Response to “Expressivism, Subjectivism and Reasons”

Note: This post is written by Katarzyna de Lazari-Radek and Peter Singer and is a response to Jussi Suikkanen's post Expressivism, Subjectivism and Reasons

Response to Suikkanen

In The Point of View of the Universe [OUP 2014; henceforth PVU] we seek to bring Sidgwick’s most important ethical ideas into dialogue with contemporary ethics.  This means that the book has a wide range, covering metaethics, the nature of justification in ethics, the dualism of practical reason, ultimate value, and some practical questions about utilitarianism.  We could not go into detail about each one of these issues, and so we focused on those that seem most important to us.

Suikkanen takes us to task for our discussion of expressivism, in particular for failing to acknowledge that it is a semantic or meta-semantic view about the meaning of normative sentences.  But in the section that precedes the one from which he quotes, we do note that expressivists claim that there is a minimalist sense of truth in which moral judgments can be true or false.  We then say: “In the following section we focus on what we see as the core of the dispute between subjectivism and objectivism: whether there can be objective reasons for action.” [PVU 44]  In the section that follows, we argue that Blackburn is a subjectivist “in the sense in which we have been using the term,” while also noting that he thinks it makes sense to say that moral judgments are true.

These statements still seem to us to be defensible.  To explain why, we need to consider Suikkanen’s claim that “…there is no connection whatsoever between existence internalism and expressivism (other than that they are both Hume-inspired).” 

 We think it to be quite an extraordinary interpretation of Hume to claim that the connection between the two lies only in the fact that Hume, quite independently, came up with both views.  Hume argues that reason is and ought to be the slave of the passions and then he argues that “Since morals …have an influence on the actions and affections, it follows, that they cannot be derived from reason.” [Treatise, Bk III, Pt I, Sec i.]  It would be hard to put more plainly the link between his view of reason as the slave of the passions (the inspiration for what Suikkanen, following Williams, calls “existence internalism,”) and his view of morality, which is presumably the inspiration for expressivism.

Nevertheless, Suikkanen tells us, expressivism is completely neutral about whether the fact that a given agent has a reason depends on that agent’s current desires.  He follows this statement with a paragraph giving expressivist readings of the statements “Charlie’s reasons depend on his current desires” and “Charlie’s reasons do not depend on his current desires.” The paragraph concludes with the statement: “Expressivism doesn’t decide which one of these first-order normative views is true, full stop. You can be an expressivist and endorse either.” 

Like Campbell Brown, in his comments on Suikkanen’s post, we have some doubts about whether the statements in question are normative.  But even if they are, what does this establish?  One could just as easily give expressivist readings of the statements “Expressivism can be supported by sound reasoning” and “Expressivism cannot be supported by sound reasoning.”  If Suikkanen seriously maintains that expressivism is neutral between these two claims, we can reasonably ask why we should bother with a theory the proponents of which do not affirm that it is supported by sound reasoning.  If, on the other hand, Suikkanen more sensibly denies that expressivism is neutral between these claims, then obviously the fact that one can give an expressivist reading of contradictory claims does not show that expressivism is neutral between the claims, let alone establish that there is “no connection whatsoever” between expressivism and Hume’s view of practical reason. 

So we cannot avoid the real question of substance: once it is granted that there are objective reasons for action, independently of the desires of the agent (in other words, once we reject the premise of Hume’s argument for his theory of morality), does expressivism remain an interesting and defensible position?  Note the conjunction: in our view, expressivists can’t have it both ways.  Perhaps expressivism can be made defensible, a la Schroeder, as a semantic or meta-semantic theory, but only at the cost of emptying it of substantive claims so that it becomes trivial.

In conclusion, we will allow ourselves a comment on the level of invective Suikkanen employs in his post. Philosophers are rightly proud of the fact that in our discipline we can tell those who hold different views that their premises are implausible, their arguments unsound, their conclusions absurd.  None of that troubles our friendship or our collegiality.  Nevertheless, we have to wonder if there  is something about expressivism that makes its proponents feel they must express their feelings more strongly than the rest of us.  Are our claims really so “outrageous”?  Could they not be published in a peer-reviewed journal?  Oddly enough, we have a counter-example: Suikkanen’s own 2009 paper, “The Subjectivist Consequences of Expressivism,” in which he defends exactly the position he now derides.  Here is a sample passage:

If expressivism has this awkward consequence, this should count seriously against the view. Expressivists themselves have claimed that their view and their other commitments do not have this implausible implication.  If my argument is correct, then that claim is false. At least normatively speaking, the commitments of their view have subjectivist implications. Of course, I cannot prevent Simon Blackburn from repeating over and over again, as he does, that it is not the case that phi-ing is wrong if and only if I (or anyone else) disapprove of phi-ing (or believe that I disapprove of it). If the argument given above follows from materials he is committed to, then all this implies is that he is committed to an inconsistent set of claims or attitudes. [Pacific Philosophical Quarterly, 90 (2009) p.364-387 at 377.]

We particularly enjoy the reference in this passage to Blackburn, for which Suikkanen is now apparently trying to atone.  We think that, in writing in his post that Blackburn “really must get tired of objections like this,” Suikkanen owed it to the reader to disclose that he himself was among those responsible for Blackburn’s weariness.  Suikkanen will no doubt say that five years ago he was young and naïve, and now he knows better, thanks in particular to Schroeder, who argues against this paper in his unpublished paper “Does Expressivism have Subjectivist Consequences.” But if the claims we make in PVU are so “outrageous” now, they were presumably outrageous when Suikkanen made them too, because all of Blackburn’s key writings were already out by then. 

~Katarzyna de Lazari-Radek and Peter Singer

70 Replies to “Response to “Expressivism, Subjectivism and Reasons”

  1. I’m not entirely a fan of expressivism anymore, but still it’s important to distinguish between the first-order normative claims Jussi is correctly claiming that expressivism is neutral between and the second-order metasemantic claims about what is going on when we make a moral judgment, when those judgments are assessed at a level of abstraction for correctness, and suchlike.
    As I said in the prior thread, it is a difficult question whether or not expressivism has subjectivist consequences given its metasemantic treatment—this turns, in part, on whether we can interpret claims like “Kicking dogs is wrong if and only if I don’t like dog-kickin'” in anything other than a first-order way. Blackburn says “no”. Dworkin, from the other camp, agrees “no”. Many other folks, like myself, tend to think yes. But this is, again, a non-trivial question not to be shrugged off. And talking as if there is no difference between the two interpretations simply muddies the waters.
    [Also, note that there are many ways to give an adequate treatment of expressivism. Mark’s work is a great example of just one of many ways to do this. And it is by no means clear to me that doing so in any of these ways ”empties expressivism of its substantive claims”. It had better not as expressivist treatments abound in other areas than morality, especially in areas where I think (and have argued) it is far more plausible, and the general semantic plan in these areas tends to be modeled on the moral case.]

  2. I don’t understand the central dilemma offered here:
    “Perhaps expressivism can be made defensible, a la Schroeder, as a semantic or meta-semantic theory, but only at the cost of emptying it of substantive claims so that it becomes trivial.”
    If ‘substantive claims’ means ‘first-order normative claims’ then it hardly follows from the fact that it doesn’t entail any of those that it is trivial – unless you think any purely meta-ethical theory which is neutral on first-order questions is automatically “trivial,” which is surely a somewhat unfortunate attitude to have towards meta-ethics as a discipline of interest in its own right. Compare: Moore’s non-naturalist realism is also meant to be neutral on the first-order normative questions. Does this mean that the semantic, metaphysical and epistemological questions raised by Moore’s view are all “trivial.” Hardly.
    If, on the other hand, ‘substantive claims’ is read more broadly to include purely meta-ethical theses which might be of interest in their own right, then there is no reason to think expressivism does not entail any substantive theses in that sense – even if it does not entail any substantive first-order normative theses.
    It therefore seems that expressivists can and should sit quite comfortably on the second horn of this dilemma.

  3. Also, on the statement “Expressivism can be supported by sound reasoning” and whether expressivists can be neutral about it. Note at least two things:
    (1) These are claims about epistemic norms, and so one could be an expressivist about practical normativity and a cognitivist about these claims. There are arguments (e.g. by Cuneo) that this is an unstable combination, but it isn’t *obviously* unstable. Whereas claims about one’s reasons depending on one’s desires are clearly claims about practical norms, and so an expressivist about practical normativity will have to be neutral about these. This combination of views would, whatever its other vices, clearly evade the force of the argument developed above.
    (2) Note that insofar as expressivism isn’t neutral about its own epistemic credentials, this isn’t because expressivism *conceptually entails* its own epistemic justifiability. “Expressivism could have been true even if nobody had good reason to believe it” isn’t a conceptual contradiction, after all. It is rather that in putting forward expressivism, one takes on a commitment to defending its epistemic credentials.
    That is a very unique sort of case, and is just a special instance of the general fact that whenever one puts forward any proposition in an assertion one takes on a commitment to defending its epistemic justifiability. An epistemic expressivist needs to explain why this connection holds, of course, but the point is that this is a very special sort of connection, and part of the pre-theoretic data to be accommodated – an ex ante constraint on theorizing. Whereas it is *not* an ex ante constraint on theorizing that our reasons depend on our desires – indeed, as a universal generalization it is more like a constraint that they don’t. Unless some special *argument* can be given that expressivism can’t be neutral about this, I don’t really see how the fact that they’d better not be entirely neutral about the justifiability of their view is germane. From the fact that expressivists must take on some normative commitments to argue for their view it hardly follows that they must take on any old arbitrary normative commitment whatsoever! So what is needed is an argument that the view entails existence internalism, and an argument that respects the resources available to quasi-realist expressivists. This is really just to echo what Jack Woods said above, though.
    Some people have tried to develop such arguments (Egan in his paper on fundamental error comes to mind for one of the better tries), but on no account is this any kind of *obvious* entailment of the view. In fact, I don’t think any of the attempts to show expressivists are committed to existence internalism are at all convincing, for the reasons Jussi canvassed. But pointing to the special case of being committed to the epistemic justifiability of whatever you assert hardly is an argument that expressivism entails subjectivism about reasons.
    It is almost as if you’ve gone from “Expressivists can’t be totally neutral about every single first-order normative claim (assuming the epistemic is normative), and still defend their view” to the conclusion, “therefore they are also committed to existence internalism.” That is a rather breathtaking leap, frankly; a lot more in the way of an argument is needed to support the conclusion that this very specific commitment is incurred by expressivists – particularly given how hard they have worked to demonstrate that it is not.

  4. I am almost a virgin blogger, and as Jussi rightly says quite weary of getting stung by the dust that people manage to kick up around the issues, some interesting, some not, connected with expressivism. But since my name is some kind of focus in this thread I hope I can say a couple of things.
    First of all I am pained on Hume’s behalf to see him carelessly paraded as a “reasons internalist”. It is true that Hume thought that the (full) explanation of an agent’s doings will need to cite more than his sensory awareness of the situation in which he is to act, coupled with whatever logical or mathematical inferences he manages to make from that awareness. It will need to cite some engagement of his desires or aversions. But Hume did not think that the reasons that apply to an agent in a situation derive from and end with the profile of those desires and aversions, and it is an elementary mistake to hold that any such absurdity follows from what he did hold. I fear that Bernard Williams is responsible for much confusion on this issue. Anyhow, for more on this I would draw attention to my paper ‘The Majesty of Reason’ in Practical Tortoise Raising, particularly pp. 287 – 8.It is also on my website, whose URL is below
    What are expressivism’s “substantive claims”?
    Expressivists, like the later Wittgenstein, think that although the levers on the engine all look alike, they do very different things. That is, the surface similarities (the propositional appearance) of language may disguise a functional plurality underneath the surface. Together with contemporary pragmatists like Huw Price, Bob Brandom, or Michael Williams, we think that the use of moral (modal, conditional, hypothetical…) language, indeed perhaps of all language, is most perspicuously theorised about in terms of use, and use can be given in more perspicuous terms than just saying “we use it to represent (describe etc.) the moral (modal etc.) facts.
    Hence an alternative meta-semantic story, which I developed in terms of attitudes, Gibbard in terms of planning, and which could also be given in other terms.The substantive promise of this is that it consorts with naturalism, it avoids metaphysics, and it avoids the creepy epistemology and complete failure to provide any theory of why we might be interested in morality, that are the attendants of alternative views.
    In call the theory ‘meta-semantic’ I mean that while it will become comfortable with attributions of truth, fact, representation, to sayings in all these areas, the explanation of how we get there, or as I have put it the explanation (or if we don’t like the word, the perspicuous representation) of why we think in these terms at all, is given in other ways.
    The only subjectivity that comes in is that if you are sincere it will be the attitudes or plans to which you are committed that you are voicing (there will be plenty of social and worldly pressures against licentious thinking). I have explained my reasons for denying that this implies anything along the lines of “if we thought differently the truth would be different” in many, many places. In the bits of this thread that I have read Singer and Lazari-Radek do not engage the fairly precise issues surrounding such conditionals, so, not knowing at all what they themselves intend by such labels as “objectivism” or “subjectivism” I simply want to go back into my tent and cover my eyes against the dust.

  5. Maybe Suikkanen should have mentioned explicitly that he had contributed to the mass of error that tires Blackburn. I read that line as a subtle mea culpa, myself, but the church likes penance to be more pure, public, and painful.
    Still, when Schroeder explained the mistake to him, Suikkanen displayed a virtue that is unfortunately rare in philosophers: he changed his mind. If the error now smells to him like tobacco to an ex-smoker, that’s understandable. Maybe rather than mock him for it we could just admire Jussi’s rare virtue, and even aspire to it ourselves.

  6. On the substance of the posting, I agree entirely with Mike Ridge and largely with Jack Woods, and I’d like to add one more point.
    Singer and de Lazari-Radek write:

    If Suikkanen seriously maintains that expressivism is neutral between these two claims[viz., “Expressivism can be supported by sound reasoning” and “Expressivism cannot be supported by sound reasoning”], we can reasonably ask why we should bother with a theory the proponents of which do not affirm that it is supported by sound reasoning.

    But de Lazari-Radek and Singer should distinguish between what a theory says and what its proponents affirm. Relativistic mechanics does not say that it is supported by sound reasoning, although presumably its proponents generally affirm that it is when asked. Even so we should take relativistic mechanics seriously.

  7. I’m with Mike Ridge. I really don’t get the force of the response.
    First about Hume interpretation – of which I said nothing in the original post and which I really don’t see the relevance of. I see nothing that commits Hume to reasons internalism in the claims that [the faculty of] reason is and ought to be the slave of passions and “Since morals …have an influence on the actions and affections, it follows, that they cannot be derived from reason.” [Treatise, Bk III, Pt I, Sec i.] Rather than making a link between reasons internalism and expressivism, these claims say nothing about whether an agent’s normative reasons depend on her desires. They do say something about the function of our mental faculties that react to reasons, which is quite different subject matter. So, I agree with Simon Blackburn very much on reading Hume.
    About the main argument in the form of the dilemma, Mike Ridge pretty much said what I would have said and so I don’t want to add anything on that either [thanks Mike!]. So, as far as I see, nothing in the response supports the idea that externalists are committed to reasons internalism and I find it odd to think that as a result of this expressivism would be trivial and uninteresting. After all, it is a deeply explanatory view about the meaning of normative language, the nature of normative thinking and the nature of moral properties.
    About my own paper, I think there’s quite a difference between the argument in it and the relevant sections in your book. I’ve also somewhat changed my mind as Jamie says but this requires explanation.
    At the time when I was writing that paper, there was a much discussed argument that expressivism collapses to metaethical subjectivism (by Jackson and Pettit). There were many very good responses to that paper, but I wanted to see whether something weaker could be salvaged. So, after a lot of work in the paper, I think I was able to show that expressivists are committed to material biconditionals of the form (in the case of reasons):
    (RS) That people live in extreme suffering is an objective object-given reason to give money to charity for Frank if and only if I am for Frank using the fact that people live in extreme suffering in giving-money-to-charity-friendly-way in deliberation.
    More than that subjectivist consequences I doubt we can get from expressivism as a metaethical view. This material biconditional does not commit the expressivist to reasons internalism (Frank’s reasons are not a function of his desires). It is also compatible with objective object-given reasons. The material biconditional finally doesn’t commit the expressivist to any kind of mind-dependency thesis – it’s purely a claim about extensions of two things in the actual world. So, from my argument, nothing like the views you endorse follow.
    I still think that expressivists are committed to material biconditionals of this form (I’m not fully convinced by Mark Schroeder’s attempt to argue to the contrary). But, here’s what I have changed my mind about (and this really is thanks to Schroeder). I used to think that the material biconditionals are a problem for the expressivists. Now I am less certain about this. The reason for this is that it seems like everyone is committed to biconditionals like this – including the most non-naturalist realists. The material biconditional is true if both sides are either true or false together. I presume that the left hand side is true for you (human suffering is a reason for Frank) and so is the right hand side (you are for him using the relevant consideration in deliberation). So, we all are committed to the material biconditional (in fact, denying it seems Moore paradoxical). So, if everyone has to accept these biconditionals, then they cannot be more of a problem for the expressivists. And I mistakenly thought that they were (or at least more general versions of them).
    One word about the strong language – and I do apologise for this. It is just extremely frustrating that the same objections keep getting made to expressivism – ones that expressivists successfully deal with. To my defence, I tried to formulate a new argument in my own paper. But, the arguments in The Point of the View have been made often and responded to by the expressivists. Parfit made similar attempts and there was a good response from Blackburn available. Had the book explained what is wrong with that response (in a way that you begin to do here), then I would have been much more sympathetic.
    I want to finish with a paragraph from Brad Hooker’s 1991 paper “Theories of Welfare, Theories of Good Reasons for Action and Ontological Naturalism”:
    “We can make parallel points about the relation between expressivism and the Present-Desire Theory of good reasons for action.The Present-Desire Theory maintains that there being a reason for S to A depends upon a connection between S’s A-ing and S’s desires. Expressivism, in contrast, maintains that judgments about what there is reason for agents to do express the appraiser’s attitudes about the doing of those acts. When one says ‘There
    is reason for Jack to A’ one is expressing a pro-attitude towards Jack’s A-ing, or at least towards one’s A-ing were one in Jack’s situation. And such a version of expressivism is perfectly compatible with the judgment that what there is reason for Jack to do is not merely a matter of what would best satisfy Jack’s present desires. So expressivism about judgments concerning normative reasons for action certainly does not support the Present-Desire Theory”
    This was written in 1991. More broadly, expressivists have been going on about this for decades now – Blackburn in a series of papers, Gibbard in variety of books. I think they and their views do deserve to be taken more seriously and given more respect than what your book showed, and this is what led to my frustration. Even if my own (perhaps failing) argument was against expressivism, it really at least tried to address expressivism on its own terms very hard. I also think that nothing in what I said goes as far as saying that ‘for expressivists nothing matters’ – that for them human suffering isn’t an issue. This is quite an attack on person.

  8. I think it is fair enough to be frustrated by this, given how often the error crops up no matter how many times it is shot down.
    In fact, it is a good sociological question why this mistake keeps being made over, and over and over again, often by very smart people, and equally often with no serious indication of any awareness and appreciation of the careful and extensive arguments by proponents of expressivism that it is a mistake. For a nice attempt to diagnose the persistent allure of this particular error, it might be useful to take a look at Sebastian Koehler’s recent paper, “Expressivism and Mind-Dependence: Distinct Existences,” *Journal of Moral Philosophy*, 2013 (http://booksandjournals.brillonline.com/content/journals/10.1163/17455243-4681039 [I think this paper is relevant to what Jussi still thinks is plausible in his old argument, actually]

  9. Mike’s link is broken (because the url includes the final period). Perform the necessary punctuectomy or use this.
    I haven’t read it, but just from the abstract it does look helpful!

  10. Perhaps expressivist’s frustration here is partly explained by their being forced to admit again that they have a merely contingent connection to a view as deeply attractive as subjectivism.

  11. Hi All,
    I am less sure than many others here about expressivism and externalism being happy bed-fellows.
    Here goes…
    I suspect that if we embrace expressivism we will be hard pressed to avoid the Williams/MacIntyre “bluff” objection to modern moral claims about there being authoritative moral obligations – roughly obligations of the sort that Kantians aim to defend. To explain, let me quickly make some set up points.
    First, Williams’ internalism is a view about the truth conditions for *possessed* reason claims (about reasons agents have); he is not giving an account of the truth conditions for *bare* reason claims (i.e. about reasons agents don’t have but in one way or another “apply” to them). Second, Williams’ himself defended internalism as a general view about all reasons, but I believe his main target was really the modern moral view that all rational and mature agents are subject to moral obligations and that they may be rightly held accountable for failures to live up to said obligations (see, e.g., his references to the internal reasons paper in his discussion of the morality system).
    Third, Williams’ arguments for reasons internalism are notoriously tortured and I want to focus on the connection to his attack on the morality system, so here are two claims that I think are plausible and that I take to embody the spirit of Williams’ internalist attack on the morality system.
    (1) If you want to claim someone is subject to a moral obligation and hold her morally accountable for living up to it, then you need to be able to justify the claim that she had compelling reason to live up to this obligation.
    (2) If the relevant reason claim is just a bare reason claim (i.e. it does not even purport to identify a reason they *had* to be moral) then the claim about her being subject to the relevant obligation is mere bluff and any attempts to hold her accountable are, in the ways Darwall has highlighted, abusive.
    Now given these claims, I want to explain why I think expressivism makes it hard to vindicate the modern moral view that all rational and mature agents are subject to moral obligations and that they may be rightly held morally accountable for failures to live up to said obligations. I will focus on Blackburn to try to get my idea out there.
    Blackburn seems to flirt with the idea that Williams could be right about possessed reasons but wrong about bare reasons (see Majesty of Reason, III). Let’s grant that this move makes sense when it comes to advisory reasons, which ground claims about what is advisable and which allow us to advise and grade others without lapsing into abuse or unfair assessment. But counsel is not command, and when we make obligation claims and hold people morally responsible we need to attend to this difference. If you are going to tell me I am subject to an obligation and then, say, blame or punish me for failing to live up to it, I think you need to be able to justify the claim that I *had* compelling reason to live up to the obligation. This is at least a plausible starting thought. So we need to look at how expressivism has us think of our reason claims in a context involving moral obligation and make sure this does not trigger worries about bluff and abuse.
    Here is where my worry hits the road, and I am open to being shown that I am making the same old mistake everyone is lamenting. If you claim I had compelling reason to help Joan, you are expressing that you are for my helping her. Since you think it is morally required, you are also a hard core advocate of my helping: you are for people who are for that, against people who are against, for my being punished or blamed if I don’t do it, etc. But if I could deliberate in excellent procedural and coherent fashion but fail to end up being for these things myself, then it is hard to see how your claim that I am *subject* to the obligation can be made in good faith; it looks like coercive bluff intended to change my attitude. And if you hold me accountable after the fact, or demand that I act as obliged before hand that does look like abuse.
    So while Blackburn can claim that it is *true* that I have an obligation to help Joan in a situation like this, the worries about bluff and abuse remain. In other words my worry is less about the truth conditions for the obligation claims than about whether expressivism allows people to, in good conversational faith, keep on with their practices of assertion and accountability.

  12. Brad. Thanks for this. A few points.
    (1) I don’t think anybody here is ruling out ex ante there being some compelling but non-obvious argument from expressivism to existence internalism. The main point is that this isn’t an obvious entailment and you do need an argument of some kind – and one that doesn’t beg the question against the accounts Blackburn et. al. have provided. Your argument here in my view falls into the category of such arguments (like Egan’s).
    (2) Note that the conclusion here is very specific – it is only when we are talking about moral obligation, which bring with them the apparent connection to blameworthiness and censure – that this argument kicks in. There is no suggestion here that whenever one has a practical reason or when some sort of practical ‘ought’ or other is true of you that this must answer to some desire.
    (3) Although I am myself sympathetic to the stuff from Darwall, the main premise that wrongness entails blameworthiness is, I think, rightly controversial. But I don’t think it will be useful to open a huge debate on that in this thread, to be honest; just worth flagging as this feeds into my point about the argument from expressivism to even a limited form of internalism is going to have to be highly non-obvious and controversial at this stage of the game.
    (4) Most importantly, I’m not seeing how this is espeially a problem for expressivists. Won’t your argument equally put pressure on cognitivists (whether naturalist or non-naturalists) to accept some form of existence internalism? After all, even if cognitivism is true, people are generally motivated to avoid the disapproval of others, and they also generally infer disaproval from negative moral judgment – this is true even if Moorean non-naturalism is true or if e.g. some form of reductive naturalism is true, I’d have thought. So *all* views are in the same boat here, if your argument works – either admit existence internalism (of some kind – the connection to possible motivation might be quite weak and counterfactual, after all) or admit that many claims of moral obligation are mere brow-beating, etc. I don’t see why the expressivist should be singled out as especially open to this line of argument, really. It actually looks like a direct argument for existence internalism about obligation, no matter what your meta-ethical sympathies. And, if it works then the expressivist can endorse it – they are neutral on this question, not committed one way *or* the other!

  13. I guess the problem is supposed to be just how an expressivist can make sense of *both* existence internalism about obligations *and* the thesis that every rational agent has sufficient reason to do the morally right thing. The worry will then be that their Humean philosophy of mind means that they can accommodate existence internalism only by holding that many agents don’t have sufficient reason to be moral – and so on this view don’t have any obligations. That is an interesting challenge, but it sounds like a challenge for *a lot* of views – certainly a Moorean non-naturalist will have trouble accommodating all of that, as will many forms of naturalism, for that matter. It is also open even to expressivists to allow that there are certain motives essential to being a rational agent – basically the sort of move that some Kantians make (Velleman, at one point, as I recall, and Michael Smith on some readings). This won’t make the expressivism otiose unless you think that *all* reasons must be derived from this essential motive, but why think that? So insofar as there are ways certain cognitivists can accommodate all of this, some of those moves are open to an expressivist too. And in any event, we are well beyond the idea that there is some obvious entailment from expressivism to any sort of unqualified existence internalism!

  14. Hi Mike,
    Thanks. That all makes sense and is very helpful! I agree that the premise is contentious and that related, but different, worries crop up for Mooreans etc. (for Parfit and company!).
    I guess I was mainly trying to push back against the idea that it is silly to think that expressivism is in tension with externalism, as that doctrine shows up in Williams. Some people (not you!) keep suggesting this is an old mistake that people are tired of calling out. I can see this when the worry is developed in terms of counterfactual conditionals with our having different attitudes in the antecedent, but maybe that is just a mistaken way that people keep trying to develop a real worry. For example I suspect the Kerstien should have said something more in the vein I did instead of trotting out the old conditional claim.
    I don’t think the same worry arises for the cognitivist, because I think the worry about bluff and abuse is especially strong when we bring to mind the expressivist picture of what we are doing in claiming that the person is under an obligation. But I have said enough!

  15. Brad
    That is an interesting argument worth thinking about more. I also think that Mike again makes very good points in response. One observation in general. I wonder if the argument is internal to moralising. It seems to me that there is a step of the kind you ought not to abuse or browbeat or bluff when you ascribe reasons (and that you ought to be able to justify your claims about obligations on grounds that hook up to people’s motivations). From this it might follow that what reasons agents have are relative to their motivational set. But, what does the work here is not expressivism’s meta ethical commitments but rather a specific normative view about holding people responsible. In this way, this argument would not show that expressivism entails reasons internalism but rather that some reasonable normative truths do. And when I say that expressivism can of course explain what I mean. The point is that the argument could not contain any normative claims as premises or expressivism itself would be off the hook. But maybe there is a way of making the arugument in a way that avoids this.

  16. For whatever this may be worth, I don’t find de Lazari-Radek and Singer’s suggestion that expressivism is obviously in tension with externalism frustrating, per se. But what I do find frustrating is that they don’t engage with the relevant literature on this topic. This sort of neglect of contemporary literature has become alarmingly common in many high-profile contemporary works in ethics. In his 1200+ page book, Parfit (2011, OUP) cites ~20 post-2000 works, and seriously engages in much less. In Kagan’s (2012, OUP) 600+ page book, he cites less than ~10 works, and only two of them were published after 2000. So, I think Suikkanen is right to worry about, among other things, OUP’s review practices. What’s going on over there?

  17. On this (from Brad): “I don’t think the same worry arises for the cognitivist, because I think the worry about bluff and abuse is especially strong when we bring to mind the expressivist picture of what we are doing in claiming that the person is under an obligation.”
    I take it that this rests on the idea that for an expressivist when we claim someone is under an obligation we are exhorting her to act accordingly or performing some such speech-act, and that this is a crucial part of the expressivist’s theory of meaning.
    But: (1) The idea that we typically or often do this when we address someone to tell them that they have an obligation should be common ground as between expressivists and cognitivists, since it is a datum that we do indeed make such claims in order to encourage the relevant sort of behavior. Even if Moore or Railton were right, it would still be true that people use moral claims to encourage pro-social behaviour by those who antecedently appropriately motivated.
    (2) The idea that we are always or typically doing this (exhorting our interlocutors) if expressivism is true may depend on a dated view of the positive theory offered by expressivists – e.g. a speech-act sort of theory of meaning (closer to Hare’s view) that most contemporary expressivists would reject. E.g., on a meta-semantic interpretation of expressivism, moral claims get their meanings from expressing distinctive states of mind, but the sort of state of mind expressed may well not be transparent to ordinary speakers (Cf. my reply to Terence Cuneo on moral assertion) – and I take it that without some such transparency assumption, the idea that expressivism makes the connection to prescriptive speech-acts stronger than on other views is unmotivated.
    (3) Quite apart from the worry about bluff and speech-acts, there is the fact that we hold people responsible for moral wrongdoing, which your argument rightly highlights. If the argument above works then unless those we hold accountable have some suitable motivation in their motivational set then this is a dubious practice. That looks like a fully general argument for existence internalism that doesn’t have much to do with expressivism in particular. Whether it is our speech acts or actually imposing more substantive sanctions on people that comes out as dodgy, if your argument is right then anyone who denies existence internalism is going to be left characterizing our practices as dodgy in some way or other. So again the point seems general.
    While I don’t think it is silly as such to think expressivism is in tension with externalism – if, e.g. you have an interesting argument for that tension, say! – I do think it is silly (and an old mistake) to think expressivism is somehow *obviously* and on its face incompatible with externalism – or, anyway, silly at this stage of the debate, given the extensive literature on this. Moreover, I take it that this is what Jussi was complaining that Singer and de Lazari-Radek had done. That suggestion of an obvious entailment relation, not standing in need of argument, is in sharp contrast to the fairly subtle and complex argument (one which relies on some pretty hefty premises!) you are offering for a more subtle and indirect connection between expressivism and (a somewhat qualified/limited) form of existence internalism in this post! Worlds apart, from the sort of mistake Jussi was complaining about, really.

  18. Hi Jussi,
    First, I don’t think it makes sense to have an internalist restriction on all bare reasons or possessed reasons claims. If I am counseling and not commanding you, then I can quite respectfully say “Jussi you have reason to eat better” even if I don’t think you can appreciate that fact without my intervention. And if you don’t follow this sensible advice I can think you are foolish without being disrespectful. Roughly attributive responsibility and advice can be appropriately grounded by external reasons in a way that moral accountability and command cannot. I guess my view here hinges on intuitions about cases in part, and on the observation that advice and command contexts are attended by very different normative presuppositions and burdens. Williams really muddied the waters, I think, when he ran together contexts of advice and contexts of warranted blame.
    Mike,
    Thanks again. Fair and helpful points all around. I will certainly have to look at your response to Curneo! I am a fan of the radical virtue ethical idea that we should try to jettison the deontic moral concepts from our thinking and practice, so I was not all that worried about the worry generalizing…but you are right that I should be careful about making it seem like this is just expressivism’s problem.
    I guess I would only add that expressivists like to make a big deal about their views not having any substantive normative consequences. But if I can get my argument to work (no easy feat) it might show that expressivism has normative consequences of the sort that Kantians and people like MacIntyre have suspected. Maybe cognitivist Mooreans are stuck with the same consequences, but that that just shows their views are normatively neutral either. At least their normative neutrality would hinge on the truth of my (admittedly contentious) premises.

  19. Brad: I wonder if you would say more about what the distinction between “possessed” reasons and “bare” reasons is supposed to come to. I worry it threatens to trivialize the debate if possessed reasons are just stipulated to be subjectivist-friendly reasons.

  20. Brad, this is quite possibly my fault, but I just can’t see how expressivism figures as a premise in your argument. To me it just looks like Williams’s “obscurity of blame” argument, which doesn’t have expressivism as a premise. So it just looks like an argument that all reasons must be internal reasons (if they support the practice of blaming those who have them and don’t act for them, I mean).

  21. Eugene: you are casting aspersions on OUP’s reviewing practices by counting the number of references to post-2000 publications in the books they publish. So just fyi, the bibliography of The Point of View of the Universe lists about 175 post-2000 publications. That doesn’t mean, of course, that our discussion is fully up-to-date with the literature in all of the areas we discuss. As we said in our original post, it is a wide-ranging book, with a focus on Sidgwick and the issues that were important to him. (We think he had a good sense of what issues are of enduring importance.) Accordingly, we only use the terms “expressivism” or “expressivist” six times in the book, and expressivism is one area in which, as Suikkanen’s original post and the subsequent comments have shown, we regrettably did not cover some of the current literature as well as we should have.
    We have been given a lot to think about by everyone’s comments. We will not be able to respond to all of them, but we hope to post, within the next day or two, a defence of the dilemma we posed in our original post, and, time-permitting, respond to one or two other comments.

  22. Hi Jamie,
    This is an rough idea & thanks for the push-back.
    I guess my basic argument looks something like this:
    (1) Gaugain denies that he has compelling reason to do the morally right thing
    (2) We have warrant to blame G only if we have warrant to claim that he is subject to a moral obligation to do the right thing
    (3) We have warrant to claim that G is subject to a moral obligation to do the right thing only if we can claim that he has compelling reason to do what is right and not fall into bluffing him/browbeating him.
    (4) If Humean expressivism is true, then our claim that he has compelling reason to do what is right will fall into bluff or browbeating.
    (5) If Humean expressivism is true, then we have no warrant to blame him.
    It sounds like Mike thinks that 4 *might* be plausible given the old school Hare semantics, but that it might not be plausible given the new meta-sematics.
    The main differences from Williams is that I do not introduce truth-conditions for reasons in a way that leads to predictable expressivist and externalist responses.

  23. Dave – that is a tough one. I guess for the purposes of my argument I am assuming that possessed reasons should just be logically linked to warranted claims about possessed reasons, and that we should assess warrant by appeal to (among other things) intuitions about bluff/browbeating and abusive speech. So someone who favors an objective view might have the intuition that it is not brow-beating for me to tell Gaugain that he has compelling reason to give up painting and be an unfulfilled but minimally decent family man. I don’t share that intuition so we disagree about whether Gaugain possesses compelling reasons. But there is no in principle reason to think this approach rules out an objective account of the reasons we possess.

  24. Thanks Brad, though I think that reconstruction packs all of the work in answering Jamie’s question into premise (4). Now the real question becomes, “What is the argument for (4)?” A closely related question is how specific that argument for (4) will be to expressivism – that is, whether an analogous argument won’t work for pretty much any view (many forms of reductionist naturalism, non-naturalism, e.g.) which doesn’t itself entail existence internalism.

  25. Expressivism, as I understand it (perhaps naively), is not merely a semantic view. It also has a metaphysical component. The expressivist agrees with the error theorist on metaphysics, but disagrees on semantics. Both views deny that there exists an objective moral reality. But the expressivist also denies, while the error theorist affirms, that moral language presuppose the existence of such a reality. (I’ve probably put this too crudely. But something like this general picture is, I hope, correct.) We can fairly call this shared metaphysical view “subjectivism”, as Mackie does, for example. So there is a form of subjectivism that is entailed by expressivism.
    I wonder whether de Lazari-Radek and Singer might be more charitbly interpreted as targeting the metaphysical part of expressivism, rather than the semantic. Some of what they say suggests that the issue is whether there exist any “objective” reasons, which sounds metaphysical. One way to put my question would be to ask: Do their objections apply equally to error theory? Or do they depend on some distinctively semantic aspect of expressivism?

  26. Hi Campbell
    it’s true that the expressivists will need to say something about metaphysics of normative properties. However, I’ve yet to come across an expressivist who agrees with error theorists about metaphysics. Allan Gibbard thinks that normative properties metaphysically just are constituted by natural properties (chapter 5 in Thinking How to Live). Blackburn has defended projectivism (Morals and Modals, for example). Neither one of them deny the existence of objective moral reality. So, I am sceptical about the claim that there is a shared metaphysical view between error theory and expressivism. And, I would not call that metaphysical view ‘subjectivism’.

  27. Hi Jussi
    I didn’t mean to suggest that there is complete agreement. Isn’t there some negative metaphysical thesis that both expressivists and error theorists accept? Simon Blackburn says above that an advantage of expressivism is that it “avoids metaphysics”. Doesn’t error theory share this advantage?
    My suggestion is that whatever this negative metaphysical thesis is — and, I agree, “subjectivism” might not be the best name for it — perhaps this is the target of de Lazari-Radek and Singer, not the exclusively expressivist semantic stuff.

  28. Hi Campbell
    yes – sharing an advantage seems right. It’s tricky, however, whether there is a negative metaphysical thesis they would accept. It seems to me that there is no easy way to formulate such a thesis. Perhaps a positive one – they are both views that are compatible with a naturalist, scientific view of the world – would be better. I’m not sure there is more than this that error theorists and expressivists will agree on. You might also be right that a metaphysical view is the target of their argument.

  29. Campbell: I think the difference is meant to be in the order of explanation. The realist and the quasi-realist will both say that there are mind-independent moral facts. The quasi-realist, however, begins the explanation of our practices (and the contents of our claims) with a sparser set of facts – the natural world and our responses to bits of that world. He then uses the quasi-realist machinery to “earn the right” to talk of mind-independent moral facts. Having earned that right, though, there is meant to be nothing wrong with deploying that talk – it need not be seen as defective, second-rate or systematically false. The realist, by contrast, starts with some priviliged set of properties and then explains how our moral predicates get their meanings by explaining how our moral language somehow manages to track or refer to those properties. The idea is that the latter order of explanation is less promising than the former for all sorts of reasons. The upshot is that it is not at all clear that there is a difference in the ontology, as opposed to a difference in what the parties to the debate take to be the correct explanation of the meanings of our moral predicates, etc. I fear you are projecting your penchant for what you like to call “bad-ass” philosophy onto theorists who prefer what you like to call “wussy” philosophy…a distinction I believe we owe to Meg Wallace…http://www.unc.edu/~megw/WussyBadAss.html

  30. Campbell: See, e.g. the second page of “How to Be an Ethical Anti-Realist” by Blackburn for one statement of this point about the order of explanation, though it crops up in several places in his writings.

  31. Jussi:
    That sounds right. So my suggestion — just to elaborate a bit — is that de Lazari-Radek and Singer are making an argument like this:

    Expressivists endorse a naturalistic, scientific view of the world. Such a view has no place for reasons, except insofar as they are somehow reducible to desires. So expressivists are committed to the rejection of objective (or external?) reasons. And this is bad because … (fill in as appropriate).
    The expressivist’s added layer of fancy semantics doesn’t help here. The expressivist might explain how we can use reason-sentences (e.g., “Jack has a reason to avoid pain on future Tuesdays”) without presupposing the existence of objective reasons, and therefore without falling into error. But that’s not enough. We need to say that there really are objective reasons, not merely that we can talk as if there are.

    That argument isn’t absurd, is it?

  32. Mike:

    The realist and the quasi-realist will both say that there are mind-independent moral facts.

    The realist and the quasi-realist will both say “there are mind-independent moral facts.” But I’m not sure they mean the same thing by this.

  33. Hi Campbell
    no – that argument is definitely not absurd (partly depending on how we fill in the (fill in as appropriate) part). Of course, the second sentence “such a view has no place for reasons, except insofar as they are somehow reducible to desires” is the bit expressivists are likely to reject (and for good reason). With that, the third sentence is denied as well.
    Also, expressivists would reject the very last sentence. There is nothing ‘as if’ about the reasons claims according to expressivists. Expressivists say that there really are objective reasons.
    In any case, I think that is a nice charitable reconstruction of the argument. It is not an absurd argument thus reconstructed. But, there are many claims in the argument which it’s not clear we should accept.

  34. Campbell: It is probably important to distinguish sentence meaning from speaker meaning a la Grice here. I am sure you are right that the realist and the quasi-realist speaker-mean to be conveying different thoughts when they make remarks about mind-independence. But of course you can speaker-mean to say something that your sentence doesn’t sentence-mean (in context) – since that is fixed by the relevant linguistic conventions (and context of utterance). The realist could just be using language in a deviant way. The debate then is about how we best explain the sentence meaning of sentences like “It is a fact that kicking dogs is wrong, and it would be even if we approved of it.” The quasi-realist has one story, and the realist has another. The crucial thing is that if the quasi-realist story is right then it captures what we mean in ordinary langauge when we make remarks about mind-independence. You can dispute that, but then you need an argument that they fail to accommodate the pre-theoretical data in some way.
    If the quasi-realist account does best explain the sentence meaning then I’m not sure how interesting it is that the realist can speaker mean to refer to some other sort of hypothetical moral fact or proposition.
    What is worse, it is not entirely clear how a non-naturalist realist could latch on to a determinate non-natural property to refer to even for purposes of speaker-meaning if they are wrong about the sentence meaning. “Which non-natural property to you mean to refer to, Derek?” can now not be answered with “the one picked out by ‘morally good’ in English” without begging the questions about sentence meaning. So it is not so clear how determinate meaning could be fixed even qua speaker meaning (for a non-naturalist, at least) if the quasi-realist is right about the sentence meaning.
    The naturalist could of course speaker mean something by ‘morally right’ like ‘maximizes happiness’ but then he will just be speaking his own little piece of local code if the expressivist is right (e.g. the Open Question Argument and similar arguments are meant to show that this couldn’t be what ‘morally right’ means in English). I mean, I could use ‘morally good’ stipulatively in the speaker-meaning sense to mean ‘done in Edinburgh’. If that has no plausible connection to what ordinary people mean by the phrase then I’m not sure why such purely stipulative speaker-meanings should be of any interest. The same goes for ‘morally right no matter how anyone feels about it’ as for the simpler ‘morally right’, in my opinion.

  35. I dont’ know Campbell, that argument seems pretty lame to me, to be honest! I’m not sure what the criteria are for ‘absurd’ and I don’t want to debate that, but at least one crucial problem is this.
    A key premise is, “Such a [naturalist] view has no place for reasons, except insofar as they are somehow reducible to desires.” But why think that? Any naturalistic property whatsoever can be the property of being a reason, compatibly with naturalism. Perhaps reasons can be glossed in terms of natural facts which explain why an action would accord with the Categorical Imperative, or maximizing the total amount of happiness in the world, or promoting knowledge is the relevant property, for all that has been said so far. Not all of those properties will entail mind-dependence in the usual senses in play here, much less subjectivism.
    The expressivist will of course have an account of what it is to hold that any one of those properties is (or, constitutes, if you are Gibbard) the property of being a reason for action or of being morally right, but so what? If they are right then that is the right account, and there is no ‘as if’ about it – there is no *other* sort of meaning associated with talk of moral properties in natural language if their view is correct, so you can’t just help yourself to the ‘as if’ idiom without some argument against that view in the theory of meaning.

  36. Campbell, perhaps you know this exchange, recorded in Ruling Passions (p. 319) between Simon and a graduate student at Kansas during a Q&A:
    “Q18: Aren’t you really trying to defend our right to talk ‘as if’ there were moral truths, although in your view, *there aren’t any really*?
    Ans. No, no no. I don not say that we can talk as if kicking dogs were wrong, when ‘really’ it isn’t wrong. I say that it is wrong (so that it is true that it is wrong…)…This misinterpretation is curiously common. Anyone advancing it must believe themselves to have some more robust, metaphysically heavyweight conception of what it would be for there to be moral truths REALLY, and compared with this genuine article, I only have us talking *as if* there are moral truths REALLY. I deny that there is any such coherent conception.”
    I don’t think you can help yourself to this more robust conception without some *independent* argument – and for the reasons I gave above, that argument will need to engage with the “fancy semantics” (meta-semantics, actually).
    [Also relevant is Simon’s reply to David Lewis who tried to paint quasi-realism as a form of fictionalism]

  37. Compare the following objection to a mathematical intuitionist:
    “But look, the twin prime conjecture is either true or its false. What you’re talking about is whether you can prove it or not. But that’s not related to its truth or falsity.”
    And, certainly, if you look at semantics for intuitionistic logic, especially with classical eyes, it will look as if the intuitionists are just talking about provability as opposed to truth. But what the intuitionists (at least some of them) crucially deny is that we can make sense of a notion of truth which exceeds provability (in some some sense of provable). So the above objection completely misfires by saddling intuitionists with a classical view of truth and falsity. It amounts to foot-stomping.
    Something similar seems to me to be going on when we saddle the quasi-realist with a ”metaphysically heavyweight” account of ‘objective reasons’ and the like. It amounts to a form of foot-stomping in favor of a view of the world the quasi-realist , in that good Dummettian phrase, ”prescinds from”.
    (NB: You can be an expressivist without being a quasi-realist. You can be a quasi-realist without being an expressivist. One is a view about semantics, the other a view about objectivity and reality. Blackburn and co tend to be both, of course. But one can have an expressivist semantics and an error-theoretic view of the metaphysics of so-called “normative properties” in the fashion suggested by Campbell Brown above. It’s just not the normal package the expressivist accepts. And it’s a rather problematic view when we think of the motivations for expressivism. But it exists in logical space.)

  38. Hi everyone,
    This comes with a little delay (and I am very unsure about it) but maybe here is an idea of what *might* drive the dispute between those who think it obvious that expressivism implies existence internalism and those who think the two positions to be entirely independent. The idea is that there is somewhat of an ambiguity in the notion of a reason that leads to different answers to the relevant question. Consider T.M. Scanlon’s recent suggestion that the basic reason-notion designates a relational property, R(p,x,c,a) – reading ‘p is a reason for agent x in circumstances c to do (or adopt) a’.
    Now here is the potential ambiguity. On the one hand, one might take p to be the ‘reason’, where p might or might not be a desire or desire-fulfillment-related fact. In that case – if one takes p to be the reason – expressivism will be neutral on existence internalism. In that case, the expressivist interpretation of statements like ‘p is reason for x to do (or adopt) a in c’ will supposedly very roughly be something along the lines of ‘Hooray for x counting p in favor of doing (or adopting) a in c!’ And this would hold for all cases of p, desire-related or not.
    Yet, one might find that take on the matter unsatisfactory, not to the least due to the then ‘factive’ charcacter of reasons. Thus, there is the ‘on the other hand’-part of the story. Instead of taking only p to be the reason, one considers THE FACT THAT the relation R holds, if it holds, to be the reason (maybe allowing for the vulgar to speak as if only p was the reason). In that case, the expressivist interpretation of reason-statements should accordingly, and again very roughly, be along the lines of ‘Hooray for it being the case that R(p,x,c,a) holds!’
    In the former case – i.e. in the ‘on the one hand’-part of the story – it appears that the expressivist story is indeed merely (meta-) semantic. ‘All’ it does is claiming that if we voice normative statements, we express our approval of some reason(s). And whether the reason is subjective, objective or what-not should be in an entirely different ballpark. In the latter case – in the ‘on the other hand’-part of the story – one might consider expressivism to do more than ‘just’ that (meta-) semantic job and instead as providing, in addition, an account of the truth-conditions of normative judgments, such that the reason-relation R holds iff one/we/some relevantly idealised actor voice(s) a pro-attitude towards it being the case that R hold. Thus, one might conceive of expressivism as here providing an arguably subjectivist account of the truth-conditions of reason-judgments and thus of reasons themselves. As a consequence one might think expressivism to imply existence internalism.
    Dunno. Probably the above goes wrong somewhere. But if so, it appears to me that this of all places will be the place to learn where.

  39. Hi Roland,
    I’m a little confused about the proposal. If, as Scanlon suggests, “the basic reason-notion designates a relational property, R (p,x,c,a)” then doesn’t that just mean that R is the property of being a reason (for an agent to perform an action in a circumstance)? Or, if you prefer, the property of favouring the action’s being performed by the agent in the relevant circumstance? If so then it would seem odd to say that the real reason is that R obtains – for that will amount to saying that the reason (or the “real reason”) to perform the action is that there is a reason to perform it – which hardly seems right.
    I think the right way to think about this sort of proposal is to think about various ordinary language claims about reasons for action and then see which ones you think demonstrate the relevant ambiguity. With that pre-theoretical data in place, you could then try to mount an argument that the best expressivist treatments will work on only one of the two relevant senses suggested by that data. You could then try to argue that only a realist construal will work for one of the two readings. But without some such anchor in the practice of ordinary normative judgment, I’m not getting the motivation for the ambiguity, quite apart from the worry I raised above about reasons not being facts that there are reasons. But I may well have misunderstood your suggestion.

  40. Hey,
    sorry for pitching in this late. I’ve been following this blog for some time now and thought it was time for my first post. I am also a little confused about Roland’s proposal. Specifically, I am a little confused about what relation R is supposed to be on the second proposal. You say that the fact that R obtains is the *real* reason to phi. If that is supposed to mean that one’s reason to phi is not p, but instead the fact that R(p,x,c,phi) holds it now seems like R cannot be the reasons relation anymore: the reasons relation has to be whatever relation holds between the fact that R(p,x,c,phi) (whatever it is) obtains and phi. But now, can’t we just give the normal expressivist interpretation about reasons statements for the second statement? Or am I missing some crucial feature here?
    I mean, I *can* see another reading of this proposal, namely that in there are some cases in which R is the reasons relation *and* the relation holding is one’s reason. Maybe there are cases in which someone has a reason to phi and the fact that this person has a reason to phi is a reason for me to psi. For example, maybe the fact that we have moral reasons to give to charity gives us reasons to blame people who don’t give to charity–I am not sure if this makes sense, my intuitions are very shaky when it comes to such statements. Are these the kinds of cases you had in mind?
    Supposing these cases make sense, it seems pretty clear that the expressivist should be able to give an account for these statements and that it would be relatively bad if the expressivist couldn’t do so. But, while I *could* see that the expressivist has some trouble giving an account of such judgements, I don’t see why the expressivist should be forced to accept existence internalism here. In general it seems like this version of the challenge amounts to just another instance of the embedding problem. Here the problem is: explain the contribution that the relational predicate R(p,x,c,a) makes when it itself is one of the relata of another relational predicate. But, whatever way the expressivist should go here it seems relatively clear that the expressivist should not let the predicate make in one instance the contribution that an attitude is expressed while in another instance it makes the contribution that we talk about the attitude. That would commit the expressivist to unmotivated equivocation! But, this latter option is the only way how I could see the expressivist being committed to existence internalism. Or am I missing something?
    Of course, independently of this, I am also not convinced that the expressivist cannot account for the above judgements in an appropriate way. For example (though I don’t have access to my copy of Impassioned Belief right now, so I cannot verify) I am relatively sure that Mike Ridge’s Ecumenical Expressivism can easily account for these judgements in a way that is not committed to existence internalism. Unless Mike (who should know his view best!) pitches in before me, I’ll see if I can look it up later.

  41. Yes I’d try something like this. Take the sentence, “The fact that there was reason for Joe to give to charity but he didn’t is a reason for us to blame Joe.” This sentence gets its meaning via the expression of a hybrid state, partly constituted by what I call a ‘normative perspective’ and partly constituted by a representational belief. Normative perspectives are complex non-cognitive states which (among other things) rule out standards of practical reason. The above sentence will have its meaning in virtue of its expressing (via robust linguistic conventions) the relevant normative judgment, where this judgment is constituted by a normative perspective and the representational belief with the following content: The notion of ‘admissibility’ is then cashed out in terms of what the corresponding normative perspective does not rule out. Because the judgment is multiply realizable across an indefinite number of normative perspective/representational belief pair, and no particular normative perspective is privileged by the meta-semantic story, there is no threat of a collapse into subjectivism. Note that no particular representational belief is going to be associated with a given token utterance of the sentence – it will rather be the indefinitely long list of normative perspective/representational belief pairs which is associated by the theory with the utterance. Hence no truth-conditions of any kind, much less specifically subjectivist truth-conditions, follow. Structurally the view is similar to what Schroeder calls ‘relational expressivism’ and also similar to Teemu’s version of expressivism in this respect.
    Obviously there are worries you can raise about this story, but they will be standard worries about the theory’s approach to the embedding problem. I don’t think there is a credible worry that the theory entails subjectivism (I discuss the subjectivism issue in the book in my chapter on truth). Sorry that is a bit of a digression from the main point, in some ways, but perhaps it helps illuminate Sebastian’s last comment.

  42. This discussion has been helpful. I’d like to try out another tack for filling in the argument from expressivism to subjectivism, which I’d be happy to be set straight about. It starts from Cian Dorr’s “wishful thinking” objection to non-cognitivism, which I’ll first describe. Dorr claims that a non-cognitivist cannot accept that it is rational to infer the conclusion of an argument like this having come to accept its premises:
    P1. If lying is wrong, then lying will be punished in the afterlife.
    P2. Lying is wrong.
    => C. Lying will be punished in the afterlife.
    Here is Dorr’s argument: Suppose Edgar starts out confidently accepting P1 but has no other evidence for C. Suppose now that Edgar comes to accept P2. Intuitively, we would say that Edgar may rationally infer C on the basis of P1 and P2. But Dorr contends that non-cognitivists hold that coming to accept a moral judgment like P2 can consist purely in undergoing a change one’s non-cognitive attitudes. But if this is so, then coming to accept a moral judgment like P2 cannot licence a change in belief about attitude-independent facts like C – that would just be wishful thinking!
    Contrast the above argument with this one:
    P1′. If lying is wrong then getting your little brother to lie is wrong.
    P2. Lying is wrong.
    => C’. Getting your little brother to willfully murder is wrong.
    Setting the Frege-Geach problem aside, non-cognitivists can unproblematically endorse someone’s coming to accept C’ as a result of coming to accept P2, because there is no wishful thinking involved here. The non-cognitivist can understand this second inference as undergoing a change in one non-cognitive attitude as a result of undergoing a change in another of one’s non-cognitive attitudes.
    Dorr’s premise about the nature of non-cognitivism seems right (at least it is true of pure non-cognitivists like Gibbard and Blackburn), and he seems right about the “wishful thinking” involved in making up one’s mind about attitude-independent facts on the basis of one’s non-cognitive attitudes. But it seems to me that we can derive a non-cognitivist commitment to subjectivism from these premises, with an argument placed in the non-cognitivist’s mouth as follows:
    1. Coming to accept a claim about attitude-independent facts purely as a result of a change in one’s non-cognitive attitudes must be irrational (this is the wishful thinking point).
    2. Coming to accept a moral claim purely as a result of a change in one’s non-cognitive attitudes need not be irrational (cf. the inference to C’ above).
    => C1. Moral claims are not claims about attitude-independent facts.
    3. “One has moral reason not to lie” is a moral claim.
    => C2. “One has moral reason not to lie” is not a claim about attitude-independent facts.
    => C3. It is not an attitude-independent fact that one has moral reason not to lie.
    => C4. If it is a fact that one has moral reason not to lie, it is an attitude-dependent fact.
    4. It is a fact that one has moral reason not to lie.
    => C5. It is an attitude-dependent fact that one has moral reason not to lie.
    Of course many pure non-cognitivists will also want to deny C5, interpreting it as a first-order normative claim. But how can they avoid also asserting it via this argument, and thereby endorsing subjectivism?

  43. Hi Simon,
    thanks for the suggestion. It seems to me that this argument feeds on an equivocation of “attitude-independent fact.” There is one reading of premise 1. on which the expressivist would deny 1. and one reading of premise 1. on which the expressivist would not deny 1.
    The reading the expressivist accepts is the one where claims about attitude-independent facts are those claim which express representational beliefs. The reading the expressivist rejects is the one where claims about attitude-independent facts include certain moral claims, such as “Its a fact that murder is wrong, and this would be so even if I didn’t believe it.” Given that this is a moral claim on the expressivist view, it wouldn’t be irrational to change your views on this on the basis of a non-cognitive attitude (as per 2). Unfortunately, one’s you’ve plugged in the version of 1. that expressivists accept, namely the one where a claim about an attitude-independent fact just being a claim expressing a representational belief, no problematic conclusion follows:
    1. Coming to accept a claim that expresses a representational belief purely as a result of a change in one’s non-cognitive attitudes must be irrational.
    2. Coming to accept a moral claim purely as a result of a change in one’s non-cognitive attitudes need not be irrational.
    C1. Moral claims do not express representational beliefs.
    3. “One has moral reason not to lie” is a moral claim.
    C2. “One has moral reason not to lie” is not a claim which expresses a representational belief.
    The conclusions that follow seem to me just parts of expressivism’s commitments. But now from this I don’t see how C3, C4, or C5 would follow.
    Another way to look at the argument and why I won’t work is as follows: let us distinguish a notion of a fact which is such that meta-ethical realists think moral facts are like that, as well as ordinary descriptive facts are like it. The (traditional) expressivist only thinks that ordinary descriptive facts are like that. We use “Fact” with a capital “F” to label those facts. Here is the argument you give:
    1. Coming to accept a claim about attitude-independent Facts purely as a result of a change in one’s non-cognitive attitudes must be irrational (this is the wishful thinking point).
    2. Coming to accept a moral claim purely as a result of a change in one’s non-cognitive attitudes need not be irrational (cf. the inference to C’ above).
    => C1. Moral claims are not claims about attitude-independent Facts.
    3. “One has moral reason not to lie” is a moral claim.
    => C2. “One has moral reason not to lie” is not a claim about attitude-independent Facts.
    => C3. It is not an attitude-independent Fact that one has moral reason not to lie.
    => C4. If it is a Fact that one has moral reason not to lie, it is an attitude-dependent Fact.
    But now note that the expressivist will not accept the relevant reading of 4. which is:
    4. It is a Fact that one has moral reason not to lie.
    This is so, because for the expressivist when you talk about moral facts, you aren’t in the business of talking about Facts. At most you are talking about facts (where talk about “facts” is accounted for e.g. by way of minimalism about facts). And so, for the expressivist C5 will not follow. But note, if we read the whole argument in terms facts rather than Facts, then the expressivist will block the argument at the stage of premise 1, because then she won’t accept the that premise anymore.
    I hope this is relatively clear and I am not missing anything crucial. But it seems to me that this argument won’t work.

  44. Hi Simon
    I think what this shows is that expressivism itself does not lead to subjectivism but rather expressivism and additional premises can be argued to do so. Then the question is are the expressivists committed to the additional premises. Here I assume that expressivists will deny the wishful thinking point – the first premise (and the additional one) in your argument. And, there is of course extensive literature on this (see responses from Enoch, Lenman, Schroeder, Mabrito and others).
    So, yes, with additional premises you can get an argument going, but it is less clear whether expressivists need to accept them. What we are looking for is to just start from the theoretical commitments of expressivism and see whether you can derive reasons-internalism from them. This I doubt.

  45. Jussi, thanks – I did not mean to challenge your objection to treating reasons internalism as an obvious entailment of expressivism. But I do think you went too far in saying that there is “no connection whatsoever between existence internalism and expressivism”. That remains to be seen. If expressivism together with some other undeniable or extremely plausible premises entails existence internalism, that would certainly be an important connection.
    Sebastian, I don’t think the argument rests on an equivocation on “attitude-independent fact” at all – you are simply tollensing my modus ponens, while still accepting as much of my first premise as possible. Of course, such a move is always logically available. Expressivists could similarly respond to (part of) the Frege-Geach problem, for example, by denying our intuition that the relevant arguments are valid or anything analogous to valid. My response is simply that my fist premise, understood in the broader way you describe, is extremely plausible (and in the literature, some of which Jussi cites, I haven’t found any good arguments against it, but I’d be happy to find out if I’ve missed one). For how *could* changes in one’s non-cognitive attitudes alone make it epistemically rational to change one’s beliefs about attitude-independent facts?

  46. Simon, what about a change in one’s plans, or intentions? Do you think a change like that could make it epistemically rational to change one’s beliefs about attitude-independent facts?
    Also, did you mean to endorse the label “wishful thinking” for the sort of process Dorr describes? That always seemed like a mistake to me. Wishful thinking is coming to believe that p because you want it to be the case that p.

  47. Hey Simon,
    sorry for being thick-headed, but could you please clarify what exactly the “broad understanding” of the first premise is?
    Here is how I perceived what is wrong with the argument: We are trying to saddle expressivism with a commitment to subjectivism purely on premises the expressivist herself ought to accept. For the expressivist to accept claims of the form “It is a fact that one has moral reason not to lie” though, she needs some understanding of what it is to talk about moral facts. Here’s what many expressivist (I think) would say: talk of moral facts is just more moral talk. To say that its a fact that giving to charity is obligatory is just to say (roughly) that giving to charity is obligatory.
    Now think about premise 1: we know how the expressivist thinks about moral facts, but what does she think about claims about “attitude-independent facts”? Well, that depends on how you read the notion of an “attitude-independent fact.” Here’s what Blackburn says: at least *some* claims about attitude-independent facts just are more moral claims. But now, the expressivist cannot jointly accept premises 1 and 2 anymore. But that just shows that the argument presupposes that the Blackburn move doesn’t work, rather than show this. What the argument needs to work from the very beginning is that expressivists can’t say that some claims about attitude-independent facts are moral claims. But that’s just to deny from the very beginning that Blackburn’s response to the mind-independence challenge works. Maybe you think that this is what is shown by the wishful thinking objection? Namely that no claim about a mind-independent fact could be a moral claim? I don’t see how Dorr’s argument can show this.
    I’ve always thought the problem with the “wishful thinking” problem was that its irrational to go from conative attitudes to descriptive belief (or whatever the expressivist thinks is expressed by ordinary descriptive discourse), which means that the expressivist has to deny the rationality of certain perfectly rational inferences. It’s not irrational to go from conative attitudes to conative attitudes, though, right? But, if Blackburn is correct that the expressivist can construe at least some claims about an attitude-independent facts as moral claims expressing conative attitudes, then the expressivist need not accept premise 1 as you state it. But note that this response is perfectly compatible with “wishful thinking” (at least as I understand it) still being a problem, on a modified reading of premise 1, so I am not presupposing, in my response, that there is no problem from “wishful thinking”.
    So, it still seems to me that the argument rests on an understanding of many of the relevant notions which expressivists will simply reject. Maybe I am still missing something, but consequently I can’t see how there’s a reading of all of the premises which an expressivist ought to accept (and this includes theses which *anyone* ought to accept – the point is simply not to beg the question against the sort of expressivism on offer) which leads to the desired conclusion that expressivists are simply subjectivists in disguise. So maybe you can help me along to see where my reasoning goes wrong?

  48. Jamie: I agree that the ordinary term “wishful thinking” is a bit narrow for what is being described. I understood it and adopted it as a technical term in this context, as no convenient term exists for the broader class of epistemic irrationality that consists in modifying one’s beliefs about attitude-independent facts to correspond to one’s non-cognitive attitudes.
    I can’t see how it makes a difference that the non-cognitive attitudes might be plans or intentions. How can changes purely in one’s plans or intentions can make it epistemically rational to change one’s beliefs about attitude-indepedent facts? If I form the intention to drink this glass of water, it may not be irrational to accept the belief that this glass will soon be empty. But this fact will be attitude-dependent in an important respect. (Similarly, if I believe that my mother only cooks what I like for dinner, and I come to dislike eggs, it may not be irrational to accept the belief that my mother will not cook eggs for dinner. But this is an attitude-dependent fact in an important respect.) Plans or intentions also presuppose facts about possibility. But it would be epistemically irrational for me to form the plan or intention to fly to the moon by flapping my arms, and then infer that I can fly to the moon by flapping my arms on the basis of my having adopted the plan or intention.
    Sebastian: the broader reading I had in mind was the one in which “attitude-independent facts” refers to just attitude-independent facts, and not only to “attitude-dependent Facts”, as you helpfully put it. I agree with you that expressivists’ further commitments may force them to deny that premise. But one cannot straightforwardly conclude that my argument begs the question! In fact, the expressivist reply you suggest looks like a question-begging insistence that coming to accept a claim about attitude-independent facts purely as a result of a change in one’s non-cognitive attitudes may be rational. I think expressivists need an argument for this point that does not presuppose the truth of expressivism itself.
    Facts about non-cognitive attitudes are obviously not attitude-independent facts, and what makes attitudes non-cognitive is that they are not belief-like states. Arguably, belief-like states constitute evidence for what is believed, and what is believed of course may be an attitude-independent fact. But desire-like states cannot similarly be evidence for what is desired (and, going back to Jamie’s point, plan-like states cannot similarly constitute evidence for what is planned, except insofar as what happens may be explainable by reference to our plans). This asymmetry between beliefs and non-cognitive attitudes is what makes it quite mysterious how it could ever be rational to come to accept a claim about attitude-independent facts purely as a result of a change in one’s non-cognitive attitudes. I think expressivists owe us an argument against this claim, if they want to deny it.

  49. Yes, those are examples of reasoning from plans to plan-free beliefs that seems irrational. But the question is whether it is irrational in general.
    Suppose Rhoda is a student driver. She has learned this rule: stop only when the light is red. She has adopted that plan, and accepts that conditional imperative.
    Rhoda is driving with her teacher when her teacher says, “Rhoda, stop.” She accepts this, without reasoning or deliberation. I would say that in so doing she has adopted a plan. As she leans gently on the brake pedal she infers, “The light is red”.
    Do you think Rhoda’s inference was irrational?

  50. Jamie: so you think Rhoda’s inference looks like this?
    Plan: Stop only when the light is read
    Plan: Stop
    => Belief: The light is red
    I think that’s irrational. What’s her evidence for the belief? *Maybe* it would be rational if she also had evidence for the belief that she only forms consistent plans. But that doesn’t seem to be part of your case. I think it’s hard to imagine your case without attributing to Rhoda a bunch of other beliefs that might rationalize the conclusion (that the instructor has commanded her to stop, that the instructor only commands stopping when the light is red, etc). But we should exclude these from consideration.

  51. …here’s an inference which may be analogous.
    Plan: Abstain from activities that are hazardous to my health.
    Plan: Have just one more cigarette.
    => Having just one more cigarette isn’t hazardous to my health.
    Looks like wishful thinking, no?

  52. Rhoda’s reasoning looks good to me. I would be very worried about someone who didn’t reason that way under the circumstances Rhoda is in.
    I agree that whether it’s good is going to depend on what else she believes, but that’s true for all reasoning.

  53. Jamie: I am interested but I don’t think I understand your point about Rhoda. If Rhonda’s reasoning is rational (absent some further implicit premise), so must my smoker’s be. But there is something wrong with the smoker’s inference. You may be right that with the appropriate background beliefs, it is rational to form beliefs about attitude-independent facts based on changes in one’s non-cognitive attitudes alone. But these background beliefs must provide for an explanatory connection between the facts and the relevant non-cognitive attitudes. For example, if I justifiably believed that natural selection had made humans desire just things that historically conferred a selective advantage, and I formed a desire for pomegranate, it may be rational to infer that pomegranate historically conferred a selective advantage. But this can only be rational because of the supposed explanatory connection between the fact (selective advantage of pomegranate) and the non-cognitive attitude (desire for pomegranate): my explanation of my desire invokes the fact that I then go on to infer. (We could even formulate a hypothetical case in which the very fact that p explains the desire that p, thus licensing a special case of “wishful thinking” in the everyday sense.) Similarly, maybe Rhoda justifiably believes that she won’t plan to stop if the light weren’t red. Then, when she forms the plan to stop, maybe she can rationally infer from this that the light is red. (I’m not sure, but I’ll grant that all such inferences are rational).
    But all that this point shows, as far as I can see, is that the first, “wishful thinking”, premise of my argument needs to be spelled out quite carefully to ensure it allows for this sort of special case. Have you got some reason for thinking that non-cognitivists always have recourse to explanatory connections like those that exist in the pomegranate case, whenever they are committed to treating inferences from changes in their non-cognitive attitudes to moral claims as rational?

  54. Simon,

    If Rhonda’s reasoning is rational (absent some further implicit premise), so must my smoker’s be.

    I don’t think so. Your smoker’s problem has to do with the way he acquired his plan, right? That’s what makes his case genuine wishful thinking, and that’s why I said (thinking I was agreeing with you) that the soundness of the reasoning depends on what else the reasoner believes.
    Compare Ralph and Sam. Sam believes that everyone who commits a crime will be punished. Watching his rival juggling, consumed by jealousy, Sam comes to believe that juggling is a crime. Happily, he infers that his rival will be punished.
    Ralph thinks that all red lights are stop signals. His teacher tells him there is a red light at the intersection ahead. Ralph infers that there is a stop signal ahead.
    Ralph is analogous to Rhoda; Sam is analogous to your smoker. The difference (between good reasoning and bad reasoning that share a logical form) shows up just as well when the reasoning is indicative. I think it has nothing to do with the plan-ladenness of your smoker’s bad reasoning.

    You may be right that with the appropriate background beliefs, it is rational to form beliefs about attitude-independent facts based on changes in one’s non-cognitive attitudes alone. But these background beliefs must provide for an explanatory connection between the facts and the relevant non-cognitive attitudes.

    That’s a bit vague. I don’t think it’s true, but in any case my real point is that whatever version of it is true is true also when the reasoning is, say, modus ponens with indicative sentences (that don’t express plan-laden beliefs).

    …Similarly, maybe Rhoda justifiably believes that she won’t plan to stop if the light weren’t red. Then, when she forms the plan to stop, maybe she can rationally infer from this that the light is red. (I’m not sure, but I’ll grant that all such inferences are rational).

    Maybe. As I was thinking of the case, Rhoda never stopped to think about whether she will plan to stop if the light isn’t red. She just accepts the conditional imperative, “Stop only if the light is red,” then on the basis of habitual (and let’s suppose justified) trust of her instructor, she accepts “Stop”, and from these she infers that the light is red. I think someone who can draw this inference in similar circumstances is in better shape than someone who balks.

    Have you got some reason for thinking that non-cognitivists always have recourse to explanatory connections like those that exist in the pomegranate case, whenever they are committed to treating inferences from changes in their non-cognitive attitudes to moral claims as rational?

    I’m not sure; I don’t know what the range of cases is that you have in mind. I think some modus ponens reasoning is terrible reasoning – everyone thinks that. (Mark Van Roojen pointed this out in a reply to Dorr.) So surely there is some terrible modus ponens reasoning involving imperatives or other plan-expressing sentences.
    I thought your claim, following Dorr, was that every instance of modus ponens in which the non-conditional premise expressed a plan was an instance of wishful thinking. I think Rhoda’s reasoning is fine, so the very general claim is false. But you think there are some examples that are not okay and that non-cognitivists must say is okay. This could be right; right now I’m claiming only that whether reasoning is (or is like) wishful thinking is not settled by whether one of the premises expresses a non-cognitive state. I think you agree, right?

  55. Mike, Sebastian,
    That’s been helpful. Though I also have to admit that I had a hard time to follow some of the more technical remarks – which is probably due to the fact that my conception of expressivism is a bit ‘dusted’… (i.e., I am not too well acquainted with the more recent literature). Be that as it may, I am not sure whether your remarks undermine my suggestion.
    Maybe another way of putting my suggestion would be in terms of the distinction between normatively relevant and normative properties. Picking up Mike’s advice and starting with some ordinary language claims about reason for action, I suppose that in everyday life (especially when it’s summer…) we will want to say things like: ‘Tim’s desire for ice-cream now is a reason for Tim to buy some ice-cream now’. Tim’s desire for ice-cream will be p, let ‘now’ roughly describe c, Tim is x, and ‘buy some ice-cream’ is phi. As the example suggests, in ordinary language, it appears we speak as if p was the reason.
    However – and that is what I was suggesting, I suppose – I am not sure whether we should import that way of speaking into philosophical discourse (and to that extent, of course, my proposal would be revisionist). There is no doubt, it appears to me, that in the given example p will designate a normatively relevant property. It picks out a property which, it appears to me, ordinary speaker are looking for when they are looking for normatively relevant features of a given situation. I.e. the property of Tim having a desire for ice-cream now will appear to be normatively salient to ordinary speakers. Nevertheless, I am inclined to doubt that p is here also, not the normatively relevant, but the – if you wish ‘genuine’ – normative property in the sense as being a property the instantiation of which licenses normative conclusion.
    Lurking in the background is, of course, Hume’s Law. From p alone in combination with an indefinite number of other descriptive statements, no normative conclusion will be licensed. This is the reason why our speaking as if p were the reason appears to me to eliptical (if not confused). Reasons, it appears to me, should be normative in the above sense (licensing normative inferences). Thus, it appears to me that instead of saying that, in the given example, p is the reason, we should say that the fact that p speaks in favor of x’s phi’ing in c is the reason. (Or to fill in the example again: we should say that the fact that Tim’s desire for ice-cream now speaks in favor of Tim’s buying an ice-cream now is the reason [for Tim to buy the ice-cream]; and maybe should even add that that relation’s obtaining plus p’s being the case in combination are the reason (?).)
    Now, I can see how this might sound odd (and, to be sure, I am myself not sure about whether it is odd or not…). It might seem odd because now, it seems, the notion of a reason appears twice and/or is self-referential. It appears in the form of R’s obtaining; and it appears in R itself in the form of the phrase ‘speaks in favor of’. But maybe we should not buy into the identification of ‘being a reason’ and ‘speaking in favor of’. Second, we might – and the functionalist might applause?! – just get rid of the description ‘speaking in favor of’ by claiming that the property of being a reason for someone just is the relation R(p,x,c,phi) – where we find out more about this relation by way of conceptual analysis (?!?; at this point my ‘convictions’ become very shaky).
    Against the backdrop of such an understanding of the notion of a reason, why should expressivism (about reasons in this sense) imply existence internalism? I think my original thought was that the expressivist would somehow be committed to claim that a speaker when ‘expressivistically applauding’ (and here my dusted views might shine through again…) for the fact that R obtains (such as when uttering sentences like ‘Tim’s desire for ice-cream now gives him reason to buy ice-cream now’) simultaneously ‘provides’ the truth-conditions for R’s obtaining. However, I now see that this is not the case; and that an expressivist could in any case be neutral on what provides the truth-conditions for R’s obtaining and just stick to his (meta-) semantic analysis of statements like the one in the example?!?
    In sum, I forgo my earlier suggestion on how expressivism might imply existence internalism. But would still be interested what your thoughts on the “what we should conceive to be a reason properly understood”-stuff are.

  56. Mike, Sebastian,
    That’s been helpful. Though I also have to admit that I had a hard time to follow some of the more technical remarks – which is probably due to the fact that my conception of expressivism is a bit ‘dusted’… (i.e., I am not too well acquainted with the more recent literature). Be that as it may, I am not sure whether your remarks undermine my suggestion.
    Maybe another way of putting my suggestion would be in terms of the distinction between normatively relevant and normative properties. Picking up Mike’s advice and starting with some ordinary language claims about reason for action, I suppose that in everyday life (especially when it’s summer…) we will want to say things like: ‘Tim’s desire for ice-cream now is a reason for Tim to buy some ice-cream now’. Tim’s desire for ice-cream will be p, let ‘now’ roughly describe c, Tim is x, and ‘buy some ice-cream’ is phi. As the example suggests, in ordinary language, it appears we speak as if p was the reason.
    However – and that is what I was suggesting, I suppose – I am not sure whether we should import that way of speaking into philosophical discourse (and to that extent, of course, my proposal would be revisionist). There is no doubt, it appears to me, that in the given example p will designate a normatively relevant property. It picks out a property which, it appears to me, ordinary speaker are looking for when they are looking for normatively relevant features of a given situation. I.e. the property of Tim having a desire for ice-cream now will appear to be normatively salient to ordinary speakers. Nevertheless, I am inclined to doubt that p is here also, not the normatively relevant, but the – if you wish ‘genuine’ – normative property in the sense as being a property the instantiation of which licenses normative conclusion.
    Lurking in the background is, of course, Hume’s Law. From p alone in combination with an indefinite number of other descriptive statements, no normative conclusion will be licensed. This is the reason why our speaking as if p were the reason appears to me to eliptical (if not confused). Reasons, it appears to me, should be normative in the above sense (licensing normative inferences). Thus, it appears to me that instead of saying that, in the given example, p is the reason, we should say that the fact that p speaks in favor of x’s phi’ing in c is the reason. (Or to fill in the example again: we should say that the fact that Tim’s desire for ice-cream now speaks in favor of Tim’s buying an ice-cream now is the reason [for Tim to buy the ice-cream]; and maybe should even add that that relation’s obtaining plus p’s being the case in combination are the reason (?).)
    Now, I can see how this might sound odd (and, to be sure, I am myself not sure about whether it is odd or not…). It might seem odd because now, it seems, the notion of a reason appears twice and/or is self-referential. It appears in the form of R’s obtaining; and it appears in R itself in the form of the phrase ‘speaks in favor of’. But maybe we should not buy into the identification of ‘being a reason’ and ‘speaking in favor of’. Second, we might – and the functionalist might applause?! – just get rid of the description ‘speaking in favor of’ by claiming that the property of being a reason for someone just is the relation R(p,x,c,phi) – where we find out more about this relation by way of conceptual analysis (?!?; at this point my ‘convictions’ become very shaky).
    Against the backdrop of such an understanding of the notion of a reason, why should expressivism (about reasons in this sense) imply existence internalism? I think my original thought was that the expressivist would somehow be committed to claim that a speaker when ‘expressivistically applauding’ (and here my dusted views might shine through again…) for the fact that R obtains (such as when uttering sentences like ‘Tim’s desire for ice-cream now gives him reason to buy ice-cream now’) simultaneously ‘provides’ the truth-conditions for R’s obtaining. However, I now see that this is not the case; and that an expressivist could in any case be neutral on what provides the truth-conditions for R’s obtaining and just stick to his (meta-) semantic analysis of statements like the one in the example?!?
    In sum, I forgo my earlier suggestion on how expressivism might imply existence internalism. But would still be interested what your thoughts on the “what we should conceive to be a reason properly understood”-stuff are.

  57. Thank you everyone once again for an interesting discussion. We have been enlightened by some of your comments and if our book ever goes into a second edition this discussion will help us improve our discussion of expressivism.
    We do not have time to respond to all of the comments made, but we do want to thank Campbell Brown for restating our argument (see his comment of September 09, 2014 at 08:06 AM) in a manner that Jussi Suikkanen agrees is “not absurd.”
    In response to Brown, Suikkanen maintains that “Expressivists say that there really are objective reasons.” They may say this, of course, but it isn’t at all clear how, on their view, “objective reasons” differ from reasons that are reducible to their present desires. At least as far as those expressivists who are quasi-realists are concerned, since they are not, after all, realists, there must be a difference between these positions. One crucial difference is how realists and quasi-realists understand objectivism.
    The passage Mike Ridge quotes from Simon Blackburn’s exchange in Kansas is something Blackburn has said many times – including when he gave a talk in a seminar series one of us runs at Princeton. He responds to a metaethical question by going normative. He finds it a “curiously common” misinterpretation, but there is nothing curious about wanting an answer that addresses the question at the level at which it was asked, or failing that, provides substantive arguments against the possibility of coherently framing such a question.
    It is worth noting that in this passage Blackburn acknowledges that his position rests on a denial of “a more robust, metaphysically heavyweight conception of what it would be for there to be moral truths.” Doesn’t this undermine the claims made by Suikkanen and his supporters that Blackburn’s expressivism is purely a semantic or meta-semantic theory with no implications for meta-ethical subjectivism? It has, Blackburn is here conceding, the implication that there is no coherent robust metaphysically heavyweight conception of moral truth.
    Sidgwick held a robust view of moral truth, and Parfit has also offered one, although he would deny that it requires heavyweight metaphysics. In our book we explain and endeavor to support Sidgwick’s view of moral truth. Perhaps Suikkanen’s initial post has given the impression that our argument is based only on Parfit’s case of the person with Future Tuesday Indifference, and then on the attempt to link the rejection of Hume’s view of reason with the rejection of expressivism, but for Sidgwick the argument also rests on his claim that there are moral truths, or axioms, that are self-evidently rational. We seek to defend this claim in our book. Our defense of it may or may not be sound, but it needs to be considered, as do Parfit’s arguments.
    Finally, we return to our earlier claim that “expressivism can be made defensible, a la Schroeder, as a semantic or meta-semantic theory, but only at the cost of emptying it of substantive claims so that it becomes trivial.” Some of the comments above deny that there is any such dilemma. We think that one illuminating statement of it comes from Schroeder himself. With his permission – but without implying that he endorses the connection we are making between it and the issue under discussion here – we will quote the conclusion of his “What matters about metaethics,” written for a forthcoming collection of essays on Parfit’s On What Matters, to be entitled Does Anything Really Matter? Parfit on Objectivity, edited by one of us, and to be published next year by OUP. (Schroeder’s paper has also now appeared in his Explaining the Reasons We Share, OUP 2014.) The passage refers to Parfit’s claim that if subjectivism is true, his life has been wasted:
    “… I’m inclined to think that the important issue about which Parfit cares is not quite the same as the issues that have been pursued in contemporary metaethical inquiry under the headings of reduction or noncognitivism. Rather, if what Parfit cares about is right, then though many metaethical views are indeed false, there is still a striking range of what I’ve called conservative metaethical theories – views which share a relatively common picture of the data, but offer competing explanations of it. Though all but one of these views are false, which one turns out to be true would not affect whether Parfit’s life has been wasted, and will have no consequences for Parfit’s arguments in the core chapters of On What Matters.
    Like the convergence between Kantian, Consequentialist, and Contractualist approaches to normative theory, the conservative approaches to metaethics which I’ve been discussing here share a common conception of some of the data. But unlike them, I don’t believe that they could merely be complementary paths toward the same truth (although contrast Gibbard [2003]). Rather, they are loosely like different orogenies for the same mountain – different theories about where it came from.
    If what you are primarily interested in, like Parfit, is how to get to the top of the mountain, then you may not care where the mountain came from. And if most of the people you talk to who do care where it came from are mostly concerned to try to convince you that that since they can’t understand where it came from, it must really be a flat plain, or that since they can’t understand how you could have gotten so high, you must not be climbing the same peak as anyone else, you are not likely to find orogeny very worthwhile. But it doesn’t follow that the mountain has no history. Even fellow climbers can pause, every once in a while, to admire the sweeping vistas, to rest up for the next leg of the journey, and to ponder whether this mountain was formed by subduction, volcanic action, or in some other way. It is true that many contributions to metaethics are like the orogenist telling Parfit that there is no mountain, or that everyone has her own mountain. But at its best and most interesting, metaethical inquiry needn’t be like that at all. It has room for many questions which can be pursued with an open mind even by mountaineers who share Parfit’s quest for the peak.”
    Like Parfit, we think that in comparison to the importance of getting to the peak (that is, discovering the truth about how we ought to act) questions about how the mountain came about (that is, about the best semantic or meta-semantic account of moral judgments) are much less significant. We concede that the term “trivial” was too strong. So we now restate our dilemma: “Expressivism can be made defensible, a la Schroeder, as a semantic or meta-semantic theory, but such a defense indicates that the questions it addresses are much less significant than the questions about the objectivity of ethics and of practical reason that Sidgwick and Parfit address, and which we discuss in our book.”

  58. Hi Katarzyna and Peter,
    Thanks for coming back on this. The restated version of the dilemma is more plausible in its claims, but perhaps not all that damaging. Not every philosophical question worth pursuing for its own sake will be as weighty as the profound question, “How ought one to live?” so I’m not so sure that those working on pure meta-ethical questions should lose too much sleep over this dilemma, to be honest. Though I’m not sure from the context whether you thought they should, or just how damaging the dilemma is meant to be. If it were damaging then I expect a huge percentage of philosophy, both contemporary and historical, would fall within its very broad scope. It reminds me a little of the issue dividing those who care about purely scientific research with no likely practical applications or “spin-offs” and those who think we should stay focused only on scientific questions we might eventually be able to transform into useful technology and the like.
    In fact, though, I think you may be too quick to dismiss the relevance of purely meta-ethical inquiry on these grounds. For one of the weighty and profound questions which would meet your criterion of significance is presumably whether nihilism is true – whether there “is a mountain” to stick with your metaphor, or whether we should be taken in by someone like the child in The Matrix who reminds us that there “is no spoon” and believe there “is no mountain.”
    Crucially, one of the potentially profound points of quasi-realist expressivism is that we don’t have to believe in any of the queer facts that Mackie went on about to believe in objective reasons in the only sense in which such beliefs can intelligibly be understood at the end of the day. If those arguments work (a big ‘if’, of course) then one of the upshots of quasi-realism is to provide an antidote to a kind of argument for nihilism, an argument which (like Mackie’s own argument) starts off with some heavy weight metaphysical assumptions about moralizing. So quasi-realism might, in the end, be very relevant to *insulating* our moral and normative practice from a kind of otherwise powerful nihilist challenge. That seems to meet the criterion of significance laid down by your dilemma above, or so I’d have thought. Again, you may not be convinced by the view for some reason, but that doesn’t mean that it isn’t offering a view which speaks, albeit indirectly, to the profound question of whether a good case for radical nihilism about objective reasons could be made out.
    In fact, I think that charitably understood quasi-realism also provides a good insulation against certain bad arguments for relativism too, and so helps guard against those who say we “each stand on our own mountain.” Such arguments often implicitly trade on a tacit cognitivism which fixes content in terms of speaker’s attitude, whereas the expressivist insists that attitudes come into the picture in a very different way, and in a way which does not entail relativism of any kind. By diagnosing this misunderstanding of the role of desires in meta-ethics, certain bad arguments for relativism (or, anyway, subjectivism) are disarmed too.
    So I’m not entirely sure that even this much more modest dilemma actually achieves its more modest aims.
    A few points of detail from earlier in your comments. At one point you say: “They may say this [that there really are objective reasons], of course, but it isn’t at all clear how, on their view, “objective reasons” differ from reasons that are reducible to their present desires.” Well, I’m an expressivist, and I say that the fact that it hampers the growth of knowledge is itself a reason not to put forward creationist ideas, and that it would be a reason even if nobody cared about knowledge. I think that is true, and I don’t know what in my expressivism suggests otherwise. Since facts about creationism hampering knowledge are not facts about desires at all, much less about my desires, and since the claim is that those facts are reasons even if nobody cares about knowledge, I have trouble seeing why that reason is somehow equivalent to my present desire. Of course, in saying that there is such a reason I am expressing a desire (actually a belief/desire hybrid on my own view) but so what? The whole point of expressivism is to offer an account of what it is to make a normative judgment which does not try to reduce normative facts to something non-normative. Why should our answer to ‘What is normative judgment?’ entail *any* view about the correct answer to ‘What is a reason for action?’ given the kind of expressivist answer we offer to the former? At any rate, the whole point of the expressivist gambit of changing questions is to avoid giving such an account. It seems to show a bit of a tin ear to find it unclear why their account doesn’t imply that reasons just are desires or some such.
    Finally, on this: “He finds it a “curiously common” misinterpretation, but there is nothing curious about wanting an answer that addresses the question at the level at which it was asked, or failing that, provides substantive arguments against the possibility of coherently framing such a question.”
    There would be nothing curious about it the first time it happened, perhaps. The point, though, is that the misinterpretation keeps happening no matter how many times Simon and others explain, in great detail why certain claims which might initially seem like purely meta-ethical questions, are actually normative questions, and hence open to an expressivist analysis.
    Moreover, the same misinterpretation keeps cropping up without any indication that there have been rather extensive efforts to explain why you can’t just assume questions about mind-independence are not first-order questions without begging the very question at issue. That this mistake would keep being made after having been addressed and diagnosed so many times is indeed rather curious.

  59. I find this increasingly frustrating as no progress is made and none of the original arguments are being addressed. I’ll just make one comment. If you deny that there is a coherent robust metaphysically heavyweight notion of moral truth (when you are not in the mood of climbing Ramsey’s ladder sideways), it doesn’t follow that you reduce reasons to your present desires (for one, if reasons reduce to your present desires, there is a coherent robust metaphysically heavyweight notion of moral truth). Neither does it follow that existence internalism is true or that you cannot give object-given reasons to your moral views.
    I also have to say that I am not sure I can find an explanatory and illuminating account of coherent robust metaphysically heavyweight notion of moral truths from the book and I think I fail to find such an account from Sidgwick or Parfit either. Anyway, time to do other things.

  60. Oh, also, on this bit: “It is worth noting that in this passage Blackburn acknowledges that his position rests on a denial of “a more robust, metaphysically heavyweight conception of what it would be for there to be moral truths.” Doesn’t this undermine the claims made by Suikkanen and his supporters that Blackburn’s expressivism is purely a semantic or meta-semantic theory with no implications for meta-ethical subjectivism? It has, Blackburn is here conceding, the implication that there is no coherent robust metaphysically heavyweight conception of moral truth.”
    Why think there being no robust notion of moral truth implies subjectivism? Simon endorses deflationism about truth, so he thinks there is no robust notion of truth, full stop, even about questions in physics. Does that mean he (or, say, Paul Horwich?) should think that the fundamental facts of physics are mind-dependent in some way? I don’t see the connection implied here between the right theory of the concept of truth and one’s view of what truths there are in some local domain (e.g. morality or physics). “Kicking cats for fun is wrong, and its wrongness in no way depends on our attitudes” – securing the legitimacy of that ought to be enough for mind-independence in the needed sense – regardless of whether one then goes on to endorses a correspondence or deflationist conception of truth.
    For what it is worth (shameless plug alert!), in *Impassioned Belief*, I develop a way for expressivists to think about truth that will allow them even to incorporate notions as robust as ‘corresponding to reality’ in whatever sense the correspondence theorist has in mind into their account of truth. That approach depends on the hybrid structure of my own form of expressivism and it would be way too much of a digression to lay it out here. But it does mean that quasi-realists can go even further in accommodating realist notions than this passage from Blackburn suggests. Whether they need to do so or not is another question, of course.

  61. Jussi,
    [this won’t be on expressivism though it will be an expression]
    As a hedonist utilitarian I’m becoming worried with you getting more and more frustrated! [joking] We won’t post anything more on that, I promise. [not joking, unless Peter wants to add anything]
    IF we are so limited in our responses it is because in order to treat you and others fair, we would need to go through a lot of literature. Neither of us has time for that now. We try to understand you and think of the problem but this will take us [and especially me] some time. Neither of us has been interested in expressivism much and this is our limitation. You have one too, I’m sure.
    I’m sorry you do not like our book. This happens. We worked on it hard and did try to make it as good as time and other things allowed. I think we did well enough, even if not perfect.
    I like seeing things in perspective. Philosophy is fascinating but what we are discussing here will have limited impact on really important things in life.
    Therefore, thank you Jussi [and everyone else] for exciting discussion! Thank you for spending time on teaching me something; I’m grateful you have exercised my brain! But let’s treat each other with kindness and be more relaxed.
    Thanks everyone and I hope to meet you at another soup!

  62. In the spirit of Katarzyna’s last post, let me offer a more constructive comment. In your previous post, I liked the fact that you were willing, in effect, to say ‘mea culpa’ and indicate that you’d like to rewrite the discussion of expressivism, a point you reiterate here, should the book have a second edition. Since you also don’t really want to engage with the complex literature on quasi-realist expressivism in the book, it seems (and perhaps with good reason, as this is a tangent from your main aims in the book), let me suggest another dilemma which might be sufficient for your purposes.
    “Simplifying a little bit, expressivism historically has come in two forms, the old school non-cognitivism of A.J. Ayer which denies that moral sentences are as much as truth-apt, and the fancier forms of quasi-realist expressivism from people like Blackburn which insist not only on moral truth but on mind-independent moral truth in the only sense that can be made of that notion – but within a broadly expressivist meta-theory. The former would be a threat to our project in this book, since ‘no moral truth’ trivially entails ‘no mind-independent moral truth’. But, fortunately, that old-school expressivism is not very plausible for a variety of by now familiar reasons and pretty much nobody endorses it. The latter, though, if it all works as its proponents claim it does (we are not sure it does, ourselves, but we don’t need to take a stand on that here) is no threat to objective, mind-independent moral truths, and so is not a view we need to oppose in any way. It is but one of a range of meta-theories which if otherwise sound could be combined with our most cherished theses about objectivity, etc.” I don’t know if you’d find that congenial, but it seems to me like a good way forward if there should be a second edition, myself, given where you’ve ended up on this.

  63. Thanks, Mike, for the suggested alternative dilemma, which is a much more precise and therefore better way of putting the point we were trying to make with our use of the passage from Schroeder.
    In his earlier comment Mike writes that if our dilemma is sound, then “a huge percentage of philosophy, both contemporary and historical, would fall within its very broad scope.” That may well be true, but is it an objection to our dilemma, or to philosophy as a discipline? That would be a worthwhile topic for another discussion, but like Katarzyna, I think it’s time to move on to other things now.
    Thanks Jussi for starting this off, and sorry that you don’t think any progress has been made. That is another point on which we are going to have to disagree.

  64. Tim Scanlon and Derek Parfit deny that there is any robust, metaphysically heavyweight conception of what it would be for there to be moral truths. So they seem to be impaled on the same horn as Blackburn and Gibbard (and Ridge and Schroeder), if indeed there is some dilemma. But I doubt there is.
    What is truth? What is a belief? What is representation?
    How do we meaning-makers and meaning-consumers stand in relation to our meanings, and to the world we represent?
    What makes ethical thought practical? How do our plans and emotions figure in what we say and how can they clash when we disagree?
    I would have thought no argument was necessary for the idea that these are profoundly significant questions. The opposed thought that only first-order ethics is important seems to me to be a fairly outrageous and academically parochial one, and prima facie extremely unlikely to be true.

  65. James, we didn’t say that only first-order ethics (I presume you mean normative ethics) is important. If that were our view, why would we have just published a book of which the first four chapters are about meta-ethics?
    Just to remind you, here’s how we restated our dilemma in our comment of September 15:
    “Expressivism can be made defensible, a la Schroeder, as a semantic or meta-semantic theory, but such a defense indicates that the questions it addresses are much less significant than the questions about the objectivity of ethics and of practical reason that Sidgwick and Parfit address, and which we discuss in our book.”
    Although Parfit denies that any heavyweight metaphysics is needed to defend the idea of objective truth in ethics, he does not reject a robust view of truth. We also drew that distinction in our earlier comment, although Jussi’s comment muddied the waters again by bringing back the reference to “heavyweight metaphysics” in connection with Sidgwick and Parfit.

  66. Mike p0inted out that not every philosophical question worth pursuing for its own sake will be as weighty as the profound question, “How ought one to live?” and added that if that were the standard then “a huge percentage of philosophy, both contemporary and historical,” would turn out to be relatively insignificant. Peter then asked, apparently seriously, whether this might just be an indictment of philosophy as a discipline.
    I was responding to that. I agree it doesn’t outright claim that only normative ethics is important, but if that wasn’t the general idea I wonder what was.
    I know that Parfit says he has a robust conception of truth that has no metaphysical commitments, but the question is what that is supposed to mean.

Comments are closed.