Monday, November 4, 2013

Error theory

Error theorists in ethics think that claims like "Murder is wrong" are all false. But this seems to me to be a self-undermining position. For if there were no true moral claims, our moral predicates would have no meaning. They would simply be nonsense.


Heath White said...

I think the consistent error theorist needs to say that "Murder is not wrong" is true. (For the very reason you point out.)

Alexander R Pruss said...

Even that's problematic.

Compare: "Quine is not mimsy." Or, equivalently: "Quine is not 982ifjf9." (Meaningless terms can be intersubstituted, no?)

Kolja Keller said...

I don't see the inconsistency.
The error theorist says the following three statements are false:
"Murder has the property of being wrong"
"Murder has the property of being not wrong"
"Murder has the property of being and not being wrong"

They endorse the negation of all three.
This is simply the result of claiming that the extension of all three of these properties is empty.

A good expressivist will instead want to say that these are not the propositions we mean when we say things like "Murder is wrong." They will want this to mean something like Mark Schroeder's semantics of being for blaming for murder.

But back to the original point: What makes them meaningless? They refer to properties with empty extension in this world. But this need not be the case in all worlds. Suppose you believe in divine command theory, but you are a polytheistic atheist (that is, you believe God's like the ones of Greek mythology to be perfectly possible. Moreover, what is right and wrong is just agreement among the gods on the matter. However, these supervene on the judgments of the gods at rah world, but this world happens to have no gods in it) Now moral predicates are perfectly meaningful, but applying any of them in this world is always false.

Alexander R Pruss said...


Nice to hear from you!

I think the better error theorist will say that murder is neither wrong nor permissible.

But anyway, that's not the issue. The issue is one of how to gain semantic access to alien properties. There are worlds containing fundamental physical properties that are irreducible to the physical properties of things in our world, and that do not figure in our world's laws of nature. Those fundamental physical properties are ones that we cannot gain semantic access to. Given Platonism (or a successful nominalist reduction of Platonism) we can quantify over such properties, but we cannot name any particular one.

Imagine aliens living in a world without electric charge and with laws of nature that are completely different from ours. How could those aliens gain semantic access to the property of being positively charged? After all, how would they distinguish it from the property of being negatively charged? We do so by ostension--say, we point to an atom and say that the stuff orbiting the nucleus is negatively charged. But the aliens can't ostend into our world.

Now, if one has a successful reduction of moral properties to non-moral ones, like the divine command one, then one can around this problem, much as you point out. But I think that all reductive theories of moral properties we have fall into one of three non-exclusive categories:
1. The theory is simply implausible. (E.g., divine command. Sorry if you like it!)
2. The theory together with obvious facts implies that some actions are wrong. (E.g., the utilitarian reduction of the right to the good.)
3. The reductive base will have the same semantic problems as "right" and "wrong" had given the kind of things that motivate error theory. (E.g., the reduction of deontic to virtue concepts. Reasons to be sceptical of the applicability of deontic concepts are--pace Anscombe--reasons to be sceptical of the applicability of virtue concepts.)

Kolja Keller said...

Thanks for the detailed reply. After reading through what I've written it seems that I mostly agree with you, but this helped me organize my thoughts on the matter some more.

Semantic access to alien properties is a good point, thanks. I think this needs to be clarified though. I think we can have semantic access to alien properties, we just can't have access to alien natural properties. We have access to the property "being a one cube mile block of uranium," even if it is impossible for anything in this world to have that property. Its natural alien properties like shmass and shmarge that we don't have access to. Thus, I think we would have semantic access to "approved by Zeus," even though nothing has this property.

But I think everything past this is just me agreeing with you and writing out why your option 1 doesn't help the error theorist.

This brings out that error theory really makes both a semantic and a metaphysical claim. The semantic claim needs to specify what our moral properties mean, and the metaphysical claim specifies that nothing (in this world?) has them.

Here is a worry though: Even if divine command theory is implausible, it seems perfectly possible that some people actually operate on it and really mean "God approves" when they say "right." Here is one way to argue that this doesn't help the error theorist: Error theory says that all attributions of moral properties are false. But the person that means "God approves" by "right" is not attributing a moral property because "God approves" isn't a moral property. Thus, people who use moral words to mean divine approval properties are not the kind of people that the error theorist talks about.

Maybe the error theorist could respond by saying that their theory is actually that all sentences in which moral words occur are false, even if the people uttering them don't mean moral properties by them. But this theory is false. Think of the flat footed biblicist that means by "x is wrong" simply "the Bible says x is wrong" without any claims about divine authorship or authority of the Bible. They may also believe this, but in their actual psychology and semantic moral properties just reduce to the text of the Bible. But in this case "murder is wrong" is true! The Bible does say that murder is wrong, so what the flatfooted biblicist means by "murder is wrong" is not false, so the revised error theory is false.

Now about points 2 and 3:
The error theorist would of course need to endorse that reductions to truly attributed properties fails for some reason or another. There is pain and pleasure in the world, and if morality reduced to it, error theory would be wrong. Thus, the error theorist must argue that when the utilitarian says "x is good" he does not mean anything moral by it, either.

It seems that the error theorist instead needs to say that there are sui generis natural primitive moral properties which don't apply to anything. The theorist cannot say that there are no moral properties or they would directly endorse that moral claims are meaningless, unless they had some theory how a predicate could have meaning without a corresponding property. Hence, an error theorist needs to be a moral realist, but then he faces the semantic access problem.

Which brings us back to option 1: Argue for the coherence of divine command theory and atheism. We have semantic access to properties like "God approves of x," and if the error theorist is successful in arguing that these can in fact constitute moral properties but that God, due to His nonexistence, doesn't approve of anything, the error theorist gets meaningful false moral claims.

Does that seem right?

Alexander R Pruss said...

I agree that some non-natural alien properties we can have access to. I used "fundamental" rather than "natural" in what I said, but the idea is similar or maybe the same.

I think your Zeus case, though, is problematic, because for Kripkean reasons I am sceptical that "Zeus" makes sense. Which of the many possible powerful supernatural beings does "Zeus" mean after all?

Here's one possible move for the error theorist. Go for a meta-level claim. Find a second-order "functional" property W such that it is fairly clear that if there were any wrongness-type property, it would have W. For instance, perhaps, a first-order property w has W iff an action's having w is always rationally a decisive consideration against performing the action. It is fairly clear that any wrongness-type property will have W. If one could then argue that no first-order property that has W is exemplified, that gives some sense to error theory.

And then the error theorist could take "Nothing is wrong" to simply be a pithy way of saying "Nothing instantiates an instantiator of W."

Compare this. We are all in some sense error theorists about unicorns. But "There are no unicorns" is in danger of being nonsense, since "unicorn" doesn't refer to any particular natural kind. Still, there is a second-order property U such that any unicornicity-type property has u. For instance, we could say that u has U iff u is a natural kind of equines that naturally have a single horn.

Then, while "There are no unicorns" may be nonsense taken literally, maybe we can take it as a pithy way of saying "No instantiator of U is instantiated."

There is also a supervaluationist move that might help the error theorist about morality or about unicorns. I wonder if you think that would work?