File this under “meta-meta-ethics”
Don Loeb and Michael Gill currently defend a ‘variability thesis’, the view that ordinary moral thought and language contains both cognitivist and non-cognitivist elements.
As Gill puts it, in a recent paper, “there really are cognitivist aspects to our moral discourse, which the cognitivists have accurately analyzed, and … there really are non-cognitivist aspects, which the non-cognitivists have accurately analyzed.” Moral discourse contains a mix of these elements. The thesis can be expanded to other areas, internalism, and so on. An earlier proponent of a similar idea was W.D. Falk, in “Morality, Self, and Others”: some parts of moral practice are social; other parts are self-regarding. The advantage of the view is that it comports well with the mongrel historical heritage of our actual practices, and also explains why certain debates in moral theory are so intractable. One disagreement between Loeb and Gill is that though Gill denies, that the variability implies ‘incoherentism’ about ordinary moral thought. However, there are a range of possibilities I can see, and I wonder what Soupers might think of the idea, and the alternatives. (And I do not exhaust them here.)
i) Ordinary moral thought contains, in addition to its normative claims, its own ‘folk theory’ of itself, a folk metaethics. The alleged variability could be here: the character of first order normative claims is homogenous and invariable (say, all expressions of attitudes or all expressions of belief), but the folk metaethics is variable: It construes those first order normative claims in incompatible ways. About some areas, for whatever reason, what we ordinarily think we’re doing when we moralize is expressing attitudes; about others, we think we’re expressing beliefs. Philosophers have latched onto these two different aspects of folk metaethics, not seeing that there are, in fact, two different folk theories. Philosophers mistakenly that folk metaethics is invariable when it isn’t. This alternative, if true, doesn’t seem to be all that damaging to moral practice. Sure, folk metaethics is screwy. But so is folk metaphysics and folk epistemology. That’s why we all have jobs!
ii) Another position is that ordinary moral thought contains the view that “folk practice (or metaethics) is invariable”. It needn’t say *how* it’s invariable (all cognitivist, all noncognitivist, etc.) All that it needs to claim, in order to create incoherence, is that folk practice is variable, yet folk practice be mottled–either in first order terms or in its metaethics. There would be a problem explaining how such a mistaken thought might ever have arisen, at least about first order moral claims. How could it have come to be a part of ordinary moral thought that its practice is invariable when it isn’t? It seems that eventually, reality would intrude.
iii) A quite different possibiliy is that first order normative practice is itself variable. In some places it is (say) noncognitive, in others it is cognitivist. Various metaethical camps have each latched onto one of the parts and mistakenly tried to use this as a Procrustean bed in which to ram the rest of moral practice. This appears to be Gill’s view (it was, I think, Falk’s as well). A priorism among philosophers explains why philosophers have overlooked this, ignorance of what moral practice is really like.
iv) Suppose there’s a twofold variability: first order practice is in some places noncognitivist, others, cognitivist; and our folk metaethics likewise is cognitivist about some practices, noncognitist about others. If so, one might further expect these variabilities to line up, the fault lines to match up, more or less. One would expect it, that is, if one thinks that folk theory isn’t dumb. Here, the only ‘incoherence’ would be in metaethics itself, in the assumption Loeb and Gill challenge, of invariability. On the other hand, there might be a mismatch, and for good reasons: there may be pressures toward a folk metaethical (but false) assumption of the invariabile character of moral thought. It may be, that is, that this kind of incoherence is actually good–not in the first order normative thought, but in the folk metaethics. For instance, it might be good that what we think we’re doing when we’re thinking in moral terms is always the same sort of activity–independently of how that activity is cashed out (cognitivist/noncognitivist, etc.). A belief in a seamless practice will reinforce its apparent rationality–or at least might.
V) A final thought: Anil Gupta argues that incoherence in a concept is not incompatible with the truth of propositions that use that concept. This seems to be one way in which Loeb’s view may be challenged. Even if there is an incoherence in moral thought, it doesn’t impede the possibility of the truth of moral claims.
Sorry, but I don’t understand what the puzzle is that requires solving. You wrote:
As Gill puts it, in a recent paper, “there really are cognitivist aspects to our moral discourse, which the cognitivists have accurately analyzed, and … there really are non-cognitivist aspects, which the non-cognitivists have accurately analyzed.” Moral discourse contains a mix of these elements.
Isn’t this just to say that, for example, cognitivists are right that a sentence like “lying is wrong” expresses a proposition and that non-cognitivists are right that sometimes people say “lying is wrong” as a way of expressing their disapproval of lying and furthermore, we can both assert the proposition that lying is wrong and express our disapproval of lying at one and the same time? But is that news? Have cognitvists ever denied this?
More broadly, we use the assertion of propositions as a way of expressing feelings all the time – even in non-moral contexts. The happy person who yells “I won!” repeatedly is asserting a true proposition each time, but also is properly understood to be expressing her joy each time. (It’s not as if she thinks saying the same proposition repeatedly will give us new information.) So, of course, the non-cognitivist is right insofar as she says that sentences that look like propositions are often uttered in order to express feelings. But cognitivists have never denied this, so I don’t see what is new about this so-called “variability thesis” and why it raises any puzzle that needs to be solved.
Isn’t this *the Moral Problem* as introduced by Michael Smith in the Introduction to *The Moral Problem*?
I share some of David’s confusion: just what is the variability thesis saying that cognitivists didn’t already know and acknowledge?
But a couple of possibilities occur to me. Here’s one: maybe the thesis is that while some moral assertions aspire to truth, others don’t; so a cognitivist analysis is correct when applied to the first sort, but not when applied to the second sort.
Another possibility would be that, while every moral assertion can be evaluated for truth, the criteria for a successful or justified assertion differ from case to case: in some cases what makes a moral assertion successful or justified is the fact that it is true, while in other cases its truth is irrelevant to the evaluation of the assertion.
These would be interesting views (though I don’t as of yet see any reason for thinking that either of them are true!). Moreover the statement that “the non-cognitivists have accurately analyzed” the non-cognitive elements of our discourse at least suggests that this may be what the variability thesis means to be getting at. But this is pretty much speculative on my part — can anyone who has read the paper in question confirm or deny it?
Perhaps I’ve not been as clear as I could have been in expressing the idea. Without getting into the finer details of the debate, non-cognitivists deny what cognivists assert, viz., that moral claims or judgments express states that are,or at least are a lot like, beliefs –states with truth conditions. The variability thesis is the view that the territory over which cognitivists and noncognitivists are at odds — whatever comes under “moral claims or judgments” — is not homogenous. Some moral claims and judgments do express beliefs, others do not. The point of the post was to lay out a few alternatives about where the variation might be thought to lie. The view is that noncognitivists and noncognitivists share a false assumption, that moral claims and judgments are predominately one way or the other. That does seem to me at least to be an assumption. Perhaps it’s the right assumption. But it is worth looking at the alternative.
Here’s a worry. There has to be some criteria for when we correctly attribute a moral judgment to someone and when we would be incorrect to do so. Someone who just says that ‘torture is wrong’ hasn’t necessarily made the corresponding judgment. And, whenever we attribute the judgment to such a person it looks like we can be right or wrong about whether the person really has made the judgment.
The debates in moral psychology can be understood as disagreements over the norms of correct use of the term ‘moral judgment’. Externalist cognitivists would say that the term correctly applies to persons who are in a certain (true or false) belief state no matter what the motivations of the person is. Expressivists would say that the term correctly applies to persons in a certain conative state. Some internalist cognitivists think that the norm of correct use latches onto a certain sui generis besire kind of state which the agent must have in order for us to be able to correctly say that she has made a moral judgment. Ecumenists want to say that some combination of both desires and beliefs are required, and so on.
Now, the worry is that unless there are some ‘shapeful’ norms of correct attribution of moral judgments, the term ‘moral judgment’ is meaningless gibberish. These norms will have to give the concept some shape for guiding the application of it. If any mental state goes for moral judgment, then saying that someone has made a moral judgment has no content – no thought is communicated. This is a challenge for someone who accepts the variability thesis. It won’t just do to say that whatever, desires or beliefs or something, go for moral judgments. Something more has to be said about when it is correct to attribute someone with a real moral judgment.
Hi, Robert. I’m still not getting the bottom line. Is the worry that noncognitivists might be right about ‘wrong’ but wrong about ‘good’? Or is it that they might be right about Scandinavians but right about up-state New Yorkers? The former doesn’t sound that interesting, and the latter sounds bizarre.
Like Mark, I’m still not getting it. Let me try a question different from his to see if it helps: What does someone who holds the variability thesis believe that a cognitivist denies (or vice versa)?
Well, I wanted to nail down the thesis itself, because I think there are a number of interesting alternatives. Someone who would hold the variability thesis about (for instance) first-order normative claims would believe that some claims are as noncognitivist think they are (not truth evaluable), but some other moral claims are as cognitivists think they are (truth evaluable). Sometimes when you think or say “That’s not right” you’re expressing an attitude, other times you’re expressing a belief. Judging from the reactions so far, however, apparently this is either obviously true, something everyone already acknowledges and understands, and so is uninteresting, or it is not only not true, but incomprehensible. Noncognitivists are right about some uses of ‘wrong’ and ‘good’, but not about other uses of ‘wrong’ and ‘good’. Obvious? Or does no one understand?
Robert,
Your clarification helps. But I am not sure if the VT is obviously true or if it is obviously false. Let me explain.
When I say to someone “That’s the dumbest thing I have ever heard” I usually don’t literally mean it. Sometimes the reason I say it is to indicate that what has been said is quite dumb. Sometimes I say it just to insult the person I am talking to – to indicate my feelings about them. And sometimes I literally mean that it is the dumbest thing I have ever heard. But the fact that I use the same sentence in several different ways on different occasions does not seem to me to change the fact that it has a literal meaning which is the expression of a proposition which can be true or false. In fact, when I don’t use the sentence straightforwardly, someone could reply, “Is that really the dumbest thing you have ever heard?” To that I might reply, “Well, no, not really” or “Oh, you know what I mean”. This is because although I am not saying the sentence to assert the proposition it expresses, I do not deny that it does express that proposition.
So when I say “That’s not right” I am sometimes saying that to assert a proposition and sometimes I say it in order to express my feelings. But even in the latter cases I do not deny that the sentence does assert a proposition. It’s just that this fact is not relevant to the story of why I said it.
All of this is something that any cognitivist can say without any trouble. So if that is all the VT is saying, then it is not news and saying something that is obviously true. If it is saying that in the cases where I say “That’s not right” or “That’s the dumbest thing I have ever heard” for the purpose of expressing how I feel then the sentence does not actually assert a proposition, then the VT is not news and saying something that is obviously false. Either way, I don’t think the VT is news.
Maybe the title of this post better describes it’s content than the view it’s content attempts to describe. I’ve clearly mangled the exposition, since it’s clear to not just me that it is an alternative. But bear with me.
Noncognitivists have an intelligible position that is not obviously false. They deny that any moral judgments or claims express states (that is, the states characteristic of their being moral judgments) that have truth conditions. They all express states that lack truth conditions. Cognitivists, also holding a view that is not obviously false and intelligible, argue all moral judgments or claims express states (again, characteristically moral judgement states) that do indeed have truth conditions.This all needs refinement, no doubt. But we must be on the same page so far.
If so, the problem here must be with the next step. So let me try it again. This step takes issue with the ‘all’ part of both claims. It that says that, if you look at what people actually do –actual moral practice–you’ll see that sometimes their moral judgments express states without truth conditions, and other times their moral judgments express states with truth conditions.
Now, if that is the thesis ( and I’m not claiming that it is true, just that it’s worth considering), then it is neither obviously false nor old news. Couldn’t be either since it would require more observation of actual moral practice than philosophers have typically engaged in to determine whether it is true.
It may be that many think metaethical issue must be a priori, and I’m not necessarily denying that. But, at least, I don’t think that can be taken for granted. In any case, none of the issues about this possible position that I find interesting can get going if no one finds this intelligible.
Robert,
I don’t know if this is useful, but might it help to see things in modal terms, i.e., to consider whether cognitivists and non-cognitivists have wrongly supposed that their claims about moral judgments are necessary truths about such judgments? As Jussi notes, the debate between these camps can be seen as a debate about the logical, psychological, etc. properties of moral judgments. (Jussi expresses it in terms of norms concerning the correct use of the *term* ‘moral judgment; I think metatethicists aren’t investigating linguistic norms but the properties held by the specifically moral components of our psychology.) And one thing that makes the debate interesting is that many of these claims are put forth as modal claims, albeit often implicitly, so that expresivists hold that moral judgments necessarily express conative attitudes, etc. So is the variability thesis the denial that there are any (interesting?) necessary features of moral judgments? That could well be true, but we would still need some way to demarcate the class of moral judgments (as Jussi says). Likewise, David’s comments indicate that moral judgments cannot be identified simply by their expression in terms of ‘moral’ language since this either conflates the judgments with the utterance expressing it or is circular.
Robert,
Now that I think I understand the claim, I am fairly well convinced that it is old news. As I understand it, once upon a time everyone was a cognitivist and thought that all moral judgments or claims express states that have truth conditions. Then, about a century ago, the noncognitvists appeared. They pointed out (correctly) that people very often make moral utterances in an empassioned way, with a goal of persuasion, or in other ways that are much more like yelling “Ouch!” than saying “The sky is blue.” They concluded from this that no moral judgments or claims express states that have truth conditions.
Cognitivists have replied that people do use moral sentences to express feelings, but they do that with all kinds of sentences (like “That’s the dumbest thing I have ever heard” or even “I’m so happy!”) and those uses need not negate the fact that all moral judgments or claims also express states that have truth conditions.
Here’s some evidence that this is old news. In Mark Eli Kalderon’s book Moral Fictionalism he writes that in 1981’s After Virtue MacIntyre says that “even if emotivism provided the wrong account of the meaning of moral sentences, it might still provide the right account of their use.” Decades before this Charles Stevenson wrote, “my two patterns of analysis in Ethics and Language do not give an account of what ordinary people normally mean when they use ethical terms…. I am inclined to think of Ethics and Language as a ‘deliberately prescriptive proposal.'”
Nonconitivists have, from the very start, made too much of the fact that we can and do use moral sentences to express emotions. The MacIntyre insight shows that this fact can be accomodated by cognitivists without threatening the basic cognitivist claims. Stevenson’s admission shows that even noncognitivists can recognize that the emotivist claims go too far. The VT manoeuvre seems to be, at best, an attempt to preserve the noncognitvist’s overreach without giving everything back as Stevenson did. Either that or it is a proposal based on a straw man version of cognitivism that presents what cognitivists actually believe as if it were a new alternative.
I think some of us may still be talking past each other. In his most recent post, Robert says that the variability thesis “says that, if you look at what people actually do –actual moral practice–you’ll see that sometimes their moral judgments express states without truth conditions, and other times their moral judgments express states with truth conditions.”
Now, this could be true in an obvious sense, and I think this is how David is reading it. Sometimes the main purpose of a moral utterance is to express what is clearly a non-cognitive state. Thus, if a friend tells me she was the victim of some serious injustice and I say “What that person did to you was awful!” my main purpose is not to state a fact — she already knows that what the person did was awful — but rather to express my feelings, console her, etc. I take it this is agreed on all hands.
But although in this case the speaker’s primary purpose is to express a state of mind that is not fundamentally cognitive, there is still a cognitive state of mind (the belief that what the unjust person did was awful) that is also being expressed. Or at least, I as a cognitivist think that there is.
But perhaps this is what the variability thesis means to deny. So the thesis is not just that moral claims sometimes express non-cognitive states (obvious), or that sometimes the expression of the non-cognitive state is the primary purpose of the utterance (also obvious), but that at least on some occasions there is ONLY a non-cognitive state being expressed, and no cognitive state (or, state with truth conditions) AT ALL.
If this is the right reading, then I’m at least intrigued. Robert, can you give examples of the sorts of utterances that the variability theorist thinks do not express cognitive states? And is there any indication of a general way of distinguishing between the two sorts of utterances, or must this be done on a case by case basis?
Two points. First, it’s false that cognitivists necessarily think that moral sentences express beliefs – mental states with truth-conditions. ‘Express’ is a technical term that is part of the theoretical apparatus of expressivism, the dominant contemporary version of noncognitivism. It’s only part of the mythology of contemporary metaethics – a mythology encouraged by expressivists – that ordinary cognitivists think that moral sentences express mental states at all. I certainly don’t think that they do.
Second point: noncognitivism and cognitivism are supposed to be competing hypotheses about the semantics for moral sentences. So it can’t be that one is right sometimes and the other is right at other times, for one and the same word, with one and the same meaning. They’re competing hypotheses about the kind of meaning that that word has. That’s why it’s not a ‘hidden assumption’ to assume that what works for one ‘use’ must work for all. Since it’s a proposal about meaning, if it works for one use, it works for all of the other uses that mean the same thing.
So to say that cognitivism is sometimes right about ‘good’ and sometimes wrong about ‘good’ is just to say that ‘good’ is ambiguous, and cognitivism is right about some disambiguations, but not about others. On this view, ‘good’ is like ‘grand’ in ‘that’s grand’ and in ‘grandfather’. But that’s not exciting, either. Noncognitivists never aspired for their view to be correct about all words; just about some of them. Moreover, if this is the view, then it ought to be testable by ordinary tests for linguistic ambiguity.
Mark,
Hmm. I didn’t say or imply that cognitivists *necessarily* think that moral sentences express beliefs. Nor did I say or imply that noncognitivists aspired for their view to be correct about all words. And although the word ‘express’ has a technical meaning for non-cognitivists, it also has an ordinary meaning which I was relying on–though I admit, poor choice of words given the audience. I was deliberately trying to be as vague and non-committal as possible characterizing these views as addressing moral claims and judgments. I did that precisely because I wanted to avoid picking any fights over what these views are. I didn’t want to get responses that were defenses of some particular version of these views, or some particular view of what the meta-meta-ethical division between them. Indeed, the view that noncognitivism is a view about moral sentences is itself a particular meta-meta-ethical position about that view. Perhaps its the right one. But it is not universal. But pehaps there is no ecumenical or non-contentious way of stating what divides these camps.
The question of whether any of this is exiting or not depends (as I’ve tried to say) on where the alleged incoherence in moral practice is supposed to lie, and one’s taste for empirical work. This isn’t what I had in mind, but suppose for the moment that we discover that ‘good’ is like ‘grand’ in the way you say. But suppose also that we discover moral practice contains some supposition that what we’re doing when we say ‘good’ is always the same thing, one way or the other. We ordinarily think what we’re doing when we talk in moral terms is always the same sort of thing, but it isn’t. That would seem to make for a certain kind of incoherence in moral practice, wouldn’t it? Interests me, in any case.
One further thing. A lot of care would be needed to even state the hypotheses I was considering in a way that wouldn’t beg lots of questions. In particular, when a social scientist goes out to ask what we’re doing (and what we think we’re doing) when we make moral judgments or engage in moral discourse, and gets answers to that question, this all may not match up very well to the philosophical categories of ‘noncognitivist’ or ‘cognitivist’. So there’s admittedly a further question of their relevance to philosophical issues.
Troy,
I guess I see what’s bothering people now (seems so clear when it’s in one’s own head). I mentioned that the states over which I was taking the two camps to be differing over were which states it is that (lets say, roughly) make a judgment or claim a moral one. That is, I think, enough to imply that non-cogntivists claim there is no cognitive state that is the state that is characteristic of moral judgments and statements. So there is only a non-cognitive state being expressed (or whatever the hell we’re calling this relation now), and no cognitive state (or, state with truth conditions), that is the state that is the “moral-making” characteristic of moral judgments claims.
Since as I understand it, the thesis is a sort of contextualist empirical hypothesis, the expectation is that it will turn out that there are, as it were, superficially the same words, phrases, and the like, behaving in different ways in different contexts. It’s as if there are different standards for what is or is not a moral claim or judgment depending on the context. Certain claims will be moral claims or judgments in one context because they involve non-cogntive states, in other contexts, it is a cogntive state that make it a moral claim or judgement. So we will end up saying something of the form: this claim over here is a moral claim, but there is no cogntive state that was involved just a noncognitive one, and it is this last one that makes it a moral claim in that context. And we will also end up saying that another claim over there is also a moral claim, but does involve a cognitive state, and that cogntive state makes it a moral claim in that context.
I suspect that this view will have to hold that in contexts in which the claim is moral because it expresses a belief, the belief is false. A kind of partial error theory, perhaps.
Mark,
Sorry, I see you weren’t responding directly to me!
Michael,
(You can see I’m moving backwards up the thread. Maybe I’m wrong but I think Troy’s point responds to David’s worry.)
I guess I’d have to side with Mark about the modality issue. And, again, I didn’t want to wade into the meta-meta-ethics lanscape and pick any fights.
Hi Robert,
Does the following sound like I’ve captured the gist of the view?
Suppose we have a range of terms that fall within a certain domain of discourse D1 (say, the domain of terms that refer to acts of commission, for example) and another range of terms that fall within a different domain of discourse D2 (say, the domain of terms that refer to acts of omission, for example). Is this the view as applied to the sentence S, ‘A is right’?
(Understanding, of course, that acts of omission and acts of commission are just examples I made up to try to understand the view.)
Mark S. has captured nicely here just the point I was going to make next (though he’s done it better than I would have):
Mark,
You write,
Assuming this is correct, is it any more than an historical point? If contemporary metaethicists by and large say that cognitivism is the view that moral claims express beliefs and noncognitivism is the view that moral claims do not express beliefs, then for contemporary metaethics that’s what the terms mean, right? We’d only run into a problem if contemporary metaethicists used as paradigm cases of cognitivism views that deny that moral claims express beliefs. But does that happen?
I’m similarly puzzled by the passage that Dan approvingly cited. If conventional usage dictates that ‘noncognitivism’ names the view that moral claims never express beliefs, then when ‘good’ is used in a moral sense, it can’t express a belief, according to noncognitivism.
This is all just to say that if the “myth” has taken hold, then it’s no longer a myth, unless you’re a prescriptivist about labels like ‘noncognitivism.’ So the claim must be that on conventional usage ‘noncognitivism’ does not name the view that moral claims never express beliefs. My command of the metaethical field is less secure than either Dan’s or Mark’s, but my understanding was that it’s not unconventional to understand ‘noncognitivism’ in that way. Do those of us who read the terrain that way have an erroneous understanding of how terms like ‘cognitivism’ and ‘noncognitivism’ are conventionally used among people who make their living working on metaethics?
Hi Josh,
As I was understanding the view that Robert was describing, it holds that what we normally take to be moral predicates, e.g., the predicate ‘is right’, are not always moral predicates. Sometimes they are used sincerely and properly to make moral judgments, sometimes they are used sincerely and properly to make nonmoral judgments. So, on such a view, ‘right’ is semantically ambiguous; it has a moral sense and a nonmoral sense. But noncognitivists (and cognitivists, I take it) have never been interested in providing a semantics for any nonmoral sense that these predicates may have.
I just wrote, in response to Josh, “So, on such a view, ‘right’ is semantically ambiguous; it has a moral sense and a nonmoral sense.”
Just to avoid any confusion–Robert has already made it clear that the view he has been describing does not have this implication. Though the view holds that predicates such as ‘is right’ are, in fact, semantically ambiguous (sometimes they pick out a property, sometimes they do not), all judgments of the form ‘A is right’ are moral judgments.
Ah, right, Dan: I completely agree that terms like ‘right’ and ‘good’ have nonmoral senses. It sounds like we had different understandings of Robert’s original post. As I understood it, the novelty of the view he discusses was supposed to be that it diagnoses specifically moral discourse as containing both cognitive and noncognitive claims.
Dan,
I wonder if this is true:
‘But noncognitivists (and cognitivists, I take it) have never been interested in providing a semantics for any nonmoral sense that these predicates may have.’
For instance, I recall discussions from Blackburn about the cases where all kinds of agents make moral judgments without being motivated accordingly – amoralists, psychopaths, and so on. Seemingly these cases are problematic for externalists like Blackburn for whom moral judgments necessarily require having some motivation to act accordingly. Thus, the amoralists judgments do not count as moral judgments but rather only as ‘moral’ judgments. The semantics for this claims is relativist – these claims are made true by what is accepted in the community.
In addition, it is not clear that either party needs to give alternative semantics for the non-moral uses – pragmatics seem to suffice. For instance, if I was a teacher, I could use the claim ‘everyone will bring their hats tomorrow’ to command by students to wear a hat tomorrow. That sentence still has the same meaning and semantics but I can still use it not to make an assertion but a prescription. In the same way cognitivists and non-cognitivists can give semantic views and allow that moral sentences can be used for other purposes on some occasions.
Hi Jussi,
On your first point, let me just admit that what I wrote was too strong. Certainly, anyone, including cognitivists and noncognitivists, may be interested in giving the semantic story for some nonmoral sense of ‘right’.
About this:
Unless this sentence is ambiguous, I don’t see any way in which this sentence could be used sincerely and properly without the speaker performing a direct assertive illocutionary act. Of course, it can be used sincerely and properly to perform a direct assertive illocutionary act and an indirect prescriptive (directive) illocutionary act. I take it that the semantics of a sentence is supposed to, in some relevant sense, “track” the direct illocutionary acts the sentence is conventionally used to perform.
I’m not sure about that. If I used that sentence to make a command and not an assertion in the situation (I don’t really present the students coming to class tomorrow with hats as true – a hallmark of assertions), I don’t see what insincere or improper there is in my utterance.
I agree with the last claim you make. But, I thought that on occasion the speaker-meaning and the sentence-meaning can come apart. That is, what the speaker intends to communicate and what she is understood to communicate can differ from what the sentence typically would be taken to communicate by default. I was just wondering if cognitivism and non-cognitivism are semantic accounts of the sentence-meaning, whether there could be occasions, as the variability thesis would predict, where the speaker-meaning is something different. However, even if we accept this, I thought we wouldn’t need another semantic account to account for the speaker-meaning in these cases but rather do work in the pragmatics – what other thoughts besides the sentence meaning the speaker is able to communicate with the moral sentence given the conventions of use there exists. This might be interesting for some but it wouldn’t come into conflict with cognitivism and non-cognitivism as semantic accounts.
Josh – I take it that there are noncognitivists, and then there are people who are not noncognitivists. Those people are called ‘cognitivists’. Contemporary noncognitivists of the expressivist variety use a word, ‘express’, in order to say what their theory is. They say that we can understand the meaning of moral sentences by understanding that they express desire-like states of mind.
Cognitivists disagree with this. They think that you can’t understand the meaning of moral sentences by seeing that they express desire-like states of mind. It doesn’t follow that they think you can understand the meaning of moral sentences by seeing that they express beliefs, or even that they think moral sentences express beliefs at all.
I called the idea that everyone agrees that moral sentences express some kind of state of mind a ‘myth’, because it is something that expressivists would like everyone to believe, in order to make it sound like their semantic project is less radical than it really is. According to Gibbard, for example, everyone thinks that moral sentences express judgments, and all we disagree about is what kind of judgment they express – belief-like states of mind or desire-like states of mind.
You say: ‘if the “myth” has taken hold, then it’s no longer a myth’. Really? The myth is part of a sales pitch for expressivism. It’s part of expressivists trying to get away with stipulating that everyone else agrees with more of their view than they really do. Gibbard doesn’t get to stipulate what I believe. I’m a cognitivist, and there’s no sense of ‘express’ that does the work that expressivists need it to do in their theory and on which I think moral sentences express beliefs. I’ve explained this here at length.
You’re right, that every standard ‘flow-chart’ of metaethical views starts by assuming that everyone thinks that moral sentences express some kind of state of mind, and asking which – then you move down the flowchart in different ways for the cognitivist and non-cognitivist answers. That’s just to say that the myth is well-entrenched.
Hi Jussi,
You say you agree with this claim,
If so, then it’s not clear to me how you could consistently hold this claim:
Hi, Mark. You write:
Just a couple of thoughts. To say that if the myth has taken hold, or is well-entrenched, then it is no longer a myth, is simply to make what I think of as a relatively uncontroversial presupposition about the meaning of terms like ‘noncognitivism.’ This presupposition is that the meanings of terms (especially, but not exclusively, terms of art) is constrained by either ordinary usage or by the usage of the relevant group of experts. (Note that this is not to say that the meaning is exhausted by ordinary or expert usage). With a term of art like ‘noncognitivism,’ its meaning is set by the relevant group of experts, in this case those who make their living working on metaethics. So if (as your flowchart comment seems to suggest) it’s standard for people working on metaethics to carve the non/cognitivism distinction as a distinction about what states of mind moral claims express, then that sets the meaning of ‘cognitivism’ and ‘noncognitivism’.
Now there might be an argument to be made that this (a certain claim about what mental states moral claims express) is not what ‘cognitivism’ has always referred to, but that’s the historical point, and is consistent with the idea that ‘cognitivism’ now does refer to that (a la Kripke’s Madagascar example). And there might be an argument to be made that it would be more profitable if we were to use ‘cognitivism’ to refer to something other than what we now use it to refer to. But that too is consistent with the idea that ‘cognitivism’ means (now) that moral claims express beliefs (or at least non-desires).
Just a final point. This is simply about how terms of art are used. Gibbard can’t stipulate what you think, but he can stipulate how he’s going to use his terms. I’ve been suggesting that if his kind of stipulations have caught on, then they’re no longer mere stipulations. But even if you don’t like that, I take it from your comments that you would agree that a not insignificant part of the metaethical debate has taken place between rival positions on what states of mind we express when making moral claims. Whatever we call those rival positions, those discussions seem to often presuppose that moral claims all express something like either desires/emotions or beliefs. It’s on the assumption that such a presupposition is not very uncommon that I found the position Robert discusses more novel than have some others in this thread.
Mark,
Not that you should divulge any more of your own views (published or forthcoming) than you wish, but could you say a little more about how the ‘myth’ seems to bias the debate in the expressivists’ favor? (As you can gather form my earlier comment, I apparently succumbed to the myth quite unwittingly!) How, for instance, does the myth make for rough sledding for cognitivists?
I’ve certainly skirted close to saying what Mark S objects to in some of what I’ve written, but I think there are two points to be made against letting what most people or experts say just settle such matters.
(1)It is a notorious problem of philosophical taxonomy that it is nearly impossible to divide logical space in two unless you define one region as the cmplement of the other. If you start with non-cognitivism as a view which says that the meaning of a moral term is to function to express a non-cognitive state of mind, then you’d better define cognitivism as just the denial of that claim. If you say anything more substantive you’ll wind up leaving room for a view which counts as neither cognitivist or not. If you do that you’ll wind up leaving those like Mark who deny that the terms function to express belief (in the relevant sense of express) in the position of being neither cognitivist nor non-cognitivist. With some ways of dividing up terrain that’s not so bad, but with these terms it looks like they ought to divide the terrain with no remainder. (Why define cognitivism in terms of non-cognitivism? Think about who first explained their position by talking about expression.)
(2) Josh’s point’s that the use by the philosophical population and by experts help determine the meanings of terms is in one sense unobjectionable, but when read as requiring that such terms must comport with popular or expert understandings of their meanings it is being taken too far. Experts and non-experts alike can be mistaken about how the alleged designatum of a term has to function to be the semantic value of the term, even if expert or conventional use of the term plays a role in making this so. (Importing a favorite example, atoms were divisible despite what Dalton thought.)
More specifically about ‘express’, it may be fine in many contexts to say that indicative moral utterances are apt for the expression of belief, or something in that ballpark. But that is a different claim than the claim that they express beliefs in the very same sense of ‘express’ that the non-cognitivist needs to be using when s/he says that we can understand what normative language is doing by noticing that if functions to to express non-cognitive states of mind. And as I read Mark’s overall point here, he thinks it is false that cognitivists should think that moral judgements express beliefs using the very same sense of ‘express’ that the non-cognitivist should employ when she says they express non-cognitivist states in the course of trying to offer an account of the meanings of moral terms. So as I take Mark’s point he’s saying that if non-cognitivists are to say anything helpful when they tell us that moral judgements express non-cognitive attitudes, they had better not also go on to say that cognitivists think they express beliefs in the very same sense of the word.
Obviously, this still leaves it to Mark to fill in the reasons why the same term cannot be used to say both things at once without homonymy. But pointing us to his paper on the topic is a fair start on that even though I agree with Michael C that the discussion here would benefit from a short summary of the main points.
Robert,
Can we get a reference to the Loeb and Gill papers regarding these issues. I remember a Gill APA session on something like this issue, but my memory is too foggy to do me much good in reconstructing the position.
The papers are in press.
Mark vR,
I agree with most of what you said, in particular that the non-cognitivism/cognitivism distinction should be characterized exhaustively and that we can mistakenly represent the meanings of our terms. So I’m not sure that I would have to disagree with your points in order to disagree with Mark S’s earlier comment. (Not that you said I would, but I just wanted to be clear that what I said was meant to be consistent with those claims.)
Like you, I normally think of the mistaken representation phenomenon as best shown with the case of ‘atom’ and other standard illustrations. This is what motivated me to add the qualifier that meaning is not exhausted by usage. But often those cases are ones in which the world partly determines the meanings of our terms (‘atom,’ ‘water,’ and so on).
Just out of curiosity, do you think this world-dictated dimension of meaning is likely to impact names for categories of philosophical views, as opposed to natural kind terms? With ‘atom’ scientists could say, “Oh, we thought the atom wasn’t divisible, but it turns out those things we’ve been calling atoms are divisible.” Or for another case, we can say that we didn’t know that H2O was essential to water, but now we do. But how likely is it that with non-natural kind terms, such as ‘non-cognitivism,’ we will say, “Oh, we thought non-cognitivist views denied that moral judgments express beliefs, but it turns out that they don’t deny this,” while still meaning the same thing by ‘non-cognitivism’ as we do now?
On that point, you characterize Mark S’s position as follows: “it is false that cognitivists should think that moral judgements express beliefs using the very same sense of ‘express’ that the non-cognitivist should employ when she says they express non-cognitivist states in the course of trying to offer an account of the meanings of moral terms.” My initial reaction to this (I haven’t read Mark S’s paper yet) would be that, assuming (!) that mainstream usage — flowcharts, etc. — says that cognitivism holds that moral claims express beliefs, and assuming (!) that in mainstream, expert metaethical discourse ‘express’ is being used in the same sense as is used when non-cognitivists talk about expressing non-cognitive attitudes, then Mark S is using ‘cognitivism’ in an idiosyncratic way. Not that there’s anything wrong with that, of course.
Mark vR has me correctly, so I’m just going to repeat the highlights and then fill in a couple more details, as requested.
1) The experts agree that the cognitivism/non-cognitivism distinction is supposed to be exhaustive, and that Moore is a paradigm cognitivist.
2) Contemporary noncognitivists think that moral sentences bear a certain relation – which they call the ‘expression’ relation – to desire-like states of mind.
3) Charity requires the assumption that this relation be understood in such a way that it can do the work that contemporary noncognitivist theories require of it. After all, if anyone gets to have a say about what ‘express’ means, it is the people who advocate theories that make use of this notion.
4) There is no relation that can do the worth that contemporary noncognitivists require of the expression relation, such that Moore believed that moral sentences stand in that relation to beliefs.
5) So if cognitivism is the view that moral sentences express beliefs in the same sense that contemporary noncognitivists hold that moral sentences express desire-like states of mind (call this Characterization C), then Moore was not a cognitivist.
6) Since Moore was not a noncognitivist, either, it follows that if Characterization C is right of cognitivism, then the cognitivism/noncognitivism distinction is not exhaustive.
7) So it follows from either 5 or 6 (pick one) that Characterization C is wrong of cognitivism, no matter what metaethicists say – and because of the very commitments of metaethicists to be using ‘cognitivism’ in such a way that it exhausts the terrain along with ‘noncognitivism’ and in such a way that Moore counts.
This argument obviously relies on premise 4; I’ll say something further about that shortly. The point is that I think people have been (relatively) sloppy, and accepted the way of framing the debate that has been offered by contemporary noncognitivists.
Josh says that Gibbard can’t stipulate what I think, but he can stipulate how he is going to use his terms (namely, ‘cognitivism’). But Gibbard has tried to sell the idea that ‘that words express judgments, will be accepted by almost everyone’ (p 84 of Wise Choices, I believe). He and Blackburn have both worked hard at selling this presupposition, which is precisely what people need to be making, in order to think that Characterization C can be true of cogntivism, but the distinction between cognitivists and noncognitivists can still be exhaustive and include Moore on the cognitivist side. Gibbard and Blackburn tried to sell that picture because it gave people a way of thinking about expressivism on which it sounded like a less radical idea than it really is – like it was only a view in the philosophy of mind, and not a totally radical idea about semantics. And people accepted the idea partly because it was pushed by Michael Smith as an alternative way of understanding what is at stake between cognitivists and noncognitivists, if the issue is not about whether moral sentences are truth-apt. So you put Blackburn and Gibbard with their agenda together with Smith, who was trying to construct a new way of saying what was at stake and accepted this idea, and that gives you the contemporary orthodoxy. That’s my rough take on it, anyway – I wasn’t around when it happened.
Michael C asks: well, why does framing the debate in this way help expressivists? The answer is that it paints the picture that expressivism is really just a view about moral thought. We all think that sentences express mental states, and only disagree about which kind, after all. But that’s wrong. Pure expressivism – the sort of view advocated by Blackburn, Gibbard, and Horgan and Timmons – has a totally radical semantic project. Though it may still be possible to say things in a language with an expressivist semantics such as ‘moral sentences have truth-conditions’, an expressivist semantics does not work by assigning sentences to truth-conditions, or to anything else that determines truth-conditions. An expressivist semantics works by assigning sentences to mental states. That’s a radical idea in semantics whose radicality goes almost completely unremarked by Blackburn and Gibbard. Why? They’re trying to make their view easy to think about. They’re trying to give people a picture of the field on which their view occupies a natural and central place.
So what do I think the expression relation is, that leads to all of this fuss? Why isn’t whatever will do the work for expressivism something that Moore thought that moral sentences bear to beliefs? Well, to answer that, you have to go through different candidates for how you might understand the expression relation. For example, you might think it is causal, following Ayer’s references to ‘ejaculations’. You might think it is an illocutionary act kind – following Dan Boisvert, and Terence Cuneo in volume one of OSME. You might follow Gibbard’s Gricean idea from Wise Choices and think that it is a matter of what mental state you are intentionally trying to communicate to your audience that you are in. I don’t think any of those – not even Gibbard’s – does the work that expressivists require of the expression relation. (And Moore certainly didn’t have Gibbard’s Gricean view of assertion, so if that’s what expression is, then Moore wasn’t a Characterization C cognitivist, anyway.)
I think that if you want to make expressivism work, you have to get over your hangups of thinking that ‘express’ is something of which you have a pretheoretical understanding, and accept that for a sentence to express a mental state is simply for it to be the mental state that is assigned to that sentence by the semantics. Similarly, when ordinary semanticists say that sentences express propositions or truth-conditions, they’re not talking about any theory-antecedent sense of ‘express’; ‘express’ is just a convenient shorthand for ‘has as its semantic value’ or ‘is assigned by the semantics to’. I don’t think that it is the job of semantics to assign sentences to mental states, so I don’t think that any sentences express beliefs, in this sense.
Are there other senses in which I think moral sentences express beliefs? Sure – but I don’t think you’ll get very far if you try and construct an expressivist theory on their basis. I think you’ll get the wrong results about the meanings of insincere utterances, and that you’ll have problems making sense of how your compositional semantics works – for example.
Okay, now I’m starting to see the appeal of Mark S’s view. (I only wish you’d said so sooner!) To (ahem) quote my earlier post, “If contemporary metaethicists by and large say that cognitivism is the view that moral claims express beliefs and noncognitivism is the view that moral claims do not express beliefs, then for contemporary metaethics that’s what the terms mean, right? We’d only run into a problem if contemporary metaethicists used as paradigm cases of cognitivism views that deny that moral claims express beliefs.”
A paradigm case of X counts as much as anything as evidence for the meaning of ‘X,’ so (assuming 4) I’m happy to get on board with the claim that ‘cognitivism’ does not unambiguously mean the view that moral claims express beliefs.
(Although, I’m curious what Mark S would say to the following kind of response. A deeply conceptually embedded part of expert usage — the flowcharts, etc. — says that cognitivism holds that moral claims express beliefs. Another deeply embedded part of expert usage — the identification of Moore’s view as a paradigm case of cognitivism — is inconsistent with cognitivism holding that moral claims express beliefs. Granting the inconsistency, why should we pick one of these over others?)
Josh – I don’t think that Characterization C of cognitivism is ‘deeply embedded’. It’s clear that people came to accept it because of a faulty inference – the one that I pointed out.
More explicitly, here’s why not to take the ‘flow chart’ as authoritative. The whole point of having a flow chart is to divide up the logical space. So a flow chart whose first question has a false presupposition is a faulty flow chart. The standard flow chart has a first question with a false presupposition. So the standard flow chart is faulty and can’t do its job anyway.
One alternative would be to take the flow chart as authoritative and conclude that Moore doesn’t even occupy a place on the contemporary palate of metaethical positions. That’s crazy – most people see the last century of metaethics as a reaction to Moore.
The other alternative is to admit that philosophers make mistakes sometimes – which I think ought to be pretty far from being surprising, given the history of the discipline.
Keep in mind, definitions – like the definition of ‘cognitivism’ – don’t come from thin air. Smart but fallible people propose them because they’re trying to capture something with a certain kind of philosophical importance. They can be right about its philosophical importance while being wrong about whether they’ve succeeded in capturing it with their definition. So then what should we go by? Well, if you’re evaluating the details of a particular argument whose author defined the terms of her conclusion clearly, then charity requires noticing that she showed what she said she would, even if that isn’t so philosophically interesting. But if you’re trying to divide up the philosophically interesting positions, you have to admit that people can have been wrong about how to describe them.
I’ll make one last stab at this, since I now see that what has made the variability thesis look incoherent in the minds of some of you is also the perspective that makes Blackburn and Gibbard’s project look like some sort of philosophical chicanery. I have no light to shed on ‘express’ and I think Mark S is right that there is a problem with whatever that is supposed to mean.
As I see it, expressivists ask us to be willing to entertain, not merely the semantic pressupostion Mark S alludes to (that words express judgments), but a somewhat different question than the one traditional metaethics asks. One can approach ethics as philosophy has traditionally approached it, by trying to give the right analysis of moral language. If these are the questions the expressivist proposals are meant to answer, they do indeed look radical and difficult to accept. It also makes the idea that, say, what counts as a moral judgment might vary between between a belief and a desire look incoherent. If one is more naturalistically minded, however, as expressivists are, then one’s primary questions are not about the analysis of moral language. They are about how to explain that area of human behavior that falls under the heading “morality”. It’s much like the questions a social scientist would ask, except that social scientists aren’t out also to make it clear how this thought and behavior is ok from the perspective of the sciences. Of course you might refuse to countenance this explanatory project as even a philosophical one, I suppose. Or you might think it’s changing the subject. But if you are willing to entertain the question of how, in a very general way (not specifically psychological or sociological ways), to explain all this stuff we do and say, then the claims of expressivism are not radical, though admittedly the approach itself is. Their proposal just isn’t an analysis of moral terms. And, I think, the same approach — the approach that is looking for an explanation that will fit in with the rest of what science tells us about the world — makes the possibility that these explanations of moral talk and behavior might go in different ways in different contexts, as the variability thesis claims, a live option.
I should say that this is a sympathetic rendering of the expressivist approach. It’s sympathetic because they often seem to be trying to turn the analytical issues into empirical, explanatory ones (as some of what Blackburn and Gibbard have to say about the Geach problem shows), and I’m not sure that makes sense. But I think the most promising project, the one that seems to have a shot at succeeding, is the more naturalistically motivated project.
Josh,
You ask:
“Just out of curiosity, do you think this world-dictated dimension of meaning is likely to impact names for categories of philosophical views, as opposed to natural kind terms?”
Actually I do because I think that in philosophy we aim to pick out theoretically fruitful divisions with our words. And we can be mistaken about how what the term designates needs to work to do the theoretical work it needs to do. So we introduce a term to capture an important distinction, while missing how to draw the line properly. This is so even if philosophy is a partly a priori discipline since we are not a priori omniscient, let alone omniscient about the empirical stuff that helps make certain a priori issues interesting and others not so much.
But, I think that this might better be an issue for another thread. Robert’s posting immediately above suggest another way in to understanding what Loeb and Gill are up to and I don’t want to drag us off that track.
Hi, Robert.
I agree with you that there are various strands to the noncognitivist tradition. Some philosophers who get classified as being part of the noncognitivist tradition were really asking ‘what do we do with moral language?’, rather than really trying to offer a different kind of semantics for moral language. For example, I might put Paul Edwards in this category. But Blackburn, Gibbard, and Horgan and Timmons definitely don’t go into it. Ayer doesn’t go into it, Stevenson doesn’t really go into it (although he’s more complicated), Hare and Smart don’t go into it, and Carnap doesn’t go into it.
I also agree with you that empirical questions about how we use moral language and what goes on in our heads when we do moral thinking are interesting. But a couple of things. First, they don’t address many of the questions noncognitivism is supposed to address – for example, can we do without an ontology of moral properties? If there are moral properties, then how do we manage to talk about them or know about them? Why does thinking that something has a certain property have to motivate you to do anything? Why does there seem to be an ‘is-ought’ gap? Second, they cross-cut ordinary cognitivist views. As I think David White made admirably clear, ordinary cognitivists can agree that we use moral language to do all sorts of things – those things just aren’t what give moral language its distinctive meaning.
And third, I don’t think they tell us anything direct about pretty much any of the traditional questions in ethical theory. Suppose, for example, that empirical science discovers for us that moral thinking makes use of the emotions – as suggested by research done by Haidt, Greene, and others. What kinds of moral theory does that rule out? It doesn’t rule out Aristotle – he thinks moral judgments are mediated by the emotions. Does it rule out Kant? Well, no – if you look at the Doctrine of Virtue, it’s compatible with the view that emotions do important work, isn’t it? Is it inconsistent with Moore? Well, Moore doesn’t say how intuition works, so it doesn’t contradict his view, either. Hume? Butler? Obviously not. So it’s a great thing to know, and an important constraint on a theory – maybe it rules out Price or Clarke or something – but not ultimately so exciting as people like to make out.
I agree, of course, that expressivists invite us to ask a somewhat different question than is asked by others – that they are not ‘providing an analysis of moral terms’. But this isn’t because they’re not interested in or committed to a semantics for moral language, but rather because they have a different idea of how to go about it. According to expressivists, you go about giving a semantics for a sentence by saying what mental state it expresses, rather than by saying what makes it true – and that’s why this is a way of making sense of moral language that doesn’t involve the same ontological commitments, and hence doesn’t involve the same semantic or epistemological puzzles, either. If someone calls herself an expressivist and doesn’t think that she has any story about the kind of meaning that moral terms have, then it’s not clear that she is asserting anything that ordinary cognitivist realists deny, and she certainly hasn’t done anything to mitigate her ontological commitments.
Finally, you seem to be suggesting that an empirical or naturalistic research project contrasts with the project of trying to understand what kind of meaning moral sentences have. But look – if linguistic semantics is an empirical discipline, then understanding the semantics of moral language is an empirical project. It doesn’t follow from the fact that Blackburn and Gibbard are substantive or methodological naturalists that they’re not committed to views about how moral language works. My own view is that expressivism is empirically false – it is simply a demonstrably bad account of the data, failing to account for the empirical facts about how many sentential constructions behave semantically.
The dilemma for the incoherentist was supposed to be this: if cognitivism and noncognitivism are competing hypotheses about the meaning of moral terms, then they can’t both be right about a single term with one and the same meaning. So the only way that they can be sometimes right and sometimes wrong, is if they’re some other kind of thesis. But that other kind of thesis, though interesting, won’t solve any of the kinds of problems that noncognitivism is supposed to solve, which it solves because it is a hypothesis about a kind of non-truth-conditional meaning that moral sentences have.
Mark,
I don’t dispute that Blackburn, Gibbard, and Horgan and Timmons for that matter, don’t also offer, or at least appear to be offering, analyses of moral langage. But where you categorize them is at odds with much — though of course not all — of what they themselves say they are doing and the nature of the proposals they offer. What they repeatly *say* they are doing is trying to “place” ethics within the natural world. They *say* that analyzing moral language isn’t the only theoretical project, that we need an explanation of what we’re doing when we think in moral terms. And they even go so far as to offer sociobiological just so stories about the origins of ethical thinking.
You may be right that these kinds of theses, if true, won’t “solve any of the kinds of problems that noncogntivism is supposed to solve”, if those problems are the traditional problems about the analysis of moral language. But that is the kind of theses they claim to be offering. So I expect that this is because they’re primarily interested in solving other problems. They may not be consistent about this in practice. But a less uncharitable approach would evaluate their proposals in their own terms, not solely in terms of how well the answer the issues having to do with the analysis of moral language. Gibbard and Blackburn sometime seem to try to change those traditional problems into different problems, problems their views might be able to solve. That’s why the program as I’ve explained it may well seem irrelevant to the traditional philosophical questions about moral language.
Linguistics is an empirical project. But how well does what linguists say about moral language comports with what traditional metaethicists say about it? I suspect there’s a very limited amount of research that would even begin to address the kinds of questions philosophers might be interested in regarding ethical language. In any event, whatever they have come up with, has metaethics really been guided by empirical findings on these issues? I’m not qualified to answer that question, but perhaps someone out there is.
I just noticed that some of you have been talking about moral incoherentism, and thought some of you might be interested in reading the paper. You should be able to find it, along with my reply to comments by Gill and Sayre-McCord at my public site:
http://www.uvm.edu/~dloeb/
I hope that works!
Sorry, I forgot tell you. The original paper is listed as “Semantic Hat III” and the reply as “LoebV2C7reply.doc”.
(If I’ve done things right, Gill and Sayre-McCord’s comments are unavailable until they grant permission for me to share them.)
Robert – I apologize for this point getting out of control, but I don’t know why you think I’m being uncharitable to Blackburn and Gibbard. I agreed that they’re not analyzing moral terms in the ordinary way – they’re not telling us what it is to be wrong. They think that you can’t give a semantics for moral terms in that way. They think that to understand what moral language does, you have to do something else – to understand the mental states it is used to express. So far, so agreed.
Their hypothesis about the nature of moral language, is that you understand what ‘P’ means by understanding what it is to think that P. So yes, most of the work they do is all about trying to understand moral thought – not explicitly about moral language. In fact, Blackburn says extremely little about moral language, and Gibbard spends something like less than a chapter per book on it.
Yes, they both want to understand both moral thought and moral language in a naturalistic way, and without committing to any analysis or reduction of moral properties. This means that they’re dedicated to not understanding moral thought as representing the world as being a certain way, morally, and not understanding moral language as having moral properties for its semantic values.
A story about moral thought alone only carries out half of this naturalistic project. If you make sense of moral thought in a way that doesn’t commit to moral properties, but your semantics still requires them, then you’re not off of the hook. Blackburn and Gibbard want to be off of the hook. So it’s a central part of their naturalistic project to be offering an account of the kind of meaning that moral sentences have that does not appeal to moral properties. It isn’t traditional philosophical analysis, because it’s not a traditional kind of semantics. But it’s still a view about semantics – simply in a non-standard way.
This shouldn’t be a surprise. They (particularly Gibbard) call their view ‘expressivism’, and expression is a relation that holds between sentences or utterances or speech acts and mental states. If you only had a view about moral thought, expression would never come into it. The fact that ‘expressivism’ is the name for the view is evidence that the view has something to do with language. So is the fact that Gibbard tells us explicitly that the project of expressivism is to understand the meanings of sentences in terms of the mental states they are used to express. I’m assuming that charity requires taking Gibbard at his word, when he says what his view is supposed to be doing.
You also make it sound like I’m claiming that we’re to evaluate Blackburn and Gibbard’s views on the basis of old-fashioned metaethical questions that they’re not interested in. I’m not. I’m simply claiming that we’re to evaluate them on the basis of the metaethical questions that Blackburn and Gibbard themselves say that they are interested in. They both say that they want to understand moral language and moral thought without letting moral properties in ‘on the ground floor’. That’s part of their substantive naturalism. You can’t do that if you don’t have a semantics that allows you to avoid appeal to moral properties.
In response to Robert’s question about linguistics: how well does linguistics connect up with metaethics? Well, it hasn’t had a very large impact. That’s because most metaethicists aren’t, in fact, very well trained in linguistics, and in fact often don’t have lots of up-to-date background in the philosophy of language generally. Does that mean there is no direct connection? Of course not – many claims made by metaethicists are claims about how language works. Those are evaluable in part in light of the whole of linguistic evidence. The fact that metaethicists haven’t paid enough attention to linguistics isn’t evidence that they shouldn’t.
As I said above, I myself think that expressivism fails substantively – not in principle or a priori – to be an adequate account of natural languages. I show in Being For, for example, that expressivism could be correct about very simple languages with ordinary logical properties and the expressive power of the predicate calculus, but that the very tools that make it possible to develop an expressivist semantics that allows for this preclude an adequate treatment of tense, modals, and binary quantifiers like ‘most’. I take it that this is part of an a posteriori argument based on observations about the properties of natural language sentences that expressivism is an inadequate theory.
But the payoff of interactions between metaethics and linguistics and the philosophy of language has the potential to go both ways: topics like the semantics of mood and in particular of mixed-mood sentences are difficult issues in linguistics and the philosophy of language, and as Hare and Smart both illustrated and as Dan Boisvert has been showing in a number of threads, is closely connected to issues about expressivism. Expressivism is also a strong candidate for the semantics of epistemic modals, another difficult topic in both linguistics and the philosophy of language, with broad implications all across philosophy, given that it has also been used to motivate relativist theories of truth. Stephen Barker’s work shows that expressivism may be related to complicated questions about quantifiers in natural languages – he advocates an expressivist treatment of quantifiers, disjunction, and negation, and claims that it solves difficult problems about ‘donkey’ anaphora. The treatment that he gives also shows how a naive truth-making theory can be made to work without allowing negative or disjunctive facts into one’s ontology. Allan Gibbard and Jamie Dreier’s work on conditionals suggest that there is a lot that is fruitful to be gained by looking at connections with expressivism. And work on ‘hybrid’ expressivist/cognitivist views by David Copp, Daniel Boisvert, my colleague Stephen Finlay, Michael Ridge, and others shows that issues in metaethics are closely connected with the complicated question of how to understand the semantic properties of pejoratives. And getting away from expressivism and related views, recent work by John Broome and Stephen Finlay shows how the way we think about ‘ought’ can be and needs to be framed by an understanding of syntax. That’s just to name a few examples of things that I think are particularly exciting issues right now on the border of metaethics, linguistics, and the philosophy of language more generally.
Hi Mark,
I’ve only got two last thing to say here.
First, I think challenges to B & G’s work such as you’re offering are certainly important, but they’re important because they force them to take a stand on what it is they’re doing. Still, if you’re evaluating their proposals simply as analyses of moral sentences, you are indeed evaluating them on the basis of an old-fashioned metaethical project that they do, very clearly, want to move away from. You’re not really looking at the big picture, the totality of moral talk and behavior overall.
Second, charity. You say:
“I called the idea that everyone agrees that moral sentences express some kind of state of mind a ‘myth’, because it is something that expressivists would like everyone to believe, in order to make it sound like their semantic project is less radical than it really is.”
Does that really seem like a charitible interpretation? The claim that they’re trying to sell some idea of ‘expression’ in order to win the argument doesn’t seem so charitible to me. They do indeed try to change the playing field to one more amenable to their view. But that’s no objection to their views.
Let me just finish by trying to say what, as it appears to me, Robert and I appear to be in disagreement about. I take it that Expressivism is a view both about how moral language works, and about what moral thought is like. Call expressivism’s commitments about moral language ‘L’ and its commitments about moral thought ‘T’. So expressivism is committed to L&T.
Robert and I agree that Blackburn and Gibbard have done a lot more work to elaborate on T than to elaborate on L. (I can explain why.) That’s great – T is really interesting. Robert suggests that what I’m doing in my work that tries to evaluate L is ‘important’, but isn’t ‘really looking at the big picture’.
Here’s why that sounds like a non-sequitur, from where I’m coming from, at least: if expressivism is committed to L&T, and L is false, then it doesn’t matter whether T is true – expressivism is still false. That’s the beauty of conjunctions; if one conjunct is false, then the whole thing is false. So while Robert thinks I’m failing to grasp the big picture, I think I’m grasping it correctly, observing that it is a conjunctive picture, and observing that conjunctions are false if one of their conjuncts is.
Robert sees things differently. He either thinks that expressivism is not really committed to L, but only to T, or else that the important part of expressivism is just the commitment to T, and Blackburn and Gibbard really, ‘very clearly, want to move away from’ their commitment to L – and hence that I’m either concerned about something that is not really part of expressivism at all, or else something that really shouldn’t be, once we simply read Blackburn and Gibbard charitably.
That’s an interesting position to take; I had an argument against it, above. Blackburn and Gibbard’s view is called ‘expressivism’, so I assume that expression must play an important role in an important part of their view. But it doesn’t play any role in T – only in L. So I infer that L is an important part of their view. This is corroborated by their direct statements – Gibbard is quite clear about this – and by the literature critical of expressivism. It also makes sense, as I indicated above, of why Blackburn and Gibbard think that expressivism preserves their substantive naturalism.
Finally, I didn’t, of course, say that it was an objection to their view that they were trying to change the playing field to be more favorable to their view. I just said that we don’t have to accept their way of changing the playing field, and that we need to understand why the playing field is set up as it is, before we go around saying that the fact that people say things when they set up the playing field entails that those things are true.
Hi Mark. My apologies for getting back to this discussion so late. You write:
I agree with this, of course. But here’s a claim that I’ve always understood as distinguishing every view that we would intuitively want to include as being in the tradition of Ayer, Stevenson, and Hare from every other view that we would intuitively want to include as being in the tradition of Moore:
I think that every view that we would want to intuitively include as being in the tradition of Ayer, etc. would accept this claim, and every view that we would want to intuitively include as being in the tradition of Moore, etc. would reject this claim. Nothing in this claim says anything about expressing the desire-like state. So, I think what you say next is true, though misleading:
This is true, but your inclusion of 2) here seems to imply that views in the tradition of Ayer *need* to rely on an clear exposition of this ‘expression’ relation in order to distinguish their views from those in the tradition of Moore. Now, as a matter of fact, I think that views in the tradition of Ayer, etc. have a clear exposition of this expression relation—”Utterances of atomic ethical sentences express desire-like states in the sense that such utterances count as sincere only if the speaker is in a particular desire-like state. However, as I just mentioned, I don’t think that views in the tradition of Ayer need to rely on a clear exposition of the ‘expression’-relation in order to distinguish their views from those in the Moorean tradition. I think you would actually agree with all this so far. Next you write:
I agree with you that this assumption is required in order to present a charitable view *of some current semantic projects*. But this assumption is not required in order to present a charitable view *of the sense in which these views (the ones offering up these positive semantic proposals) are to be distinguished from* views in the tradition of Moore. So, I guess I see your next claim as a Red Herring: