More thoughts on analytic metaphysics

I've been rather hard on metaphysics in a couple of recent posts. I'm not trying to bash metaphysics just for the hell of it. I started out grad school doing metaphysics, still do some of it, and think there's a very important place for it in the heart of philosophy. Still, there are aspects of metaphysics that have long bugged me, and I've been trying to work out in my head what bugs me ever since I read Chalmers' awesome new paper, "Verbal Disputes." So, if you all don't mind, I'd like to continue working out some of those worries here. If I'm way off, by all means, please save me. I'd love to have my worries put to rest!

Anyway, in my earlier post on "counterfactual theories of causation", I argued that these theories seem to me to have the wrong direction of fit. A genuine metaphysical theory, I want to say -- a theory about what is really out there in the world (i.e. "carving the world at its joints") -- should begin with the world. So, for instance, when we look at the world, there are things about it in the realm of what we call "causes" that seem intrinsically interesting. As Hume pointed out, certain things in the world -- the things we call "causes" -- are constantly connected. That constant connection -- all F's being followed by Gs, no exceptions -- is clearly something interesting in the world itself (beyond our conceptualization of it). Are those constant connections mere regularities (as Hume thought), or is there some kind of necessitation between them (as Kant thought)? These seem to be genuine metaphysical questions prompted by observation of the world itself.

Now turn to counterfactual theories of causation. Here, the direction of fit seems exactly the opposite way around. Counterfactual theories begin with our concept of a cause to the world. To see how this is the case, consider how many counterfactual relationships there are in the world. There are, quite literally, an infinite number of counterfactual dependencies. How, then, do "counterfactual theories of causation" get off the ground? Answer: by showing that certain counterfactual dependencies fall under our concept of a "cause."  The counterfactual theorist notices that we classify certain things as "causes" (e.g. "Billy's rock caused the window to break"). They then note that certain counterfactual dependencies supervene on Billy's throw (but not, say, Suzy's), and so they try to figure out precisely what it is about Billy's rock that makes those properties fall under our concept of a "cause" (but not, say, Suzy's).

Notice how different this is than the earlier case (where, instead of starting with our "ordinary everyday concept", we began with observing interesting stuff in the world itself). Concepts are just that: concepts. Suppose I called this (and only this) computer a "schmauze." Then suppose I wanted to figure out what a schmauze is. Well, I notice that one and only one thing answers to my concept of a schmauze: this computer! Viola, I've done metaphysics. But of course I haven't. All I've done is show what falls under a concept. There isn't anything metaphysically interesting or special about schmauzes. There's only one schmauze in the world, and it's just this darn computer. Boring.

That, in a nuthsell, is the problem I'm having these days with analytic metaphysics. Real, genuine metaphysics, I want to say, starts with the world and tries to make clear sense of the really interesting stuff in it. Analytic metaphysics, in contrast, begins with concepts and then just classifies which things in the world answer to those concepts. It may well be true that our ordinary concept of a cause "picks out" a class of counterfactual dependencies. Okay, but what of it? There are, again, an infinite array of counterfactual dependencies in the world. Why think this one -- the one we denote with the word "cause" -- is any more special than the phenomenon (this computer) I denote with "schmauze"?  In principle, as far as I can tell, the counterfactual theorist can't give any good metaphysical answer. They can say, "This counterfactual dependency explains our causal judgments" -- but that, again, is just to refer back to our concept.  By a similar token, I could say, "Schauzes explain my schmauzal judgments." Yes, all fine and good -- but schmauzes aren't interesting. In order to explain why the things we call "causes" are metaphysically interesting, the counterfactual theorist must point to something else: namely, the metaphysically interesting fact in the world itself that there are constant conjunctions between events. Which is just to say: they must give up a counterfactual analysis in order to explain what's interesting and unique about causation.

I have the same problem with debates over "material constitution." When you look at the world, there's just stuff: quarks, bosons, hadrons, leptons, etc. Now, of course, that stuff is sometimes arranged in the form of what we call "tables" and "chairs" -- but what of it?  It's just our way of carving the world up. It's not that tables and chairs have persistence conditions. It's that there's stuff we sometimes call "tables" and "chairs." End of story. Why think there are special "constitution" relations or "problems with different modal properties", as opposed to just stuff that sometimes we call "tables" and other times we call "chairs"?

Finally, turn to analytic metaphysical analyses of beliefs, reference, and knowledge. Begin with, say, Kripke's Godel/Schmidt case. When you look at the world itself, here's what there is: (1) a word, "Godel", which (2) is causally related to some things (Godel) in certain ways and other things (Schmidt) in other ways, and which (3) the human brain can use to refer to Godel (if we like) or Schmidt (if we like). Once we do the science -- once we see how human brains use words, how those words causally relate to things in the world (e.g. to Godel, and Schmidt) -- then, or so it seems to me, we've done metaphysics. We've discovered what's in the world. But then suppose we ask which of those things fall under our concept of a "belief", "referent", or "knowledge." Classifying things may have pragmatic import -- there's a point to calling something knowledge (it's the norm of assertion!). But genuine metaphysical import? As though our concepts make objects? (A strange idea. I call this computer a "schmauze" and I've made something: a schmauze!). Wouldn't just talking about the world itself -- about how the human brain uses words, how those words causally relate to objects in the environment, etc. -- tell us everything we need to know *metaphysically* about the world?

I'm sure I'm being terribly unfair with all of this. If I am, I want to learn why. Please, teach me, convert me. Make me a true believer again!

Comments

[Rob Gressis]

Hi Marcus,

I can't convert you, because I know next to nothing about metaphysics, and haven't really thought about justifying my (at least somewhat) intuition-based approach to philosophy. I'd say you'd probably end up converting me, but I doubt that too, based on a particular (strange) meta-philosophical view I hold (in a nutshell, the view is: most philosophers I meet are better at arguing for philosophical claims than I am; therefore, if I always accepted claims on the basis of their arguments, I'd always be ping-ponging around from claim to claim; that's very unpleasant, so I'm just going to read everyone's arguments in a detached manner for a while until I figure out what to do about this).

That said, I'm confused by a pair of claims you make. On the one hand, you write about material constitution:

"When you look at the world, there's just stuff: quarks, bosons, hadrons, leptons, etc. Now, of course, that stuff is sometimes arranged in the form of what we call "tables" and "chairs" -- but what of it? It's just our way of carving the world up. It's not that tables and chairs have persistence conditions. It's that there's stuff we sometimes call "tables" and "chairs." End of story."

But on the other you write about cause:

"for instance, when we look at the world, there are things about it in the realm of what we call "causes" that seem intrinsically interesting. As Hume pointed out, certain things in the world -- the things we call "causes" -- are constantly connected. That constant connection -- all F's being followed by Gs, no exceptions -- is clearly something interesting in the world itself (beyond our conceptualization of it). Are those constant connections mere regularities (as Hume thought), or is there some kind of necessitation between them (as Kant thought)? These seem to be genuine metaphysical questions prompted by observation of the world itself."

I don't understand why the Marcus who makes the material constitution claim also makes the claim about causes. Surely if you look at the world, you don't see anything called "necessitation". All you see are things (or perhaps events) and then other things (or perhaps other events). Why would you think there is necessitation rather than regularity?

I have a feeling your answer has to do with the allegedly exceptionless nature of these regularities. But I'm not sure, so I won't put words into your mouth.

[Tuomas]

Marcus,
I understand your frustration, but it seems that your conception of "analytic metaphysics" is very narrow. You say that:

"Real, genuine metaphysics, I want to say, starts with the world and tries to make clear sense of the really interesting stuff in it. Analytic metaphysics, in contrast, begins with concepts and then just classifies which things in the world answer to those concepts."

This may indeed be the view of, say, Frank Jackson, who famously defends conceptual analysis, but this is by no means a generally accepted view in analytic metaphysics! There are plenty of metaphysicians who do more or less exactly what you suggest that "genuine metaphysics" should be about. To name a few recent examples, consider L.A. Paul's account of metaphysics as modeling (http://philpapers.org/rec/PAUMAM-2), Kit Fine's "edictic" approach (http://philosophy.fas.nyu.edu/docs/IO/1160/met.pdf), E.J. Lowe's full-blown Aristotelianism (http://philpapers.org/rec/LOWTRO), or indeed my own approach on these lines (http://philpapers.org/rec/TAHIDO). None of these give much credit to the idea that metaphysics is conceptual analysis.

[Marcus Arvan]

Hi Rob: Thanks for your comment. Yes, it's the exceptionless nature of the regularities which I think makes it a legitimate metaphysical question whether the regularities are all there is to causation or if there has to be some form of brute necessitation to account for the regularities. For my part, I'm on the side of the necessitarians, as I think the regularities are incomprehensible without positing some such relation. In any case, that's why I think it is a legitimate metaphysical debate, whereas counterfactual theories merely classify phenomena under a concept (which I consider to be philosophy of language more than metaphysics).

[Marcus Arvan]

Hi Tuomas: thanks for your comment, and for the very helpful links!

I agree that there are people out there who do the kind of metaphysics I consider "real metaphysics" as opposed to mere classification, which all I think conceptual analysis does. I guess I meant the term "analytic metaphysics" to refer to the latter only. Although this may not a standard way of using the term, I guess I want to say there are two types of "metaphysics":

Analytic metaphysics: which proceeds by way of conceptual analysis (this is what I object to).

World-based metaphysics: which simply aims to understand what phenomena there are in the world itself beyond our mere conceptualization of it (which I think is great).

I want to say: if people want to do the former, fine. But realize its philosophy of language more than anything else. There may be something funny in *language* about how we classify tables vs. lumps, but it is a mistake to think that it is a *metaphysically* interesting question. Metaphysically, there are just hadrons, bosons, leptons, and we just call some arrangements of them "tables" and "chairs." Not much metaphysically interesting there. Linguistically interesting? Perhaps -- but not metaphysically.

[Flaffer]

At the core of your issue seems to be a materialism somehow tied to science, no? What would make the concept of a 'lepton' more worthy of explanation then 'table'? Why is the distinction between 'table' and 'lumps of atoms' not talking about the world at the world-level, as it were? So you seem to bring into the debate a certain metaphysical view which justifies certain talk over certain other talk (or really, says one way of looking at the world is 'talk' and the other is about things-in-themselves, to borrow a Kantian way of talking.

Having said that, I think there is much confusion in some metaphysics and I find myself agreeing with some of the general points but for other reasons. But if one assumes that language reaches out to the world, looking at the world from within language does not seem so wrongheaded to me. And assuming that one way is somehow purely about things-in-themselves (distinguished by scientific empiricism?) while other ways are "merely language" seems to be an untenable distinction.

Or am I mischaracterizing your view here?

[Tuomas]

(I notice that some of the links I provided above broke down due to the brackets, but it should be easy to fix them.)

Marcus,
Indeed, I'm largely in agreement with you, although I do consider myself to be doing analytic metaphysics of the non-linguistic sort. Some (I wouldn't say *most* though) of the work done under the label of "analytic metaphysics" is much too linguistically oriented to be "genuine metaphysics", even though there may be a place for it. This is something that has been forcefully argued in the (neo-)Aristotelian camp in particular, but of course it's also a theme familiar from the recent metaontological debate.

I notice that Carrie Jenkins has a very recent related paper in Erkenntnis, 'Merely Verbal Disputed' out on this topic as well: http://link.springer.com/article/10.1007%2Fs10670-013-9443-6

If I'm allowed some further self-promotion, I have a paper forthcoming in Erkenntnis myself, in which I criticise Scott Soames's account of modality exactly because of its linguistic as opposed to metaphysical nature -- it struggles to explain metaphysical modality: http://philpapers.org/rec/TAHSDA

[T. Parent]

I would like to register, very respectfully, that some of Tuomas' recommendations could be contentious... I'm thinking of Thomas Hofweber's paper "Ambitious, yet Modest Metaphysics, where Lowe is the paragon of "immodest metaphysics," and Fine is mentioned as a case of "esoteric metaphysics" (along with Schaffer, an Aristotelian).

Less contentious examples of legit metaphysics, I would think, include Maudlin's _The Metaphysics in Physics_ and Ladyman et al. _Everything Must Go_. I'm not sure they are the clearest writers, but at least their metaphysics is grounded in the latest physics. Maudlin does an impressive job, especially, at rooting out the outmoded physics implicit in David Lewis' views.

I might also note that Frank Jackson's notion of conceptual analysis is not what you might expect. He thinks of Gettier and Chisholm as exemplars of the method, even though they are talking about knowledge rather than "the concept of knowledge." Jackson's rationale is that Gettier/Chisholm are interested in giving necessary and sufficient conditions, in the manner of an old school Oxford-style analysis.

Besides, there is the familiar point about disquotation. The concept of F applies to x iff x is F. If so, then to have a view about what is F is to be committed to a view about what falls under the concept of F (and vice-versa). So I am not sure whether it matters whether metaphysicians phrase things in terms of concepts or in terms of things.

Still, I share the worries about conceptual analysis--but more because it is usually practiced from the armchair. If x's being F is mostly an empirical matter, then why wouldn't x falling under the concept F be empirical too? (My concept externalism is creeping in here...sorry Marcus!)

[Marcus Arvan]

Hi Ted: Thanks for your comment. I think the disquotation move is made too casually, and that it *does* matter whether we talk about concepts or things. Let me explain.

I think the problem with disquotation in metaphysics is this. When you disquote, you *can't* assume that the thing you're disquoting is metaphysically *interesting*. For example, I call this computer a "schmauze." I then do conceptual analysis and arrive at the conclusion that all schauzes are this computer. Great -- I've done metaphysics. But, of course, the point is what I've done is completely trivial and boring.

I want to say the same thing goes for other disquotational approaches to metaphysics that don't start with the world *beyond* our concepts.

My example: the counterfactual theorist shows that our concept of a "cause" applies to such-and-such counterfactuals. Well, that's great -- but my point is that those counterfactuals aren't *intrinsically* interesting (they're just one counterfactual relation among zillions!). The only thing interesting about them is that they answer to our concept "cause."

Other things in the world -- constant regularities, and perhaps, necessitation -- *are* intrinsically interesting. They seem to be *unique* properties in the world unlike anything else (whereas, again, there are zillions of different counterfactual relationships).

This, in a nutshell, is why I think the disquotation misses the point of what metaphysics should aim to do. It *shouldn't* aim to just "disquote". It should aim to figure out which sorts of things in the world are unique and interesting beyond our language.

[Jonathan Livengood]

I've been thinking a bit about your earlier post and what it means to carve nature at its joints. I want to try out a somewhat sloppy pragmatist thesis (I can already see lots of problems, but I don't want to get hung up on them right now).

Thesis. A classification scheme carves nature at its joints just in case the scheme is useful for whatever purpose one had in introducing it.

Illustration. I would like to assign responsibility to agents for the various things that they do, which I think have various effects, some good and some bad. I also think (let us suppose, at least) that one cannot be responsible for anything that is not an actual effect of one's actions. In other words, I need to know under what conditions the action of an agent is an actual cause of an arbitrary outcome.

The problem is certainly one of classification. And in one sense, I am free to label things as I like. For example, I might say that an agent's action is an actual cause of an outcome just in case the agent whistled at 437 Hz three seconds before the outcome. Now I know exactly when something is an actual cause.

But presumably, my goal of assigning responsibility -- itself having some purpose that might be satisfied to various degrees -- would not be helped at all by the proposed scheme for labeling actual causes. Whereas, some other scheme would be objectively better suited to my purposes. Moreover, there will be empirical tests of the goodness of my classification scheme, specifiable in terms of the satisfaction of my goals, so long as the scheme is understood relative to my purposes in proposing it.

The reply, then, to the accusation that this or that pattern of counterfactuals -- taken as an analysis of actual causation -- is only interesting insofar as it answers to our concept of actual cause is that the pattern of counterfactuals is interesting insofar as it helps us to do whatever it is we are trying to do with a concept of actual causation. (Also, it will be quite possible for the result of a metaphysical inquiry in this sense to have non-trivial divergences from our actual concepts, since our actual concepts, understood here as classification schemes, may not serve our purposes as well as alternative classification schemes.)

[Tuomas]

Ted,
I'm quite aware of Hofweber's point of view, but I'd maintain that it's equally controversial -- as is the Ladyman et al approach. I've had a public chat about these matters with Hofweber: http://www.philostv.com/tuomas-tahko-and-thomas-hofweber/

One of my papers I linked to above directly engages with Hofweber's and Ladyman et al's critique of Lowe and Fine: http://philpapers.org/rec/TAHIDO

I do think that there's an important role for science in metaphysics, but the Ladyman et al approach, for instance, would pretty much reduce metaphysics to science, and that's just as bad as reducing it to conceptual analysis. There's a middle way: metaphysics may be guided by empirical and purely conceptual or linguistic considerations, but it shouldn't be reduced to them. I think L.A. Paul's work is a good example of this (and perhaps less controversial than Fine and Lowe). Unfortunately, I haven't read Maudlin's book (yet), so I'll refrain from commenting on that.

[Lee Walters]

Hi Marcus,

I completely agree with you, as I’m sure you’re being completely unfair with all of this too.

You say that counterfactual theories of causation start with our concept of a cause. But this just seems wrong. What theories of causation do, counterfactual or otherwise, is start with causation and seek to give an account of it. This has nothing to do with concepts and I don’t know why you would think otherwise. An account of constant conjunction does not start with the concept of constant conjunction, but rather with constant conjunction, and theories of causation are no different in this respect.

Now perhaps you think there is no non-gruesome relation of causation, that we have to take it as primitive or eliminate it, or it simply reflects certain perspectival facts. Well perhaps you’re right, but what is the argument for that? In any case, let’s say you’re right, well then you are giving a theory of causation just as others are, and we find primitivist, eliminativists, and perspectival positions on a range of issues in metaphysics.

The metaphysicians you criticise take seriously the idea that we are tracking genuine/interesting phenomena in the world such as causation, knowledge, goodness, beauty, material objects, shadows, etc. Perhaps that is wrong. But why do you think that? If we are correct that there is such a thing as causation it is legitimate to ask for an account of it, although perhaps no non-circular one can be given.

You also claim that constant conjunctions are obviously interesting. But I would have thought it very likely that these do not form a unified group. Some of these constant conjunctions will be mere coincidences, whereas others will reflect something deeper, something like, say, causal structure. We can and do distinguish between these two explanations and it is only in causation that we are interested in giving an account of. Moreover, if we can make sense of indeterministic causation, then what is of interest is not just a subset of the constant conjunctions, so to focus on them seems to be doubly missing the point.

We can also argue about how important causation is – does it feature in our scientific, folk, and philosophical theories. If the answer is no, like it is for schmauzes, then no one will be interested in it. But causation is here importantly different from schmauzes since it does feature in at least some of these theories.

The presupposition that you seem to be going in for is something like the following: we can reach out and *see* what the world is like independently our thought and language as if we had some “ontologiscope” which revealed the structure of reality: it really has leptons but doesn’t really have chairs. But we don’t have any such device. Rather there are certain truths about the world that we can ask about. Some of these may be mere perspectival truths whereas others may not. But this is what metaphysics is in the business of deciding.

[T. Parent]

Hi Marcus,
Yes, I agree that your schmauze example does not result in anything interesting. But for my part, that's because the theorizing is limited to apriori theorizing, rather than it being directed at a concept. After all, what falls under the concept "schmauze"--i.e., your computer--*is* quite an interesting thing, once we start investigating it empirically! You might object: But that wouldn't be metaphysical inquiry. My reply: Perhaps, but that just means there's nothing interesting for metaphysicians to do in this particular case. (Still, they may have things to contribute in other cases.)

Hi Tuomas,
Sorry--I didn't mean to insinuate that you were unaware of the controversy! I just wanted to note it for nonspecialists. And yes, Hofweber's view is contentious as well. Indeed, I'm teaching a grad seminar right now, and we're reading your defense of Aristotelian metaphysics later on in the semester. (It seems to be one of the few responses to Hofweber's paper that's currently available.)

[Marcus Arvan]

Lee: Thanks for your stimulating and challenging comment. I have to demur, though, with respect to your claim that theories of causation, counterfactual or otherwise, begin with causation itself not its concept.

I want to say there is an important difference with respect to how counterfactual theorists constructed their theories as compared to, say, regularity theorists. Regularity theories begin with perceptual observation: the closest thing we have to an "ontoloscope." They then proceed as follows: we perceive certain puzzling regularities in nature, and try to make sense of them. So the perception of the world comes first, the conceptualization and theorizing second. The direction of fit is world to theory.
Now, I suppose you will say (as your remarks imply) that all observation is theory-laden. And perhaps that's right. But I take it that we can also distinguish between degrees of theory-ladenness, in a way that enables us to more clearly distinguish the world itself from our conceptualization of it. So, for example, if we wanted to understand the metaphysical nature of this computer, we should aim to do so in terms of concepts that can be *operationally* defined that make empirical predictions (this, I think, is what the logical positivist a had right). This, as far as possible, enables us to test our concepts against the world itself. If, as in the case of "phlogiston", the predictions made by the concept don't pan out, we reject the concept as not mirroring nature.
Here, then, is what I want to say. Metaphysics should, for the very same reason, aim to penetrate *as* far as it can to the world itself, so that we can be confident that our metaphysical theories are really about *its* nature rather than merely how we choose to carve it up. I think regularity theories do this by "sticking" to perceptual observation. One literally does *see* constant conjunctions in nature that appear to be *unique*: every time there's an F, next there's a G. But this is not how counterfactual theories were generated (i.e. on the basis of direct perceptual observation). No, there were developed in response to "what we would say" about people throwing rocks at windows and such. It's this move -- the development of the theory on the basis of how we simply classify things in ordinary language -- that concerns me. I'm then further concerned by the fact that whereas fundamental regularities in nature seem metaphysically *special* -- parts of nature itself that call out for explanation -- the counterfactual relations focused on by counterfactual theorists seem to be "just one more counterfactual", one that cannot explain anything metaphysically special about causation above and beyond the fact that we call *those* counterfactuals causes.
Which brings me to your point about usefulness. I agree that our ordinary concept of a cause -- the one that counterfactual theorists begin with -- is useful. So too is our concept of money. But there's nothing metaphysically special about money. It's just stuff that we use for a certain purpose. So too, I want to say, is our ordinary macro concept of a cause in its "covering" certain counterfactuals. It's useful, but unless we want to be pragmatists about truth, it's important not to confuse the usefulness of a concept in ordinary speech with metaphysical truth.

Anyway, thanks again for your challenging comment. I think I see more clearly the case for claiming that I'm being unfair to counterfactual theories, but I'm not yet convinced. Maybe more discussion can convince me. ;)

[Douglas Kutach]

Marcus,
I was struggling for several years with how best to explain the worry you are having. My best attempt at a clarification and resolution is at http://sagaciousmatter.org/Kutach_Causation_Intro.pdf
and the supplementary material at http://sagaciousmatter.org/causation.html

[Marcus Arvan]

Hi Douglas: your work on this looks super interesting. Can't wait to take a closer look -- thanks for sharing!

[Eve]

Hi Everyone, Metaphysics for me is going beyond the boundaries of what is presented in order to find the what, the why and the how, this is the basis of science and philosophy.For me Metaphysics is not about trying to work out what we already have & know, it is about what we don't know. To me it is more than a debate needing to be scientifically proven, it is about the possibilities. It is about pushing the boundaries of the mind and seeing beyond. The shame of it all these days is that Metaphysics has become confused in new age & religious quarters to give gravity to the God question.