Even Pyrrho believes in composite objects

Here’s an argument against hydrological scepticism. Given the way we’ve been using the word ‘water’ all this time, it’s going to refer to whatever is the wet stuff rivers and lakes are made of. Now we might have got our chemistry wrong, and actually the wet stuff isn’t H20 at all; it’s XYZ. In that case there wouldn’t be any H20, and considered as counterfactual that’d be a situation without any water. But considered as actual, if the wet stuff is XYZ then when we've been saying 'water' we've been talking about XYZ, so water is XYZ, so we still know there’s water. None of that’s very controversial. Even if you’re fan of (what I understand to be the views of) Pyrrho and think that while everyday knowledge is fine our claims to scientific knowledge are pushing it, you can still be sure there’s water.

Now think about a mereological nihilist who thinks composition’s not identity and there aren’t any composite objects. There are electrons and quarks, but no atoms, molecules, skyscrapers or galaxies. Here’s an argument that this combination of views doesn’t hang together well. Start from the position of someone who thinks composition isn’t identity, and it happens but contingently. Now consider the situation, which they think is possible, where composition hadn’t happened. That’d be a world with just simples and no composite objects, just like the mereological nihilist thinks. Of course the simples arranged peoplewise wouldn’t have noticed, because some simples arranged skyscraperwise look just like a skyscraper. What would the part-whole-talk have referred to? If there’s no eligible referent that matches use well enough, then it wouldn’t have referred to anything. Is there such a referent? What about plurality inclusion? It looks reasonably eligible to me. The simples arranged chairlegwise would be among the simples arranged chairwise. Plurality inclusion is a partial ordering like parthood’s meant to be. I’m not saying that it's a more eligible referent than the part-whole relation, but in the absence of the latter I think plurality inclusion is a pretty good candidate.

Plurality inclusion is what composition-as-identity fans think part-whole-talk is actually about. So my suggestion is that if you don’t think composition’s identity, then you should think that if composition didn’t happen then part-whole-talk would be about plurality-inclusion and identity. This is inconsistent with the combination of composition-isn’t-identity and mereological nihilism.

So if the argument works, we can be surer that composition happens than we can be that non-identity composition happens, just as we can be surer that there’s water than that there’s H2O. Also, we shouldn’t think both that water’s H2O and that there’s no water in the rivers. If thinking composition isn’t identity and it doesn’t happen is like that, then we shouldn’t think both those things. But some people do.

Now, this argument makes use of the Lewis-style metasemantics whereby reference is determined by use plus naturalness. It also supposes that our concept of composition is the kind of concept which, like that of water, doesn’t tell you all about what composition’s like. But if you don’t think composition’s identity, then you should think that. And even if you do, maybe you should still think that. Look at Peter van Inwagen’s despair at solving the general composition question. In spite of the theoretical background it relies on though, I think it’s an interesting argument, one I’ve not seen written down before and one I’d like to see mereological nihilists address.