Introduction
Huemer writes Ethical Intuitionism with the exasperated air of a biology teacher at a Christian school, who is having to explain, once again, why the fact that apes do not give birth to humans does not, in fact, mean that evolution is false. And there’s a reason for that: Heumer is defending the much-maligned doctrine of ethical intuitionism, according to which we come to know moral truths by intuitions—by their intellectual appearances. The same way that we can see with our intellects that the shortest distance between two points is a straight line (assuming space isn’t curved, you pedants!), there can’t be married bachelors, and that modus ponens is a valid rule of inference, we can see with our intellect that torture is wrong, pain is worse than pleasure generally, and that if A is better than B and B is better than C, then A will be better than C.
Why is Huemer frustrated? Well, because the doctrine of ethical intuitionism—which seems to be the way that most people do ethics most of the time—is generally not taken especially seriously, usually for utterly terrible reasons. And this was especially true when he was writing the book—it’s made something of a comeback in recent years. The arguments against it, and the objective morality he derives from it, are completely terrible, and yet most people dismiss it based on these confused criticisms.
Huemer’s favorite example of this is the argument from disagreement. People assert that the fact that there is moral disagreement shows that ethical intuitionism is false. But this is an utterly bizarre claim—people sometimes give conflicting reports about what they saw, but that doesn’t mean that you should never trust your vision. Disagreement proves that some way of knowing things is not infallible, but it doesn’t prove that it’s totally unreliable.
Huemer’s methodology for ethics is relatively straightforward and seems like the methodology that most people apply to most things: think hard and then believe whatever is most obvious. This seems like a good approach; if you have a conflict between two of your beliefs, you should believe whichever one is more plausible. If you discover that your belief that Bill Clinton was a good president conflicts with your belief that it’s wrong to bomb pharmacies—as a matter of fact, it does—then you should give up whichever is less obvious, namely, the belief that Clinton was good.
Huemer is arguing for moral realism in his book—the idea that there are some moral statements that are true that aren’t made true by anyone’s attitude. He also argues for moral non-reductionism—the idea that moral statements are not explicable purely in non-moral terms. His basic argument is pretty straightforward, all of the specific kinds of anti-realism have to say things that are really implausible. Take error theory—the view that all moral statements are false—as an example: the error theorist has to think that all of the following statements are false:
You shouldn’t torture innocent people absent a really good reason.
Pain is bad.
Being virtuous is generally better than being vicious.
If you see a baby on the street, you should not press a needle into its eye.
The holocaust was bad.
Serial killers do immoral things.
The Nazi Odilo Globocnik might be, as one historian claimed, "the vilest individual in the vilest organization ever known".
But these are all very obviously true. Because moral realists say that there are statements about morality that are true, and that aren’t made true by our attitudes, the anti-realist has to think that every specific moral proposition is either neither true nor false, false, or true but made true by someone’s attitude. But this isn’t believable. Take the sentence: it’s wrong to put little babies in the blender and turn it on.
That doesn’t seem neither true nor false. If it were neither true more false, you wouldn’t be able to say things like “because it’s wrong to put babies in the blender, if I did that it would be wrong,” any more than you can say “if, don’t put babies in the blenders, then I should don’t put babies in the blender.” You can’t have if sentences about things that are neither true nor false.
It also doesn’t seem false. It seems true that it’s wrong to torture babies. And it doesn’t seem like its wrongness is caused by anyone’s attitudes; if everyone approved of torturing babies, it would still be wrong. No one’s disapproval of torturing babies is what makes it wrong, instead the wrong-making feature is the fact that it inflicts lots of suffering on babies.
Huemer’s assault on non-cognitivism
Huemer’s second chapter is dedicated to non-cognitivism, the idea that moral sentences are neither true nor false. He argues that it’s just wrong as a semantic account. Moral sentences have various features that are only had by propositions, including:
Moral sentences take the form of declarative sentences—sentences of the form A is B. We say things like “torture is wrong,” but we don’t do this with non-propositions. We don’t say things like “go is for the dodgers.”
Moral properties can become abstract nouns—the word goodness exists, as does the word rightness. There is no analogous abstract noun for non-propositions—no one says “I am questioning the acts hurrayness,” the way they’d say “I am questioning the acts rightness.”
You can have sentences of the form “it is true that X,” where X is a moral statement. You can’t do this with things that are neither true nor false; no one says “it is true that go dodgers,” but you do say, “it is true that I exist.”
One can attach propositional attitude verbs to evaluative statements. One can say, “I hope that abortion is fine,” while they can’t say, “I hope that ouch!”
One can ask questions about moral properties, in a way one can’t do for things that are neither true nor false. You can say, “I wonder if abortion is okay,” while you can’t say, “I wonder if boo! redsocks.”
You can have complex arguments involving moral premises. You can say “if abortion is wrong, then God will punish people who get abortions.” This would be impossible if abortion being wrong was neither true nor false—one can’t have conditional claims about sentences that describe nothing about the world.
Non-cognitivists have various ways of trying to explain the fact that their theory, on its face, is obviously radically out of accords with how most people use language. In order to do so, they make their theory more like realism, wherein the moral statements are able to be true or false, but their truth doesn’t mean what it doesn’t in other domains. Huemer argues that none of these solutions can avoid the sixth problem I listed, and that they each run into other problems.
Finally, Huemer advances a pretty interesting argument against non-cognitivism: We are just introspectively aware that moral sentences are truth-apt—capable of being true or false. When you say sentences like “torture is wrong,” it feels like you are saying something is true or false. You’re not giving commands—when I say “Hitler shouldn’t have killed people,” I am not telling Hitler what to do. When I say, “pain is bad,” I’m not instructing pain on how it should behave. And it’s clearly not an emotion—I have no emotional attachment to the claim that if A is better than B and B is better than C then A is better than C or that tax fraud committed in 350 CE that ended up harming others was wrong. None of the accounts of non-cognitivism account for the conditions in which we make moral statements, which are often conditions of cool reflection rather than emotion, and which certainly don’t involve telling anyone what to do. Huemer notes:
our emotions about some things are stronger than about other things. If the core explanation for the correlation between moral judgment and emotion is that 'good' and 'bad' are just emotion-expressing terms, then what should we predict about how the strength of moral emotions should vary? It seems that the strength of the emotions associated with moral judgments should be roughly proportional to the perceived level of goodness or badness of a thing and to the confidence of one's moral judgment: the worse one takes a bad thing to be, and the more confident one is in one's judgment, the more strongly one should feel about it. Is this true? In some cases, yes: I feel more strongly about the wrongness of murder than I do about the wrongness of jaywalking. In other cases, no: I feel more strongly about someone's stealing my stereo (and this is a moral emotion, indignation) than I do about Emperor Nero's execution of Octavia in 62 A.D. Truth be told, I have little if any feeling about Octavia's murder. But Nero's execution of Octavia was far more wrong than the stealing of my stereo, and I am quite certain of that.
…
Third, we can distinguish between the degree of confidence of a moral judgment and the degree of rightness or wrongness that the moral judgment attributes. Imagine someone who thinks abortion is likely wrong but lacks confidence in that judgment. He sees that there are plausible arguments on the other side. However, he says, if abortion is wrong, it is very seriously wrong, since the only way it would be wrong would be by being akin to murder. Here, the perceived degree of wrongness is very high, but the confidence in the judgment is very low. On the emotivist theory, what could these two dimensions correspond to? On the emotivist theory, one simply experiences a feeling about abortion. This feeling could come in degrees of intensity-people can feel more or less strongly about something. But how could this give rise to our sense that there are two things here that come in degrees and that may have radically different degrees in a given case?
Skewering subjectivism
Huemer’s next chapter is on subjectivism. He rehashes the standard objections: individual subjectivism means that we’re morally infallible and that if I approved of torture and uttered the sentence “torture is great,” I’d be right. But that’s obviously false. His discussion of ideal observer theory is the most interesting.
The ideal observer theory says that morality is about what an ideal observer would judge. So if we say murder is wrong, that means that some ideal observer would disapprove of X. This ideal observer is supposed to have all relevant factual knowledge, be perfectly good at vivid imagining, and possess various other things that make one ideal. So he’s supposed to be sort of like me, but a tad bit better at vividly imagining scenarios, albeit probably less handsome. Note we cannot say that the observer is morally ideal, because the ideal observer theory is supposed to account for morality, so we can’t build morality into the ideal observer.
Now, one version of ideal observers says that something is wrong if all ideal observers would disapprove of it. But given that there’s nothing constraining their desires, there’s nothing that all ideal observers would agree upon. So this means nothing is wrong. A better thought is that the relevant ideal observer is the ideal version of the speaker: So if John says torture is wrong, that means “the ideal version of John would hate torture.” But this has a lot of problems.
First, there’s the problem of horrible desires. Suppose that Hitler, even after knowing what his victims experienced and being informed of all the relevant facts still supported killing them. On this account, when Hitler says “ze Jews should be killed,” that would be true. But that’s obviously false.
Second, there’s the problem of disagreement. If I say, “torture is wrong,” and that just means that my ideal self would disapprove of torture, someone whose ideal self could approve of torture could say “what you said is true, but torture isn’t wrong,” and this might be true! But this couldn’t be true. It can’t be true that when I say torture is wrong it’s true and when he says it isn’t, it’s not true.
Third, there’s the problem of fallibility. It seems like even non-morally ideal agents can be morally mistaken—if you know all the non-moral facts, you might still be wrong about the moral facts. But this means that the non-moral facts don’t settle the moral facts.
Fourth, there’s the problem of arbitrariness: why does the ideal observer want particular things? He can’t desire it because it’s good, because the ideal observer theory is supposed to account for goodness, not suppose it. But then what he wants has to be arbitrary—not based on things that are worth desiring. But arbitrary whims can’t ground morality.
Now, the ideal observer theorist can stipulate that he is kind, for example, meaning he’ll only like good things. But if you stipulate that ideal observers have to have particular kinds of desires because only a particular type of ideal observer counts, then that’s a kind of moral naturalism. If you think ideal observers only count if they value X, Y, and Z, then you’re just stipulating that morality is about valuing X, Y, and Z.
Reducing the prominence of ethical reductionism
Huemer’s chapter on ethical reductionism is, in my mind, the least convincing. While he replies to some of the ways that people might be moral naturalists, he does not reply to what is, in my mind, the most convincing version of moral naturalism. Ethical reductionists are those who believe in moral facts but think that they are just a kind of natural fact. Thus, when one says “torture is wrong,” that is just a statement about various natural facts. Perhaps it means something like “torture causes lots of severe pain.”
Huemer argues convincingly that many of the ways that people claim we can know the moral facts according to moral naturalism are wrong. For example, he claims that we cannot observe the moral properties. When people say that they see that something is wrong, they do not literally mean that its wrongness emits photons that they observe. So the moral facts can’t be used to explain what we see.
But I think the best version of moral naturalism relates to conceptual analysis, and Huemer doesn’t spend any time addressing that. On this account, when we say X is wrong, we’re referring to some hard-to-define natural property. We determine which things are right or wrong by reflecting on our moral intuitions, just like we determine what the word knowledge picks out by introspecting on our intuitions about which things we know. Now, while this means that moral claims aren’t discovered through empirical evidence, they’re discovered through empirical analysis. And while this is maybe sort of kind of analytic naturalism, it effortlessly avoids the traditional problems with analytic naturalism.
The most common issue—and the one Huemer addresses—with analytic naturalism is the open-question argument. You can wonder whether pleasure is good, but you can’t wonder whether pleasure is pleasure. Therefore, goodness can’t just mean pleasure. But you can coherently wonder whether some type of conceptual analysis is correct. We can, for example, coherently wonder whether knowledge is justified true belief. Therefore, conceptual analysis seems to just easily resolve all of Huemer’s objections to moral naturalism, except for one.
This last argument is sort of interesting, and I think it represents the fundamental reason to think ethical naturalism is false:
1. Value properties are radically different from natural properties.
2. If two things are radically different, then one is not reducible to the other.
3. So value properties are not reducible to natural properties.
If this argument sounds thin to you, it is probably because you accept one of the following two responses.
First, a reductionist might say that we cannot simply see the nature of a thing merely because we have a concept of it. We could not see, merely from our concept of water, that water was identical with H20. But my argument doesn't assume that we can see everything about the nature of those things we have concepts of; it only assumes that we can see some things about them, and that in the case of physical and psychological properties, we can see that they are not evaluative. To illustrate, suppose a philosopher proposes that the planet Neptune is Beethoven's Ninth Symphony. I think we can see that that is false, simply by virtue of our concept of Neptune and our concept of symphonies. Neptune is an entirely different kind of thing from Beethoven's Ninth Symphony. No further argument is needed. Indeed, if a person couldn’t see that Neptune is not a symphony, we would say he either had no idea what Neptune was, or had no idea what a symphony was. Second, a reductionist might say that the argument, if accepted, would also 'refute' the theory that Water =H20, for 'water' and 'H2 0' just seem very different on their face. Similarly, the argument would seem to refute the theory that sound consists of compression waves in the air, because 'sound' and 'compression waves' sem very different. But this is not so. On the face of it, water doesn1t seem to be H20; but it is not the case that water seems to not be H20. Our pre-scientific concept of water takes it to be a clear, odorless, tasteless, etc., liquid. We cannot, on the basis of this concept, discern anything about what its micro-structure might be like. We have no experience of just seeing that there are no tiny particles composing it, and so on. Nor do we have the sense that water is a different category of thing from H20; on the contrary, we see that they are the same category of thing, namely, physical substances.
This argument is, I think, convincing, and basically no one ever makes it. But I think it’s right!
The rest of the book
The rest of the book is pretty standard; Huemer replies to almost all of the major arguments for moral anti-realism. He points out that a lot of the arguments for moral anti-realism are just utterly unconvincing. For example, take the argument from queerness. This argument says that morality can’t exist because it would be too strange, utterly different from everything else in the universe.
Why is morality supposed to be weird? Well, the answer is that it’s different from other things. But so are a lot of things. Here is a nice list of things that Huemer gives that are radically different from other things
time
space numbers
propositions
substances properties
relationships
mental states
physical states aesthetic
properties
fields (e.g., the gravitational field)
the past dispositional properties
moral properties
For this argument to be convincing, one would have to think that things that are very different from other things probably don’t exist. But this is false. There are loads of things that are very different from other things. So we have no reason to think that unique things are different. It’s also not clear why morality is so weird; it just seems to be a brute intuition. But that’s not a good argument!
Huemer’s treatment of the evolutionary debunking argument is interesting. He starts out by saying that the moral realist explanation of how we come to have moral beliefs is pretty standard; the same way we reason about other things. This account is, I think, broadly right—and it allows us to vindicate our moral beliefs, however we acquired them. He then on top of this remarks at just how poor the evolutionary debunker’s explanation is. It always involves some extremely amorphous just-so story appealing to vague norms like “cooperation.” It never is specific or predictive—if our moral beliefs were totally different from what they are, it could “explain” them just as well with some implausible just-so story. Our moral beliefs that people on the other side of the world matter, for example, don’t seem to directly enhance our evolutionary survival. I think this is right—the evolutionary debunking story ends up being terribly unconvincing on purely empirical grounds.
All in all, Huemer’s book is extremely convincing, and very worth reading. It is, in my mind, probably the best defense of moral realism, with the possible exception of On What Matters. One other thing: the book is absolutely hilarious. I have never laughed harder at a book. I don’t want to spoil any jokes, but the book is sufficiently funny that people in adjacent rooms assumed that I was watching comedy. So even if you hate the philosophy, read it just for the humor.
Wait, if moral realism is true, doesn't it mean some philosopher one day will just pull a Moses and descend from the ivory tower bearing morality itself, that is, the correct moral positions to hold on every issue, and everyone will have to accept it, the way we are forced to accept the conclusions of science? And if that can't happen, then how can moral realism be true?
If we are going to try to make intuitionism work, we need to deal with the problem that sometimes our intuitions conflict (even within a single person). I guess it's fashionable to mention concepts like "reflective equilibrium" and then move on with life, but not many people appear to be really working on how to think about these things.
By the looks, Terry Hogan raises some interesting concerns about Bayesian epistemology:
https://thorgan.faculty.arizona.edu/sites/thorgan.faculty.arizona.edu/files/Troubles%20for%20Bayesian%20%20Formal%20Epistemology.pdf