(part 1 of an n-part series about meta-ethics in the 20th century)
Inspired by Dana’s series on Leibniz and Spinoza, I thought I’d write a series of posts about some philosophical issues in my vicinity. Though this series will be different from hers in many ways (viz., neither interesting or true) I hope it will serve two useful functions: first, it will help give historians and other non-philosophers an idea of what it is we do, and second, it will make them feel ok about ignoring us altogether. I should also note that this is more or less a top-of-the-head exercise, so while I’m pretty sure the broad lines of reasoning are accurate I’m not going into the bookstacks to check any of this stuff. Undergraduate plagiarists take note.
What I want to do in this series is discuss some of the metaphysical and linguistic concerns that drove a lot of the metaethical discussion of the last hundred years. In particular I want to focus on the relationship between two questions: first, do moral properties– properties like rightness or goodness– fit into our best picture of the world, in particular with the picture drawn by the natural and social sciences? Second, how do moral judgments and concepts function? That is, what’s going on when we make judgments about, for example, an act’s being right or wrong? If you think that moral properties aren’t metaphysically problematic, you can give a straightforward answer to the second question: moral judgments try to get the moral facts right. If you suspect, as many have, that the answer to the first question is no, the second question becomes more pressing, because you need to provide some alternative account of what’s going on in moral judgment. What are we doing, if not describing moral reality?
It’s typical to begin a discussion of meta-ethics with a look at GE Moore, the British philosopher active at Cambridge over the first half of the 20th century (we missed the anniversary of his birth on November 4th, 1873), and, in particular, with Moore’s famous “Open Question Argument,” developed in Principia Ethica (1903). This is not to say that Moore’s argument is sound: there is widespread agreement that the argument as given fails, though there’s disagreement about whether it’s just a bad argument that happened to be very influential* or whether it’s a bad but influential argument that helps us to see important things about moral judgments and properties. On the other hand, regardless of its philosophical merits, the argument serves as a useful organizational tool, as we can use different responses to the OQA as a way to see the contrasts between various meta-ethical positions.
The OQA concerns the property good: Moore is interested in “the nature of that object or idea” that the word “good” stands for. His view:
good is good, and that is the end of the matter…[good] cannot be defined, and that is all I have to say about it.
Well, that helps. (There’s a reason Moore chose Butler’s remark “Everything is what it is, and not another thing” as the frontispiece for his book.) More precisely, Moore thought that the property of goodness was simple, non-natural, and indefinable. Making sense of this is hard, because Moore’s discussion is, or at least seems to be, really sloppy about the differences between questions about linguistic definitions, concepts, and modes of presentation, on one hand, and questions about properties, on the other. But roughly, Moore thinks that goodness is simple, because it cannot be broken down to component parts; it’s non-natural, because, again very roughly, it’s not the kind of thing studied by the natural sciences; it’s indefinable because it can’t be defined in terms of some other property.
But why think this? It’s an argument from elimination. Moore sees three possibilities: (i) his view; (ii) the view that goodness is analyzable in terms of other properties, and (iii) the view that “good” has no meaning. He argues against (ii) and (iii), leaving his view as the only option.
(iii) looks bad because we do understand claims about goodness, and that would be hard to account for if “good” were meaningless.
(ii) looks bad because…ah, finally, the Open Question Argument. Suppose goodness were analyzable in terms of some other property or properties. Suppose, to use Moore’s example, that something is good just in case we desire to desire it. Consider the question
I see that I desire to desire x, but is x good?
This, Moore says, is an open question: competent speakers can ask it, wonder about it, and so on. (Contrast “I see that Tom is a bachelor, but is he married?” or “I see that I desire to desire x, but do I desire to desire x?”) And in general, for any natural property N, we can ask an open question: “I see that x is N, but is it good?” This “shews,” Moore says, that “we have two different notions before our minds.” Moore concludes that any attempt to equate goodness with a natural property will fail. People who make this mistake, Moore says, are guilty of the naturalistic fallacy.
So, in a nutshell: Moore’s argument uses the “openness” of a question involving purportedly identical properties M and N– “I see that it’s M but is it N?”– to show that M and N are not identical.
(An aside on the naturalistic fallacy: it’s neither naturalistic nor a fallacy. Not naturalistic, because Moore’s argument works just as well against attempts to define goodness in non-natural (e.g. supernatural) terms. If you defined goodness in terms of God’s will, the open question argument would be just as effective. Not a fallacy, because the OQA is not actually compelling.)
So what’s wrong with the argument? One common response is to roll one’s eyes and chant the mantra “Water=H2O,” the standard example of a non-obvious property identity. Water is H2O, kinda, but at one point the question “I see that it’s water, but is it H2O?” was open, even if it no longer is. Point: property identities can be surprising, the sorts of things that can be discovered via empirical enquiry, and, most important for OQA purposes, the sorts of things that cannot be determined merely by reflecting on the concepts involved. If we think the OQA generalizes– that is, no property put in place of N produces a closed question– then we think that “good” isn’t synonymous with a term for N, or that the concept [good] is not the same as the concept [N]. But this is different from saying that the property of goodness is not identical to the property N. To put things another way: Moore’s test seems to rule out only the claim that G and N are identical by definition. It doesn’t rule out the sort of surprising, not-true-in-virtue-of-meaning identity that holds between water and H2O. Goodness could thus be identical to some N, even though the Moorean question is open.
Second, even as a claim about concepts, Moore’s argument is on shaky ground: it might be, after all, that there are true claims about two concepts being identical that are subject to doubt by competent speakers.
But suppose it’s 1903 and these things don’t occur to you. If you buy the argument, you’re left with (i): Moore’s view that goodness is a non-natural, non-analyzable property.
This is a problem, because Moore’s view is pretty weird. It leaves us without much to say about goodness, for one. Moore thought (or is widely believed to have thought) that we grasp moral principles directly via a faculty of intuition, which means moral argument comes down to whoever squints hardest into the mysterious realm of value. And non-natural stuff is less and less appealing to philosophers who are more and more enamored of the natural sciences. (See also: circle, Vienna.)
Suppose we don’t like Moorean intuitionism. What are the other options? Let’s return to (ii) and (iii). Eventually people will start going with option (ii): goodness is identical to some natural (or non-natural) property. But in the immediate aftermath a lot of people thought option (iii) was the way to go. They didn’t like Moore’s positive view, but they thought the OQA ruled out other options. So they concluded that “good” was, in a sense…meaningless.
More accurately, they thought statements about what’s good and right aren’t talking about some aspect of reality. Instead, they’re in a different line of work altogether– the function of moral claims isn’t to state how things are, but to express emotions, or make commands, or influence action, or…the positive proposals varied, but the core idea is that these terms are misleading. They look like they’re used to assert facts, but really they’re doing something else. “Stealing is wrong” is actually functioning as “boo stealing!” or “don’t steal!” or something like this.
This sort of view is usually called expressivism (because it says moral claims are expressing mental states like emotions or condemnations or whatever) or noncognitivism (because it says that moral claims express states that are not cognitive, i.e., not evaluable in terms of truth or falsity, i.e., not beliefs). And for a long time this view was really influential for a variety of reasons. Interesting questions: why does the OQA seem particularly appealing in the case of value even though it also looks like a bad argument? Is the OQA getting at something interesting even though it doesn’t really work on its own terms? And what should we make of this idea that moral claims don’t assert anything? Tune in next time to find out!
*Keynes reportedly described Moore’s work as “better than Plato.” Then again, Keynes single-handedly caused the Great Depression, so what would he know.