According to some (but not all) ‘hybrid’ metaethical theories, moral sentences like ‘stealing is wrong’ express both beliefs and desires, but different beliefs for different speakers. I think Paul Edwards was a forebear of this position, but it has recently been defended by Stephen Barker and Michael Ridge.
I understand these kinds of views to work something as follows: every speaker is assumed to have some property, P, such that she disapproves of P-actions. Then, for any given speaker, S, who disapproves of P-actions, ‘Stealing is wrong’ expresses the belief that stealing is P, and expresses disapproval of P-actions.
For example, if Max disapproves of actions which fail to maximize happiness, then ‘stealing is wrong’, when spoken by Max, expresses the belief that stealing fails to maximize happiness, and disapproval of actions which fail to maximize happiness. Similarly, if Jax disapproves of actions whose maxims are non-universalizable, then ‘stealing is wrong’, when spoken by Jax, expresses the belief that the maxim of stealing is non-universalizable, and disapproval of actions whose maxims are non-universalizable.
Now, I think there are many problems with this sort of view, but the one I want to focus on arises when you consider third-person belief reports. Suppose that Max and Jax both believe that the maxim of stealing is non-universalizable, both believe that stealing does maximize happiness, and both believe that each other believe these things. Jax says, ‘stealing is wrong’, and Max believes that Jax is sincere. Max says, ‘Jax thinks that stealing is wrong’.
What I want to know is, whether Max has just said, inter alia, that Jax thinks that the maxim of stealing is non-universalizable, or whether he has just said, inter alia, that Jax thinks that stealing fails to maximize happiness. What should this sort of theory say?
That seems like a hard question (which I gather is part of the point of asking it). We generally presume that a sincere utterance of a sentence ‘P’ is good reason (other things equal) to attribute the belief that P to the person making it. On that way of going on, it seems like it makes more sense that the sentence attributing the belief in question attributes the belief with a content predicating the property that the subject of the attribution disapproves of with the sentence P, rather than the belief whose content attributes the property I disapprove of. So if we’re talking about Jax, it seems like the thing to attribute is the belief that stealing is non-universalizable. At least that way of going has the virtue of having the content of the sentence justified by the sort of thing we take to be good evidence for uttering it.
But then normally if you and I each think that an action is wrong, but for different reasons, we say stuff like, “We agree that punishing Jack is wrong, but we disagree about what makes it wrong.” And it seems like the last clause is often meant precisely to capture a difference between us that is like the difference between Max and Jax. So that cuts the other way.
I’m sure you’ve thought of these considerations and more besides.
A great question. I’m with Mark. I think the account could be slightly less specific. The account of what Max said would not have to pinpoint down the specific property that is disapproved. I think doing that would get the view to all sorts of epistemic problems.
So, When analysing Max’s utterance ‘Jax thinks that stealing is wrong’ you could have a choice between:
A)Jax believes that stealing has some property he disapproves of [and]
B) Jax believes that stealing has some property I (that is, Max) disapprove of.
I think the account would have to go for A. It would have to find room for accounting for utterances like ‘Jax thinks that stealing is wrong but it actually isn’t’. Here you want to say that Jax has an attitude towards something he shouldn’t have and for the expressivist it is the latter that is an expression of my attitude. If the first part of the utterance was related to my attitudes too, the whole utterance would not make sense. The propositional attitude ascription is an intensional context, so ‘stealing is wrong’ cannot be understood in the same way there as in the simple utterance where my attitudes are expressed. I think this would be motivation for A. And, I’m too sure you’ve thought of this. What would be the problems of A?
This is indeed a brain tickler of a question, but I have a subsiduary question, (perhaps a question hidden in the one above):
Why do we connect “sincere assertion” with belief?
While this does indeed hold over some examples, I simply cannot see why this is the mark of the generalised on it.
Surely, not everything that is uttered can be attributed to being believed?
Also, it seems that assuming the belief nature of sincere utterances and the substitutivity of co-referential terms can lead to paradoxes (Frege and Kripke) in language, so surely there must be doubt thrown on the assumed correlation of “sincere assertion” and belief?
I have read your question a number of times and I do not see the problem. If both Jax and Max believe that stealing maximizes happiness and both believe that the other believes this to be so then how can Max entertain the belief that Jax thinks that stealing fails to maximize happiness? This would go against what you claim Max believes (to know) about Jax. What am I missing?
Hi, John.
I didn’t say yet what the problem is supposed to be. You, like Jussi, seem to think that the answer has to be that Max has said (inter alia) that (1) instead of that (2):
(1) Jax thinks that the maxim of stealing is non-universalizable.
(2) Jax thinks that stealing fails to maximize happiness.
After all, you note – and this is what Mark vR pointed out in favor of this view – it makes perfect sense for Max to report what Jax thinks in this way, but it wouldn’t make any sense for him to do so, if he was saying (2) – which he knows to be false. Hence, you say, there is no problem; the answer has to be (1). Jussi tentatively agrees with you, and Mark vR sees the pressure to go this way, but also sees problems with the other side.
Here is a first reason to suspect that things are not so obvious. According to this theory, ‘wrong’ has a dual, expressive-and-descriptive content, but its descriptive content is an ordinary, context-dependent one. If Jax says, ‘I like stealing’, it does not make sense for Max to report this, even if he thinks that Jax was sincere, by saying, ‘Jax thinks that I like stealing’. Contra Jussi’s remark, context-dependent terms do not work differently inside intensional contexts – they work in the same way both inside and outside of intensional contexts. If Max wants to report what Jax thinks, he needs to say, ‘Jax thnks that he likes stealing’.
So if in Max’s mouth ‘wrong’ picks out that property, whatever it is, that Max disapproves of actions for instantiating, then the hybrid theorist would need a totally new theory about the compositional semantics of attitude-ascriptions, in order to be able to say that the answer is (1), as you and Jussi think it should be.
This is at least some kind of problem, because hybrid views are attracting people in large part because they promise to provide answers to traditional problems ‘on the cheap’. They are supposed to be unlike traditional expressivism in not needing to provide a creative and revisionary semantics for natural-language constructions. So this, I think, is grounds to think that the answer to my question should not be obvious.
Still, I can put my question again: suppose that the details of a hybrid semantics for attitude-ascriptions could be worked out that would yield the result that the answer is (1) rather than (2). Is that what this kind of hybrid theorist should want?
Hi Mark,
Thanks for the challenge, which is a bracing one. I won’t speak for defenders of other hybrid views, as different sorts of hybrid views will, I suspect, need to give slightly different answers.
My own stab at the general cluster of issues is in a footnote to my ‘Best of Both Worlds’ paper from the Madison conference a few years ago, published in the Oxford series. Instead of trying to summarize that, I’ll just cut and paste it in here and see what you think.
“Actually, one very important qualification must be added here, which raises issues I must put to one side here. For the general account of the meaning of normative predicates laid out in the text does not plausibly extend to contexts in which normative predicates figure in the contents of a propositional attitude attributed to someone (e.g., when I say, “She believes that abortion is wrong.”). The point is that in these contexts we are not typically assuming that the person to whom we attribute the propositional attitude associates the same cluster of descriptive properties with a given normative predicate that we do. For example, when I, as a utilitarian, say that Jones believes that abortion is wrong and need not be presuming that Jones believes that abortion fails to maximize happiness.
So we should understand such attributions in terms of the attribution of a suitable belief/desire pair without taking a position on whether the speaker shares our conception of the good. So when I say that she believes that abortion is wrong I am making a purely descriptive claim, namely that she has the belief that abortion is wrong. It turns out (though a given speaker may not realize this, of course) that the belief that abortion is wrong is really in one sense a belief/desire pair – a general pro-attitude of the right kind and a belief which makes suitable anaphoric reference back to the content of that pro-attitude.
In itself this need not pose any special problems. The meaning of normative predicates in propositional attitude ascriptions is connected in obvious and systematic ways to their meanings in other contexts. Moreover, there need be no special problem about the validity of arguments employing such ascriptions as premises, since we cannot in general draw any inferences about the contents of such ascriptions from the ascription itself (apart from the fact that someone believes or desires that content, and that inference is valid on the account developed here).
An instructive analogy is with pejorative terms (also discussed in Copp (2001)). Plausibly, to call someone a ‘nigger’ is at least in part to express contempt toward certain people in virtue of their race. However, intentional attitude ascriptions need not involve any such expression of contempt. For example, someone who sincerely says, “David Duke just thinks of me as a nigger,” certainly does not thereby express contempt for people in virtue of their race. Instead he ascribes to Duke an attitude of contempt and a belief that he (the speaker) has the features to which this contempt is cued. Here we have a nice parallel with the account developed here, for on the Ecumenical Expressivist account we should also say that such contexts involve the ascription of a suitable attitude/belief pair. Moreover, this shift in expressive meaning (in the case of pejoratives) from intentional attitude ascriptions to other contexts seems to create no insuperable problems in this context, and this point should be common ground. So if there are general problems lurking here then they are problems for everyone and not just the expressivist.
The only real difficulties emerging for Ecumenical Expressivism on this front arise when we combine ascriptions of normative beliefs with claims about the truth of what the subject believes, which should allow us to infer a normative conclusion. For example, we have inferences like, “She thinks abortion is wrong, and everything she thinks is true, so abortion is wrong.” However, I shall not here go into the details of how Ecumenical Expressivism is best extended to deal with these further cases. For this would require a full theory of truth (for a start) and would therefore take us too far afield from an outline of the basic ideas and advantages of the ecumenical approach. I explore these issues in my “The Truth in Ecumenical Expressivism.” Thanks to Timothy Williamson and John Hawthorne for pressing me on this point.”
Obviously there are further issues lurking here, but that is roughly how I would respond. Applying this to your example, Jax’s assertion expresses the belief that Max believes that stealing is wrong, and this is, inter alia (on this version of my view at least – I now favour a version that involves the idea of an ideal prescriber) the belief that Max approves of actions (and refrainings) insofar as they have some property p and at the same time believes that refraining from stealing has that property. Note that Jax may not know that this is precisely what his speech act expresses – he may not know the correct metaethical theory – but this needn’t be an objection in itself.
This is really just a slight variation on Jussi’s option (A), but with a few more bells and whistles added.
All the best,
– Mike
Consider the following argument:
P1 Jax thinks that stealing is wrong.
P2 Stealing is not wrong.
C Jax thinks something false.
In answer to my second question, Mike has just said that he would take the option where the descriptive content of the attitude-report is determined by the subject’s attitudes, rather than by the speaker’s. He didn’t have an answer as to how this result is generated by a compositional semantics, but I didn’t ask for one; I had set that worry aside, noting that it makes his idea of solving the problems of metaethics ‘on the cheap’ suspect, but I’m willing to go with him so far as to imagine that we have a semantic theory that generates the results that he desires.
On this theory, the descriptive contents of premises P1 and P2 are given as follows:
P1* Jax thinks that stealing is P (where P is the property Jax disapproves of actions for being).
P2* Stealing does not fail to maximize happiness.
No matter what descriptive content we give to C, this argument will therefore equivocate. There is a slight complication here, which is that Mike wants to say a bunch of complicated things about how the word ‘false’ works. But this problem should not turn on them. One of the central points of Mike’s hybrid account is to offer a theory which accounts for logic and inference ‘on the cheap’.
I’ve just given an example of a perfectly fine argument for which the descriptive contents that Mike assigns equivocate – in fact, for which the descriptive contents are not even truth-preserving. So even if he were to have something fancy to say about ‘false’ that would in some sense explain how the conclusion followed, that explanation could not work by appeal to the truth-preservingness of the descriptive contents. He has just endorsed the view from which it follows that those descriptive contents equivocate. So if he has any explanation of why the conclusion follows at all, it will have to be of the sort that ordinary, pure, expressivists might offer – which would appeal to attitudes, and not just descriptive contents. This, I think, is a bad position to be in.
So the problem I’m posing is a dilemma: either (1) you say that the descriptive content of an attitude ascription is fixed by the speaker, in which case you fail to license attitude ascriptions when they are obviously appropriate, or (2) you say that the descriptive content of an attitude ascription is fixed by the subject, in which case you create an equivocation in inferences that involve moral sentences both inside and outside of attitude ascriptions, like the one that I offered.
Either way, this looks very bad to me. Notice that Mike’s references to pejoratives in the footnote that he provided for us do not help, because pejoratives have a fixed descriptive content that is the same for different speakers. The problem that I’ve raised is a problem only for hybrid views according to which there is a differentdescriptive content for different speakers. So, for example, Dan Boisvert’s very nice hybrid theory is designed precisely in order to avoid these kinds of problems. Dan’s theory models moral terms closely on pejoratives, and it has precisely the features that Mike’s footnote notes about pejoratives. Mike’s theory, unfortunately, doesn’t have those features.
Here’s a related problem that’s occurred to me. Consider:
A) Arnold thinks stealing is wrong.
B) Betty thinks stealing is wrong.
Therefore, C) Arnold and Betty agree on something.
I take it this inference is valid. But if Arnold’s and Betty’s beliefs have different descriptive properties, the inference is not valid. And if the premises have the structure that Mike suggested (I think):
A*) Arnold thinks stealing is P (where P is a property of which Arnold disapproves)
B*) Betty thinks stealing is Q (where Q is a property of which Betty disapproves)
the inference is also not valid.
The only way I can see to get a common descriptive content for A and B is to use the speaker’s descriptive content, but (as Mark notes) that will get Arnold and Betty’s beliefs all wrong.
Just two quick points. First, notice that the problem Mark is raising here is not essentially one for *hyrbid* theories that advocate a speaker-relative account of descriptive content. It is a problem for *any* speaker-relative account of the descriptive content of moral sentences. For example, the problem seems to be applicable to ethical subjectivism as well as to the speaker-relativist theory Jamie suggested some years ago in “Internalism and Speaker Relativism” (which I’m assuming is not a hybrid theory).
Second, towards the end of that paper, Jamie discusses an interesting example that appears to show that the expression ‘nearby’ might be an example of an expression that is speaker-relative when unembedded, but subject-relative when embedded. If I remember the example correctly, it is something the following. Someone (John) is coming to dinner at my home, runs out of gas while still quite a distance from my home, calls me, and says, “Don’t worry, there is a gas station nearby.” I turn to my wife and say, “John says not to worry, there is a gas station nearby.”
I’m wondering what folks think about Jamie’s example.
Hi, Dan.
Imagine the conversation had started like this: you’re on the way to John’s house, and run out of gas. You call him, and he says, ‘where are you? Oh, don’t worry – there’s a gas station nearby.’ You turn to your wife and say, ‘John says not to worry – there’s a gas station nearby.’
‘Nearby’ isn’t an example of a word that is speaker-relative outside of attitude reports but subject-relative inside them; it’s an example of a word that is context-dependent in some way that is more flexible than ordinary examples of ‘pure’ indexicals like ‘I’, ‘here’, and ‘now’. It can sometimes pick out the location of the speaker, and sometimes some other contextually salient location. Regardless of whose location is picked out by ‘nearby’, however, the attitude-report can disquote only if the speaker refers in her context to the same thing as the subject of the report did, in her context.
This makes me think that Jamie’s move was really just away from a strictly speaker-relative view to a more flexible contexualist view, like the one explicitly advocated by my colleague Steve Finlay. A view like Steve’s has the resources to say, in answer to my original question: ‘well, either one – it depends on the context’. But even a flexible view like this one can’t allow for a single context in which Max can both disquote what Jax says, voice his own view, and then go on to deduce that Jax thinks something false – a piece of data that both non-contextualists and pure expressivists can explain.
Incidentally, I’d be interested to hear from Mike about this, but I take the greater flexibility of the Dreier/Finlay sort of view to be strictly ruled out by Mike’s other commitments. Mike thinks that a speaker will express a desire-like attitude that that very speaker has, and that the descriptive content of a sentence will be determined by the desire-like attitude that that very sentence expresses. So in general, on Mike’s view, the descriptive content of a moral term is not flexible in the way that it is on Steve’s view. It is always fixed by facts about what the speaker’s attitudes actually are – though as he pointed out, he would like it to work differently inside of attitude reports.
Hi Mark,
Yes, I should have made it clearer in my comment above that *Jamie* does not take the example ‘nearby’ to be one in which the descriptive content of that expression is speaker-relative when unembedded, but subject-relative when embedded. It’s just that I have claimed (elsewhere) that there is *no* such natural language expression whose content is speaker-relative when unembedded, but subject-relative when embedded. Jamie’s example ‘nearby’ is the closest I’ve seen to anything that could serve as an effective counterexample to my strong claim, so I was wondering whether anyone thought it could indeed serve as an effective counterexample.
As for whether Jamie’s Speaker-Relativism is not *strictly* speaker relativist, I’m having a hard time seeing how it can avoid being so. As I understand his view, the descriptive content of a sentence like ‘Stealing is wrong’ would be something like:
“Stealing is F, where F is the property of being approved of by a moral system that best balances the content and subject matter of a standard and the motivations and affective connections *with the speaker*.”
Is this, in fact, what Jamie’s view comes to? (By the way, does Jamie still hold this view?) If so, I’m not quite seeing how it avoids “strict” speaker-relativism and the implications you have described in your post.
Dan,
I think Jamie’s view was, ‘Stealin is wrong’ expresses that stealing has the property that is disapproved of by the most salient moral system in the context of conversation. Among the things that make a moral system salient are the attitudes of the speaker, listener and agent. (It’s easy when they all agree, harder when they don’t.) But there could be other features that make a moral system salient. And the system need not be the one fixed just by the actual beliefs/attitudes of the relevant persons as to what was right and wrong. It could also depend on the most attractive way of cleaning up their commitments to get a coherent and defensible moral view. So, for example, the salient system even when the speaker, listener and agent are all carnivours could include a commitment to vegetarianism if one of the things they do hold in common would make that further commitment more coherent with what they have in common than its denial would be.
But I’ll bet we’ll be hearing from Jamie soon.
I love the smell of Pea Soup in the morning! I just hope its not Apocalypse Now for Ecumenical Expressivism [Ok no more bad jokes for this post, anyway…]
Seriously though, thanks again for the challenge. These uses of normative vocabulary in intentional contexts are especially tricky for me, so it is good to keep my heels to the fire here. There was no way to respond to this very difficult (one of the most difficult, I think) challenge for my view in a short span, so apologies in advance for the unusual length of this post.
I’d like to concede at the outset that the example of ‘nigger’ and other pejoratives is not a wholesale save of my view (not that I really had meant it to be, though no doubt I could have been clearer about that) – I concede what Mark says about that example. The only point I meant to make there is simply that we should not be too quick to dismiss the idea that the expressive meaning of a term could shift from non-intentional to intentional contexts – and on that point, Copp and Boisvert and I agree. This might be relevant to the issue about compositionality that Mark raises but puts to one side, but I agree that it does not meet the main challenge he is actually pressing here; fair enough.
Mark is also right that I will need to advert to a ‘fancy’ account of discourse about truth and falsity to deal with this problem. On the other hand, he is not right to claim that this solution will force me to go for a more standard non-ecumenical expressivist solution in these cases, or so I shall argue. At any rate, unlike some versions of non-ecumenical expressivism, my account won’t explain logical validity in terms of a ‘logic of attitudes’ or a ‘fractured sensibility’ and the idea of a descriptive content will figure in my solution, although not in any simple way. However, my views here are rather complex, so it will take a bit of philosophical spadework to explain how I meet the challenge.
One last concession: I should at the outset also say that given the complexity of the account I am about to offer that the ‘on the cheap’ rhetoric, which I have sometimes used, is perhaps itself a bit cheap. I think that my account can deliver certain results more elegantly than more standard forms of expressivism, and is on balance superior, but if by ‘on the cheap’ one meant something like ‘with less fancy conceptual machinery’ then that wouldn’t really be apt, given my take on uses of ‘true’ and ‘false’. OK, enough preliminaries; begin proper reply.
The specific challenge is to explain how my account can accommodate the logical validity of the following argument:
P1 Jax thinks that stealing is wrong.
P2 Stealing is not wrong.
C Jax thinks something false.
The more general worry is that my account implausibly entails that such arguments fall prey to the fallacy of equivocation. If this could be sustained then it would be a riff on the Frege-Geach problem, which also alleges the imputation by the expressivist of equivocation where there is none. Since one of the main arguments for Ecumenical Expressivism’s superiority to other forms of expressivism is its ability to solve the Frege-Geach problem, this is a very serious objection indeed.
Since the challenge is to explain the validity of certain sorts of arguments, it is useful to put my definition of validity on the table:
An argument is valid just in case any possible believer who accepts all of the premises but at one and the same time denies the conclusion would thereby be guaranteed to have inconsistent beliefs.
This account combines with Ecumenical Expressivism to deal quite nicely with arguments employing normative predicates without any intentional contexts. For example:
P3 Jax’s stealing was wrong.
P4 If Jax’s stealing was wrong then he will go to jail.
C Jax will go to jail.
On the most recent version of my account, P3 expresses endorsement of a certain sort of prescriber, and the belief that that sort of prescriber would be disposed to insist on Jax’s not stealing. P4 expresses endorsement of a certain sort of prescriber and the belief that if that sort of prescriber would be disposed to insist on Jax’s not stealing then he (Jax) will go to jail. The conclusion simply expresses a belief that Jax will go to jail. For someone simultaneously to accept the premises and deny the conclusion of this argument would (in part) be to believe that a certain sort of prescriber would be disposed to insist that Jax not steal, believe that if that very sort of prescriber would be disposed to insist that Jax not steal then he (Jax) will go to jail, but at the same time believe that Jax will not go to jail. This is transparently inconsistent, so the original argument is valid. So far, so good.
Now, things get much more complicated with intentional contexts combined with the use of ‘true’ and ‘false’. The next step is to see what I want to say about uses of ‘true’ and ‘false’. Other expressivists adopt a deflationist theory of truth discourse, and if deflationism is defensible then I think I could adopt these solutions as well, though this would mean that Mark is right that in these cases my account would be in the same boat as other forms of expressivism. That needn’t be a fatal objection so long as the view has other advantages, but it would be a little disappointing in some ways.
However, I have some technical worries about deflationism about truth. I also think it would be unfortunate if expressivism’s tenability were tied to the tenability of deflationism about truth. It would seem to give unnecessary hostages to fortunes. I also wanted to see whether the move to an Ecumenical form of Expressivism didn’t provide an opportunity to be more Ecumenical about the right theory of truth in its most primary sense, anyway. What follows is an abbreviated (believe it or not!) account of my discussion of these matters in ‘The Truth in Ecumenical Expressivism’.
One way to think about Ecumenical Expressivism is as a kind of surprising semantic discovery. Common sense may tacitly suppose that both normative and purely descriptive assertions express beliefs, and beliefs in exactly the same sense. Expressivism (both ecumenical and non-ecumenical) entails that this is false. In a strict and philosophical sense of ‘belief’, normative utterances express belief/desire hybrids where the belief has a descriptive content, and not simply beliefs in the strict sense with normative contents. I have elsewhere compared this to the discovery that ‘jade’ did not denote a unique natural kind. The same might turn out to be true of ‘belief’, with ‘normative belief’ and ‘descriptive belief’ being the analogues of ‘jadeite’ and ‘nephrite’.
Given this discovery, we could just decide that there really are no normative beliefs, but there are good quasi-realist reasons not to do this, or so I have argued. Instead, we should retain a deflationist sense of ‘belief’ according to which whatever state(s) of mind expressed by an ordinary assertion are ipso facto beliefs (well, it’s the hybrid of those states which is the normative belief in this sense – its not that each individual state of mind in the strict sense is a belief in this more deflationist sense). So now we have two senses of ‘belief’ – a strict and philosophical one (spelled out via some notion of ‘direction of fit’ – functionally on the version I’d prefer) and a deflationist one. In the strict and philosophical sense there are no beliefs with normative contents, and in the deflationist sense there are. Note the analogy with jade. On discovering that ‘jade’ did not denote a single natural kind we could have inferred that there is no such thing as jade, but given our practical interests in jade discourse (the jeweler’s purposes didn’t depend on natural kindhood) we sensibly did not do this.
OK, so suppose we accept this story. To answer Mark’s challenge we need to get back to claims about truth and falsity. However, what I want to urge is that the germane question now is not ‘What does ‘true’ as we now use it actually mean?’ as that meaning might not fit well with the expressivist’s surprising semantic discovery. Instead, we should ask, ‘What should we mean by ‘true’, assuming we want to continue to treat normative judgments as beliefs in a quasi-realist fashion?’
Well, lets take whatever is the most plausible theory of ‘true’ and ‘false’ quite apart from the issues raised by expressivism. One nice feature of this approach is that we need not suppose that this theory will be deflationist – it could even be the correspondence theory. Now the expressivist proposal (and please note that this is a proposal about what we might now come to mean as opposed to what we already do mean) about the use of ‘true’ and ‘false’ in contexts which range over normative claims is as follows [in my paper, I develop this more slowly and motivate it, but this post is getting very long already!!!]:
Take any sentence ‘p’ in which locutions of the form ‘q is true’ are used. On the proposed account, an utterance of ‘p’ expresses two states of mind:
(1) A suitable pro-attitude to a certain sort of prescriber
(2) The strict sense (direction of fit sense) belief that r, where r is what you get when you take ‘p’ and replace all uses of ‘q is true’ with ‘there is a proposition s, the strict sense belief in which would (at least partly) constitute the deflationist belief that q for anybody who endorses that sort of prescriber[anaphoric reference back to the content of the proattitude in (1) here], and s is true’.
The idea is that the use of ‘true’ in (2) can be understood in the primary sense of ‘true’ which applies only to literal descriptive contents. That could be a deflationist view, the correspondence theory, an epistemological conception, or whatever. One advantage of this approach is that it is consistent with a wide range of theories of truth in the most primary sense of ‘true’ (as it applies in the most strict sense to literal descriptive contents) and then just offers a way of generalizing on that base notion to a broader sense of ‘true’ which can apply across normative contexts as well.
Ok, so that does mean that in contexts in which ‘true’ and ‘false’ range over normative contexts that I am offering an expressivist theory of the use of ‘true’ and ‘false’ at least insofar as those predicates range over normative contexts. That may generate some incredulous stares from those who are more traditional in their tastes about the theory of truth, but remember I am not claim that this is what we already in fact mean by ‘true’ and ‘false’ but what we should come to mean by them in light of the Ecumenical Expressivist’s surprising semantic discovery – otherwise we would have to give up on normative discourse as we now know it – again, all of this is assuming we agree that the expressivist really has discovered something true about ordinary practice, but I am just assuming this here.
Now lets (finally!!!) return to Mark’s test case argument, which went as follows:
P1 Jax thinks that stealing is wrong.
P2 Stealing is not wrong.
C Jax thinks something false.
To test for logical validity, given my psychologized definition of validity, we need to see what anyone who accepted the premises but denied the conclusion would thereby believe (in the strict sense) and then see whether those beliefs are contradictory.
(1) Acceptance of P1 is constituted by the belief that Jax believes, in the deflationist sense of ‘believes’, that stealing is wrong. [which is, inter alia, the belief that Jax endorses a prescriber which he takes to be disposed to insist on not stealing].
(2) Acceptance of P2 is constituted by endorsement of a certain sort of prescriber and the belief that that sort of prescriber would not be disposed to insist on not stealing.
(3) The denial of (C) would be the belief that it is not the case that Jax believes something false. This, in turn, would be constituted by (a) the endorsement of a certain sort of prescriber (the same one endorsed in (2), given simultanaety of accepting (P2) and rejecting (C)) and (b) the belief that it is not the case that Jax believes (in the deflationist sense of ‘believes’) something such that his belief (in the deflationist sense) would be constituted by a belief (now in the strict sense of ‘belief’) that q by anyone who endorses that sort of prescriber, and such that q is false.
Someone who satisfies 1-3 does, however, have contradictory beliefs, and in the strict direction of fit sense of ‘belief’ (no logic of attitudes here), and so the argument does come out as logically valid on my account. Here is the proof:
Consider someone S, who satisfies 1-3.
Step 1. By (1), S believes (in the strict sense of ‘believes’) that Jax believes (in the deflationist sense of ‘believes’)) that stealing is wrong.
Step 2. By (2), S believes (in the strict sense of ‘believes’) that a certain sort of prescriber P would not be disposed to insist on stealing.
Step 3. By (3), S also believes (in the strict sense of ‘believes’) that Jax does not believe (in the deflationist sense of ‘believe’) something such that anyone who believes (in the deflationist sense of ‘believes’) that and endorses a prescriber of type P would have a belief in the deflationist sense which is partially constituted by a belief in the strict sense that q, such that q is false.
Step 4. However, from the correct semantic theory (Ecumenical Expressivism), it follows a priori that someone (say, Jax) who believes (in the deflationist sense of ‘believes’) that stealing is wrong and at the same time endorses a prescriber of sort P must also thereby believe (in the strict sense of ‘believes’) that a prescriber of sort P would be disposed to insist on not stealing.
Step 5. So, from Steps, 1, 3, and 4, it follows that the belief (in the strict sense) that S believes (in the strict sense) that the belief (in the strict sense) that a prescriber of sort P would be disposed to insist on not stealing is not false.
Step 6. However, from Steps 2 we know that S believes that a prescriber of sort P would not be disposed to insist on stealing.
Step 7. From Steps 5 and 6 plus the assumptions of bivalence (not trivial, but, I think, defensible) and the assumption that truth in its primary sense (correspondence, or whatever) is at least disquotational, we get that S at one and the same time believes that (i) a prescriber of sort P would be disposed to insist on not stealing, and (ii) a prescriber of sort P would not be disposed to insist on not stealing.
Step 8. Since this line of reasoning is perfectly general (no special assumptions about S were made) it follows that anyone who accepts the premises but rejects the conclusion of the argument would thereby hold contradictory beliefs (in the strict sense of ‘beliefs’).
Step 9. From Step 8, plus my psychologized definition of logical validity, it follows that the original argument is logically valid.
QED
Now this proof is quite elaborate and complex, so if my theory suggested that this is what goes through the mind of an ordinary speaker in looking an argument like Mark’s then that would be ridiculous. However, that is not my claim. This proof should be seen as a piece of meta-theory which vindicates a very simple form of logical reasoning given a certain semantic theory. The ordinary person need not be seen as thinking about this at all in assessing arguments as valid any more than they need to be seen as thinking about the proofs of soundness and completeness in assessing arguments’ validity. So I don’t think the admittedly complex proof that this form of reasoning comes out as logically valid in my system is in itself any kind of objection. It does, however, force me to admit that along this dimension, at least, my solution to the Frege Geach problem is probably not so well glossed as being ‘on the cheap’!!!
Now return to Mark’s objection in his last posting on this:
“So even if he were to have something fancy to say about ‘false’ that would in some sense explain how the conclusion followed, that explanation could not work by appeal to the truth-preservingness of the descriptive contents. He has just endorsed the view from which it follows that those descriptive contents equivocate. So if he has any explanation of why the conclusion follows at all, it will have to be of the sort that ordinary, pure, expressivists might offer – which would appeal to attitudes, and not just descriptive contents.”
False dichotomy. I can’t simply invoke the entailment of the descriptive contents in any straightforward sense, so the first point is right. It doesn’t follow from this that my account ‘will have to be of the sort that ordinary, pure expressivists might offer…’ In fact, given that my account makes essential use of the ecumenical machinery and the beliefs (in the strict sense) associated with normative judgments, the solution I offer is not even available to other, ordinary expressivists. Given the complexity of the solution, you might not think that is a problem for them (!), but I won’t try to convince you about that one way or the other here. My point is just that the dichotomy Mark presented is a false one, given the actual details of my account.
If I am being completely honest, I would admit that I would prefer a much simpler solution to contexts in which normative predicates occur in intentional contexts. That said, I don’t think the complexity of the account I have developed is, in itself, a decisive objection to it.
OK, that is it for now. Sorry for the ridiculous length of this post.
– Mike
On Heath’s challenge, which is as follows:
A) Arnold thinks stealing is wrong.
B) Betty thinks stealing is wrong.
Therefore, C) Arnold and Betty agree on something.
I take it this inference is valid. But if Arnold’s and Betty’s beliefs have different descriptive properties, the inference is not valid.
————————————
Well, here I need to invoke something like Gibbard’s agreement in plan, or Stevenson’s agreement in attitude. In one of *those* senses, suitably spelled out, we do have agreement between Arnold and Betty.
On my story, (1) and (2) entail that Arnold and Betty both are committed (on pain of irrationality) to not stealing, and that they are disposed to feel certain negative feelings toward themselves if they do steal, and so on. On my account this is sufficient for agreement in one important sense of ‘agreement’. Here, of course, my account is more like traditional expressivist theories, (in particular, it is very close to Gibbard’s account), but I don’t mind that. The Ecumenical idea is that by bringing in both beliefs and desires you can use each where it can do the most effective work. In appealing to the conative states to explain in what sense we have agreement here, I am just being true to my Ecumenical roots!
Thanks for the objection,
Mike
Mark vR, thank you. That helps me understand the view a lot better.
Mike,
I’m not sure your response meets Mark’s challenge.
Your position, as I understand it, is this: if the conclusion of the argument (the one about Jex) meant something other than what it in fact means — i.e. if it meant what it would on your revisionary semantics for ‘true’ and ‘false’ — then the argument would be valid. But Mark’s challenge is to explain why the argument is valid as it is, not to propose a way of changing the argument so that it would become valid.
Campbell,
Well, that is a fair point, but then you can see my reply as going in two stages. First, if the sort of expressivism about normative discourse I am defending is correct then we need to rethink our understanding of truth discourse anyway. Second, once we do that in a plausible and principled way, it turns out that the arguments in question are indeed logically valid.
Also, as Gibbard has plausibly suggested, the distinction between conceptual analysis on the one hand and proposals for how we might come to use various predicates is not a sharp one. Any analysis, as he puts it, will strain our pre-existing concept in various ways. My idea is that my proposal about ‘true’ is not arbitrary (which would rob the point about how it saves validiy of its interest!) but is one which can accommodate most of what we want to say anyway in a plausible way that fits in with what, according to Ecumenical Expressivism, anyway, is the truth about what we mean when we discuss matters normative.
Best,
Mike
Mike,
You suggested that
C) Arnold and Betty agree on something
above is true just because A&B agree “in plan.” I’m thinking that if you say this, agreement “in plan” is going to do all the work. Consider
(1) Arnold and Betty agree about whether stealing is wrong.
(2) Stealing is not wrong.
(3) Therefore, either (Arnold and Betty both believe something true) or (Arnold and Betty both believe something false).
I wish I could put this point more formally, but your logical pyrotechnics are too intimidating. So let me put it informally: What makes (1) true, according to you, is that Arnold and Betty agree “in plan”; nothing descriptive about their views on stealing comes into it. Then nothing descriptive in (2) is going to interact with (1) in order to produce (3). The only thing that matters about (2), for this inference, is the “plan”.
So the inference is valid wholly in terms of the non-descriptive elements of the respective plans involved.
But this first inference is not importantly different in kind from the one Mark brought up before, namely
Jax believes stealing is wrong.
Stealing is not wrong.
So Jax believes something false.
If plans can do the work in the Arnold and Betty case, they should be able to do it in the Jax case too. And the contrapositive.
Mike,
I may be getting lost in the negations, but I’m not understanding (3) as a gloss of the denial of (C). It looks to me that the denial of C ought to cover all beliefs including non-moral beliefs and so that it should not essentially involve endorsing any prescribers. But maybe I’m just confused about what’s going on here. Can you help?
Whew. And I thought I had been made fun of in public for the length of my posts…
First, I take Campbell’s response to Mike as decisive given the revisionary flavor of Mike’s view of ‘true’ and ‘false’; if the data are that arguments like the one I cited are valid in the language that we already speak, then offering a helpful way that we could come to use ‘true’ and ‘false’, once we recognize the truth of Mike’s view, does not explain why those arguments are already valid. It just explains what else we could mean by them that would make them valid.
But in fairness to Mike’s proposal, I take its revisionary flavor to be non-essential; he could instead have offered the same view as one about the semantics of ‘true’ and ‘false’ in our actual language. If that’s the view, then I think that this account of ‘true’ and ‘false’ is false, but it’s not fair to assume that it is, in posing this problem for him.
Mike is correct to note that I cheated when I stated the problem; I emphasized the equivocation in the descriptive content of the two appearances of ‘stealing is wrong’ in the premises, and I did that because on a naive view, the validity of this argument turns on the two occurrences of ‘stealing is wrong’ not being equivocal in their descriptive content. But this was cheating, because I ignored what the descriptive content of the conclusion was going to be, and there are some descriptive contents that could follow from the premises even though the two appearances of ‘stealing is wrong’ do not have the same descriptive content. Compare, for example, the following ordinary descriptive argument:
P1 Jax thinks that he is tall.
P2 I am not tall.
C There is some sentence, S, such that Jax believes what he could say by uttering S, but S is false relative to my context of utterance.
So despite the fact that ‘he is tall’ and ‘I am tall’ have different descriptive contents in the premises, we can still draw conclusions from them. Mike’s move is like this one – though of course he would urge me to add, for completeness, that he’s said a lot more to explain why this is not totally ad hoc.
I use this example to illustrate, partly in order to bring out that there are a lot of things that Mike has not told us about his proposal. For example, he tells us that he has a ‘deflationary’ account of ‘believes’, but he has not told us what, exactly, are the objects this deflationary sense of ‘believes’ relates us to. Yet his statement of the descriptive content of ‘Jax believes something false’ quantifies over things believed in the deflationary sense.
Presumably it is not propositions that are believed in this deflationary sense. Maybe it is sentences. He’s going to have to answer questions like this, however, in order to be able to deal with extensions of Heath’s problem. Consider, for example, the sentence, ‘Arnold and Betty agree on something and it is false’. If the thing they agree on is a plan, but it is sentences that are false, sentences like this won’t make sense. Moreover, they might not agree on any sentence, because Arnold might be a monolingual German speaker and Betty might be a monolingual anglophone. There are reasons, after all, why philosophers generally treat propositions as the objects of agreement, disagreement, truth, falsity, belief, desire, hope, anguish, and so on.
Incidentally, I believe that I’ve shown how pure expressivists can handle these sorts of constructions in Being For.
I have more to add in support of Dan’s conjecture that natural language constructions can’t interact in the way that Mike claims that ‘thinks that’ and ‘wrong’ do, but I have to get ready for Steve’s seminar!
Hi, Mark –
I think Mike just endorsed the view that all sentences containing ‘true’ or ‘false’ express endorsement of the prescriber. You might think that he has to say this, in order for sentences like ‘what Al said is true’ to express states that will be guaranteed to motivate in case Al said something moral.
Moreover, you might think that this creates strong pressure to say, along with Mike, that all normative sentences whatsoever express the same desire-like attitude – approval of an ideal prescriber. For example, I think that on Mike’s view, ‘skiing is swell’ and ‘torture is impermissible’ will express the same desire-like attitude for the same speaker, and this is a big place where he departs from other hybridists, who will usually associate each normative predicate with a different attitude. If you had that view, but Mike’s account of ‘true’, then you’d have to think that every sentence with ‘true’ in it expresses the desire-like attitude corresponding to each and every moral predicate, so Mike’s view might seem like moderation in comparison.
Okay, now preparation for real.
Quick question: What is the motivation for building into the semantics of evaluative claims some content or other about the speaker or evaluator’s desires?
I have a hard time getting on board. I would have thought that possessing the relevant desire is not part of the content of what is asserted, but rather, it is a supposition for attributing the content of what is asserted. If this correct and it seems to me it is, then the types of problems raised above, e.g. problems associated with validity, seem to me to vanish.
Hi, Christian. Hybrid theories don’t build any content about the speaker or evaluator’s desires into semantic content. But some hybrid theories – specifically those of Ridge and Barker – build a context-dependent factor into the descriptive semantic content of a sentence which varies systematically from speaker to speaker, depending on what that speaker’s desire-like attitudes are.
So, for example, you can think of the general formula for such a view (which is very similar to Barker’s view) as saying that ‘stealing is wrong’ has the same semantic content, relative to a given context of utterance, as ‘stealing instantiates dthat(the property I disapprove of actions for instantiating)’, using Kaplan’s semantics for ‘dthat’. This content does not have anything about the speaker’s desires built into it, but it does have its content determined by what the speaker’s desires are.
Similarly, I understand Ridge’s view as saying that relative to a given context of utterance, ‘stealing is wrong’ has the same descriptive content (expresses the same belief) as something very much like, ‘stealing instantiates the property that dthat(the ideal prescriber I express approval of by uttering this very sentence) prescribes that we not do’. Again, given Ridge’s assumption that the ideal prescriber that you express approval of is whichever one you actuall approve of, this means that the descriptive content of the sentence is fixed by the attitudes of the speaker, but those attitudes are not part of the content.
Hi everyone,
This has been very useful for me. Let me make just a few more (mercifully shorter!) comments:
Heath:
I think I oversimplifed things a bit in my original reply in a way that created some confusion. Disagreement in plan can supervene on both plans and belief. If, for example, I plan to do whatever Sue tells me to do, and I believe that she has told me to jump up and down in circumstances C, whereas you plan to do whatever Joe tells you to do and believe that Joe told you not to jump up and down in circumstances C, then we disagree in plan, and this disagreement supervenes in part on our beliefs. This is because our plans plus our beliefs can rationally commit us to further plans/intentions, though. However, as you say, the argument and the challenge in your last post is really the same as Mark’s, so my response will be the same. I am not yet entirely sure why what I say about normative disagreement gets me into trouble with that reply, though. If the machinery (which, I admit, is elaborate) works in Mark’s case it will work here. Perhaps the worry is that the solution to the Frege problem doesn’t map so neatly onto the account of agreement/disagreement. There is probably some truth in that, but I am not sure its anything like a knock down objection either.
Mark: First you are right that its not essential to my reply to Campbell that the semantics offered is revisionary, though I think it would be a very hard line to say that the elaborate machinery maps exactly onto what we already mean, so I am not sure how much help this will be for me, the Gibbard point about analysis always stretching a comment that I made before notwithstanding. Perhaps I should go this way, though. To be honest, this is one aspect of my view where my take on things is less settled than in other areas, so I will think some more about that; thanks.
On the other hand, its not so clear to me that the revisionary approach does preclude an answer to the objection. It does, however, mean that I need to say more. I suppose what I would have to say that the logical validity of the argument assumes that the use of the truth predicate *as we already understand it* is as much as coherent when quantifying over normative contexts. If the semantic discovery I am postulating about first order normative claims is correct then, depending on the theory of truth on offer, then the coherence of the use of truth in these contexts is indeed dubious. So the first stage of the reply is to contest the validity of the argument, but this is a misleading way to put it – its only lacking in validity insofar as the conclusion involves a subtle incoherence or deeply failed presupposition or something like that.
This might seem like a lot to swallow on the strength of expressivism, the arguments in favour of which I admit are not absolutely decisive (!), but then we rarely get such clear and decisive arguments anywhere in philosophy. Furthermore, I do think the fact that there is a strategy on offer for ‘saving the phenomena’ by revising our use of ‘true’ and ‘false’ should take some of the sting out of this aspect of the dialectic.
Christian: The problems with validity still arise on my view, anyway, though I think your comment was intended more for speaker relativism than ecumenical expressivism, right?
– Mike
Hey Mark. You can see how I might have been confused given the following: “Then, for any given speaker, S, who disapproves of P-actions, ‘Stealing is wrong’ expresses the belief that stealing is P, and expresses disapproval of P-actions.”
I was understanding ‘expresses the belief’ to mean expresses the proposition that is the content of the belief and ‘expresses disapproval’ to mean expresses the proposition that is the content of the desire that P-actions not be done. And if the content of the belief and desire just is the proposition expressed the sentence, then the claim that a desire, or some feature of someone’s desires, is part of the semantics of the sentence would naturally follow.
In any case, the proposal you’ve suggested is interesting. Let me make sure I understand it, okay.
You said; “‘stealing is wrong’ has the same semantic content, relative to a given context of utterance, as ‘stealing instantiates dthat(the property I disapprove of actions for instantiating)’, using Kaplan’s semantics for ‘dthat’. This content does not have anything about the speaker’s desires built into it, but it does have its content determined by what the speaker’s desires are.”
You’re saying here that these two sentences express the same proposition in some context:
(1) ‘stealing is wrong’ and,
(2) ‘stealing instantiates dthat(the property I disapprove of actions for instantiating)’
Is the reason you think this proposition does not, that is “this content does not have anything about the speaker’s desires built into it” that the reference-fixing description prefixed by ‘dthat’ in (2) does not contribute those reference-fixing properties expressed by the description into the proposition itself?
I think the semantics for ‘wrong’ cannot be the semantics for ‘dthat(…x…)’ but I first want to understand the idea.
Christian –
It’s not my idea; I think I’m firmly on the record as thinking that it is a bad idea. But yes, the way that ‘dthat’ stipulatively works, is it semantically contributes only its referent to the proposition expressed. I might be reading too much into Barker and Ridge, but Mike’s official formulations say that the belief expressed is ‘anaphoric’ on the desire expressed. If ‘anaphoric’ reference is directly referential, as it in fact is in the central cases in which ‘anaphora’ is defined, then his view ought to be statable more precisely as I did.
Mike –
I think it’s important to reserve ‘logically valid’ for arguments that are truth preserving in virtue of their form. The argument that I gave is informally valid, because it turns on the meaning of ‘false’. I don’t know what you think the constraints are on something being what a speaker actually means by a word, but by your own account, ordinary moral modus ponens arguments have conclusions whose descriptive content is inconsistent with the descriptive contents of their premises in a way that could only be recognized by someone who had the background knowledge that all moral sentences express the same attitude (which pure expressivists like Gibbard and cognitivists like me explicitly deny). So I don’t see how this is any worse position for you to take. You might think that it is a constraint on understanding ‘false’ that you understand that arguments like the one I mentioned are informally valid, but that the exact reasons why this is so are not something that speakers have to be able to grasp.
If you say something like this, then my objection to the fork of my dilemma that you seized is going to collapse back into Dan’s point that natural languages don’t have constructions that behave in the way that you need ‘wrong’ to interact with ‘believes that’ in order to be able to seize this fork of the dilemma in the first place.
Here’s a way of spelling out at slightly greater length what the problem is: Kaplan called construction which operate on character rather than content ‘monsters’, and hypothesized that there are none in natural languages. (And Jason Stanley and Jeff King have recently offered an argument that would explain why – arguing that a compositional semantics always contributes only the content of each expression compositionally, so that complex expressions don’t even have characters that do any compositional work.) But the fork of my dilemma that you seized involves thinking that attitude ascriptions operate on characters just in case their arguments are normative, and operate on contents otherwise. Such constructions would be even weirder than monsters. Moreover, even if there were some of them, what would explain why all and only attitude verbs work this way, or why it is always normative contents that trigger the operation on character and always non-normative contents that trigger the content operation? That’s part of why, in the space of things, it is such an extraordinary thing to think.
It’s worth noting that this sort of problem is precisely what pure expressivism was originally motivated to solve.
Mark and Dan:
Interesting. I was just at a workshop in Zurich with Kaplan and when we were talking over a meal the topic of ‘monsters’ came up and he told me that he was a bit embarrassed to learn after he’d published that stuff that there actually are some languages in which there are ‘monsters’ in his sense. I don’t remember which language or the details of the example, but perhaps this is at least a bit useful for me dialectically here.
I am not entirely sure I understand why my position is well characterized in terms of character in precisely Kaplan’s sense though. Perhaps you could elaborate a bit? Sorry; I am probably just being slow.
Best,
Mike
Hi, Mike; just this: the putative advantage of a hybrid view over a pure expressivist view is that you’re supposed to be able to take advantage of an ordinary, descriptivist, compositional semantics. Dan’s view works precisely that way, with the addition of a clause in the semantics for attitude reports that he argues is needed anyway, if we’re to treat pejoratives correctly. (He and I disagree on this, but it’s an interesting question, and one you raise in your footnote.)
But if you take the descriptive component of moral sentences on your account, it has a context-dependent element. The exact property that the descriptive component contains, for a given utterance, is fixed by facts about what desire-like attitude that very sentence expresses, which is in turn fixed by facts about which ideal prescriber the speaker approves of. So the descriptive component has a non-trivial character.
We can simplify and skip ‘belief’ reports and stick to indirect discourse reports to raise all of the same problems; ordinarily an indirect discourse report of ‘Jax said that P’ attributes to Jax having said something whose descriptive content, relative to Jax’s context of utterance, was the descriptive content of ‘P’, relative to the speaker’s context. So ‘said that’ operates on content, not character. You want it to work that way for ordinary indexicals; obviously if Jax says, ‘I’m awesome’, I can’t report this by saying, ‘Jax says that I’m awesome’. But you want it to work otherwise for a special class of context-dependent terms – the normative ones. For those, you want the semantics of ‘Jax said that P’ to pay attention not to the actual descriptive content of ‘P’, but to the descriptive content that it would have, if spoken by Jax. I don’t think that the weirdest part of your proposal is that it would require monsters; I think the 2nd-weirdest part is that it would require constructions that are partial monsters, operating on character but sufficing with content in all and only the non-normative cases. The weirdest part is that it would turn out that every construction that can take a clausal complement would have to work this way, and coincidentally, they would have to all only need to work on characters for normative complements.
Mark, thanks for the explanation. Very nice. Things are quite a bit clearer to me now and I think this hybrid view is interesting. Actually, the idea that desire like attitudes of a speaker might determine reference fixing properties for moral terms is very intereresting to me. You’re raising one kind of worry, but here is another, given what you have said in your post and in your explanation to me.
Data (from Star Trek) could believe that lying is wrong. If the Hybrid view is right, then he could not believe that lying is wrong since he has no desires let’s assume, though he has beliefs. Thus, the hybrid view is wrong.
Yes, Christian, but I don’t think any proponents of hybrid theories are going to worry very much about your example; it looks like just an ordinary example of an amoralist, and hybrid views are motivated in order to explain judgment internalism of one sort or another. I take your point to be that what is false doesn’t need to be explained, with which I’m sympathetic, but it’s nice to be able to evaluate views on independent grounds.
M, I didn’t realize that “hybrid views are motivated in order to explain judgment internalism of one sort or another.” In any case, I do think the amoralist objection is a bit different than the one I’m suggesting. I think a person could, at least intuitively and superficially, mean what we mean by saying “lying is wrong” without having desires of a certain sort. This intuition is stronger than one which says a person must have similar desires in order to believe what we believe when we believe that lying is wrong. In my counterexample I said “believe”, but only meant “mean” and so I take that back (though I’m not an internalist).
“I take your point to be that what is false doesn’t need to be explained.” Oops, I didn’t mean to convey that. I just meant to give a counterexample to the hybrid view, from intuitions about meaning per se, rather than those concerning speech reports. The idea is that people could mean what we mean by ‘wrong’ though they do not have the same desires, in this case, any desires at all.
This point hangs quite a bit upon whether we are assessing ‘wrong’ in an idiolect or a language, but insofar as we have intuitions about sameness of meaning, I think these intuitions count against the hybrid view.
Hi Christian,
Thanks for your objection. I am a Star Trek fan, and I agree that Data has normative and even moral beliefs. Let me see whether I can’t accommodate this intuition about Data.
What is most essential to normative judgment on my view is having a will – being able to plan, and form intentions (intentions to follow the prescriptions of a given sort of prescriber, say). In that respect, my view is a bit like Gibbard’s.
I think it is plausible to suppose that Data does have a will, and can form intentions and plans, and act on them. Moreover, if you describe a being which is utterly incapable of forming intentions or planning in any interesting sense, then I am much less sure that it is pretheoretically plausible to say that this being is yet capable of making normative judgments.
Basically, I think your objection has more sting against classical emotivists, who put more weight on certain sorts of subjective feelings or emotions than my account does. Its more plausible to put more weight on executive states like intentions for this reason.
Incidentally, there are some contemporary theorists who bite the bullet here. Jesse Prinz just gave a talk in which he specifically argued with Andy Clark about Data, who on his view does not make moral judgments. This, though, is because Jesse does put a lot of weight on the emotions in particular when it comes to normative judgment.
Best,
Mike
Mike,
Ha! It seems that Data is a popular counterexampler…
I’m not sure what to say about your response. It seems reasonable enough to me. We would then need some different reference fixing description. Do you have any suggestions?
My other worry, and perhaps a more fundamental one, is that desires, beliefs etc. are contingent states. As such, they could exist without one another and so we could have normative beliefs, without the existence of any state distinct from it, including desires, a will, an intention (in some sense) and so on. What do you think about that kind of general worry.