My apologies about the delay and I do recommend also checking out Richard's Arneson's interesting post on moral luck below, but we are finally able to begin the discussion of Jack Woods's Philosophers' Imprint paper 'Expressivism and Moore's Paradox'. To start things off, here is Teemu Toppinen's critical intro to the paper. I'm looking forward to know what you all have to say about Jack's wonderful paper and Teemu's equally brilliant discussion of it!
[Addition – 01/07/14: I've just been informed that a version of Teemu's critical notice will be published by JESP. Wonderful news and I'll add the link to the published paper later.]
Thanks for asking me to kick off the discussion of this terrific paper. Woods offers a neat, sharp argument against expressivism and discusses (and refutes) a number of potential objections. I first quickly outline Woods’s argument, and then explain why I don’t think that it succeeds.
Woods’s argument goes roughly as follows:
(i) If expressivism is true, then the parity thesis is true, that is, sincere moral assertions express desire-like states (e.g., being in favor of or against something) in exactly the same way as sincere non-moral assertions express beliefs.
(ii) The way in which non-moral assertions express beliefs explains why sentences of the form ‘p, but I don’t believe that p,’ where ‘p’ is a non-moral sentence, are Moore-paradoxical.
(iii) If the parity thesis is true, then the way in which moral assertions express desire-like states should explain why sentences such as ‘Murder is wrong, but I’m not against it’ are Moore-paradoxical.
(iv) These latter kinds of sentences are not Moore-paradoxical. So:
(v) The parity thesis isn’t true. So:
(vi) Expressivism isn’t true.
Woods’s focus is on arguing against the parity thesis. The conclusion that expressivism isn’t true is drawn only very tentatively. More precisely, then, the conclusion of the paper is that given the plausible assumption that expressivists should accept the parity thesis, expressivism isn’t true. (Two minor things: (a) Woods at one point (p. 6) seems to suggest that the familiar expressivist accounts of the expression relation, such as the one proposed by Mark Schroeder (e.g., in Being For), would involve rejecting the parity thesis. I’m not sure that this is right, but this doesn’t seem to matter much, as I agree with Woods that something like his objection would, in any case, apply to (at least most of) these accounts. (b) It is perhaps worth pointing out that Woods’s argument also seems to threaten some ‘ecumenical’ forms of cognitivism and besire-views.)
My main worry about this argument is due to its focus on moral assertion, in particular. To anticipate my main concern, I think that we should reject premises (i) and (iii). I’m fine with assuming that expressivists should accept a parity thesis of the relevant kind. If (i) and (iii) were formulated as claims about normative assertion, then perhaps expressivists should accept these theses. But if we make the required modifications to (i) and (iii), (iv) does not seem to be true anymore. And so the modified parity thesis and expressivism can both be saved.
It is best to begin with premise (iv), which Woods mostly concentrates on. Consider the following claims (p. 5):
(11) Murder is wrong, but I don’t believe it is wrong.
(3) Fuck the Yankees! I have no negative attitude toward the Yankees.
(8) Murder is wrong, but I’m not against it.
Woods suggests that (11) and (3) are incoherent in a way in which (8) is not. I’m inclined to agree. A possible, quite plausible explanation for why this is so is this: ‘Murder is wrong’ expresses a belief that murder is wrong, but might not express, in the same sense, any desire-like attitude against murder. This explanation might sound like it would be a bad fit with expressivism, but it seems to me that actually expressivists can accept this explanation. They can do this, assuming that ‘wrong’ is a context-sensitive term, which is often, but not always, used to make a normative judgment, consisting (at least in part) in a desire-like attitude. If something along these lines makes sense, and if the non-normative uses of ‘wrong’ are central and common enough, then it’s no wonder that sentences such as (8) are more easily given coherent readings than, say, (11) or (3).
In discussing possible objections, Woods acknowledges the possibility of broadly this kind of defensive move. He notes that someone might suggest that ‘Murder is wrong’ can be used in an inverted-commas sense, as making a claim to the effect that murder is wrong by the standards of the prevailing society. He also considers the possibility that this sentence could be used in a ‘theoretical’ sense to make a ‘cold and impersonal’ moral judgment. I’ll say some more on these suggestions and Woods’s objections below. But it is helpful to note, already at this point, that these ideas do not exhaust the options that expressivists have with regard to giving descriptive, non-normative readings of a sentence such as ‘Murder is wrong.’ The expressivist could suggest that when we use the sentence ‘Murder is wrong,’ we always, very roughly, characterize murder in relation to standards or norms of a certain kind. Perhaps the relevant kinds of norms are norms for when to feel guilt and resentment. (Perhaps they should be understood quite differently, but this should suffice for illustrative purposes.) However, just which norms of the relevant kind are relevant might vary from one context to another. In some contexts, ‘Murder is wrong’ might be used to make a claim about the relation of murder to the norms that have currency in the speaker’s own society. In others, this sentence might express, for example, a belief about the norms of some other, salient, society, or about the norms that the speaker used to endorse (perhaps before having suffered a severe brain damage). A very central, plausibly primary, use for this sentence would be that of relating murder to standards that the speaker herself endorses, or subscribes to. This, the expressivist might say, would be the genuinely normative use of the sentence. It would be this specific sense of ‘wrong’ which the expressivist would wish to give an expressivist account of. What’s been suggested here is very much inspired by Michael Ridge’s recent, relatively detailed and sophisticated account of the meanings of ‘good,’ ‘ought,’ etc., in his Impassioned Belief (OUP, 2014), but the basic idea of there being a variety of descriptive uses for terms well suited for making normative judgments is, of course, a familiar one.
If this line of response were correct, then there should also be senses of ‘wrong’ which would make sentences like (8) come out as incoherent. But this seems like a nice prediction to me. Let’s begin with some normative claims that aren’t specifically moral, and that are not so easily given non-normative readings. Consider:
(12) All things considered, we ought not to murder, but I’m not in any way against murdering.
(13) There’s some reason not to murder, but there’s nothing about murder that I’d be against.
These seem incoherent to me, just like, for instance, (11) or (3), above. That’s a neat prediction of expressivism. How about wrongness in a normative, ‘reason-implying’ sense?
(14) Murder is wrong – that is, wrong in a sense which implies that there are extremely weighty reasons not to murder –, but I’m in no way against murder.
Again, this seems clearly incoherent. So, it seems fine to allow that there are such readings of ‘Murder is wrong’ which would render something very much like (8) incoherent. (‘Murder is wrong [in the reason-implying sense], but I’m not against it’ might still sound coherent, because this claim could perhaps be understood as saying that even though there are extremely weighty reasons against murder, it’s still, on balance, okay.)
So, again, (i) and (iii) should only (possibly) be accepted once they’ve been modified so as to make claims about the relationship between normative assertions and desire-like states. However, once we modify these premises in an appropriate way, (iv) does not seem to be true anymore.
As noted, Woods (p. 7) considers some objections roughly along these lines. First, he acknowledges that one might object that the “relative felicity of the examples [such as (8)] is due to an inverted-commas use of ‘wrong’. When so used, ‘Murder is wrong’ means something like ‘Murder is wrong (by the moral standards of the prevailing society)’.”
Woods isn’t happy, at all, with the inverted-commas response. He suggests that “such an aberrant interpretation of the meaning of ‘Murder is wrong’ is implausible without conditions suggesting such an interpretation” (p. 7). I do not think, however, that this is a very aberrant interpretation. For instance, when I teach ethics, there often are people in the audience who have some difficulty seeing how ‘Murder is wrong’ could have any other kind of meaning. So, it seems to me that an inverted-commas reading of sorts is very much available for this sentence. In any case, I’ve suggested that there’s a wider range of descriptive ways of using ‘wrong,’ and not all of these seem aberrant. It also seems quite plausible that utterances of (8) themselves suggest a descriptive interpretation exactly because, in using this sentence, the speaker is reporting that she doesn’t have the attitudes that one would take ‘Murder is wrong’ to express when used normatively, as it is most commonly used.
Now, Woods does consider the idea that “it is the mere availability of an inverted-commas reading that renders the examples coherent” (p. 8). He notes that “inverted comma readings are often indicated by stress, but [(8) does] not require stress to be felicitous.” However, the inverted-commas (or the other non-normative) readings of sentences such as ‘Murder is wrong’ need not be indicated by stress, so this seems irrelevant.
Woods also argues that “the mere existence of an inverted comma reading […] is not sufficient for the coherence of such examples” because “predicates of personal taste, for example, clearly have an inverted commas reading, but ‘Broccoli is delicious, but I don’t like it’ is still strikingly incoherent” without an appropriate stage-setting. Now, it’s true that there just being a sensible inverted-commas reading for the relevant sentences is not sufficient to make them sound coherent. However, such readings may be more or less easily available or accessible to a hearer. ‘Wrong,’ it seems to me, is quite often used in an inverted-commas sense, or, anyway, in some descriptive sense. So, here a descriptive reading is easily available. ‘Delicious’? Not so often, it seems. Now that I have given the sentence ‘Broccoli is delicious, but I don’t like it’ some thought, I can easily hear it as coherent (perhaps the speaker has a history of being a broccoli-lover, or perhaps she’s deferring to those with a more ‘refined’ taste in order to be more helpful). However, this sentence was, at least in my case, somewhat more resistant to a coherent interpretation than ‘Murder is wrong, but I’m not against it.’ I wouldn’t say that one sentence sounds more incoherent than the other, but it’s easier (for me) to hear the latter as coherent. (As for the standard Moore-paradox sentences such as ‘Snow is white, but I don’t believe it is,’ these would seem to require truly aberrant interpretations in order to make any sense.)
So, I don’t think that Woods’s response to the objection from inverted-commas uses succeeds. Also, the objection from inverted-commas uses can be further strengthened by acknowledging a wider range of descriptive uses for terms like ‘wrong.’
Another related objection, discussed by Woods, distinguishes “between cold and impersonal moral judgments issued in the course of theorizing and heated moral judgments issued in the course of criticizing others’ actions or guiding our own” (p. 9). Woods suggests that if uses of terms like ‘wrong’ don’t always express desire-like states, “it is unlikely that such attitudes will play a serious role in the account of their meaning” (p. 9). His proposal is that moral assertions often conversationally imply that the speaker has the relevant desire-like state. However, as argued above, there are other, more expressivist-friendly, options available for capturing the distinction between those uses of terms like ‘wrong’ that express desire-like states and those that don’t.
I’ve focused on what I found to be problematic about this very nice paper. The rest seemed pretty convincing to me. Finally, I cannot resist noting that I appreciated the decadent air of p. 5, which contained, within one spread of a philosophical article published in a top notch journal, uses (or, anyway, mentions) of ‘Yay for drinking a lot of beer tonight’ and a sentence of the form ‘Fuck the x,’ plus, in a footnote, a description of drug-induced hallucinations.
Teemu Toppinen
I would like to suggest a different kind of response to Jack’s argument, namely one that could be offered by a *minimalist* expressivist. Here’s how it goes:
Normative claims express conative attitudes, and also beliefs (understood in a minimal sense) – what Jack calls “small-b beliefs”. In this minimal sense, any discourse that meets certain syntactic constraints (and maybe some pragmatic ones too) expresses beliefs. Ordinary descriptive claims express small-b beliefs, but also big-B beliefs, i.e. robustly representational beliefs. (It’s not entirely clear how to make sense of such a notion of robust representationality within a minimalist framework, but let’s assume our minimalist expressivist has a good account of this. I take it that Jack’s argument is not supposed to hinge on the absence of a solution to the problem of creeping minimalism.)
Now, the minimalist expressivist could go on to argue, ‘p, but I don’t believe that p’ is paradoxical or incoherent only if belief is understood in the minimal sense. The same kind of Moore-paradoxical sentences can be found in both ordinary descriptive discourse and normative discourse, and they involve the notion of small-b belief. Facts about small-b beliefs are so intimately tied to the assertoric surface of the discourse and hence transparent to speakers that one cannot deny that the (minimal) assertions one makes express small-b beliefs without running into some kind of incoherence.
However, the minimalist expressivist might say, the following is not Moore-paradoxical: ‘p, but I don’t have a robustly representational belief that p’. Some philosophers (e.g. Huw Price) deny that any beliefs are robustly representational, presumably without incoherence. A fortiori, I can deny that any of *my* beliefs are robustly representational, even while expressing them. No paradox here.
Similarly, if expressivism is a claim about the deep semantics or psychology of our normative language, the fact that normative claims express conative attitudes need not be transparent in a way that would make ‘p, but I don’t have the conative attitude X’ paradoxical (where the conative attitude X is the attitude actually expressed by p).
Minimalist expressivists could thus account for Moore-paradoxical sentences across the board, in terms of small-b beliefs, without any asymmetry in this respect between ordinary descriptive discourse and normative discourse. And they can still endorse a relevant parity thesis: Moral assertions express conative attitudes in exactly the same way that ordinary descriptive assertions express big-B beliefs.
Jack addresses a similar objection in his paper (Objection 6 on p. 9), but (1) I believe the objection as stated there isn’t fully developed or articulated in its strongest form, and (2) I’m not sure I understood his response to it. So it would be great to hear more from him on this issue.
Camil,
Really interesting and thanks for the comment. I say my piece about this in the paper, but I’ll add a bit here. I suppose I think that it would be weird if cognate expressions such as ”think”, ”hold that”, ”view the world as being such that … is true of it”, ”would describe the world as …-ish”, and such didn’t pick out big-B beliefs. But each of these produces Moore-paradoxicality in the appropriate sentential context. ”Robustly representational belief” doesn’t, of course, but one might just think that robust isn’t the sort of adjective to apply to a psychological state. Or have other some sort of disagreement or lack of understanding with a bit of philosophical jargon like ”robustly representational belief”. The homely ways of picking out representational beliefs yield Moore paradoxicality. The homely ways of picking out desire-like states do not.
Also, though this is less rigorous, I think it guts the expressivist program to say that it is not essential to moral discourse that by engaging in it you commit yourself, in the sense of commit I spell out in the paper, to possessing big-D desire-like states. But this is essentially what you have to say to run the minimalist solution you favor.
I’d like to start by thanking Teemu for his wonderfully apt set of comments. I have some disagreements, of course, but I think he has honed in on an interesting problem for my argument. A few remarks:
(1) My target is specifically Moral Expressivism, not Normative Expressivism. Some bits of discourse that have been given expressivist treatment have no problem with my argument—for example, “It might be raining, but I believe it’s not raining” really is incoherent in exactly the same way as Moore’s paradox. More flatly normative cases such as: “I ought to not murder, but there’s no reason to not murder” or “I ought to not murder, but I plan on murdering” have a slightly different feel. To my ear, the first is borderline incoherent, the second coherent. What about Teemu’s (12-14)? These sound somewhere between borderline incoherent and incoherent to me, but note that all 3 contain extraneous material such as “all things considered” and “in the sense that implies that there are extremely weighty reasons” (13 is the cleanest example in this regard.) I view such material as containing instructions to the hearer to interpret the statement in a certain way—a way that perhaps produces a form of incoherence by means of strongly suggesting or implying that the speaker has the relevant desire-like states. So, for example, consider:
Jack is tall—that is, tall to the degree that implies he shouldn’t live in buildings with low ceilings, but I’m in no way against Jack living in such a building.
This sounds borderline incoherent to my ear, much like (14), but I would be very hesitant to say that predicates like `tall’ have, as part of their competence conditions, a connection with being for or against. Rather, I’ve forced an interpretation of my use of `tall’ which is more directly connected to having reasons for and against, and, consequently, assuming I’m a normal sort of fellow, with my being for or against.
On balance, I’m undecided on the explanation of cases like Teemu’s where you potentially force a more expressive interpretation of ‘ought’, ‘wrong’, or ‘reason’. One reason is that they still don’t sound incoherent in the same way that Moore-paradoxical assertions sound incoherent—a fact that requires an explanation which I believe will not be forthcoming. At a minimum, this is something a hermeneutic expressivist now needs to supply. Another reason is our interpretation of expressions like ‘I’d be against’ can seem rather cognitive—i.e. as stating norms or consequences of norms I agree with, not picking out a desire-like state (see fn. 15 of my paper). This is especially true when they are prefaced with claims about reasons! A final reason for my hesitation will appear in my next comment.
(2) I have no serious objection to the story that Teemu suggests, drawing on Mike Ridge’s Impassioned Belief, where terms like `wrong’ are interpreted as something like “wrong by the relevant standard” and where the relevant standard is fixed by the conversational context. This, as yet, is no expressivist story. What makes it expressivist, presumably, is that there is some connection between my asserting that something is wrong, interpreted in this contextual way, and some desire-like state, and where this connection forms part of the competence conditions of such assertion—at least for certain uses of `wrong’. This latter condition is crucial. Consider, as a way of illustrating its importance, the predicate `poisonous’. What `poisonous’ means is fixed by a set of standards and my assertion that something is poisonous is quite often connected with a desire-like state. Why? Because by the standards present in most contexts, poisonous things endanger our health, it is quite generally assumed that people we’re dealing with have an interest in preserving their health, and ordinary conversational contexts are cooperative. My assertion, then, of “That drink is poisonous” typically informs the target of my assertion that I think they ought not to drink it and, often, that I want them to no not drink it. It has the perlocutionary effect of a recommendation. But these features are indirect, routing through assumed facts about the conversational context. The content of `that drink is poisonous’ is still purely descriptive. And the primary function of my assertion is descriptive, even if the intended effect is to recommend.
The success of the sort of picture that Teemu and Ridge paint, as an expressivist story, will depend on how the connection between moral assertions and desire-like states is spelled out. If it is spelled out like the story I just gave about poisonous, I have no objection, but it is not an expressivist story in any interesting sense. If it is spelled out in a way which makes it part of the competence conditions of a predicate like ‘wrong’ that there are desire-like states expressed, then I still need an answer to why seemingly competent speakers can explicitly deny possessing these states without incoherence. My view is roughly that this indirect method of generating desire-like states is the most plausible explanation of the connection between assertion and desire-like states in the moral case; it is possibly the most plausible explanation in the normative case as well—hence my hesitation in (1).
(3) Finally, on the question of inverted-commas readings of `wrong’, I’m worried that if inverted-commas readings are as common as Teemu suggests, in what sense are they inverted-commas readings? Wouldn’t they just then be the meaning of wrong? The Ridge-style story mooted by Teemu certainly suggests so.
Maybe the best sense to be given to this defense is that if wrong is interpreted Ridgely (certainly not Rigidly), an inverted-commas reading would be a case where I do not endorse or accept the prevailing standards and hence do not have the required desire-like states. Teemu suggests that when I hear a case like my (8), I infer that this is how ‘wrong’ is being used, which is why it does not sound incoherent. But the speaker’s intentions and endorsement of various norms and conventions aren’t sufficient to block incoherence, even if easily inferred. Compare a similar case—I am from Boston, a place slightly notorious for blunt, bordering on rude, assertions. When I say “I feel like crap” to a polite inquiry about my health, people balk; this balking is not undermined by their easy inference that I reject the convention of saying something like “I’m good” to “How are you?”. I need to signal in some sense that ordinary conventions aren’t in force. This was my point about inverted commas readings—they need to be signaled, otherwise they produce incoherence when the ordinary role of an assertion is contravened. So I do not think that the availability of these readings is sufficient to explain the coherence of my examples.
I want to thank Teemu again for his really great commentary and for getting me to rethink some of these issues.
Thanks for the paper Jack, and the summary Teemu.
My short reaction is that (8) is Moore-parodoxical, and that premise (iv) of the argument is false. I’d be interested to hear if this reaction is esoteric.
Of course, there are various ways of reading (8) where it looks fine, such as where we understand ‘wrong’ in an inverted-commas sense (cf. pp.7-8). Further, principles of interpretive charity might encourage us to read in it one of these more acceptable ways. Further still, if there are very many ways to interpret (8) as coherent, and comparatively fewer ways to interpret (11) as coherent, (11) might seem worse. But these facts are all just distractions from the central question of whether the most straightforward reading of (8) is incoherent, and it seems to me that it is. To repeat, I may be esoteric.
Hi Alex
I don’t think that you are esoteric. I think I have the same kind of an intuition about those sentences. Jack has a response to this – Objection 1. He says that we may find the examples to have ‘a whiff of paradox’. He then says that this supports his view because we don’t find the cases in the same way strongly incoherent. But, I’m not sure I’ve got a grasp of the difference between a statement feeling strongly paradoxical and it only feeling slightly paradoxical.
Teemu,
I very much liked your response. I had one question about it. I think it’s fine for everyone to accept that moral (and perhaps also normative) sentences have different uses. So, the expressivist can claim that many uses of these sentences express plans but at the same time many other uses can express beliefs/describe. Likewise, the cognitivist/realist can say that many uses of these sentences express beliefs/describe the world but at the same time many other uses of these sentences are expressive of desire-like states.
At this point there is no disagreement. Describing these theories as conflicting requires taking some of the uses to be *core* uses and then explaining the other uses as one’s that are parasitic on those uses. I guess my slight worry is that the more the expressivists acknowledge the presence of the ‘non-expressivist’ usages the harder it will be for them to argue that the expressivist story captures something essential about the core meaning of moral/normative language.
Thanks for the great paper Jack, and the insightful comments Teemu.
My response is in line with the part of Jack’s main response to Teemu listed as #2. The basic worry is that, the more content you pack into a sentence beyond the relevant moral word (and the alleged contrast, obviously), the harder it is to discern whether any resulting problems come from the moral word alone or the moral word + the rest. If you’re trying to thread the needle in saying that the competence conditions for words like “wrong” are as the expressivist says, intuitions get marred when you add in additional material.
Additionally, while I do find Teemu’s (12), (13) and (14) more problematic than Jack’s (8) (which I don’t find problematic), I still find them in some way different from the straightforward “P, but I don’t believe that P.” In keeping with Jack’s paper, I’d say for Teemu’s sentences I find the whiff of paradox stronger but not overpowering, but I find the paradox in standard Moore-paradoxical sentences to be sulphur-egg level. This leads me to suspect that the additional content in (12), (13), and (14) alters, rather than reinforces, how I hear the word “wrong.”
Hi Jack,
So, I am wondering about the Parity Thesis, and whether the expressivist really needs it. I think that there are actually two versions of it in your paper. The first (PT1) is the ‘natural’ thought that “sincere moral assertion expresses states like approval or disapproval in exactly the same way as sincere non-moral assertion expresses belief.” (3)
The second, stronger version (PT2) involves the following requirement: There is a kind of linguistic intuition we have when we consider certain Moore-paradoxical claims in the descriptive domain. And the Expressivist (allegedly) needs the following condition to obtain: when we consider what appear to be Moore-paradoxical claims in the normative domain, we have to have a linguistic intuition of precisely the same strength.
Gibbard only seems to be committed to PT1: he notes that the same basic account of how assertions express beliefs will be available for the expressivist in explaining how moral utterances express plans. Does this entail PT2? It seems not. We might still be confronted with moral Moorean paradoxes even if the confrontation is not phenomenologically identical to our encounters with nonmoral ones. Perhaps there is a ‘strength’ threshold that has to be met if the intuition is to indicate a Moore-paradox, but your Objection 5 shows precisely how that threshold can be crossed in the moral case (the strength of the intuition definitely increases when one actually hears the sentence used).
Moreover, the citation you give for Blackburn (1984 p. 169) does not point to any discussion of the Parity Thesis, and I am not aware of Blackburn having endorsed it anywhere else. In fact, he is rather fond of emphasizing various discontinuities between normative and descriptive discourse. So perhaps there are coherent expressivist views which do not require either version of the Parity Thesis.
Anyway, I am certain that you have more to say about this, and that you will probably point me to something in the paper that I’ve missed. Thanks for a thought-provoking piece!
Hi Nick,
I take it that while PT1 doesn’t entail PT2, PT1 + the explanation of Moore’s paradox that I suggest in the paper does entail PT2. It is possible to deny the explanation of Moore’s paradox I suggest while maintaining the parity thesis. This would require showing why wishes, boo/hooray talk, non-moral assertions, promissory assertions, and expressions in the optative all give rise to clearly Moore-paradoxical phenomena, whereas moral assertions do not. Especially given that by PT1, they expresses states like approval or disapproval in exactly the same way . At this point I’d press on what ”in exactly the same way” is supposed to mean.
Now, you may say that things like my (8) are Moore paradoxical; they just do not have the same feel as the ordinary case. I disagree that they’re Moore-paradoxical, but I also think that they should feel the same. Or, they should feel the same given PT1 and that so many other cases yield familiar feeling Moore-paradoxical feelings of similar strength to the ordinary Moore-paradoxical construction. Absent an explanation of this, it feels like special pleading on behalf of the expressivist view. I discuss this a bit in objection 1.
re: Blackburn. It’s definitely suggested, not asserted, but that’s not surprising given the source. It’s further suggested by the discussion on pg. 167 of Ayer: ”what it is to accept a moral remark; it is to concur in an attitude to its subject, rather than in a belief.”
Thanks for the interesting comment!
Alex,
I guess I’m inclined to agree with you (and Jussi). My first reaction to something like (8) would be that it’s incoherent. But I can easily come up with interpretations on which what has been said is not incoherent, after all.
Jack,
Thanks for the thoughtful replies (and for the kind words)! Here are some responses.
(1) Okay, so the target isn’t normative expressivism, but moral expressivism. A moral expressivist, I take it, says, roughly, that the meaning of moral sentences is to be explained with reference to their functioning to express certain kinds of desire-like states. Now, we could say that a sentence such as “Murder is wrong” is a moral sentence. But in this case the expressivist is likely to say, somewhat more carefully, that this sentence can mean many things, depending on the context, and that her expressivist story is not meant to apply to all of its possible meanings. Indeed, the expressivist might very well suggest that what she’s interested in explaining is the meaning of this sentence when it’s being used in a normative sense (I say a bit more on what this amounts to in response to Jussi, below).
You suggest that sentences such as “All things considered, we ought not to murder, but I’ve got nothing against murder” are not incoherent in the same way that the the standard cases of Moore’s paradox are. I guess my intuitions simply differ here. I tried forcing a normative interpretation for “Murder is wrong”, yeah, and when this is done, the relevant Moore-sentence seems incoherent to me. The case that I offered seems to me different from the tallness-case, as it’s not plausible that there’s a sense of ‘tall’ which implies shoulds, but it is plausible enough, I think, that there is a sense of ‘wrong’ in which it does imply some normative, oughty stuff.
(2) You’re absolutely right in that the Ridge-style ‘contextualist’ framework for ‘wrong’ is quite compatible with rejecting expressivism. However, my idea was, exactly, that with some uses of the term (the normative ones), the connection with a desire-like state is part of the competence conditions. So, the story really must be very different from that which is plausible in your nice example of ‘poisonous’. In these normative cases we should find the relevant Moore-paradoxical sentences incoherent, but, as noted above, this doesn’t seem problematic to me.
(3) Yeah, that was basically the idea: very roughly, there are cases in which the relevant standards are not being endorsed (the descriptive readings) and cases in which they are (the normative readings). I’m not sure that I completely understand the “I feel like crap” analogy, but I’m very pleased to learn that we Finns should do okay in Boston! Anyway, so, there’s a convention (rejected by those from Boston/Finland), which dictates that we should respond to “How are you?” with “I’m good”, and likewise, if expressivism is true, there’s a convention which dictates that we should be against what we think is wrong. Is that the idea? And even though, when I respond “I feel like crap”, it is easy enough to infer that I’m from Finland (my accent will rule out the Boston option), my utterance still sounds inappropriate. So, (8), too, should sound wrong, even if it is easy enough to infer that the speaker is using ‘wrong’ in an inverted-commas sense. (I wonder if I’m getting this at all right.) Actually, the cases seem quite similar to me, as I’m inclined to think that (8) tends to sound initially wrong. But perhaps more importantly, if there are descriptive readings available for ‘wrong’, (8) doesn’t have to sound inappropriate in all contexts. It should only sound wrong in a context where the normative reading would be expected. And in those contexts, it sounds (initially) wrong, or so it seems to me.
Jussi,
Thanks for that. Now, the cognitivist can agree that there are senses of “Murder is wrong” where the sentence expresses a desire-like state in some interesting sense. But I do not think that they can accept that in some of its uses the meaning of this sentence can be explained just by its expressing the relevant kind of desire-like (or relational) state. To accept this is to accept that expressivism is correct in some cases. There is a certain class of uses of moral sentences (with vague boundaries) which has been puzzling metaethically inclined thinkers for familiar sorts of reasons (that’s the normative uses! – this minimal way of characterizing the normative is in part Gibbard-inspired). If expressivism is taken to be true of this class of uses of moral sentences, this seems to me to amount to saying that expressivists pretty much got it right all along. (And if expressivism would be taken to be true of some other class of uses, this would seem like a rather curious view to me.) So, while cognitivists shouldn’t accept that expressivism is the right story about some of the cases, expressivists can happily agree that some cases (e.g., inverted-commas uses) should be given cognitivist treatment.
Whitney,
Thanks, that’s helpful! I don’t think I’ve got anything to add to what I’ve said above, though.
Great. That’s helpful in understanding where you’re coming from. Obviously I disagree about the felt incoherence of sentences like (8), but I also want to note three things.
First, I’m with Jussi’s second comment: this strategy seems to me to throw the baby out with the bathwater. Surely the game is to show that core moralizing is expressive. Second, and connectedly, I still say that the pragmatic explanation does a better job at predicting the variety of cases—cognitivism plus pragmatic effects is a very powerful explanation of why certain cases yield desire-like states and some don’t. Expressivism plus inverted commas is not a promising explanation, in my view, for reasons like that spelled out in my point (3) (which you’re right about the interpretation of, by the way.)
Third, the point with ‘tall’ was supposed to be that plugging ‘in the sense which implies…’ into the sentential context forced the reading you intended, just like plugging something similar into the ‘tall’ sentence forced a normative interpretation. We can modify the examples to get cases where interpreters need to infer that the speaker has various desire-like states or beliefs about reasons, or the like, but the expressivist view is supposed to be a view about ordinary moralizing. I think this could have been clearer in my earlier post, and I apologize for that.
Thanks to all. This has been very useful already!
Hi, sorry if I’m coming in on this too late. Time zone difference. I just want to ask if (8) really sounds *Moore-paradoxical* to people, as opposed to just weird, which is how I hear it. Remember, Moore-paradoxical sentences sound like the speaker is literally contradicting himself. That’s the paradox: it’s not a literal contradiction, but it sounds like one. This makes the sentence uninterpretable as a piece of communication except outside of very special contexts (maybe on the therapists couch). (8) doesn’t sound like a contradiction to me. I have a pretty good sense of what the speaker is saying. He thinks murder is immoral, but doesn’t care about morality. Or something like that. It sounds weird. But the guy who would say it is a pretty weird guy. (An extremely forthright and self-aware sociopath, maybe.)
Anyway, if others really hear it as contradictory, I’ll accept my intuitions have been corrupted by philosophy. But I’ve noticed in some of these types of discussions, anything that sounds a bit strange gets treated as slightly Moore-paradoxical. But that’s not what Moore-paradoxicality is.
Thanks for the discussion everyone. I have three comments, which I will try but largely fail to keep short.
1) I think it is a mistake to see (the best versions of) modern expressivism as committed to the incompatibility of the following claims: normative judgments express conative states and they also express beliefs. If they express beliefs, they are *normative* beliefs. If they express normative beliefs and also conative states, then normative beliefs are different from nonnormative beliefs in this (perhaps fundamental) way.
2) Here is Jack’s response to Teemu on “inverted comma” readings (which I don’t think is a good way to describe them–“nonnormative” or some such is much better):
“inverted commas readings … need to be signaled, otherwise they produce incoherence when the ordinary role of an assertion is contravened. So I do not think that the availability of these readings is sufficient to explain the coherence of my examples.”
But it isn’t true that nonstandard, in this case, nonnormative readings have to be explicitly signaled, right? The signal to interpret nonstandardly is contained in the fact that not doing so delivers an interpretation of incoherence. We generally try to understand our (intelligent) interlocutors as not being incoherent. We can’t always do this, and so sometimes we can’t see how what they say is coherent. But when a coherent interpretation is available, we normally do and should at least tentatively accept it.
We cannot see what someone could be doing in (sincerely!) asserting Moore-paradoxical claims, but we can see (multiple different things) that someone could be doing in saying that stealing is wrong, but that they are not against it. Given the option of interpreting the person as incoherent vs. interpreting them as employing the term “wrong” in a nonnormative way, the choice is easy (and of course not a matter of conscious deliberation). For example, in response to the question, “Is murder wrong?”, if a relevantly educated person answered, “It is and it isn’t,” I wouldn’t interpret them as saying something incoherent, but rather preliminary. So I do think the ready availability of such interpretations is sufficient to explain the coherence of such examples.
3) Finally, we needn’t interpret such claims as nonnormative in order to explain their coherence within an expressivist framework. Expressivism (stripped of its unfortunate “noncognitivist,” “nonrepresentationalist” and “antirealist” baggage) holds that normative judgments express conative attitudes. There are many such possible attitudes, and many of them are conflicting. Something can be morally wrong but prudentially right; wrong from the conative perspective associated with friend-loyalty, but right from the perspective of citizen-duty, and on and on.
The overwhelming majority of the time, the perspectives from which normative judgments are given are not made explicit (in my view, speakers are normally not well-equipped to even recognize them, much less make them explicit). This leaves hearers to interpret (though again, not explicitly). So even on a normative interpretation of “murder is wrong,” an expressivist can happily allow that a conative state is being expressed, despite the person following it with the claim that they are not against it. The interpretation goes something like this (not in these words of course). There is a conative perspective from which the speaker judges murder wrong (it might be a very powerful one–the speaker might be in genuine tears when speaking). Nevertheless, the speaker doesn’t stand behind, or identify herself with, this perspective.
For the expressivist, there must be some other conative perspective with which the speaker is (attempting to make herself) more deeply practically identified. So again, we could even change the example to “Murder is wrong but it isn’t” and, *even if we interpret the first part normatively*, we still wouldn’t be forced into a Moore-paradoxical interpretation because there are different normative perspectives form which wrongness judgments can be made. In the case at hand, the expressivist could (and the details of context really matter) interpret the person as making a normative judgment from one conative perspective, but taking a (normative) practical stand relative to some competing perspective. Of course this sort of linguistic behavior is nonstandard, which is precisely why nonstandard interpretations are and ought to be sought by interpreters. And unlike the case of “It’s raining but I don’t believe it is,” there are, it seems to me, plenty of non-incoherent and expressivist-friendly interpretations available.
Eric Campbell
Eric,
Thanks! Two minor things. First, not that “Is grass green? It is and it isn’t” is likewise coherent. So there’s no substantive difference here between the ordinary non-normative case and the normative case. The coherent interpretation strategy isn’t sufficient, I think, to save the expressivist here; at least not by itself.
Second, I deal with your third (and I think very sensible!) worry somewhat in the paper. Here’s what I say:
“Most of the cases given above used very generic language
for the attitudes involved. If we run the ordinary version of
Moore’s paradox with an expression for a more generic attitude
(say, ‘think’) we continue to get incoherence. Since disapproval and
being against encompass a number of possible negative affective
and conative attitudes just like thinking encompasses a number of
cognitive attitudes, we should expect incoherence with ‘disapprove’
and ‘being against’ just like we find incoherence with ‘think’. We
do not find such incoherence. This objection also conflicts with a
desideratum of a successful expressivist account of moral discourse.
A successful expressivist account should give a compositional account
of the attitudes expressed by moral assertions. The more diverse the
attitudes expressed by moral assertions, the weaker the prospects for
a straightforward compositional account of the attitudes expressed.”
The idea was that just as there are many conative attitudes, there are many cognitive attitudes. But we still get Moore-paradoxicality when we use the most generic language behind them. If there isn’t something more generic to be said about the conative attitude expressed, then the prospects of giving a compositional account of expressivist meaning—and hence of solving the Frege-Geach problem, looks just about impossible. Even the best strategies for accounting for inconsistency in an expressivist framework (which Derek Baker and I argue in a recent paper forthcoming in Ethics involve either Davidsonian interpretational incoherence or Velleman-style functional role accounts) will falter if the platter of conative options for an ordinary speaker is so broad.
Now, the identification with a normative standpoint strategy is interesting. But I think that even here there are problems with the supposed disanalogy with the case of ordinary assertion. There are epistemic standpoints as well and yet we do not get to eliminate Moore-paradoxicality by thinking that the speaker doesn’t identify herself with this standpoint. When I say, tears in my eyes, that there are propositions, but I don’t believe it, we get incoherence even though there is a similar interpretation available about me asserting something from one epistemic perspective, but taking a stand on another (Quine is a hero of mine, after all.)
But I should, for full disclosure, note that something along these lines was my first and biggest worry with my examples, so it’s useful to have someone push them so nicely. Thanks again!
The first “not” should be “note”, obviously.
Sorry to come so late to this party, but I’ve found this all very interesting! I can’t resist weighing in on several points:
(1) I like the fact that in the paper you take the trouble to distinguish revolutionary expressivism from hermeneutic expressivism, and in passing mention that there are serious problems in arguing for reforming definitions for normative discourse in general – I think this is a neglected but very interesting and deep problem. Sebastian Koehler and I wrestle with some of the methodological problems that arise for any form of global revolutionary view about normative discourse in general in our “Revolutionary Expressivism” (Ratio).
(2) Applying the objection to Ecumenical Expressivism – I can’t resist seeing how the objection plays out against the backdrop of my own preferred version of expressivism. On the version of the view I defend in Impassioned Belief, only certain uses of ‘ought’ (e.g.) are normative, but you (Jack) are right that for my view to be worthy of the name ‘expressivism’ there must be some sense in which at least these uses are associated with pro-attitudes. For me, they are: they express hybrid states, one component of which is a “normative perspective,” where normative perspectives are global Bratmanesque policies against certain kinds of standards of practical reason and a commitment to reason and act only in ways not ruled out by all of those standards you have not (yet) ruled out. This suggests that the analogue of your (8) for me would be something like the following, where the context somehow makes clear that ‘morally wrong’ is not simply an anthropological comment on what some or other culture or person’s moral standards say, but is an engaged first-order normative remark:
(8*) Murder is morally wrong, but I am totally open to endorsing moral standards which permit murder.
This, to my ear, has quite a bit more than a “whiff of paradox,” but I admit I may be in the grip of my theory. I’d be curious to see what others think of this and other test cases as applied to my approach, which has this standard-oriented structure.
(3) Core versus non-core uses of ‘ought’, ‘good’, etc. The issue of whether the expressivist theory must provide the “core” meanings for these terms. If by ‘core’ we mean ‘most common’ or even ‘most central to the meaning of’ then I don’t actually think that expressivists need to agree that their theory, which I understand as a meta-semantic theory, provides the core meanings of these terms. It may be that the majority of uses of ‘good’ trigger conventional standards of some time, as with ‘good toaster’, and these may be no more or less central than ‘pleasure is good as an end’. What is crucial is that there are certain distinctive contextually specified contents for these terms such that these contents are normative in some interesting sense, and that the normative plays a special role in our lives, and that expressivism provides the right meta-semantics for these uses. I’m not sure this commits me to some of the claims made here about expressivism capturing the “core” of the meanings of these words in English, as the comment thread here at some points suggests. This is also relevant to Teemu’s original line of reply, which I think does have promise. After all, if normative uses of ‘good’ (etc.) aren’t core in these senses, then it may not be so hard to get non-normative readings into play. Moreover, I’d add, that considerations of charity may strongly encourage interlocutors to hear uses of ‘good’ (etc.) as non-normative when they are followed by relevant clauses like ‘but I’m not against murder in any way’ or whatever. In fact, that may even happen for ‘all things considered’ which usually works well to trigger a normative reading, since we can hear this in terms of considering all of the things deemed relevant by some conventionally recognized set of standards we don’t actually endorse. Ditto for ‘weighty reasons’ (as there can be weighty reasons of etiquette I don’t give a damn about – they are still weighty qua the norms of etiquette). So the point about context-sensitivity does still have a lot going for it, given the impetus from charity to hear the first half of these sentences in a non-normative way given what is said in the second half.
(4) The parity thesis, and the argument as Teemu reconstructs it and as Jack defines it. Teemu’s first premise is: “(i) If expressivism is true, then the parity thesis is true, that is, sincere moral assertions express desire-like states (e.g., being in favour of or against something) in exactly the same way as sincere non-moral assertions express beliefs.” (emphasis added) The use of a definite description here assumes that there is only one way in which sincere non-moral assertions express beliefs, but that seems highly dubious. I think that by itself this doesn’t cut all that deep, but it at least means the argument should be reformulated – presumably with a universal quantifier? That is, something like: “(i) If expressivism is true, then the parity thesis is true, that is, sincere moral assertions express desire-like states (e.g., being in favor of or against something) in all of the same ways that sincere non-moral assertions express beliefs.” I think this may matter to the soundness of the argument, since there may be certain ways in which an ordinary assertoric use of ‘grass is green’ expresses the belief that grass is green which entails a kind of transparency, such that the expressivist need not endorse that the parity thesis holds up for that sense of ‘express’. Indeed, I’d be inclined to say just that. This brings me on to my last point.
(5) Transparency: I think Camil’s remarks about transparency get to the heart of the matter. Expressivism is a theory of (a) the nature of normative judgments, (b) the meanings of normative claims, such that the latter is explained in terms of the expression of the relevant judgments. This theory, as I understand it, is compatible with the real underlying nature of these judgments as outlined in (a) not being transparent to ordinary speakers. What must be transparent to those speakers, perhaps, is what commitments one undertakes in making such judgments – including certain practical commitments. So on my view, they must perhaps realize that judging that abortion is morally wrong somehow rationally commits them to rejecting moral standards which allow abortion. Must they realize that this judgment so commits them in virtue of being (very roughly) a hybrid state with a belief-like component and a desire-like component? No, I’d have thought. They might have some false theory of the nature of their judgment – they might think it is a besire, or endorse a view like Michael Smith’s, each of which would vindicate the assumption that there is some such commitment issuing from the judgment. Or, like the vast majority of ordinary thinkers, they may have no theory of the nature of these judgments. That is fine too, so long as they recognize what they are committed to in making the judgments – they need not know how the judgment grounds this commitment. The upshot is that the expression of a hybrid state is not transparent. This does require a lot of the work in the theory of semantic competence to be done by know-how rather than know-that, but I think that is independently a very plausible (and fairly orthodox) view, anyway.
Note that this story does not give up on the thesis that “by engaging in it (moral discourse) you commit yourself…to possessing big-D desire-like states.” It may give up on the thesis that by so engaging you commit yourself to possessing such states in precisely the sense Woods has in mind – if that sense entails transparency of a certain kind. But that seems OK to me.
By contrast, there is a sense in which an assertoric use of a declarative ‘p’ does transparently express the belief that p. It is this transparency which grounds the obvious paradox associated with Moore sentences. This might be the ‘same content’ sense of expression that Schroeder discusses in Being For, e.g. But that isn’t the notion of expression I deploy in defending my version of expressivism, nor is it Blackburn’s or Gibbard’s preferred version. Note how this point relates to my concern in point 4 above about the parity thesis and how we need to be more careful in formulating it.
Jack at one point in the original paper anticipates this move, I think, with the following comment:
“Even without the parity thesis, it is difficult to imagine how this coordination of expressions and avowals could proceed if the expressing or avowing character of moral assertion were not at least tacitly recognized by ordinary moralizers. Since expressivists like Blackburn hold that expression of practical states is what we do with moral assertion and that coordination of practical states is the point of moral discourse, they must hold that ordinary moralizers are in some position to recognize that moral assertions express practical states.”
I think it is enough that speakers recognize what their judgments commit them to, though, and that their semantic know-how means that they navigate these commitments by in fact accepting the right kinds of hybrid states. I don’t think they also need to realize that this is how they navigate these commitments.
Even more obviously I don’t think competent speakers need to recognize the function of their discourse – functions here can be specified in Millikanesque terms – in terms of what sociologically explains why the discourse persists. I don’t know why ordinary speakers would need to know that. Perhaps certain kinds of religious discourse actually functions in this sense to reinforce patriarchy. It doesn’t follow that users of the discourse will recognize this – in fact, the function may be better fulfilled precisely because it is covert.
Just before the passage I quoted above, Woods himself quotes Blackburn to underscore his argument. Blackburn there does talk about ‘making public’ our practical stances, but there is a crucial de dicto/de re distinction we can usefully invoke here. What we make public are states which are in fact practical states qua their direction of fit. But we needn’t make them public “under that description.” We might rather make them public only qua ‘the belief that abortion is wrong’ and the like.
Another way to see the point about transparency would be by seeing how the argument on offer easily threatens to prove too much. Ordinary speakers must recognize that the judgment that grass is green is, in some sense, about grass. Do they have to recognize how the judgment manages to do this? E.g. is it an objection to functionalism in the philosophy of mind that ordinary speakers wouldn’t find ‘grass is green, but I’m not in such-and-such functional state’ paradoxical, where ‘such-and-such functional state’ spells out what the best functionalist theory tells us about the nature of that belief? I’d have thought not. Similarly, competent users of ‘abortion is wrong, but I don’t disapprove of it’ needn’t find that paradoxical either even if the best philosophical theory of the belief that abortion is wrong adverts to disapproving of abortion, say.
Mike,
Thanks! I’m going to have to respond in detail tomorrow, but thanks for the great comment. Two very quick things. First, I wouldn’t be tempted by that quantified version of the parity thesis, since there will surely be small differences between the cases. What I actually think is that if (hermeneutic) expressivism is true, then the parity thesis is the best extant explanation of what ”expression” is supposed to mean. And, I think, the key part of the parity thesis is that the commitment to a desire-like state works in the same way as the commitment to a belief works. This is buried a bit in my early explanation of Moore’s paradox. I was trying to be very general about the parity thesis, following tradition, but it might be better to get a bit more precise about it. I’ll have a think.
Second, though I’ll say something more about this tomorrow, I think it would be very weird if all the cases I mentioned in my response to Camil were transparent in this way, but normative language, this very central bit of our ordinary talk, was not. And, as I say in the paper, we suss out our grasp of grammar by means of our feeling that something ain’t right about a certain sentence; if we have a bad explicit grasp of grammar, that doesn’t actually affect our judgments of grammaticality much—at least for first languages. Likewise, I think, with the cases of Moore’s paradox and like types of incoherence. The jarring feeling we feel in these cases is indicative of something deeper about the commitments we incur by engaging in certain communicative practices. Believing we’re committed to besires by our judgment something is wrong should have about as much effect on the presence or absence of such feelings as believing that sentence adverbs are ungrammatical has on whether we find ”Luckily, I went to the store” grammatical.
More tomorrow!
Thanks Jack. On your second point, it would only be weird if there were an asymmetry between the cases, right? I think they are symmetrical. In both cases, statements of the form, ‘p, but I don’t believe it’ come out as weird, and rightly so. Similarly, I’d have thought, ‘p but I’m not in such-and-such functional state S’ where S is type identical to the belief that p do not come out as weird at all – whether ‘p’=’grass is green’ or ‘abortion is wrong’. It is just that the functional type in the one case constitutes a robustly representational state whereas the functional type in the other case constitutes (on my view) a hybrid state which is such that any token of it is only partly constituted by a representational belief. But I may have misunderstood your comment and look forward to your elaborations! Thanks very much for engaging with my comments, and for writing such an intriguing paper, for that matter.
Yes, only then, but I think they are asymmetrical. I don’t have much to say about to your (1-4) other than what I’ve said above, so I’m going to focus on the transparency issue—some of my remarks bear on the other points as well.
I suppose I think that the functional-state description of the belief is significantly different than the coarse description of the desire-like state I was after. The claim that’s needed here is:
It is sufficient for a hermeneutic expressivist view that the underlying functional type can be a hybrid state; that we somehow commit ourselves to having that hybrid state; but that the normative aspect of it isn’t transparent to speakers in such a way as to be captured by an expression like `being against’ or `want’ or what have you.
I think this claim is false – we have to have the ability to recognize what we are committed to by fundamental bits of our linguistic practice like normative talk, at least at a fairly broad level of description. Otherwise, it is almost impossible to see how we could be engaged in the practice of externalizing our (endorsed) moral standards.
For a useful comparison, consider the case of conventional locutionary acts like promises. We do not need to have a detailed view of intention, how the locutionary act of promising works to generate commitments, etc in order to be competent promisers. But one thing we need to grasp is that if we sincerely promise, we incur a commitment to intending/trying to keep it. Along with other things, such as a belief that I’ve promised. By ‘grasp’, I do not mean that we have a belief about commitments—much like the case of grammatical relations or conversational norms governing implicature, it is far more tacit. It is sufficient that we expect that promisers intend to keep their promise and we do not know what the speaker is up to when they deny such immediately after promising, absent special conditions which make this sensible.
Luckily, we do grasp that promisers commit themselves to having certain intentions – which is born out by the fact that locutions like “I promise to meet you for lunch, but I have no intention of keeping it” are incoherent in very much the same way as “I’m having lunch, but I don’t believe I’m having lunch.” And we get this incoherence whether we’re Hart or Joe or Josie Sixpack.
I think we agree that this commitment-oriented account of expression is the central notion of expression for the expressivist – or, anyways, the most promising one. And, on it, all the cases I describe in my response to Camil express various mental states and when we conjoin them with a denial of being in those states, we get Moore-paradoxical constructions. But not with the moral case. So I think the cases are asymmetrical.
The problem here is one of balancing the transparent cognitive content of moral assertions with the deep practical content. If the practical content is deep enough, I’m going to say that the account of commitment starts to break down…we need to be able to recognize what people commit themselves to by making moral assertions in order for them to function correctly—and it would be a miracle if the commitments we incur don’t have a common coarse lexical expression like ‘being against’. Otherwise I’m not exactly sure what you could mean by ‘commitment’ – though see below. I think it also gets more and more tempting to go with (conversational) ecumenical cognitivism, in your vocabulary, at this point. As I remarked to Teemu, this seems a much more powerful explanation than the corresponding expressivist requirement. On the other hand, if the practical content is shallow enough, then like promises, optative expressions, and so on, we should find Moore-paradoxical constructions. Which I hold we do not.
On the last suggestion of your comment: I’m intrigued, but I still think that it’s insufficient to accommodate the central role of practical discourse in our ordinary lives to require only commitments to behave in certain ways dictated by possession of normative standards. Why then would promising have such an obviously expressed mental state (described as an intention), but moral assertion not. After all, they both involve commitments to behave in certain ways. And we have a word to describe the states you mention – policies, or, even better, plans. I suppose I’m not sure what, on this view, would remain of the idea that we ordinary moralizers can recognize the commitments taken on by someone’s moral assertion. But it’s an intriguing suggestion; I will think about it more.
The main goal of my paper was to argue that (hermeneutic) expressivism is not true, but there were also secondary goals. First, to get expressivists to be clearer about whether they were advocating hermeneutic or revolutionary views and in which sense. Second, to give a better account of expression that didn’t run headlong into my problem – i.e. to stop simply poaching the relation between sincere assertion and belief to develop the expressivist view. I think the behavioral account of expression you’re pushing here is an attempt to do this second thing; if successful, it would skirt my problem. I have my doubts as to whether it succeeds, but it’s all to the good to formulate more and better options here!
Thanks very much for the interesting reply and the detailed comments.
Thanks Jack. Promising is an interesting test case, but in some ways the analogy is imperfect. To understand ‘promise’ you have to understand a certain speech-act, namely promising, and plausibly you can only understand that speech-act by understanding its role in assuring others about how you are going to act. That role for promising seems *very* close to the surface for competent speakers. Moral/normative uses of ‘ought’, ‘good’, ‘reason’, ‘must’, etc. are not tied in this way to a specific speech-act (unless that speech-act is assertion, but then all normal predicates will be tied to that speech-act in some special way!).
In relation to this point, note that “I promised you I would meet you for lunch, but I’m not going to (something came up).” Seems fine, at least in terms of linguistic competence – depending on the details we may think an utterance of this betrays immorality but that is a different matter. It is only when someone is making a promise and in the same breath says they won’t fulfil it that we get incoherence. Again, that connection to this special sort of speech-act suggests a useful disanalogy.
I agree with you that we need some level of understanding of the commitments we undertake – not just to act in certain ways, but to count some considerations as reasons. But then I think the following *do* sound incoherent:
“I am sure that morally I must tell the truth in Court tomorrow, but I am totally open to accepting moral standards which would not require my telling the truth in court tomorrow.”
“I am sure that the fact that it will be painful is a reason not to bang my head in the wall, but I am totally open to accepting a fully general set of standards of practical reason which do not count that fact as a reason.”
These strike me as clearly problematic in terms of coherence, but, again, perhaps I am just idiosyncratic. Perhaps we disagree more about which sentences sound incoherent than we do about some of the more subtle issues of how much must be transparent.
Also: On the superiority of Ecumenical Cognitivism on this front. I discuss what I consider to be the problems facing Ecumenical Cognitivism in the book; I won’t rehash all of that here. Short version: The best versions of the view cannot simultaneously accommodate the full range of fundamental disagreement by the conceptually competent without implausibly severing the needed links between truth/falsity talk and agreement/disagreement talk. I do admit, though, that simply in terms of accommodating the role of moral/practical thought and discourse in settling the thing to do and how to reason about the thing to do that certain forms of ecumenical cognitivism do very well. That isn’t my objection to those views! However, I don’t think the best versions of Ecumenical Cognitivism are implicature versions, as they can’t adequately explain the irrationality of judging one must do something while failing to intend to do it. Anyway, that is a whole other conversation, really.
Thanks again.
Great. One quick point: “I believed grass is green, but it wasn’t” is likewise fine – it’s characteristic of Moore-paradoxical constructions that third person cases, including diachronic cases, are not paradoxical. So this part of the disanology doesn’t hold up.
I suppose I would hold that moral assertion is a form of speech act…at least that’s what pure expressivists should say plus or minus a. It of finagling. Hybrid views like yours are trickier, which is way I’m softer on them in the paper. It all depends on how the connection between moral assertion and practical states is made.
On your examples: I’m inclined to agree that they sound ungreat, but it’s hard to hear them as distinctively practical since “accepting moral standards yadda yadda” sounds a bit like “believe it is wrong yadda yadda”. You can stipulate that they’re to be heard practically, but our triggers for commitment aren’t tied to our surface beliefs like that. We hear what we hear. And if they’re heard cognitively, then they ought to sound paradoxical. I address this a bit in fn. 16 of the paper.
I can’t resist an example of Wittgenstein’s on this point: “Mr. Scot is not a Scot.” Try to hear the first as a common noun and the second as a proper name.
On ecumenical cognitivism, let’s take the chat offline as you’re right it’s a different conversation.
Thanks again!
*a. It => a bit
*way => why
Hi again, Jack!
I take your point about how one might here ‘accepting a standard’ as something cognitive, though I guess that just makes it hard to formulate a fair version of your argument against a view like mine (or Gibbard’s in *Apt Choices, Wise Feelings*) which is formulated in terms of accepting or rejecting norms/standards. I also think the following also sounds terrible, if that helps at all (it might seem harder to give the second clause a non-practical reading):
“All things considered, I definitely must tell the truth tomorrow, but I as a practical matter I am totally open to not telling the truth tomorrow.”
Oh, and I am not sure even non-ecumenical expressivists of a quasi-realist stripe should allow that moral assertion is a special sort of speech-act, as opposed to just being an instance of the speech-act ‘assertion’ with a moral content. I discuss this in more detail in my reply to Terence Cuneo’s nice challenge to expressivists in his Madison paper in my “Moral Assertion for Expressivists.” (*Phil. Issues*) I won’t rehash all of that here either, though!
I completely agree it’s harder to lodge a version of my argument against views like yours or Gibbard’s ACWF view for exactly this reason; in the case of Gibbard, the norm-acceptance story is part of what motivates charges that his view can look more or less like a cognitivist view. When it’s read directly in terms of plans, it’s more clearly an expressivist view. And, in my view, more clearly vulnerable to the sort of argument I give. So this issue seems to me to bleed into the issue of what counts as an expressivist view and what’s necessary for a speaker to grasp in order to count as expressing practical states. I’m not that all that interested in rectification of names, but I think the second issue is of paramount importance. At a minimum, the available coherent positions for an expressivist seem to me to be significantly narrowed in light of my argument.
Jack,
Thanks for your response, and sorry to have tuned out of the discussion for so long. I just wanted to second a point made by Michael above, adding a minimalist touch to it: insofar as the expressions you cite in your first response — ”think”, ”hold that”, ”view the world as being such that … is true of it”, ”would describe the world as …-ish” — give rise to Moore-paradoxical sentences, the minimalist expressivist can claim that they do so only if understood in a minimal sense. Such minimal cognitivist notions, she will say, equally apply to (and equally give rise to Moore-paradoxical sentences in) ordinary descriptive discourse and normative discourse. But the deep functional features of *any* kind of discourse are not similarly transparent, so we shouldn’t be surprised by the lack of Moore-paradoxical sentences involving them.
I also agree with many of Michael’s other points regarding expression and commitment. But let me end by saying that I fully agree with you that the available coherent positions for an expressivist are significantly narrowed in light of your argument.
I should say as well that I also agree there are certain otherwise tempting ways of understanding expressivism that your argument puts nice pressure on, Jack – they just aren’t the forms of expressivism dearest to my heart! Gibbard’s latter plan-laden view is, I agree, in some ways a more apt target, for example. One caveat about that, though: That view is about the normative more generally, and not specifically about the moral, so some of the moves you made much earlier in the thread about your focus being specifically on the moral may in another respect make the plan-laden view less apt as a target. Even so, I’m sure you are right on balance that it is a better target for you than the WCAF view.
I agree about Gibbard; I think a version can be lodged against it, but it has to be done carefully—it’s not nearly as straightforward. And,as you note, I’m less sure about the straight normative case. I hope to explore it in future work.
I should be clear here about one of my disagreements with Camil and perhaps Mike. As sketched in my response in the paper (Objection 6),minimalism isn’t sufficient to defuse my argument; by itself, it just explains in a deflationary way the incoherence of ”Murder is wrong, but I don’t believe it’s wrong”. What is needed is the additional claim that any way of indicating the relevant practical state committed to is either like robustly representational belief in being too esoteric or opaque to trigger Moore-paradoxical reactions OR that it does trigger Moore-paradoxical reactions. My suggestion is that the claim that all such ways are too esoteric sits badly with the idea that our moral talk indicates to ordinary speakers commitments to practical states; thus I think that there are non-esoteric ways to indicate such states, at least at a coarse level of description, for most expressivist programs—I cite in the paper a number of cases in the literature where this is clear. And these ways seem to me to not yield Moore-paradoxical reactions. I thank Camil and Mike for putting pressure on this line of thought. I still think it holds up, ultimately. But it’s worth pushing on it as hard as possible given how pervasive minimalism is for expressivist programs.
I suppose my thought is that the relevant states of mind don’t need to be esoteric for the nature of the state of mind expressed to be less than fully transparent. Suppose that what is transparent is that the belief that one must not kill commits one, on pain of irrationality, to not killing. It turns out that the best theory of rationality (a broadly instrumentalist/coherence based one) combines with an expressivist theory of the relevant must judgment to explain this commitment, but this explanation need not be obvious to ordinary speakers. They might be equally open, under Socratic pressure, to a reasons-responsivness story plus some form of realism, or some other story again. So long as they implicitly recognize that the state is one that rationally commits a thinker in this way, and they also are competent in their semantic know-how with ‘rational’, ‘irrational’, and cognates, this will be enough to explain their navigating these commitments using the right states of mind without understanding their true underlying nature (e.g. as hybrid states). I think I hadn’t made this clear before – it isn’t that the building block states of mind which go into the hybrid state that is a normative judgment on my view are necessarily all that esoteric, taken one by one, anyway. It is rather that the relevant competence/know-how only requires a recognition of what someone is rationally committed to doing (and thinking, and reasoning in accord with, etc.) – this competence/know-how doesn’t require knowing *in virtue of what* these commitments are undergirded.
Compare: To know how to play chess, I need to know the point of the game and how the pieces move. I don’t need to know in virtue of what these count as the rules of chess (what *makes* this move legitimate or required by the rules), or indeed in virtue of what these count as a game – at least, not in the sense of being able to articulate this discursively if I had to do so.
Excellent; this is much clearer to me. I still reckon that if the proper theory of the must judgment is expressive, then there should be some recognizable practical component of the ‘must’ judgment available to a competent speaker under some description. But, this pushes us back to your examples on which we differ, somewhat, on our coherence judgements about. As I said above about Gibbard, this is why hybrid views are more difficult to apply my argument to. The judgments are less clear, perhaps, which could possibly be explained by us reading them as cognitive. It also pushes us onto the disputes we pushed to the side about ecumenical cognitivism vs non-cognitivism. And it raises the question above about what sense in which belief-expression and must-expression are functioning in the same way—that is, how much of the parity thesis we are preserving here.
Two brief side-notes. First, I do not like the ‘on pain of irrationality’ locution since it can suggest that the irrational do not have such commitments. But I think this is a mistake; irrational folk engaged in ordinary assertion still commit themselves to believing what they assert and, probably, very immediate consequences of it. Likewise I’d think with must-assertions. Second, the chess analogy puzzles me a bit. I agree that we do not need to know the explanation of what makes a rule of chess a rule of chess to be able to play chess; but it seems to me plausible we need to know that must-judgments have an expressed practical component in order to have an expressive theory of ‘must’…though I agree not in order to have a view where our moral assertions commit us to act and behave in different ways.