In this paper, Bennett distinguishes three different versions of the ‘dismissive’ attitude towards metaphysical questions, and asks whether any of them are appropriate in the case of the debates over composition and colocation. She (rightly, we thought) argues that we shouldn’t automatically put all metaphysical debates in the same category – dismissivism might be appropriate for some debates, but inappropriate for others.
The three kinds of dismissivism discussed are ‘anti-realism’, which claims that there is no fact of the matter about the answer to some metaphysical question; ‘semanticism’, which claims that some metaphysical question is ‘merely verbal’, and that the answer to it is analytic in our language (whichever language that is); and ‘epistemicism’, which claims that while a metaphysical question does have a non-analytic answer, we are not currently in a position to judge either way on it. Bennett goes on to argue that, for the debates she considers, semanticism is implausible and epistemicism is a live option. But because she doesn’t say much about anti-realism, the positive arguments for epistemicism seem pretty weak. (I wasn’t really convinced by the negative argument against semanticism either – it boils down to the claim that we can’t define things into existence, which will be denied by anyone with neo-Fregean sympathies.)
The argument for the disjunctive conclusion that either anti-realism or epistemicism is true about the debates considered goes via the claim that these debates are ‘difference-minimizing’. I wasn’t entirely sure what this was meant to mean – does whether a debate is difference-minimizing depend on the intrinsic properties of the issue being debated, or on the participants in the debate, or both? For the argument to lead to any substantive conclusions, I think it must be that the issue is intrinsically such that rational philosophers debating it will tend to difference-minimize – but Bennett on various occasion mentions philosophers (Burke, Rea, Cameron, Parsons) who don’t difference-minimize. Couldn’t this form the basis for a counter-argument? I suppose Bennett has to rely on the claim that these philosophers are just badly mistaken and have misjudged the intrinsic properties of the issue the debate is about. Either way, I thought the notion of ‘difference-minimizing’ was too vague and weak to have a strong metametaphysical conclusion founded on it.
Part of the argument that these debates are intrinsically difference-minimizing seems to be that structurally symmetrical problems arise for both sides of the debates. This feature of the dialectic, if genuine, does seem to be of real metametaphysical interest – someone who wanted to defend a form of structuralism about metaphysics might argue that the different sides agree on the structure of the correct view, which is all there really is to a view, so that they’re not really disagreeing at all (I take it this would amount to a form of anti-realism.) But it’s not clear how this feature gives us much motivation for epistemicism – if the debate really is symmetrical in nature, then the claim that there is an unknowable fact of the matter about which side is right seems dubious. Such a fact of the matter would be ‘metaphysically arbitrary’.
In any case, I wasn’t convinced that the debates are totally symmetric. Bennett argues by induction from 4 cases where a ‘twin’ argument can be given against one of the arguments used by one side, but that’s a pretty weak inductive base. Moreover, one of the examples looks flawed. Bennett argues that the ‘causal exclusion’ or ‘overdetermination’ argument used by the nihilist against the believer in composite objects has a twin argument which works against the nihilist – where a believer would say that a ball broke the window, even though the simples arranged ballwise were causally sufficient for the breaking, the nihilist must accept many pluralities of simples, all of which are causally sufficient to break the window. It doesn’t matter exactly which plurality we settle on. But this doesn’t look like a twin for the causal exclusion problem, it looks like a twin for the problem of the many.
Consider the following case – two simples travelling together jointly break a window. Neither of the simples by itself would have been sufficient for the breaking. The believer, who says that the pair which the simples composed was the object which broke the window, seems vulnerable to the causal exclusion argument; the simples were jointly sufficient, so why postulate the ball as a cause? (I’m assuming the simples aren’t many-one identical to the ball.) But the nihilist seems vulnerable to no analogous argument. There’s only one plurality of simples sufficient for the breaking – both of them. Thus, no causal overdetermination. And the reason there’s no argument against the nihilist here is just that, as I’ve set the case up, the problem of the many can’t get a grip. Hence my suggestion that while the nihilist does face an analogue of the problem of the many, he faces no analogue of the causal exclusion argument.