Let us suppose that the main conclusion of the previous chapters is correct: moral discourse consists largely of untrue assertions. Those arguments have primarily targeted deontological notions like obligation and prohibition. (p. 175)Such notions, Joyce believes, are connected to all normal moral discourse, so that we cannot just get rid of them and carry on with moral discourse as we know it. (This reminds me of the introduction to Tal Brewer's The Retrieval of Ethics, specifically the part where he says that the radical critique of modern moral philosophy presented by Anscombe and MacIntyre has been normalized so that virtue ethics is quite consistent with standard philosophical ideas of how to talk about morality.) "Moral discourse," Joyce writes on p. 177, "is a house of cards, and the card at center bottom has "categorical imperative" written on it." If we use Anscombe's preferred concepts, such as 'unjust,' 'unchaste,' etc., then this doesn't really help, Joyce says, because these concepts include or presuppose the idea that we ought not to be unjust or unchaste or whatever. And these 'oughts' are categorical (says Joyce). He mentions Anscombe's proposal that we jettison the language of categorical imperatives, but does not consider that to be the only option. So he explores fictionalism as another alternative.
What non-moral reason might there be to continue to engage in moral discourse even after we have seen through it? It must be useful. (This is starting to look like a utilitarian defense of talking and thinking like a Kantian. Didn't Hare try something vaguely similar?) The previous discussion of evolution suggests that moral discourse (as Joyce understands it) is useful, and so we might expect him to conclude, as he does, that we should keep using this kind of language.
We should, that is, use it, but not believe it. Indeed, for the most part we should live just as if we do believe it, perhaps even saying (and believing) at times that we believe it, but in reflective moments acknowledging that we know it isn't really right:
what a person believes cannot be simply read off her actions, speech and thought -- rather, the matter is determined by what she will say in a particular kind of context, of which "when doing philosophy" is offered as a familiar setting well towards one end of the continuum. (p. 192)Which starts to make moral fictionalism seem a little irrelevant. It's something to be ignored more or less all the time except when one is doing philosophy.
It is also important to note that Joyce's fictionalism involves thoughts, just not beliefs (where beliefs are understood as what you defend in a philosophy seminar, not something that informs one's normal behavior). If I engage in the fiction that I am a bear then I will say "I'm a bear" and make what I imagine to be bear noises, rather than "I'm pretending to be a bear. Imagine that I am making bear noises." I might well even think to myself (i.e., roughly, say to myself) that I'm a bear. If I really get into the role perhaps I will believe that I'm a bear. But I won't believe this upon reflection. Similarly, a moral fictionalist will think that slavery is wrong, say that slavery is wrong, act as if slavery is wrong, and possibly sometimes even believe that slavery is wrong. But she will not really believe it.
"Real belief" looks pretty unimportant, but I suppose it matters to philosophers. And one thing that matters in all this is that fictions have to be joint efforts if they are not to be lies. We aren't playing cops and robbers if I'm the only one who knows I'm not a real cop when I tell you you're under arrest. So "there can be no honest 'lone fictionalist'..." (p. 204). So I guess we need to get a group of people to pretend to take morality seriously without doing so immorally. Or something.