A Counterexample to the 'Belief in Ability Thesis'

I’m gradually reading my way through the recent issue of Midwest Studies in Philosophy, on free will and moral responsibility. I’ve read over the half the volume now, and I’m pleased to say that it’s all excellent. My impression is that the standard is actually higher than in comparable general journals. Some of you may have seen the discussion of all invitation journals and their problems on Brian Weatherson’s blog. I share some of the worries expressed there, but issues like this one go a long way towards allaying them.

Here, I want to address Coffman’s and Warfield’s paper, ‘Deliberation and Meatphysical Freedom’. The topic will be familiar to most of you: it’s the link between  deliberation and our belief in freedom. Specifically, Coffman and Warfield want to defend what they call the Belief in Ability Thesis (BAT), which they attribute to van Inwagen and Searle, against putative counterexamples.

As they explicate it, BAT states:

If S tries to decide which of (mutually excluding actions A1 …An to perform, then S dispositionally believes of each of A1 …An that he is metaphysically free to perform it.

Coffman and Warfield deliberately leave ‘metaphysically free’ unanalyzed. They don’t want to commit themselves, here, to compatibilism or incompatibilism (they claim that van Inwagen has been misunderstood on this point: though he is an incompatibilist, he doesn’t build his incompatibilism into BAT). They say a little more about dispositional belief. The only thing that matters here, I think, is that on their analysis an occurrent belief is a limiting case of a dispositional belief. They also have a little more to say about deliberation. Their definition is deliberately minimal: deliberation is a “trying to choose” that ‘occurs before action but after reasons for various actions have been weighed and evaluated (but have not decisively favored one course of action over all others)’. I think this definition is too minimal, but I don’t think that that matters here. I’m willing to accept the definition for the purposes of the exercise.

Coffman & Warfield claim that there are no clear counterexamples to BAT. My purpose is to present one. Let me start with one of van Inwagen’s own cases. Van Inwagen asks someone who doubts BAT to

Imagine that he is in a room with two doors and that he believes one of the doors to be unlocked and the other door to be locked and impassable, though he has no idea which is which; let him them attempt to imagine himself deliberating about which door to leave by.

Bok modifies this case to get a (putative) counterexample to BAT. I agree with Coffman and Warfield that her case fails as a counterexample. Here’s my attempt. I borrow from Newcomb’s problem for the case.

Suppose that Sally finds herself in a room with two doors. She knows that one of the doors is locked and the other unlocked, but she doesn’t know which is which. She also knows that the unlocked door is the door that an alien superscientist, who has a perfect record at predicting the decisions of human beings, has predicted that Sally will choose.

Sally believes, correctly, that she is not metaphysically free to open either door, since one door is locked and will remain locked whatever she does. But she seems free to deliberate about which door to pass through. Therefore, BAT is false.

Comments

[Kip Werking]

"Sally believes, correctly, that she is not metaphysically free to open either door, since one door is locked and will remain locked whatever she does."

Just to be clear: the door will remain closed because she is predetermined to choose the unlocked door, and not because the other door is locked. The predetermination, and not the lock, is the relevant feature here, right (or else what is the point of mentioning the superscientist?)?

I agree wholeheartedly with that interpretation. The idea that deliberation is compatible with prediction (or with our entire lives being designed) cuts to the heart of the free will problem: the difference between hard compatibilism and free will denial. Neil and I (if I understand you correctly) feel the force of this argument: if my entire life was designed by someone else, then I can deliberate all I want, but I do not have metaphysical free will.

But not everyone claims to feel this intuition. Although Watson says that a hard compatibilist is the only kind of compatibilist to be, he remains one. I think (although I've never verified) Fischer is a hard compatibilist in this sense too.

I talk about Newcomb's paradox and hard compatibilism in footnote 30 of my new article Who's Afraid of Creeping Excuplation? (revised here):

"The final footnote of Watson’s “Soft Libertarianism and Hard Compatibilism” notes that “it is hard to see how a designer could determine a creature’s every thought and action without designing a whole world.” See Watson. (1999). “Soft Libertarianism and Hard Compatibilism.” The Journal of Ethics. 3: 353-368. Turing’s halting problem may also suggest a limit upon the predictability of agents. Although there are differences between brains and Turing machines, it is not clear that these differences would be relevant to making humans more predictable in principle. Even Turing machines, however, are only unpredictable upon their first execution. Similarly, Newcomb’s Paradox shows that there is a severe limit upon the kinds of lives a designer can design if the designer also wishes to inform such agents, during the middle of their lives, of its predictions. A sufficiently defiant agent might feel motivated, by learning of these predictions, to rebel against the designer’s predictions and do so. But this contradicts the hypothesis that the designer has designed the agent’s entire life. So a designer cannot design at least these defiant agents and also inform them of its predictions. A designer might, however, design a defiant agent and keep its predictions secret or limit itself to designing sufficiently docile agents who, when they learn of the designer’s predictions, happily oblige. This unpredictability has nothing to do with the concept of free will which this article argues does not exist; nevertheless, it may offer some measure of comfort to certain philosophers."

http://people.wm.edu/~ktwerk/exculpation.htm

[Fritz]

1. About whether our understanding of "deliberation" is too minimal: there are *many* we think perfectly acceptable senses of *deliberation*. We work with the one we do because it is the one we find in the work of those (Taylor, van Inwagen, Searle) who we take to be asserting BAT.

2. About this attempted counterexample: I wonder what you think about our commentary near the end of the paper about PvI's case? My first reaction is to extend what we say there to your case. Of course deliberation is possible in *both* your case and van Inwagen's case. So is the satisfaction of BAT: you would believe of each door that you're metaphysically free to leave through it while also believing that there is a door (one of the two) that you're not metaphysically free to leave through. These are just simple double-minded cases aren't they?

In other words - I am objecting to the *form* of the counterexample (and EJ and I do so often in the article). You can't get counterexamples to BAT by sketching cases in which you deliberate but believe you aren't free with respect to some alternative about which you are deliberating. These won't be counterexamples because examples like this are *consistent* with BAT (because consistent with the deliberator having contradictory beliefs).

[Randy Clarke]

Sally might think about the matter as follows:

"Door A is either locked or unlocked. If it's locked and I choose it, my choice won't unlock it; rather, my choice will falsify the alien's prediction. (That will be a first, but stranger things have happened.) And likewise with regard to door B. So I'm effectively in the same boat with van Inwagen's agent."

Where would she be mistaken?

[Neil]

The disadvanatge of distant time zones: the conversation takes place much slower than I'd like...

Kip:

the door will remain closed because she is predetermined to choose the unlocked door, and not because the other door is locked. The predetermination, and not the lock, is the relevant feature here, right (or else what is the point of mentioning the superscientist?)?

The door will remain closed whatever else happens. One door is locked, the other open, and nothing and no one will interfere with them. I don't like the word 'predetermined', because to my ears it suggests constraint. Sally is determined to choose, the way we all are.

Fritz:

Aren't you simply changing the case? You say that Sally could be deliberating in a double minded way, the way that the agent in PVI's case is (in your retelling). But I'm stipulating that she isn't. She believes that one door is locked, and she fails to believe anything inconsistent with that belief. I don't see that there's anything incoherent or implausible about this.

The reason that I don't like the definition of deliberation is that it implies that when we come to a decision about our how we ought to act by way of assessing reasons, such that we see at the end of assessment that we ought, all things considered, to A, we haven't deliberated. And I think that's extremely odd.

BTW, I read the relevant sections of the Bok last night, and it seems to me that you misinterpret her slightly. You take her to be offering a counterexample to BAT with her 2 doors case. She isn't. She's offering it as a description of what deliberation is usually like from the POV of a consistent determinist. Your first set of criticisms are the right response here: that she assumes that PVI is arguing that determinism is incompatible with BAT.

Randy:

Can't I stipulate that Sally believes (indeed, that Sally truly believes) that alien superscientist will get it right? (If you don't like the case, substitute God).

[Kip Werking]

"The door will remain closed whatever else happens. One door is locked, the other open, and nothing and no one will interfere with them. I don't like the word 'predetermined', because to my ears it suggests constraint. Sally is determined to choose, the way we all are."

OK, just to make sure I understand: the relevant feature of the lock (a constaining device) is that it serves as a secret distinguishing mark between the doors, and not that it would prevent Sally from opening the door (if she tried to hypothetically)? The constraining-ness of the lock is irrelevant to the agent's sense of lacking free will? Your example would work just as well if you replaced the "lock" with a hidden identifier?

Suppose that instead of the lock distinguishing between the doors (one is locked, the other isn't), they are distinguished by one leading a prize and the other not. Sally does not know which one leads to the prize. She knows, however, that superscientist/God predicts that she will choose the door that leads to the prize. This is just as much a counter-example to BAT as your example is, or am I misunderstanding the relevance of the lock? Sally doesn't feel free because superscientist/God assures her that her fate is specified before she chooses, not because the lock would prevent her from opening the door if she tried?

If this interpretation is correct, one might prefer the prize version to the lock version, because prizes have nothing to do with constraint, and the lock might pump the wrong intuitions...

[Neil]

Kip,

It might help if you look at the exchange between Matt Weiner and me over at TAR: http://tar.weatherson.net/archives/004567.html#comments

I want to distinguish between causal determination of our choices - which Fritz claims should not be seen, even by an incompatibilist, to be incompatible with BAT - and actual constraint. Sally is physically prevented from opening one door, in the sense that should she try, she will fail. But superscientst (or God) predicts that she won't try. She doesn't know which door is locked, so from her perspective she is free to try either (and a compatibilist will want to say that she is free to try either).

[Neil]

I want to go back to Randy's comment. I have modified (or precisified) the case so that Sally truly believes that the scientist will predict correctly. I think that response is adequate. But I'm not sure what to say about the case as interpreted by Randy. Can't Sally nevertheless deliberate about what to do, even knowing that one door is locked, if the situation is as described?

[V. Alan White]

Say Sally knows that only one door is unlocked and that she knows that the Superknower knows what she will do. Assume nothing about the metaphysical ground of Sally's decision-making. Now have the content of her decision-making only influenced by the factors mentioned above such that neither door is rationally favored, and thus the final decision is equally weighted for both doors. Sally has a quantum device that randomly outputs 1 and 0 (by 50-50 probability) and she thus associates each of those outputs with a door. She activates the device and acts accordingly with its output (and thus ideally rationally).

Q: Does Sally "try" to decide here at all, except to rely on a random rational process? Does the device "try" to decide?

Q: Could one kick the question of Sally's "trying" back to the original wish to leave the room? Might the example then trivially depend upon a tacit assumption that Sally is trying to leave the room? Does my example also show that Sally is predisposed to try to rely on rational processes? Do these points show as well that the concept of "trying" in this example is recessive in a way that the example itself cannot analyze? (Shades of Galen Strawson?)

[Neil]

Hi Alan,

I'm not sure what you're doing - are you asking some interesting questions about the case, which are orthagonal to BAT, or are you suggesting a problem with the counterexample? If the latter, you'll have to help me see where the problem is. Anyway, I'll answer your questions:

1. It seems to me that Sally is not deliberating, in the sense that Coffman & Warfield specify. She has resorted to a non-deliberative decision procedure.

2. I'm not sure what the point of 2 is. We are assuming that Sally's only options (that is, the only options she takes herself to have reason to perform) are to open door A or to open door B. S far as the concept of trying is concerned, there may be a problem with it (indeed, I have a problem with the notion of volitions, which I take to be equivalent). But it's not a problem for me ; it's a problem for everyone. So I don't see why I need to solve it here.

[Fritz]

There's a one stipulation maximum -- one can't stipulate both that she is deliberating (in the Taylor/van Inwagen sense) *and* that she lacks the BAT demanded "can"-beliefs. Which stipulation do you want?
If it were ok to stipulate both things then you would have thereby stipulated your way to a counterexample to BAT. That doesn't seem fair now does it....

If the former, then it sure looks like she has the relevant "can" beliefs (the case is just like PvI's and we'll say the same thing about it we said about his case near the end of our paper). If the latter, then though she may be deliberating about something, I don't see why the content of her deliberative act includes anything about the alternative in question.

There is regular misunderstanding about PvI's door case (quoted above). People think PvI and Taylor think that deliberation is impossible in this case. They do not. They think that deliberation about a variety of things is possible but that any deliberation about which door to leave through will reveal an inconsistency in belief on the part of the deliberator (she will be "double-minded" as we explain in our paper). Consistent deliberation is also possible in the door case, but not consistent deliberation about which door to leave through -- vary the content of the deliberation and everything is fine, eg, she can deliberate about what door to try to leave through....

[V. Alan White]

Thanks Neil--you did pick up on my puzzlement about "trying", though I do think it's pertinent insofar as the original BAT relies on some sense of "tries to decide". In fact my point in my remarks was to show that that phrase (in BAT) either covertly does too much work ("trying" is part of the claimed ability) or that it explicitly does none (so conceiving of the two doors as options in the example alone carries the weight of believing in one's ability).

I have to run to class now--I'll try to say a little more later when I have the chance. Thanks for struggling with my sometimes "orthogonal" mind though! (I like that!)

[Neil]

I don't agree that there is a one stipulation maximum. Better, there is a one controversial stipulation maximium per thought experiment. By my count, I have zero stipulations in my case. Here's how I count 'em:

1) The double-mindedness stipulation. Double-mindedness is, I take it, the exception to the normal case. For the great majority of my beliefs, I believe that p, and I fail to believe that not-p. So stipulating that this is the case here doesn't count. It's a background condition of normal cases: rather than me having to do work to show that it oughtn't to hold, you have to do some work to show that I'm not entitled to it (not that it can't be the case; your task is less demanding - to show that I have to count it as a stipulation).

2. The sense of deliberating. I thought that you assumed that the sense at issue was so minimal that any notion of deliberation presupposes it. In any case, nothing turns on this here: for any plausible notion of deliberation you care to use, the counterexample will go through.

You say that in case like the one I sketch, deliberation is only possible if the agent is double-minded. But without an argument to back that up, that's just begging the question. It seems to me that deliberation is clearly possible in the case, without double-mindedness, and other people share my intuition (eg. - careful now, appeal to authority coming up - Brian describes it as 'decisive'). So the burden of proof is on you to show that double-mindedness is necessary.

[V. Alan White]

Neil's counterexample pits objective fixity against rational deliberative process, and much in the way Alvin Goldman did some time ago with his famous Book of Life case. Can someone who knows a future outcome of deliberation truly deliberate about that outcome? It seems either we thus oddly believe that we have a real ability to deliberate despite knowledge of its eventual outcome, which seems to gut the whole idea of an ability to choose by deliberation, or we simply cease believing that we are able to deliberate in this way since the outcome is determined (and/or known for sure).

We could preserve the status of deliberation here by conceptual stipulation, such as defining it as a process that has no known outcome, or by stipulative constraint, such as insisting that knowable outcomes are metaphysically impossible. But those strategies (aside from arguments that back them) salvage BAT against Neil's counterexample by petitio only. It's certainly not clear a priori that Neil's scenario is logically impossible (which was Goldman's point about the Book of Life). So Neil's challenge stands.

Which is why I offered the example I set out above, in my stupid and truncated attempt at clarification. If Sally knows that one outcome is inevitable (and not in the nontrivial sense of the givenness of one result instead of two, but that the open door will be the one chosen, and which of the two is not known before the deliberative process is undertaken)-then it does not matter how the process itself proceeds metaphysically. If it is deterministic or indeterministic, the outcome is given as in the example. My example intends to remove prejudicial elements of reason and cause-if one particular door is favored in the deliberative process, then there are factors involved that could stand as influencing the final result in a strictly causal or noncausally inclining way. But if the doors are rationally equal in terms of which should be favored, then these factors are removed. So the deliberative process can equivalently proceed deterministically or indeterministically with a result which is known to be inevitable in kind, and yet terminate with that known result without invalidating the need for the process itself to take place.

But note-it still requires that the final result be not known in terms of which door is eventually selected. Because if that is known, the process itself cannot take place-the result is known and hence there is no process of coming to know!

So what is “being free” with respect to deliberation here? That the process of deliberation takes place at all. That is, the sense of “freedom” in this example is contrasted with the case of there being no process of decision at all. (In Goldman's case the protagonist does not know how the Book of Life brings about its predictions exactly-he/she must go through the psychological motions of fulfilling them.)

Is a decision of any type/kind “free” with respect to there being no such thing at all? Compatibilists might well agree. I doubt incompatibililists would, since the metaphysical nature of the process is what interests them. But then I would refer them to using a quantum device to resolve a rational 50-50 decision process, and ask what of any real philosophical significance (control/responsibility) is added by that device. Now add in the influence of any purely probabilistic factor in another decision that is not 50-50-what is added by that? And yet this is what incompatibilists seem to need to make their views appear to have ultimacy (as Kane puts it).

[Neil]

I'm not sure if this is a response to Alan, or a remark inspired by Alan... Here's an incompatibilist worry: if Sally's deliberative process is indeterministic, then it is false to say that the results are predictable. Of course, if this is the best response to my case then I have still shown something significant. Fritz wants to claim that BAT is a thesis that is neutral between, and ought to be embraced by, all players in the debate. If the only way to block my objection is assuming indeterminism, then he and Coffman are wrong about that.

[Fritz]

Well, I'll try, in one more way, to say something.

1. If we can just stipulate whatever we want, then here's a counterexample to BAT. I stipulate both that an agent deliberates about options A, B, C, etc.... and I also stipulate the the agent does not satisfy BAT's consequent. Counterexample.

2. The notion of deliberation in play is not so trivial that any notion of deliberation presupposes it. We said, and I've repeated, that the Taylor/PvI sense of deliberation is "minimal" in that some people want more from a notion of deliberation, and others want to acknowledge thicker senses of deliberation. For example, one might sensibly think that there is a legit sense of deliberation that involves the weighing of reasons for and against various alternatives leading to a decision. It seems to me that is a fine notion of deliberation but it is not the Taylor notion, *nor does one deliberating in this new sense thereby deliberate in the Taylor sense*.

3. Here's what's easy - describe an agent who is deliberating in the Taylor sense. Here's what's harder: settle both that this agent is deliberating *about alternatives A B C etc...* AND that the agent doesn't satisfy BAT's consequent with respect to each alternative within the content of deliberation. This is where Coffman and I think so many people mess up attacking Taylor. Describing cases in which an agent is clearly deliberating about A B C etc... and in which an agent believes of, eg, A that he is not free with respect to it is not going to get anwhere as a counterexample: it's simply not of the right form to be a counterexample to BAT. A counterexample requires that one not believe that one is free (not that one believe that one is not free) with respect to at least one item withing the content of the deliberative act. Attempted counterexamples of the wrong form are not going to settle the dispute.

4. So far as I know, all Coffman and I tried to do was show that counterexamples have not been produced (most offered are of the wrong form to be counterexamples). We made no serious effort to argue for BAT, offering only an indirect argument at the end of our paper.

5. Your preferred approach currently seems to me to take this beginning: stipulate that I lack the BAT-implied "can" beliefs. Well, then what, it's easy to see that an agent can *deliberate* but why think the content of the deliberation includes the option for which you've stipulated the absence of the relevant BAT implied "can" belief? In PvI's case, for example, if we stipulate that the agent lacks at least one of these beliefs, we need one more thing to complete a counterexample: we need that the content of the deliberation includes the component in question (rather than being a deliberation about some set of options that does not include that component). What we've said is: we've seen no example that makes it clear that these conditions are satisfied together and THAT is what one needs to have a clear counterexample to BAT.

[Fritz]

PS --

1. I assume that the appeal to "others [eg, Brian] being convinced" was a joke. His own alleged counterexample seemed to indicate that he wasn't really clear about what BAT said didn't it?

2. My appeals to the possibility of double-minded deliberators have, I fear, been misunderstood. That approach is another way of illustrating that some of the offered "counterexamples" are of the wrong form to serve as counterexamples. Stipulating that an agent believes that he can't do something is helping construct a counterexample to BAT, because BAT is silent about what beliefs about *inability* -- BAT makes commitments only about beliefs *in ability*. So I don't need to "beg any questions" and/or argue that certain agents really are double minded -- it's enough to point out that alleged counterexamples aren't of the right form precisely because they don't rule out double mindedness [and stipulations to rule it out will be one stipulation too many as discussed above].

That's all for me on this topic for now.

[Kip Werking]

Neil,

I’ve taken a look at the comments in that other thread, but I still think it would be most helpful if you could answer a few short questions, at your convenience. In particular, I would like to know:

1. If you think my “prize” thought experiment also succeeds as a counter-example to BAT?
2. If it doesn’t succeed, how is the difference between them relevant?

Some of your comments in the other thread suggest answers. There you wrote:

“If your case is a successful counterexample to BAT, then it should be true that the belief in determinism is by itself (rationally) incompatible with belief in BAT.”

Do you also think that if my prize thought experiment were a successful counter-example to BAT, then determinism is by itself incompatible with BAT? You also wrote:

“Now, the nice thing about my case is that I do have an actual physical constraint, whereas in Brian's it is instead a future event that is determined.”

My question to you, again, is *why* is this a nice thing (perhaps you don’t want your example to commit to determinism or indeterminism?)?

It seems to me that the superscientist (like the future self in the time traveler example) is doing the work in your example, and not the lock.

[Neil]

Fritz,

It seems to me that I'm missing something: I just don't get why you think that the counterexample is of the 'wrong form'. We agree that if you can stipulate whatever you like, then counterexamples are both easy to produce and do not settle anything. I'm claiming that my stipulations are reasonable: that single-mindedness, in particular, is the normal case so that it counts as background condition. For that reason, anyone who wants to claim that I can't have it in my case needs to provide an argument to that effect. Maybe I have just stipulated the notion of deliberation, and it's a controversial one. Fine, but so what? For *any* plausible notion of deliberation, the counterexample succeeds (suppose that Sally is assessing which door would be better to go through: then she deliberates*, and the counterexample goes through). In any case, it seems that I'm entitled to the stipulation: since you and PVI assume it, I'm allowed to take it over as an assumption (indeed, I'm *required* to take it over, on pain of changing the subject).

That takes care of your (1) and (2). Now (3) I'm puzzled by:

A counterexample requires that one not believe that one is free (not that one believe that one is not free) with respect to at least one item withing the content of the deliberative act. Attempted counterexamples of the wrong form are not going to settle the dispute

The consequent of BAT states: S dispositionally believes of each of A1 …An that he is metaphysically free to perform it . Now, if S believes that he is not free to perform one of the alternatives, then S falsifies the consequent, because it is not that the case that he believes of each that he is free to perform it.

4. You say that you have not attempted to give an argument for BAT. Yes, I know (though you do give some reasons why it is intuitive). You say that no counterexample has succeeded. That's why it is appropriate for me to offer a counterexample.

5.

Your preferred approach currently seems to me to take this beginning: stipulate that I lack the BAT-implied "can" beliefs. Well, then what, it's easy to see that an agent can *deliberate* but why think the content of the deliberation includes the option for which you've stipulated the absence of the relevant BAT implied "can" belief?

Two points, First, I have not *stipulated * that the agent lacks the "can" belief: I have described a situation in which the agent has in her possession certain facts that lead her reasonably and truly to believe with respect to one of the options that she lack the relevant ability. Second, I have said that she deliberates with regard to the doors which to open, and once again this is not just a stipulation. The set up makes it reasonable for her to deliberate which door to open, despite her knowing that one of the doors is impassable. Why is it reasonable for her to deliberate about these two options, and not about something else? Because she truly believes that which door is locked is sensitive to her deliberative processes.

I do fear that we have talked past one another in this thread. But *I* (at least) have found it useful. Thanks for the input. It may be that we have reached a dialectical stalemate.

[Neil]

Kip,

The prize case isn't (as Fritz would say) a *clear* counterexample to BAT. In your case, to say that Sally can't open one door. b/c superscientist predicts that she doesn't, seem to presuppose an incompatibilist analysis of 'can'. So I think that your case will succeed just in case, at the end of enquiry, an incompatibilist analysis of ability terms turns out to be the right one. I wanted to follow Fritz in leaving ability terms unnalysed, and thereby not begging any questions against anyone.

[V. Alan White]

After rereading the stuff above, I think I can finally see the essential nature of this dispute over BAT, enough to appreciate both Neil’s and Fritz’s (if I may) perspectives. And enough to apologize to all parties for my lumbering interloping. Pledge: If my thought starts to fleet, I MUST delete.

The dispute concerns how Sally’s knowledge of the Superknower’s prediction/knowledge affects her beliefs about her abilities. From Neil’s perspective, it nullifies any such belief about any particular alternative of deliberation, but without fixing on particular alternatives to do so; from Fritz’s, it fails to nullify any particular belief about an ability associated with any particular alternative of deliberation. That’s a subtle difference, but a real one. So Neil’s counterexample works by making the epistemic effect of Sally’s Superknower operate upon the alternatives collectively, whereas Fritz insists that this is of the wrong logical form because a BAT counterexample must work for an alternative separately (and at least one separately).

Put another way, Neil’s perspective treats the ability question as a type/general claim, where Sally believes only that she is able to choose and do only one of several alternatives, but not all, so the others, whatever they are, she is not able to choose and do. Fritz treats the ability question as something of a token/specific claim of belief in ability of each separate alternative of choice and action.

Looking back at BAT, I see Fritz’s point—Neil’s example doesn’t apply to each of the deliberated alternatives. On the other hand, given that real deliberations involve the known fact that only one of several mutually exclusive alternatives can be realized, why shouldn’t the ability question be interpreted as a general one that applies indifferently across all of them? So—is BAT a correct thesis about ability claims as quoted above?

Again, forgive my ramblings.

[Neil]

Thanks Alan, that's useful. If I understand your point, BAT should be expressed as follows:

An agent can deliberate about what to do iff she believes with regard to each alternative that she is free to do that alternative.

So Sally must believe that she is free to do A and free to do B. But she doesn't believe that: she believes that she is free to do A or B.

However, I still think my CE succeeds. Sally *can* deliberate, even though she doesn't believe that the alternatives between which she chooses are open to her. She can deliberate because she believes that which door is locked is sensitive to her deliberative process. Deliberation is not a sham for her, for that reason.

There is a very important and difficult question lurking in the background, about how to understand ability claims. Unfortunately, as soon as we address that question directly, the pretense of neutrality between compatibilism and incompatibilism has to be dropped.

[Kip Werking]

Neil:

But BAT (as written above) doesn't say "can." It says "metaphysically free." Personally, I think "metaphysically free" implies more than conditional-can, it implies possibility.

More importantly, I think that interpretation *captures* what is important about BAT. Suppose that we formulated BAT in this less controversial way: "then S dispositionally believes of each of A1...An that it is possible for S to actually perform S in this world." Let's call this BAN'.

A compatibilist might nit-pick and protest that some lingering compatibilist interpretation of ability remains in BAN'. But I am trying my best to avoid that. My point is just that: my prize counter-example should be a real counter-example to BAN', if not BAN. Furthermore, that the prize counter-example is a counter-example to BAN' is an essential point in this free will debate. It is important to realize that all of the deliberation we do in this life does not, for example, prove that our entire lives were not designed by a superscientist.

Originally, I suggested that the superscientist, and not the lock, is doing the work in your example. But on your interpretation (according to which prize and the lock examples are not functionally equivalent), then what work is the superscientist doing?
Consider the superscientist-less version:

"Suppose that Sally finds herself in a room with two doors. She knows that one of the doors is locked and the other unlocked, but she doesn’t know which is which."

This version is just like the original, except that Sally does not know that the superscientist has predicted she will pick the unlocked door. But this leaner example seems to disprove BAT just as well as your superscientist version. I could even repeat your explanation for its effectiveness as a counter-example:

"Sally believes, correctly, that she is not metaphysically free to open either door, since one door is locked and will remain locked whatever she does. But she seems free to deliberate about which door to pass through. Therefore, BAT is false."

There is no mention of the superscientist, because (I think), the superscientist is gratuitous. Either the lock does the work in the example or the superscientist does. But I do not see the reason for both of them.

But there is a greater problem. I think your suggested counter-example equivocates between the actions Sally is considering. BAT says that if Sally is deliberating between several options, then she must believe that she is metaphysically free to perform those same options. In your example, Sally deliberates about which door to *try* to open. She deliberates about *trying* to open either door. But when you consider which options Sally is metaphysically free to perform, you consider whether Sally is metaphysically free to actually open the door. If Sally were not uncontroversially free to even try opening either door, she would not deliberate about trying. In this way, I don't think your example is a clear counter-example to BAT.

Ultimately, whether this sense (in which your example succeeds to disprove BAT) or the other (in which it doesn't) wins, doesn't strike me as terribly important. The importance of BAT should be, I think, that it is false. This is captured by BAT’. BAT’ is false because people can still deliberate about options even though they know they are fated/determined to only perform one of them. Indeed, I do this every time I deliberate.

[Neil]

Kip,

First, I doubt you can restate BAT (or BAN) so that it (a) captures something intuitively true about our freedom and (b) can't be given a compatibilist interpretation. Second, if you do, you change the subject: PVI, and Fritz, want BAT to be accepted by all participants in the debate (see Nelkin's article on this point, which is here http://www.ucl.ac.uk/~uctytho/dfwNelkin.html).

Here's the work the superscientist is doing (and the reason the "trying" interpretation won't work). In your modified case w/o superscientist (actually PVI's original case), Sally can only deliberate about which door to try to open. But in my case, she can deliberate about which door actually to open, because she knows that whichever door she tries to open she will *actually* open: this is so because which door is unlocked is sensitive to the upshot of her deliberative processes.

[Kip Werking]

Neil,

Thanks for your patience. I am finally starting to understand the motivation behind your counter-example, although I'm not sure yet that it succeeds.

[Neil]

I am finally starting to understand the motivation behind your counter-example, although I'm not sure yet that it succeeds.

Me and you both!

[Kip Werking]

Ok, after a moment's reflection:

1. BAN' satisfies both of those conditions. It captures something intuitively true about freedom: that we can deliberate about options even though, for example, our entire lives were designed by a superscientist (and in my opinion, this sense in which BAN is false is the sense it is interesting, perhaps because I lean incompatibilist). And it is formulated to specifically deny a compatibilist interpretation. Or rather, it is silent about compatibilism. BAN' makes no reference to freedom or free will. It only makes reference to possibility in this world.

2. I think I'm convinced that your example is a true counter-example to BAN. It is a peculiar and fascinating situation in which an agent seems simultaneously so enslaved (because fated/determined as predicted by the superscientist) and yet so free (because nothing conspicuous constrains Sally's deliberation, except perhaps Sally herself). It is almost as if, by deliberating, Sally *makes* the door locked or unlocked.

In hindsight, I think the spirit or thrust of your counter-example is much like that of the prize example. But your example includes the lock, which may satisfy those who wish to keep BAN silent about compatibilism.

[Mark Smeltzer]

What's BAN?

I see Kip's definition for BAN', but I can't find one for BAN in the thread...

[Kip Werking]

Oh, my mistake. I meant to write BAT', not BAN'.

I used to call TNR TRN too.

[Mark Smeltzer]

That solves the mystery!

No worries :)

[V. Alan White]

Neil et al--

Does BAT capture the mundane sense of ability that ordinary choosers have? First, as Neil correctly interpreted my last remark (disregard the previous s**t), BAT distributes ability belief to each of all deliberated alternatives, and apparently conjunctively. From PvI's example, one can plausibly believe that "I can choose and open door 1, and I can choose and open door 2"--let's call this D1&D2. As a conjunction, both appear to be logically possible as subsequent events to one's preceding deliberative situation. But then one must add in the principle of the exclusivity of choice and action (PECA)--that of any number of logically possible alternatives of choice and action, only one actually occurs, thus finally ruling out others. This is an axiom of action based on the law of excluded middle and is unquestionable. So a real, self-reflective deliberator about the doors believes (D1&D2)&(PECA). But then that translates into the deliberator (my god I'm starting to sound like the governor of California) believing D1orD2, and given the relevant truth conditions, with an exclusive understanding of that "or". So the distribution of belief across the alternatives is that only one is realizable, and thus the other(s) is (are) false as a logical consequence. So though Sally may not know beforehand which of D1orD2 is false as an eventual act, she certainly does not inclusively believe that both are eventually true. So Sally believes that (at least one) of the alternatives is eventually false.

This counts against Fritz's and Coffman's account, which holds (against PECA) that beliefs about ability distribute over all conjuncts of choice/action equally. As purely logical possibilities and in some weak sense that is plausible. But add in PECA--which is a realistic constraint--and one must (in type fashion, distributed over all alternatives) believe that only one alternative is that which one is able to choose and do. So even though one does not know which of logically possible alternatives one is not able to do, there is a belief that this is true of all of those other than that which one finally does. And this favors a sense of Neil's counterexample against BAT that has realistic legs. In fact, the fact of PECA counts against BAT itself as a correct account of how our beliefs work on our conceived alternatives of thought and action.

[Neil]

Hi Alan,

I'm not sure I understand the point (not your fault; I'm logically challenged). I take it that you think that I am claim that Sally must believe that at t she can choose A and B, whereas Fritz thinks that she need only believe that she can choose A or B - so that if she has the ability to choose just one, then BAT is true. But consider PVI's original case: the agent is in a room with two doors, one of which is locked and the other unlocked, but she does not know which. PVI says that (unless she has inconsistent beliefs) she cannot deliberate, and Fritz concurs. But in this case, she can choose either A or B.

[EJ]

First off, thanks to all who have contributed to this thread so far for very interesting and helpful discussion of our paper! Because it contains stuff that still needs discussion, I’m going to address Neil’s reply (October 22, 5:07pm) to Fritz’s most recent post. I’m going to argue two claims: (1) Neil’s example doesn’t engage BAT, and (2) one natural “tightening” of Neil’s example (i) engages BAT but (ii) isn’t a clear counterexample to BAT. Many (if not most) of my remarks will reiterate or amplify points already made by my distinguished partner-in-crime.

First, Neil’s example doesn’t engage BAT: it leaves open the question whether Sally believes of each door that she’s (metaphysically) free to open it. The relevant fact in Neil’s case is that Sally believes [It’s false that I’m free to open either door]. Writes Neil:

“I have described a situation in which the agent has in her possession certain facts that lead her reasonably and truly to believe with respect to one of the options that she lack[s] the relevant ability.”

But it’s perfectly consistent with Sally’s holding such a belief that she ALSO believes of each of the relevant options that she’s free to perform it. I worry that this modal fact hasn’t been fully appreciated. Writes Neil:

“The consequent of BAT states: S dispositionally believes of each of A1 …An that he is metaphysically free to perform it. Now, if S believes that he is not free to perform one of the alternatives, then S falsifies the consequent, because it is not that the case that he believes of each that he is free to perform it.”

The inference involved in the second sentence is plainly invalid: it’s perfectly possible that S has inconsistent beliefs. More fully, it’s perfectly possible that S

- believe of each of A1…An that he’s free to perform it
AND
- believe [It’s false that, for each of A1…An, I’m free to perform it].

So, Neil’s example simply leaves open the question whether Sally believes of each door that she’s free to open it. But then Sally may satisfy BAT’s consequent in Neil’s example. That example doesn’t engage BAT.

A natural way to “tighten” Neil’s example is to stipulate that, throughout the duration of the case, it’s false that Sally believes of each door that she’s free to open it. (Incidentally, such a stipulation is perfectly appropriate. For one thing, it’s clearly possible: surely there’s a world in which Sally fails to believe of each of those doors that she’s free to open it.) From here, what we need for a successful counterexample to BAT are further details the consideration of which gives one a strong sense that Sally is trying to decide which of the doors to leave by (open, pass through), or perhaps a strong sense that she could so try to decide.

For what it’s worth, my honest (not to mention somewhat convenient) reaction to the "tightened" case is this: I just don’t get a strong sense that Sally is trying (or, could try) to decide which of the doors to leave by – nothing like, say, my sense that Henry doesn’t know there’s a barn when considering the (in)famous “Fake Barn Country” cases. And I’d be mildly surprised were I to learn that most folks who consider (a suitably “tightened” version of) Neil’s case get a strong sense that Sally is trying (or, could try) to decide which of the doors to leave by (open, pass through). (Perhaps I just set myself up to receive a mild surprise the next time I check this blog.)

On the other hand, were Neil to claim (e.g.) that Sally could now try to decide which of the doors to try first, or that Sally could now *reason* about (as opposed to try to decide) which of the doors she’ll leave by (“I’ve formed an intention to open that one – But, according to the superscientist, the chosen one is the open one – So, I’ll be leaving by that one”), then I wouldn’t suffer from a similar “phenomenological block.” But then we still wouldn’t have a successful counterexample to BAT.

One final remark. I (this is me talking, NOT Fritz – I haven’t run this by him, as he’s “jumped ship” for Michigan this semester) am now thinking a more promising strategy for “counterexampling” BAT is to begin by stipulating that a given agent is trying to choose among actions A1…An. Then add further details that make plausible the claim that the agent doesn’t believe of each of A1…An that she’s free to perform it. If memory serves, Randy Clarke employs something like this strategy in his recent, excellent book on free will. Fritz and I don’t explicitly tackle his case in our paper. Briefly (and again, assuming memory serves – I don’t have Randy’s book here), one of my worries about Randy’s case is that it putatively involves an agent who avoids holding the relevant “free to” beliefs by exercising a kind of “doxastic self-discipline.” I worry that there’s implausible Doxastic Voluntarism (DV) lurking here. In any event, I think an instance of the strategy suggested above that stays away from DV is, other things equal, more likely to succeed than one that doesn’t.

[Neil]

Thanks EJ. I, too, am grateful for the discussion. For a variety of reasons, I'm a little isolated down here (the Australian philosophical community is small; Australia is a long way away from the good conferences; I work in an applied ethics centre where people tend to have interests other than the metaphysics of free will). So the opportunity to engage with people like you and Fritz, and the other contributors to this blog is really important for me.

On to BAT. I've already given reasons why I don't think that I need to stipulate that Sally doesn't have inconsistent beliefs. To me, the claim that she might is a cheap shot: it's like saying that a case involving one agent shooting another might fail because the case as described is consistent with the physical laws being different or the agents being immortal (alright, it's not that bad). There is, I said, a stipulative burden here, and it falls on those who would avoid the CE by insisting on double-mindedness. In any case, this is not important since you (but not, I think Fritz) accept that I can have the 'stipulation'.

So let me have it. You don't get the intuition that Sally then deliberates about *which door to open* (as opposed to *which door to try*). I'll just give you my reasons for thinking that Sally deliberates about exactly that. Then maybe people can record their responses (in another virtual incarnation, I'm an experimental philosopher).

Sally deliberates about which door to open because she reasonably (and probably truly, depending on exactly how the case is spelt out) that the door which she tries she will open. She believes that because she knows that which door is in fact unlocked is sensitive to the upshot of her deliberative processes. So she knows that, with a probability at or near 1, the door she selects will in fact be the door she leaves by. Hence she deliberates not about which door to try, but which to open. In a similar fashion, I do not deliberate about whether to try to move my arm, but about whether to move it: because I know that the trying is extremely likely to be successful.

The polls are open....

[EJ]

Thanks for the very helpful, “debate-advancing” post, Neil! Two things.

One: Looking back at your initial description of the case, what Sally knows is (something like) [I will choose the unlocked door]. I think such a description is more felicitous than one according to which the content of Sally’s knowledge involves something about deliberation. Basically, the latter is vulnerable to a “question-begging” charge, whereas the former isn’t.

Two: You seem to think that Sally’s knowledge that she’ll choose the unlocked door clearly positions her to try to decide which door to open. That’s just not obvious to me. Moreover, I think there’s a plausible error theory in the neighborhood for your intuition. To my mind, what Sally is now clearly positioned to do is *reason* about which door she’ll open on the basis of (a) her knowledge that she’ll choose the unlocked door and (b) her knowledge about which door she’s chosen (once she makes a choice/forms an intention, which of course one can do without “trying to decide” in the relevant Taylor/PvI sense). I hypothesize that you’re conflating (i) Sally’s being positioned to try to decide which door to open with (ii) Sally’s being positioned to reason about which door she’ll open. Because this distinction is subtle, it wouldn’t be shocking were excellent philosophers to run these things together.

[Kip Werking]

EJ writes:

"For what it’s worth, my honest (not to mention somewhat convenient) reaction to the "tightened" case is this: I just don’t get a strong sense that Sally is trying (or, could try) to decide which of the doors to leave by – nothing like, say, my sense that Henry doesn’t know there’s a barn when considering the (in)famous “Fake Barn Country” cases. And I’d be mildly surprised were I to learn that most folks who consider (a suitably “tightened” version of) Neil’s case get a strong sense that Sally is trying (or, could try) to decide which of the doors to leave by (open, pass through). (Perhaps I just set myself up to receive a mild surprise the next time I check this blog.)"

But this is possible, right? Isn't this what I do every time I deliberate? I feel/think that the course of my entire life is fixed by (for example) the state of the universe before I was born. I feel, in other words, that my life is more or less fated (I certainly don't think that any quantum noise in the system will enhance or enable my metaphysical freedom).

Yet I deliberate all of the time. I deliberated about whether to write this comment. Nothing about deliberation itself contradicts the idea the destiny. Determination shows that the outcome of deliberation is fixed, but knowing that the outcome is fixed does not prevent us (me, if not Sally) from deliberating. Perhaps, however, I am being irrational, and Sally is more rational than I am. Maybe the perfectly rational agent wouldn't (couldn't!) deliberate?

The critical distinction here (again) is between epistemic and metaphysical freedom. What enables Sally to deliberate is her *ignorance*. She is metaphysically enslaved but epistemically liberated: she knows not yet which door she will choose. As long as she does not know, she can deliberate.

I actually feel this is a critical point to appreciate in the free will debate. If deliberation ensures that we believe that we are free, then (by contrapositive) those who do not believe they are free must not deliberate. But those on the dark side of the free will debate profess to deliberate, so we must we deluding ourselves. Something has gone wrong.

EJ also writes:

"I hypothesize that you’re conflating (i) Sally’s being positioned to try to decide which door to open with (ii) Sally’s being positioned to reason about which door she’ll open. Because this distinction is subtle, it wouldn’t be shocking were excellent philosophers to run these things together."

The distinction between "try to decide" and "reason about" is subtle, too subtle for me to appreciate how it is relevant here. Could you elaborate?

[Fritz]

I'll take your last point Kip:
the distinction between (Taylor's) deliberation and other senses of deliberation is fundamental to this discussion, because BAT is formulated using the Taylor undertanding which is neither standard nor immediately intuitive (or so it seems to me). As we explain in the article, Taylor's deliberation is what sometimes happens after we have reasoned about what to do. Sometimes reasoning *settles* what to do and the next step is acting in accordance with the verdict of reason (ceteris paribus). But sometimes, at least, the reasons don't settle the matter in this sense and we have to decide what to do after reasoning ends but before the decision is implemented. This involves a trying to decide that Taylor stipulatively identifies as his sense of deliberation. That's the notion in play in BAT. van Inwagen followed Taylor. I *think* Searle arrived at the same idea independently (I'm not 100%certain that Searle's is the same idea and also not certain it was arrived at independently).

[Neil]

I'm with with Kip: I'm not sure that I get the distinction EJ and Fritz are trying to make. One thing Sally might be doing is attempting to predict her future behavior. Now, that seems, intuitively, quite different from deliberation about what to do. But focus on the kinds of considerations that Sally thinks about when she engages in the thinking about her future action: she thinks (or at least *can* think) about such questions as: which is the best door to open? Which door is the door I feel like opening? Am I somehow inclined to opening one door rather than the other? Should I just plump for one door? For *any* notion of deliberation, Sally can engage in the appropriate thought processes: she can try to work out which door would be best to open (which would be to deliberate in a sense which is not the one that EJ & Fritz want to employ) or she can try to decide after having exhausted the reasons. Since she knows whatever decision procedure she uses will result in her selecting the open door, she can use that procedure.

[Fritz]

EJ and I are not personally wedded to the Taylor sense of deliberation. That is, we don't think that is is the only acceptable sense of deliberation, nor do we think that it is preferable to other acceptable senses. In fact, at least as compared with ordinary and contemporary use of the term (driven, no doubt, by the commonplace "jury deliberations" sense of the term) I find the Taylor sense a bit strained even. That said, in evaluating BAT, it is the *Taylor* sense that is in play: I know of no one who has seriously asserted BAT or a BAT-like thesis for any other sense of deliberation. I suspect that all or most theses in this family of possible BAT-like theses would be vulnerable to counterexample. If one really doesn't understand the Taylor sense of deliberation, I would think the appropriate response to a defense of BAT would be "sorry, I don't understand it becaust I don't understand the sense of deliberation in play" rather than "here's a counterexample to BAT".

[V. Alan White]

Before presenting my (I hope and I bet YOU hope too) final thoughts on this, let me say how much I appreciate the work of everyone in this blog thread. Thanks so much for such good food for thought, even if I haven’t always properly digested it. (But I did order the Midwest anthology this past week largely because of this interesting discussion.)

I will *try* to say everything below starting from the minimal context of what I see as a Taylor-style analysis of deliberation; if I stray from that perspective too early in my remarks I’m sorry and it’s unintentional and please correct me.

Compare the influence of a Franfurt counterfactual intervener to that of Neil’s Superknower. Whereas the victim of a Frankfurt-style case does not know about any restriction the intervener could impose, Sally does know some specific kind of effect of the Superknower, though not specifically as it relates to one possible action.

Does the typical Frankfurt victim fulfill the conditions of BAT as cited above? Presumably. Any Frankfurt victim can think of alternatives of action, and with a belief about each that any such alternative is one that he/she is able to try to do. And so he/she believes he/she is able to try all of them at least disjunctively—but the actual effect of the intervener is to block all intervener-opposed actions in terms of opportunity for success. Should the Frankfurt victim try to do otherwise than what the intervener wishes, the opportunity will be blocked. In terms of conceived ability to try, the Frankfurt victim (before the actual effort of trying) fulfills BAT as a condition of particular belief about each alternative presented in thought (this is I take it the basis of the familiar “flicker of freedom” claim in terms of conceived ability to try to do alternative things). And now note—this is even true of all reflective Frankfurt victims who know that it is logically possible that interveners might affect their actions. After all, all that is required is that any Frankfurt victim does not know that he/she actually is such a victim even in light of reflective knowledge that this kind of situation is possible.

So in the typical Frankfurt case, BAT is fulfilled completely—even the actions sanctioned by the intervener are believed to be possible actions (as “try-able”) though they are actually impossible (intervener-blocked opportunities upon trying).

But in Sally’s case she knows much more. She knows prior to undertaking an action that some opportunity is blocked by the (predetermined) fixity of her not trying to do the action attached to that opportunity in the first place. She is like a reflective Frankfurt victim but in a stronger sense that she knows she will not even attempt (at least) some action, though she does not know what it is. Even so, prior to actually trying, Sally can attach a particular belief to each alternative of action that it is one that she can try, consistently with BAT.

Neil’s counterexample thus shows that Sally has a meta-belief about particular beliefs about abilities and opportunities that even a reflective Frankfurt victim cannot have. Since such a Frankfurt victim does not know that several alternative opportunities of fulfillment are blocked, he/she can plausibly believe that each of all conceivable alternatives of actions can be tried. Sally knows differently—all alternatives of action other than the one allowed are blocked not just by prohibition of ultimate accomplishment, but by the prohibition of the attempt itself. While this is not specific knowledge of which alternative(s) of trying is (are) prohibited, there is at least knowledge of some alternative of action that one cannot try at some specific time to do.

So how do Sally and a Frankfurt victim differ? In meta-belief about particular beliefs about abilities to try to do certain things. So is Neil’s example a counterexample to BAT? It depends on whether BAT is only about particular beliefs about particular abilities to try to act, or whether BAT includes meta-beliefs about those particular beliefs. Clearly Neil’s counterexample works in the latter interpretation; clearly it does not in the former interpretation.

So then the issue boils down to whether meta-belief about particular beliefs about abilities (and opportunities) is an essential part of the full concept of deliberation. If it is, then BAT as an indefeasible thesis about particular beliefs about abilities is only trivially true as it relates to the issue of deliberation. (Again, as Warfield seems to hint at above.)

[Mark Smeltzer]

Am I the only one who disagrees with Neil and thinks that Sally, as stated, isn't deliberating?

The strength of Neil's point seems to rest on the strength of the hearer's intuition that Sally is indeed deliberating (without further qualification). In analyzing how I react to this case, I try to put myself in the situation and become boggled down with the oddness of it all.

In the end, the only way I can imagine being able to actually deliberate about which door to open is to put the thought of the Superknower out of my mind. In order to make a decision, I have to pretend like I don't know about the expected outcome (since I don't have the luxury of being a Superknower, I can't simply forgo the process of choosing).

Perhaps this is why Fritz brought up the point about double mindedness? It seems to me that Sally is capable of deliberation insofar as her actions are going to issue from her own deliberative processes -- the Superknower, in this case, is not choosing for her.

Neil's case seems generalizable to any "prophecy" scenario -- even the self-fulfilling kind, and the even the kinds where the knowledge associated with the meta-beliefs are quite "soft".

For example, I know that I am going to wrap up this message and get back to work within a few minutes and that I am not going to do A1 ... An number of things during that interval, but that knowledge does not seem to militate against my ability to deliberate about which words to select, whether to check for typos, whether to preview the message first, etc.

Moreover, even though I *know* that I am the kind of person who is highly unlikely to check for misspelligs, grammar errors, or even whether I effectively made my points, that knowledge is not relevant to the question at hand: should I check for typos? Answers of the sort, "Well, we all know you're not going to" are irrelevant.

The same seems to be true of the meta-belief in Sally's case -- it seems entirely irrelevant to the choice Sally has to make (viz., the meta-belief that "I will open the door I decide to try" does not militate against the belief that "I can try to open either A or B". And I think that is why it seems like she is deliberating... if the meta-belief logically ruled out a particular alternative, there's no way she could be deliberating with respect to that alternative.

Regarding Neil's specific case, Neil seems to be trying to produce a counterexample by describing a situation in which "then S dispositionally believes of each of A1 …An that he is metaphysically free to perform it" is false, on a hard read of "metaphysically possible", yet we believe that S is deliberating. Neil seems to be advocating a revision similar to this:

If S tries to decide which of (mutually excluding) actions A1…An to perform, then S dispositionally believes that he is metaphysically free to perform A1 or …An.

This revision would have "really wierd" consequences if it proved true! However, I think the problem is not with Neil's case, but with his interpretation of BAT -- it just seems too hard.

My guess is that BAT was intended to read something more like the following,

If S tries to decide which of (mutually excluding actions A1…An to perform, then S dispositionally believes of each of A1…An that it is logically possible that he is metaphysically free to perform it.

Neil's case may be a counter-example to the "hard" read of "metaphysically free", but it clearly is not a counter-example to this softer read. (Which read is actually the one that PVI and/or Taylor prefer is question best left to the scholars.)

[Fritz]

Notice that if BAT is read with the modifier "logically possible" added to the consequent instead of in the way that we suggest (following, we assert, Taylor and PvI) then BAT requires that deliberators have the concept of "logical possibility". We doubt that all deliberators have this concept.

[Neil]

in evaluating BAT, it is the *Taylor* sense that is in play: I know of no one who has seriously asserted BAT or a BAT-like thesis for any other sense of deliberation

Well, if that's right, then I'm even more confused than I thought. In discussing her own BAT-like thesis(R) Nelkin considers a variety of such claims, from, among others Kant, Aristotle, Reid, Dennett, Kapitan, Clarke, Bok and Pettit. So far as I can see, *all* these philosophers defend a BAT-like thesis, and *none* of them assume Taylor's notion of deliberation. Indeed, it is *because* this is the case that BAT matters.

[Fritz]

As I understand Dana's interesting work on these issues, she, like Peter, finds most everyone discussing the issue to endorse the idea that deliberation carries with it "the sense of freedom". I think that this is right too -- others surely know more about the history than I do so my agreement isn't so important I suppose.
As we descend into the *details* of this "sense of freedom", what I think we find are many different senses of "deliberation" in play and many different understandings of the "sense of freedom" idea. We get a proliferation of specific views by mixing and matching various senses of deliberation with various specifications of what the "sense of freedom" comes to -- each combination of instances of these two variables could be used to generate a BAT-like thesis. We think (and, oh yes, argue a bit in our paper!) that the Taylor/PvI specification is the one that has the best chance of coming out right.
So that's why I take what I said and defend about BAT to be consistent with Dana's remarks (and Peter's, and others') about the widespread "shared" sense of freedom. Hope that clears up at least what in the world I'm talking about...

[V. Alan White]

Ok, so that was my penultimate comment. I will try not to make it the antepenultimate one.

After Sally opens the door, as she must, Sally can deduce from her previous meta-belief that the opposite door was one that she should not have BATtily believed she could have tried. No one else, including Frankfurt victims and except hard incompatibilists, could have deduced likewise, and in the case of the hard incompatibilist the premise would not be one that attached by special knowledge about ability to a single set of alternatives. There is a special epistemic status for Sally's meta-belief that revokes or withholds full force of belief for each of the conceived alternatives in a way that is "harder" than even that of the hard incompatibilist. Say that Sally's action X occurred and Y didn't. Then counterfactually Sally could say nothing--she could not even begin a relevant "If I had chosen Y. . ." because no very close possible world includes the trying to choose Y since the Superknower is included in them. But that's not true for a similar chooser in a hard incompatibilist scenario, for there are some close worlds where through tiny adjustments of cause the counterfactual antecedent is made plausible. Sally simply knows strongly that BAT beliefs for her given alternatives are more tentative than even hard incompatibilists would allow. If those BAT beliefs are thus "hardened" over what even hard incompatibilists can believe, is that not an indication that her meta-belief has produced this hardening? This is another way of seeing the force of Neil's example. . . Can Sally really believe in any intelligible metaphysical equality of her alternatives given her knowledge of the Superknower?

[Neil]

Alan,

You say that Sally has a meta-belief unavailable to an agent in a FSC. What is the content of that belief? Is it that she first-order believes 'I can open A' and "I can open B', but second-order believes that one of those beliefs is false? I don't think she has those first-order beliefs: I think she believes 'I can open either A or B', and 'I will open whichever door I choose'. I don't see what higher-order belief she has, with regard to those beliefs, than any other agent cannot have.

I think there is a mistake in the way you set out the case. You write:

She knows prior to undertaking an action that some opportunity is blocked by the (predetermined) fixity of her not trying to do the action attached to that opportunity in the first place

But Sally knows no such thing - not unless all believers in determinism who actually live in a deterministic world know it. Sally is not (compatibilist) preventing from trying anything; unfalsifiable prediction is not prevention (sayeth the compatibilist: recall, we are trying to be neutral between all sides here). The point of the lock is to make it the case that Sally is prevented from opening the door; and superknower's role is to allow her to deliberate despite this fact.

[Mark Smeltzer]

Fritz,

I think Neil's example does work against the hard read that is implied by a denial of the softer read I proposed.

Let F(S,X) stand for the sentence, "S dispositionally believes he is metaphysically free to X".

BAT can thus be stated as,

If S tries to decide which of (mutually excluding) actions A1…An to perform, then F(S,A1…An).

Let's say there are only two options, like in Neil's case, so BAT instantiates as:

If S tries to decide which of (mutually excluding) actions A or B to perform, then F(S,A) and F(S,B).

Neil's case demonstates that the conjunction of F(S,A) and F(S,B) is false! So, if this is the proper way to understand BAT, then Neil's counterexample succeeds.

That is why I suggested a softer read of F(S,X). I agree with your point about whether most agents would have a concept of logical possibility, and I think that condition is too strong as well -- but it is in the right vein.

If we read "metaphysically free" in F(S,X) in to mean that S simply believes that X is metaphysically possible, and not that S believes that X is categorically actualizable, then we have a way around Neil's case.

Let F*(S,X) stand for the sentence, "S dispositionally believes it is metaphysically possible for him to X".

Given F*, the conjunction of F*(S,A) and F*(S,B) is presumably true in Neil's case, and I think that accounts for why we get the intuition that Sally is deliberating.

(This is not something I've done much work on, and if I'm wasting time traversing well known territory, feel free to give me the smack down!)

[V. Alan White]

Thanks Neil, but now since my answer above is the antepenult, I've run out of cute Greek nomenclature monikers for any subsequent blogs. :-)

I'm obviously not expressing myself well. I take BAT as originally expressed to be a statement about each thought alternative of contemplated action to encompass first, a logical possibility of action as conceived in deliberation (that's how Sally thinks about the two doors as conceivable events that follow the supposition that she has chosen), and second, such a possibility then also conceived as being metaphysically possible--that for each such possibility she is free (has the ability and opportunity) to do it. So as/after she tries to deliberate each of these individual possibilities, she attaches a belief to each that she can do it (and thus try to do it--if she cannot believe that she can try to do anything, she cannot believe that she can accomplish it). She sees two doors; just on the face of it she believes that she can try to open door 1, and/or she can try to open door 2. She cannot form these beliefs if she believes that she is physically incapacitated, that the doors are both locked, etc. But she does not believe these things. So she believes she can try to open each of the doors. (Neil, I now see that these beliefs must attach to each alternative whether thought of as conjuncts or exclusive disjuncts--the additional idea of the exclusivity of choice is added in one way or another. With the exclusive "I can try to open D1 or D2" the relevant truth condition is included automatically as T/F, F/T but not T/T; with the conjunctive "I can open D1 and D2" it is added by "Oh, and not both D1 and D2".) That is what I take by BAT's claim to belief in metaphysical freedom to open the doors.

But that's consistent with an ordinary Sally who knows nothing about the state of the doors, an informed ordinary Sally who knows one door is locked, a FSC Sally who reflectively knows that one door is locked and that she can always be cancelled in her decisions by a potential intervener, but does not actually believe that obtains in this case, and even a Superknower-informed Sally, who knows that she will open only the unlocked door, but does not know which before she decides and tries to open it. Now none of the Sallies up to the latter case has her first-order BAT beliefs about her freedom to try to open each door significantly modified--there is no reason to believe otherwise. But Superknower-informed Sally does believe otherwise. She knows that one of her BAT beliefs about her ability to try to open a door is false--when she does finally try to open one, she will know that she could not in reality have tried to open the other (otherwise she would not have proper Superknower information). And that knowledge about her BAT beliefs can only be called a meta-belief.

[Neil]

when she does finally try to open one, she will know that she could not in reality have tried to open the other

Shouldn't a compatibilist deny that? Certainly a compatibilist about deny foreknowledge and freedom? So far as I can see, nothing about the situation is relevant to the semantics of 'try' which doesn't hold in a determinstic world.

[V. Alan White]

But isn't that (what I said) just a consequence of the unusual content of Sally knowledge? She presumably knows nothing about how the Superknower's prediction comes about (determinism, mystical fatalism, superknowledge of indeterministic occurrences. . .)--just that, given the infallibility of the prediction, nothing else but opening the unlocked door can occur. The Superknower's role here is special in that it metaphysically "locks down" what Sally can do in all possible worlds where that role obtains. No mere deterministic cause can do anything like that.

[Neil]

Consider Superknower*. She differs from SK in that her predictions are only very, very good. The set-up is otherwise the same: one door is locked. Given that one door is locked, Sally is not metaphysically free to open either door. But Sally knows that SK* is a very, very good predictor of human actions, and therefore knows that there is a very high probability that the door she selects will in fact be the door that is unlocked. Does this example lead you to share my intuition that Sally (compatibilist) can try either door?