Here is a worry for (pure) expressivism. I think it is new. I am not sure how cogent it is, and this is
probably not its most elegant formulation.
So help is welcome. Consider
(1)
No ought-claim is true.
Construe ‘ought-claim’ to include (1) itself. Then (1) is necessarily false: if it is true, then it says of itself that it
is false. (You might think (1) is a
pretty good statement of error theory; I will leave that as a problem for the
error theorist.) Consider now
(2)
No ought-claim ought to be accepted.
This is contingent.
It might be, for example, that one ought to do what has the best
consequences and the best consequences happen when people never think about
ought-claims. If someone finds
themselves accepting (2), of course, the first thing they ought to do is figure
out how to abandon it. That doesn’t
affect its truth however.
Suppose, however, that you are a pure expressivist (like
Gibbard) who thinks of ‘ought’ as planning language. How do you construe (1) and (2)? Here is my best guess. You think of (1) as quantifying over plans
and rejecting all of them. So you can
sincerely assert (1) if you have no plans. You think of (2) as itself a piece of
planning: the plan to have no
plans. Of course, this plan immediately
fails in a self-undermining way. Still,
you could have this plan, if you are a little confused. You can sincerely assert (2), then, if you
have the plan to have no plans.
Now the worry is this.
If my glosses are right, the expressivist can distinguish between the
contents of (1) and (2). The problem is
that the contents seem wrong. (1) is
self-refuting. But the expressivist
gloss I suggested is not self-refuting; it is a simple piece of quantification
which doesn’t seem to quantify over itself in any way. If (1) is true then (1) is false, but I don’t
see how this entailment survives under the expressivist interpretation. On the other hand (2) is not
self-refuting. It might or might not be
true. (Granted, it could not be
rationally accepted.)
But the expressivist gloss I suggested for (2) is self-refuting: a plan that cannot possibly succeed. My expressivist glosses for (1) and (2) seem
reversed in their self-refuting properties.
Maybe I should just reverse the interpretations, then? I don’t think this can be right. Consider
(3)
“I ought to take out the garbage” is true.
(4)
“I ought to take out the garbage” ought to be
accepted.
The friendliest gloss on (3) is just having the plan to take
out the garbage. The best gloss I can
come up with for (4) is having the plan to have the plan to take out the
garbage. (I don’t necessarily have the
plan to take out the garbage, yet, if I accept (4).) If those glosses are right, then “is true”
applied to ought-claims is just a matter of accepting or rejecting plans, while
“ought to be accepted” applied to ought-claims is a matter of planning to
accept plans. That supports the original
reading of (1) and (2).
But that reading seems incorrect. If it is incorrect, then pure expressivism
has the wrong semantics for (1) and (2)…unless I’m mistaken. Thoughts?
Hi Heath,
Can I ask a clarificatory question first? I am having a little trouble understanding what (1) says, especially given your instruction that (1) is supposed to include itself.
Some ways of construing (1):
1*: No sentence in which the word “ought” appears is true.
2*: No proposition expressed by a sentence in which the word “ought” appears is true.
3*: All propositions in which it is either claimed that something ought to be the case or it is claimed that it is not the case that something ought to be the case are false.
I don’t see how 1* or 2* are self-refuting. Maybe 3* is? Even that is not clear to me. Could you say a little more about how you are understanding (1)?
Seems non-problematic. Couldn’t a Gibbardian expressivist think of (2) as expressing a plan not to have plans (i.e, as plan not to plan)? This plan seems like a plan that can never succeed as long as you have that plan itself. This self-underminingness seems to capture the self-defeatingness of (2) pretty well.
Should add that it is not enough for self-refutingness that you have a plan that cannot succeed. I think self-refutingness would require that there is something contradictory within the scope of the planning attitude. For instance, if you had a plan to (p and not-p), this would be self-refuting. But this doesn’t seem to be the case when you plan not to have plans. The kind of self-underminingn that you get there seems to match the idea that (2) is something that you cannot rationally accept.
Kris,
I see your difficulty. I would think 1* and 2* are self-refuting unless we think “ought” (with quotes) is a name of the word ‘ought’, not incorporating that word itself. (Maybe we could get around this with a fancy definite description.) I had something like this in mind. (I didn’t want to help myself to the notion of ‘proposition’ given that I was heading into expressivism.)
I would think 3* is not self-refuting, as it makes a claim about what is false, not about what ought/not to be the case.
Jussi,
I agree that the expressivist ought to construe (2) as the plan to have no plans. The point was that the plan to have no plans is self-refuting in a way that the proposition “No ought claim ought to be accepted” is not. Granted one cannot rationally accept it; the point remains, it is only contingently false. (I assume it is false.) Whereas the plan to have no plans necessarily fails.
Heath,
I think I might be repeating here, but I still wonder if we could distinguish different kinds of self-refutingness to solve the problem. A claim is self-refuting in a robust sense when it is self-contradictory. Hence, logically speaking it could not be false.
The expressivist could argue that such claims express plans with self-contradictory content – plan (to phi and not to phi). And, we agree that (2) is not self-contradictory in this sense.
Thus, (2) should express a plan which has a non-contradictory content. This is true – not having plans is something that could be the case.
(2) should also express a plan which cannot be rationally accepted. This also seems right on the expressivist reading given that having the plan undermines its own aim. So, the expressivist could say that the plan is self-refuting but deny that this means amounts to self-contradictoriness instead of more practical self-defeatingness.
Like Kris, I’m not sure what (1) is supposed to mean.
Let me offer a suggestion. In Gibbard-speak, it’s
(1G) No plan-laden claim is true.
There is a problem, though. It’s very unclear whether (1G) is itself plan-laden. With what plan is it laden?
Here is a way around that problem. I think both of these turn out to be plan-laden:
If so, then (1G) entails that neither of them is true. But they are contradictories, so (1G) entails that each of a pair of contradictories is untrue. That’s a reductio on (1G). And as far as I can see, it works just fine in Gibbard’s semantics and logic.
Is that good enough for you, Heath? Or did you want something other than a reductio as a pathology for your (1) to suffer from?
Shouldn’t Gibbard say only that (2) is the plan to accept no ought-claims, and (4) the plan to accept “I ought to take out the garbage”?
To interpret these claims further as you suggest seems as though it might overlook the use/meaning distinction.
Compare:
(5) I ought to say, “I ought to take out the garbage”
Which is not equivalent, in Gibbard’s terms, to:
(5′) I plan to say that I plan to take out the garbage.
I am becoming convinced that there is less here than I initially thought. I wanted (1) to quantify over itself but I think it doesn’t, on ordinary construals. And I think Jussi has the right take on (2). That removes the difficulties I thought were there.
Ah well. Thanks for the help!
Heath (cc. Jamie),
Given your last comment, I may be beating a dead horse, but I still think there’s something important here, at least with respect to (1):
“Ought-claim” can be understood either to refer to all claims involving ‘ought’ or to claims that entail the existence of some ought (that someone really ought to do something). I will refer to the latter as “substantive” ought-claims. If (1) is about all ought-claims, it is self-defeating; if about only substantive, it is not. This is why (1) is not really a problem for the error-theorist: Error theory is a view about substantive ought-claims and so the error-theoretic reading of (1) is not self-defeating. Consider: “One ought not murder” vs. “It is not the case that one ought not murder.” Many error theorists would, I think, deny the former but accept the latter, because they are, respectively, substantive and non-substantive.
Your suggestion seems to be that because the expressivist will understand (1) as the rejection of plans, he will be unable to make sense of (1)’s self-defeating nature (on one reading). Jamie’s contention is that this is not necessarily true; the expressivist can, instead, claim that both plans and their rejections count as “plan-laden.” Thus, Jamie suggests that we read both (P1) and (P2) as plan-laden. So, the problem is not that the expressivist cannot make (1) self-defeating; Jamie has demonstrated that he can.
Now, I know very little about the details of “plan” talk, so I apologize if I’m missing something obvious and ask you to bear with me (again, I’m afraid, as this is similar to what I was arguing here). The question, as Jamie said, is whether (1G) is plan-laden. This, I take it, is the expressivist analogue of the question about how to read (1). Asking whether (1G) is plan-laden is like asking whether (1) is an ought-claim. It seems to me that the expressivist may answer either way, but in either answer he loses something important. If (1G) is not plan-laden (which seems to be your initial reading), then the expressivist runs into your objection; he cannot express what we mean by a self-defeating (1). Alternatively, he may render (1G) self-defeating by reading it as plan-laden, as per Jamie’s suggestion. He thereby avoids your objection. But doesn’t this make it impossible to even express the other, error-theoretic reading of (1), that no oughts exist? Now, of course the expressivist isn’t worried about things in those terms, but shouldn’t the expressivist at least allow for the possibility that the error theorist be able to state (P2)—just as I suggested she would do with the contradictory of “One ought not murder”—without its being plan-laden? Doesn’t Jamie’s solution make certain kinds of error-theoretic or skeptical or nihilistic expressions impossible? These views may be wrong, but surely they are not wrong because they can’t even be expressed.
David,
I can think of two ways an expressivist might respond to your worry (though I’m not sure if either is very good). First, he might read (1G) as plan laden, but propose “there aren’t any plans” as expressing error theory. (Admittedly error theory then turns out to be quite a silly view on expressivism, but it can be expressed.)
Second, he might bite-the-bullet, acknowledge that an expressivist cannot without contradiction assert that “no plan-laden claim is true,” but show that the expressivist can capture the “flavor” of error theory. Call a plan about which plans to have a “second order plan,” and a plan the subject of which is which actions to perform, or states of affairs to bring about, etc. a “first order plan.” It seems possible to have a second order plan not to have any first order plans. The expressivist might then say that someone who has this second order plan is an error theorist, and he can express his error theory by saying “no first-order plan-laden claim is true.” This captures the flavor of error theory insofar as the distinction between first-order plans and second-order plans seems to map onto the distinction between judgments of practical reason and (a certain subset of) judgments of theoretical reason (those that concern claims of practical reason). Because error theory is a theoretical position about practical judgments (albeit one that has practical implications), expressivism can accomodate it just as well as cognitivism.
Angus,
Consider an error theorist. She says, “No substantive ought-claim is true.” The expressivist replies, “No, you can’t say that, because all ought-claims are plan-laden, and so what you just said is self-refuting, since it is itself an ought-claim. You must mean that there aren’t any plans.” “But,” replies the error theorist, “I don’t believe that.”
Now what are we to say? Unless “there aren’t any plans” is understood in terms of some non-planning non-cognitive attitude, it is, presumably, a proposition that one may or may not believe. So what’s going on in the error theorist’s head? The expressivist saddles many of us (cognitivists) with false beliefs about what our ought-claims are really expressing, and thus about what we believe. Now we have someone—the error theorist—who has made a claim she took to be entirely normatively sterile, but the expressivist must maintain that her claim is actually plan-laden and therefore self-refuting, and that the only way to make sense of her is to attribute to her a belief she takes herself not to have. She may not only take herself not to have this belief, but not even understand it, given that she may know nothing about expressivism, and thus not even understand how plans could be relevant to what she’s talking about! This way of understanding the error theorist seems wholly implausible to me.
As to second-order plans: I worry that your suggestion doesn’t really capture the “flavor” of error theory as well as it might seem. What you suggest seems to me the expressivist analogue to the cognitivist mistake of thinking that all error theorists are committed (incoherently) to the belief that we ought not to countenance talk of oughts. But, of course, a true error theorist is committed to nothing of the sort; she would reject that ought like any other. Similarly, I can’t see that she would be comfortable being told that she just plans not to plan. She does not mean to reject planning by planning not to plan; she means to reject it in a way that is entirely external to planning altogether. It seems to me quite easy to see, in cognitivist terms, what this “external rejection” means, what the error theorist is after. But I don’t see how the expressivist can make sense of it.
I think you’re right, David, that according to expressivism, error theorists don’t have anything coherent to say. But that might be a feature, not a bug. Think of Hare in “Nothing Matters” (he says he can’t figure out what the issue is supposed to be between someone who believes there are values woven into the fabric of the universe and someone who believes there aren’t).
Jamie,
There’s incoherent and there’s incoherent. There’s a difference between talking about my friend’s bachelor husband and talking about my downtown char Tony-wise. The trouble with expressivism thus understood, it seems to me, is that it can make no sense of error theory; error theory is incoherent in the latter manner above. Consider someone who isn’t an error theorist, but in a moment of reflection wonders whether there’s anything she ought to do at all. The expressivist seems to have no way to capture what she takes herself to be wondering. He can try with the (self-refuting) plan-laden (1G) but, as I’ve tried to argue, I don’t think that really captures what’s going on inside her head. But isn’t such wondering a real part of moral reflection? Shouldn’t any good theory of moral language be able to make sense of it, even if just to dismiss it as obviously and necessarily false?
David,
I don’t think that’s right. When an ordinary person wonders whether there’s anything she ought to do at all, it’s like this: if she suddenly remembers she said she’d pick up her husband after bowling, then her question is answered.
When an error theorist says that there is nothing whatsoever that one ought to do, under any circumstances (and he has to say more, too, because he isn’t just saying that every action is permissible), he’s using the language in a pretty non-standard way. I think an expressivist might be entitled to demand that he say what he means – and then it might be just fine for the expressivist to answer, “Okay, I understand, but you didn’t say that with your original sentences.”
Suppose I am an error theorist about mathematical language. A formalist has an ingenious translation manual that she uses to take sentences apparently quantifying over numbers and assign to them propositions with no such quantification. So I say, “There are no prime numbers greater than 6, because there are no prime numbers, because there are no numbers at all.” The formalist can’t translate this into anything that could be what I meant. But she doesn’t care. She thinks I’m just confused when I try express skepticism about numbers in this way.
Jamie,
“When an ordinary person wonders whether there’s anything she ought to do at all, it’s like this: if she suddenly remembers she said she’d pick up her husband after bowling, then her question is answered.”
The woman might not take what she said to her husband as an answer to her worry; it seems she makes no error in continuing to wonder whether there is anything she ought do (surely her saying she would do something does not entail that she ought to). I’m not sure what to say beyond that I still think it is important that moral language be able to capture that possible moment of thorough-going skepticism between the thought of one’s promise (or whatever) and the acknowledgment of some (or any) ought.
“[The error theorist] has to say more, too, because he isn’t just saying that every action is permissible.”
I have always taken it that this is what the error theorist maintains. Permissible is the contradictory of ‘impermissible’. So, if nothing is impermissible (or obligatory) then everything is (merely) permissible. What am I missing?
“The formalist can’t translate this into anything that could be what I meant. But she doesn’t care. She thinks I’m just confused when I try express skepticism about numbers in this way.”
Let me preface by saying that I know next to nothing about philosophy of mathematics, and so what I say here may be completely wrong. That being said, I’ll just try to state my worry:
It seems to me that there is a subtle, but fundamental difference between expressivism and formalism. Expressivism is a theory about the content of moral language; formalism is a theory about the truth-conditions for mathematical language. When I say “Two plus two equals four,” the formalist maintains that what makes my claim true is some set of deductive rules within the formal system of mathematics (or something like that—again I’m completely out of my element here). But the formalist need not deny that I am expressing my belief that the abstract entity ‘two’, combined with itself, produces the abstract entity ‘four’. So, while the formalist would be unable to translate error theory about numbers into his formal system, he could still make sense of the error theorists’ position on the error theorist’s own terms. It seems to me that because expressivism is a view about the content of people’s claims, it does not have this same luxury.
David,
That’s what I think, too. But the error theorists that I know insist that their error theory is not a substantive normative view. The view that it is permissible to eat the flesh of animals, say, is a substantive normative view. Mackie, e.g., went to great lengths to explain that his error theory has no normative implications. He denied the existence of permissibility as well as the existence of obligatoriness.
Now I must be missing something.
When you assert the sentence, “Two plus two equals four” the formalist says that you are expressing a belief with an existential commitment to a platonic object, which belief he thinks is false. And he has given truth conditions for the sentence, and it turns out, to nobody’s surprise, that the sentence is true.
So, a formalist has to say that by sincerely asserting a true sentence you are expressing your false belief? I hope not. That would be very bad news for formalism.
Jamie,
I think that ‘permissible’ can either be the contradictory of ‘impermissible’ or be substantively normative, but not both. If ‘permissible’ and ‘impermissible’ are contradictories, then both’s being substantively normative would mean that there are, necessarily, true substantive normative claims. This would seem clearly to beg the question. Since I take their being contradictories as primary, I take claims about permissibility not to be substantively normative. Now, you might say that this is just a terminological issue—that nothing rests on which way we choose to use ‘permissible’. I think, though, that there are good reasons for taking ‘permissible’ to be the same as ‘not impermissible’. If we don’t do this, and we take permissibility as substantively normative, we must say what this substantive content is. The idea seems to be something like: if something is permissible, then we ought, all things considered, to permit it. (Of course, maybe you take the content to be something else entirely, in which case the rest of this argument won’t follow.) Now consider a case in which something is morally permissible but, for non-moral reasons, shouldn’t be permitted. Do we have to maintain, still, that even though all things considered the action shouldn’t be permitted, we still have some moral reason to permit it? This just sounds implausible to me. That speaks in favor of treating ‘morally permissible’ as just ‘not morally impermissible’. For the sake of consistency, we should treat ‘permissible’ (full-stop) the same way. So, ultimately, I think that the error theorist’s position can be both (a) there are no true substantive normative claims and (b) everything is permissible. (I am indebted to Christian Coons for some aspects of this argument.)
As to formalism: I should have said that the Platonic realist has two beliefs: one with the content ‘two plus two equals four’ and one about the truth-conditions for that content, namely some connection to abstract entities called “numbers.” The formalist will say that the first belief is true while the second is false. So, what of error theory? The formalist will think the position absurd, because for him, numbers are just part of a formal system, and of course that exists. But this doesn’t prevent the formalist from making sense of what’s going on in the realist’s head. According to the formalist, the realist has a false belief about the truth-conditions for claims about numbers (Platonic realism) and that false belief has lead him to another belief—one that the formalist might himself agree follows from Platonic realism—namely mathematical error theory. The formalist doesn’t need to be able to express the error theory in his formal system because he acknowledges that it is a view that only makes sense given assumptions he denies. But none of this prevents him from understanding those assumptions, or positions that follow from them.
Now, contrast this with the expressivist. I say, “Murder is wrong.” The expressivist doesn’t merely have a view about what makes that true or false (or whatever, if he doesn’t talk about truth). He has a view about the content of my expression: I am expressing a plan that I have. So far, so good. Now I say, “There are no true normative claims.” He has no choice but to interpret this within his expressivist framework; he can’t just take it on my terms and call it false the way the formalist can for the realist in mathematics. But, of course, the suggestion has been that he can’t make sense of the error theorist within the expressivist framework, so it seems he fails in his interpretation of the error theorist in a way the formalist does not. Does that make sense?
When you say ‘Murder is wrong’, the expressivist agrees (because he also plans to feel guilty should he murder). When you say ‘Eleven is a prime number’, the formalist agrees. When you say ‘There are no true normative claims’, the expressivist doesn’t agree because neither he nor anybody can really have ruled out every plan for every situation. And when you say ‘There are no numbers’, the formalist disagrees because he can easily find a counterexample! But the expressivist can understand what Mackie is thinking (I think – Hare claimed not to be able to understand it, but that was in the way that philosophers often say they don’t understand someone else’s metaphysical position), and the formalist can understand (with the same caveat) what a mathematical eliminativist is thinking: each is thinking that ordinary talk commits one to weird objects that don’t, in fact, exist.
So, in short, I don’t see the contrast.
On the interpretation of expressivism we have been considering, any statement containing normative language (e.g., “There are no true substantive normative claims”) is plan-laden. Assuming that any statement of error theory will, similarly, use normative language (I don’t see how it could fail to), it will be impossible to have an error-theoretic claim that doesn’t turn out, for the expressivist, to be plan-laden. Ultimately, it seems there is no way for the expressivist to understand someone as saying that “ordinary [normative] talk commits one to weird objects that don’t, in fact, exist” because any statement taken by its speaker to express this idea will instead be understood in terms of plans by the expressivist. Since formalism is not a view about how to interpret mathematical claims but only a view about what makes them true or false, it does not have this problem.
I thought we went through this already. That’s why I said this above:
So, a formalist has to say that by sincerely asserting a true sentence you are expressing your false belief? I hope not. That would be very bad news for formalism.
I don’t get this:
Huh. What about if the person says “Ordinary normative talk commits one to weird objects that don’t, in fact, exist”? Why wouldn’t an expressivist understand that sentence in exactly the way that you do?
“So, a formalist has to say that by sincerely asserting a true sentence you are expressing your false belief? I hope not. That would be very bad news for formalism.”
This is why I distinguished between the first-order mathematical belief and the second-order belief about the truth-conditions for that first-order belief. Given that, I think I can maintain that “formalism is not a view about how to interpret mathematical claims but only a view about what makes them true or false” without running into your objection. The formalist need say nothing about how to interpret (first-order) mathematical claims.
“What about if the person says “Ordinary normative talk commits one to weird objects that don’t, in fact, exist”? Why wouldn’t an expressivist understand that sentence in exactly the way that you do?”
“Ordinary normative talk,” seems to refer to a set of claims, all of which the expressivist understands in terms of plans (“ought-claims” refers either to this same set or to some subset of it). It seems, further, that if claims containing “ought-claim” are themselves ought-claims, then claims containing “ordinary normative talk” are themselves ordinary normative talk. I don’t see how the expressivist could plausibly maintain that “Ought-claims commit one to weird objects that don’t, in fact, exist” is plan-laden while “Ordinary normative talk commits one to weird objects that don’t, in fact, exist” is not.
It looks to me like the expressivist is in exactly the same position with respect to
Ought-claims commit one to weird objects that don’t, in fact, exist.
as the formalist is in with respect to
Mathematical claims commit one to weird object that don’t, in fact, exist.
What is the contrast supposed to be?
There is often a problem about how to distinguish first-order claims from second-order claims. But whatever problem there is seems to me to show up just the same in both (mathematical and normative) discourses.
Ah ha! I think I found the problem: I misconstrued (sorry!) part of your original solution to Heath’s objection. I thought that you agreed that “No ought-claim is true” was going to have to turn out to be plan-laden in order for the expressivist to render it self-defeating. Given that, I took it that any claim about “ought-claims” would likewise turn out plan-laden, and thus the expressivist would be unable to hear the error theorist as expressing a belief when she says “Ought-claims commit one to weird objects that don’t, in fact, exist,” because that, too, would turn out plan-laden.
Your solution, rather, I take it, is that “No ought-claim is true” is not plan-laden, but is nevertheless self-defeating because it entails the truth of contradictories. Now, as you might guess given what I said about permissibility, I don’t think we can say that two contradictories are both plan-laden, and so I find that solution unsatisfactory. But as it stands, given your solution, I see that you can maintain that the formalist and expressivist can interpret the error theorist equally well.