Moral Fictionalism and Mixed Arguments

In his wonderful MORAL FICTIONALISM, which I’ve just happily read, Mark Kalderon defends a very interesting metaethical view. Kalderon holds two positions that are not usually conjoined. But I am not sure they can both be coherently held, although I don’t know whether I can articulate my worries very clearly.

First, Kalderon defends moral factualism.  The content of a moral sentence is a truth-valued proposition, not an attitude or a commitment or anything else typically identified by nonrealists as the content of a moral sentence.  Kalderon thus faces no Frege-Geach problem, since moral predicates in embedded contexts and those in freestanding contexts have identical meanings.

Second, Kalderon defends moral noncognitivism.  To accept a moral sentence is not to believe it, but essentially involves a certain desire (the details of which don’t matter here).  Likewise, to utter sincerely a moral sentence is not to assert it, but to quasi-assert it.

Putting these two points together, consider this chestnut:

P1  If lying is wrong, then getting one’s little brother to lie is wrong.
P2  Lying is wrong.
C    Therefore, getting one’s little brother to lie is wrong.

Since Kalderon is a moral factualism, he (correctly) regards this argument as valid, just as ordinary moral cognitivists do.  But since Kalderon is a moral fictionalist, he doesn’t think that in accepting P1, P2, and C, I am believing P1, P2, or C.   I am accepting them in some other way.  Still, it would seem that if I accept P1 and P2, and am entitled to do so (whatever that ends up amounting to), then I am entitled to accept C.   (I hope that I have characterized his view accurately.)

My objection:

Consider the following ‘mixed’ argument

P3    Murder is wrong and illegal.
C*    Murder is illegal.

This argument is valid.  So, if I accept P3, and am entitled to do so, I am entitled to accept C*.  But the form of  acceptance that’s involved in accepting P3 is very different from the form involved in accepting C*: the former involves a desire, the latter is a straightforward belief.  But it strikes me as odd that I might be entitled to believe something on bases I am not entitled to believe.

Likewise, if I sincerely utter P3, and am entitled to do so, I am entitled to sincerely utter C*.  But the kind of speech act that’s involved in uttering P3 is very different from the kind involved in uttering C*: the former is merely a quasi-assertion, the latter is an assertion.  But it strikes me as odd that I might be entitled to assert something on bases I am not entitled to assert.

Kalderon emphasizes that we shouldn’t confuse the content of a moral sentence with the psychology of the speaker uttering it or accepting it.  This is true.  But there is still some connection between the content of a moral sentence (such as its logical relations to other sentences) and what the speaker is entitled to do with respect to it, given the speaker’s other entitlements.   

So I am wondering whether moral fictionalism really does overcome the Frege-Geach problem faced by ordinary noncognitivists.   Although it can explain why P3 implies C*, it seems unable to correctly explain why one is entitled to believe C* when one is entitled to accept P3.  Or, that’s how I see it, but I wish I could articulate my concern better.

31 Replies to “Moral Fictionalism and Mixed Arguments

  1. Eric,
    couple of points. First, Kalderon’s fictionalism is a version of non-cognitivism but this does *not* imply that relevant attitudes are anything like desires. They are not like desires at all. Desires have world-to-mind direction of fit and fictive attitudes do not. In his view, there are moral propositions which have truth-conditions and could be believed in the same way as sentences in Sherlock Holmes stories have truth-conditions and could be believed. But, Kalderon thinks that we do not believe moral propositions anymore than we believe Sherlock Holmes propositions. We rather entertain them. The difference is supposed to be revealed by the fact that we do not subject these states to the same epistemic norms as in cognitive domains. If we subjected these states to similar norms, then we would begin to have moral beliefs.
    The moral claims can be true or false in the fiction and these truth-in-the-fiction values can be presumably be used to account for truth-preservation in valid inferences. The moral quasi-beliefs do not aim at truth [full stop] but they do aim at truth-in-the-fiction and can be critised for failing to do so (as well as incoherence). And, for certain claims we have more warrant that they are true in this way.
    I guess in the example you give we should construe P3 as a shorthand for
    Murder is illegal and murder is wrong.
    Accepting this conjunction commits one to two things – believing that murder is illegal and quasi-believing that murder is wrong (in the moral fiction). So, P3 does not express a straight-forward atomic belief all the way through but rather a combination of two distinct states. Of course, if there is warrant for the attitudes expressed by P3, this warrant carries over to the both conjuncts separately.

  2. Oh yeah. One more thing. I don’t think he believes in moral factualism. That would suggest that there are moral facts – which he probably would deny. The view does commit him saying that there are moral facts in some possible worlds in order to get truth-conditions for the moral propositions – and thus content. But, he doesn’t have to say that there are moral facts in our world. And, according to him none of us would be committed to moral FACTS in our world – we just project them to the world as a fictive extra layer on the top of the naturalistic features.
    And, the sui-generis moral propositions have truth-conditions but not truth-values (except true-in-a-fiction and false-in-a-fiction). The fictionalists I take it want to say that only real beliefs and assertions that aim at truth can succeed or fail to be true really and therefore only they have truth-values.

  3. Jussi,
    It’s been awhile since I read the book, but I recall it pretty much as Eric describes. First, doesn’t Kalderon say that the relevant (non-cognitive) attitude we adopt in accepting a moral claim is desire in the directed attention sense? Second, isn’t his brand of fictionalism factualist? You gloss factualism as the view that “there are moral facts” (a not uncommon use of the word ‘factualism,’ to be sure), but on Kalderon’s use, ‘factualism’ names not that view, but instead the view that the contents of moral claims are propositions. At least that’s how I remember it. I also remember thinking that I found it more intuitive to think of the position as descriptivist (but still non-cognitivist), rather than as factualist, but maybe I’ve got quirky intuitions.

  4. Josh,
    well, he does spent awful lot of time arguing against all sorts of expressivist views according to which moral utterances express desires, i.e., pro-attitudes. You are right that he does bring up Scanlon’s desires in the directed attention sense. I think he says something like that moral acceptance is a matter of structuring your consciousness so that certain features are salient or appear as reasons. In moral attitudes things have seeming normativity for the agent.
    But, it doesn’t seem to be true that when things appear as normatively salient for the agent she necessarily desires to act accordingly. It may be true that she is irrational unless she does so. But, for this reason I’d hesitate to call the make-belief an instance of desiring anything. And, in the context of non-cognitivism debates in general it is quite misleading.
    You are also right that Kalderon uses factualism quite strangely. According to his gloss on it error theory would be factualism and naturalist versions of realism non-factualism. Problem with saying that he is a descriptivist is that I’m not sure that he would say that our moral attitudes describe our world – they do describe a fiction but they do not aim at correct description of reality.
    Getting back to Eric’s original post, I was thinking of the following inference:
    1. London is the capital of UK and the city where Sherlock Holmes lived.
    2. Therefore, London is the city where Sherlock Holmes lived.
    My intuition is that we would want to call 1. true and the inference sound. However, the second conjunct is only true in the fiction and not ‘really’ true like the first one. The attitude we have towards the first conjunct is then different from that which we have towards the second. The point is that this inference is exactly same as the murder inference. We should try to account for them in the same way. Maybe one way to do it is to say that the truth of the conjunction consists of the truth of the both conjuncts and their truth consists of different things – correspondence to the reality for the first and correspondence to the fiction for the other.

  5. Jussi wrote: “Problem with saying that he is a descriptivist is that I’m not sure that he would say that our moral attitudes describe our world – they do describe a fiction but they do not aim at correct description of reality.”
    I think it’d be better to say that he thinks moral sentences express propositions which represent (or describe) the world, but the attitutes constitutive of accepting these sentences don’t represent (or describe) anything. In particular, it seems problematic to say that these attitudes describe the fiction…at least if that means they are beliefs about what the fiction says (and what else could it mean?). Kalderon clearly thinks the attitudes are noncognitive states involving the affect that Scanlon describes as a desire in the directed attention sense.
    So maybe a better problem case is something like the inference from Dorr’s “Non-Cognitivism and Wishful Thinking”:
    (1) If lying is wrong, the souls of liars will be punished in the afterlife.
    (2) Lying is wrong.
    (C) The souls of liars will be punished in the afterlife.
    Or, lest we go fictionalist about the afterlife too, we could try:
    (1) If lying is wrong, then liars will be punished here on earth.
    (2) Lying is wrong.
    (C) Liars will be punished here on earth.
    Because he’s a factualist (in his sense), Kalderon can say that the inference is valid because of the relation between the propositional contents of each of the sentences. However, can he explain why it is rational to transition between the noncognitive states that constitute accepting (1) and (2) to the cognitive state–i.e. belief–that constitutes accepting (C)? It seems like he needs a story about the rational connections between these noncognitive states and the belief. If the noncognitive states have a desire-like direction of fit, then Dorr will say that this is a case of wishful thinking.

  6. I understand Kalderon’s views exactly as Josh characterizes them.
    Jussi,
    Thank you for the Sherlock Holmes example. It does seem that Holmes argument and the murder argument should be handled similarly. Yet accepting P3 essentially involves a desire, while accepting “London is the capital of UK and the city where Sherlock Holmes lived” does not. So maybe there is still a difference.
    I need to read the Dorr essay, pronto.

  7. Matthew,
    well, this is getting awfully complicated. First, I don’t think sentences are even in the business of expressing anything. Agents can express by doing things – they can utter sentences in order to make their attitudes known. Sentences might have propositions as their content. Second, I don’t think propositions describe or represent *the* world if this means our world. Desires have propositions as their content. Those propositions do not describe our world – they do describe a state of the world we would like our world to be in. Propositions might rather be sets of worlds where the corresponding sentence is true or functions from the sentences to truth-values in different worlds.
    I don’t see the problem with these attitudes describing a fiction. I think there is an answer to the question what else could that mean. There’s couple of ways to go on about this:
    1. Imagine I start to write a piece of fiction and entertain the relevant thoughts at the same time. ‘Once upon a time, there was a boy called Mark, who lived in a very big city, and so on and so on’. I am describing a fiction here in text as well as in my thoughts. But, these thoughts that describe the fiction do not aim at corresponding what is true according to this fiction – rather they constitute what is being case in this fiction. Unlike beliefs they cannot fail to be true of the fiction.
    2. Kalderon later on in the book characterises the truth according to the moral fiction being as what a virtuous person would see as a salient for her actions. If the reactions of morally virtuous agent gives the truth of the moral claims, then the acceptances of moral sentences can aim at moral truth and be true or false depending on whether they fit what the virtuous person sees as salient. The virtuous agent sets the frame of the fiction. He admits that different people might have different versions of the virtuous agent in mind and therefore be immersed in different fictions. This rather resembles some of the Mike Ridge’s view.
    The funny thing about Scanlon’s desires is that he thinks that when we talk about desires in everyday language we mean states that keep directing our attention to features of things that seem to be reasons for us – that seem to require actions from us. In most cases such thoughts come with affect – motivational pull. But, it seems an open question whether these desires really are as mental states desires in the philosophical sense of being states that have the world-to-mind direction of fit. Given what Scanlon says it seems just as appropriate to construe these desires as unreflected believes (‘that looks to me like a reason’ – ‘that looks to me like a duck) about reasons and say that in rational agents such believes about reasons create the corresponding desires. But, it seems possible to be failed to be motivated by these desires.
    I also think that mere propositional content of the sentences cannot solve the Frege-Geach problem. The content itself cannot explain why one is guilty of a logical mistake if one does not make the valid inference (Eklund’s review of Kalderon’s book is good on this). I do think he needs the fictive attitudes and a story about why some of them with certain combinations are logically impermissible. It is so easy to say that the explanation is that given the propositional content of the premises the form of the inference preserves truth (in-the-fiction). If this is not what is used to account for logical validity, then we need some alternative – and we need a story why that alternative is not available for the expressivists who too think that the expressed attitudes have propositional content.
    What would I say about the mixed inference you give? I would say that when we are immersed in the moral fiction we project a layer of moral facts to the world through our make-believing. The premises and the conclusion of inference would be about this compound world and the validity would be explained in normal way with truth-preservation.
    I just don’t think that Kalderon is committed to the idea that the non-cognitive attitudes have a world-to-mind direction of fit. If that was right and the moral propositions had sui generis content, then I would be wanting the world to have certain sui generis moral features. This seems rather non-sensical. We should ask him though – he often reads this blog.

  8. I’m not sure I see either how Kalderon has to worry about Dorr’s wishful thinking objection or the Frege/Geach/Searle problem.
    As I understand the latter, it is the problem of explaining the contents of embedded moral sentences in such a way that (among other constraints)they have the various logical relations that they do while deny oneself certain materials available to the cognitivist. As long as you think there is no problem in general with explaining logical implication for even ordinary descriptive sentences, the fictionalist can just give the standard story using the same story about the contents of the judgements that a cognitivist employs. ‘P or q’ used fictively has the same content as ‘P or q’ used to express belief, so it ought to have the same logical relations to other sentences in either case.
    That is unless I’m missing something. Where is the Eklund review?
    As for wishful thinking, my take on the worry there was that you should not form new beliefs based on something that is not evidence for what you come to believe. And I don’t yet see why a fictionalist would have to do that, even when reasoning in a way that employs sentences that contain both fictive bits and bits that are true in a straightforward way. (Lots of ordinary fiction has sentences of that sort as well.)
    Suppose I employ a ‘plumbing model’ when I do electical work, thinking of voltage as water pressure, amperage as the volume of water moving past a point, and resistance as kinks in the hose that make it take more pressure to push the same volume past that point. And I do so because I believe that thinking of it in this way will allow me to put the circuitry together so that it works. Now suppose I say, “Either the pressure has dropped, or this resistor is open.” What I say is not something I believe. But I say it precisely because I think reasoning in this “as if” way will lead me to true beliefs about what’s going on in the circuit. Now suppose I find that there is nothing wrong with the voltage and I say, “Well, it’s not the pressure.” And from this I conclude that the resistor is open.
    I don’t see anything especially worrisome here. It may be that what I’m really doing is changing my mind based not just on the steps of the argument, but also on the basis of the background thoughts. If I make sure to only introduce the “mixed” conditionals when I think reasoning using them will lead to reliable inferences, and I believe I have done this, it seems like a rational change of mind.
    If the fiction is employed for a certain rational purpose, and that purpose is one which is best served by using them as premises in arguments that will lead to change your non-fictive beliefs (when they contradict the consequents of conditionals with antecedents true in the fiction) you ought rationally to see to it that you only introduce conditionals which take you to true conclusions when you use them in this way. But supposing you do that, I’m not sure that you are being irrational to change your mind on the basis of using such conditionals.
    The plumbing/electronics story is a toy analogue of how constructive empiricists are supposed to think of talk of scientific entities that are postulated by certain theories. They think that until you get to talk of empirically accessible claims the talk is not literally true, but that using it will be empirically adequate, by which I take them to mean that the empirical beliefs formed on that basis will be true. And I believe that Mark Kalderon’s models for his own theory included such constructive empiricist views about science.

  9. Mark,
    sorry it wasn’t a book review – but rather Eklund’s Stanford entry on fictionalism that is here;
    http://plato.stanford.edu/entries/fictionalism/
    He discusses quickly Kalderon’s fictionalist solution to the Frege-Geach problem. I think the point is that there is more to that problem than the logical relation between the content of the sentences – the problem is also about how to explain the right kind of a mistake the reasoner makes in making invalid inferences.
    In explaining the latter, one needs to show that the relevant kind of attitudes are critiseable for failing to obey the norms of logic, i.e, for failing to track the logical relations between the descriptive sentences.
    I guess the worry is something like that the mental states ‘I desire [that I comment on Pea Soup and that I don’t comment on Pea Soup]’ and the state ‘I believe [that I comment on Pea Soup and that I don’t comment on Pea Soup]’ have the same propositional content [p&~p]. Only in having the latter attitude, I am making a logical mistake for which I am critiseable. I take it that the fictionalist will have to say that there is something about the fictive attitudes that make them more like believes in this respect than desires. Maybe contradictory contents would fail to constitute fictions or something like that. But, in any case this goes beyond saying things about the content.

  10. Just a footnote on Mark’s point about similarities between constructive empiricism and Kalderon’s fictionalism. As I remember van Frassen’s view, he held that belief in the non observable posits of an empirically adequate scientific theory was rationally optional, whereas no doubt Kalderon urges more skepticism about morality.

  11. Thanks for the reference Jussi. Eklund contrasts what is “literally expressed” by what is “actually expressed,” in setting up the problem. I’m not sure I understand the contrast entirely, but it looks like a distinction between what the sentence literally says, and perhaps the state of mind that a person uses the sentence to express (in the sense of ‘express’ postulated by non-cognitivists when they say such and such judgements express concognitive attitudes.
    I guess I think that viewed that way the worry is really very closely connected to Dorr’s wishful thinking worry – that this kind of change of mind (the change from the first state to the last via the intermediate state expressed) is not rational change of mind.
    I guess that leads me to two thoughts. (1) if the reply about Dorr I gave above works then Kalderon is OK.
    (2)One strategy for solving the Frege-Geach problem for non-cog is to explain why the states of mind are such as to require such patterns of attitude revision in light of new commitments, and then to generate the logic of the sentences that express these states out of those materials. For this approach, handling the problem suggested by Eklund is the first step to solving Frege Geach. But the fictionalist is taking a different approach to the logical relations between the sentences – one that does not require solving this problem first. This is not to say you don’t want to answer the question, but it is to say that fictionalism seems to make it less pressing to answer it before we solve Frege-Geach.
    David,
    You’re right about the differences between CEs and Kalderon’s actual views, I think. But it would be possible for someone who was not sure the anti-realist arguments were all that strong to suspend belief in the truth of the sentences uttered and be like some constructive empiricists in her first level attitudes.

  12. I say some stuff about both (a) the desire/make-believe issue, and (b) the taxonomical issue (factualism and fictionalism) in my NDPR review of Kalderon’s book, here:
    http://ndpr.nd.edu/review.cfm?id=6321
    The following excerpt might be helpful:
    Moral attitudes, then, have a phenomenological character — they involve certain factors in a circumstance seeming to be reasons. Kalderon claims that the best explanation of this phenomenology is as a particular kind of affect: a desire in the directed attention sense (following Scanlon). To have a moral attitude is to structure your experience so that certain features are salient or appear to be reasons; it is “literally” to decide how you feel about the circumstance (50-1). This is a form of make-believe because it involves structuring our consciousness as if we really believed the moral sentence. (It is unclear what Kalderon would say about the amoralist, who accepts moral claims without motivation or recognition of reasons for action.)

  13. Hmmm. I find Eklund’s paragraph on Kalderon on Frege-Geach kind of cryptic, but I recall that when he was working on his entry, he wrote and asked about my reply to Mike Ridge’s hybrid account, and said that the problem I was pointing out for Ridge was very similar to the one he was pointing out for Kalderon for his SEP entry. (I’m no longer sure this is true.)
    Neither is related (at least directly) to the Wishful Thinking problem, which arises in the case of arguments with descriptive conclusions; the argument Eklund considers has a normative conclusion, and the problem for hybrid views that I discussed in my reply to Ridge requires an argument with a normative conclusion.
    I’m not sure what it is that is worrying Eric about the argument with the descriptive conclusion that doesn’t worry him about the argument with a normative conclusion, though, unless it is akin to Dorr’s Wishful Thinking problem. If the problem is just that merely going fictionalist, while it allows you to retain formal, semantic properties of arguments, doesn’t yet explain why accepting their premises commits you in some way to accepting their conclusions, then I think it is akin to Eklund’s worry, and can be stated just as well without the restriction to arguments with descriptive conclusions.

  14. Mark,
    You’re right that Eklund’s worry can’t be precisely Dorr’s. I was being very coarse-grained when I said Eklund’s concerns seemed to me to Dorr’s rather than Frege-Geach. What I meant was that he was concerned with reasoning and the relations between states of mind than with logic per se. Standard non-cognitivist accounts use the former to explain the latter, but the fictionalist doesn’t seem to need to do that. Obviously the distinction between logic and reasoning is somewhat controversial and I’m showing Gil Harman’s influence by employing it here.
    In any case, you’re right to point out that Dorr’s objection is to reaching non-normative conclusions from normative premises given the non-cognitivist construal of normative claims as expressing desires. And Eklund’s example is not of that sort.

  15. About the Sherlock Holmes analogy:
    Consider the following three arguments.
    A1. London is in the UK and is the city in which Sherlock Holmes lives.
    A2. Therefore, London is the city in which Sherlock Holmes lives.
    Next,
    B1. London is in the UK and is the city in which Sherlock Holmes lives.
    B2. Therefore, London is in the UK.
    But, now, pretend that for some reason, Doyle wrote the Holmes stories such that they were set in a parallel world in which London is in France rather than the UK, and consider this argument:
    C1. London is in France, and London is the city in which Sherlock Holmes lives.
    C2. Therefore, London is in France.
    Clearly, in the case of the third argument, we have to be careful to understand the conclusion and its premise as applying to Doyle’s fictional world.
    But then, why not understand arguments A and B as being meant to apply to Doyle’s fictional world as well? Its easy to think this about argument A, of course–but why not also understand argument B this way? The only relevant difference between argument B and the other two is that the conclusion of the second argument is expressed by a sentence which, in other contexts, would happen to be literally true rather than fictionally true. But “other contexts” shouldn’t be relevant to our interpretation of the argument in its “present” context.
    Now what about this argument:
    D1. Charity is wrong and illegal.
    D2. Therefore, charity is illegal.
    Again, clearly, this argument’s conclusion (and premise) should be understood as acceptable only if interpreted as applying to some fictional world’s morality–and hence, as applying only to a fictional morality. But according to fictionalism, _our own_ morality is fictional. So, what should we say about the following argument now:
    E1. Murder is wrong and illegal.
    E2. Therefore, Murder is illegal.
    ?
    Here’s what I’d say: That when we accept this conclusion, we accept it only as a fictional one–but that we also immediately recognize it as expressing a non-fictional truth as well. To be clear, as non-fictionally true, we do _not_ accept the sentence as a conclusion to _this_ argument. Rather, as a conclusion to _this_ argument, we accept it as fictionally true. Its just that we can barely help but notice that the conclusion is expressed using a sentence which can also express a very salient fact about _non-fictional_ legality as well.
    -Kris

  16. Kris,
    I’m not sure why we would think that. The Frege-Geach lesson is that we should understand the propositions of the premises in the same way as the propositions in the conclusion. p&q; Therefore, q is valid – p&q; Therefore, r is not. I think the presumption is that we should understand by default sentences in a non-fictional way.
    Unless we have special reason to think otherwise we can assume that people talk about the world as it is. If someone says that London is in France they would be making an obvious mistake and charity would require us to in that special case to assume that the are talking about Doyle’s fictional world in which we now are assuming this really is the case.
    I think this carries over to the mixed moral inferences. The assumption is that the people mean the non-moral premises to be taken at face value. I’m not sure what reasons there would be to assume assume otherwise. And, the moral fiction some might say we are immersed in does not specify the whole world all over again but rather add extra features to our actual world. So, I would think that we have no reason to think that the non-moral conclusions we get from the mixed inferences are nothing but claims about the actual world.

  17. I was trying to show how a fictionalist about morality (or about anything else for that matter) might answer Eric’s question: Why are we entitled we _believe_ C* when we only entitled to _accept_ P3? My answer is that we are only entitled to _accept_ C* as a conclusion to the argument. When we are entitled to believe it, we are so entitled not because of the argument “P3 thereforce C*” but rather because of some other fact.
    On the account I gave, conclusions are understood only in the same way that premises are understood (just as you say they should according to the lesson we learn from Frege-Geach). When premises are fictional, conclusions are fictional. When premises are non-fictional, conclusions are nonfictional.
    This raises the question what is supposed to happen when there are both fictional and non-fictional premises. I was trying to point out that it is available to a fictionalist to plausibly insist that this can not happen. If there is any fictional element in the argument, then the whole argument should be taken as being about a fiction.
    Against this point, you bring up the following consideration: “…the moral fiction some might say we are immersed in does not specify the whole world all over again but rather add(s) extra features to our actual world.” In other words, you are effectively denying what I just claimed, namely, that when we are talking about a fictional world, everything we say about it is to be understood as fictionally true. Rather, you say, some of the things we say about it should be understood as _not_ fictionally true _but instead_ actually true.
    I do not think a fictionalist would (or at least, would have to) agree with this. In the fictional world of Doyle’s stories (now I mean the ones he actually wrote, not my strange alternative version,) it is true that London is in England. Importantly, it is also true non-fictionally that London is in England. But this sentence’s (proposition’s?) being true nonfictionally is different than its being true fictionally. Doyle’s story is set in a fictional world, and so, everything true in Doyle’s story is true fictionally. The fact that something is true non-fictionally does not exclude the possibility that it might be true fictionally as well.
    So if the above is correct, then if we are going to be fictionalists about morality, then what we can say about the inference from P3 to C* is that P3 and C* are both to be understood as true fictionally, because P3 is true only fictionally. If we accept C* on the basis of accepting P3, then we’ve accepted C* as fiction. If we accept C* non-fictionally, then we’ve accepted it on some basis _other_ than the argument from P3.
    Does that make any more sense?
    -Kris

  18. My apologies for the poor editing in the previous post.
    It should have been addressed to Jussi, of course.
    “Fact” at the end of the first paragraph should have been “consideration” instead.
    There are some misspellings as well, but I think none that make the post unclear.
    Again, my apologies.
    -Kris

  19. Kris,
    Your proposal seems like a possible option for MF. Yet I think Kalderon actually thinks that if one accepts C*, one believes C*–belief is the form of acceptance appropriate for sentences like C*. This is what puzzles me.
    Having just read Dorr’s short paper, I now see that my original post was not that, um, original. Or maybe it is only in the sense that fictionalism, which _seems_ to account for the role of moral predicates play in arguments better than other forms of noncognitivism, still may not be entirely out of the woods.

  20. Kris,
    I think you have located our disagreement. I don’t think moral fictionalist should claim that when we use moral language we are talking about some other fictional world than the actual one. Rather, we are talking about fictive features of our world. I think this is much more plausible – it’s also the way in which Blackburn construes fictionalism in his criticism. We shouldn’t get carried away with the analogy from novels. The analogy is only supposed to tell us something about the nature of the attitudes and not so much of the content.
    I do think that in ordinary discourses we need special reasons to understand any statement as fictive. Kalderon tries to argue that we can find reasons to think that our moral utterances are fictive – namely from the kind of epistemic norms that govern these mental states.
    Anyway, take the kind of mixed conjunction we have talked about – ‘London is south of Paris and murder is wrong’. What do you have to show this conjunction false? Well, one think that looks enough is to compare where the sun is at noon during the same day in London and in Paris. If it is located higher above the horizon the the first conjunct is false. I think we would accept this evidence as falsifying the conjunction as uttered by someone in ordinary talk. If the fictionalist account was that the whole conjunction was about some other world then this would not do as evidence. The evidence could not be found from our world but rather from the fiction where-ever it is framed.
    The fictionalist wants to claim that things are different with the other premise. Listing the naturalistic features of which our world consists does not falsify the moral premise even when wrongness is not amongst those features.
    I don’t think there is a mystery here as long as we think that accepting a conjunction is not a distinct attitude but rather consists of accepting each conjunct. The fictionalist can then say that the acceptance of the conjuncts can consist of having different attitudes. Of course in the inferences we have talked about if we have warrant for accepting the conjunction, i.e, for accepting each conjunct, this warrant transfers in the inference to the conjuncts individuals.
    The warrant is in addition for the kind of attitude we had for the conjunct in the first place – whether it was believing or make-believing. What kind of warrant is required for the conjunction depends on what kind of warrant is required for each of the conjunct. If the conjuncts are different, then the warrant is different too – in the mixed moral cases the warrant for the non-moral claim differs from the warrant for the moral claim. I do think that we should retain the idea that warrant transfers across valid non-question begging arguments.

  21. Eric,
    I guess, then, against Kalderon, I’d say that the form of acceptance appropriate to a sentence is not determined by the sentence itself, but rather, by the sentence and its context (at least).
    Jussi,
    To clarify, I think a moral fictionalist should say that moral talk is fictional talk about the actual (non-fictional) world. But I also think that we should say that Doyle’s stories constitute fictional talk about the actual (non-fictional) world.
    On reviewing the posts in this thread, I think you and I are reading the two example arguments in different ways. I’ll write the arguments down here for clarity:
    Q1. Murder is illegal and wrong.
    Q2. Therefore, murder is illegal.
    R1. London is in the UK and Sherlock Holmes lives in London.
    R2. Therefore, London is in the UK.
    Now, I’ve been reading the premise of the second argument as being fictional through and through. It may be _about_ the actual world (as I think it is,) but only _fictionally_ so. For this reason, I read the conclusion R2 as fictional as well.
    I’ve been reading the first argument analogously. On behalf of the MFist, I’ve said the premise of the first argument should be read as fictional through and through–fiction _about_ the actual world, but fiction nonetheless. This means, I think, that the conclusion Q2 should also be read as fictional.
    If this is the way to read the example arguments, then what does it tell us about the answer to Eric’s puzzle? Looking at the R-argument, we see that the answer might be, you’re _not_ entitled to believe the conclusion, qua conclusion. You’re only entitled to accept it. If you believe it, (and you should!), you should be doing so on some basis _other_ than that offered in the R-argument itself.
    I argued that an MFist can plausibly say this about the Q-argument as well. Eric says he thinks Kalderon would not say we should accept the conclusion “Murder is illegal,” but rather, believe it. If this is true, then what I am saying is corrective to Kalderon–if he wants to maintain his fictionalism, and if I’ve read the two example arguments correctly, then I suspect he’ll have to say something like what I’m saying. Is there some reason Kalderon _must_ say that the conclusion is to be believed rather than accepted? (Of course the sentence “Murder is illegal” should be believed, not just accepted, but it can be that we should believe a sentence in some contexts, but only accept it in others. The R-argument illustrates this, on my reading of it.)
    Meanwhile, you’ve been reading the arguments differently. You read them as, in a sense, changing topics during the course of the premise. (I don’t mean that as a criticism.) They start out talking non-fictionally, but end up talking fictionally.
    On this reading of the arguments, I think what you’ve said is right. We can believe the conclusion, because the way we’ve accepted it when we accepted the premise is by belief, not mere acceptance as fiction.
    Which is the right way to read the Q-argument (the one about Holmes)? I think both are acceptabe readings. In some contexts, my reading would have been meant, and in other contexts, your reading would have been meant. So we both have something to say that applies in some actual situations.
    But the Holmes argument is supposed to be analogous to the Murder argument. Does the Murder argument have two alternative readings as does the Holmes argument? Or does the Murder argument only have one reading, corresponding to just one of our alternative readings of the Holmes argument?
    What should the MFist say here?
    On a first run, it seems to me an MFist might go either way. On both your analyisis and mine, Eric’s puzzle is answered, though in different ways. So if the MFist wants to avoid saying that we “really” only “accept” murder is illegal in this context, then perhaps he should try to say the argument should be read your way. If the MFist wants to avoid saying that the premise of the argument has both a fictional and nonfictional part, then perhaps he should try to say the argument should be read my way. Or if the MFist doesn’t find either of these to be things he wants to avoid, then he might (among other possibilities) argue that in different contexts, the Murder argument should be read differently, such that the conclusion is sometimes to be “accepted” and other times to be “believed.”
    Well, that’s what I’ve got.
    -Kris
    (By the way, everyone, I know I’ve sort of just jumped in here without warning, and that none of you have any idea who I am. In some contexts, this would be considered a breach of some kind. If there is any reason that I should be posting less, or not at all, please do feel free to send me an email to that effect.)

  22. Hi, sorry to come to the party so late, I have had some pressing deadlines. A number of issues have been raised, let me briefly comment on some of them.
    Jussi complains that my use of ‘factualism’ is strange. I find the worry itself strange since I never used the word ‘factualism’. I do talk about ‘nonfactualism’, but that is an established term of art. I avoided using ‘factualism’ for the converse notion because it is potentially confusing. The labels don’t matter so much as the distinction they mark. The distinction was a semantic disitnction, between contents that purport to represent a subject matter and contents that do not so purport.
    Acceptance of a moral sentence, a sentence that expresses a moral proposition (i.e. a proposition whose truth or falsity depends on the obtaining of the relevant moral fact), needn’t be belief, or so I contended, but could rather be noncognitive. I also proposed a substantive thesis about these noncognitive attitudes—that they were a kind of affect, what Scanlon called desire in the directed atttention sense. Jussi worries whether these are really desires because they lack world-to-mind fit (don’t you mean mind-to-world?). I guess I am unmoved, since like Johnston and Scanlon, I regard philosopher’s talk of desires as a fantastic abstraction comprising a variety of different and differently related states. Scanlonian affects at least has many of the important properties that philosophers attribute to these desires and seemed to fit the bill for the kind of attitude that could constitute moral acceptance.
    I agree with Mark van Roojen, that Eric’s problem, if it is one, has nothing to do with the Frege-Geach problem. As long as moral sentences have representational contents, as long as they express moral propositions that are truth-evaluable, then there is no problem about logical relations among moral sentences. There *may* be a problem, but if it is, it is about moral *inference* and not about moral *entailment*. (Like MvR I am down with Harman about the logic and reasoning distinction.) Moreover, if there is a problem about inference, it is not due to the representational semantics I endorse otherwise it would be a problem for the realist as well. And I take it that the problem is meant to be peculiar to moral fictionalism.
    Again, like MvR my first reaction was to think of cases of reasoning involving metaphor. We reason with metaphorical premises all the time, and unproblematically reach warranted beliefs on this basis. If ordinary cases like MvR’s plumbing example works, then more would need to be said about what the source of the problem is for the fictionalist. BTW I think that cases of reasoning with metaphors are better cases to think about than reasoning with premises from an explicit fiction since as Kris Rhodes points out, there are potential complications. But since the analogy between moral fiction and explicit fictions like “Naming the Rose” is really abstract (all it really came down to was that acceptance in these domains needn’t be belief), I can’t see how MF is drawn into these complications. (OT, but we should really give Sherlock Holmes a rest lest people think we are illiterate nerds. Nerds, maybe, but illiterate? Perish the thought.)
    I think there is a real problem stating the alleged problem about reasoning with mixed premises explicitly. Eric seems to assume some principle like if you accept the premise and the premise entails the conclusion, you should accept the conclusion. Seems plausible at first, but of course it is false. Suppose you accept some claims and they entail some evident absurdity. If you have good reason to accept these claims, and are clueless about how to revise so as o avoid the problematic commitment, the rational thing to do is to continue to accept them and to take care not to infer everything they entail. Think of your initial reaction to the semantic antinomies, or to the sorities. If you work on these, you may come to revise in certain ways, but if you are an intelligent non-specialist and are unsure about how to revise, then you kept on believing the problematic claims and took care not to believe everything they entailed. This is just one problem; others could be generated. The main worry I have is that this style of objection must advert to principles of reasoning that too closely mirror logical relations in a way they could not so mirror—at least if Harman’s lesson of distinguishing logic and reasoning is right and properly digested.
    I haven’t thought through Eklund’s objection. It is a bit telegraphed. If I have anything useful to say about it, I will return with a report.

  23. Kris,
    I think I agree with what you said. I do think I want to hang on to my reading on the basis of Davidsonian principle of charity. We want people to have as many real true beliefs as possible.
    Mark,
    thanks for all the clarifications. I wasn’t complaining about your use of ‘factualism’. I was uneasy about you being labelled as a factualist in the original post. I first said that this cannot be right. Josh then argued that you are one given by what you mean by factualism. I then said that that would be an odd use of the term.
    I did mean world-to-mind direction of fit. This is the direction of fit desires and pro-attitudes have. Their aim is to fit the world to the mind. Truths have mind-to-world direction of fit. They aim to fit the mind to the world. I am usually confused about this but this time I did check this.
    I guess I am still troubled by this. There is both cognitivist and non-cognitivist understanding of Scanlonian seemings to be reasons. And, if you are an internalist cognitivist the fact that these seemings come with affect don’t decide the issue either. In order to come to one side on the question in moral psychology we do need something more precise.
    Given what Mark says, I am also starting to think why expressivists cannot use the same solution to the Frege-Geach problem. They could say that moral predicates are like the truth-predicate according to Horwich in that they refer to a logical properties. The sentences then have some truth-conditions – doesn’t even matter which. This enables us to account for the logical relations between the moral sentences and explain the entailments.
    But, they could say that these sentences are never used to refer to the property but rather to express different kinds of pro-attitudes. If the Frege-Geach problem is merely the problem of the entailments, it would not be a problem for this sort of neo-expressivist. I think this is the line Matthew Christman and Dorit Bar-On are actually taking.
    I do think that this is not quite right. The original Frege-Geach had to be a requirement to explain the inferences too in addition to entailments. This is what pushes Blackburn et al to the logic of attitudes.

  24. Mark is right that he does not the term “factualism” in his book; I used that term to characterize his opposition to nonfactualism. Perhaps there is a better term? Nonnonfactualism? Representationalism?
    One clarificatory point: I do not think or assume that if you accept the premise and the premise entails the conclusion, you should accept the conclusion. That would indeed be crazy.
    Now I do seem to have assumed that if you accept the premise, and you are entitled to do so, and the premise entails the conclusion, you should accept the conclusion. And the semantic antimonies and related paradoxes cast doubt on this assumption. But they do so because, in these cases, the conclusion of the argument conflicts with _other_ things we believe (and are entitled to believe). So these cases are tricky. But nothing like that is going on in cases of mixed arguments, so I don’t see how the existence of semantic antimonies is relevant to whether and why one is licensed to believe something whose content is logically implied by content one (merely) accepts.
    I think Jussi might be correct to say that “I don’t think there is a mystery here as long as we think that accepting a conjunction is not a distinct attitude but rather consists of accepting each conjunct. The fictionalist can then say that the acceptance of the conjuncts can consist of having different attitudes.” That provides a neat way of disentangling the conjunction.
    But this leads me to wonder whether there are _other_ cases that cannot be disentangled. It is often purported that thick ethical concepts cannot be entirely disentangled without remainder into a purely descriptive part and a purely normative part. And yet it seems we are entitled to infer both purely descriptive (but partial) and purely normative (but partial) conclusions from them. Consider:
    TP1. His behavior was rude.
    TCN. His behavior was prima facie wrong.
    TCD. He violated the norms of his society.
    TP1 is some reason for accepting (but not believing?) TCN, and some reason for accepting (in the form of believing?) TCD. And yet according to MF (as I understand Mark understands it), one does not believe TP1 when one accepts TP1. So it seems (to me) that Mark will say that one is (at least prima facie) entitled to believe something (TCD) on grounds (TP1) one is not entitled to believe.
    I hope I haven’t mischaracterized what his position is, or will be, so I of course welcome correction.

  25. That’s very interesting. I’m not sure that disentangling is required in this case for fictionalism about rudeness. The inferences in this case do not seem to be deductive in the same sense as earlier. I wonder if the fictionalist can still accept that thick moral properties are not disentangable. This would mean that can classify rude actions correctly only when one is immersed in the fiction shared by the linguistic community and one could not say reductively on which natural properties the given fictive thick property, rudeness, supervenes.
    The fictionalist could still explain what lies behind the inferences you give. About the first, she could say that the moral fiction comes as a package of different moral qualities that are inferentially interrelated. One of these relations in the fiction can be that the fictive rudeness property is projected over only actions on which we also project the fictive prima facie wrongness property.
    The second inference is slightly odd. I’m not sure all rude actions violate the norms of any society. To violate norms of society is probably rude as such. Of course, in many cases they do. But, maybe this relies on an inductive generalisation. We can think of particular cases of rude actions and the naturalist features on which we typically project rudeness. We find out that in most cases the same actions break social norms. On this basis we generalise and this generalisation grounds the inference from TP1 to TCD and not merely the concept of rudeness as such.

  26. Hi everyone!
    I just finished reading Mark´s illuminating book and have a few small questions.
    Let´s start with this one. Given Mark´s distinctions between use (acceptance) and content I am somewhat puzzled about how Stevenson´s emotivism fits in.
    If I remember Stevenson correctly then by his account when making a claim like “abortion is wrong” you report your own attitude towards it (as with cognitivist subjectivism) and then also issue an implicit demand “do so as well”.
    I guess this would mean that the content of a moral sentence according to Stevenson is a proposition that represents how things are, how your own attitudes are. When I utter “Abortion is wrong” then the sentence has something like “I dislike abortion” as content. The acceptance of that sentence seems to be belief. When I accept the sentence I believe that “I dislike abortion”. This would mean that according to Mark´s distinction Stevenson would be a full-blown cognitivist and it is hard to see how the non-cognitivist element can be fit in.
    Or did I misconstrue something?

  27. Indrek,
    If that’s what Stevenson means then yes, I think he is cognitivist. There is an interesting exchange with Moore in the Schilp volume on Moore where Moore suggests that Stevenson is not enough of a non-cognitivist because of just this feature.
    But, this may not be the only way to read him. Stevenson always offered what he called “patterns of analysis” if I recall the phrase correctly, and they did not seem to be exact translations of the target claims, insofar as he offered differing such patterns for the same targets. So I wonder if the relevant point of the ‘I dislike X, do so as well’ was just that such utterances had both an expressive and exhortative (horatory?) function. So maybe Stevenson just had no better way to capture the expressive bit in English.

  28. Indrek – let me echo Mark vR. I think Stevenson is pretty clear that “I disapprove of stealing; do so as well” is not in any way supposed to be equivalent to “stealing is wrong”; it is simply supposed to be as close to a paraphrase as you can get while also being enlightening. He doesn’t say that the first part of the paraphrase is a “report” of your disapproval; rather, he says that it is expressing your disapproval.
    Of course, that leaves open the possibility that Stevenson meant something more like reporting by “expressing” than like what contemporary expressivists mean by it. It could also be that his other main “pattern of analysis” is harder to fit into any simple classification of cognitivism vs. non-cognitivism, as well – see Ethics and Language. In general I agree with the drift of your comment, that it is hard to state a version of the cognitivism/non-cognitivism distinction which clearly gets all of the paradigm views to turn out to be on the right sides of the line.

  29. Indrek,
    Let’s forget, for a moment, what Stevenson actually meant. Let’s see how the view you describe might be classified as a species of noncognitivism.
    There’s no problem with there being a cognitive component, so long as it does not concern facts about morality. So far so good, since the cognitive component concerns the speaker’s attitudes and that’s a psychological and not a moral fact.
    The natural thought is that the normative component of a moral utterance, on the account you describe, is the prescriptive component, the “do so as well”. The idea, roughly, would be that moral utterances convey *endorsed* attitudes where the endorsement consists in the speaker’s readiness to prescribe the attitude (and perhaps be justified in so doing.) In a way, it would not be a million miles away from Gibbard’s idea that morality concerns the rationality of sentiment.
    So, moral acceptance, the attitudes involved in accepting a moral claim and expressed in making a sincere moral utterance, would not involve belief in a moral proposition and would involve a noncognitive attitude (the prescription). But that straightforwardly counts as noncognitivist at least on the understanding of the cognitive/noncognitive distinction described in *Moral Fictionalism*.

  30. Thank you for the replies and clarifications. I have a few more questions but will pose them when I´ll be back from a trip after a week.

Comments are closed.