Expressivism, or more generally moral anti-realism, is an area of tremendous philosophical endeavor these days, despite being counterintuitive to plain persons and a research program which has a history of making progress only at the cost of eliminating distinctions between itself and its rivals. (OK, so I tipped my hand there.) Why is it so popular? What are the fundamental motivations behind it, and which is/are primary? Or in other words, why do so many smart people think they can’t be moral realists? I can think of four motivations, but I couldn’t say which is uppermost in people’s minds. Since there are no doubt plenty of anti-realists who read this blog, maybe they can improve my understanding.
- Naturalism. Since the scientific world view contains no values, our moral discourse better not refer to any.
- Motivational considerations of a broadly Humean sort. Since moral commitments motivate and realist beliefs never do, moral commitments are not realist beliefs.
- The agent-centered perspective. We are agents and not merely spectators. Moral realism gives us a false, spectatorial picture of moral thought. Anti-realism gives us the true, agential (e.g. “planning-laden”) view of moral thought.
- Realism ressentiment (operating at the Freudian level, if at all.) It’s somehow distasteful/oppressive/square for there to be objective values which constrain the rightness of our choices.
Survey says…? (And feel free to debate the merits of any of these motivations.)
To put my cards on the table, I’ve always wanted to be a realist, but I’m finding myself defending various kinds of anti-realist views more often. I don’t think that expressivist views are the best ones in the anti-realist camp but that is another question. I also feel more and more sympathy towards different kinds of naturalist, realist views. What motivates me towards the anti-realist side of things? Mainly the difficulty of understanding the ‘strict’ realist positions in the first place. I find it increasingly difficult. The stories I read about what the moral properties are supposed to be and how we are supposed to know about them all give analogies and metaphors that rely on good faith and usually are full of holes. If there was a comprehensible realist position, I’m sure I would be willing to endorse it wholeheartedly. I do find McDowell’s position appealing but many realists don’t.
Couple of other motivations result from supervenience and epistemological considerations. I’m sure there are plenty of others.
Heath,
The options you offer seem to me to suggest a very simplified view of why people might reject realism as opposed to understanding more complex reasons they might have for accepting anti-realism. The last one of the suggestions is actually a borderline ad hominem, suggesting psychological reasons why anti-realists are unable to accept realism. The anti-realist’s counterpart analysis of the realist would be to suggest that people have a psychological need to believe in realism that parallels the need some have to believe in religion. As Mackie puts it, “The denial of objective values can carry with it an extreme emotional reaction, a feeling that nothing matters at all, that life has lost its purpose.”
Kant offers this self-diagnosis, “I inevitably believe in the existence of God and in a future life, and I am certain that nothing can shake this belief, since my moral principles would thereby be themselves overthrown, and I cannot disclaim them without becoming abhorrent in my own eyes.” Kant admits here that his belief in realism is because it would be too awful to believe otherwise.
But most people – realist or anti-realist – will tell you that there complex reasons for holding the view they do that don’t simply boil down to a single, simple reason. PEA Souper Michael Huemer’s Ethical Intuitionism is a good example. He identifies five possible meta-ethical positions: non-cognitivism, subjectivism, nihilism, naturalism (reductionism), and intuitionism. He gives at length arguments against the four views he rejects (including one realist alternative) and, at even greater length, arguments for the one he accepts. There are many considerations which can’t easily be summed up in a single line as the reason he is a realist (although if he thinks it can, I certainly would be interested to hear that from him). Similarly I don’t think an anti-realist’s reasons for holding the view he does can be in any one case boiled down to one line, let alone the reasons for anti-realists in general so easily being summarized.
It is odd to me that you put the question as “why do so many smart people think they can’t be moral realists?” First, it is not that they can’t be realists, it is that they find arguments against realism compelling. Second, most (all?) realists find arguments against some forms of realism compelling, so even realists accept arguments against realism. Anti-realists might have some different, more general arguments that cut against any sort of realism (Mackie thought he had that), but all one needs to reject realism is to accept arguments that different realists also accept – perhaps just buying Huemer’s arguments against naturalism/reductionism and buying naturalist arguments against intuitionism.
Or, just as realists think they primarily have arguments for a particular realist view and only secondarily arguments against other views, the anti-realist might think the issue is not “why can’t I be a realist” so much as it is finding compelling positive arguments for a particular anti-realist view. Those arguments might be complex, and certainly will vary from one anti-realist view to the next.
I think it would be useful for Heath to define or at least characterize realism. According to the way I would define realism, Kant most likely qualifies as a constructivist non-realist, but it seems David White wants to classify him as a realist. I don’t know whether this is because David White interprets Kant differently than I do, or because he has a different understanding of realism than I do. A definition might help to avoid this sort of confusion, and so might make the discussion here a little easier.
Let me second David’s remark that there are bound to be a plurality of reasons for adopting an anti-realist attitude towards morality. Nonetheless, I think that it is useful to divide these reasons into those that arise from considerations internal to moral practice and those that arise from considerations external to moral practice. Whereas motivational considerations and considerations about agential perspectives seem to be the former kind of reason, naturalism, as you characterize it, seems to be the latter kind of reason. Though even that can be a subtle matter, since a naturalist might draw attention to a special feature of the evaluative, say, that makes it somehow problematic that the evaluative has an intelligible place in nature. For what it is worth, my opinion, which I shall not defend here, is that the reasons internal to moral practice for being an anti-realist are more interesting than the reasons that are external to it. At least for a moral philosopher (though perhaps this last qualification trivializes the claim).
David K,
The fact that no one has a good taxonomy in metaethics these days doesn’t help matters. Let’s say that an anti-realist, for present purposes, is one who thinks that moral judgments are not like ordinary beliefs and are more like ordinary desires or intentions (the “direction of fit” criterion). Claims that moral judgments are in some ways like both beliefs and desires we’ll count as anti-realist. That definition could be improved on, no doubt, but precision isn’t vital.
David W,
I recognize that many anti-realists’ assessment of their own reasons could be “there are a zillion considerations on both sides, and after weighing them up, I just come down on [some particular position on] the anti-realist side.” It could be that complex. But it might not be, at least for everyone, and at least for some kinds of considerations. For example, no one (yet) seems to embrace the naturalist line of reasoning I suggested. That’s at least a little informative. I also recognize that we could ask the same kinds of questions about realists, but that would be a different post.
Mark,
Thanks for the constructive suggestion. That’s the sort of thing I was looking for, to help me sort out my own thoughts.
Hi Heath. Thanks for the post. You originally said this,
and I was going to caution against conflating expressivism (a positive thesis about moral judgments) with anti-realism (a negative thesis about metaphysics). But then you said this:
so it seems as though you are aware that there are views that combine expressivism with—well, I was going to say “realism,” but you are considering such views anti-realist. So, it seems to me that what you are really after is why folks would be expressivists.
The reason I find expressivism plausible has something to do with your third option. Moral judgment surely has *something* to do with pro- or con-attitudes. To me, the connection is so strong that I find that the notion of an amoralist (def., one who makes sincere moral judgments, but lacks the correlative pro- or con-attitude) to be conceptually incoherent. So I take this as strong evidence for expressivism. But you don’t have to go all-out like me; that a change of attitude reliably follows in the wake of a change in moral judgment is sufficiently strong evidence, some think, for expressivism. A third reason is that moral judgments are normative, that is, they are intended, in the end, to guide behavior, and expressive or directive illocutionary acts, as opposed to assertive illocutionary acts, are usually far more apt to help achieve that objective. These are, of course, all defeasible reasons, but they should certainly carry some weight.
Oh, by the way, do you (and do others) really think that expressivism is “counterintuitive to plain persons”?
I offer this characterization of realism, although it was disputed in another post:
Value realists think there are objective facts that make (some of) our evaluative statements true. For example, “Torturing puppies is wrong” is made true by the objective fact that torturing puppies is wrong.
How complex are the motivations for realism/anti-realism? As I think David White suggested, one might well answer Heath’s opening question with a very long list of all the arguments that have been offered for realism, against realism, objections to those arguments, replies to those objections, etc. . . . And if we philosophers were purely rational beings, perhaps that would be the only appropriate response.
But here’s why I don’t think that’s the most illuminating sort of response:
In other words, if a graduate student decides to write a dissertation defending expressivism, say, then he will almost certainly “find” that there is at least one defensible form of expressivism, and he will almost certainly wind up for the rest of his career defending that. The probability that a graduate student deciding to write a dissertation on theory X, after doing the research, concludes that theory X isn’t true after all, is almost 0.
Corrolary: the detailed arguments are almost entirely epiphenomenal.
I say something like this at the end of Ethical Intuitionism. Of course, this runs contrary to what we’d like to think about ourselves. But it’s still true.
I’m less sure than Michael H. that initial sympathies determine where a philosopher ends up. Here are three further factors that could diminish or eliminate the importance of initial sympathies:
1. Advisor bias: A grad student might have an advisor whose views differ from the student’s initial sympathies. The student might find herself defending some or all of the advisor’s views — either because the advisor is persuasive, or because she thinks she must defend the advisor’s views in order to succeed, or because she thinks things will be easier for her if she does, etc. In that case, the student’s initial sympathies don’t play much of an explanatory role.
2. Original contribution bias: A student might come up with a new line of argument for a view with which she did not initially sympathize. She might think that the new line of argument is more interesting or impressive than the arguments she would give to support the view she arrived to grad school having; so she might end up defending a view that differs from her initial or even current sympathies.
3. Attraction to perplexity: Someone might think that metaethical questions are genuinely perplexing, with seemingly good arguments on all sides, and this might be enough to attract her to metaethics even if she doesn’t start out feeling strongly attracted in any particular direction.
I wrote a dissertation defending moral relativism. I now doubt that moral relativism is true, though I still think it’s worth taking seriously.
My dissertation advisor was (and is) a relativist, but I chose him because of my topic rather than vice versa.
Michael Smith also changed from being a kind of relativist, I think, though he didn’t call it that.
Hi Michael and David Killoren,
That is a pretty depressing line of thought you guys are pushing! (Doesn’t mean that it isn’t true, though.)
Michael: why do you think people have these “vague sympathies” prior to graduate school?
The way you guys seem to be spinning the story… well, it sounds like what views we end up defending and fully articulating is largely determined by non-rational (or irrational?) gut instinsts (Michael’s suggestion) as well as social pressures and the desire to say something new and cool (David’s suggestions). And that sounds pretty bad. Maybe even shameful.
Is this an accurate description of what happens?
In my case, the sympathies were not vague: I’m a hardcore moral realist, always have been, probably always will be. Torturing babies for fun is just wrong, independently of what anyone else thinks or feels about it, etc., etc.
There are cool and fun puzzles about how I know that torturing babies for fun is just wrong, independently of what anyone else thinks about it, but I do know it. And I knew it before I started taking philosophy classes in college. Maybe you did do?
So I am a little unhappy with how you represent the progression through graduate school. In my case, and I suspect for many others, things went like this:
I started off prior to intensively studying philosophy in college and graduate school knowing that certain things were wrong, and that they were wrong independently of what anybody thought or felt. In college and graduate school, I had a finite amount of time, and so I couldn’t read everything that was of interest to me. I was aware that people wrote whole books and long articles defending views I antecedently knew to be false. These works looked pretty interesting, but if you need to cut something from the reading list, defenses of views antecdently known to be false seems reasonable. Sometimes I would come across a cool or interesting challeng to my moral realism, and I would try to figure out how to answer the challenge. (Sometimes I came across cool and interesting challenges to the claim that there are tables, and I tried to figure out how best to respond to the challeng.) Since I knew the conclusion was false, I wasn’t going to be convinced by the challenge — but it was still worthwhile, and fun, and perhaps illuminating to figure out where the challenge went wrong, i.e., to diagnose the flaw in the argument.
So I didn’t start with “vague sympathies.” I started with some concrete, first-order moral knowledge, and some concrete, meta-ethical knowledge about the first order judgments. A lot of our anti-realist colleagues will be unhappy with my way of describing my situation; I understand this, and respect them — but I’m going to stick to my guns and stand by this description.
Michael: Given your own meta-ethical views, do you think I am not entitled to describe my situation in this way? (I have purchased your newest book, but it hasn’t arrived yet, so forgive me if you discuss this in the book.)
Perhaps this is clash of intuitions territory but I found your 1, 2 and 3 to be pretty powerful motivations (I don’t think I understand 4, and share the worry it contains an ad hominem element). I’ll grant you that it’s not clear the natural or best way to respond to these motives is anti-realism (cf Crispin Wright’s often undervalued remarks on how to frame this debate in Truth & Objectivity). Nevertheless, your 1-3, plus a broad feeling that moral values exhibit *some* degree of relativity or response-dependence (pretty much a platitude among new undergraduates I find) seem to me overpowering motivations for anti-realism or at least a ‘qualified’ realism, as Mark Johnston put it.
2 and 3 look pretty good to me. I’m less impressed with 1 and 4, but two strong reasons is better than one often gets in philosophy.
I also question whether meta-ethical anti-realism is counter-intuitive to the “plain person.” It seems to me that the plain person has contradictory tendencies in this respect, favoring either realism or anti-realism depending on how the issue is raised (Plato was perhaps not the first to notice this, but it is well presented in the views of his character Euthyphro).
I just want to say that Kris M’s explanation sounds totally plausible to me. And I also want to point out that the explanations I offered aren’t all cynical. I think if someone got into metaethics because they were “attracted to perplexity”, that wouldn’t be shameful.
My characterization of how philosophers come to their positions in my first post is oversimplified: I agree with David K’s comments about further factors that influence our positions. Also, “vague sympathies” may not have been the right term. What I meant was that in college or graduate school before your dissertation, you might have realist or anti-realist sympathies, but you probably do not embrace any very specific or detailed theory. So Kris McDaniel, in college, had a very strong belief in realism, but he probably did not have a specific or detailed version of realism, he probably did not have responses to the objections to realism, and so on.
Kris, I think you probably do *know* that baby-torture is wrong, and also know that that’s objective. You probably even knew it when you were in college. But just to be extra cautious, and nice to the anti-realists, let’s say that you strongly believed in moral realism. So, if you decided to make metaethics your main area, you would almost certainly have wound up producing more elaborate defenses of some more elaborate version of realism.
At the same time, there were other, intelligent people in college who strongly believed in anti-realism (they would have said they “knew” that there are no objective moral facts). And if one of them were to go into metaethics, they would almost certainly wind up producing elaborate defenses of some more elaborate version of anti-realism.
If you were to read their arguments, or they were to read yours, we can already predict (even without knowing any of the details of how your or their arguments would go) that neither of you would be convinced by the other.
So, in this sense it seems to me that the detailed philosophical arguments are largely epiphenomenal.
But, this doesn’t mean that we’re irrational, nor that considering arguments is pointless. Rather, it means that we need to pay more attention to the general, non-technical, original motivations behind realism and anti-realism, rather than to the more detailed, technical arguments that most academic discourse focuses on. Which is why Heath’s opening post is a good one.
I think the underlying motivation for realism, for most realists, is indicated by Kris’ post: Some things just seem obviously wrong, and it doesn’t matter how anyone feels about them (etc.). It doesn’t seem like the badness of torture is just a matter of some convention, or my just happening to prefer non-torture over torture, like how I prefer chocolate over vanilla ice cream. (Of course, this characterization of the motivation is pretty close to just saying: anti-realism seems obviously wrong.)
I think one of the underlying motivations for anti-realism is that moral properties are “weird”. This isn’t discussed much in print (except for Mackie’s famous discussion of “queerness”), but I think philosophers are constantly moved by the sense of what is weird. Some philosophers would rather say almost anything than wind up with a “weird” metaphysical view.
One other, small comment: Heath’s mention of motive #4 isn’t “ad hominem”. Heath isn’t arguing that anti-realism is false because anti-realists have bad motivations (that would be an argument ad hominem). Hypothesis #4 might be (slightly) *insulting* to anti-realists, but then that’s no reason to think it isn’t true.
Michael Huemer says: “I think the underlying motivation for realism, for most realists, is indicated by Kris’ post: Some things just seem obviously wrong, and it doesn’t matter how anyone feels about them (etc.). It doesn’t seem like the badness of torture is just a matter of some convention, or my just happening to prefer non-torture over torture, like how I prefer chocolate over vanilla ice cream. (Of course, this characterization of the motivation is pretty close to just saying: anti-realism seems obviously wrong.)”
But this can’t be right. Anti-realists like Gibbard and Blackburn also say that some things just seem obviously wrong and that it doesn’t matter how anyone feels about them. They also say that it doesn’t seem like the badness of torture is just a matter of some convention, or my just happening to prefer non-torture over torture, like how I prefer chocolate over vanilla ice cream. Moreover, they say these things wholeheartedly, not tongue in cheek or with fingers crossed.
What makes them anti-realists, rather than realists, is not the fact that they deny truisms like this. What makes them anti-realists is that they say, in addition, that what they are doing when they say these things is expressing their non-cognitive attitudes. So what makes realists realists must be the fact that they say, in addition, that what they are doing when they say these things is expressing their cognitive attitudes. Their respective motivations for being realists and anti-realists are therefore their respective motivations for saying these additional things, whatever they are.
Professor Smith writes:
“What makes them anti-realists, rather than realists, is not the fact that they deny truisms like this.”
This seems to be saying that realists and anti-realists are on the same page when it comes to claims like ‘torture is wrong’; that neither denies these “truisms”. But do anti-realists take these kinds claims to be “truisms” that can be denied? ‘Denial’ sounds to me just like another word for ‘claiming to be false’, which ,by virtue of their non-cognitivism, they’re obviously not doing. The sense in which anti-realists aren’t denying these claims is quite different from the sense in which realists don’t deny them, no?
Michael Huemer wrote:
“I think the underlying motivation for realism, for most realists, is…: Some things just seem obviously wrong, and it doesn’t matter how anyone feels about them (etc.)…. (Of course, this characterization of the motivation is pretty close to just saying: anti-realism seems obviously wrong.)”
Both parts of that seems right to me. The parenthetical also sounds a lot like what we might say about people who are religious – It just seems obvious that where there is something there must be a maker of that something, thus the universe must have a creator. And both of those ideas do have a pull to them, but just as one might question religious assumptions one can also question realist assumptions, especially when there seems little evidence beyond the “logical” appeal.
“I think one of the underlying motivations for anti-realism is that moral properties are ‘weird’.”
This also seems right to me and to have a natural parallel to religious belief. While the idea of there being a creator seems obvious at first, it starts to shake under scrutiny. Conceptions of God can be logically impossible or incoherent. The idea that there is a creator does not explain how God was created, thus not providing an ultimate explanation. In the end, religious stories become just too weird to believe. But that seems to be a reasonable complaint, at least as a justification of a shift of the burden of proof squarely onto the theist.
“One other, small comment: Heath’s mention of motive #4 isn’t ‘ad hominem’. Heath isn’t arguing that anti-realism is false because anti-realists have bad motivations (that would be an argument ad hominem).”
I agree. The reason I called it “borderline ad hominem” is because his list consists of three reasons one might have to reject realism followed by one possible cause for rejecting it. Since he did not clearly distinguish the reasons from causes, it seemed that one might read his list and interpret them all as claims about reasons. I don’t think that is what he intended, thus my use of the qualifier “borderline”.
What makes them anti-realists, rather than realists, is not the fact that they deny truisms like this. What makes them anti-realists is that they say, in addition, that what they are doing when they say these things is expressing their non-cognitive attitudes. So what makes realists realists must be the fact that they say, in addition, that what they are doing when they say these things is expressing their cognitive attitudes.
Wouldn’t Mike H’s response be that, were realists to express other cognitive attitudes to torture (e.g. asserting that it is permissible), it would still not be the case that torture is permissible. On the other hand were anti-realists to express other non-cognitive attitudes to torture (e.g. approving of torture in some way or other), it might be the case that torture is permissible. Isn’t this the worry, or something close to this?
It’s pretty incredible that one still sees this kind of comments repeated:
“But do anti-realists take these kinds claims to be “truisms” that can be denied? ‘Denial’ sounds to me just like another word for ‘claiming to be false’, which ,by virtue of their non-cognitivism, they’re obviously not doing.”
“On the other hand were anti-realists to express other non-cognitive attitudes to torture (e.g. approving of torture in some way or other), it might be the case that torture is permissible. Isn’t this the worry, or something close to this?”
It’s not like anti-realists aren’t aware of these objections. Blackburn wrote a reply to this line of thought against expressivism already 1988 (How to be an Ethical Antirealist). Of course people kept repeating the objection so he had to write another paper making the same points. That paper, ‘Must We Weep for Sentimentalism’, was published this year in Dreier’s ‘Contemporary Debates in Moral Theory’ collection. Highly entertaining and recommended. It would be nice to hear what’s wrong in Blackburn’s reply before we proceed and keep on repeating the old objection.
Suppose we were to ask Gibbard and Blackburn whether torture would be permissible if they were to express approval of torture. I take this to be equivalent to asking them whether torture is permissible in the nearest possible world in which they have and express an attitude of approval towards torture. Would they answer yes? Plainly not. For in answering this question they would still have to express their actual non-cognitive attitudes. Since they presumably actually disapprove of torture even in that world–they disapprove of torture-whether-or-not-they-disapprove-of-it–I am quite certain that they would vehemently deny that torture is permissible in that world.
“On the other hand were anti-realists to express other non-cognitive attitudes to torture (e.g. approving of torture in some way or other), it might be the case that torture is permissible. Isn’t this the worry, or something close to this?”
Jussi is surely right, but can I also put in a word for Stevenson here. He was clear if moral judgements express attitudes, then what justifies the moral judgement is just what justifies the attitude. I morally disapprove of toture because of the unnecessary pain it causes. I express this by saying ‘torture is wrong’. What justifies this claim is just what justifies the attitude it expresses, namely the unnecessary pain of torture. Since this fact hasn’t changed in a world in which my attitude has, I will not make a different judgement about the wrongness of torture in that world. That claim was always present in expressivism – indeed it was what defined the movement from a naive subjectivism to expressivism. So its a very old point.
Also regarding this comment:
“But do anti-realists take these kinds claims to be “truisms” that can be denied? ‘Denial’ sounds to me just like another word for ‘claiming to be false’, which ,by virtue of their non-cognitivism, they’re obviously not doing.”
I guess minimalism about truth might sort that out (which again is nothing new – it was suggested by Stevenson in the final essay of his ‘Facts and Values’).
It’s not like anti-realists aren’t aware of these objections. Blackburn wrote a reply to this line of thought against expressivism already 1988
Jussi, I’m glad they’re aware of it. My question was simply trying–off the cuff, since I’m not familiar with all of the exchanges on this issue and presumably you are–what Mike H’s worry was. That is, I was trying to locate what Mike H. found troubling about anti-realism. I wasn’t presuming to offer any new criticism of anti-realism. Now, I guess if it’s ok with you, I’m going to try to relocate it.
Michael Smith writes,
. . . I take this to be equivalent to asking them whether torture is permissible in the nearest possible world in which they have and express an attitude of approval towards torture.
That sounds right. Let that world in which they express and attitude of approval to torture be w.
Would they answer yes? Plainly not. For in answering this question they would still have to express their actual non-cognitive attitudes.
Right. In the closest world to ours in which my attitudes have changed on torture, there is probably something wrong with my attitudes, and torture is still wrong. But consider worlds that are not close to ours. Consider distant worlds where my attitudes are different and there isn’t anything wrong with them. Maybe you’re claiming that there aren’t any such worlds or that it begs the question to assume there are such worlds.
Since they presumably actually disapprove of torture even in that world . . . I am quite certain that they would vehemently deny that torture is permissible in that world.
I’m not sure what you mean by “they actually disapprove of torture in that world”. I think this means that they disapprove of torture in that world. But recall that that world is w, by stipulation. So now they also disapprove of torture in w, which seems contrary to the hypothesis.
Mike Almeida asks what I meant by “they actually disapprove of torture in that world”.” What I meant is that in the actual world they disapprove of torture in w (where w is, by stipulation, a world in which they approve of torture).
Mike,
I’m sorry. It’s just that one becomes slightly frustrated of hearing the same objections so many times. The issue has been discussed earlier here on Pea Soup too like in the Huemer thread few days ago. But, I’m sure we all have similar cases. I guess the non-naturalists are just as sick of hearing that science proves that there are no moral properties so aren’t you believing in similar things as fairies? Of course the non-naturalists have replies to this too but it’s not like many people are listening. The same seems to be true of expressivism and the dependence of moral obligations of the attitudes of the appraisers.
Everyone will judge torture to be wrong in a world in which they are stipulated to judge torture to be wrong, whether that judgement is expressive of an attitude or belief. So that can’t be the problem (which would be the same for expressivists and anti-realists alike). The problem is whether we – in the actual world – have to consider that judgement in the stipulated world to be appropriate. Expressivists (and realists) do not, since for expressivists the wrongness depends on the pain (which is unchanged) not the attitude.
Of course in my last post I meant to say “which would be the same for expressivists and realists alike”.
Mike also says: “Consider distant worlds where my attitudes are different and there isn’t anything wrong with them. Maybe you’re claiming that there aren’t any such worlds or that it begs the question to assume there are such worlds.”
Expressivists need say neither. Expressivists will ask: what must the world be like if there isn’t anything wrong with (moral) approval of torture? The only thing I can think of is a world in which torture doesn’t cause unnecessary pain (since pain is what the moral disapproval depended on).
So there are such worlds, and expressivists will say that torture is acceptable in them, but then they are also worlds in which it’s no longer clear that torture isn’t acceptable.
(Perhaps then it wouldn’t be “torture”, depending on the load you think that term bears.)
Jussi, I haven’t been reading a lot of Pea lately and so not paying much attention to what’s come up on earlier posts. I do understand the frustration.
Neal, you write,
Expressivists will ask: what must the world be like if there isn’t anything wrong with (moral) approval of torture? The only thing I can think of is a world in which torture doesn’t cause unnecessary pain (since pain is what the moral disapproval depended on).
There are other things you can think of. Maybe the psychology of these agents is radically different. Maybe pain does not have quite the same qualitative feel (though it plays the same functional role) in other worlds. Maybe pain is so celebrated for its incredible biological utility in some worlds that agents find it difficult to disapprove of it.
Hi everyone,
Just to weigh in on this on a few points.
First, I am expressivist. What do I take to be good reasons for believing expressivism? That is a long story, but since at least some of my reasons do not appear on the initial list of four reasons given, I shall at least indicate what they are.
Rejecting expressivism plausibly commits one to some form of cognitivism (some could object even here, but I think the objections can be answered). Cognitivist views typically come in reductionist and anti-reductionist versions (there are versions try to avoid the idea that moral predicates refer to properties at all, but while remaining cognitivist – I have in mind e.g. Horgan and Timmons’ stuff, among others; I am putting those views to one side for these purposes to avoid unnecessary complexity – H & T themselves at some point come close to conceding that their view might reasonably be characterized as a form of expressivism anyway).
Back to the main plot: Reductionist forms of cognitivism face a challenge from Moore’s Open Question Argument and its successors, where its successors include Hare’s missionary and the cannibal tale and Horgan and Timmons style ‘moral twin earth’ arguments and the like. I think these arguments can be supplemented and strengthened in vaious ways so that they still make some trouble for ‘Cornell style’ forms of realism and the like.
Anti-reductionist views face different objections, but one I think is often not fully appreciated is the challenge to explain supervenience.
So I’d recommend adding a fifth reason to your list, which is really a dilemma: In rejecting expressivism, you are led to either reductionist or anti-reductionist cognitivism. On the first horn, the OQA and its more sophisticated successors cause trouble. On the second horn, the challenge to explain how irreducible and sui generis moral properties supervene (as a necessary truth) poses problems. This is all very quick and simplistic, but hey, this is just a quick post on a blog which is inevitably going to be a bit elliptical.
Then again, I suppose this could all be epiphonemenal. I could just be an expressivist because I wasn’t breast fed or something. I’ll put those Freudean hypotheses firmly to one side. For what its worth, my main advisor (Geoff Sayre-McCord) was a pretty hard core realist. On the other hand, I did work with Simon Blackburn a fair bit.
Oh, and Michael Smith and Jussi and the rest are absolutely right about the attitude dependence stuff. We expressivists don’t believe for a minute that right and wrong is attitude dependent in the way critics have often assumed. It is a bit frustrating that this objection keeps getting wheeled out again and again, when its been so firmly and effectively dealt with so many times for decades now – Simon Blackburn’s latest paper on this is the last in a long line of expresivist tracts explaining why this objection rests on a conflation of expressivism with subjectivism. But enough said about that.
– Mike
While I agree with Jussi and others that Blackburn and Gibbard have heard and rebutted the “But surely killing babies would be wrong even if I didn’t disapprove of it!” objection about seven million times, I’m not sure if the fact that expressivists have a good response invalidates the original point. The question at this point in the dialectic was why some people have an initial bias for realism. The fact that the most sophisticated anti-realist views combined with minimalism may license us to say “It is part of the eternal furniture of the universe that torture is wrong” (which, of course, is for them just another expression of complex con-attitudes toward torture) is far from obvious to someone beginning metaethics. (Unless they happen to be taught by a sophisticated expressivist, of course.) For psychological explanatory purposes, it’s enough that it appears as if expressivists can’t say these things, or take them at face value. So the desire to say such things may well lead to looking for a defense of industrial-strength realism, and if Michael H.’s pessimistic view is correct, that basic pro-attitude toward realism (note the irony!) may well remain after the student comes to understand all the resources of contemporary expressivism.
Moreover, it remains controversial whether expressivists can indeed handle all the truisms, arguably especially if we include explanatory ones like “I believe that torture is wrong because torture is wrong”. Unless they go minimalists about explanation, at which point we can forget about the realism/anti-realism debate…
Antii, is I believe that torture is wrong because torture is wrong really a truism?
To me it sounds very strange, not obvious at all. I doubt very much that it is an ‘ordinary thought’, one that occurs to anyone outside of self-conscious philosophy.
What’s a truism?
Michael Smith points out that anti-realists claim to be able to accomodate some of the intuitions that I said motivate realism. But that doesn’t show that those intuitions are not in fact a large part of the motivation for realism. It shows that anti-realists dispute whether the motivations for realism are good motivations. If you think the anti-realists are right, then that shows that realism is poorly-motivated.
I, of course, think the anti-realists are wrong. I think the expressivists, in particular, cannot accomodate common sense moral thoughts like, “Torture would be wrong even if I approved of it.” I also think they have no right to speak of moral facts, moral properties, or moral truth. I’m fully aware that they claim to be able to speak of those things, but I disagree. I think they are merely trying to change the meanings of words like “property”, “fact”, and “truth”, to enable them to talk like realists. And the reason they’re doing this is that if they stated their views forthrightly (like Ayer and Stevenson did), people would see that they were obviously wrong.
Jussi points out that anti-realists such as Blackburn have responded to some simple and obvious objections to their views. That’s right. But here’s an objection the expressivists haven’t responded to yet: the Frege-Geach problem. It’s an extremely well-known, potentially fatal objection to expressivism/non-cognitivism; they’ve had 50 years to respond to it; and they still haven’t solved the problem. Yet expressivism remains perhaps the most popular view in meta-ethics, and expressivists still go on talking as if they’d solved the problem, using the very sentences that Peter Geach pointed out 50 years ago that their view couldn’t make sense of.
Now someone (perhaps Jussi?) is going to point out that Blackburn, Hare, et al. made some efforts in the direction of solving the Frege-Geach problem. These efforts generally overlook the generality of the problem, i.e., they give an ad hoc account of how to understand one or two specific kinds of sentences. But Frege-Geach is a general problem involving *any and every* context in which a proposition can be embedded (since all of these contexts can take moral claims in place of the proposition).
Geach’s original example–
–is just one illustration of this problem. How would an expressivist understand what that sentence means?
Michael,
I was surprised to see this criticism in your book. Here’s one basic line one could take.
If p, then q
is logically equivalent with
not-p or q.
Now take the set of planned worlds from Gibbard semantics. Not-p is a set of planned worlds that is ruled out by the plans expressed by the moral claim p. q is the set of planned worlds that include the plans expressed by q. not-p or q a.k.a. if p, then q is the aggregate set of these two sets of planned worlds. So, you are committed to either plans that do not involve the actions that are mentioned in the p or the actions that are involved in the q. Of course the fact that your p is an instance of a sentence including an existential quantifier makes the sets more complicated. But, there is nothing different in this semantic analysis than in the possible world semantics which a realist would give.
Your claim in the book was that the complex claim expresses a belief in a sui generis non-structured proposition. There is a worry that this is denying compositionality which is problematic in many ways.
Jamie,
Here’s an exchange:
A: You believe torture is wrong because you’re so emotional: you’re disgusted by the thought that anyone would ever suffer.
B: No, I believe torture is wrong because *torture is wrong*.
I think that conversation could occur among non-philosophers.
David, maybe you’re right.
Some moral realists (Nick Zangwill in particular) deny that moral facts can cause moral beliefs, though, so if it is a common sense intuition that they do (and I still doubt it is), it isn’t one that cuts between realism and anti-realism.
Jamie:
I don’t know what Nick Zangwill believes in this matter but one need not accept that there is a causal relation between “A believes that p” and p in order to accept the truth of “A believes that p because p”. Especially if the belief in case is a case of a priori knowledge, one can think that the fact that p is at least part of the explanation of why the agent believes that p without thinking that there is a causal relation between p and the subject’s belief that p. I take it that one can say, presumably correctly at times, “John believes that ‘if p then (if q then p) is a tautology’ because ‘if p then (if q then p) is in fact a tautology” without thereby committing oneself to a causal relation between the fact that a certain sentence is a tautology and John’s belief.
Jamie,
To add to Sergio’s points, which I think are correct, I just want to say that although not all realists think that moral facts have causal/explanatory powers, it’s probably the case that only realism is compatible with the view that moral facts have causal/explanatory powers. (This is especially likely if we accept the very broad definition of realism that many here seem to accept.) Thus, I think that although any common-sense intuition that moral facts have these powers wouldn’t “cut between” the realism/anti-realism distinction, it remains that the intuition would support realism unless it were explained away.
Michael H.,
On the Frege-Geach problem, it is a serious problem, and I think every expressivist realizes that dealing with that problem is a key challenge. Indeed, I have myself recently taken a stab (shameless plug: see my “Ecumenical Expressivism: Finessing Frege”) at it. My solution, if it works, is perfectly general, and so meets your concern about the generality of the problem. I won’t rehearse that solution here though.
Of course, realist views face serious problems too (some of which I mentioned in my previous post). I am not sure we make much progress just by pointing out that opposing views face certain well-known objections. That is true on both sides of the dispute.
You also dispute expressivist construals of a variety of claims, but without any further argument (apart from F-G) that these construals must be incorrect. You then offer the following diagnosis:
“And the reason they’re doing this is that if they stated their views forthrightly (like Ayer and Stevenson did), people would see that they were obviously wrong.”
Hmm. Seems a bit uncharitable, somehow.
Also, this accusation presumes that Blackburn’s and Gibbard’s *real* view is not their stated view, and that they secretly think that torturing babies would be OK if everybody approved of it – or think something which, when stated forthrightly obviously entails such conclusions. I see absolutely no reason to believe this. You may be right that their analysis of what it means to say ‘Torturing babies would be wrong even if everybody approved of it’ is incorrect, but that is another matter. That doesn’t imply that Blackburn and Gibbard (and Hare and myself, for that matter) secretly believe some *other* semantic view, according to which these statements make no sense or are incoherent or some such.
Regards,
– Mike
Mike H.,
I don’t see why expressivists have had a harder time explaining the Frege-Geach problem than cognitivists have had explaining the motivating nature of moral beliefs. Both views have their burdens to bear. And there’s no better solution for cognitivists than the efforts expressivists have made toward solving F-G. Or at least I don’t see why we should think cognitivism has the upper hand.
Sergio, good point.
I have to admit that I have a lot of trouble seeing how the fact that if p then (if q then p) is a tautology could explain someone’s belief that it’s a tautology.
You weren’t careful with the quotation marks, but I assume you meant a conditional proposition to be the tautology, and not some sentence. But in that case, that a certain proposition is tautologous is itself a tautology. And if we want a sense of ‘explanation’ with oomph, then it’s hard for me to see how tautologies can explain things.
David, again, if we’re talking about explanatory powers, so that it’s a kind of metaphysical explanation, then I don’t see any clear piece of common sense that tells us that moral facts have them.
Your idea that only realism is compatible with the explanatory powers of moral facts seems plausible to me, but I’m not confident that it’s true. (I’m pretty confident that it’s in the right neighborhood.)
Jamie,
Yeah, I was trying to be careful to leave open the possibility that common sense doesn’t tell us moral facts have these powers; that’s why I used phrases like “the intuition would support realism” rather than “the intuition supports realism.”
As for the other point: I’d be interested to know which versions of non-realism you think have a chance of giving a causal/explanatory role to moral facts. It seems very unlikely to me that expressivism could do this, but I’m not so sure about other kinds of non-realism. (Of course, if realism = non-expressivism, as Heath’s post seems to suggest, then there aren’t any other kinds of non-realism; but I’m guessing you’ve got a narrower definition of realism in mind than that.)
I don’t have much to add at this point to David and Sergio’s responses to Jamie. Except maybe this: while I think it’s commonsensical to say the sort of things David suggested, I’m not sure what the force of ‘because’ is in that sort of claims. *Perhaps* it is the sort of thing an expressivist could say. All I wanted to point out is that that’s not obvious even after we’ve taken in the lessons of contemporary expressivists on other realist-sounding tropes.
Mike writes:
But here’s an objection the expressivists haven’t responded to yet: the Frege-Geach problem. It’s an extremely well-known, potentially fatal objection to expressivism/non-cognitivism; they’ve had 50 years to respond to it; and they still haven’t solved the problem.
Actually it isn’t quite 50 years since the first mention of the point by Geach (I think it was 1957-8), though Mark Kalderon tells me the point is in Ross much earlier.
But anyway, while I’m not an expressivist, two points seem fair to raise on their behalf. (1) In philosophy 50 years is not all that long of a time, and (2) in that 50 years several expressivists have tried to handle the problem at some level of generality. (For that matter some non-expressivists have worked on approaches to the problem as an earlier thread here highlighted about a week or so ago.) I’ve argued in various places that some of the extant attempts don’t work, but there surely have been serious attempts that look/looked promising. Gibbard’s world/norm and fact/prac proposals are surely and attempt to deal with the issue in a relatively wide-ranging way, though it is controversial how successful it is in handling all the relevant embedded constructions. Hare, was in many ways worried about related issues even before Geach published his objection, and Hare pursued the issues involved at a quite general level. I could mention others, but you get the idea.
The development of expressivism seems a lot like the development of other philosophical theories. Someone proposes a theory in a relatively simple form to capture a certain set of philosophical motivations. Objections are put forth by detractors. Fans of the theory find ways to modify it to avoid the objections while retaining the key ideas. Some of these modifications generate new objections, or there are new versions of the old objections put forth. New modifications are made. And so on.
I don’t think the modifications are meant to hide what the theorists really think. Rather I think they are an attempt to come up with a theory that captures the original motivations for the view while avoiding the most telling objections. I think this means that it is always worth going back to the simple versions of the theory when you are trying to get clear on the motivations because they typically get captured in the most simple way by those early theories. But I don’t think that means that you can refute the whole range of theories that develop out of this theory just by showing that the theory is subject to some objection in its most simple version.
I am a bit dismayed that some have found it obvious that expressivists flout firm moral intuitions—in effect that expressivism would constitute a debunking explanation of morality. This just misses the insight, if it is one, that is common between contemporary expressivists and moral sense theorists like Hume (sensibility theories can but need not be construed along expressivist lines). When a hungry child asks for bread and someone give him a stone, Hume would complain of that person’s want of a natural affection. The complaint seems both sound and sensible. What’s so bad about giving to those in need because of your compassion for the needy? This is not the same as saying something is right because I wanna or wanna wanna or whatever.
Mark,
that’s a good point. I think though what drives the opposite side is Kantian insight of the blindness and contingency of natural affections. We see people whose affections are more towards cruelty and come to think that it could have just as well happened that our inclinations would have been targeted towards giving the stone instead of the bread. Such causal contingency in explanations for our natural affections seems to make us want to look for grounds for why some affections are better than others. I take it that this drives the realist’s motivation for moral facts that give standards for the blind compassions. I’m not saying that this is an argument for the realist side but it does give an explanation for wanting to resist the purely sentimentalist picture.
I have some questions for the expressivist, plus the customary apology for the author’s own ignorance of the details of the expressivist position.
According to Michael Smith, “What makes them anti-realists, rather than realists, is not the fact that they deny truisms like this. What makes them anti-realists is that they say, in addition, that what they are doing when they say these things is expressing their non-cognitive attitudes. So what makes realists realists must be the fact that they say, in addition, that what they are doing when they say these things is expressing their cognitive attitudes.”
Question 1: Is this a fair characterization?
Question 2: Suppose the answer to question 1 is “yes.” Suppose I say the words, “Torturing babies for fun is wrong.” Call this sentence “P.” When I say P, I have not expressed a cognitive attitude, but rather a non-cognitive one.
Belief is a paradigmatic cognitive attidude. Here is a sentence:
I believe that P.
Is this true (by the expressivists’s light)?
How can this sentence be true , if my saying P only expresses something non-cognitive?
Question 3: Suppose the expressivist says that “I believe that P” is true. Suppose I say:
1) I believe that torturing babies is wrong.
2) I believe that the democrats have retaken the House and the Senate.
Does the word “believe” have the same meaning in sentence one and sentence two?
Question 4: If the answer to question 3 is “yes”, why is the view a non-cognitivist view?
Question 5: If the answer to question 3 is “no”, what does the expressivist say about:
I believe that torturing babies is wrong and 2+2=4.
I hereby apologize if the answers to these questions are well-known and tedious to provide. (Although a good reference would be appreciated.)
Kris,
good questions altogether. It’s always difficult to put words in someone’s mouth so I’ll start by quoting Blackburn: ‘I think that naturalism demands this view of ethics (quasi-realism), but in any case it motivates it. It does so because in this package the fundamental state of mind is not located as a belief (the belief in a duty, right, value). We may *end up* calling it a belief, but that is after the work has been done. In fact, we may end up saying that there really are values (such as the value of honesty) and facts (such as the fact that you have a duty to your childre).”
So, given this passage, I think expressivists would accept that ‘I believe that p’ is true even in the moral cases. What would be true is that I accept that p and that this acceptance is technically constituted by desire-like attitudes. We have just ended up calling acceptances of moral judgments beliefs and there is no reason to abandon this first-order practice even though in doing metaethics we come to realise that the attitudes are not what philosophers mean by beliefs. In this way the ordinary discourse is a minimalist about ‘beliefs’.
3 is a good question. Much depends on what is meant by mean the same and whether the claims are made in the internal or the external perspective. I don’t think that the second claim either commits to any technical notion of beliefs but only to the acceptance of the claim as does the first to the acceptance of the moral claim. At least the utterances would be used for the same purpose. We don’t usually say these kinds of things to describe our own mental life but to say that torture is wrong or who won the election.
Of 5, I wonder if the expressivist would say that when we say things like that we are not really saying that ‘I believe (torture is wrong & 2+2=4)’ but rather ‘I believe (torture is wrong) & I believe (2+2=4)’. The natural language can hide such subtle differences.
Hi Jussi,
Thanks for the helpful replies. I have some follow up questions to what you’ve just said.
You wrote in your response to my first question:
“In this way the ordinary discourse is a minimalist about ‘beliefs’.”
Why is this the right thing to say? I’m inclined to think that what we’ve learned (given that expressivisim is true) is that there isn’t a unique *non-disjunctive* mental state that fits the ordinary notion of belief. There are mental states — call them “acceptances” which apparently are desire-like– and there are mental states that philosophers have called beliefs — call thse states “P-beliefs”. Something is a belief in the oridnary sense if and only if it *either* an acceptance *or* a P-belief.
(And isn’t this a surprising result? Wouldn’t we have been inclined to think that beliefs formed a unified psychological kind if we weren’t expressivisits?)
You wrote: “3 is a good question. Much depends on what is meant by mean the same and whether the claims are made in the internal or the external perspective.”
I didn’t understand the second sentence, so this follow-up question is just a request for clarification.
“Of 5, I wonder if the expressivist would say that when we say things like that we are not really saying that ‘I believe (torture is wrong & 2+2=4)’ but rather ‘I believe (torture is wrong) & I believe (2+2=4)’. The natural language can hide such subtle differences.”
Suppose the expressivist grants that I believe that torturing babies is morally wrong. I also believe that 2+2=4. Doesn’t it follow then that I have two beliefs? So if the expressivist feels pressure to say what you suggest the expressivist could say — well, I worry that what you say doesn’t help with the problem.
(But if the expressivist does take someone to have a belief that P just in case that person either accepts that P or P-believes that P, then I guess the expressivist can happily say that I have two beliefs.)
You didn’t answer Q4, so I want to push a little harder here. Why is expressivism a version of anti-realism? Expressivism says that:
(1) Moral claims can be the objects of beliefs *in the ordinary sense of “belief”*.
(2) Many moral beliefs are true, and moreoever are true independently of what people wanna or wanna wanna. (And since the minimalist story about truth is the whole story about truth, moral beliefs are even true in the same (minimal) way in which every truth is true.)
And finally, since expressivism is all about respeciting the platitudes, the expressivist will presumably grant:
(3) The true moral belief that Hitler was a really bad guy is far more reasonable to believe than its denail.
So why not also grant:
(4) We know that Hitler was a really bad guy.
This is why I’m puzzled: in what sense is this view an anti-realist view?
This is in response to Kris.
Jon believes that grass is green just in case Jon stands in a certain relation to the proposition that grass is green. Call this relation ‘B’. According to a common view, B is the semantic value of ‘believes that’, and so ‘Jon believes that P’ can never be true unless ‘P’ picks out a proposition to which Jon stands in this relation.
Expressivists do not think that in order for ‘Jon believes that murder is wrong’ to be true, ‘murder is wrong’ must pick out a proposition, and Jon must stand to it in B. Does it follow that they must think that ‘believes that’ is ambiguous?
No. The ambition of expressivism is to offer a univocal semantics for ‘believes that’ which yields as a special case that when ‘P’ is a descriptive sentence, ‘Jon believes that P’ is true just in case Jon stands in B to the propositional content of ‘P’, but when ‘P’ is a normative sentence, ‘Jon believes that P’ is true in virtue of Jon’s having some non-cognitive attitude. So – different realizers for descriptive and normative belief, falling out from a single semantics.
That is the ambition, I say. Can anything like this work? Well, here is a first pass: ‘Jon believes that P’ is true just in case Jon is in M, where M is the mental state expressed by ‘P’. If descriptive sentences express mental states that consist in bearing R to their propositional contents, then if ‘P’ is descriptive, ‘Jon believes that P’ is true, on this view, just in case Jon stands in R to the propositional content of ‘P’. While if normative sentences express noncognitive attitudes, then if ‘P’ is normative, ‘Jon believes that P’ will be true, given this view, just in case Jon is in that noncognitive attitude. So that gets us what we wanted.
Your question #4 is a good one, and I think many people have been confused about this. Given this semantics, the expressivist will continue to say things like ‘there are moral beliefs’, ‘moral sentences express beliefs’, and so on. So what makes it expressivist?
The answer is that someone who holds this view thinks that though descriptive beliefs consist in bearing a certain relation, R, to a proposition, moral beliefs do not consist in bearing that very relation R to any proposition. This is a very theory-laden thesis, about the nature of moral thought. But no surprises there – expressivism is a _theory_ – a theory about the nature of moral thought and moral language.
Is the toy semantics for ‘believes that’ offered just above a plausible one? That’s something that we can test. You test it the way that you test any other semantic theory. For example, does it license the inference from ‘Jon believes that murder is wrong’ and ‘Mary believes that murder is wrong’ to ‘there is something Jon and Mary believe’? Or from ‘Jon believes that murder is wrong’ and ‘it is true that murder is wrong’ to ‘Something Jon believes is true’? I don’t know how to construct an expressivist semantics for ‘believes that’ which licenses the second inference in any natural way, and the one that I gave above doesn’t explain either one.
My point is: from the fact that expressivists have the ambition to explain these things, it does not follow that they will be able to explain them. That’s why it’s important both to take expressivists at their word about what they would _like_ to construct a semantics in order to be able to do, but also to test them on whether any theory that they construct is actually able to do these things.
Mark Schroeder, could you clarify something? You wrote
My question goes back to Kris’s Question 4 (how is a view that endorses the claim that when we accept moral claims we have an attitude of belief properly called non-cognitivist?) and Michael H’s charge of changing the meanings of the relevant terms. If we took it as a given that belief is a cognitive attitude, how are we supposed to simultaneously render this view non-cognitivist and make sense of the claim that ‘Jon believes that P,’ where P is normative, is true in virtue of some non-cognitive attitude Jon has?
Just to take a stab in the dark here, we can’t say that the true-in-virtue-of relation holds because belief is identical to the non-cognitive attitude. After all, we’ve stipulated that belief is a cognitive attitude.
But we also don’t seem to get what we want if we say that the non-cognitive attitude bears some non-identity true-in-virtue-of relation to ‘Jon believes that P.’ For Kris’s question was how expressivism can endorse claims such as ‘Jon believes that P,’ where belief is stipulated to be a cognitive attitude and ‘believes’ is used in the ordinary sense, and still be called non-cognitivist. To say that the non-cognitive attitude bears some relation other than identity to the belief doesn’t go any distance towards explaining how acceptance of a moral claim can be mere belief and yet also be non-cognitive. (In fact, it doesn’t say anything about that at all; rather it only says something about that in virtue of which it is true that we have our moral beliefs.) But only an explanation of how Jon’s belief that P could be non-cognitive would seem to render the view non-cognitivist, and I thought that was the rabbit that Kris wanted to see pulled out of the expressivist hat.
As a way out of that dilemma, I suppose one could say, ‘To be properly called non-cognitivist, a theory need not deny that when we accept moral claims, we believe them. Rather, it must only hold that when we believe moral claims, that in virtue of which we have moral beliefs (where the that-in-virtue-of relation is not identity) is itself a non-cognitive attitude.’ But that seems to depart from the ordinary understanding of ‘non-cognitivist.’
Any clarification would be much appreciated. Like Kris, I apologize in advance if I’ve missed some well-known, standard response to these questions.
Hi, Josh.
The charge of changing the meaning of other terms is highly relevant. The way that expressivism responds to the kind of problems that have been discussed here, is to provide a non-standard semantics for the sentences that are used to state things. If that non-standard semantics is false, then expressivists don’t count as believing that there are moral beliefs, for example, even if _they_ think that they do. Similarly, someone who thinks that ‘God’ denotes my wristwatch, and believes in my wristwatch, does not thereby count as a theist.
But how do we evaluate the expressivists’ semantic claims? We can’t just insist that they are false. If expressivists have a complete non-standard semantics for English (which they don’t, but they aspire to), and it makes the same predictions about all of the ordinary things that we say in ordinary life – the kinds of things about which we have linguistic intuitions – then how do we know it is wrong?
Fortunately, we’re not actually in this position. Expressivists have said what they would _like_ to be able to devise a semantics to do, but they haven’t gotten very far, yet, at doing it. So it’s a very open question whether there will turn out to be good evidence that their semantics is false, and hence that they have changed the topic.
Still, go back to ‘believes that’. What is pre-theoretically uncontroversial, is that we say things like ‘Jon believes that murder is wrong’. Other things are pre-theoretically uncontroversial: it is okay to infer ‘there is something Jon and Mary believe’ from ‘Jon believes that murder is wrong’ and ‘Mary believes that murder is wrong’. An adequate semantic theory for ‘believes that’ needs to make sense of these and other sorts of data. But no theory about the semantics of ‘believes that’ is pretheoretically uncontroversial – semantics is itself theory.
An _objection_ to expressivism can only be based on some piece of data it fails to accommodate – it isn’t an objection to expressivism that it is inconsistent with cognitivism. Given that, it’s fair game for expressivists to offer their semantics for the sentences that are used to formulate the data, and to explain why they are assertable or not. You can’t ‘stipulate’ that what ordinary people mean by ‘believes that’ satisfies the semantics that cognitivists would use – though you can, of course, point to the usual set of data that lead semanticists to treat it in that way, and point out that expressivists have yet to offer a serious competing account that explains all of the same data.
That leaves open, however, whether there will still be other sentences – theory-laden ones – that expressivists will not be able to capture. Presumably, there must be, or expressivists would not have managed to say what makes their view different. It is just that I don’t see why any of those sentences have to be things that we have commonsense views about.
For example, as I noted in my previous post, according to cognitivism there is a single relation, R, which Jon stands in to the proposition that grass is green when he believes that grass is green, and stands in to the proposition that murder is wrong when he believes that murder is wrong. This is not, I think, pre-theoretically compelling, even if it is true. No claims about propositions, relations, etc. are going to be pretheoretically compelling. So failing to capture this wouldn’t be an objection to expressivism.
Finally, as I also insisted on, above, and pointed out last week in a previous thread, that expressivists have the ambition of developing a semantics for ‘believes that’, or any other construction, which has certain properties, is not to say that there is any expressivist semantics which has those properties. It is useful to wonder about what expressivists would get to say, if their view were fully developed and the details worked out, but it is a very live question whether the details do work out, and one I am myself highly skeptical about. The toy semantic theory that I offered above shows that expressivists need not think that ‘believes that’ is ambiguous. But if it is a genuine theory about the semantics of ‘believes that’, then it competes with other theories in the philosophy of language, and is subject to the same criteria, which it may not meet.
Thanks, Mark, that’s helpful. I’m still hung up on a couple of things, I think. I’m still not sure exactly where the mistake is in my earlier comment, for example, but first a couple of clarificatory points. First, it definitely wasn’t supposed to be an objection that expressivism is inconsistent with cognitivism (!). I also didn’t mean to stipulate anything about what ordinary people mean by ‘believes that.’ What I’d thought had been stipulated, though, is that belief is a paradigmatically cognitive attitude. Maybe that’s where the hang-up is. I read Kris as suggesting, reasonably, that much of the conventional dialectic in the literature takes cognitivism just to be the view that acceptance of a moral claim is belief. Notice that this is an account of cognitivism, not of belief. If that’s what cognitivism is, then if some theory says that acceptance of a moral claim is belief, it’s hard to see how that theory could be non-cognitivist.
So maybe the kernel of the question behind this isn’t (just?) whether the sophisticated expressivist can offer a novel account of ‘believes that,’ but whether in doing so there’s also a reconfiguration of what it means to be a cognitivist. Kris’s Question 4 requires an answer to that question, it seems, since as soon as that novel account is given, expressivism seems to license the claim conventionally assigned to cognitivism, namely that acceptance of a moral claim is belief. At least that’s how I understood Kris’s question as to why this view is non-cognitivist.
So I guess that it would be nice to hear a further story about what cognitivism is, if it isn’t the (unique) view that belief is the attitude we take when accepting a moral claim. In your earlier comment, you said that “Expressivists do not think that in order for ‘Jon believes that murder is wrong’ to be true, ‘murder is wrong’ must pick out a proposition, and Jon must stand to it in B.” The first of those two theses sounds to me like descriptivism or factualism, not cognitivism. The second might get closer to cognitivism. In terms of carving terrain this way, I think I’m operating under the influence of recent books like Kalderon’s. But maybe I have funny views about these labels? In any case, it sounds like you want to say that expressivism is non-cognitivist and that expressivism can endorse the claim that the attitude we take in accepting a moral claim is belief, in which case I guess you want to deny any account of cognitivism that characterizes it as the view that the attitude we take in accepting a moral claim is belief.
This is a follow up to Josh and Mark’s comments, wherein I indicate my basic agreement with Josh.
In this thread, the notion that everyone — including the expressivists — wants (and should) accept certain moral truisms, such as “It’s morally wrong to torture innocent babies just for kicks.” A view that is inconsistent with a truism is a view that’s in trouble.
In general, a view that is inconsistent with something obviously true – a truism – is a view that is obviously problematic. And it actually doesn’t matter whether the truism is a moral truism, right? If you’ve got to respect the plattitudes, you’ve got to respect the plattitudes.
Expressivists tell us that there are non-cognitive attitudes and cognitive attidueds. Here is a *truism*:
(1) Belief is the paradigmatic cognitive attitude.
It seems to me that it follows from (1) that:
(2) All beliefs are cognitive attitudes.
But it also seems to me that, given expressivism and the semantics Mark has sketched on behalf of the expressivist, (2) is false. Hence, if expressivism is true, then (1) is false. So expressivism is false. Expressivism is false because it denies a truism about belief.
Kris,
I think this is a reasonable worry, and I agree that it shouldn’t matter whether the relevant platutide/truism is moral or non-moral.
Here is a reply: In the past, we have taken our use of various terms to be causally regulated by stuff with a theoretically defined and uniform (along certain dimensions). It then turned out, on further investigation, that the stuff doing the causal regulation was disjunctive, and that not all of it fit with our theory. Our use of ‘jade’ turned out to be causually regulated by jadeite and nephrite.
When we learned this fact about the underlying stuff, we had a choice. We could have said, “Well, its a platitude that jade is a natural kind – a single sort of stone (or whatever). So there is no jade.” Instead, we went on talking about jade even though it picked out a disjunctive set of stuff.
If the expressivists are right then we have (through philosophical, as much as empirical, investigation in this case) learned that our use of ‘belief’ is causally regulated by disjoint states, some of which fit well with our prior theoretical assumptions, and some don’t. On making this discovery (if that is what it is – itself controversial, of course – but there are expressivist arguments for this), we again have a choice. If the quasi-realist project succeeds, then we have good practical reasons to go on treating these strictly non-cognitive attitude as beliefs – so that their expression can figure in rational inferences, planning, argument about what to do, etc., etc.
On my own view, the jar with ordinary thought is arugably less extreme. I argue that normative judgments are partially constituted by a non-cognitive attitude and partially constituted by a belief (in the same sense that descriptive beliefs are beliefs – remember, we have to allow that the underlying stuff is disjunctive). This makes referring to these hybrid states as falling under the heading ‘belief’ less jarring, or so I’d argue.
Anyway, that is all rather elliptical, but its meant to show that there are some further interesting moves to be made here, or so it seems to me anyway.
Best,
Mike
Hi Mike,
I’m worried that your reply to my worry doesn’t help. Here’s what’s worrying me now. (I have a lot of worries. I’m a worried man.)
First of all, it isn’t a platitude that jade is a natural kind. It isn’t a truism. When we learn that Jade isn’t a natural kind, we don’t learn that a truism is false.
But it is a platitude that belief is a cognitive attitude. In fact, when we try to glom on to the difference between expressivists and everyone else, the expressivist say things to us like “Moral utterances express non-cognitive attitudes, not cognitive ones, like, you know, belief.”
The expressivist relies on the platitude in order to get us to even understand her position. But expressivism (given Mark’s suggested semantics) is incompatible with the platitude, and thus false.
That’s my worry.
Here’s a second but related worry.
Consider that last speech.
“Moral utterances express non-cognitive attitudes, not cognitive ones, like, you know, belief.”
I don’t see how the expressivist can say this speech with a straight face. I don’t really have an independent grasp of the difference between a cognitive and a non-cognitive attitude — I get some help in understanding the distinction when I am told that belief is a paradigmatic example of a cognitive attitude, and that whatever we are expressing when we make a moral judgment, we are not expressing something like a belief.
The techincal terms “cognitive attitude” and “non-cognitive attitude” have never, to my knowledge, been rigorously defined. They aren’t bits of ordinary language, unlike the term “belief”, which is a perfectly respectable bit of ordinary language. So I doubt that anyone else has an independent grasp of the distinction between a cognitive and a non-cognitive state either.
So if we abandon the platitude, we lose our grip on what made the expressivist theory distinctive. We abandon the platitude, we lose our grip on what it is to be a cognitive state. If a belief can fail to be a cognitive state… well what’s the point of entertaining a counter-factual whose antecedents is the denial of something obviously necessarily true? But were I to learn that a belief can fail to be a cognitive state, I think I would have to conclude that I have no idea what “cognitive state” means. I think I would have to conclude that no one had given a clear sense to that term. And in any theory formulated in terms of it would be at least as unclear as the term in which it is formulated.
(In an earlier comment in the thread, I did raise a worry that belief turns out to be a disjunctive kind if expessivism is true. So i want to briefly say something about this worry as well. I don’t think it’s a truism that beliefs should form a natural kind. So I wouldn’t take the mere fact that expressivism denies this to constitue a devastating objection to expressivism. It’s more of a minor worry: isn’t it surprising to learn that this is the case? I think in your comment, you’ve pretty much answered this worry satisfactorily.)
I’m really confused now. How can “Belief is a paradigm cognitive attitude” be a platitude if “cognitive attitude” isn’t ordinary language?
What’s a platitude?
Here’s how I see the situation. There seems to be a kind of minimalist sense of ‘belief’, according to which a belief that p is any state that is correctly expressed by sincere assertion of p. Maybe this is wrong; maybe there is no such sense; but I think Crispin Wright has made a good prima facie case that there is.
Now, we also have a concept, for which we have no better ordinary term than ‘belief’, that is a functional kind term. Beliefs, in this sense, are states that play a certain hard-to-define role in our cognitive economy. Expressivists say that moral judgment does not play this role but some other role. They are therefore not ‘beliefs’ in this theoretic sense. But there is nothing wrong with calling them ‘beliefs’, because we also have the perfectly good minimalist sense.
There may still be other problems saying what expressivism is — my own view is that there is now a problem saying what divides realism and irrealism and the problem is not especially a problem for either realists or irrealists but a general metametaethical problem.
Just a great discussion. Thanks everyone!
Trying to flesh out Kris’ and Josh’s worries, Mark writes:
I’d like to make one large point by way of making two supporting points. The large point is that, while Mark’s worry articulated here is trenchant, it is a worry for *everyone*, and not just for expressivists. How so?
I take it that by ‘natural’ is meant either (i) ‘intuitively compelling’, or (ii) ‘makes sense of the fact that “expressive” predicates (like ‘is wrong’ on expressivist accounts) play a role in determining that some kind of noncognitive/nonrepresentational state is attributed to a person by the use of attitude-attribution verbs, like ‘believes that’. But, no semantics is going to be natural in sense (i) if it has to use the notions of propositions or possible worlds, for these notions are just, well, quite unintuitive.
How about ‘natural’ in sense (ii)? Suppose ‘is Christian’ and ‘is a Jesus-Freak’ pick out the same property, only the latter is also conventionally used to express contempt for Christians. (Use any example: ‘is from Quebec’/’is a Canuck’; ‘is an American’/’is a Yankee’; etc) Suppose also that Bettina believes that Carrie is Christian, but has no contempt whatsoever for Carrie for being Christian. Here are two reports:
(1) Bettina believes that Carrie is Christian.
(2)Bettina believes that Carrie is a Jesus-Freak.
Intuitively, (1) accurately reports Bettina’s psychological state, but (2) misreports Bettina’s psychological state as one consisting of, in part, contempt. However, (2) would not be a misreport if ‘believes that’ were used to report *only* a representational or cognitive state. (One response here might be that the contempt expressed in uttering (2) is the speaker’s, and that (2) does not actually attribute contempt to Bettina. I think there are good reasons to think otherwise that are too deep to go into here.) Thus, if my example is not misleading in any way, the semantics of ‘believes that’ must somehow reflect that beliefs, at a minimum, are not always purely cognitive. But if so, then providing such a semantics is a challenge that falls on everyone, not just expressivists. (Unless, I guess, if one wanted to deny that ‘is a Jesus-freak’, ‘is a Canuck’, etc work expressively like expressivists think ethical predicates work.)
Here is another way to make the latter point. Consider these inferences: From ‘Bettina believes that Carrie is a Jesus Freak’ and ‘Donna believes that Carries is a Jesus-freak’ to ‘There is something that Bettina and Donna believe’; and from ‘Bettina believes that Carrie is a Jesus-freak’ and ‘it is true that Carrie is a Jesus-Freak’ to ‘Something Bettina believes is true’. Mark, I’m assuming you’d have to be an expressivist about ‘is a Jesus-freak’ (or ‘is a Canuck’, etc). How would you construct a semantics that licenses these inferences in a way that makes sense of the fact that ‘is a Jesus-freak’ actually plays a role in attributing some kind of mental state that has, at least in part, an element that is noncognitive/nonrepresentational. If you can do so, then an expressivist (especially a complex expressivist like Mike Ridge, David Copp, Kyle Swan, or me) would likely be able to adopt that semantic story, or at least a part of it.
Great discussion everyone – very helpful. Not much to add except couple of small points.
First, I think it is interesting to look at the mathematics analogy. When we do mathematics we say things like ‘I believe that 345*12=4140’ and that ‘I think this is true’, and so on. And, in an obvious way we think that our attitude is a cognitive one – we need to give proof for it and if someone shows that our calculation is mistaken we change our opinion. This is how things look from the internal perspective of mathematising. In many ways, things look the same and we say the same things when we are moralising, from that internal perspective.
When we do philosophy we stand back from the practice. Some of the mathematical claims seem to be too referring to objects and describing the mathematical facts. There is a temptation to think that those claims are true if and only if the corresponging mathematical objects exist and false otherwise. However, it’s proven to be difficult to find the mathematical facts and objects from the world. So, one way would be accept that our mathematical claims were false like ordinary beliefs would be in the analogical case.
But, another way would be to say that the mathematical mental states are not really in the business of describing the mathematical facts of the world but doing something else. This would be a claim from the external meta-perspective. If we accept this, does it mean that we should revise the practice of speaking that we believe that 2+2=4? I don’t see why we should. I think this would be another case where we found out that the states we call beliefs in ordinary life can consist of different types of attitudes. I guess for the purposes of classification in the ordinary life the responsiveness to reasons is more important than robust representation. So, in the same way in metaethics from the external perspective we can accept that moral judgments are not robustly representing like beliefs in that sense but they are still beliefs in the more minimal sense of the ordinary discourse.
I’d also like to recommend for everyone interested in the topic a correspondence in Analysis in mid-nineties between Smith and other Canberrians versus Alex Miller and John Divers. The ANU/Princeton group objected that minimalism in ethics would not be so minimal after all because it would rule out expressivism. They would say that minimalist accepts the platitude that the ordinary moral assertions express beliefs. Then they would say that accepting this platitude commits the minimalist to another platitude that connects beliefs to representation and aim at truth, and by same token we have already ruled out expressivism. Beliefs cannot be understood in a non-committal way. The Miller reply on the behalf of the Wright minimalism was that if there is a platitude that connects beliefs to descriptions and facts then the descriptions and facts in these platitudes must be minimal as well – they do not include any metaphysical package which the expressivist could not accept. If on the other hand the facts and descriptions are not understood in this minimal way, then there are no *platitudes* that connect the FACTS and REPRESENTATIONS to beliefs in the ordinary sense. It’s a great debate and for once I think that the Canberrians might not have ended winning it.
First let me echo Jamie in response to Josh and Kris, and then a reply to Dan.
If by ‘cognitivist’ you stipulatively mean, ‘like a belief’, then I don’t think that you can use the word ‘cognitivist’ to distinguish between cognitivism and expressivism. But if by ‘cognitivist’ you mean, ‘the very same kind of underlying attitude that someone is in when she believes that grass is green’ (or some other paradigmatically descriptive belief), then I think you _can_ use ‘cognitivist’ to distinguish cognitivism from expressivism. It’s true – people are often sloppy and just say that cognitivism is the view that moral utterances express beliefs. But clearly they mean _the very same kind of state as descriptive sentences express_. That is, as I have been saying all along, where they differ from expressivists. I personally think there is great evidence for this view, but it is not one of the pretheoretical data.
Second, to Dan: I agree that your pejorative cases are interesting. I myself think that it’s important to classify like views together and unlike views separately, and your view is both sufficiently unlike Ridge’s and sufficiently unlike Gibbard’s that I think it deserves separate discussion. I almost mentioned your treatment of attitude-ascriptions in my initial comment. It’s more detailed and better motivated than other expressivist treatments I know about, but I think its advantages are also peculiar to the ways in which your view is different from all of the others. I also think that the costs for your view are going to be significantly different from those for, say, Ridge or Gibbard or Copp or Barker or Blackburn or Horgan and Timmons, so I didn’t mean to be addressing it.
To (partially) answer your challenge, though, I’m not personally convinced that pejoratives require any different treatment than presuppositions. For one thing, their color is not pure evaluative tone. ‘Yankee’ connotes certain descriptive features associated with a certain kind of prejudice. So does ‘redneck’, but different features. And the kind of contempt that is associated with these different terms varies widely. It seems to me that the evaluative tone associated with these words is probably connected with them in the same way that this other descriptive information is connected with them. If it is a presupposition, compare to ‘the present king of France is bald’, which carries, I’ll assume, the presupposition that there is one and only one present king of France. This embeds in valid arguments, and belief-attributions attribute the presupposition to the subject of the attribution.
This treatment seems only to work if the presupposition of pejoratives is itself some kind of normative claim, and that seems to suggest that the same strategy can’t work for every normative claim. That’s just a baby theory and I’m not committed to it. Another baby theory is that, like using ‘ain’t’ or swearing in polite company may reveal your vulgar upbringings, use of ‘nigger’ or ‘wop’ reveals that you are a racist, but not because of any expressive conventions associated with the language at all – just because those words are the ones in circulation in subcultures that are generally racist.
Maybe neither of these theories is promising, but I throw them out just to show that a hybrid expressivist-cognitivist view about pejoratives is not inevitable. Both toy theories I threw out are purely cognitivist about pejoratives. Though I agree with you, it is a good case, and your view can build on it, though I think there is no reason to think that Ridge or Gibbard can; their views are too different in structure.
Okay, this is helpful. So if I’m following properly, it sounds like expressivists should say one of the following two stories in response to Question 4.
Mark S’s story. There are beliefs and BELIEFS. With apologies to Jamie, it is, or should be, platitudinous that BELIEF is a paradigmatic cognitive attitude. Further, it is, or should be, a platitude that cognitivism is the view that our attitude in accepting a moral claim is BELIEF. Some people play a little fast and loose and render this platitude in terms of beliefs rather than BELIEFS. (Alternatively: the platitude should be that BELIEF is a paradigmatic cognitive attitude, but some people play fast and loose and say that belief is a paradigmatic cognitive attitude.) But the fast and loose version begs the question against some kinds of sophisticated expressivism. If we go with the former (slow and tight) platitude, we can make sense of the claim that expressivism is non-cognitivist.
Mike R’s (more jarring) story. To render the platitude in terms of beliefs, rather than BELIEFS, is not sloppy. That is, we can legitimately say that it is platitudinous that cognitivism is the view that we have the attitude of belief when we accept a moral claim. But, in the time since this became platitudinous, we’ve learned (surprisingly) that ‘belief’ actually refers to a disjunct, in which case we need to revise our platitude, because it’s false. On the revision, beliefs can be non-cognitive, and therefore when sophisticated expressivism allows that we believe some moral claims, it can remain a non-cognitivist view.
Some of the comments, such as Jamie’s and Dan’s, seem consistent with both stories, but I’m not sure if they’d agree with that.
I’m sympathetic with Kris’s worry: once we start to divorce the platitudinous connection between cognitivism and belief, I start to lose my grip on what cognitivism is, and therefore what non-cognitivism is, and so on. I also worry that the traditionalist might just as well be able to hold the line and say
I suppose maybe it’s six of one, half-dozen of the other with this traditionalist rejoinder.
One last thought: Jamie, I’d like to know what a platitude is, too. But in response to the question, “How can ‘Belief is a paradigm cognitive attitude’ be a platitude if ‘cognitive attitude’ isn’t ordinary language?” couldn’t we say that some platitudes are limited to certain audiences, namely specialist platitudes? For example, if there are some things that all rock stars agree on about the crazy life on the road, when they get together and drink too much and destroy hotel rooms, it might make sense to say that ‘Ever since they started bolting down the TV sets in hotel rooms, it’s a platitude among rock stars that you just can’t throw a good TV out the window any more.’ I saw the cognitivism platitude as a (much less exciting) version of that kind of thing: the claim that belief is a cognitive attitude is a platitude (a claim that has been rendered false, as evidenced here?) among those who work on ethics for a living. And some specialist platitudes use terms that aren’t part of ordinary language.
Josh –
Look at it this way: standard presentations of the distinction between cognitivism and noncognitivism assume that ‘judges that’ can be given a neutral semantics, and treat ‘believes that’ as picking out a distinctive kind of attitude, stipulated to hold in paradigmatic descriptive cases such as the belief that grass is green, and assumed to have such theoretical properties as mind-to-world direction of fit, being a representation, metaphorically describable as a kind of ‘map’, and so on.
Since in all such standard discussions, it is clear that ‘belief’ is being used to mean all of these things, ‘belief’ appears to be used in a technical sense in such discussions. That leaves it open to expressivists to grant that technical sense of ‘belief’ in order to distinguish their view from cognitivists, but deny that the relation it picks out is the semantic value of ‘believes that’ in English.
Why shouldn’t they be able to do this? After all, the very setup presupposes that ‘judges that’ can be given such a semantics. So whatever that semantics is, the same one can be given for ‘believes that’. In that case, ‘platitudes’ or stipulations about the techincal sense of ‘belief’ on which it is stipulated to pick out a state with mind-world direction of fit, etc., etc., aren’t really platitudes about the word ‘belief’ in ordinary language. It’s not like they are stipulations about _real_ BELIEF, either. They’re just the definitions that we accept for a technical term, in order to be able to use that term in order to distinguish expressivism from cognitivism.
The moral is that you can’t stipulate that the state of mind that someone is in who thinks that murder is wrong is representative, or has mind-world direction of fit, by stipulating that you are going to use the word ‘belief’ to mean those things, and noticing that in ordinary language, people use the word ‘belief’ to describe the thought that murder is wrong. Your stipulations may coincidentally be correct about the semantics or metaphysics of ‘belief’ in ordinary language, but they aren’t authoritative.
Mark,
I think I’m losing track of the dialectic a bit here. Your last comment seems like it’s supposed to disagree with something I said in my previous comment, but I’m not sure what that something is.
Again, I’ve made no pronouncements about ordinary usage for ‘believes.’ And I’ve just tried to summarize two views (one from you) given here, plus a line that a traditionalist might prefer, all of which are put forth in order to satisfy the desiderata that we should change something in the standard map of the meta-ethical terrain to accommodate the possibility of sophisticated expressivism and that we don’t want to beg the question against sophisticated expressivism.
Your solution, I thought, was to change the platitude about the attitude cognitivism says we have from belief to BELIEF, or if you prefer, from ordinary-usage-belief to stipulated-technical-belief. Mike R’s (more jarring) solution was to render beliefs non-cognitive. The traditionalist solution, which was meant to speak to Kris’s intuitions that belief is a paradigm case of a cognitive attitude (against Mike R’s solution) and that it is a platitude that cognitivism uniquely holds that the attitude we take in accepting moral claims is belief (against your solution), is to hold that there may be another, closely related attitude, called ‘E-belief,’ that we can have when we accept moral claims, and a meta-ethical theory must find room for either beliefs or E-beliefs. While these are all different solutions, which may have different costs and benefits, just in terms of ability to satisfy the two desiderata considered above, they seem to be equal.
None of this says anything about how expressivists must analyze ordinary English uses of ‘believes that.’ And I of course understand that stipulations might not coincide with ordinary usage (indeed, I’ve made some of my bones–meager as they are–exploiting that very point!). So I’m not sure where you’re disagreeing with me. Or maybe I’m confused, and you’re not disagreeing.
Is there any way to frame this issue without the stuff about platitudes? (Although I did like the expression “platitudes about attitudes”.) I’ve completely lost track of what platitudes are supposed to be and what role they’re supposed to play in philosophizing.
Specifically, I don’t see why any metaethical theory should feel in any way constrained to make your rock-star ‘platitude’ come out true. A ‘commonplace’ plainly may be false, after all, and a theory may explain why it is false; a famous instance would be the commonplace that all mammals are viviparous, which turned out to be incorrect. I hereby suggest that we call such a commonplace a “….-…… ………”.
Josh –
The datum, I take it, is that ordinary speakers say things like ‘Jon believes that murder is wrong’. Jamie is right – expressivists can rule that Jon is wrong. But expressivists can also rule that Jon is right – provided that they can provide the right semantics for ‘believes that’ to fit their view. That was my point all along. If we agree on that, then we agree.
I just don’t understand what further point there is to saying that there must be some further data that we would describe by saying, ‘beliefs are cognitive’ or ‘beliefs are not cognitive’ or ‘BELIEFS are cognitive’ or ‘thinking that murder is wrong is either a belief or an E-belief’. I don’t see what anything like that adds. These aren’t data, under any reasonable construal of the data. They’re highly concocted claims which as far as I can tell, you’re stipulating to be analytic. I say that it simply doesn’t matter – as long as we’re making up terms, ‘E-belief’ does just as well as ‘BELIEF’ or ‘belief’ – they’re all equally well made-up, so there’s no dispute between people who describe things in different choices of made-up vocabulary like that. I don’t think there’s any real choice to be made, if the terms are all made up, anyway, and they allow us to describe the same situations.
There is only a dispute if people think they are using the ordinary senses of those terms. But there, I’ve been insisting that sentences involving the ordinary senses of those terms either 1) do not discriminate between expressivism and cognitivism, or 2) are theoretically laden, anyway, and hence nothing expressivists should care about ‘capturing’.
Right, Jamie, I agree that a theory might debunk, as opposed to accommodate, a platitude or a commonplace.
I thought, though, that we were considering sophisticated expressivists’ attempts to accommodate platitudes, such as ‘We believe moral claims’ (isn’t that–the desire to accommodate rather than debunk–in some ways the raison d’etre for making expressivism sophisticated?). And I thought that Kris’s question was whether such accommodation means violating another platitude, namely the one that says ‘Expressivism is non-cognitivist.’
But, if those are the questions, can we avoid platitude-talk? I don’t know. How about “thesis” talk? That avoids dependency on widespread-ness, etc. Here are some theses:
1) We believe moral claims.
2) Belief is a cognitive attitude.
3) Cognitivism is the unique view that holds that we believe moral claims.
4) Expressivism is non-cognitivist.
5) Expressivism can agree that we believe moral claims.
So, I might have misunderstood some folks, but here’s a rough repackaging of the issues in a way that avoids platitudes. These theses are jointly inconsistent as written. A version of Kris’s Question 4 is how to render them consistent, in a way that doesn’t beg any questions against sophisticated expressivism. In order to preserve Kris’s (and, I guess, my own) intuitions about preserving (2) and (3), I suggested a ‘traditionalist’ rejection of thesis (5), so that when expressivists point to some non-cognitive attitude that looks just like belief, and say that this non-cognitive attitude is belief, they’re really talking about E-belief; plus a modification of (1) to read (1*) We either believe or E-believe moral claims. If I correctly follow Mike R. and Mark S., Mike R. suggested that a way out of the inconsistency is to reject (2), and allow that belief can be a non-cognitive attitude; Mark S. suggested (I think) that a way out is to reject (3) and hold that non-cognitivism can allow that we believe moral claims (just not that we BELIEVE or believe-in-a-technical-cognitivist-sense moral claims). And, sure, old school expressivists could debunk, by denying thesis (1).
I don’t know if all of that is right, but at least it avoids talk of platitudes.
Wow, things are moving fast here. Okay, Mark, it sounds like we pretty much agree. One thing, though, is that you say
That’s not true. I’m not stipulating them as analytic. Instead, I was taking them as given, in order to make sense of Kris’s Question 4 (a question that seems reasonable). So, what’s the “further point”? It’s that we want to preserve some intuitions that some of us have, namely that (3) Cognitivism uniquely holds that we believe moral claims, and that (2) Belief is a paradigmatic cognitive attitude. You disagree with (3), from what it sounded like before (although now I’m thinking that you might prefer to reject (2)–your way out of the inconsistency I described in the last comment is less clear to me now), but I think it would be unfair to say that I am just stipulating it. It seems safe to say that (3) is part of the tradition–one that you called ‘sloppy’ but recognized nonetheless. The point of introducing E-beliefs, then, is so that, charitably, we don’t have to say that the tradition is being sloppy, and yet we can still preserve, in a non-question-begging way, room for sophisticated expressivism.
Anyway, that’s all just by way of sorting out the details in a charitable manner. It sounds like we’re in agreement about the substance.
Posting late and speaking as an outsider in all this, I would have to say that I came to philosophy as an undergrad believing in realism full stop, but found myself progressively unable to state the position in a coherent and believable way that did any justice to what would be required for my conceptions of morality to obtain. I became a reluctant Wittgenstinian metaethicist.
Of course I think that my preferred projectivist metaethical view has independently persuasive arguments, but since the original post asks about motivations, I would say that a broadly naturalistic outlook looms large. I just find it impossible to believe that in the nanoseconds when the proto-matter was condensing out of the primordial explosion of the big bang, rules about whether the use of the small fork or the big fork constitutes proper and true ettiquette for eating your salad were somehow ineluctably woven into the fabric of the universe in the way that the laws of gravity and magnetism were a part of the universe. Much as I would love to believe that reality cares about enforcing my opinions on American foreign policy, I’m just incapable of crediting the notion as a serious metaphysical option.
Josh, I’m sorry, this is probably fruitless. For some reason I cannot understand what issue you’re getting at; reframing it without ‘platitude’ talk didn’t help.
You gave five numbered propositions in the comment that’s three comments above this one (I hope; I’d better not wait too long to hit POST). You say that they are inconsistent. You say that Mike Ridge’s suggestion is to reject (2).
But you have offered Mike a really bad position, since the remaining four propositions are still inconsistent.
I cannot figure out whether the idea that belief is cognitive is supposed to be relevant, important, a platitude, a commonplace, a hinge proposition, or what. It appears to have no role to play at all in the story; the inconsistent set is still inconsistent without that claim. And it is still inconsistent without (1), so that one too seems to be irrelevant. I can’t reconstruct the story.
Also, nobody got my joke.
Whoops–you’re right, and that’s a bit of an embarrassing mistake. My apologies for any confusion. I hope this isn’t fruitless, though. At least, I hope we’re getting closer to sorting out Kris’s Question 4, although maybe I’m alone in that hope. Anyway, in the hope that it’s not fruitless, let’s step back: it would be nice to reconstruct some theses that led to Question 4, and then see how the different proposals sort that out. So how can we do that? I’m really kicking myself here for making the mistake you pointed out, but I think general idea behind the earlier comment might be salvageable. Here’s how it might go.
(3), (4), and (5) are jointly inconsistent. This is the inconsistency that Question 4 originally picked up on. Mike’s solution is to deny (3) (cognitivism is the unique view that holds that we believe moral claims), on the grounds that belief can be non-cognitive (against (2)). The traditionalist solution is to preserve (3) (and (2)), deny (5), and avoid begging the question by reformulating (1).
(The more I think about it, I must have misunderstood Mark’s view–maybe he’s been pursuing that same overall strategy as Mike, with different details. But I’m not really confident in this, and I’m just going to talk about Mike’s view and traditionalism for now.)
Last thought: I don’t think that (1) and (2) are irrelevant. (2) is superfluous with respect to the inconsistency, but if the denial of (2) explains the denial of (3), then it’s not irrelevant. And both (1) and (5) were required for formulating the traditionalist line: (1) is there to be modified into (1*) so that the denial of (5) doesn’t beg the question against sophisticated expressivism.
Anyway, I hope that’s less of a mess than the previous comment, and I hope that the point of all this still is clear enough, namely that there’s something to the intuition behind Question 4, since while perhaps expressivists can arrange the pieces to allow ‘Jon believes that slavery is wrong,’ we could also (without begging any questions, allowing expressivism to be as sophisticated as it wants) maintain that any view that says that we believe moral claims is cognitivist.
I didn’t understand Mike to be saying something different than I was, although I’m happy to stand corrected. I just meant (originally) to be illustrating how an _expressivist_ could go about accepting Josh’s 5).
Does that mean that I actually advocate 5)? No. But issues about how to describe what is at stake, when the views in question themselves differ on what our words mean, are very complicated.
Take my favorite analogy. Suppose that a long-time atheist friend of ours tells us that he has realized that he believes in God after all. For he now realizes that God is love, after all, and that is just a two-place relation that he stands in to his wife and his cat, and he has never been tempted by nominalism. Our friend says things like ‘God exists’. But does he believe in God? It’s not clear that he does. Even if he uses ‘God’ to refer to the relation he stands in to his cat, and he believes in that relation, and thereby expresses things he believes when he says, ‘God exists’, it’s natural to think that he is still an atheist, albeit one with a funny idea about what ‘God’ refers to, agapic Christian metaphors aside.
What does that have to do with expressivism? Just this: ‘sophisticated’ expressivists may accept a semantics for ‘believes that’ according to which they can say things like ‘Jon believes that murder is wrong’, and not commit to anything inconsistent with their view. But it is a live question whether their semantics is correct.
Of course, if we are having a conversation with our friend about what is at issue between our view and his view, he won’t agree with our characterization that he is an atheist. Since he thinks that ‘God’ really does refer to love (and always has – his is not intended to be a revisionary view), he thinks that he thinks what every other theist thinks. We disagree. We think that other theists believed in God, but our friend only believes in love. Among the things we disagree about, are what it is that he thinks.
Similarly, if we (say) are cognitivists, we may think that among the things (sophisticated) expressivists get wrong, is the semantics of ‘believes that’. So we may not really think that expressivists _do_ think that there are moral beliefs. But we can’t trot that out in evaluating sophisticated expressivism, because _according to sophisticated expressivism_, that semantics _is_ correct.
Since the semantics of ‘belief’ is itself at stake, we can’t use the ordinary language word ‘belief’ in a way that both sides of the debate can agree to, in order to state the issues. But it doesn’t follow that, conditionally on one or another side of the debate being right, there aren’t further ways that we could describe what is at stake. It is just that those ways of describing it aren’t theoretically neutral.
It’s a tricky issue, trying to formulate the difference between views that also differ on what our words mean, in a vocabulary that both sides can accept. It’s not always clear that doing so is possible. For example, David Lewis has a nice discussion of this kind of problem in connection with Meinong someplace, and how to describe Meinong’s view, given that he had a different idea than Lewis did about how quantifiers work.
I want to emphasize two things: first, that even if we disagree with expressivists about how ‘believes that’ works, we have to accept their view, for the purpose of evaluating their view as a package. If we can’t come up with any objections to how things work, conditional on their assumptions, then how do we know that _isn’t_ how they work? But secondly: that it is a live question whether or not the semantic assumptions that expressivists make can be filled out in a way that does avoid further objections that would come up, even granting their assumptions.
Our friend’s view about the referent of ‘God’ makes little sense, for example, of the thesis that God is omniscient or created the universe. Similarly, it may (or may not) turn out that the semantics required to allow expressivists to say things like ‘there are moral beliefs’ will have funny results elsewhere. So I _don’t_ think these things are untestable and we can’t do _anything_ but take expressivists’ word for it. I think the project is all about expecting expressivists to ante up their promised semantic theories that are going to let them say everything that we ordinarily say, and to test those theories by ordinary semantic criteria, and independent sorts of data.
In short (apologies for the length), I might agree that 5 is wrong, and there might ultimately be evidence of that (and I suggested how to go about looking for it), but given that according to some versions of sophisticated expressivism 5 is true, it’s not fair to assume that 5 is false in evaluating those versions of expressivism, and hence not a fruitful way of trying to classify different metaethical theories.
Wow, I went away for a couple days and the discussion exploded. (I’m at the metaethics conference at Bowling Green, btw, which is great.)
Jussi wrote:
I didn’t mean to say that, so let me try to clarify. Compositionality is exactly what I’m trying to preserve. For simplicity, let’s consider a statement,
where E is evaluative and D “descriptive” (i.e., non-evaluative). My view is that there is a general ‘disjunction’ concept or function, which can be applied to any pair of propositions and yields a third proposition (the disjunction of the first two) as output. The disjunction function can be applied to descriptive propositions, and it can equally be applied to evaluative propositions, because evaluative propositions are equally legitimate propositions (they are merely propositions in a different subject matter from the misleadingly-named “descriptive” propositions). Propositions are suitable objects of belief, and so the proposition expressed by “E or D” is apt to be believed. And it’s belief in that that you express by saying “E or D.”
The problem for expressivists is twofold. First, it seems like they don’t have any object to be the content of the statement “E or D”, because while D expresses a proposition, they think E doesn’t. It’s unclear what the disjunction of a proposition with some non-propositional thing (such as a course of action) is. Second, even if they found a content for the statement, they don’t have a suitable mental state to direct at it, or to be the mental state expressed by the statement “E or D”.
Now, Gibbard might have solved the first problem, by proposing that the contents of statements are these abstract objects, sets of factual-normative worlds. But, as I argue in the book, he hasn’t made any progress on the second problem. You can have an attitude of *acceptance* toward a norm (earlier expressivists would have appealed to the attitude of “approval” toward a practice), and you can have the attitude of *belief* toward a factual proposition. But there’s no attitude, or collection of attitudes, available to the expressivist to direct at the set of factual-normative worlds that is supposed to correspond to the sentence “E or D”.
To see that, what you have to do (as I tried to show in the book) is just to consider the possible different combinations of beliefs and norm-acceptance states. You find that no possible combination rules out exactly the set of factual-normative worlds that “E or D” is supposed to rule out. (Incidentally, it matters what example you take. If the example was (E and D), then Gibbard is fine.)
Michael,
first, here’s what you wrote in the book:
‘The cognitivist need not try to construct belief that *if stealing the candy was wrong then God will punish you* out of separate opinions about the candy theft and about God’s possible future punishment. Instead, the cognivist has a single object, the proposition *that if stealing the candy was wrong, then God will punish you*, and a single mental state (a belief) that can be directed towards that proposition. That is slightly misleading way to preserve compositionality and construct ‘if p, then q’ truth-functionally from separate parts of p and q. If the single propostion is structured, then the expressivist can use the same structure to specify the relevant attitudes.
I don’t also see anything problematic for the expressivist with the type of claims E or D. Say I utter ‘Pleasure is good or pigs fly’. Why can’t this complex utterance express commitment to either approving pleasure or believing that pigs fly? I take it that this is what Gibbard and Blackburn are after with their commitment-theoretic semantics (see Ruling Passions ch. 3). If you add minimalism about truth, saying that such utterance is true is to make the same commitment, saying that it is false is to express the lack of the combinatory commitment. In other words, acceptance is supposed to cover norms and beliefs, not norms and propositions.
This is also supposed to be perfectly general account of the semantics of logical terms so using ‘and’, ‘or’, ‘if, then’ is supposed to always express structured commitments. Thus, ‘snow is white and grass is green’ expresses commitment for believing that snow is white and commitment for believing that grass is green. So, I don’t think the expressivist needs a general single attitude towards collections of possible worlds, if the complex commitments can include attitudes of certain plans in other worlds and of things being in certain way in other ones.
Sorry I haven’t been able to keep up with all the comments in this thread. But I want to respond to a couple of comments directed to me:
(1) Mike Ridge says (responding to me):
Actually, what I think is that they have a view that entails that “torturing babies would be ok if everybody approved of it” *doesn’t make sense*. (That’s because I think the subjunctive conditional requires a proposition as consequent.) They may not know that their view entails this, however, because they have false views about conditionals, or something like that.
MR continues:
I didn’t mean that Blackburn, Gibbard, et al. secretly hold some other semantic view–that is, I didn’t mean that they’re just straight-out lying. What I meant was that they hold views like (that would, in fact, be correctly expressed as):
(1) “Torturing babies is wrong” does not express a proposition.
and that these views entail (in a pretty obvious, direct way) things like:
(2) There’s no sense to be made of statements like “If P, then torturing babies would be wrong” (where P expresses a proposition).
(3) No one has any moral beliefs.
The expressivists see this apparent implication. They also correctly perceive that (2) and (3) are obviously wrong. So they experience cognitive dissonance. Ordinary folks would relieve the dissonance by relinquishing (1). But they instead cleverly devise a convoluted alternative: alter their semantic theories, i.e., give up the most natural (and *in fact* correct) semantics for sentences like (2) and (3).
So, that’s the nicer way of expressing what I had in mind when I said, “And the reason they’re doing this is that if they stated their views forthrightly (like Ayer and Stevenson did), people would see that they were obviously wrong.”
No doubt the expressivists believe they have good reason for doing this. I think it’s a misguided methodology, into which all too much of contemporary philosophy has fallen. But that’s too much to go into now.
(2) Robert Johnson says:
Of course I can’t say enough about that here to be convincing (for more, see Ethical Intuitionism!). But my view is that the Frege-Geach problem is a much sharper problem initially than the problem of moral motivation for realists. That’s partly because I think the Frege-Geach problem relies on more objective data (in an epistemological sense of “objective”)–the claim that “The verdict was unjust, or S didn’t kill N” makes sense is a more objective piece of evidence than the claim that moral judgements are intrinsically motivating. I think the Frege-Geach evidence is pretheoretically uncontroversial. I think internalism about moral motivation is a very disputable assumption.
I also think it’s false that cognitivists have a hard time explaining the motivating nature of moral beliefs even *if* you assume internalism. I think the idea that cognitivism has trouble here derives from a philosophical theory (which I happen to think is an awful theory…), rather than from some prephilosophical datum.
(3) Mark van Roojen writes:
Yeah, your point is well-taken. I may just be expressing my disbelief that any of the attempts to escape Frege-Geach work, together with my judgement that the objections to realism (and in particular, to ethical intuitionism) were never all that good. This may be a somewhat subjective judgement on my part, but it just seems to me in general that far, far more effort has been expended, and far greater lengths gone to, in the direction of trying to save expressivism than was ever expended investigating ethical intuitionism. It seems to me that the profession was eager to reject intuitionism and did so quickly and with little investigation, while they have doggedly persisted in trying to keep expressivism alive despite initially more serious difficulties and despite little success in dealing with them.
Of course, I can agree that the situation is much like that with many other philosophical theories, without being comforted by that. It seems to me that philosophers in general have made very little progress, and I wonder if this is due to some basic methodological problem.
Exchange between me and Jussi:
ME (in book): the cognivist has a single object, the proposition *that if stealing the candy was wrong, then God will punish you*, and a single mental state (a belief) that can be directed towards that proposition.
JS: That is slightly misleading way to preserve compositionality and construct ‘if p, then q’ truth-functionally from separate parts of p and q. If the single propostion is structured, then the expressivist can use the same structure to specify the relevant attitudes.
ME: Not sure what you mean by your first statement. I thought I was preserving compositionality in the standard way. I think I just have the standard view of conditionals, and then I apply it also to the case where you have a moral proposition as one of the components. Where we disagree is that I don’t think the expressivist can just “use the same structure to specify the relevant attitudes.” I don’t understand the idea there. I’m not sure what you mean by “structure”–you mean the logical structure of the proposition? But that doesn’t specify the *attitudes* at all (in either case). The relevant *attitude* in the descriptive case is *belief*. But presumably the expressivist is not going to say, “yeah, just take the same attitude in the case where you have an evaluative component,” is he? (Yes, I know I’m ignoring the big debate above about “belief”, “BELIEF”, and so on. I’m assuming that beliefs require propositions, and that expressivists deny that there are evaluative propositions.)
(Second issue)
JS: Say I utter ‘Pleasure is good or pigs fly’. Why can’t this complex utterance express commitment to either approving pleasure or believing that pigs fly?
ME: Not sure I understand that either. I was assuming that “E or D” was going to express a mental state that you have. Is the “commitment” you speak of a mental state that you have? Or are you saying that you’re only telling the audience that you have one mental state or the other, while not telling them which you have? Or something else?
What Blackburn says is roughly this: In saying “A or B”, one is expressing a disposition to accept A, if B should be closed off to one.
This strikes me as absurd. I believe that cows have four stomachs. This entails that either cows have 4 stomachs, or space aliens will soon invade Denver (inclusive or). So I also accept (Cows have 4 stomachs, or space aliens will soon invade Denver). But obviously I have no disposition to believe that space aliens will soon invade Denver, should I discover that cows don’t really have 4 stomachs.
Michael,
I found that dispositional interpretation of Blackburn in the book and here too really odd and can’t find any evidence from Blackburn’s text for it. It’s different to say that one is committed to believing one of the disjuncts if one were to reject the other than to say that one is disposed to do so. The former is to say that one oneself thinks that one ought to do so whereas the latter is a prediction of what will happen if one finds oneself disbelieving the other conjunct. In this way it makes perfect sense to say that if I accept the disjunction that ‘either cows have 4 stomachs or the aliens will soon invade Denver and I find that the first disjunct is false, either I must accept that aliens will soon invade Denver or stop accepting the disjunction. Otherwise I am committed to a contradiction. But, this is not a prediction of what will happen.
Also, something I’d like to ask about your view about disjunctions (and conditionals). If they are single propositions and we believe them, wouldn’t it imply that by believing a disjunction we are in a mental state that aims at representing the world in the one, true way the world is. I mean propositions are representations of the world and having a belief is thinking that world just is in some way. Now, in what we do we believe that the world is if we believe a disjunctions?
The problem is that there are many ways in which the world can be to make a disjunction to be true and by believing the disjunction we haven’t committed ourselves for the world to be in any one way of these. So, you would think that something else must be going on than single ordinary propositions and beliefs with that content.
If someone were to assert that subjunctive conditionals require propositions for consequents, would that be sufficient grounds to suspect that he was not a native speaker of English, or would we think he was joking?
Hi,
As others have said – Wow, I go away for a few days and this place goes nuts!
A few further follow-ups.
(1) Kris: You may be right that it was never a ‘platitude’ that jade was a natural kind. Partly like Jamie, I am not entirely sure what people mean by ‘platitude’. Its an empirical question whether people at one point even generally believed (forget about whether it was taken to be analytic) that jade was a natural kind, so I am willing not to insist on this point.
I suppose my main point is that it seems to me that it *could* turn out that the stuff which causally regulates our use of a word like ‘belief’ might be disjoint, and that some of the stuff doing the causal regulation might not fit well with some of our theoretical assumptions about the nature of belief. We might in spite of this (for quasi-realist sorts of reasons) have good reason to go on using the word ‘belief’ to refer to all and only those things in the relevant disjunctive set.
Now it *may* be a platitude (in some sense of ‘platitude’) that belief is a cognitive state (in *some* sense of ‘cognitive state’). As far as that goes, though, the expressivist can defend a minimalist reading (as one, but perhaps not the only, reading of ‘cognitive state’). What does seem possible, though, is that the states of mind which causally regulate our attributions of moral beliefs (and which we express when making moral assertions) are in some fundamentally theoretical important way more like desires than they are like descriptive beliefs. Here the expressivist *does* need to tell a story about what this difference is, but its not as if nothing has been said about this. Typically, expressivists invoke Anscombe’s idea of different ‘directions of fit’ and then try to spell out this metaphor in one or another way – Lloyd Humberstone has a nice paper in which he reviews lots of different ways in which the metaphor can be fleshed out. Gibbard has a detailed account of what he calls ‘norm acceptance’, and in some of his work he gives an account in terms of the biological function of the states of mind in question. These are all highly controversial claims, and I think it would be too much for this post to try to resolve them (!) one way or the other. My only point is that there are theoretical accounts of how we should understand ‘cognitive attitude’ in these contexts, and its not clear to me that ordinary folks will turn out (in the relevant technical sense) to accept as platitudinous that belief is a cognitive state *in the relevant techinical sense on offer*.
I assume that you don’t consider the theories and proto-theories of what distinguishes these different accounts of ‘cognitive’ versus ‘non-cognitive’ satisfying, but I am not sure why. If they are coherent, then it does seem like a real possibility that our use of ‘belief’ is causally regulated in the relevant way by both cognitive and non-cognitive attitudes in the relevant technically specified sense. Spelling out the technical sense in more detail and defending the claims made about moral judgments in terms of it is a tall order, but then just see Gibbard and Blackburn, who do go into some detail on these fronts.
(2) Mark S.:
I did take what I said to be in the same spirit as your early posts and the reply you gave there, actually, just for the record.
I like Dan Boisvert’s pejorative examples too, but I think you underestimate the extent to which a view like mine or Gibbard’s or Blackburns could profit from such examples. I agree that my view is different in interesting ways from Boisvert’s, and that our views therefore face different objections. Nonetheless, it does seem that his examples strongly suggest (your alternative readings notwithstanding) that at least some ordinary ascriptions of beliefs do plausibly refer to hybrid cognitive/non-cognitive states, and moreover that the folk would not have a problem with this in these cases. In which case, at least some of this thread’s objections are at the very least overstated.
I agree that my view cannot reap all the dividends from these analogies that Dan’s theory can. That’s because dan’s semantics for moral claims is much closer to the sort of semantics most naturally suggested for sentences with pejoratives than my own (since I agree that sentences with pejoratives are, typically, anyway, better understood in terms of what I call ‘Ecumenical Cognitivism’ (like Dan’s view) rather than ‘Ecumenical Expressivism’, like my own view. So I agree that he can perhaps reap a mored direct dividend from these examples than me. I deny only the claim that expressivists cannot use his examples to fend off at least some sorts of objections as far too quick.
(3) Michael H.:
You say:
“Actually, what I think is that they have a view that entails that “torturing babies would be ok if everybody approved of it” *doesn’t make sense*. (That’s because I think the subjunctive conditional requires a proposition as consequent.)”
Well, of course, if it were true that such sentences made no sense given expressivism, that this would be a serious objection indeed. However, expressivists do give a semantics for such sentences, and they do argue for them, so you’d need to say more about where those accounts go wrong.
Your objection, as put here, is just the assertion that the consequent of a subjective conditional must be a proposition. So far, I am not convinced, since an expressivist can say that the consequent expresses a proposition in one sense of ‘proposition’ – namely the proposition that torturing babies is OK. Its just that we need to countenance a suitably deflationist reading of ‘proposition’ too. Some of the earlier comments on Wright’s work in this area is relevant here. This is related to your exchange with Jussi, though, and I won’t go into all of that now. Its also not at all obvious to me that its any kind of prior constraint on a semantic theory that it accomodate such a theoretical claim, anyway – we may well need a separate argument for your claims about the semantics of subjunctive conditionals too.
It does seem that the crux of your objection relies on the thesis that expressivists must deny that moral utterances express propositions, but they have taken great pains (most notably in Blackburn’s case, but in some of Gibbard’s work too) to explain how they can say just these things. So to avoid begging central questions we’d need some account of why they cannot consistently say them.
You end by calling into question the methodology of expressivists. Here is a methodological question for you: Suppose it did, contrary to your own arguments, turn out that moral utterances in ordinary languages expressed non-cognitive states of mind after all (in one of the senses of ‘non-cognitive’ to which I alluded in my reply to Kris – you can, of course, give different answers corresponding to each precisification of ‘non-cognitive’ – I know I gave more than one so you may give different answers depending on which is the most promising for expressivists). What methodology would you advocate for deciding whether to infer that (a) there are no moral beliefs, or (b) moral beliefs turn out to be rather different, in the ways expressivists have insisted, from ordinary descriptive beliefs.
Or do you insist that it simply couldn’t turn out that moral utterances expressed non-cognitive states of mind in any of the relevant senses? That seems hard to believe, somehow; the expressivist semantics seem at least like a possible semantics, after all (and Gibbard devotes a long time in THINKING HOW TO LIVE to showing this is so – he gives a ‘possibility proof’ in the first part of the book). If it is possible, then a good and thorough philosophical methodology should give us some guidance on how to proceed in that case. I guess from what you’ve said so far that you’d advocate that we just conclude that moral discourse is too infected with error to be worth keeping, but that is a big guess. Basically, I am very unsure about your own philosophical methodology on this point.
It seems like a fair question if your complaint is methodological, for the expressivist moves to which you object are in fact motivated by the idea that moral utterances do in fact express non-cognitive states of mind. Your worry is that it would be bad methodology to try to preserve a sense in which there are still moral beliefs if this turned out to be true. I wonder why you think this is so.
I suspect your real worry is not so much methodological as substantive. But that depends on what methodology you would advocate should it turn out that moral utterances in ordinary language do conventionally express non-cognitive states.
Regards,
– Mike
I’m going to insist on the middle road, again.
I agree with Mike Ridge, against Mike H., that the correct semantics for counterfactual conditionals is itself one of the important issues in play in the debate between expressivists and cognitivists. And I agree with Mike that Blackburn and Gibbard have been explicit about this. It’s question-begging to simply assert that their semantics is false, because it is part of their view. We need independent evidence to reject it.
But I disagree with Mike’s assertion that either Blackburn or Gibbard has offered any well-developed expressivist proposal about the semantics of counterfactuals. I’d be enormously grateful if someone could point me to one. What Blackburn has insisted on most clearly, is that however it works, counterfactual conditionals with normative consequents will express some kind of noncognitive attitude. That is the only assumption he needs in order to fend off these kinds of objections. But he hasn’t really, I think, offered a serious compositional semantics for counterfactuals. Nor has Gibbard, even though in THTL, his ‘proof’ of strong supervenience strictly speaking requires some kind of semantics for the sentence stating strong supervenience. I think we can read some ideas into the kinds of thing that they say, but I don’t see any clearly developed proposal.
If there were an explicit expressivist proposal for the semantics of counterfactual conditionals on the table, then presumably it would yield predictions about how counterfactual conditionals with purely descriptive antecedents and consequents work, and we could test it on the basis of those predictions. As far as I know, no one has done this. (Although, just to put in my plug, I discuss two proposals about how it could be done in the final chapter of Being For.) The same problem arises for tense. In fact, at the end of his 1973 paper ‘Moral Realism’, Blackburn considers the problem for the past tense, rather than discussing the modal version. Since tense is plausibly even more complicated than alethic modals (see Jeff King’s recent paper in Philosophical Perspectives), I’m even less optimistic about how an expressivist treatment of tense will go.
Here are two rather different objections to expressivism.
1. We use (here there seems to be some dispute about the proper term but I’ll go with Mark and call them) counterfactual conditionals with normative consequents, so expressivists owe an explanation for how these work and nobody has yet given one.
2.1 Counterfactual conditionals require propositions in their consequents.
2.2 But according to expressivism normative sentences do not have propositions as their semantic values.
2.3 There are counterfactual conditionals with normative consequents.
2.4 So expressivism is false.
I fully agree with (1). But Michael Huemer appears to be offering (2). Mike Ridge explains why (2.2) is doubtful. But I gave a (very obvious) counterexample to (2.1). (It looked like I was asking a question, but actually I was giving the counterexample — sorry if that was too cute.)
As to (1): while this is a good point, it would be a lot more worrisome for expressivism if there were some established and secure semantics for counterfactual conditionals. But there isn’t. In my opinion, the closest we’ve got is Dorothy Edgington’s account, which is itself a kind of ‘expressivism of conditionals’ (Edgington argues that conditionals in English standardly do not have any truth conditions).
Hi, Jamie.
I don’t think your counterfactual embeds a question in the consequent; it looks to me like you’re asking which counterfactual is true.
In his book Renewing Meaning (p 38), Stephen Barker has some examples of questions embedded in conditionals, where he has specific arguments that the ? can’t be understood as scoped over the whole conditional. The first one is: ‘If this coin is tossed, how likely is heads?’ And the second is, ‘If it is raining outside, why is it that we cannot hear the rain?’
He thinks the answer to the first question should be a conditional probability, but that if the ? was scoped over the conditional, then the answer would be the probability of the conditional. Since they’re not the same, that’s the wrong reading for the question. His case for the second one is less persuasive; it turns on assuming that the conditional is material and that pointing out that the consequent is true suffices to answer a ‘why’ question about a material conditional.
Here are two unsatisfying disclaimer-style replies:
(1) Yes, it’s true that I didn’t give an argument for the statement that counterfactual conditionals require propositions in their consequents. That was intended as (and was) a mere expression of opinion. I haven’t spent much time thinking about counterfactuals in particular, but my thought was that problems like the ones that (I argued in Ethical Intuitionism) confront Gibbard, Blackburn, et al. for indicative conditionals and disjunctions would also occur for subjunctive conditionals. Also, I misspoke (or misthought). My considered opinion was that you need a proposition for the antecedent of a conditional. (Thus, you can say, “If the roast is done, then turn off the oven,” but not “If turn off the oven, then the roast is done.”)
(2) When I said “… philosophers in general have made very little progress, and I wonder if this is due to some basic methodological problem,” I didn’t have a worked-out alternative methodological proposal, or even a very specific methodological critique in mind. Rather, I had two vague thoughts. One: Something must be going wrong in philosophy, because it is so much less successful than science, mathematics, and the like. Two, slightly more specifically, it seems to me that in science, adding more epicycles onto a theory is strongly discouraged, and at some point (not very far along) it leads other scientists to reject one’s theory even if the theory accomodates the data; but in philosophy, one can add epicycle after epicycle and we say you’re just doing good philosophy. But I don’t have a lot to say about that right now.
However, I do want to say something more (and hopefully more satisfying) in response to Jussi.
(3) How to understand Blackburn’s “commitments”? Perhaps a commitment to A is a disposition to A; or perhaps it is a belief (or whatever the attitude is) that one should A.
I think that things were portrayed differently in different of Blackburn’s writings. I think he changed his view (between Spreading the Word and Ruling Passions, and I think that the disposition view was the later view. Unfortunately, I haven’t got the text in front of me now to enable me to pick out the specific words of Blackburn’s that made me think this. However, what I can say that you might be of interest to you, is that I think Blackburn’s view is problematic on either reading. Also, if I’m wrong about its being problematic on one reading, then I’m also wrong about its being problematic on the other reading. So it just doesn’t matter whether the commitments are dispositions or normative beliefs.
The reason is that I don’t have a commitment in *either* sense–nor, indeed, in any decent sense at all–to accepting that space aliens will soon invade Denver, if I should discover that cows don’t really have 4 stomachs. Just to generalize: my accepting P commits me (just by propositional logic) to accepting (P v Q), no matter what P and Q are. They need have no interesting relation to each other, and I need have no interesting attitudes about their relation. But it strikes me as very implausible to claim that, for any P, if I accept P, then I am (if I just follow the rules of logic) going to be committed to accepting any arbitrary other claim Q, should P be refuted. I don’t have such a commitment in the dispositional sense, nor in the normative-attitude sense.
Now, you can add on the clause (following Blackburn), “or withdraw that very commitment,” resulting in the claim that when I accept any claim P, I am (if I am logical) committed to accepting any arbitrary claim Q *or* to withdrawing this very commitment, should P be refuted. But then, as you know, my argument would be that this renders the hypothesis of such a commitment empty. We could talk more about that if you want.
(4) Jussi asked:
One might respond to this by just saying there are disjunctive ‘ways of being.’ For example, suppose I believe that Jon’s hair is either brown or black. [Jon’s hair being brown] is a way the world might be. So is [Jon’s hair being black]. And I might just say: here’s another way the world might be: [Jon’s hair being brown or black]. (In this particular case, I might say that this amounts to Jon’s hair having the disjunctive property, being-brown-or-black.)
But you might not like that. Suppose that you reject disjunctive “ways of being”. In that case, I would instead ask you to question the assumption that a proposition always represents the world to be some (unique, specific) way. Some propositions represent the world, not as being some one particular way, but as being *either* one way or another. (Something similar can be said about negative propositions: rather than representing the world as being a certain way, they represent the world as *not* being a certain way.)
Lastly, apologies to those to whom I didn’t respond sufficiently. I’ll try to find time to respond more adequately later.
Michael,
thanks a lot for the comments. Now, I’m even more puzzled though so I guess I do want to talk more about it.
We started looking at the idea of what accepting claims like
PvQ
commits us to. The Blackburn proposal was that in the case that one comes to believe that one of the disjuncts is false on has to by one’s own lights either accept P, full stop, or stop accepting the disjunction PvQ. So, in the case you have accepted, as you claim, that either cows have 4 stomachs or aliens will invade Denver, if you find out that cows have 5 stomachs, then you are committed to either believing that the aliens will invade Denver or stop believing the disjunctions. But, you cannot go on accepting the disjunctions. By your own lights. Otherwise you think that it is okay to accept contradictions (P from (PvQ)and~Q and ~P). If you accept that then logic goes overboard anyway. That sounds very good to me and nothing you say here:
“The reason is that I don’t have a commitment in *either* sense–nor, indeed, in any decent sense at all–to accepting that space aliens will soon invade Denver, if I should discover that cows don’t really have 4 stomachs. Just to generalize: my accepting P commits me (just by propositional logic) to accepting (P v Q), no matter what P and Q are. They need have no interesting relation to each other, and I need have no interesting attitudes about their relation. But it strikes me as very implausible to claim that, for any P, if I accept P, then I am (if I just follow the rules of logic) going to be committed to accepting any arbitrary other claim Q, should P be refuted. I don’t have such a commitment in the dispositional sense, nor in the normative-attitude sense.
Now, you can add on the clause (following Blackburn), “or withdraw that very commitment,” resulting in the claim that when I accept any claim P, I am (if I am logical) committed to accepting any arbitrary claim Q *or* to withdrawing this very commitment, should P be refuted. But then, as you know, my argument would be that this renders the hypothesis of such a commitment empty. We could talk more about that if you want.”
seems to contest anything in the Blackburn view. It seems to be just to be saying that you haven’t accepted the disjunction we started with.
I don’t have a problem with disjunctive ways of being but that seems to be just for the expressivist who is going to say that some of these disjunctive ways of being are planned ways of being.
Mark,
I don’t think the (any) counterfactual question asks which counterfactual is true, because I don’t think counterfactuals have truth conditions. So we could say that counterfactual questions ask which counterfactuals to accept. But that, it seems to me, is consistent with the idea that in a counterfactual question, the question is being asked on the (suspected false) condition; that is, when you answer it you’re supposed to pretend or assume that the antecedent is false, and then decide what the answer is to the consequent question. (This is a version of ‘trying on’ for conditionals.)
I think this approach handles Barker’s conditionals correctly (I should definitely read his book). I imagine the coin is tossed and ask myself how likely heads is. I imagine that it’s raining out and ask myself why we can’t hear the rain.
I completely agree with Barker that the answer to the coin question must be a conditional probability. The way we make judgments of conditional probability is (roughly) by supposing the antecedent and assessing on those grounds the probability of the consequent. I couldn’t understand this:
Do you mean that to end: “… that the consequent’s truth suffices to answer a ‘why’ question about the material conditional”? But it doesn’t, so I don’t follow.
In any case, I say (for now) that the consequent is indeed a question, and that we are supposed to answer it ‘on condition’ of the antecedent, meaning, while pretending the antecedent is true.
Quick question:
Doesn’t it seem obvious that
(1) It is necessary that ~P or Q.
entails:
(2) If P were the case, then Q would be the case?
(1) Clearly has truth-conditions. So doesn’t it follow that (2) must as well? (Given that (1) entails (2)?)
What’s the motive for thinking that counterfactuals don’t have truth conditions?
Kris, briefly:
No, I don’t think that being entailed by something with truth conditions entails having truth conditions. Why should it? (Here’s a formal example. Let ‘e’ be an empty name and ‘R’ a monadic predicate, so that ‘Re’ has no truth conditions, at least on one standard view. But ‘Re’ is entailed by ‘(x)Rx’, which does have truth conditions.)
The reason to think that counterfactual conditionals have no truth conditions is that it looks like the probability of a counterfactual conditional is a conditional probability, and there conditional probability is not the probability of any conditional.
See Dorothy Edgington’s ‘state of the art’ paper in Mind 104:235-329.
Hi, Jamie –
I’ll let you interpret and assess Barker’s argument yourself; as I was reading his argument about the second conditional, it goes: suppose that the question is about the conditional. Then people in a closed room who can’t hear any rain who asked, ‘if it’s raining outside, why is it that we can’t hear the rain?’ could answer that question by pointing out that they can’t hear the rain, and _that_ is why the conditional, ‘if it’s raining outside, then we can’t hear the rain’ is true. But people in a closed indoor room who can’t hear any rain would _not_ be satisfied with this answer, so they can’t be asking a question about a conditional. Hence, in the example sentence, the question is embedded in the consequent, rather than scoped over the whole conditional. I wasn’t convinced by this; I don’t see why we should think that pointing out the truth of the consequent counts as an explanation of a conditional truth.
Hi Jamie,
This is probably pretty far off topic, and I apologize…
You wrote:
“Here’s a formal example. Let ‘e’ be an empty name and ‘R’ a monadic predicate, so that ‘Re’ has no truth conditions, at least on one standard view. But ‘Re’ is entailed by ‘(x)Rx’, which does have truth conditions.)”
Two thoughts: I believe that logical systems for which ‘(x)Rx’ entails ‘Re’ are also systems in which we assume either that (i) all names are non-empty, and hence “Re” is true or (ii) sentences with empty names can nonetheless be true or false. (I’m happy to be schooled on this.)
Second: What do you mean by “entail”? You clearly don’t mean that P entails Q iff Q is true in all worlds in which P is true.
Mark,
Ohhhh, I see. Well, no disagreement from me; I never would have thought that the conditional question was asking about the truth of a material conditional.
Kris,
Sorry, I thought we were talking about logical consequence. You mean semantic entailment, as in, no world in which the premises are true and conclusion false. (So am I right that the premise, “(1) clearly has truth conditions” is doing no work? Anything that is entailed, in your preferred semantic sense, must have truth conditions?)
Then I guess my answer is that what seems obvious to you (that the necessary disjunction entails the counterfactual conditional) doesn’t seem obvious to me; it seems like a theoretical question. What do you think about indicative conditionals? Do you say that they must have truth conditions because they are entailed by necessary disjunctions?
Jussi,
Sorry, I think I need to explain more explicitly what I had in mind. You say:
Right, I’m not disputing that. I’m not disputing that, in a perfectly good sense, if I accept (P v Q), then rationality commits me to accepting Q or giving up (P v Q), should P be refuted. What I’m disputing is that this commitment explains the meaning of disjunction.
Suppose someone says to me: “Mike, in exchange for $1, will you agree that if cows turn out to have 5 stomachs, you’ll either jump off a cliff, or renounce this very agreement?”, I guess I would say “Yes.” Now, the point of this example is that the “agreement” here is an essentially empty one. I haven’t really committed myself to anything. Obviously, I’m not committed to anything if cows have 4 stomachs, but also, if cows turn out (surprisingly) to have 5 stomachs, I’m still not really committed to anything. If I want to jump off a cliff when that happens, I can do so, and say I fulfilled the agreement. But if I want to not jump off a cliff, I can refrain from jumping, and still say I fulfilled the agreement (because I fulfilled the second disjunct of renouncing the agreement). No matter what I choose to do, I won’t have violated the agreement. And so, as long as the terms of the agreement take this form (with the escape-clause disjunct on the end), I can embrace any number of such empty “agreements” or “commitments” for free. It’s sort of like undertaking a commitment to either do A or not.
Now, the commitment that you have when you accept (P v Q) can be rendered non-empty, if there is some independent, further substance to the accepting of a disjunction. The commitment to either accept Q or do X, should P be refuted, is a meaningful (non-empty, substantial) commitment, if X is something more than just giving up this very commitment.
So, suppose that accepting (P v Q) is something more than just having the commitment Blackburn speaks of — suppose, for example, that it is a matter of believing a specific proposition, the proposition that is true in exactly the union of the set of possible worlds in which P is true and the set of possible worlds in which Q is true. Then committing to [accepting Q or ceasing to accept (P v Q)] is committing to something more than just [accepting Q or going back on this commitment]. So, on my view of the disjunction, the commitment we’re talking about becomes non-empty.
It now occurs to me that, as I neglected to say in the book, there’s a further problem here: on Blackburn’s view, it would seem to be (virtually?) impossible to violate the commitment in question. On my view, it is straightforwardly possible. That’s because it’s (virtually?) impossible to violate a commitment to [do A or withdraw this commitment], since if one doesn’t do A, one will presumably (?) count as withdrawing the commitment to do A. But on my view, one clearly can violate the relevant commitment — one just has to continue to accept ~P and (P v Q) while refusing to accept Q.
I take it that my view is the right one — it’s not impossible to accept ~P and (P v Q), but deny or withhold Q; it’s just irrational. (Side note: set aside the kinds of cases Foley discusses in “Justified Inconsistent Beliefs”; assume the relevant probabilities are all high enough.) It’s also not impossible to reject disjunctive syllogism; again, it’s just irrational. But on Blackburn’s view, it looks like it’s impossible to reject disjunctive syllogism, since rejection of this inference pattern precludes one’s genuinely using the disjunction concept.
Now I hope I didn’t just create more confusion.
Brief comments on conditionals:
(a) I don’t find the Barker argument about the conditional probability question all that interesting, because it’s straightforward–as the argument itself admits–to interpret the question as a non-conditional question. That is, the question, “If A, then how likely is B?” can be interpreted as, “What is P(B|A)?” The latter seems to me to be a categorical question. (A couple of rephrasings might make it seem more so: “About the function P(*|A): what value does this function assign to B?” Or: “What is P(A&B) divided by P(A)?”)
(b) Jamie, could you clarify this:
? Suppose (A -> B) is a counterfactual conditional. P(A -> B) is equal to what conditional probability? (Surely not P(B|A).) Also, I thought it was a common view (at least Adams defended it) that the conditional probability P(C|D) in fact is the probability of the indicative conditional, P(D > C). (N.B., of course not of the subjunctive conditional and not of the material conditional.)
(c) I find this a more plausible example of a conditional question: “Jamie, do you think there are conditional questions? If so, what would be an example of one?” The second quoted question is a conditional one, because the question is only posed on condition that the respondent thinks so-and-so. (In the example in (a), the question is just posed straight-out, not conditional on anything.)
Michael,
That’s very helpful and clear. I’m not sure though what the problem is supposed to be for Blackburn anymore. It looks to me like when it comes to the commitments he is on a par with the traditional account. I take it that according to him acceptance of a disjunction is the commitment to ‘really believe’ one of the conjunct in the case one rejects the other. You say that this does not capture the content of believing a disjunction. That may be but I don’t see the problem you refer to.
This commitment seems violatable. It seems possible to accept not to disbelief both disjuncts at the same time (what counts as such acceptance? Can’t be much. Sincery saying that ‘I do’ should suffice), but not ‘really’ to believe the other disjunct when coming to reject the other. And, I don’t see how refusing to believe the other disjunct would as such count as withdrawing the conditional commitment unless believing contradictions is impossible.
So, this is not impossible but, as you say, irrational. Of course you can get yourself out of the contradiction by withdrawing the commitment of the disjunction (accept one disjunct, when rejecting the other). But, this seems to work similarly in the traditional account. I don’t see how this would make the commitment empty. The actual agreement seems just like a bad analogy.
Michael H.;
(b)
You asked two questions; I’ll answer the second one first.
Sorry, I put it in a confusing way.
The acceptability (believability, assertibility…) of an indicative conditional is its conditional probability (that’s Adams’ thesis, and it’s hard to deny). But there is no proposition (that’s a function of D and C) whose probability is P(C|D). David Lewis proved this in 1976 (“Probabilities of Conditionals and Conditional Probabilities”, Philosophical Review 85, 297-315) and his proof has been improved several times since then (for intelligibility and generality). So, indicative conditionals do not express propositions.
Now the first question:
It depends on which counterfactual conditional you mean, I think. As a general rule, had/would conditionals are past tense do/will conditionals. This should be obvious from the grammar, though it struck me as a revolutionary idea when I first read it. For instance, suppose I tell you:
(d/wl) If you go in there you will be hurt.
Alarmed, you stay out and a moment later the ceiling collapses. So I was right:
(h/wd) If you had gone in there you would have been hurt.
Surely (h/wd) is the past tense of (d/wl).
Therefore, the acceptability of (h/wd) ought to be some temporally shifted conditional probability, and in the central cases it is: it’s the probability at the time of the antecedent of the consequent given the antecedent. (Thus, the probability at the time I warned you of your getting hurt given that you went into the room.)
There is a problem for some non-deterministic cases, which Stephen Barker pressed in the late 1990s, but I’ll skip the complications that his cases require. (In a nutshell, they force the conditional probability to incorporate information about causes.)
(c)
Yes, that’s a good conditional question.
About conditionals:
Jamie, you say Lewis showed that “there is no proposition (that’s a function of D and C) whose probability is P(C|D).” I’m not sure I understand the parenthetical. Do you mean a proposition that’s a truth-function of D and C, so that it could be defined by a truth-table? If that’s what you mean, then I can see how Lewis might have proved what you say (though I’m still curious about how the proof goes). But if that’s not what you mean, then I don’t see how one could prove what you say Lewis proved. Perhaps I’m confused, but it seems to me defensible to think that the probability of an ordinary English indicative conditional just is the conditional probability (while of course denying that the indicative conditional is truth-functional). That seems especially defensible if the believability of the indicative conditional is determined by the relevant conditional probability. Of course, you might just tell me to read Lewis’ paper.
I’m also puzzled about the claim about subjunctive conditionals. So you say to me, at time t2, “If you had gone in there (at time t1), you would have been hurt.”
The claim is that the acceptability (at t2) of this counterfactual varies with the value, at t1, of P(you will be hurt|you go in there). This initially seems wrong to me: suppose that at t1, you had no reason to think that I would be hurt if I went in there; you had no knowledge or evidence of the impending roof collapse. But now, at t2, you of course know about the roof collapse. So at t2, you rationally assign a very high probability to “If you had gone in there, you would have been hurt.” But at t1, P(you will be hurt|you go in there) was very low, and you (at t2) know that.
Of course, I’m using epistemic probability in saying that. So you might want to say, no, you have to use physical probability. At t1, there was a high physical probability of my being hurt given that I went into the building. But then the *acceptability* at t2 of the counterfactual can’t be determined merely by this *physical* conditional probability, since (if we modify the case a little) you need have no epistemic access to this physical probability, even after the fact.
Another problem–maybe this is the Barker objection?–is that the physical probability at t1 of the roof collapsing might have been only, say, .1 (this is an indeterministic world); and yet, at t2, having seen the roof collapse, you can say for certain that if I had gone in there, I would have been hurt.
These things are all pretty complicated and deserve more than a blog comment’s worth of explanation, but I’ll give mini explanations. The Edgington paper I cited should be a good source for the developed arguments, and I think I can find a good presentation of the Lewis proof. Hm. Al Hajek has a paper in the Skyrms & Eels volume Probability and Conditionals. And Jonathan Bennett’s book has a good presentation, too.
First: no, not just a truth function. There is no connective, *, such that pr(A*C) = pr(C|A). No connective, truth functional or otherwise. Yes, David Lewis proved this. The proof is not obvious, to say the least, so the fact that the contrary seems defensible to you does not really show that you are confused.
Second: right, for the counterfactuals it’s the objective probability, not the rational credence that you or I might have had. When I note that you would have been hurt had you entered the room, I find the conditional acceptable (and expect that you will) because I have new information.
Right.
Here’s the official story:
I think this paper is available at her web site (if not then I can’t imagine where I got it!).
Thanks, Jamie. The Edgington view sounds good. She doesn’t appear to have a web site (other than a brief bio page at Oxford), but I think I discovered where you found her paper. In case anyone else is interested, it’s at:
http://www.nyu.edu/gsas/dept/philo/courses/content/papers/
edgington.pdf
Jussi,
Suppose you meet someone whom I shall call Contraman. Contraman says, “I think one of the two propositions, either A or B, is true.”
You say: “Oh, so if you discovered that B was false, you’d approve of (or be disposed to, or whatever) your accepting A, or giving up the belief you just mentioned?”
Whereupon Contraman replies, “Oh no. I feel perfectly fine about accepting contradictions. I wouldn’t approve of accepting A, or making any other change.” Or he might say, “You know, I don’t particularly care what I would do in such a situation. I don’t feel any particular sentiments, either of approval or of disapproval, toward counterfactual actions in that scenario.”
Now suppose, if you can, that Contraman is sincere. Does Contraman (now, in fact) accept the disjunction (A v B)? Seems to me that he does (and that this whole scenario is possible). But on Blackburn’s view, you’d have to say that Contraman doesn’t really believe the disjunction, because he doesn’t have the right sort of second-order approval attitude.
So, I’d say Contraman merely has unreasonable attitudes (or fails to have some reasonable attitudes). But I think Blackburn would have to say Contraman can’t possibly have the attitudes he reports.