Metaethics, a relatively short intro
(This post contains jargon, which I’ve tried to explain as we go along, but it can probably be safely ignored if you don’t understand it. Everything important is explained in plain text. Metaethics is the study of what we mean when we discuss morality, the nature of moral facts and properties, and how we come to know them. It’s a topic I’ve been interested in for a long time, but never got around to writing the big post about. Before now.)
David Hume observed that, in all the theories of morality he had encountered, the author makes an invisible jump: at one point, he is talking about what the world is like, and in the next, he’s talking about what the world ought to be like, and nowhere is the jump between the two sorts of propositions accounted for. He didn’t say it was impossible to account for, only that he’d never seen anyone do it; certainly, he never provided any good justification for this jump himself. This little observation has spawned a whole literature. The problem of accounting for the jump from is to ought is appropriately called the Is-Ought problem.
Objective morality, which philosophers take to mean “mind-independent” morality, depends on bridging the gap between is and ought. Despite much bravado, no one’s actually done it.
Henry Sidgwick, a Victorian philosopher, noted that sometimes prudence and morality clash. Prudential reasoning is what you do when you consider your goals and the way to get there, and decide on the course of action that best fulfills your goals, whether they be the murder of a million people, the establishment of world peace or something in between. Sometimes, having considered all your goals, and the best way to achieve them, you conclude that the best course of action is something that’s generally held to be immoral: smuggle and steal and lie, say. If you want to be a crime lord, smuggling and stealing and lying may be your best bet. That, then, is what prudential reasoning says you ought to do. This is a familiar sort of ought: it is what Kant had called a hypothetical imperative. If you want x, you ought to do y; you want x, therefore, you ought to do y. But smuggling and stealing and lying is morally wrong: therefore, you ought not to do those things. Sidgwick couldn’t resolve the clash between prudential and moral reasoning. He called it the Duality of Practical Reason, and it was a real problem. Few paid it much attention, sadly.
This was simply the Is-Ought problem in another guise. To have objective (i.e., mind independent) morality, you need to provide a way for moral oughts to arise. You can’t just assume them. This is what most “solutions” to the problem do: they smuggle in normative assumptions disguised as descriptive is-statements. Take, for example, the idea that somehow God can provide a foundation of objective morality. Plato, in his dialog Euthyphro, demolished this idea thousands of years ago. Socrates, in that dialog, asks: is the pious loved by the gods because it is pious, or is it pious because it is loved by the gods? If the pious (good) is loved by the gods because it’s pious (good), that must mean that the gods’ opinion of whatever is pious isn’t what makes something pious, and hence the gods fail as a source of piety; and if something is pious because it is loved by the gods, then piety is simply arbitrary. There is no reason why murder is bad, and prayer is pious; it simply so happens that the gods like it. Couldn’t the gods have liked murder? Would that make murder ok? This doesn’t bridge the gap that Hume identified a long time after Plato: simply because something is loved by God (or the gods), that doesn’t explain why we ought to do it. At this point, some people will appeal to authority: God made us, dammit, so he’s got the right to decide what’s good and bad! But this is a moral principle — whoever makes a universe has the right to command beings in it, or some such — and an explanation or foundation of morality can’t depend on a moral principle, because that would be circular. It should be clear that by now that God lacks what it takes to found objective morality: either, on one horn of Euthyphro’s dilemma, he must appeal to an external standard of morality (God commands what’s good), or he’s simply another voice in moral discourse. (What’s good is what God says; in which case, why ought we do what’s “good”, so defined?)
Bertrand Russell, in the 1920s, toyed with the notion that maybe morality was truly corrupt. It was simply based on assuming that there was an objective moral quality, some foundation for moral oughts, that simply didn’t exist. Later, in the 1970s, Australian philosopher J. L. Mackie wrote a book called Ethics: Inventing Right and Wrong, in which he argued for what he called an error theory of morality. Moral discourse was fundamentally in error, he argued: specifically, it was based on the assumption that there are objective moral properties, such as “right” and “wrong”, “good” and “evil”, but these don’t exist. In other words, morality is a mistake. One of his arguments is called the argument from queerness. It says that morality must have a sort of magnetic power in it that somehow overpowers prudential reasoning (if x, then y), regardless of the speaker’s desires. This is simply too queer. Either there are weird magnetic properties like “morally right” and “morally wrong” that no one is able to account for, or, if these properties depend or supervene on natural properties, like “causes pain” or “causes pleasure”, then the relationship between these two kinds of facts is equally mysterious. Morality is based on a fairy tale, on mysterious properties that don’t make sense and whose existence no one can demonstrate either through logic or science. This is error theory: morality’s wrong, it works on false premises, if we want to be rational, we must abandon it.
The problem with objective morality can be located in this friction between prudential reasoning and moral reasoning. Prudential reasons are hypothetical imperatives: in the event of X, then Y. You ought to X if you want Y. However, moral reasons are supposed to be categorical: you ought to X, no matter what. Categorically, in every case, you ought to X; X doesn’t depend on your own desires or goals in any way. It stands above you. No one sensible accepts the excuse “I just murdered her because I felt like it, and it fulfilled my goals.” No, no matter what hypothetical imperatives tell you, categorical, moral imperatives are supposed to overpower them.
Given any supposed categorical imperative, such as “don’t kill”, I can imagine an agent who has a goal that would be fulfilled by breaking the imperative, in this case, killing. The hypothetical imperative, if you want X, you ought to do Y, in conjunction with the desire to achieve X, gives us an ought: you ought kill. Prudential reasoning, which is the ordinary kind of reasoning, has given us something that contradicts categorical imperatives. There’s a contradiction here, so you can’t have both. In the choice between hypothetical and categorical imperatives, there really is no choice. Ordinary logic, indeed any kind of practical reasoning, demands hypothetical imperatives. And if they contradict categorical imperatives, that means categorical imperatives — which, recall, are supposed to be truly categorical, meaning no hypothetical agent should be able to circumvent them — are out. Objective morality depends on categorical imperatives, which can’t exist, so objective morality is out.
This was my reasoning, and it led me to a moral error theory. I became convinced that morality was a mistake. I tried monitoring my speech for moral judgments, carefully removing anything that might resemble a moral stand. I tried to be a fictionalist, who says that morality is just a convenient fiction. I tried to find a flaw in my argument. I tried everything. The problem was that, despite my philosophical conviction that morality was corrupt, I couldn’t willingly turn myself into a psychopath. I couldn’t stop feeling wanton cruelty was wrong. It was impossible. I still felt moral emotions, even as I became convinced that all moral statements were categorically false.
Metaethics is the study of the semantics, metaphysics and epistemology of moral discourse, which means it’s the study of what “this is (morally) wrong” and similar statements mean, what kind of properties and relations these statements have, and how we come to know them. There are many metaethical theories. Those that analyze morality as objective are called moral realists. Error theory says that morality is semantically realist, meaning that statements such as “this is (morally) wrong” mean that this is objectively wrong; but, on investigating the metaphysics of morality, we find that there is just no property for “objectively wrong” to correspond to — indeed this property couldn’t possibly exist, for reasons explained above. So error theory is anti-realist in that it says morality isn’t objective, although its semantics assumes it is. This mismatch between what kind of facts we try to talk about when we talk about morality, and what kind of facts exist, account for the “error” in error theory: moral discourse (“this is wrong, this is right, this one’s good and that one’s evil”) tries to discuss and describe objective facts, but there are no objective moral facts to describe, so they inevitably fail. All morality is false. It is no more true that “murder is wrong” than it is that “murder is right”. Both are as nonsensical as “murder is blargnurg.”
I despaired. What to do when morality was in error, but moral emotions were very much still present? At this point, I discovered expressivism. Expressivism is another metaethical theory. It, too, is anti-realist, in that it says that morality isn’t objective. But that doesn’t matter, because according to modern expressivism, morality doesn’t even try. We can’t be in error for referring to non-existent properties if we’re not trying to refer to these properties. Expressivism starts by looking at the semantic dimension of morality: what does it mean to say that something is wrong, something is evil, something’s right and good and so on? According to their analysis, moral discourse consists not of plain descriptions of facts, but simply of expressions of our moral feelings, such as the ones I inevitably have upon hearing of injustice, murder, and so on. By concluding that the semantics of morality are subjective, emotional outcries (oof! yay! boo!), so to speak, they sidestep the metaphysics almost entirely: it doesn’t matter whether you can bridge that damned is-ought gap, because that’s not what morality’s about! Leave it to the same people who argue about angels on pinheads, and let’s get on to discuss real issues, seems to be the attitude.
An early version of expressivism is called emotivism, or the “boo-hooray” theory of morality. It says that when we say charity is morally right, we’re not saying that “moral rightness” is a property of charity. No, we’re saying something akin to “hooray to charity!” Just like “hooray!” or “ouch!” aren’t about properties, “charity is right” isn’t about moral properties. “Boo!” isn’t the sort of thing that can even be true or false. There’s no facts of the matter here, no question of whether “boo” is true or false. True or false simply don’t apply. This is what emotivism says of moral discourse: it’s noncognitive, which means that it’s neither true nor false, it has no truth value at all. Like primitive outcries, moral discourse is a series of highly sophisticated grunts and cheers that express our sentiments without describing how the world is supposed to be, what kind of properties murdering or raising money for charity are supposed to have, or anything of the sort.
A problem with this theory was soon discovered. It’s called the Frege-Geach or embedding objection. It concerns the embedding of moral statements in larger statements. A canonical example is this argument:
- If it’s wrong to lie, it’s wrong to get your little brother to lie.
- It’s wrong to lie.
- Therefore, it’s wrong to get your little brother to lie.
This is simply, everyday logic. But this argument can’t possibly be valid if emotivism is to be trusted. The above is an application of the fundamental logical principle modus ponens. Modus ponens, like every other logical principle, only works when the premises have truth value. In “if X, then Y”, X and Y must be either true or false. But according to emotivism, “it’s wrong to lie” means something like “boo! to lying”, and “boo! to lying” is neither true nor false. This means the argument above, which is obviously valid, can’t be valid on emotivism. Therefore, emotivism is false. We could patch up this by saying that the “it’s wrong to lie” in “if it’s wrong to lie, it’s wrong to get your little brother to lie” means something different from the “it’s wrong to lie” that stands alone on line 2), but this would be equivocation, using the same term to mean two different things, which would, again, make the argument invalid. Emotivism simply can’t account for ordinary moral reasoning, and therefore fails as a semantic theory of morality.
Modern expressivism is the successor to emotivism. The philosopher Simon Blackburn has championed one kind of modern expressivism, which he calls quasi-realism. It also analyzes morality in terms of expressions of emotion. But unlike emotivism, it doesn’t say that these emotional expressions are non-cognitive, neither true nor false. No, obviously they have truth value, because we observe that they act like they have truth value when we reason morally, such as in the example above. Inspired, perhaps, by Wittgenstein’s mantra “meaning is use”, the expressivist concludes that the emotional expressions in moral discourse, such as in “murder is wrong”, are used as if they’re true or false, and therefore, they are true or false. This, however, is a subjective kind of truth, belonging to an emotional logic. This doesn’t jive well with regular theories of truth, which posit objective criteria like “correspondence with objective facts” as criteria of truth. Subjective truth sounds comical: if I imagine the moon to be made of cheese, this can be true — and false at the same time! After all, it’s all subjective. But the expressivist insists that we must split our analysis of “truth” into pieces. First, “truth” means whatever “truth” is used to mean in a given domain. Second, “truth” clearly is used in a kind of emotional logic in morality, but in science, it is equally clearly used for correspondence with objective reality, or something like it. In other words, you must judge truth by domain-specific criteria, so that scientific statements are judged based on the criteria for scientific truth and mathematical statements on mathematical criteria, and moral statements based on criteria of moral truth.
What those criteria are, can be discussed. Generally, expressivists don’t make much of a difference between saying “murder is wrong” and “it’s true that murder is wrong”. Truth, first and foremost, means you emotionally assent; so saying “X is true” (or just “X”) means you’re emotionally disposed to think that X; but it also carries the implication that you will also assent to whatever follows from X, and it implies that X is defensible within moral discourse, and not just an arbitrary opinion.
(This isn’t a get out of jail free card for religion: there’s no “religious” truth, so much as there are religious statements in different domains, like ethics, where statements are judged based on ethical criteria of truth, and science, where statements are judged based on scientific criteria. This whole “separate magisteria” thing is dead in the water because religion tries to define its own rules, which is okay if you’re a domain unto yourself, but then religion starts making statements in other domains, while trying to impose its own criteria, which is simply wrong. “God exists” must be judged on a correspondence theory of truth: either that statement corresponds with actual reality, or it doesn’t. Religion isn’t a domain to itself. Instead, religion touches on physical science, on psychology, on ethics, on literature, and so on, and in all of these fields, there are established standards for truth that religion can’t bend simply so it gets out of its contradictions and falsehoods. A religious statement can be true, but only if it fulfills the criteria of truth in whatever domain it inhabits. “You ought to love your neighbor as yourself” might be a truth, but if that’s the case, it’s an ethical truth, and must fulfill ethical truth criteria.)
Ok, ok, I hear you saying, so this theory’s pretty neat, but how do you know it’s true? Maybe there are no glaring contradictions in analyzing the meaning of moral discourse like expressivists do, but isn’t this all theoretical? Isn’t this just a way things could possibly work? How do you know it’s how things actually work? Couldn’t error theory’s analysis, which says that “murder is wrong” means that murder is objectively wrong, be just as correct?
I brought up the above objection on a forum. The answer was a short but brilliant rhetorical question: what would be different? If expressivism describes what we do when we discuss morality in a precise and non-contradictory way, then, what would we do different if we found out that some other theory was correct? If we’d be doing all the same things, exactly the same, what difference does it make? In fact, this is just a repeat of meaning as use: we’re using morality as if expressivism is true; there’s no “deeper truth” here, moral discourse means whatever it’s used as, and it’s used as expressivism. Even if somehow you could make a case for another semantic analysis of moral discourse, if an expressivist analysis would allow us to carry on exactly as before, there’s no practical difference between switching to expressivist semantics and expressivism being correct in the first place.
Expressivism, then, says that moral discourse consists of expressions of our moral feelings, but these feelings act like they’re true and false, we have standards for them, we have arguments for them, we have a peculiar kind of emotional logic for them, so we might as well call them true or false. Voila! Morality is saved. It may be subjective, but it’s no less true. Suddenly we can say that murder is wrong without committing any metaphysical blunders.
That was metaethics, the (relatively) short introduction. I don’t feel like writing the longer one.
