(Read Pt. 1 here.)
Daniel Nolan, Greg Restall and Caroline West have an interesting article called Moral Fictionalism versus the Rest available in its longer form here and in a shorter form in the Australasian Journal of Philosophy, 83 (2005), 307–330. (Thanks to Greg Restall for popping up in the comments last time to direct people to the longer version available online, after someone noted that they couldn't access the actual AJP article from where they are. For the sake of convenience, when I refer to page numbers in what follows, they'll be for that online version.
I was initially interested to see the article mostly because meta-ethics is a strong secondary interest for me at this point, and I was delighted to see some of my favorite figures in my primary area of interest (phil of logic) tackling it. In the last post, I started out by briefly going through other meta-ethical options, setting the whole thing up in terms of how these other options make sense of our intuition that it is somehow correct to say that "torturing children for fun is wrong" and somehow incorrect or mistaken to say that "playing Wii Golf is wrong."
I went on to say that the "distinction surely has to do with our moral intuitions, and for shaved apes like us to 'have moral intuitions' is surely ultimately for us to have a certain sort of neurological event, one resulting from genetics, environment or some eccentric combination of the two. If moral properties exist, and they are non-physical (as they certainly seem like they would be), no remotely plausible causal story relates them to humans having moral intuitions about them.
"At this point, of course, one may go in any number of directions. For example, one can try to make sense of moral properties so they aren't non-physical. Volumes can be (and have been) written about this proposal in all of its myriad flavors--synthetic reductionisms, sentimentality theories and so on--but for the moment I'll just say that, while I've argued for this kind of approach in the recent past, at the moment I find the prospects for this sort of project fairly bleak."
If you're interested in why I tend to find those prospects fairly bleak, see my subsequent post on why I think typical, initially plausible-sounding naturalistic stories about moral realism tend to collapse into relativism. Of course, some sort of frankly relativistic realism of one sort or another might well be the right option (keeping in mind that not all forms of relativism are as silly as the simple variety that introductory ethics instructors the world over spend the first day of class demolishing), but, for now, let's put that to one side.
If not non-naturalistic moral realism, and not naturalistic moral realism, then what? Well, there's always outright error theory, but if we're interested in capturing our intuition that it's somehow correct to say that "torturing children for fun is wrong" and somehow mistaken to say that "playing Wii Golf is wrong," then standard forms of error theory aren't real options.
An additional choice on the market today is Blackburn/Timmons/Gibbard-style "quasi-realism." I find it unattractive mostly because I'm a minimalist about truth, and, while every figure mentioned in the last sentence claims the same thing, I think that their view amounts to a theory of truth no less "substantive" than coherence, young-Wittgenstein-style correspondence and so on. (Note that this is not a criticism that Nolan & Co. make, although I think they should.) See below for an explanation of all that.
Meanwhile, at this point, it might start to look like all the possible options are wrong, and we're in serious trouble. Indeed, it's precisely because this is how things look that I find fictionalism so interesting. It lets have our "some moral statements seem correct and others seem incorrect" cake and not only eat it too but wash it down with both a shot of Respectably Naturalistic Picture Of The World and a Plausible View About Truth chaser. What's not to like?
Before going much further, it's worth fleshing out for a moment what "fictionalism" means. Any fictionalist view about moral statements is one according to which moral discourse is like fictional discourse. What semantic status an apparently correct moral statement like "torturing children for fun is wrong" has according to any version of fictionalism, then, is a matter of what status the fictionalist in question takes apparently correct fictional statements like "Sherlock Holmes is a detective, not a carpenter" to have. There are, at this point, at least two options (actually, there are definitely a lot more, but, for the time being, let's stick with just these two, since one of them is the Restall/Nolan/West-preferred and the other helps us bring out some interesting things about it):
(1) What we can call "truth-fictionalism," which stems from the view that the natural language sentence "Sherlock Holmes is a detective" is true, although its surface grammar has to be re-interpreted a bit. When we say "Sherlock Holmes is a detective, not a carpenter," we really mean "In fiction, Sherlock Holmes is a detective, not a carpenter." Similarly, with whatever the meta-ethical version of a fiction operator may be implicitly affixed the beginning of the sentence, "torturing children for fun is wrong" is true as well, despite the fact that we're neither postulating a really existing property of wrongness nor playing quasi-realist games with the notion of truth.
(2) What we can call "falsehood-fictionalism," which stems from the view that the natural language sentence "Sherlock Holmes is a detective" is just as false as "Sherlock Holmes is a carpenter," but, while discussing fiction, it's correct to assert the former and incorrect to assert the latter. If we take the surface grammar seriously, and we take reference failures to guarantee falsehood in the typical Russellian sort of way, the falsehood of any statement one makes about Sherlock Holmes one way or the other would seem to be straightforward enough. The appropriateness or inappropriateness of asserting any such (false) sentence, however, is a function not just of their truth or falsehood but of the purpose of fictional discourse, and similarly, even if "torturing children for fun is wrong" is, strictly speaking, just as false as "playing Wii Golf is wrong", it's still correct to assert the former and incorrect to assert the latter. This difference in correctness is not a matter of the way that the world is (since, on this view, the world lacks properties like "wrongness") but simply a matter of the function of moral language.
Before we go any further, it's worth addressing one bit of taxonomy--why not call (1) a version of moral realism and (2) a version of error theory? This seems to me to be largely a matter of decision and convenience. If one describes any view according to which moral statements are true as realist and any view according to which they are false as error-theoretic, we can talk about realist fictionalism and error-theoretic fictionalism, but I think that the shape of things is clearer if we reserve the word "realism" for views according to which the world contains really existing moral properties and the word "error theory" for views according to which there's no sense in which some moral statements are more correct than others.
A more interesting question, to my mind, is the one of what the difference is between (2) and quasi-realism. Both "falsehood-fictionalism" and quasi-realism agree that (a) the world isn't obliging enough to provide us with metaphysical facts to make true statements like "torturing children for fun is wrong", but (b) the nature of our moral discourse still makes some such statements (but not others) correctly assertible. Where they come apart is on the question of whether (c) is therefore true.
My gloss on this would be that the main difference is that falsehood-fictionalism is a view that you get when you start out from quasi-realist-like assumptions, but you take minimalism about truth seriously. (I'm certainly not claiming that you have to be a minimalist to be a (2)-style fictionalist. Indeed, Nolan/Restall/West, while advocating (2)-style fictionalism in the linked article, don't claim to be minimalists, and indeed seem quite prepared to cede the term to the quasi-realist. Rather, my claim is that minimalists who agree with the starting points of the quasi-realist story necessarily have to become (2)-style fictionalists in order to remain consistent.) If truth is nothing more than T-Schema instances--or, arguably equivalently, "'P' is true" never means anything above and beyond what P means--then, if I ascribe a non-existent property to something, my statement is straightforwardly false. (If, say, we call the fictional color in H.P. Lovecraft's classic horror story "The Colour Out Of Space" 'glack,' our statement "snow is glack" is obviously and straightforwardly false.) It doesn't matter how our discourse about that property works, what useful function might be served by that bit of our language, or what the conditions might or might not be for in any sense 'correctly' or 'appropriately' asserting that something has that property. None of those things can enter into our assessment of the truth or falsehood of the statement, because there's nothing more to truth than the instances of the T-Schema. The quasi-realist's move is, inevitably, to build things about assertion into their account of truth, and at this point, I can't for the life of me see how their view about truth is any less "robust" or "substantive" than, say, the picture-theory version of correspondence. The picture-theorist, after all, accepts all instances of the T-Schema, they just tell a substantive, definitional story about what all those instances have in common.
(Of course, if the quasi-realist wants to frankly admit that they have a substantive theory of truth and frankly argue for that theory and against any kind of minimalist or deflationary option, that's their right, and any such proposal needs to be carefully considered on its merits. What's annoying is that quasi-realists tend to wrap themselves up in the banner--and inherited intuitive appeal--of minimalism, when their implicit theory of truth is really anything but.)
(As a further sidenote to all this, depending on how a (2)-style fictionalist fills out her preferred story about moral language, (2)-style fictionalism might amount to a version of expressivism, if we use that term to refer not just to quasi-realism but to a whole family of related views, like Ayer's non-cognitivism and Hume's projectivism. Hume, remember, took moral statements to be false. When we paint the world in the colors of our moral reactions to it, we are misrepresenting it, not doing something non-descriptive.)
In any case, before saying anything more about (2)--of which the Nolan/West/Restall position is a variation--we should take a harder look at option (1), "truth-fictionalism." What's wrong with that?
One objection might be that, even if the "implicit fiction operator" story is a plausible reconstruction of what ordinary speakers are getting at when they make positive assertions about Sherlock Holmes--if you say "Sherlock Holmes lived on 221b Baker Street" and I offer to pour through the records of the place to prove that no such person ever occupied it, the obvious response would be "you know that's not what I mean"--it still isn't a particularly plausible reconstruction of what ordinary speakers mean when they talk about wrongness. I'm not sure how devastating this objection is, given cases of fictional discourse whose fictionalness isn't universally agreed on. For example, if Robin is an agnostic who doubts the historicity of the Exodus, and Jane is a devout Christian fundamentalist who takes the Bible as the inerrant Word of God, and one of them thinks that Aaron was Moses' brother and the other of us was sure that Aaron was Moses' son, Jane and Jill don't seem to talking past each other. The (1)-style fictionalist could perhaps make sense of this by saying that both our statements have an implicit "according to the Bible..." operator B(...), and our disagreement about whether B(P) universally entails P is beside the point.
On the other hand, the (2)-style fictionalist might claim to have a more straightforward account of this case...and thus, by analogy, a more straightforward account of what's going on when Mark the Robust Non-Naturalistic Realist and Ryan the Fictionalist disagree about whether abortion is wrong. The (2)-style fictionalist could gloss the Jane and Robin case by saying that the statements they are asserting directly contradict each other, neither being prefaced by any sort of implicit operator, and that their disagreement about whether whatever the correct answer is is assertible because, say, they know it to be true, or because, despite its falsity, the function of fictional discourse entitles us to assert it, is quite irrelevant to the case. Similarly, in the Mark and Ryan case, the claims "abortion is wrong" and "abortion is not wrong" directly contradict each other, and their disagreement about whether the correct answer is assertible because it captures the moral facts or whether it's a matter of the function of moral language is quite beside the point.
This leads directly into another advantage for (2)-style fictionalism over (1)-style fictionalism. "According to morality, abortion is wrong" and "according to morality, abortion is not wrong" don't contradict each other. (They might implicitly contradict each other if we make the background assumption that our moral fiction is internally consistent, but that needs to be separately argued for. Famously, ordinary fictions are often quite inconsistent. "In the Sherlock Holmes stories, Watson's war wound is on his right shoulder" and "in the Sherlock Holmes stories, Waston's war wound is not on his right shoulder" are both true. There are desperate ways of interpreting away such inconsistencies, but they fail when we come to deliberately inconsistent works of fiction like Graham Priest's playful short story "Sylvan's Box.") (1)-style fictionalism, in other words, faces a Frege-Geach problem about whether the logic of moral discourse is going to end up being revisionary. The (2)-style fictionalist, on the other hand, would seem to have no such problem. Even if one is a truth-preservationist about validity, since moral statements are just ordinary statements, some moral arguments are valid and others are invalid (although, of course, none are sound) and there will be nothing revisionary about any of this. (Similarly, the inference from "snow is glack" and "if snow is glack, grass is glack" to "grass is glack" is valid.) So far, so good.
Indeed, West, Nolan & Restall put quite a bit of emphasis on Frege-Geach, and the attractiveness of their avoidance of the problem. (See, for example, p. 25 of the linked article.) There are, however, some tricky issues about how thoroughly they've escaped Frege-Geach, and I'll end on a quick look at that point.
To start to get a handle on it, we can ask a basic question:
What's the value of moral discourse, given that it doesn't have the value of "getting at the truth"?
In the case of fictional discourse with the rules of assertion that go with it (according to the (2)-style fictionalist), the whole story might simply be that the pleasure of conversationally recreating the details of our favorite novels and short stories is an extension of the original pleasure of reading them, and that might be all there is to it. Moral discourse, however, seems to serve richer human purposes. It seems to be intimately linked to all sorts of things that definitely do exist. Even if there's no such thing as wrongness, and as such it's not true that "torturing children for fun is wrong", it's still true, in a typical discussion about the matter, regardless of their meta-ethical views, all participants prefer that no one tortures children for fun, are upset by the idea, disapprove of other people doing so and plan not to do so themselves. Presumably, even if we take truth out of the equation, the remaining purpose and importance of continuing to engage in moral argumentation has something to do with all of this.
In their discussion of the advantages of moral fictionalism over standard error-theoretic accounts, Nolan & Co. write:
"A fourth advantage of moral fictionalism over eliminitavism has to do with its capacity to salvage the important role moral discourse is widely thought to play in coordinating attitudes and regulating interpersonal conflict in cases where people disagree about what they are to do, especially where collective action is needed or the proposed actions of different people interfere with each other." (p. 21)
In filling in the details of how one sort of plausible fictionalist story about this might work, they go on to say that one "reply might be to connect non-moral preferences and what is true in fiction via internalist bridge laws (though care must be taken in stating these). If the fiction is set up in such a way that it is guaranteed that what is good-in-the-story that the people engaged in the story have certain non-cognitive attitudes towards it, then coming to realize that some course of action does have certain moral properties according to the story should prompt the realization that the action is one that the agent has certain attitudes towards.
"Or the fictionalist could tell an externalist story: it might be the case that, by and large, people contingently want to bring about situations which are true according to the fiction. According to the externalist, no mere cognitive belief alone will affect people's preferences, but that does not mean that people may not alter their preferences to reflect what is true in the fiction." (p. 22)
The last point, it seems to me, could be more happily paraphrased as something like "people may have an over-arching preference for their behavioral preferences to line up with what, in the fiction, is 'right.'" (Indeed, given Motivational Humeanism--a view, which, anecdotally, it seems to me is accepted by an awful lot of philosophers with realist views about morality--there's a certain sense in which the fictionalist is in no worse boat than a robust moral realist. Even if there are moral facts, and even if we have access to them, people will only be motivated by them to the extent that they happen to have a desire to be moral.) In any case, all of this talk of regulating preferences brings us back to Frege-Geach.
It seems obvious that it's at least possible to have inconsistent approvals, preferences or plans. I can prefer for two mutually inconsistent things to happen, I can approve of them both, and I can plan to do both. My preferences and approvals will never be fully satisfied and my plans will never be fully carried out, but this can be true without inconsistency rearing its ugly head. (For example, I might plan to always carry out every promise I ever make to anyone about anything, but be psychologically incapable of making good on this.) Conversely, I might not have any preference, approval or plan one way or the other about some issue.
It might seem, though, that part of the value of moral discourse is that it helps reduce both the inconsistency and the incompleteness of my attitudes. Therein, indeed, might lie much of the usefulness of careful, rational discussion of moral issues. So far, so good.
But, given the first option sketched above (internalist bridge laws), it seems like the incompleteness and inconsistency of my plans will tend to infiltrate into the moral fiction and this useful purpose will be undermined. It could be that, even in idealized circumstances, I would still equally prefer for P and ~P to come true, or have no preferences one way or the other. If the moral story we've all been telling to each other seems humans started to develop ideas about morality is "designed" so as to line up with our preferences or plans (or even hypothetical preferences or plans formed under idealized circumstances), then won't it reflect the inconsistency and incompleteness of our (actual or hypothetical) plans and preferences?
OK, but what about the externalist option? Going with that, it could be that the fictional world of rights, duties and the rest described by our moral discourse is both complete and consistent, and that "by and large" people happen to have the happy trait of over-archingly preferring to line up their behavioral preferences with the imaginary properties that exist in that world. Outstanding! Even here, though, problems of consistency and completeness arise. Without infallibly internalist bridge laws, the fictional world of moral obligations might be as complete and consistent as external reality--there might be a "Moral Law of the Excluded Middle" and a "Moral Law of Non-Contradiction"--but why suppose that things have worked out that way? If morality is a human construction--like the world of the Sherlock Holmes stories, it's something we made up--why should it be any more complete or consistent that the Holmes stories are?
Take a typical case of reflective equilibrium, where a conflict between immediate moral intuitions is used to help us to regiment them into a consistent system, in precisely the sort of way that's important to the function of moral discourse described above (to help regulate preferences and coordinate collective action): the argument against consequentialism based on the second version of the Trolley Problem. We start by pointing out that in all morally relevant ways, the switch-pushing case and the fat-man-pushing case seem to be equivalent, and thus reason that "if it's OK to push the switch, it's OK to push the fat man." We then assert that it's not OK to push the fat man, and conclude that it's not OK to push the lever. So far, so good, and all precisely the sort of thing the fictionalist is supposed to be able to make sense of, despite the fact that they take all of the statements with the moral predicate "OK" in them to be, strictly speaking, false.
Why, however, should we still take this to be a valid argument, given that the moral fiction, being a human construction, might well contain inconsistencies? Surely, what it's correct to assert about rightness and wrongness is a function of which things are right and wrong according to the story, and, in the story, (a) if it's OK to push the switch, it's OK to push the fat man" and (b) "it's not OK to push the fat man" could both be true in the story without (c) "it's not OK to push the switch" being true in the story. After all, (d) it's OK to push the fat man" might, in principle, also be true in the world of the story if we have no guarantee that it's an entirely internally consistent story.
Of course, the fictionalist could stick to their guns and say that validity isn't a matter of correct-assertibility-preservation, and certainly not of truth-according-to-some-fictional-story-preservation, but a matter of truth-preservation in the actual case. This is fair enough, but I think that at the end of the day, the fictionalist is faced with an unpleasant choice:
(1) They could accept a correct-assertibility-preservation account of logical validity at the price of accepting revisionary conclusions about the logic of moral discourse, which is to say, re-introducing the Frege-Geach Problem, the avoidance of which was supposed to be a major selling point of the account*,
(2) They could accept a truth-preservation account of logical validity at the price of destroying the value of moral argumentation. That is to say, they could continue to say that some logical arguments are valid and others are invalid and that the rules for deciding which are which are the same as they are for arguments about anything else, at the expense of logical validity being an important virtue for moral arguments. If, after all, moral statements are always false (so the point of moral arguments can't be to convince us that the conclusions of those arguments are true), and logically valid arguments can take us from moral premises which it is correct to assert to moral conclusions that it is not correct to assert, then why should moral fictionalists care whether or not some argument is valid?
*Restall, as a logical pluralist, might have the resources to accept this with a shrug--perhaps the logical consequence relation appropriate for moral argumentation is inappropriate for other contexts--but Nolan, who has argued extensively (in his work on counterpossible conditionals) in defense of a monist view about logic, should presumably be more troubled. I'm not sure what West's position is. In any case, in the article, they jointly present fictionalism's alleged avoidance of Frege-Geach-style problems as a major selling point.