Actually, There Are Only Three Moral Anti-Realist Positions

Well, sort of, if we squint a little

I.

What is this? Two metaethics-posts in a row?! Last Thursday I decisively proved that metaethics is bunk, and that we shouldn’t care about it AT ALL. Yet see who’s swallowing his words now… Well, I still stand by what I said then, but I also just find the topic interesting. Back then I was pretty nice to

Lance S. Bush

, agreeing with him that anti-realists have no problem accounting for first-order morality. Not this time; there’s no more Mr. Nice Guy!

Lance has been going on a crusade against moral realism and moral realists for quite some time, and one thing he often complains about is that realists tend to only allow three positions for the anti-realist to take: Non-cognitivism, error theory, and relativism. This, he claims (e.g. here and here), is a mistake, as each of these positions make general claims about the meanings of moral statements. But what if these statements don’t have uniform meanings across utterances?

Crucially, the context here is usually that he’s objecting to a

Michael Huemer

-style process of elimination argument for realism. The problem is that this type of argument tries to establish realism by ruling out the alternatives, but if the alternatives ruled out aren’t exhaustive, then the process fails! However, we can simply stipulate an exhaustive set of definitions that do allow the argument to proceed (

Robert Hall

has a very similar (and good) post where makes pretty much exactly the same point—check it out!¹):

1. Non-Cognitivism: There are no truth-apt moral judgements.

2. Error Theory: There are some truth-apt moral judgements, and all such judgements are false.

3. Relativism: There are some (truth-apt and) true moral judgements, and all such judgements are stance-dependently true.

4. Realism: There are some (truth-apt and) stance-independently true moral judgements.

These definitions aren’t quite standard (though I think they’re quite close, actually). It would be very strange if there were a “correct” way of dividing up the positions, but if the point is simply to make sure that people know how the labels are usually used, then go ahead and object to this classification! However if the point is to object to the elimination-argument, then these stipulative definitions will avoid that, as we’ll see.

Not to go all symbolic on your ass, but we can formalize the positions as follows to show that they’re exhaustive:

Domain of discourse: Moral judgements

A: …is truth apt

T: …is true

S: …is stance-dependent

Non cognitivism: ¬(∃x)(Ax)

Error theory: (∃x)(Ax)∧(∀x)(Ax→¬Tx)

Relativism: (∃x)(Ax∧Tx)∧(∀x)((Ax∧Tx)→Sx)

Realism: (∃x)((Ax∧Tx)∧¬Sx)

These are all mutually exclusive and logically exhaustive (proof in footnote)²

---

II.

Typically the objection raised to non-cognitivism is the Frege-Geach problem: Moral judgements seem to be able to be used in antecedents, questions, etc. (e.g. “if breaking the human heart is wrong, then you should subscribe:”

). Statements should maintain their meaning between being asserted and being in an unasserted context (such as in an antecedent). However this isn’t the case if moral judgements express non-truth-apt attitudes. Suppose “not subscribing is wrong” expresses some non-belief attitude like Boo(not subscribing). Putting that in an unasserted context like “If Boo(not subscribing) then…” will no longer express the attitude, and so the meaning is no longer the same. But since moral judgements can be embedded in unasserted contexts, moral judgements must be something different.

The point here isn’t whether or not this objection works—let’s just suppose that it does. Then we rule out non-cognitivism as defined above: Some moral judgements must be truth-apt.

Let’s then zoom in on those judgements that are truth apt. The idea here is supposed to be that it’s very implausible that there aren’t some of these that are true. For example that killing jews in the holocaust is wrong (pulling out the Nazism right off the bat), that grinding babies in garbage disposals is wrong, and that not subscribing to Wonder and Aporia is worse than both of these combined. In fact it’s so obvious that any argument to the contrary will have less obvious premises. This means that it can’t be that all the truth-apt moral judgements are false.

Again, we’re not considering whether this is a good argument, just supposing that it works. And if it does, then that rules out error theory as defined above. That leaves relativism.

The Canonical argument against subjectivism is similar: Surely it’d be wrong to torture babies, grandmothers, and Joe on the street, even if everyone but them (including you) were fine with it! There are also more technical issues about individuation of cultures, what it means for a judgement to be stance-dependently true, etc.

Regardless, if this objection also works, then the anti-realist logical space has been exhausted! To avoid realism, one must deny at least one of these arguments. But what if—as I take it Lance roughly thinks—moral judgements aren’t uniform in what they express, some being emotings, some false beliefs about stance-independent facts, etc. Well those that aren’t truth-apt will be covered by the first argument, those that are by the second and third. Any position will fall into these categories, and objections should be to the arguments (probably mostly the one against error-theory, I suspect), not the framing.

---

III.

One objection Lance has sometimes raised is that sentences and words don’t mean anything, rather people mean things. I actually agree 100%!!! He thinks that this is a problem since dividing the positions as above assumes that the statements mean things, when it’s really the people that do. But the moral judgements quantified over in the definitions can simply be what people express with the words they use, not the sounds and letters produced, and everything should work the same.

I also suspect he might object that there isn’t any well-defined class of “moral judgements,” as it’s just a subset of ordinary discourse, with no clear boundary. This needn’t be a problem though, as we can just stipulate a minimal class of moral judgements by ostension: “Moral judgements at least include what I mean when I say “xxx,” and what she meant when she said “yyy,” and what they meant when they said “yyy,” and…” Once we have such a class, if the process of elimination works, then we have established realism, and from there the question is simply how far the domain of moral discourse should be extended.

I will admit that if claims can be neither truth-apt nor not truth-apt (as might be the case if Don Loeb’s incoherentism is right, or something like that), my classification isn’t entirely exhaustive, since it assumes that the predicates given apply consistently across the domain of moral judgements. This is something Lance is also good at pointing out, and something that I think realists making the argument should at least acknowledge and be aware of—so kudos where it’s due!

At the same time, I don’t actually see it being relevant to the process of elimination argument. This is because the argument against error theory is basically an argument to the effect that there are true moral judgements—and if there are true moral judgements, an option like incoherentism is also ruled out. If we want to be stringent, we could add a incoherentism (and similar positions) to the classification, and then call this step the argument against error theory and others. (I do think this makes the whole framing of a process of elimination a bit strange, as each step sort of subsumes the previous ones, but that’s not what we’re here to contemplate.)

Additionally, we might simply draw the class we call moral judgements to exclude all statements where the predicates in the definitions do not coherently apply. This would leave us with a class where we can again run the process of elimination. It might admittedly be a bit artificial, and could be problematic if it turns out that most of our everyday moral judgements don’t fit in this class.

---

IV.

So I don’t think the right place to object to this argument is to object to the framing. Rather objections should target the actual steps of the arguments (probably primarily the step against error theory)—something I won’t discuss here.³ When reading and listening to what Lance says, I also feel like this is mostly what he does. Objecting to the division of positions just ends up being a distraction, and an unnecessary one, if what I argue here is correct. If the goal is to make sure that people understand how metaethical positions are usually divided, then great, go ahead; but if the goal is to respond to this argument, then it should only be a side-note.

I hope this post hasn’t sounded too confrontational or something—that’s not my intent! While I tend to disagree with him on a lot of things, I think Lance does a lot of good gadflying of philosophy (especially being critical of philosophers coming up with armchair-statistics), and I enjoy a lot of his content! This was simply a point that has bothered me a few times when I’ve heard it. Also, I apologize for any point where I have misinterpreted or misrepresented his views (the devil made me do it), and I take full credit for any good points I’ve made.

You Might Also Like:

1. He frames the positions in terms of possibility. I think that’s probably actually the more accurate way to do it, but I guess I didn’t feel like it—and the difference is minor. Anyways, to revise the later formalizations, you’d just add diamonds and boxes as follows:

• ¬◇((∃x)(Ax))

• ◇((∃x)(Ax))∧□((∀x)(Ax→¬Tx))

• ◇((∃x)(Ax∧Tx))∧□((∀x)((Ax∧Tx)→Sx))

• ◇((∃x)((Ax∧Tx)∧¬Sx))

2. To test whether some set of propositions is mutually inconsistent, we just have see whether every pairing of them is contradictory:

Great!

Then to test whether they’re logically exhaustive, we just have to test whether it’s contradictory for them all to be false:

Again, great!

Yes, I have a logic exam soon, how could you tell? Is this a big part of the reason why I made this post? We may never know…

3. Also we should generally be weary of disjunctive arguments like these, as you can give plausible-sounding objections to almost all philosophical positions.