Philosophy likes to see itself as on the hunt for eternal truths, from moral facts to the deep structure of reality. But this story mistakes what metaphysical language does, argues Amie Thomasson, whose recent book Rethinking Metaphysics has caused a stir in this most ancient of disciplines. Recent work in linguistics—largely overlooked by philosophers—shows that classic metaphysical concepts arise not to describe reality but as tools for getting things done. Metaphysics isn’t about describing reality, but rather redesigning and engineering language to sharpen its tools, to enable us to do new, better things with it.

Metaphysics has been going on for a long time. It’s high time we rethink it. Those who do metaphysics ask questions about what exists, and about the natures of things, their “essences,” and whether they are fundamental or “grounded” in other things.

While metaphysics reaches back in the Western tradition at least to ancient Greece, you might understandably ask: These days, aren’t those questions addressed by the sciences? We look to biology to determine what kinds of life there are, to physics to find out if there are supernovas, to chemistry to find out what the fundamental elements are, etc.

A common reply is to say: Sure, science answers some of these questions, and where it does, metaphysics should work hand in hand with the sciences. But metaphysics (they say) is broader, concerned with “everything whatsoever.” So, metaphysics also asks questions like: are there (or are there really) numbers, properties, moral obligations, possibilities or even possible worlds? Are moral properties grounded in natural properties? And so on.

___

I begin from the simple point that we shouldn’t just assume that all parts of our language function in the same way.

___

In Rethinking Metaphysics, I ask: What if some of these words (“number,” “property,” “possibility,” “moral obligation”…) aren’t even trying to pick out some parts of the world that we can just look for, identify, and study as we can study plants, animals, or supernovas? What if these words play a very different role in our language? Assuming that they just “pick out” (or perhaps try and fail to pick out) features of our world, which we can go on to investigate, relies on assuming that all these nouns work in the same way: that they aim to refer to and describe some features of the world. This is to make what I call the “functional monist” assumption. All the recent major approaches to metaphysics make this assumption, and it leads them astray.

Metaphysicians often reply: “Functional monism is a thesis about language. But I’m not interested in words, I’m interested in the world. So, I don’t care about how our terms such as ‘possibility’ or ‘redness’ or ‘the number seven’ or ‘moral obligation’ work. I’m interested in whether there are such things, and if so, what their natures or essences are, whether they are fundamental, or what they are grounded in.”

But if these terms enter language in a different way, and function differently than our terms for observable, concrete parts of our environment—if they aren’t even attempting to pick out features of the environment—then asking questions like these may be a mistake. In Rethinking Metaphysics I begin from the simple point that we shouldn’t just assume that all parts of our language function in the same way. Before we launch into “metaphysical investigations,” we should first ask what functions these terms serve, and how they enter language.

But how can we address questions about the functions of various parts of language—including terms that have long been central in metaphysics? Here, I suggest that we look to scientific work on language itself. Work in systemic functional linguistics has long addressed questions about the functions language serves in human life. This work in linguistics has been largely unknown in philosophy, and I aim to show its relevance and significance.

___

Languages permit grammatical shifts for good reasons: not to “track” new kinds of discovered “things,” but rather to enable us to do important new things with language.

___

Some of the results from systemic functional linguistics give us reason to think that terms like “possible,” “necessary,” “might” and “must” serve primarily interpersonal functions—serving to regulate behavior, express attitudes, etc. Such terms of course figure prominently in talk about what’s necessary or possible, and modal verbs such as “may,” “ought,” and “must” are central parts of moral discourse. If that’s how these terms enter language, then asking questions about what sorts of things our modal or moral talk refer to (possible worlds? non-natural moral properties? natural features of the world?) is barking up the wrong tree.

Systemic functional linguistics also sheds light on what it calls “grammatical metaphors”—terms that are introduced via a shift in grammatical category from other terms. For example, we can move from saying the tomato is red to saying that the tomato has “the property of redness”—introducing a new noun term, redness. Or we can shift from saying there are two forks on the table to saying “the number of forks on the table is two.” Such terms for numbers and properties are not introduced by any additional observations or worldly evidence—but just by means of these permitted grammatical shifts. And languages permit these grammatical shifts for good functional reasons. For, as linguist Michael Halliday shows, adding these terms enables us to do important new things with language, such as forming the kinds of generalizations we need in order to develop generalized scientific theories and the legal and bureaucratic systems that organize modern governments and social life—and that make the rule of law possible.

So, there is no need to “justify” introducing talk of numbers or properties, by arguing that saying that there “really are” such things adds “explanatory power” to our theories. And statements about numbers and properties don’t require that we posit “truthmakers” to explain what makes them true. Instead, we can make use of these grammatical shifts to introduce talk of numbers and properties—and these entitle us to say that there are numbers and properties. And we can see why languages permit these grammatical shifts for good reasons: not to “track” new kinds of discovered “things,” but rather to enable us to do important new things with language.

If we can no longer think of metaphysics as addressing deep worldly questions about existence, essences, or grounding, does that mean we should just give it up? In the book, I suggest a different approach: that we should instead rethink it, considering it instead as properly engaged in conceptual engineering. On the view I develop, conceptual engineering includes two parts. It begins from a reverse-engineering phase of figuring out what functions various parts of language serve, and how they serve them. Doing this work can help overturn the functional monist mistake, and help us determine what sorts of metaphysical questions are and are not appropriate. This can lead us to diagnose many classic “metaphysical problems” as either easily answerable, or as pseudo-problems that arise from a mistaken presupposition. A nice feature of this way of debunking some metaphysical problems is that it is not subject to the metaphysician’s usual line that any critiques of metaphysics must be giving us “just more metaphysics.” For the approach to debunking these problems does not rely on metaphysical theses at all, but rather on an analysis of the functions of the language involved in stating the problems.

SUGGESTED VIEWING The return of metaphysics With Hilary Lawson, Galen Strawson, Patricia Churchland, Maria Balaska

The second part of conceptual engineering involves making decisions, based on that reverse-engineering knowledge, about whether we should make changes to various parts of our language. In some cases, once we uncover the functions that parts of our language (such as our talk of race) serve, we might deny that we should serve those functions. In other cases, we may find that some useful terms (say, for “privacy,” “intelligence,” etc.) no longer serve well in the current technological, natural, or social context.

A great deal of work in metaphysics—especially classic work in metaphysics before 1950—has always been directed at addressing questions about how our conceptual scheme works, or how it should work. Debates about free will, the mind, art, and many more can be easily seen as addressing questions about how our conceptual scheme does work or how it should work (what criteria should we adopt for when someone acts freely? How can we revise traditional mimetic understandings of art to allow for expressive and abstract forms of art?). Addressing such conceptual and linguistic questions is a good part of what work in philosophy—including metaphysics—has always done.

It is also work that is centrally important to human life, and that will never be finished. For we will always need to reevaluate and reshape our conceptual and linguistic scheme as the environmental, social, and technological context changes, and as our purposes change.

Rethinking metaphysics as conceptual engineering avoids making it look like an outmoded rival to the natural sciences. And it does not make metaphysics trivial or uninteresting. For this kind of work in conceptual engineering has great worldly importance—importance to how we treat each other, how we organize our societies, and how we conduct our scientific investigations. On this view, the work of metaphysics is not about discovering some set of eternal truths. It is a never-ending and central human project, of figuring out how we should think and talk—and that matters deeply to how we live.