easily explain. In short, noncognitivism runs into trouble whenever we move away from the simplest cases of moral assertion. The account of moral language that views it simply in terms of offering commands and expressing attitudes may seem adequate when we restrict our attention to simple assertions like “murder is wrong” or “inequality is bad.” But extending that account to more complex linguistic contexts (including ones where we aren’t straightfor- wardly making moral assertions at all) remains a challenge. None of this is meant to suggest that no proposals can be offered by the noncognitivist to explain what is going on in these more complex cases. There are, in fact, any number of proposals that have been made, though the details rapidly become extremely technical and complicated. Here again, then, the point is not at all to claim confidently that the noncognitivist’s difficulties are insuperable, but only to note that they are indeed difficulties. In comparison, at least, the cognitivist seems to have the upper hand. Of course, we must bear in mind that we have yet to give the noncognitivist’s attack on cognitivism a full hearing. It may be that when we examine it more carefully we will ultimately decide that cognitivism must indeed be rejected after all—and moral realism with it. I don’t believe that this will, in fact, be the most reasonable conclusion to reach; but for the time being this is merely a promissory note on my part. Still, until such time (if ever) as we find a com- pelling reason to prefer the noncognitivist’s account of moral discourse, I think it is more reasonable to take moral claims at face value: as genuine assertions about purported moral facts.