Certain big philosophical terms seem fundamental, yet cause big problems. Existence, being, real, cause, freedom, good, and true are a few examples. These terms have acquired meanings, and then more meanings, and then yet more meanings, resulting in camps of philosophy who seem to say opposite things using the same words.

Are these disputes non-substantive? True, they often revolve around terminological disagreements, but they are not about terms, or at any rate we don’t want them to be. We want them to be about the things to which they refer: about existence, reality, causation, the good, and what grounds what. The disagreements begin to look terminological when the debaters realize that they are talking past each other, using those fundamental terms in different ways. But isn’t that a false impression, as long as each person is claiming truth for their position? Doing that is very substantive indeed; it’s not a mere acknowledgment that “We’re using the terms differently.”

Borrowing some concepts from Theodore Sider (all the references are from Writing the Book of the World), I want to sketch out a way of understanding what’s going on here.

One of the key concepts Sider has endorsed is “reference magnetism.” (He attributes the term to a 1984 paper by Harold Hodes, but it’s usually associated with David Lewis.) According to reference magnetism, we don’t simply assign words to things or concepts in such a way that our statements about them come out true. Truth on an interpretation isn’t enough. We also want the references of our words to have certain characteristics, certain external constraints on meaning. Here Sider’s preferred term is “joint-carving,” borrowed from Plato, by which he means “corresponding to actual ontological structure.” (I find the term disgusting, but it’s too central to Sider’s thought to be simply dropped.)

An example of joint-carving that Sider offers: Imagine two electrons, alike in every respect, plus a cow. We could find ways of grouping one of the electrons with the cow, forming the mereological item “electron-plus-cow,”and go on to say true things about it, and the remaining lone electron. Sider’s contention is that to do this is to carve reality very badly; it’s a “bizarre interpretation.” “The three objects should be divided into two groups, one containing the electrons, the other containing the cow. The electrons go together, and neither goes with the cow.” Sider also follows Lewis in calling this a “natural grouping,” which is problematic, but let it pass. It’s enough to understand the kind of thing that joint-carving is. (Though I can’t help pointing out that “natural” itself is one of those terms with more than one eligible reference magnet.)

So the idea is that we have a selection of words (exist, real, etc.) and a conceptual field which contains opportunities for those words to refer perspicuously – to carve at the joints. Sider is saying that the conceptual field has natural structural divisions, so when we try to match words with concepts we can be more or less perspicuous. A word like “exist” can be pulled toward one or more of these “reference magnets,” and made to refer to them. How does this happen? Through historical usage, primarily, which may evolve into ordinary language as well. So there may be a number of joint-carving candidates for what “exist” ought to refer to – but Sider’s point is that it’s a limited number, we can’t just say that “exist” will hereafter mean “go for a walk.” Indeed, for most non-philosophical terms (that are sufficiently precise), there will only be one reference, and it will be uncontroversial. Example: For a while, “leopard” was probably used to refer to several large cats. But now that we’ve agreed on what the leopard species is, the only reference magnet in the vicinity is Panthera pardus. Any other non-metaphorical use of the term is wrong.

The problem is, the “big” words are so encrusted with centuries of varying uses at the hands of varying philosophies, that they now get drawn to many different reference magnets. Sider (and I) would say that trying to argue for a single meaning for a word like “good” is a non-substantive debate. It really is a wrangle over terminology. But . . . the possible “magnets” are not themselves words, and the issue is not merely linguistic. It is as substantive as can be: ontology, what the world is like. Our problem is that we can’t settle on which of our big terms ought to be coupled with which magnet. This entire picture, as you can see, assumes that reference is to some degree external to theory, that we need to do more than make an internally consistent interpretation.

So this is very far from a skepticism or relativism about metaphysics. (It’s true that philosophers unsympathetic to metaphysics might stop right here and say, “See? There’s no ‛there’ there. If you don’t even know what the words mean . . .”) Rather, it’s an acknowledgment that we don’t know which (potentially meaningful) words belong with which concepts. We don’t know, for instance, whether we ought to let “exist” be drawn toward the concept of number, and say that numbers exist. We don’t know whether “cause” should be drawn toward the concept of having reasons, and say that a reason is a cause. Etc.

Sider suggests a different approach: “Sometimes fundamental metaphysics can be conducted in ordinary language. But not always. Metaphysicians need a plan B.” Rather than continuing to try to work with ordinary language, we can “enter the metaphysics room” and coin new, improved terms that we all agree to use . . . and then disagree about those. At least we’ll know what they refer to.

Sider imagines a natural-language expression E, and a reference magnet m. (You can also just think of a reference magnet as a plausible candidate meaning, if you prefer.) Should E be paired with m? How would we know if E meant m, and only m? This question is non-substantive, because “its answer would turn on linguistic usage, not reality’s structure”:

That is, the ordinary, natural language question, phrased in terms of the ordinary, natural language expression E, would be non-substantive. But we could discard E and enter the metaphysics room, so to speak. We could replace the ordinary expression E with an improved expression E* that we stipulate is to stand for the joint-carving meaning in the vicinity. The question we ask in the metaphysics room, cast in terms of E* rather than E, is substantive. Indeed, it is superior to the original question, for it concerns reality’s fundamental structure, rather than its merely conventional or projected aspects. This is plan B. — Sider, 74.

Now I want to depart from Sider on one point. (And I should emphasize that much of the above is my own interpretation of Sider.) I’m not convinced that “reality’s fundamental structure” is the best way to talk about what Sider wants to talk about. I don’t know how fundamental the various reference magnets may be, or whether it’s necessary to drag in “reality” (one of those very terms whose ambiguity causes so much trouble). This is a version of the same question raised about “natural” groupings. I certainly don’t know whether “naturalness” or “fundamentality” are properties we can treat the same way we treat things like “yellow” or “square”. I’d rather say that words map imperfectly onto concepts, and that the structure of concepts – their relations, groundings, logics – is something we can discern regardless of the words we use. Plan B is an attempt to help everyone concerned to find a way to stop disagreeing about words and get on with doing metaphysics.

How might we come up with E*? Sider uses the example of C, a relation that is in many ways like our ordinary-language use of “cause” but, as it happens, applies only at the subatomic level.

Now the ordinary English term ‛cause’ may well not mean C. For i) C fits terribly with ordinary usage of ‛cause’ [because we use ‛cause’ to talk about macro-level events as well] . . . and ii) ‛cause’ may well be a nontheoretical term in English. Rather than standing for C, ‛cause’ may instead stand for the non-joint-carving relation that best fits our usage of ‛cause’. A debate involving ‛cause’ would then not be substantive. But we could enter the metaphysics room, and coin a new term, ‛cause*’, for the joint-carving relation in the vicinity of causation. ‛Cause*’ will stand for C – fundamental causation, we might call it – and our new debate about causation* will be substantive. — Sider, 75-76.

In some ways, this approach is familiar, even truistic: Define your terms! And if that doesn’t work, coin a new one! But the way Sider lays it out, we have the advantage of a much clearer insight into why philosophers fall into non-substantive disputes.

Consider how a certain type of philosopher might respond to this proposal: “Sorry, can’t do it. You want me to agree that there is more than one of these so-called ‛reference magnets’ in the vicinity of my term ‛exist’. Well, yes, I agree that people have used ‛exist’ incorrectly; they haven’t spotted the correct definition of what it means to exist, whereas I, along with [fill in favorite famous philosopher], have. But there’s no need to enter your metaphysics room and come up with fancy terminology. Instead, I’ll keep working to convince you that my use of ‛exist’ has indeed trumped all the other reference magnets in the vicinity, just as ‛leopard’ did.” And so the terminological/historical bickering goes on . . .

Another type of philosopher might respond, “I’m wary about this division between word and concept – between a term and its ‛reference magnet’. Are we really able to perceive structure (‛joint-carving’ or not) apart from the words we use to describe it? Does this depend on a special sort of intuition, and/or a multiplication of entities? Surely our challenge, if we’re going to do metaphysics at all, is to use the words we have in order to create the most plausible, parsimonious, and complete account we can. The words are the structure.”

Two things should be said about this latter response. First, you don’t need special powers to see the structure of the way concepts relate. And while it may require words to do so, you can notice these relations while at the same time realizing that the words you’re using to think about them are imprecise and conventional. The map is not the territory. Second, “using the words we have” does work well in some areas of philosophy, but we all appreciate the power of logical languages that can remove vagueness and allow us to clearly see what we’re talking about. The question is whether this (broadly) analytic approach is the right one to use for metaphysics, not whether it can be used at all. That said, we should expect difficulties in trying to translate familiar terms into Ontologese.

Both of these responses seem to me to invite a retreat into non-substantive disputes. The first philosopher wants to prevail in a debate about what a word ought to mean, based (I presume) on a story about what it has often meant in the past, and the successes that this meaning engendered. Of course, this individual wouldn’t put it that way. They would say that the word does mean X, not that they think it ought to. So from this position, “real,” for instance, would be like “leopard” -- there’s only one reference magnet in the vicinity.

The second philosopher doesn’t see daylight between word and reference; for them, to discuss reference can only be a discussion about how to use words, not about independent concepts or structures. But, as Sider puts it, reference is explanatory: It’s supposed to do more than pair word A with object B and show us what true things we can now say; that would be a kind of theory-internal version of reference. Rather, “one can explain certain facts by citing what words refer to.” This is why we regard “‛theories’ based on bizarre classifications as being explanatorily useless.”

Sider adds a caveat:

Reference must have the right sort of basis in the fundamental if it’s to be explanatory. It’s highly unclear what exactly the “right sort” of basis is [my itals]. . . but it’s quite clear that a relation connecting us to bizarre semantic values would have the wrong sort of basis – for the same reason that arbitrary [my itals] correlations between the motions of the planets and the stock market have the wrong sort of basis. — Sider, 29.

This indicates that we still have a lot of work to do in cashing out “the right sort of basis.” The term “non-arbitrary” looks to be important. I’d be interested to know what others (who find Sider’s view appealing) might suggest here. I’m also interested in knowing whether the idea of reference magnetism sheds any light on what happens when “big” terms are employed in philosophy. I think it does. I think it explains something I’ve noticed for years about how certain words can be pulled by different meanings. But it’s a metaphor. We need more clarity about “natural,” “fundamental,” and “right sort” if we’re going to make it plausible.