Trouble with Compatibilism

So I just finished reading Carolina Sartorio's new paper in Phil Review, "Making a Difference in a Deterministic World", and I'm puzzled. I'm not puzzled by the paper -- which is clear and well-argued enough -- so much as I am puzzled by the entire view it is concerned with: compatibilism.

Here, in a nutshell, is Sartorio's main thesis: even if everything in the universe is fully determined in advance, there is nevertheless a sense in which our actions can make a difference in the world. Sartorio's argument then is that because "difference-making" is intuitively tied to moral responsiblity, and compatibilists prior to her paper haven't made adequate sense of how our actions can make a difference in a deterministic world, she has provided a better compatibilist account of how we can be morally responsible for our actions.

But of course here's what puzzles me, and what seems to puzzle most of compatibilism's opponents as far as I can tell: how compatibilism is supposed to be at all interesting, regardless of how we spell it out. After all, suppose I said to you: "All of your actions were determined billions of years ago. But, don't worry, you can make a real difference in the world. I have a philosophical analysis of 'making a difference' that proves it!"  The obvious enough rejoinder seems to me to be this: "Well, of course. Surely there has to be a sense in which I can make a difference in the world. My actions are mine, after all -- they belong to me -- and there's clearly a sense in which they do make a difference in what happens. If the laws of nature cause me to drink a Coke, then, indeed, I have made a difference in the world: there is one less Coke to drink. But so what? It's one thing to say that there's a sense in which I can make a difference in the world. It's another thing to show that it is a sense worth philosophically caring about." 

Now, I know compatibilists think they have arguments that compatibilism is interesting and fruitful. They may say, perhaps, that Frankfurt-cases show that we can be morally responsible even in cases where we couldn't behave otherwise. Yet I have to confess that I've always been baffled by how Frankfurt cases are used in the debate. Here, after all, is the classic Frankfurt case set-up. We are supposed to (A) imagine a person making a choice P, and then (B) imagine a mechanism which, if the person tried to make some alternative choice Q, would force the person to instead choose P.  Here, for instance, is a typical case (copied from Wikipedia):

Donald is a Democrat and is likely to vote for the Democrats; in fact, only in one particular circumstance will he not: that is, if he thinks about the prospects of immediate American defeat in Iraq just prior to voting. Ms White, a representative of the Democratic Party, wants to ensure that Donald votes Democratic, so she secretly plants a device in Donald's head that, if activated, will force him to vote Democratic. Not wishing to reveal her presence unnecessarily, Ms White plans to activate the device only if Donald thinks about the Iraq prior to voting. As things happen, Donald does not think about Iraq prior to voting, so Ms White thus sees no reason to activate the device, and Donald votes Democratic of his own accord. Apparently, Donald is responsible for voting Democratic although, owing to Ms. White's device, he lacks freedom to do otherwise.

Here is the problem. Frankfurt cases are strongly disanalogous to physical determinism. In a Frankfurt case, the person's action is not determined by any actual physical laws.  The sense in which the person "cannot do otherwise" is entirely counterfactual. It is that if they tried to choose otherwise, someone (or some mechanism) would step in and ensure that they don't succeed. But this "trying" isn't even possible under physical determinism. It's not the case that if I tried to behave otherwise than I do, physical laws would step in and stop me. It's that I can't even try to behave otherwise if physical determinism is true (it is not a physical possibility). This, then, is the problem with Frankfurt cases. They push certain intuitions -- that we can be morally responsible for our actions even if we couldn't do otherwise -- because, contrary to determinism, they smuggle in libertarian intuitions. They do this because alternative possibilities are only ruled out counterfactually. For all Frankfurt cases show, the reason why we judge a person free and responsible in those cases is that (A) we judge the person had libertarian free will to make the choice (they caused their action independently of physical laws), but (B) alternative possibilities are counterfactually ruled out because, if they libertarian-ly tried to choose something else, some mechanism would force them to behave the same way.

Accordingly, Frankfurt cases don't seem sufficient to me to philosophically motivate compatibilism. They're a poor analogy to determinism. In order to motivate compatibilism, we would have to tell a story like the one above (about voting for democrats) using determinism. But when we tell such a story, it doesn't seem at all like the person is free or morally responsible. For here's such a story:

Donald is a Democrat, and his vote for the democratic candidate was fully determined by the laws of nature billions of years ago.

I take it when we look just at this case, we all have the incompatibilist intuition that of course Donald isn't responsible for his actions. His vote simply isn't his fault. There is no deep sense in which he could help it at all. And this, again, is very different than the Frankfurt case. In the Frankfurt case, we feel like there's a sense in which the person "could help it" -- but again, that's only because the forcing of their behavior is merely counterfactually triggered. In the Frankfurt case, the person's actual choice isn't determined by anything besides their (libertarian) free choice.

But now if Frankfurt cases aren't sufficient to motivate compatibilism, what can be? A Strawsonian picture of reactive attitudes (see section 4.3 here) perhaps?  I'm happy enough with this kind of compatibilist picture of moral responsibility -- with the idea that we can sensibly hold people morally responsible in a deterministic world. But now, it seems to me, we're no longer talking about free will. We're merely talking about moral responsibility. And so compatibilism about free will, it seems to me, remains insufficiently motivated.

Have I gone wrong somewhere? Am I being unfair to compatibilism about free will? Seriously, I'm willing to be open-minded. I've just never come across an argument for it that seems to me half-way persuasive. If anyone can persuade me, or even just semi- persuade me, I'd be grateful. In fact, I'd be grateful for just a nice discussion about it. :)

Comments

[Clayton]

Hi Marcus,

I'm really sympathetic to just about everything you say in your post. Back when I leaned more heavily compatibilist, I used to think that the Frankfurt cases could be used to motivate compatibilism, but a few folks I spoke with insisted that they could only play a more modest role. They insisted that they could only be used to undermine a premise in an argument for incompatibilism, not support compatibilism. People in this camp might have thought, like you, that there's an important difference between the case and what happens in a deterministic world. (My counter-response was always that I could appreciate the logical point that they were trying to make, but also though that the real reason (notice the 'the') that determinism is a threat is that it takes away alternate possibilities. If so, then the F cases could be used to motivate compatibilism.)

You might find Maria Alvarez's and Helen Steward's recent work on this to your liking. Alvarez has an AJP paper where she argues that the F cases assume a problematic conception of agency that is alien to the incompatibilist's way of thinking. Steward's case for incompatibilism seems to be similar in some respects to yours (judging on the basis of what you've said here). What's interesting about her line is that she takes the point that you seem to be making and uses it to bash both the compatibilists and the standard incompatibilists. Agency properly understood has to be something that makes a difference and she thinks that there's no interesting sense in which agents make a difference in a deterministic world.

[Marcus Arvan]

Hi Clayton: thanks for your comment, and for drawing my attention to Alvarez and Steward's papers. I will definitely check them out. I'm also glad to hear you're sympathetic. I sometimes feel like I'm out in the wilderness on this, given how influential compatibilism is nowadays (much to my chagrin).

Obviously, I'll have to read Steward's paper, but for what it is worth for the moment, I think I provide a scenario in "A New Theory of Free Will" of how agents *can* make a genuine, deep libertarian difference in a (physically) deterministic world -- as, in my view, the *appearance* of complete physical determinism in our frame of reference can be generated (as a kind of illusion) by libertarian free will in a higher reference-frame.

One could of course object that, on my view, reality isn't *truly* deterministic (it only appears that way), in which case I've changed the subject a bit (though, I hope, in an interesting way). But that's another story, and my own axe to grind. :)

[ambrose]

I like this post a lot, and also sympathize with the concern you're expressing. I have another worry, maybe incompatible with yours :) The worry is that the notion of determination has no teeth, doesn't really add anything at all to our picture or the world or our understanding of free action, or whatever. What does the claim that Don's vote was "determined" a billion years ago really amount to? I can't see how it means more than this: given all the information relevant to his vote (except the sheer fact that he votes Democrat) one could deduce with certainty that he will vote Democrat. Surely that's not even _apparently_ incompatible with freedom. In the same sense, given all the information about laws of nature and all the facts about the distant future, one might be able to know with certainty that I'm typing these words right now. (I know the actual world might not be quite like this, but there are "deterministic" possible worlds of this kind; that seems to be the only kind of determination there could be.) So I worry that compatibilism is not false but vacuous: it tells us only that freedom is compatible with a truism or tautology. And the trouble is that this kind of compatibility does nothing to advance our understanding of freedom. (It's like the fact that freedom, if there is such a thing, must be compatible with the fact that bachelors are unmarried.) Anyway, that's a long post, and maybe not directly relevant to the point you're making. But the connection is this: what is it in the notion of "determination" that you think _could_ or _would_ somehow threaten freedom? Where is the incompatibility?

[Daniel]

You're a big Dennett fan, right? Check out Elbow Room. It addresses the sorts of concerns you have, I think pretty persuasively.

[Marcus Arvan]

Ambrose: thanks for your comment, and I'm glad to hear you sympathize too. However, I still think the notion of determination can have teeth. Let me explain.

You ask: "What does the claim that Don's vote was "determined" a billion years ago really amount to? I can't see how it means more than this: given all the information relevant to his vote (except the sheer fact that he votes Democrat) one could deduce with certainty that he will vote Democrat."

To which I reply: I *don't* think that's all determination can amount to. Here's why. I think Humeanism about causation is false. I think the only sensible way to think about laws of nature is via an utterly primitive, causal necessity (or determination) relation. I could go into a long schpiel about why I think this, but perhaps that would take us too far afield. If you'd like the get an idea of why I reject Humeanism, you might want to read the second half of Greg Rosenberg's 2004 OUP book "A Place for Consciousness: Probing the Deep Structure of the Natural World." The broader story is that -- for reasons I allude to in "A New Theory of Free Will" (and which I hope to develop further elsewhere) -- I think *any* world has to be fundamentally dualist in structure. Any "physical" objects and properties are relational. But relations must have *relata* (i.e. analogues to Kantian noumena) which are simple and instrinsic. Finally, I think qualia, subjective time, and causation are all intrinsic stuff of this sort. They are the intrinsic properties "behind" the physical objects and properties measured by the sciences. And I just don't think it's (logically) possible to make a world any other way. Worlds must, at minimum, be comprised by things that bear relations to other things. But things cannot bear relations at *all* unless there are "things in themselves" to bear those relations.

Anyway, if, like me, you think that causal determination is a fundamental, intrinsic part of the world, then determinism has real teeth. It is not merely a matter of what one can deduce. The laws of nature *make* stuff happen in a fundamental, irreducible way that I think -- unless one adopts some wild theory of free will like mine -- rules out genuine free will.

[Marcus Arvan]

Daniel: hmm...I've read Elbow Room, and I didn't find it persuasive at all. What argument, in particular, do you think persuasively addresses my worries? I'd be happy to discuss it.

Also, I wouldn't call myself "a big Dennett fan." Although I studied under him as an undergrad and respect him as a person and as a philosopher, I now reject just about all of his views about mind and the basic nature of our world. I'm a dualist. I think a proper understanding of quantum *physics* -- which Dan admits to not knowing much about -- stands against any naive form of materialism (Heisenberg, he of the famous uncertainty principle, *himself* thought it implies a form of dualism -- and he is by no means alone). See e.g. Heisenberg's 1958 book Physics and Philosophy.

[Griffin]

Marcus: I wasn't quite sure what to think of your post till the paragraph distinguishing compatibilisms about moral responsibility from those about free will. To someone used to using "free will" to designate whatever capacities are required to be properly morally responsible (I think I borrowed this from Galen Strawson, initially, but I don't think it's too uncommon), this altered my reading of the whole post.

Now I understand what's going on a bit better -- it seems to be something like the line of thought leading Fischer to coin *semi*compatibilism -- but I guess I've always been a bit confused about what interesting "free will" issues remain to be debated once we settle the responsibility question. Obviously, if we define "free will" as the unconditional ability to do otherwise, compatibilist theories of free will won't get too far. Presumably, the idea is to identify the term with the ability to do otherwise in some sufficiently interesting sense...but, I guess, I'm left wondering wherein exactly the interest lies. Clue me in?

[Matt]

Marcus,

Are you familiar with Mele's Zygote Argument? (I think the original version is in Free Will and Luck)

I think he's trying to motivate the same type of worry with your line: "Donald is a Democrat, and his vote for the democratic candidate was fully determined by the laws of nature billions of years ago."

[Kristina]

Marcus: You might be interested in Christian List's recent paper in Nous entitled "Free Will, Determinism and the Possibility of Doing Otherwise". He argues for what he sometimes calls "Libertarian Compatibilism". Roughly, the argument is based on different levels of description, so that one can both be a determinist with regard to the physical level of description and also hold that there are genuine possibilities of doing otherwise at the agential level (where the latter is taken to be the appropriate level of ascription of free will). I think it provides an interesting alternative to standard compatibilist positions because, in contrast to Frankfurt style arguments, it doesn't deny that we need the possibility of doing otherwise in order to have free will.

[Marcus Arvan]

Hi Kristina: thanks for your comment, and for pointing me to List's paper. I agree that List's view is an interesting alternative. Indeed, I am of course somewhat sympathetic with List's view, as I defended a (somewhat) similar view in my recent Phil Forum article (which came out before List's paper, by the way). The problem I have with List's argument is that it depends on non-reductive physicalism, a view I take to be incoherent (though I realize not everyone does).

Anyway, if you find List's view interesting, have you read my paper in Phil Forum? I also call my view "Libertarian Compatibilism." The view is quite a bit different than List's, though, and I think (or at least hope) you might find it interesting! :)

[Marcus Arvan]

Kristina: after thinking about List's paper a bit more, I'm finding it more and more problematic for precisely the reasons I find Compatibilism problematic. List thinks that because of non-reductive physicalism, it is (A) physically true that a person couldn't do otherwise (their body couldn't do anything other than what physics commands it), but (B) the *person* modally could do otherwise because, as an agent, their decisions -- their beliefs, desires, etc. -- are not reducible to the physical. This seems to me utterly outrageous, and what's messed up with non reductive physicalism too. I just can't make any real sense of the conjunction. His body *had* to do what it did, but *he*, the agent, could have done otherwise. If anything seems incoherent to me, this does. And, at any rate, like Compatibilism, it seems to me to water down free will no nothingness. If his *body* couldn't do anything other than what it did, how is the agent *really* free?

[Roman]

I'm also perpetually puzzled by the way Frankfurt cases are used, and why they became such a huge publishing industry. But I think there's a serious worry about the "making a difference in the world" phrasing. It seems equivocal, and I wonder how much work the equivocation is doing. If you think of making a difference in the world in one sense, it sounds like you are talking about being outside the world and making changes to it. And that seems crazy. I'm in the world. So anything I do makes a difference. That's true of Coke, too: when it bubbles, it makes a difference to the world. I take it, Marcus, that this is something like what you're getting at? That this is why compatibilism doesn't seem to be saying anything interesting?

But that's not quite right, I think, although I'm somewhat sympathetic to libertarianism. You say that compatibilism isn't saying anything interesting about free will. The implication isn't that we should switch theories, though. It's that before we start talking about compatibilism and libertarianism, we first have to figure out just what *is* interesting about free will, at the very least so we can figure out what could count as saying something interesting about it. Once we figure that out, the question is: now that we've figured out what is conceptually interesting about free will, and by extension what would be interesting if it were true of actual agency, we have to ask whether it is true or could be true of agency in the world as we understand it. I think that is what compatibilism is aimed at doing, though it may not be what many compatibilists in fact end up doing. By contrast, libertarians (or so it seems to me) typically have to add some ad hoc feature to the world to account for free will, which should make the project less interesting (it may be interesting if the ad hoc feature is real, of course, but ad hoc features--like quantum indeterminacy that just happens to match up with decision-making--have the unfortunate property of being dubious).

Anyway, that's my first attempt at answering your question.

[Marcus Arvan]

For more on why I worry that List's view is a non-starter (on the grounds that non-reductive physicalism is incoherent), see Susan Schneider's 2012 paper in Nous, "Non-Reductive Physicalism and the Mind Problem." I think Schneider's refutation of NR-physicalism is spot-on.

[Marcus Arvan]

Roman: that's part of the problem I have. The other part of the problem is what you mention later in your comment: I don't think compatibilists *succeed* at showing any interesting form of free will compatible with determinism. As I mention, I think we can make sense of *reactive attitudes* of moral blame and responsibility on Strawsonian terms alone (i.e. without even concerning ourselves with free will).

Thus, the only *interesting* question I think there is about free will per se is whether we really have it in a deep libertarian sense. I just don't think compatibilists about free will have ever given half-way convincing arguments that there is any philosophically interesting form of *free will* compatible with determinism.

Make sense?

[Roman]

Marcus: two things.

First, there is a difference between saying that compatibilism doesn't succeed in addressing any interesting problems, and saying that compatibilists don't. I was trying to address only the former concern by defending compatibilism as such. If the problem isn't with compatibilism as a view, but only with the particular ways it is defended, the solution isn't necessarily to jettison compatibilism but to jump on the boat and come up with a better version of it.

Second, I don't know what you take the *interesting* question to be. If the question is, literally, "do we have free will in a libertarian sense?" then it follows by definition that compatibilism has nothing interesting to say about that. But if compatibilism is uninteresting based on a stipulative definition, that in itself isn't a very interesting claim.

What I take it you have in mind is that there is some feature of free will that is interesting, and this is a feature that *only* libertarian accounts address (or can address?). I have my doubts, just because I don't think libertarians typically do a particularly good job of making sense of free will (given that libertarianism is typically just compatibilist free will + indeterminacy). But I can't be sure of what you mean until I see what feature you have in mind. I don't see it yet. I lean toward libertarianism because I think compatibilism excludes moral responsibility and moral agency. But you don't think so. So I'm still not sure what else there is that's important enough to insist on ad hoc world-reconceptualization for.

[Marcus Arvan]

Hi Roman: I don't think you're reading my last comment as charitably as you might. I think that because (1) compatibilists have never made what I take to be a half-way persuasive argument for the claim that *compabitilism* is interesting, (2) there are no good reasons as of yet to believe that compatibilism itself can say anything very interesting.

Also, I actually *do* think we are justified in presuming (not "stipulating") that libertarianism is the only interesting view in the area, at least until a compelling argument for some other alternative is provided. For, first, libertarianism surely *is* an interesting view. It implies that we can make choices in some way unconstrained by the laws of physics. Second, it clearly does -- as you note -- seem to be the only view of free will that sits *naturally* (without a good deal of straining) with moral responsibility. Thus, I think that unless and until a good argument is provided for some other option -- and again, I don't think this has been done for compatibilism -- we are justified in presuming that the only *interesting* options on the table are libertarianism and hard-determinism (the denial of free will).

Finally, I don't think it's right to say that libertarianism is "just free will + indeterminacy." I think it has *traditionally* been understood this way, which is why -- frankly -- I hadn't been interested in it for most of my career. It always seemed to me to be ad hoc. However, as you'll see if you check out my recent Phil Forum paper, I don't think this anymore. For, in my paper, I argue that libertarianism is a central part of (A) a grand metaphysical picture, that (B) has some real support from several different philosophical and scientific angles, that (C) provides a deep explanation of quantum phenomena and an array of different problems in the philosophy of mind, time, and personal identity that *no* other metaphysical theory does. In other words, I argue that libertarianism is a part of a very powerful explanation of reality.

Anyway, so while I agree with you that libertarianism has *traditionally* been mysterious and ad hoc (which is why I basically ignored the view for a long time), I now think there are good reasons to think the opposite: that it is a view worth taking seriously, and perhaps even necessary for explaining many parts of our reality, including quantum mechanics and lots of other philosophical problems.

[Roman]

Hi Marcus,

I worry a bit about how much work is being done by the word "interesting" here, since that word is so subjective. For example, I don't think a view's implying "that we can make choices in some way unconstrained by the laws of physics" makes that view interesting: it could make it simply wrong. But to be interesting, at least to me, it must have plausibility.

Then there's the way that you use the word, when you say that "the only *interesting* options on the table are libertarianism and hard-determinism." Hard determinism has struck me as a rational consequence of some other views, but I'm not sure that it's an "interesting" view in itself; nor are the views of which it is a consequence. (If, on the other hand, implausibility is part of what makes a view interesting for you, then compatibilism is *very* interesting, since it tries to merge two things that don't seem to fit at all!) So perhaps we simply apply "interesting" to different things and there's nothing else to argue about.

But here's a suggestion: what if our rational decisions are just independent of the laws of physics, not in the sense that they can somehow violate those laws, but rather in the sense that the two are incommensurable and yet somehow parallel (as Kant argued). Some have insisted that this is compatibilism. If it is (though I do have my doubts), it is to me by far the most interesting theory of free will ever devised. Part of what I conclude from this is that compatibilism/libertarianism isn't a very interesting division in the first place.

I probably won't have time to read your PhilForum paper in the near future, since it's a bit long (though I did read an earlier version of it some time ago). Perhaps you could point me to the place where you explain why you think libertarianism (as opposed to your own view) is an interesting view?

[Marcus Arvan]

Hi Roman: I guess I don't think "interesting" is as subjective as you do. Anyway, I entirely agree with you on Kant's approach. In fact, I take the view I defend in my recent paper to be a scientifically respectable and more detailed model of Kant's approach! On my account, genuine libertarian free will is entirely *compatible* with the appearance of complete physical causal closure. The appearance of causal closure is a kind of illusion generated in part by libertarian free will in a higher reference-frame. That's why I call it "Libertarian Compatibilism". I've always found Kant's view fascinating, and I think I've found a way to provide a detailed metaphysical and physical picture of how exactly it works.

[Marcus Arvan]

P.S.: on your last question, as I mentioned in my earlier comments, I don't think *traditional* forms of libertarianism are interesting. In fact, until very recently I didn't find the entire free will debate interesting in the slightest. I always just took it for granted that we don't have free will. That only changed once the idea for Libertarian Compatibilism popped into my head.

[Marvin Edwards]

The compatibilist position goes like this:
P1: A freely chosen will is when someone chooses for themselves what they will do, while free of coercion and other forms of undue influence.
P2: A world is deterministic if every event is reliably caused by prior events.
P3: A freely chosen will is reliably caused by the person's own goals, reasons, or interests (with their prior causes).
P4: An unfree choice is reliably caused by coercion or undue influence (with their prior causes).
C: Therefore, the notion of a freely chosen will (and its opposite) is still meaningful within a fully deterministic world.

This seems to be a valid argument.