Thursday, September 08, 2005

Non-Cognitivism and Anti-Realism

Here's a question that's been bothering me for a while. (I seem to remember that someone - it could have been Gideon Rosen - raised something like it in the discussion following my talk at the last Arche Modality Workshop.)

Could one be a non-cognitivist about (say) ethics without thereby being an ethical anti-realist in the sense I favour, i.e. without believing that what it is for an ethical proposition to be true is for us to be some way?

Suppose you're a non-cognitivist about ethics because you think ethical discourse is expressive of our attitudes. Surely you'll therefore agree that what it is for murder to be wrong is for us to have a certain kind of attitude to murder?

One reason why you might not agree is that you might think that nothing is what it is for murder to be wrong. You might think that there is no way for the world to be that would correspond to murder being wrong, since ethical discourse does not correspond to states of affairs or facts in the way a cognitive discourse does. But if someone were to claim that what it is for murder to be wrong is for us to disapprove of it, we would take her as saying that there are ethical facts, it's just that they are facts about our attitudes.

But why should we take 'what it is for' (WIIF) talk as talk about states of affairs or facts, when we're dealing with a non-cognitive discourse?

Presumably one motivation for resisting the anti-realist WIIF claim supposed to be that when we say 'murder is wrong' we aren't asserting that we disapprove of murder. But to claim that what it is for murder to be wrong is for us to disapprove of murder is not to claim that when we say 'murder is wrong' we are asserting that we disapprove of murder. (Similarly, to say that mental states are brain states is not to say that when I say 'I'm happy' I'm asserting that I'm in brain state B. I may never have heard of brain state B.)

17 comments:

Andrew said...

I think the familiar kind of expressivist will (should?) say:

"What it is for "Murder is wrong" to be true is just for murder to be wrong. So the only good question I can hear you asking is: what is it for murder to be wrong?

If you intend "What is it for murder to be wrong?" as a request for something like an ontological reduction base, then there is no answer to the question, for talk about the wrongness of murder is not appropriately related to anything which can stand in ontological reduction relations. You're making something like a category mistake.

If you intend it to be closer to "Why is murder wrong?" then the answer is because it hurts people, is unjust, is non-conducive to peace, flourishing and happiness, etc.

Carrie Jenkins said...

I was thinking that the WIIF question might be heard in a way that is equivalent neither to a request for a reduction base nor to a why-question of the kind you mention.

It'd be something like asking what is essential to murder's being wrong. That needn't (at least it's not obvious to me why it must) be heard as a question about what is essential to *the fact* of murder's being wrong, or as otherwise betraying a category mistake.

Perhaps your familiar expressivist will think it's a category mistake because murder's being wrong isn't the kind of thing to which anything can be essential.

But I'd like to hear the argument for that claim, since I'd have found it fairly natural for him to say that something about human attitudes is what's essential to murder's being wrong.

Andrew said...

I think I can get some sense of what people mean when they talk about objects having essences (basically, a certain species of hyperintensional modal property or the individuating, constitutive properties that undergird such modal properties). And perhaps that notion extends in a natural way to ways objects are, even when no actual objects are that way. For example, maybe being spatio-temporally located is essential to being human (although I think I'd rather try to think of this as short-hand for conditional attribution of essential properties to objects: if x is human, then x is essentially spatio-temporally located.)

I can see that token events, like particular murders, might have essences. (Maybe neither exact location nor duration was essential to the Battle of Waterloo, but involving certain of the participant armies was.) And the essences of type events, like murders, might be treated a bit like the property case:

If e is a particular battle, then e essentially involves certain of e's participants.

But I'm getting a bit lost with the idea of murder's being wrong having an essence. Is it supposed be some kind of combination or function of e.g.

(a) x is a wrongful act iff x is essentially E1

(b) x is a particular murder iff x is essentially E2

so that the essence of murder's being wrong is e.g. the union of E1 and E2? Or is the idea closer to

(c) If x is a wrongful act & a murder iff x is essentially E3

where E3 isn't a simple function of E1 and E2?

(Obviously this latter case sounds like it contains a redundancy because it's arguably a priori that murders are wrong, so best just to think of 'murder' as picking out the relevant subvening class of killings).

If something like (c) is what is meant, I think the expressivist will hear that kind of conditional as some more sophisticated variant on e.g.

Boo! (Booing x & not holding x to be E3) and Boo! (Holding x to be E3 and not booing x)

So for example, the expressivist may express their contempt for those morally benighted people who accept that murder hurts others, isn't conducive to flourishing, etc, and yet don't judge it to be wrong, by pointing out that murder's wrongness just essentially consists in its being such as to hurt others, being non-conducive to flourishing, etc. I'm guessing they won't expect Tony Soprano to be much swayed by that consideration, but guess that's not the point.

Daniel Elstein said...

Carrie, the standard reason that for expressivists your kind of anti-realism is false is that the wrongness of acts doesn't counterfactually depend on the internal states which statements of wrongness express.

For instance, expressivists will hold that murder would be wrong *whatever* attitudes people had to it. Thus what it is for murder to be wrong cannot be for people to have certain attitudes to murder.

In fact, most expressivists would take your question to be the wrong way round: you want persuading that expressivism is compatible with realism (in your sense), but expressivists don't believe that expressivism is compatible with anti-realism (in your sense).

I agree with Andrew that (for expressivists) the wrongness of an action consists in what makes it wrong (e.g. its bad consequences).

Andrew said...

Having just re-read my last comment, there's a potential unclarity near the end. I didn't mean to suggest that the proposed expressivist account of the essence of murder's being wrong was contingent on something like (c) being what Carrie intended by 'the essence of murder's being wrong'. The same story will be given, obviously, if it's some kind of union of (a) and (b) that were intended. The point was just that I expect that the expressivist will deal with equivalences like

Act A is wrong iff ....

in a uniform way - e.g. employing their favoured expressivist construal of biconditionals involving moral terms - even when there's some attribution of essence on the RHS.

Carrie Jenkins said...

Andrew,

The idea of *murder's being wrong* (a state of affairs) having an essence is supposed to be different from the idea of any (real or possible) act of murder having an essential property. I want to talk about *what's essential to its being the case that murder is wrong*. Other ways to talk about this would include asking what it is for murder to be wrong, or what murder's being wrong really amounts to. Suppose we think that what murder's being wrong really amounts to is our booing murder. I don't see that that's going to be easily translatable into the kind of biconditionals you were describing (e.g. I think it's different from 'x is a murder iff x is essentially booed by us', or anything in that line.)


Daniel,

You write:
"expressivists will hold that murder would be wrong *whatever* attitudes people had to it. Thus what it is for murder to be wrong cannot be for people to have certain attitudes to murder."

They might - but in my 'Realism and Independence' paper I talk about ways of understanding the counterfactual which seem to make it compatible with the view I call anti-realism (basically, these are ways where the possible worlds under consideration are evaluated *by us with our actual attitudes*).

You also say:
"you want persuading that expressivism is compatible with realism (in your sense)"
I didn't mean to ask for that - my thought was that expressivists probably should come out as anti-realists, but I was exploring a way they might try to argue that they are not. (They wouldn't be realists either, if this sort of argument works - they'll think the realism/anti-realism question is the wrong sort of question to ask here.)

Andrew said...

I can think of two good prospective candidates for the essence of a state of affairs.

Firstly, that it is essential to a state of affairs that it be composed, in the relevant sense, of the very objects and properties that it is. Since expressivists don't think that wrongness is a property in the relevant sense, they won't think it is a component of a state of affairs in the relevant sense. (They might give an expressivist account of what state-of-affairs talk comes to when discussing wrongness, and an account of what essence comes to when discussing expressivist states of affairs, but these will involve expressing attitudes, and not citing them. For example

The state of affairs of murder's being wrong obtains iff murder is wrong

with the standard treatment of the expressivist biconditional, as before.)

Secondly, that it is essential to a state of affairs that it ground or necessitate the truth of the fact-stating proposition(s) that describe(s) it - for example, it being essential to the state of affairs of murder's being wrong that it is the ontological ground of the truth of the proposition that represents murder as being wrong.

For the expressivist, ascribing truth to the proposition that murder is wrong is just the same as expressing a complex of attitudes towards murder. Ditto for ascribing falsity. There isn't any proposition that describes a worldly state of affairs. Moral propositions aren't true because of the way the world is.

I'm worried that I'm missing your point, because I don't see any space for the kind of distinction you seem to want to make. Is your line of thought that we can argue as follows?:

(1) The state of affairs of murder's being wrong (what murder's being wrong really amounts to) = whatever in the world grounds the truth of the proposition that murder is wrong.

(2) The things in the world that ground the truth of murder's being wrong = a certain complex of attitudes

So

(3) The state of affairs of murder's being wrong (what murder's being wrong really amounts to) = a certain complex of attitudes.

And given some premise about it's being essential to a set of attitudes that they are a set of attitudes we would have

(4) The state of affairs of murder's being wrong is a essentially related to a certain complex of attitudes.

If so, I think the expressivist is going to just say that there's an equivocation going on. In the normal, metaphysical sense of ontological grounding - the sense in which we might say that the truth of propositions about water, or the property of being water, are typically grounded in H20, and not Paris, or the world as a whole, or XYZ, propositions about wrongness - and wrongness itself - aren't grounded in anything, any more than "ouch!" is grounded in C-fibre stimulation. They're going to say that it's simply a mistake to think that every proposition is grounded in this sense. If you have a really clear idea of what this kind of grounding amounts to, that entails that e.g. every declarative sentence is grounded in this sense, then I'd be interested to hear it. But prima facie it sounds very plausible to me to claim that only genuinely fact stating sentences/propositions are grounded in the facts, and only genuine, worldly properties are grounded in the world.

Maybe this is what you were looking for an argument for, rather than a plausibility claim. but it's very difficult to mount a relevantly strong argument when the sense of grounding that you're appealing to is already so unclear, and the relevant notions of e.g. property, proposition, state of affairs, etc, aren't set out metaphysically. My suspicion is that when you come to say what a state of affairs is, or what the relata of the grounding relation are, etc, the expressivist will be able to point out to you where she gets off the boat with respect to (1)-(3).

Of course, the expressivist will accept that the truth of propositions about murder are grounded in non-cognitive attitudes, if all that amounts to is that we appeal to non-cognitive attitudes when outlining the non-fact-stating semantics for moral discourse. But that looks like a very different sense of grounding than in the standard case above. Why, in general, should we expect this kind of semantic grounding to entail ontological grounding? It doesn't in e.g. the case of the possible worlds fictionalist, who appeals to world-concepts when setting out her semantics while not taking modal talk to be ontologically grounded in worlds. Ditto in the case of e.g. the strong eliminativist, who takes moral or modal talk to be necessarily false in virtue of its appeal to conceptually incoherent truth-conditions.

Say we follow Fine, and take a notion of 'what is real' as primitive. Then I think that what the basic expressivist claim is is that what murder's being wrong really amounts to in that sense is: nothing, and that ascriptions of wrongness to murder are none the worse for it.

Carrie Jenkins said...

I should have been clearer when I said murder's being wrong was a 'state of affairs': I sometimes (carelessly) use that expression in a way that isn't meant to be ontologically committing.

It is of course a fair question what the essence of murder's being wrong amounts to if murder's being wrong isn't a state of affairs in the sense of being a piece of the world composed of objects and properties. I think this is where the really interesting issue is.

Maybe it would help to get a notion of the relevant kind of grounding off the (ahem) ground if we thought about conditions for appropriately uttering 'murder is wrong'. (NB: I'm not proposing a *reduction* of essence to conditions for appropriate utterance - though that might be an option - just that this might point in the right direction.) In the case of genuinely fact-stating sentences we can give conditions for correct assertion in ways which are obviously of the essence or in other ways :

'7+5=12' is assertable iff 7+5=12
vs
'7+5=12' is assertable iff all bachelors are unmarried.

Maybe the fact that it seems to be of the essence of the correctness of asserting 'murder is wrong' that our attitudes be a certain way similarly flags that what's essential to murder's being wrong is something to do with our attitudes.

This doesn't buy us a metaphysical story - but then the expressivists don't think there's a metaphysical story there to be had. Hmmm. I'll keep thinking about whether more can usefully be said. This discussion's been very useful, in any case.

Andrew said...

Won't you get:

It is of the essence of the correctness of asserting "murder is wrong" that

"Murder is wrong" is assertable iff murder is wrong

which is equivalent to some more sophisticated version of:

Boo!(Recognising the assertability of "murder is wrong" and not Boo!ing murder) and Boo!(Boo!ing murder and not recognising the assertibility of "murder is wrong")

This again seems to express our attitudes, but not to cite them. But the "what the wrongness of murder essentially is" or "what the wrongness of murder really amounts to" claims originally mooted seemed to cite attitudes, not express them.

Okay, I'm done now, I promise!

Carrie Jenkins said...

But if you press for something informative, the expressivist ought to give you the thing I suggested. I think cases where there's really nothing informative available - where there's only the disquotational thing - are cases where we should be realists about the subject matter. (To tie this back to the arithmetical example, a certain kind of mathematical anti-realist can say it's of the essence of the correctness of asserting '7+5=12' that '7+5=12' is provable. She'll think that because she'll think it's of the essence of its correct assertability that 7+5=12, and she'll think that what it is for 7+5 to be 12 is for '7+5=12' to be provable.)

Andrew said...

I can't see why we should think that the disquotational case entails realism. But even if we granted that,

(1) the account needn't be disquotational - e.g. we might have

"Murder is wrong" is assertable iff S is wrong

where S is the subvening base of the class of killings we're concerned with, and where the biconditional is treated expressively as above. So a large class of moral claims can be treated non-disquotationally. Of course, you might argue that that leaves an important remainder. But I guess the expressivist will demand an argument as to why treating the 'base' claims disquotationally commits her to realism. (She might say, plausibly enough, that the expressivist semantics already display the relationship between the correct use of "__is wrong" and our attitudes which constitutive of anti-realism. This relation isn't essential dependence of wrongness on our minds, since her position is precisely that there is no property of wrongness to be the relata of such a relation).

(2) Isn't the route through the 'essence of correct assertability' just a detour, now? The real work is being done by the claim that what it is for murder to be wrong is for us to have certain attitudes. But that's what my expressivist is denying.

In the maths case, it looks like we have robust truth-conditions:

"7+5=12" is true iff "7+5=12" is provable

That's why e.g. adding the disquotational schema

"7+5=12" is true iff 7+5=12

allows us to derive

7+5=12 iff "7+5=12" is provable

which your mathematical anti-realist then reads as a left-to-right ontological reduction. But the expressivist is precisely denying that moral claims have robust truth-conditions of this sort - they think that the semantics of moral claims aren't given by associating them with worldly truth-conditions, but by displaying their expressive role. So the two cases look crucially disanalogous to me.

Daniel Elstein said...

Carrie, I read your 'Realism and Independence' paper. If I've got the distinction between 'what it takes' and 'what it is' right (and I accept it arguendo), then the quasi-realist should say that there's nothing which counts as what it *is* for lying to be wrong. That would be for there to be some state of affairs describable in non-moral terms that was *conceptually* equivalent to a moral state of affairs. In that case expressivism would be equivalent to some sophisticated form of subjectivism; but in fact expressivists think that subjectivists are quite wrong. Hence the expressivist emphasis on Moore's open question argument.

So the reason I think that your realism/anti-realism distinction is unacceptable to the quasi-realist is that it misses out the option that the quasi-realist wants to take. For you, realism about wrongness involves thinking that what it *is* to be wrong is something non-mental, whereas anti-realists think that it's something mental. But quasi-realists (and other expressivists) think that there's nothing that it *is* to be wrong, though they agree with realists that what it *takes* to be wrong is something non-mental.

You say in your article that 'the external, meta-ethical, question on which the realist and the quasi-realist disagree' is 'the question of whether or not something’s being wrong (meeting the standards for being accounted wrong) is a matter of our taking a negative attitude towards it' (p 20); but that's the difference between objectivists and subjectivists, not realists and quasi-realists.

Neil Sinclair said...

Hi Carrie, Daniel et al.

Having read your paper, I think I agree with Daniel on this one.

The expressivist will of course claim that what it is for us to accept an ethical proposition as true is for us to be some (attitudinal) way. And he will accept, with the realist, that what it takes for something to be wrong is some (usually non-mental) feature of the object of evaluation. But there is, for the expressivist, no further questions about whether what it is to be wrong is something mental or not. Thus the expressivist will not be an anti-realist in your sense.

So I think the expressivist must disagree with the claim that: "…what it is for murder to be wrong is for us to have a certain kind of attitude to murder". The expressivist will only hear this issue as immanent, that is, as asking what it takes for something to be wrong, to which he will give an objectivist (mind-independent) answer. There is then no further question of what it is for something to be wrong, only the question of what it is to think that something is wrong - to which he gives his distinctive answer.

But the expressivist will be an anti-realist in a negative sense – since he will fail to believe that what it is to be wrong depends on us. He will fail to believe this by ducking the 'external' question entirely. To paraphrase something an expressivist said once: "Questions of mind-dependence are moral [immanent] questions or nothing". (No prizes for guessing who the expressivist is).

There is a further question of whether the expressivist is an anti-realist in a positive sense, namely in the sense that anti-realists claim that what it is for something to be wrong is something mental. Again, since he can only take this issue as an immanent one, and since his immanent answer is objectivist, he will not be an anti-realist in this sense.

But then anti-realism and realism in these postive senses are not exhaustive, since they fail to consider the position of someone who ducks the external question entirely (Daniel’s point). I guess how much of a problem that is depends on what the distinction was intended to do. But I don’t think it should worry the expressivist much. He cannot give an account of what it is for something to be wrong in the same way that you give an account of what it is for something to be funny (viz. to make us laugh). But then it is not clear that such accounts (and the distinction they engender) are available for every concept.

Neil Sinclair said...

Sorry: Error in the previous post. The first line of the forth paragraph should read: "But the expressivist will be an anti-realist in a negative sense – since he will fail to believe that what it is to be wrong depends on something other than us"

Carrie Jenkins said...

Hi All,

This has been really helpful - sorry I've been too busy to respond properly to all the comments. So it looks like the majority view is that expressivists should take the line I mentioned in my original post, that '*nothing* is what it is for murder to be wrong'. I can see I'm going to need to write another paper about this ...!

Carrie Jenkins said...

My final comment on this: the main point of my paper was to argue that quasi-realists wouldn't be able to imitate realists when it came to saying what it is for murder to be wrong. If the consensus here is right, and the quasi-realist in question (we all know who we mean) is also an expressivist, then he can't do that. Where the realist says that what it is for murder to be wrong is nothing to do with us, the quasi-realist has to say that there's *no such thing* as what it is for murder to be wrong, which is just another way of betraying his non-realism.

Daniel Elstein said...

Carrie, I guess whether the quasi-realist 'betrays his non-realism' in a worrying way (for him) depends on whether the question of what it is for murder to be wrong is an ethical question or not (i.e. an internal or external question). As I understand it, the quasi-realist project involves mimicking the realist only as far as the internal questions go; otherwise quasi-realism collapses into realism. But we seem to agree that the question here is an external one. So the quasi-realist didn't *want* to imitate the realist on the question of what it *is* for murder to be wrong. It's on the 'what it takes' question that the quasi-realist wants to imitate the realist; and he can! Of course, it may be that quasi-realists *said* that they agreed with realists about what it is for murder to be wrong, but this was presumably because they didn't have before them your distinction between what it is and what it takes, and thought they were talking about what you call 'what it takes'.