2-D Semantics, 2 Problems For Moral Naturalism

Whether an action is wrong doesn't depend on whether you're in Dallas; but synthetic moral naturalism must deny that.

the west [2-D semantics] is leading Ukraine [moral naturalism] down the primrose path and the end result is that Ukraine [moral naturalism] is going to get wrecked

—John Mearsheimer

I’m currently rereading David Chalmers’ book The Conscious Mind. It’s probably the best book I’ve ever read on consciousness, and perhaps the best book I’ve ever read. It’s very, very convincing. Chalmers’ is an incredibly precise thinker. He develops (perhaps originally? I’m not super familiar with the history) a developed account of semantics.

The primary intension of a term denotes the things that are a priori knowable as being potential candidates for the term. So, for example, the primary intension of water might be “a clear liquid that has such and such nutritional properties, that evaporates, and that is important for humans to drink.” Before we conducted any empirical tests on water, before scientists figured out water was H20, we knew that about water.

When I say the things that are a priori knowable as being potential candidates for the term, I mean something rather specific. I mean that if we discovered that actual water had those properties we’d count it as water. So if we discovered that what we take to be water is really made of XYZ, but has the same properties as water, being a clear, healthy liquid that tastes like water, we’d call it water. In contrast, if it turned out that we had somehow been hallucinating drinking water, when we were actually drinking milk from a dragon—milk which is pearly white and not needed to survive—we would not call that water.

Words also have secondary intensions. The secondary intension of a word is what the word picks out given the actual facts about the world. So, because we know water is H20, the secondary intension of water is H20—if an alternative world has XYZ that has a similar chemical makeup to water, it wouldn’t have water, because facts about the actual world determine that water is H20.

Richard has a paper and a series of posts that make a very convincing post that a proper analysis of 2-D semantics straightforwardly wrecks moral naturalism. (Amusingly, when I first came across Richard’s 2-D argument against moral naturalism, before I knew about 2-D semantics, I thought that those were spatial dimensions—that somehow moral naturalism would be unable to accommodate morality in Flatland—which would be a pretty hilarious argument). I think this is most true of synthetic moral naturalism, which says that the moral facts are somehow discovered through observation rather than true by definition. These considerations are enough to cause me to think that analytic naturalism is a much better view than synthetic naturalism—though I still think analytic naturalism is implausible on account of the triviality objection, problem of substantiveness, Huemer’s various arguments, the moral equivalent of the knowledge argument, and the open question argument (Huemer has a nice reply in Ethical Intuitionism to the objection I’ve previously given to the open question argument. The very brief TLDR is that it’s plausible that if knowledge is justified true belief, then others who refer to other things as knowledge are using language wrong. But it’s not plausible that deontologists are, for example, misusing language, though it’s very plausible that they are substantively wrong).

The first problem that 2-D semantics poses for moral naturalism is unique to synthetic naturalism. Synthetic naturalism says that we discover the moral facts through observations about the world. Thus, just as we discover through empirical inquiry that water is H20 rather than, for instance, XYZ, we discover that the moral facts are whatever they are—utilitarian, perhaps, rather than Kantian. But this means that, as is true of water, the primary intensions of the moral terms differ from their secondary intensions. This is a general property of things that we discover synthetic facts about—for instance, water is H20 in its secondary intension, but its primary intension could be XYZ.

But, as Richard notes, this entails a troubling kind of parochialism. In order to know if some other world has water, we would need to know facts about the actual world. If water is H20 in the actual world, then alternative worlds would only have water if they have H20. In contrast, if water is XYZ in the actual world, then alternative worlds would have water only if they have XYZ. To tell if some alternative earth has water, you need to know facts about the actual earth. Water is parochial in the sense that to know if some other world has water, we need to know facts about the actual world.

But moral facts seem to be nonparochial in this respect. To tell whether some act is right, committed on another world, you do not need to know where you are located—and whether you’re located on a world where morality has some primary intension or some different primary intension. If the linguistic norms differ in Texas and California, and someone does something in Texas, to figure out whether it’s wrong, you do not need to know whether you are actually located in Texas or California. Whether some action is wrong doesn’t depend on where I’m physically located!

The second problem 2-D semantics raises for moral naturalism is that it allows the open-question argument to once again rear its ugly head. To avoid the open-question argument, synthetic naturalists hold that there is a difference between the meaning of a term, knowable a priori, and the secondary intension. But then the open-question argument can be made with the primary intension of the term. It makes no sense to ask “I know this is the actually existing clear liquid that fills lakes and rivers and has such and such nutritional properties akin to water, but is it water?” This is because the primary intension of a term is almost identical to its meaning. But this means that the OQI can be raised with the primary intensions of the moral terms. For example, suppose someone says that the moral facts are whatever the actually existing community of speakers says is moral. This will be its primary intension, and then the secondary intension will be whatever happens to be what actually existing speakers say. (Compare: X is whatever is the clear watery substance in a world with Y is the clear watery substance that actually exists in the world). But it seems we can coherently say “yes, I know that’s what actually existing speakers approve of, but is it right.”

I think these arguments are enough to pretty much doom synthetic naturalism. Anyone have any objections?