“Two things,” Immanuel Kant wrote in the late 18th century, “fill the mind with ever new and increasing admiration and awe, the more often and steadily we meditate upon them: the starry firmament above and the moral law within.”

The awesome starry firmament inspires plenty of controversy—about the composition of dark matter, for example. But a lot is known: the sun is composed of hydrogen and helium, the Horsehead Nebula is 1,500 light years distant, and so on.

There’s also plenty of controversy about moral law. Should we give much more to charity than we actually do? Is torture permissible under extreme circumstances? Is eating meat wrong? Could it ever be permissible to kill one innocent person in order to save five? But, again we know a lot. Throwing good taste out with the bathwater for the sake of a clear example, everyone knows that boiling babies for fun is wrong. Boiling lobsters is a matter that reasonable people may disagree about, but as far as boiling babies goes, agreement is pretty much universal. Babies suffer when boiled—they are not like the worms that live near undersea vents, who are partial to scalding water. If something goes without saying, it’s this: one ought not to boil babies for fun.

Apart from filling the mind with admiration and awe, the starry firmament and the moral law together fill the mind with a problem, which Kant’s remark obscures. The quotation suggests, misleadingly, that the astronomical and moral realms are wholly separate—the former is “above” and the latter is “within.” But they aren’t: as Moby correctly sings, “We are all made of stars.” The heavens and human beings are composed from the same physical stuff, and are governed by the same physical principles. The starry firmament isn’t really “above”—it’s everywhere. We, along with lobsters and the rest, are part of it.

Everything, in short, is a natural phenomenon, an aspect of the universe as revealed by the natural sciences. In particular, morality is a natural phenomenon. Moral facts or truths—that boiling babies is wrong, say—are not additions to the natural world, they are already there in the natural world, even if they are not explicitly mentioned in scientific theories. Fundamental sciences such as particle physics and molecular biology do not speak explicitly speak of sand dunes, or boiling water, or lobsters, but facts about sand dunes and the like are implicitly settled by more fundamental facts: arrange bits of matter a certain way and you have an eroding sand dune, or boiling water, or (here the arrangement needs to be very complicated indeed) a lively lobster. And, presumably, the same goes for the moral facts.

But how can morality be a natural phenomenon? We ought not to boil babies, but the natural world seems not to contain any trace of an “ought,” or an “ought not.” A dropped stone is under no obligation to fall, it just does. Admittedly, I might say, before dropping a stone out of the window, “This stone ought to hit the ground in three seconds,” but here I just mean something like “It is likely that the stone will hit the ground in three seconds.” If the stone doesn’t do that, it has done nothing wrong, and is not to be blamed for anything. In the natural world, nothing ought to happen, or ought not to happen, in the relevant sense of “ought.” Keeping within the confines of nature, there is no space for the fact that we ought not to boil babies. Yet since nature is all there is, there is no place left to go.

This problem is sometimes traced to David Hume’s Treatise of Human Nature, in which Hume, writing half a century before Kant, complained of an “imperceptible change” from “the usual copulations of propositions, is, and is not” to propositions “connected with an ought, or an ought not.” “This change,” Hume said, is “of the last consequence. For as this ought, or ought not, expresses some new relation or affirmation, ’tis necessary that it shou’d be observ’d and explain’d; and at the same time that a reason should be given; for what seems altogether inconceivable, how this new relation can be a deduction from others, which are entirely different from it.”

The natural world contains plenty of facts concerning what is (or is not) the case: babies suffer in hot water, boiling water is hot, Virginia will drown if no one pulls her from the River Ouse, and the like. But how do we get from these facts to what ought (or ought not) to be the case—facts that are “entirely different”? As the philosopher Simon Blackburn puts it in his Ruling Passions, “the problem is one of finding room for ethics, or of placing ethics within the disenchanted, non-ethical order which we inhabit, and of which we are a part.”

Responding to this problem, Judith Jarvis Thomson observes in Goodness and Advice, “became the central task of Anglo-American moral philosophy in the century just past.” The problem is not one in ethics, like the issue of whether we should give more to charity than we actually do, but rather is about ethics or morality. It accordingly belongs to that branch of philosophy called “meta-ethics,” which started in earnest when G.E. Moore published Principia Ethica in 1903, and which has been flourishing ever since.

Before touching on some of the high notes, as well as looking down a few blind alleys, what about Kant? He did, after all, write numerous very long sentences on both the starry firmament and the moral law. But it is no easy matter to bring Kant’s views to bear on the problem as we have stated it, and in the juggernaut of contemporary meta-ethics he has not been in the driver’s seat.

The task before us is to try to squeeze morality into the “disenchanted” natural world; as Blackburn says, this “is above all to refuse appeal to a supernatural order.” One might object that this is to stack the deck: these ground rules exclude the obvious source of morality, namely God. Although Kant himself did not hold that morality is of divine origin, the view is suggested by his phrase “the moral law.” Human laws (“Thou shalt not smoke in bars”) are made by humans; who else could have made moral laws (“Thou shalt not covet thy neighbor’s ox, nor his ass”) but the Supreme Lawgiver himself?

This “divine command” theory of morality has the rather alarming consequence that—to borrow an aphorism Sartre attributed to Dostoevsky’s Ivan Karamazov—if God is dead, everything is permitted. The more fundamental difficulty, however, was pointed out by Socrates in Plato’s dialogue Euthyphro. Do the gods love good things because they are good, or are good things good because the gods love them? Surely the former—if Zeus, Uranus, and the rest started loving pointless suffering that would not make pointless suffering good. No doubt God, if there is one, enjoins us to avoid pointless suffering, but that is not why pointless suffering is bad. It is bad anyway—that is precisely why God enjoins us to avoid it.

Divine-command theory can be watered down in various ways and in recent years has experienced a minor revival; even diluted, it remains a fringe position. A considerably more popular suggestion is that moral facts can be squeezed into the natural world with no effort at all, because moral facts are actually natural facts in disguise. And if this is right, Hume was completely wrong. “Ought” does not express “a new relation or affirmation”: an “ought” turns out to be a kind of “is.”

What kind of “is,” though? Here’s one simple idea: “Stealing is wrong” and “People ought not to steal” are fancy ways of saying “I disapprove of stealing.” And if they are, then moral facts just are natural facts (specifically, psychological facts), their naturalistic credentials obscured by language.

Unfortunately, this idea is just too simple to be true. One problem, pointed out by Moore in his 1912 book Ethics, is that the view cannot accommodate the phenomenon of ethical disagreement. If I say “Stealing is wrong,” and you say “No it isn’t,” you are denying what I am asserting: if I spoke truly, you spoke falsely. But according to the simple idea, I am in effect saying “I approve of stealing,” and you are saying “I do not approve of stealing,” and if that is so, then we aren’t disagreeing and can both be right. Similarly, if I say “I live in Cambridge” and you say “I live in Somerville,” we aren’t disagreeing and can both be right, which we will be if I live in Cambridge and you live in Somerville. And if that objection isn’t completely convincing, here’s another one. The simple idea implies that a person’s attitudes settle the truth of her moral claims, which is obviously mistaken. If someone disapproves of interracial dating that doesn’t mean that she speaks truly when she says “Interracial dating is immoral.”

What if we switch from the first person to the third? Suppose that “Stealing is wrong” means “The community disapproves of stealing,” or something similar. That would at least allow for moral disagreement. In our little dialogue I would be saying “The community disapproves of stealing,” and you would be saying “The community doesn’t disapprove,” and one of us must be wrong. But the second objection has not been deflected at all: if the community follows Leviticus and endorses slavery, that hardly makes slavery permissible.

If moral facts are going to turn out to be natural facts in disguise, a much more sophisticated approach is needed. One idea—going back to Aristotle—is to ground morality in biological functions and purposes. As defended in Philippa Foot’s recent Natural Goodness, the idea is that the naturalistic foundations of morality can be dimly glimpsed in evaluative (yet naturalistic) facts about organisms, such as “This tree has good roots.” And if that doesn’t appeal, then there are many more suggestions on the menu. Consensus on their respective merits, however, is not exactly imminent.

Modern naturalistic theories of morality are reactions to the challenge laid down in Moore’s Principia Ethica. Moore argued that although the moral facts do not have their source in any deity, neither are they facts about happiness, attitudes of approval and disproval, human biology, or any other kind of natural fact. It is true that we ought not to boil babies—but this is not a natural fact or truth. It is a “non-natural” fact, in short. To identify the moral facts with natural ones was, Moore charged, to commit “the naturalistic fallacy.” This was a miserable choice of terminology, as Moore later admitted: to commit the naturalistic fallacy was simply to disagree with Moore, not to make (as the term “fallacy” suggests) an error of reasoning.

The divine-command theory and Moorean non-naturalism both reject the demand to find room for the moral law in the starry firmament: no room is needed, because the moral law has found agreeable accommodation elsewhere. More radically, the demand may be rejected by denying that there is a moral law to begin with—that is, by denying that there are any moral facts or truths. If there aren’t any moral facts—if boiling babies for fun isn’t wrong—then the problem of finding room for them doesn’t arise.

How could it turn out that there are no moral facts? Perhaps morality is bunk, like various theories of the starry firmament—astrology, say, or the ancient Greek astronomical theory of crystalline spheres. All distinctive astrological claims about the influence of the planets on daily life are false; likewise the claim (for example) that the stars are on the largest crystalline sphere centered on the Earth—that’s false, because there is no such sphere. If morality is a similar tissue of confusion, then it’s false that boiling babies is wrong. That is, someone who boils babies is not doing anything wrong. And this doesn’t mean that his actions are right—all claims of these kinds are false too.

We will get back to this vertiginous view in a moment. Before doing that, we should look at another less obvious route to the conclusion that there are no moral facts—a garden path that 20th-century meta-ethics went down again and again.

We use language to try to state facts, or assert things. “The pub is open” is typically used to assert that the pub is open. If it is open, then the speaker has made a successful assertion: she has stated a fact, namely the fact that the pub is open. But there’s much more to conversation than the exchange of information. Someone might ask “Is the pub open?,” or give the order “Close the pub!,” or express her delight that the pub is open by saying “Yippee, the pub is open!,” to give three of many examples. Someone who asks “Is the pub open?” is not attempting to state that the world is a certain way. And although someone who says “Yippee, the pub is open!” is stating that the pub is open, she is doing much more than that, namely expressing her delight at that fact. If she had just said “Yippee!,” she would have expressed her delight without stating anything.

That is one observation. Here is another: our impression of how parts of language work is sometimes seriously mistaken. “It was Russell who performed the service of showing that the apparent logical form of a proposition need not be its real one,” wrote Ludwig Wittgenstein in his Tractatus, referring to Bertrand Russell’s seminal 1905 paper “On Denoting,” which argued that the semantics of definite descriptions (like “the author of Waverley” or “the closest planet to the sun”) are quite unlike what one might naively take them to be. According to Russell, “the author of Waverley” is not a kind of name for Sir Walter Scott, but is instead an expression like “some author” and “all authors”, which are not names at all.

Let us apply these two observations to the case at hand. On the face of it, someone who says “Stealing is wrong,” or “Silvio ought not to shoot Adriana,” is attempting to state a moral fact, just as someone who says “Tony lives in New Jersey” is attempting to state a geographical fact. If stealing isn’t wrong—if the world isn’t that way—then the speaker’s attempt failed, and she spoke falsely; similarly, if Tony doesn’t live in New Jersey. But—drawing on the second observation—appearances can be misleading. Perhaps someone who says “Stealing is wrong” is not making an assertion about some mysterious realm of moral right and wrong, but rather is doing something else. What could this other thing be, though? Drawing on the first observation, language is sometimes used not to state facts but to express the speaker’s attitudes—delight, dismay, and so on. And, as it happens, if someone says “Stealing is wrong” she typically disapproves of stealing. So, perhaps when someone says “Stealing is wrong,” she is not attempting to state any moral fact, but is rather choosing a (somewhat imperspicuous) way of expressing her disapproval of stealing.

This “emotivist” view, that moral language doesn’t have the function of stating moral facts but serves rather to express the speaker’s attitudes, was independently developed by the American philosopher C.L. Stevenson and the British philosopher A.J. Ayer in the 1930s. (The basic idea was given an important twist in 1952 by another British philosopher, R.M. Hare, in a book called—with a nod to the “ordinary language” philosophy of the time—The Language of Morals.) In saying that “You acted wrongly in stealing that money,” Ayer wrote, “it is as if I had said, ‘You stole that money,’ in a peculiar tone of horror, or written it with the addition of some special exclamation marks.”

It is crucial to distinguish emotivism from the subjectivist view discussed earlier, on which “Stealing is wrong” means something like “I disapprove of stealing.” Someone who says “I disapprove of stealing” is making a psychological claim about herself—if she doesn’t disapprove of stealing, then she has spoken falsely. On the emotivist view, “Stealing is wrong” can be roughly paraphrased as “Boo to stealing!”—and someone who says “Boo to stealing!” is not making a psychological claim, or indeed any claim at all.

Emotivism solves the problem of finding room for morality in the natural world quite neatly. No room needs to be made for moral facts, because there aren’t any. But the absence of moral facts is no strike against moral talk, because it was never in the fact-stating line of work—it serves the function of expressing attitudes instead. The whole quest for the ground of moral truth is like Ponce de Leon’s search for the fountain of youth—misconceived from the beginning, because there’s no such thing to be found.

And emotivism is the theory that keeps on giving. A little earlier we noted the connection between saying that “Stealing is wrong” and disapproval of stealing. Put another way, if someone says “Stealing is wrong” then almost invariably she is averse to stealing, and so is motivated to avoid stealing. (Think of the oddity of someone saying “I agree, of course, that stealing is wrong, but I have not the faintest inclination to curb my shoplifting habit.”) But if morality is a factual matter, how can this close connection between sincere moral assertion and motivation be explained? If someone asserts a plain matter of fact—that there’s free beer at the pub, say—then she is merely expressing her belief that there’s free beer at the pub. And, it seems, a belief by itself doesn’t motivate anyone to do anything—a point famously made by Hume. (There is no oddity in saying “I agree, of course, that there’s free beer at the pub, but I haven’t the faintest inclination to go there.”) Emotivism, by contrast, can explain the connection between saying “Stealing is wrong” and the motivation to avoid stealing very easily—it’s no more mysterious than the connection between saying “Broccoli, yuck!” and the motivation to avoid eating broccoli.

(The importance of this tension between the apparently factual status of morality and its motivational force cannot be overstated; it might well be called, as it is in Michael Smith’s book of the same title, “the moral problem.”)

Despite its great benefits, emotivism is too clever to be true. The really crushing objection was made by a contemporary of Moore’s, W.D. Ross, and much later, in expanded form, by the philosophers Peter Geach and John Searle. Emotivism works nicely for standalone moral sentences, like “Eating meat is wrong,” but of course sentences can occur as parts of larger sentences. For instance, the sentence “Eating meat is wrong” occurs in the subordinate clause of the conditional sentence “If eating meat is wrong then Meadow should have the salad.” Consider now the argument:

1. If eating meat is wrong then Meadow should have the salad

2. Eating meat is wrong; so

3. Meadow should have the salad.

This is a good (“valid”) argument: it is an instance of the logical rule called “modus ponens”: from “If P then Q” and “P,” infer “Q.” Crucially, it is only a good argument because “P” is univocal throughout—it means the same thing when it occurs alone and when it occurs as part of “If P then Q.” For example, take “pen,” which is ambiguous between “writing instrument” and “enclosure,” and consider the following argument: “If the pig is in the pen then the pig is very tiny; the pig is in the pen; so the pig is very tiny.” Assuming that “pen” changes its meaning half-way through, this argument is plainly not good—the conclusion doesn’t follow from the premises.

Now, “If eating meat is wrong then Meadow should have the salad; eating meat is wrong; so Meadow should have the salad” is a good argument. As we have just seen, if this is a good argument, then its component words must be univocal throughout. However, notice that the emotivist must tell some special story about the meaning of complex sentences like “If eating meat is wrong then Meadow should have the salad.” “Eating meat is wrong” as it occurs in the subordinate clause cannot mean “Boo to eating meat!” because “If boo to eating meat! then Meadow should have the salad” is not remotely grammatical, and makes no sense at all. In other words: if emotivism is true, “Eating meat is wrong” changes its meaning halfway through the argument, which accordingly fails. However, the argument is perfectly good, so emotivism is false.

Conclusive refutations of philosophical positions are about as rare as sightings of the ivory-billed woodpecker. Socrates pretty much drove a stake through the heart of the divine-command theory, and the Ross-Geach-Searle objection comes awfully close to doing the same to emotivism. But philosophers rarely allow theories to expire naturally; largely thanks to surgery performed by Simon Blackburn and the American philosopher Allan Gibbard, various sophisticated descendants of emotivism live on. A very impressive effort, but arguably Sisyphus had an easier job. (For a clear presentation of some difficulties, see chapter five of Mark Kalderon’s Moral Fictionalism. Gibbard’s most recent book, Thinking How to Live, and Blackburn’s Ruling Passions both represent something of a retreat from the more straightforward forms of emotivism they had defended earlier.)

Emotivists tell us not to worry that moral talk is not true: it is not false either, because it has a different purpose than describing reality. Let’s get back to the alternative idea that moral talk is bunk: it is not true because it is false. John Mackie, an Australian philosopher who taught at Oxford University for many years, wrote a book called Ethics in 1977, with the subversive subtitle “Inventing Right and Wrong.” In the first chapter of that book, Mackie argued that morality is, like astrology and the theory of the crystalline spheres, a false theory. (This nihilistic chapter is a staple of introductory classes on meta-ethics, a fact fortunately unknown to those who complain that the academy is corrupting the youth with moral relativism.)

To say that a theory is false is not necessarily to say that we should reject it as entirely useless. That did happen with the theory of the crystalline spheres, because truth-telling was its raison d’etre; deprived of that, and without being especially useful for navigation, the theory was so much dead weight. Astrology, though, despite being false, lives on even among non-believers as a harmless source of amusement. Someone who says “Aquarians are unconventional” at a party might just be playing along, simply pretending to assert in full seriousness that Aquarians are unconventional.

Indeed, Mackie himself, after he has argued in chapter one that it is false that we ought not to boil babies, that it is false that we ought to keep our promises, and so on, spends most of the rest of the book discussing which moral principles we should adopt. That might sound paradoxical, but Mackie is not recommending that we believe moral claims (which he thinks are all false), but rather that we act as if some of them are true. Why should we do that? Mackie takes a hint from Hume, and the 17th-century philosopher Thomas Hobbes: in a nutshell, playing the moral game serves our interests. In particular, it is a “device for counteracting limited sympathies”: paying lip service to a system of morality, despite its falsity, greases the wheels of social cooperation. Mackie adds that the adoption of a moral system might be an evolutionary adaptation, an hypothesis discussed at length in Marc Hauser’s recent Moral Minds.

According to Mackie, philosophy has exposed our commitment to moral claims as a gigantic error. But there is another possibility, explored in Kalderon’s Moral Fictionalism, namely that we were never seriously committed to moral claims in the first place. Admittedly, the ordinary person says “Stealing is wrong,” but someone’s commitments cannot be so easily read from what she says. Maybe when the ordinary person says “Stealing is wrong” she is rather like our skeptical partygoer who says “Aquarians are unconventional.” Or perhaps she is like someone who, in a conversation about detective fiction, says “The world’s most famous detective lived at 221B Baker Street”—even though no detective has ever lived there, the speaker is not mistaken.

Why did Mackie think morality was all false? The argument on which he rested most weight was the so-called “argument from queerness,” which can also be found in Wittgenstein’s “Lecture on Ethics.” If there were moral facts, Mackie says, “they would be entities or qualities or relations of a very strange sort, utterly different from anything else in the universe.” Moral “oughts” and “ought nots” are too “metaphysically queer” to squeeze into a disenchanted nature: we must conclude that they apply to no one.

Why is a moral “ought”—say, “You ought to pull Virginia from the river”—metaphysically queer? Because knowledge of it would provide the knower with a motive to pull Virginia from the river. An “objective good,” according to Mackie, “would be sought by anyone who was acquainted with it . . . not because he desires this end, but because the end has to-be-pursuedness somehow built into it.” And that, he thinks, is very mysterious. If I know that Virginia has just jumped into the Ouse, that by itself is not going to motivate me to pull her out—for that, I need to want to save her. Knowledge of any fact—knowledge that the world is a certain way—should not by itself motivate: one could always imagine the knower shrugging her shoulders: the world is that way, so what?

This should be reminiscent of the Humean claim, mentioned earlier, that beliefs by themselves never motivate. If I believe that Virginia has just jumped into the Ouse, that by itself is not going to motivate me to pull her out. Indeed, Mackie’s claim about the motivational inertness of knowledge stands or falls with the Humean thesis about the motivational inertness of belief—if you hold one, you have to hold the other.

But this point undermines Mackie’s own “error theory” of morality. According to Mackie, we don’t know that we ought not to boil babies because, if we did, that knowledge would mysteriously motivate us. Instead of knowing that we ought not to boil babies, Mackie thinks, we (mistakenly) believe that we ought not to boil babies—hence the error theory. But this is to forget that the motivational impotence of knowledge goes hand in glove with the motivational impotence of belief. If Mackie’s argument against moral knowledge is any good, it can be pushed further: we don’t (mistakenly) believe that we ought to boil babies because, if we did, that belief would mysteriously motivate us. Put another way: if one can learn to live with moral belief (as Mackie did), there is no obstacle to living with moral knowledge.

This is not yet to acquit morality of the charge of falsity. There are other arguments for the prosecution, some of which are due to Mackie, and some to later writers (see, in particular, Kalderon’s Moral Fictionalism and Richard Joyce’s The Myth of Morality). Still, the presumption of innocence applies.

What have we learnt from a century of meta-ethical theorizing, inaugurated by Moore’s Principia Ethica? Too much to explain here; three lessons will have to do.

One surprising lesson is that the difference between non-naturalists like Moore and those who think that moral facts are natural facts in disguise is actually quite elusive. Moore agreed that the naturalistic facts are fundamental in the sense that they implicitly settle moral claims. Once all the facts about the location of sand grains are in place, the facts about sand dunes are implicitly settled. Similarly, once all the naturalistic facts about suffering, enjoyment, and so on are in place, the moral facts are implicitly settled: an “ought” does follow from an “is.” Indeed, Moore was the first philosopher clearly to point this out! The locus of dispute between the naturalists and the non-naturalists is not, then, whether a moral “ought” follows from a natural “is.” If there is a real dispute here, it is by no means easy to characterize.

Another lesson is that the deep problems of meta-ethics are not just about ethics. To a significant extent morality is not a self-contained system with its own proprietary vocabulary and problems: it is inextricably tangled with our normative and evaluative thought and talk in general, which extends to reasons, rationality, aesthetics, etiquette, and much else besides. Concerns about the status of morality soon spread like spilled ink: if there’s no room for ethics in a disenchanted nature, most of our distinctively human form of life is also excluded.

And finally, what’s the verdict on the naive thought that, as a plain matter of fact, boiling babies for fun is wrong? A little bit of philosophy—always a dangerous thing—can make it seem an outdated relic of our Pleistocene way of thinking. But more philosophy—about 100 years’ worth—shows that it might even be right. To be so close to one’s commonsense starting point after such a long journey can seem a little disappointing, but travel is supposed to broaden the mind. Philosophy, Bertrand Russell wrote, “removes the somewhat arrogant dogmatism of those who have never travelled into the region of liberating doubt, and it keeps alive our sense of wonder by showing familiar things in an unfamiliar aspect.”