A Suggestion for Setting Up the Problem

What follows is a brief section taken from a draft of the first chapter of my dissertation -- it's basically a suggestion for how to set up the debate that we are all engaged in.  I'd be interested to see what people think.

At its heart, the problem of free will and determinism is about neither determinism nor free will. Or, rather, it’s only about determinism inasmuch as it is about a particularly troubling consequence of determinism – one that would still be troubling even if determinism were false. This troubling consequence is that what we do is ultimately a matter of luck. And it’s only about free will inasmuch as that particular term has been sufficiently disambiguated. Given the various ways the term ‘free will’ has been used, however, it’s almost better to just drop the term altogether in favor of whichever particular disambiguation one is interested in. The disambiguation I’m interested in is control. So, as I see it, the problem of free will and determinism is really a problem about luck and control. But what sort of luck am I talking about? And what sort of control?

We humans think we exercise a distinctive sort of control over our actions. What we do is up to us, in such a way that the same is not true of plants and insects, for example. It’s in virtue of this control – this up-to-us-ness – that we can be held responsible for what we do. You can blame me for what I do partly insofar as what I do is under my control. If, on the other hand, it’s clear that something I do is completely out of my control, then it’s unfair to blame me for doing it. Let’s regiment these truisms of commonsense into the following thesis – call it ‘the control thesis’:

(CT) Some of us at least sometimes have control over our actions.

Although I’m certainly not interested in defending any empirical claims about commonsense intuitions, I think it’s pretty uncontroversial that most of us endorse (CT). Indeed, our entire system of criminal punishment seems to presuppose (CT) as its basis. And many aspects of our relationships with other people would seem to presuppose (CT) as well. In any case, (CT) seems like a thesis one could reasonably take as a datum of commonsense, whether or not one ends up rejecting it in the end.

Here’s another truism of common sense: If something is just a matter of luck, then it is not under anyone’s control. Suppose you and I are about to begin a game of touch-football and we meet in the middle of the field for the coin flip. If I call heads, and you call tails, and the ref flips a fair coin that comes up heads, it’s inappropriate for you to blame the ref for the fact that the coin came up heads. After all, whether the coin came up heads or tails was just a matter of luck, and certainly the ref doesn’t have control over what’s just a matter of luck. Or, at least, it’s only appropriate for you to blame to ref for the result of the coin toss insofar as you have reason to believe that he somehow rigged the toss – that is, insofar as you think the result of the coin toss really was under his control. Indeed, the fact that luck rules out control is, I think, more than a truism of common sense. It’s plausible to suppose that it is true by definition. At least, there is certainly a salient sense of the word ‘luck’ according to which to say that something is just a matter of luck is just to say that it is not under anyone’s control. In what follows, this is precisely what I shall mean by ‘luck’. For clarity’s sake, let’s make it explicit:

‘Some event, E, is just a matter of luck’ =df ‘E is not under anyone’s control’

Given this stipulation, it’s indisputable that luck undermines control. Let’s regiment this claim as the following thesis – call it ‘the luck thesis’:

(LT) Necessarily, if an event is just a matter of luck, then it is not under anyone’s control.

Again, there might be other interesting uses of the word ‘luck’, but in discussions of moral responsibility, this seems to be the understanding that matters.

We need only one more claim before the problem of luck and control shows up in full force. This is the claim that all of our actions are ultimately just a matter of luck. Such a claim also enjoys much support from commonsense. If you don’t feel the pull of the claim, you probably aren’t paying enough attention to the word ‘ultimately’. If you really stop to think about it, it’s clear that much of who we are and what we do was shaped, if you go back far enough, by factors beyond our control. Where we were born, whether our parents aborted us when we were fetuses, how much food we got to eat when we were children, what public schools we attended – all of these things were ultimately just a matter of luck. Or, at least, they were ultimately a matter of luck for us. They might not have been a matter of luck for our parents, of course, but this just pushes the problem back to our parents. Surely who they were and the things that they did were also ultimately a matter of luck for them. And since the same seems to be true of everyone, at any time, all over the world, it starts to look as though all of our actions are, from an ultimate perspective, just a matter of luck. Indeed, human beings came to be the sorts of creatures that they in fact are as a result of various cosmic factors none of which were under anyone’s control. This is not to say, of course, that what we do on a day-to-day basis is a matter of luck – after all, much of what I do depends on decisions that I make. Rather, it is to say that what we do on a day-to-day basis is ultimately a matter of luck. Taking the ultimate perspective here is important, since it is really what makes the claim in question so plausible. Let’s regiment this plausible claim as the following thesis – call it ‘lucky actions thesis’:

(LAT) All of our actions are ultimately just a matter of luck.

Another way we could bring out the force of (LAT) is to say that, from the ultimate perspective, all of our actions originate from factors that are beyond anyone’s control. Now we can see more clearly the problem of control and luck. The problem is that the above three claims appear to form an inconsistent triad. That is, it seems that not all of the following can be true together:

(CT) Some of us at least sometimes have control over our actions.

(LT) Necessarily, if an event is just a matter of luck, then it is not under anyone’s control.

(LAT) All of our actions are ultimately just a matter of luck.

We can see the apparent inconsistency as follows. Suppose that (LAT) is true. By a simple universal instantiation, we can conclude that some particular action of mine, say E, is ultimately just a matter of luck. But (LT) tells us that if an event is just a matter of luck, then it’s not under anyone’s control. So it follows that E is not under anyone’s control, not even mine. But clearly our decision to use me and E as our examples was arbitrary – the claim generalizes to anyone and any action at all. But then no one’s actions are under their control, then (CT) is clearly false.

There are three ways to respond to this apparent inconsistency. Two involve embracing the inconsistency and thus rejecting one of the theses involved; the other involves rejecting the claim that the three claims are inconsistent. First, we could reject (CT). If we go this route, we must reject that any of us ever have control over our actions. Second, we could reject (LAT). If we take this route, we must find a way to defend the claim that some of our actions are not ultimately just a matter of luck. Finally, we could reject the claim that the theses are inconsistent by pointing out a subtle equivocation in our above proof that the three form an inconsistent triad. Notice that whereas (LAT) speaks of actions that are ultimately just a matter of luck, (LT) makes no mention of the ultimate perspective. Strictly speaking, in order to show that the three claims are inconsistent, we would need to add the following premise:

(*) If an event is ultimately just a matter of luck, then it is just a matter of luck.

The third route we are currently considering proceeds by rejecting (*), and thus by rejecting the claim that the above theses are inconsistent. Theorists of this variety will admit that, from an ultimate perspective, all of our actions are just a matter of luck. But they will deny that we need control over the ultimate sources of our actions in order to have control in the sense required by (CT).

What makes the problem of luck and control a genuine philosophical puzzle, of course, is that no matter which route we take, we seem forced to say something very counterintuitive. Either we admit that we don’t have control of our actions, or we accept that somehow we can be in control even from the ultimate perspective, or we say that we can have enough control even without having ultimate control. Each certainly is somewhat unattractive, but we must choose. How shall we choose?

(Note: This maps onto the traditional way of setting up the debate as follows.  Roughly, compatibilists are those who reject (*), libertarians are those who reject (LAT), and free will deniers are those who reject (CT). Determinism and indeterminism work their way in via the idea of luck.  As Mele points out in passing in his recent book, the problem both are supposed to pose to free will can be expressed in terms of luck -- remote luck in the case of determinism and present luck in the case of indeterminism.  This all suggests that the debate over free will and determinism is just a specific instance of the problem of luck and control.)

Comments

[Mark Smeltzer]

Distinguishing between the global (viz., ultimate) and local perspectives is a good place to start. That helps clarify that the tension hinges on whether the types of claims we want control to underwrite originate from the local or the global perspectives. That's where your (*) premise comes into play.

In other words, if I assert, "This murderer deserves your scorn," am I co-extensively and passively asserting that it is also not ultimately a matter of luck that he is a murderer deserving of scorn?

If the (*) premise is accepted, there remains the interesting order of business that involves settling what "ultimate" means here. By the reckonings of some who philosophize about time and causation, the relationship between the past, present and future might not be so heavily tipped as to warrant the claim "the present/future are ultimately dependent on the past, which is a matter of luck". Regardless, in the context of these views about the continuity of time we could still say, "the exact contents of the set of every moment in time is ultimately a matter of luck", at which point we're back to good ol' possible world semantics. However, the difference here is that each possible world in this scenario is a self-contained unit and the relevance of the trans-world comparisons usually used to motivate worries about our lack of "trans-world control" would become quite suspect. Anyhow, this isn't my area of focus, but it is rather interesting since it seems to draw out ambiguities in our concept and the consequences of global determinism, which may make us all the more suspicious about assenting to (*).

Anyway, good post. I hope it generates a good discussion :)

[Kip Werking]

Neal,

I really like the way you've set this up. It shows that you're approaching the problem by addressing the concerns of (e.g.) Nagel and G. Strawson. And, in my opinion, these are the concerns that define the debtate. There are other interesting questions and puzzles in this area of philosophy, but this puzzle is at the heart of the matter.

My one critical note is this: for me, the free will problem is about free will. It strikes me as disingenuous to say that it isn't about free will so much as it is about control, because many or most non-realists would hesitate to deny the existence of any control, even if they would deny the existence of free will. So this rhetorical move (like so many others in this area of philosophy) seems to set the burden against the non-realist too high.

Consider this example: thermostats obviously control room temperature, even if they do not ultimately control it. This sense of control is immune to worries about luck and ultimacy. But we do not think thermostats have free will, because, in part, free will and control are not synonymous.

What conclusion might we draw from this? For one, we need not say that your work is irrelevant or uninterenting. But we might say this: you are not really answering, or solving, the problem of free will and determinism. You're solving a related problem, about the compatibility of control and determinism. But these are two distinct problems.

I think you would be sympathetic to this move, because you write:

"Given the various ways the term ‘free will’ has been used, however, it’s almost better to just drop the term altogether in favor of whichever particular disambiguation one is interested in. The disambiguation I’m interested in is control. So, as I see it, the problem of free will and determinism is really a problem about luck and control."

I'm very sympathetic to your idea here, because I think free will is quite an ambiguous term. But I also think that the following inference is fallacious:

1. I'm particular interested in one aspect or interpretation of the free will problem
2. That aspect or interpretation is the correct one

One might read you as saying this, or something dangerously close it, depending upon how they read the statement "So, as I see it, the problem of free will and determinism is really a problem about luck and control." So my modest suggestion is that you be careful to state that you are exploring your own related puzzle, and this may or may not capture the essense of the free will problem.

[Neal]

Mark,

You're right that the notion of 'ultimate' at play here needs to be clarified and spelled out. I actually like your 'global' vs. 'local' terminology as well. Those too would have to spelled out, of course.

Kip,

You're right -- it's just a bit of rhetoric to say that the free will debate is not about free will. Of course it's about free will. But the problem is with the term 'free will'. So I'll drop the term. But the new term, 'control' will have to be given some sort of content to differentiate it from the sort of "control" thermostats have. This could be done (I think) simply by saying that the sort of control we're interested in is the sort that is required for moral responsibility and desert. Then when I say the debate is about control and you say it's about free will, we're talking about the same thing. Right?

Thanks to you both for your thoughts.

[Daniel]

Surely another way to respond to your inconsistent triad is to deny the truth of LT? (Or is it really an inconsistent quartet, given * ?)

While LT may seem "indisputable", I don't see why, at least according to common sense, it is any more so than CT.

[Jonathan Farrell]

I've a couple of thoughts on what's been said in this thread. The first is about the way 'ultimate' is being used in Neal's setting up of the issues here (I like the setting up, by the way). The second concerns the 'free will'/'control' issue.

LAT ("all of our actions are ultimately just a matter of luck") seems to me to be open to at least two readings. The first says something like "the ultimate source of our actions is some event(s) not under our control." The second says something like "a thorough investigation of the kinds of things that our actions are (the ultimate analysis of 'action') will tell us that they are not under our control."

It seems to me that what Neal says in the paragraph preceding the statement of LAT justifies the first reading, but it's not at all clear whether it justifies the second. And we need something like the second reading if the triad is to be inconsistent. Introducing (*) is one way to get from the first to the second readings. (But, of course, it’s controversial.)

Secondly, Neal, I'm not sure that your response to Kip's comment works. I agree that 'free will' is a rather nebulous term, but replacing it with 'control' seem to take you down one of two paths. You might say, "all that stuff we used point to by saying 'free will' - the stuff you need for MR and desert - I'm now calling 'control'." But for this to work, 'control' has to be as unclear as 'free will' was.

Alternatively, you might say "some of that stuff we used to point to when we said 'free will' I'm now calling 'control'." But the question remains whether control now includes the kind of stuff needed for MR and desert.

(I’ll add a third thought, because Daniel has posted since I started writing this. I don’t think we can reject LT, because Neal has set it up such that LT is true by definition (or have I missed something?).)

[Fritz Warfield]

About "control": dropping "freedom" and "free will" language and replacing it with "control" language is helpful only if, as you rightly agree, we attach adequate content to this notion of "control". But distinguishing your sense of "control" from thermostat "control" over temperature by saying that your sense is the sense required for responsibility (etc.) surely doesn't do this. This doesn't attach content to the notion of "control". It points to an alleged consequence of one notion of control without saying what that notion of control is. And as we all know here, this invites Frankfurt inspired worries about whether a particular notion of control really is required for responsibility (or whether it is merely closely correlated with it).

Replacing your word "control" with "free with respect to" in your presentation of your central problem doesn't seem to change things much if we're not told what control is. As you attempt to unpack this notion of control, I suspect you'll end up offering an understanding of it that turns out to be one of the unpackings of "freedom" in the standard debate. If so, the detour through control language is an unnecessary detour.

So what's the advantage supposed to be of switching from "freedom" talk to "control" talk?

[Neal]

More excellent comments...thanks.

Daniel,

I think Jonathan is right -- the way I've stated it (LT) just says that necessarily, if P then P.

Jonathan,

You point out an interesting potential ambiguity on the idea that our actions are ultimately just a matter of luck. You're right that I will need to give more content to the notion. And that's not something I've done yet.

Fritz and Jonathan,

I think you are both right that when I'm explicit about how I'm using the term 'control', it will turn out that I'm using it in *one* of the ways the term 'free will' or 'freedom' has been used in the traditional debate.

The advantage to dropping the terms 'free will' and 'freedom', however, is that they've been used in so many ways, I fear they cause confusion. Indeed, up until just a couple of years ago, I had always read 'free will' the way van Inwagen does -- as the ability to act and the ability to refrain from acting. But, as you both point out, this is no longer the standard way of using the term. Nowadays, it seems, most people use 'free will' as though it refers to a role to be filled -- whatever sort of control we need to have over our actions in order to be responsible for them. But if that's the role, then why not just talk directly about control? (Of course, we'll still need to talk about the concept that van Inwagen honed in on -- later in the dissertation, I call it 'dual-ability'.)

Perhaps 'control' also has unwanted connotations. But given the prevalence of talk about the "control-condition" on moral responsibility, my hope is that this word will allow us to go straight to this idea, without taking any detours through dual-ability or the ability to have the will one wants, etc. Again, these are important ideas, and they'll have to be confronted, but the setup of the problem itself doesn't require us to confront them.

Also, Fritz is right that the setup of the problem could be equally compelling even if we used some term like 'freedom with respect to'. But I hope that later issues to be discussed - like Frankfurt-examples, for one -- will be more lucid if we make it clear that we're interested in whatever control is required for responsibility by just talking about control. Then we can ask, along with Frankfurt, whether control requires the ability to do otherwise, rather than asking whether freedom requires the ability to do otherwise, a question that may sound like a non-starter to someone who thinks 'freedom' just means dual-ability, or something like that.

[John Fischer]

Perhaps one advantage of using the term "control" rather than "free will" or "freedom of the will" is that the latter terms evoke consternation that sometimes borders on apoplexy among some people; they evidently think that the terms imply or at least suggest some weird faculty of the will. Now I don't think the will is so weird, but why not avoid all of this?

[Alan]

There is one kind of case that combines luck with a form of local control that implies responsibility: someone who, for instance, deliberately uses a "luck" device to place someone in mortal danger, but doesn't directly cause that person's death. (I suppose beyond more exotic cases involving the deliberate use of chance devices to endanger people, there are more common instances like tobacco companies in a luck-related way endangering people's lives through an addictive product that doesn't always kill.) I don't know if this conflicts with LT as Neal intended it, but there does therefore seem to be a sense in which a luck event (either as token or type) is within someone's control, at least in the sense of knowing that luck can be a factor in a form of intentional endangerment.

[Daniel]

Your argument for the 'indisputability' of LT goes something as follows:

(1) A priori, luck implies lack of control.

therefore

(2) Necessarily, luck implies lack of control.

It seems, however, that one might doubt both the the validity and the soundness of this argument.

Firstly, regarding validity, might one not raise Kripkean qualms about the contingent a priori?

And, secondly, regarding soundness, might one not reject (1)? While you introduce (1) as a mere stipulation, I think you need to rely on a stronger claim of aprioricity to support (2).

In sum, someone who accepted both CT and LAT (and, perhaps, *) would presumably deny LT, either because she believed that the implication from luck to lack of control was a priori but contingent, or because she doubted that the implication was a priori at all.

[Neal]

Daniel,

I think maybe I'm confused. As I've stipulated the terms, the claim that luck undermines control is necessarily true in exactly the same way that the claim that all bachelors are unmarried is necessarily true. I've just introduced a term of art, 'E is just a matter of luck' and said that the way I'll be using this term is as follows: 'E is not under anyone's control'. None of this relies on anything about a prioricity, does it?

[Tom Clark]

Neal, I appreciate your focus on control. My altogether predictable 2 cents:

You said “We humans think we exercise a distinctive sort of control over our actions. What we do is up to us, in such a way that the same is not true of plants and insects, for example. It’s in virtue of this control – this up-to-us-ness – that we can be held responsible for what we do. You can blame me for what I do partly insofar as what I do is under my control. If, on the other hand, it’s clear that something I do is completely out of my control, then it’s unfair to blame me for doing it.”

In thinking about what distinctive sort of control is required for blame, responsibility and desert, we have to first specify what’s concretely entailed by being blameworthy, responsible and deserving – what are the responsibility practices we have in mind? But the practices we think are justifiable might well depend on what sorts of control we discover to exist. For example (you mention criminal justice), people’s intuitions about the justness of retributive punishment sometimes shift when they discover ultimate control is impossible. They sometimes (not always) become more consequentialist in their notions about what blame, desert and being held responsible entail about how and whether people should be punished. Consequentialist responsibility practices (constrained and guided by personal autonomy as a primary desideratum to be maximized) only need proximate control, and in fact require it.

What I’m suggesting is that the questions of control and justifiable responsibility practices (what we mean by responsibility, ultimately) are fully linked. We shouldn’t suppose our practices are unrevisable and then engage in a perhaps fruitless quest to establish the existence of the sort of control we suppose is necessary to justify them. Not that you were necessarily heading in that direction of course. So, bottom line: I don’t see what’s so counterintuitive or unattractive about your last possibility (“we can have enough control [to be held responsible] even without having ultimate control”), so long as we calibrate our responsibility practices accordingly.

[Daniel]

Neal,

Do you think that the following set of beliefs is incoherent?:

Believing (a) that some of us at least sometimes have control over our actions (CT), (b) that all of our actions are, ultimately, just a matter of luck (LAT), and (c) that if an event is ultimately just a matter of luck, then it is just a matter of luck (*).

These beliefs entail (d) that it is not the case that, necessarily, if an event is just a matter of luck, then it is not under anyone’s control (not LT). Thus, if the set of beliefs is, indeed, coherent then it is possible for LT to be (rationally) disputed.

Of course, on your stipulative definition of 'luck', it does indeed follow that the set of beliefs, above, is incoherent. My point is that, taking all terms to have their regular (perhaps inchoate) meanings, the set does not appear incoherent.

This seems important if your inconsistent quartet is to present a genuine puzzle: i.e. each horn is independently plausible, yet all four horns are, together, inconsistent. On your stipulative definition of luck, it is trivial that (asssuming *) CT is true if and only if LAT is false.

I don't think this is, intuitively, trivial, and I don't see why you need, implausibly, to arbitrate this away. My suggestion is simply that rejecting LT is a fourth possible way to respond to your inconsistent quartet.

[Neal]

Daniel,

I see, that clears things up. You're not suggesting a rejection of (LT) as I have formulated it. Rather, you are saying that there is another, perhaps more intuitive, meaning of 'luck' according to which a thesis that reads exactly like (LT) could legitimately be disputed. I agree with that.

The main reason I stipulated the particular definition of 'luck' that I did was so that I could tie down what I think is an extremely slippery notion. I had considered leaving 'luck' at a more intuitive level, and then formulating the inconsistent triad (as a "real" inconsistent triad) as follows:

(1) Some of us at least sometimes have control over our actions.
(2) All of our actions are ultimately just a matter of luck.
(3) If all of our actions are ultimately just a matter of luck, then no one ever has control of their actions.

Perhaps you like this way of setting things up better? In some ways, it is cleaner. But then we need debate what it means to say an event is just a matter of luck. Taking my way other way allows to avoid this particular debate in favor of debating about control and ultimacy.

[Daniel]

Neal,

Right - that helps clarify our disagreement. I would certainly opt for your alternative formulation of the inconsistent triad.

Here's another worry with your initial formulation. Given your stipulation about the meaning of luck, the trilemma really reduces to a dilemma, as follows:

(CT) Some of us at least sometimes have control over our actions.
(LAT) All of our actions are ultimately just a matter of luck (i.e., by definition, out of our control.)

This dilemma doesn't strike me as being as interesting as your second trilemma.

[Kip Werking]

Here is a (delayed) response to Neal:

You say:

"This could be done (I think) simply by saying that the sort of control we're interested in is the sort that is required for moral responsibility and desert."

Fair enough. The only problem with this might be that, by answering *this* question (and not the one about control, thermostats, and determinism), you become vulnerable again to the same problems that motivated you to focus on control, thermostats, etc. You initially focused upon control because it is, arguably, less ambiguous than "moral responsibility and desert."

But, one can worry, are "moral responsibility and desert" less ambiguous than "free will"? Perhaps; but if so, I think, not by much. They seem more ambiguous than the term control (in the thermostat sense). To be honest, I'm not sure how ambiguous these terms are, or what their exact meanings (if any) are.

This is not meant to denigrate your work: this problem plagues every philosopher in this area. I, too, would like to say meaningful things about free will and moral responsibility. But I find myself confronted with the ugly possibility that these terms are suprisingly ambiguous, and that others might disagree with me about them because they, innocently, use different definitions than I do.

I am also highly skeptical of the ability of philosophers to capture, from their armchair, any such meaning. I remain enthusiastic about the idea of using surveys and empirical investigations (as Eddy Nahmias suggested in another thread) to study semantic ambiguity and the definitions of terms.

[Paul Torek]

Tom,

Your point (Feb 13) that "the practices we think are justifiable might well depend on what sorts of control we discover to exist" is well taken. But even more important, in my mind, are the disagreements among us about what "our practices" actually are. As we witness in every Congressional debate, individual legislators have different motives for voting for a given law. Presumably, the same holds for voters, or the public in general. We may be largely agreed that a particular murderer should be in prison, without agreeing on why. Moreover, our different views about the whys of imprisonment may drive us to be interested in different sorts of control - rather the other way around, then, from your suggestion.

For my own part, the way some philosophers define our moral responsibility practices leaves me out altogether. If the most robust conception of self-creation turns out to be both consistent and applicable, I still would not believe that anyone's suffering (evildoers or otherwise) is inherently good. According to these philosophers, neither compatibilism nor incompatibilism are options for me, since there is no sort of control that would justify this kind of "desert" in my eyes. They want to exclude me from the debate. I refuse to be excluded.

[Tom Clark]

Paul,

Yes, intuitions about retributive punishment can drive beliefs about ultimate control, and yes again that ultimate control doesn't logically entail that the wrongdoer's suffering is a good in itself. But it's the metaphysical straw some people cling to in order to justify retribution, which is why it's worth debunking.

When people specify what kind of desert they have in mind in terms of how we treat people, the debate about free will gets more interesting, to me at any rate.