The Truth About Billy and Suzy?

I'd like to follow up on the comment I made in response to Moti's recent post on Peter Hacker's piece on the nature philosophy. As I noted in my comment, I've long worried that traditional philosophical methods -- principally, developing philosophical theories about causation, justice, etc., on the basis of our "considered judgments" about what counts as a cause, justice, etc. -- tell us more about our concepts (i.e. how we conceptualize the world) than the world itself.

Now, there are at least two obvious ways to respond this sort of worry. First, one can say that it is of great philosophical interest how we conceptualize the world -- that it is in fact of great importance what we classify as a cause, or justice, etc. Second, one can say that investigations of our concepts don't just tell us about our concepts, but about the world itself. Our concepts, after all -- well, many of them, at least -- "reach out" to the world: they refer to entities and properties within it.

To illustrate: I was having a conversation a while back about these issues with a friend who does analytical metaphysics, her primary focus being causation. I suggested to her that if we base theories of causation on what we judge to be "causes", we're really just developing theories of what we call "causes." She gave something like a standard deflationary reply. Because "X is a cause" is true iff X is a cause, she said, our judgments about the truth of various statements about causation tell us about causation. And, of course, one needn't be a deflationist about truth to make a similar move. One could just as well be a correspondence theorist and hold that statements about "causes" are true because they correspond to things in the world: namely, causes.

In my comment on Moti's post, I mentioned a new paper by David Chalmers, "Verbal Disputes", which just appeared in Phil Review (the link is to a free PDF copy). As I mention in my comment, I think Chalmers' paper represents an important step forward for philosophy, and I'd like to explain why I think this by applying its most central methodological idea to the case I've been discussing: causation.

Chalmers argues that to determine whether a philosophical dispute is "merely verbal" -- by which I take him to mean roughly, a dispute about our concepts instead of the world -- we should temporarily replace the concept in question with a description of what the concept is being used to refer to. Let's apply this idea to counterfactual analyses of causation.

People who work on counterfactual theories of causation have spent a great deal of time discussing various cases that seem to "pose problems" for such theories. One very well-known case is that of Billy and Suzy, two people who both throw stones at a window. As it turns out, the window's breaking counterfactually depends on both throws. If Billy hadn't thrown his stone, Suzy's would have broken it, and if Suzy hadn't thrown her stone, Billy's would have broken it. But, in the actual world where they both throw their stones, only one of the stones breaks the window, preempting the other one. Finally, there are different variants of the case: "early preemption", "late preemption", etc. People who work on counterfactual theories of causation suppose that if they can come up with a counterfactual analysis that correctly classifies all of these cases -- all of the cases we judge to be, and not be, causes -- then they will have given a true theory of causation.

Now let's use Chalmers' method for determining whether counterfactual analyses are "merely verbal" (telling us more about how we use the word "cause" than the world itself). Since counterfactual theorists want to understand causation in terms of counterfactual dependence, let's replace the word "causation" in the cases of concern to counterfactual theorists with "counterfactual dependence." So, here's the story: Billy and Suzy both throw stones at a window. The window breaks. The question then is: upon which stone throw does the window's breaking counterfactually depend?  The answer, obviously, is: both of them.

Now, of course, counterfactual theorists of causation want to say that only one of the throws genuinely causes the window to break (the other throw is preempted). But, here's the thing, we've just seen that this cannot be accounted for merely in terms of counterfactual dependence. The idea of counterfactual dependence -- and facts about counterfactual dependence -- by themselves give us no reason to think that one stone caused the window to break over the other. If we want to say that one chain of counterfactual dependence is "causation" whereas another one is not -- and of course this is what we want to say (it is what counterfactual theorists take as their starting point for debate) -- we cannot explain our wanting to do this merely in terms of counterfactuals.  For again, the window's breaking counterfactually depends on both stones.

Here's the rub: the counterfactual theorist has no metaphysical explanation of why we call one counterfactual chain the cause of the window's breaking over the other. They can clearly provide practical reasons for our calling one chain a "cause" over the other -- by, for example, saying that we are concerned with some counterfactual chains over others. However, they cannot provide a genuine metaphysical reason for calling one chain a "cause" over the other. For again, in their metaphysics, there is nothing in the world that corresponds to our concept of a "cause" above and beyond counterfactual dependencies, and the window's breaking counterfactually depends on both stones.

There are, it seems to me, only two genuine metaphysical candidates capable of explaining why we call one stone -- one chain of counterfactual dependencies -- the cause of the window's breaking but not the other.  We judge the one throw to cause the window to break because it is the particular chain of counterfactual dependencies because the throw:

1. Comprises a chain of constantly connected events leading to the window's breaking (i.e. Humean "regularity theory" of causation), that may also
2. Comprises a chain of events that necessitate the window's breaking (i.e. a Kantian necessitation theory of causation).

There's a simpler way to put all of this. If Chalmers' method is correct (and if I have applied it correctly), counterfactual "theories of causation" can't be genuine metaphysical theories of causation.  They at most classify certain counterfactual chains as "causes" but not others. They fail to provide any metaphysical explanation of why it is correct to call those things -- i.e. one type of causal chain but not others -- "causes." In other words, counterfactual theories provide no reason at all to think that our concept of a cause carves nature at its real metaphysical joints. In order to provide a genuine metaphysical theory of causation, one must appeal to something beyond counterfactuals to explain why one type of counterfactual chain are "causes" but others aren't.  And for that one needs to appeal to either a Humean regularity theory, a Kantian necessitation theory, or something else entirely.

Comments

[T. Parent]

Hi Marcus,
Thanks for the post, and the rec for the Chalmers paper. (I hadn't read that one, and I enjoyed it very much.) This is a topic I care deeply about, so I'm glad to have an opportunity to chat with you about it.

My principle response is that your remarks seem to concern two different topics. One is the merely verbal disputes that occur in metaphysics. The other is the nature of metaphysics, generally speaking. More precisely, the second topic is whether metaphysicians are investigating the world, or just the concepts they use to represent the world.

It seems these two can come apart. Indeed, most everyone will agree that some disputes in metaphysics are merely verbal. E.g., the Humean compatibilist may be just using 'free will' in a different sense than the libertarian. Nonetheless, even if the dispute is merely verbal, this would leave open whether these metaphysicians are really investigating a concept or the world. Thus, even if the dispute is merely verbal, one could still say that they really are inquiring into worldly phenomena. It's just that they are concerned with different worldly phenomena (perhaps without realizing it). Indeed, the different worldly concerns is what might explain the differences in their use of 'free will'. But despite being aimed at the world, the dispute might remain purely verbal. For the dispute may well evaporate once they realize that they use 'free will' to pick out different worldly phenomena.
Anyway, my two cents for what it's worth.

[Marcus Arvan]

Hi Ted: I'm glad you enjoyed the post, as well as the Chalmers paper -- and I'm glad to have the opportunity to discuss it with you, too.

I agree with you that I'm discussing two issues -- but I wonder just how distinct they are (i.e. how much they can come apart). Here's a thought: a dispute is "merely verbal" iff it doesn't carve nature at its joints.

Consider the case I discuss: causation. On the analysis I gave, counterfactual theories of causation are merely accounts of what we *call* "causes". Although they concern features of the world -- namely, counterfactual dependencies within it -- they are merely verbal in that they merely classify the things in the world we call "causes" without providing any metaphysical account of whether/how those things actually carve nature at the joints. In contrast, necessitarian and Humean theories are not merely verbal because they can both make a genuine claim to doing the latter (carve nature at its joints).

The same is true, I want to say, of free will. The compatibilist account is merely verbal because compatibilism merely classifies the things we call "free will" in ordinary life -- without giving anything resembling a compelling account of how our practice of calling such things "free" or "not free" carves nature at its joints...whereas deeper *metaphysical* theories of free will do.

In other words, I kind of want to say: a theory or dispute is merely verbal just in case the theory/dispute concerns how we use a given term (e.g. "causation", "free will", etc.) without providing any deep theory of whether/how the given use of the term actually carves nature at its joints. I may be wrong about this, but if this is correct, it collapses the two issues into one, no?

[Jonathan Livengood]

First, a small complaint. In Billy and Suzy cases of pre-emption, you need to decide whether the events are coarse-grained or fine-grained. If events are fine-grained, then there is no problem in pre-emption cases, you just index the events by time. (If the first rock had not been thrown, the event of the window breaking at the time it did would not have occurred. Rather, the window would have broken at some later time.) If events are coarse-grained, then it is not the case that the breaking counterfactually depends on both throws. Rather, the breaking counterfactually depends on neither throw. (And that's the usual treatment of the case.)

Second, a question: What exactly are you asking for when you ask for an account that carves nature at the joints? It is of course a familiar and evocative phrase, but what is the demand here?

[Marcus Arvan]

Hi Jonathan: Thanks for your comment. Here are some thoughts in reply to both of your main points. First, thanks for clearing up the preemption stuff. I wasn't meaning to suggest that counterfactual theorists don't have any good answers to preemption cases (as I understand it they've now moved on to overdetermination issues); I was just trying to use the case to illustrate.

As to your (very good, and difficult) second question, I’ve been toiling with different answers to it on and off all day. For a while, I didn’t think I could come up with a good answer. However, now I think I may have the beginnings of a good one. Although I can't pretend to give anything like an iron-clad account of "nature-carving", allow me to offer up a first-gloss analysis for discussion. If it’s completely off, I’ll be happy to see why it is – and, in any case, I’m happy that you asked the question. It has, at the very least, helped me to clarify my own thinking.

I’d like to begin with an observation on how counterfactual theories of causation are constructed. Let’s look at the world. There are, obviously, countless types of counterfactual dependencies in the world. So, let’s say there are type-A dependencies, type-B dependencies, type-C dependencies, and so on. Next, let’s suppose some counterfactual theory successfully picked out the specific type of dependency – say, type-C dependencies – as corresponding with our concept of a cause (such that all and only type-C dependencies are the type of thing that we judge to be “causes”).

Now let’s think about how this theory was constructed. There *are* a number of dependencies in the world. The counterfactual theorist then asked, “What type of counterfactual dependencies answer to our concept of a “cause”?” See what’s going on here? The counterfactual theorist *begins* with our concept of a “cause” and then simply aims to describe the things in the world out there that answer to it. The counterfactual theorist cannot point at anything in the world to *explain* why we should have conceptualized the world that way. They cannot point to anything objectively *special* about type-C dependencies to explain why should have conceptualized them as “causes” (for – absent something additional, such as necessitation or regularity [see below]) there is nothing objectively special about type-C dependencies; they are “just another” type of dependency in the world among many). The order of explanation is precisely the *other* way around: counterfactual theorists regard type-C dependencies as special *because* they correspond to our concept of a “cause.”

This is why I don’t think counterfactual theories can make a claim to carving the world at its joints. They take our *concept* of a cause – *our* way of carving the world -- as-is and ask which sorts of things in the world correspond to that way of carving it. The only answer they can give to the question, “Why do C-type dependencies count as “causes”” is a convention-based, instrumental one. Their only answer is: “Because type-C dependencies *answer to our concept*.” Let us call any theory of this structure an OUR-CARVING theory.

Now turn to regularity and necessitation theories of causation. I think they are legitimate candidates for carving the world at its joints. Why? The short answer is: because they can point at something in the world *about* C-type dependencies that makes them special. According to a regularity theory, type-C counterfactual dependencies are unique among all dependencies in the world – and indeed, unique among *all* things in the world – in virtue of possessing an additional world-based property: the property of being constantly connected with other events in specific ways. Similarly, according to a necessitation theory, type-C dependencies are unique among all such things by *necessitating* other things.

Notice what’s different here. Unlike the counterfactual theorist, who simply *begins* with our concept of a “cause” (OUR CARVING) and attempts to determine which things in the world answer OUR-CARVING, regularity and necessitarian theories look for some unique property in the world itself to *justify* carving the world that way. Call any such theory – one that seeks *whether* a given concept (e.g. “cause”) refers to any objectively unique property in the world – a WORLD-CARVING theory.

Now we have a the beginnings of an analysis. A theory is not genuinely metaphysical (an OUR-CARVING) theory just in case it begins with nothing more than our concepts and seeks to provide an account of the kinds of things in the world that answer to our concept. A theory is genuinely metaphysical (a WORLD-CARVING) theory just in case, independently of our conceptual scheme, there are truly unique properties in the world that *justify* conceptualizing the world one way over another (viz. if there really *is* a necessitating relation, there is a unique, robust property in the world that *distinguishes* the kinds of things we call “causes” from all other things). I realize I probably didn’t put this last part very well, but I have to run to watch the Super Bowl. Hopefully what I have written is at least worthy of discussion. If not, oh well. ;)