Blog Coordinator

« Libet and No-Choice Closure | Main | Joe's Taxonomy: Why Can't We All Get Along »



Feed You can follow this conversation by subscribing to the comment feed for this post.

The first thing I want to say is that you're really showing up all of your predecessors as featured bloggers. We all manage to post every week or so. Everyone one of your posts is at least as interesting as any of mine (not to pick on anyone else ), but they come so think and fast!

I think this is one is the coolest yet. I want to wait until I see the full argument before thinking it through fully. Just a couple of points come to mind. First, epistemological skepticism usually targets continent propositions, so it's not clear that a priori claims should fall within its scope. Second, it's true that that worries about luck play an important role in epistemology but not in arguments for skepticism. Rather their home is in the Gettier debate, and therefore the analysis of knowledge. Certainly luck as I analyse it - an analysis heavily indebted to Pritchard's *Epistemic Luck* - concerns the difference between the actual world and nearby possible worlds! rather than between the actual world and skeptical scenarios (which are very distance from the actual world).


I’m unsure if premise #1 is needed, but I like your argument and I believe your conclusion is true.

It’s difficult to prove that X doesn’t exist, especially if X is related to a human reference issue (i.e., the inability to quantify and measure new emergent forces before those new forces add together with other preexisting forces and form a new net sum).

FW skepticism is one thing, but to *prove* FW doesn’t exist is something altogether different. The real challenge as I see it, is to develop arguments that prove FW does exist.

Seeing the different parallels laid out this way, they are genuinely suggestive. (By the way, I would have been more easily persuaded if you'd just introduced your approach with this post, but I am far from the best person to listen to about matters of presentation.)

In response to your closing question on the other thread, if the epistemological versions are less persuasive, it is just because the premises are less appealing for non-structural reasons. E.g. for argument (I), you need one premise in addition to the closure/transfer principle. In the epistemological version, this is something like "I don't know I'm not in a skeptical scenario". In the MR version, it is something like "I'm not responsible for the laws or the distant past". I think that the former is more easily rejected (and the Moorean inference from "I know I have hands" to "I know I'm not in a skeptical scenario is easier to swallow than an inference from "I am responsible for posting this letter" to "I am responsible for the laws and distant past", not that that means either is easy to swallow). For argument (II) (which, I note, contains no reference to incompatibilism, so I have no problem with it), the epistemological version has the usual three responses of foundationalism, coherentism and infinitism, or whatever you call them; for the MR version, the linear nature of time rules out a response parallel to coherentism, and human finitude rules out one parallel to infinitism, so only a parallel to foundationalism, i.e. a claim that there are some events for which I can be responsible even if they aren't the results of others for which I am responsible, is available. Strawson's initial assumption is essentially just that this MR parallel of foundationalism is false. I'm tempted to foundationalism as a response to both forms of the argument, but I suspect that the MR version is less appealing on average, to both philosophers and laypeople.

Do you think there are viable MR parallels to responses to the epistemological version of argument (I)? E.g. if we think of contextualism about knowledge as a response to argument (I), is there an MR parallel to contextualism (presumably, that the operator 'could do something to make false', or whatever the experts are using these days, is not invariant in meaning across the premises) that could be made to work?

I also wonder if there is a parallel to be made with respect to evolutionary arguments against the possibility of moral knowledge in particular. If the epistemological version is something (very crudely) like 'we evolved to be disposed to certain moral beliefs for pragmatic reasons, so we can't expect them to track the truth', I think the MR version should be something like 'we evolved to be disposed to certain moral behaviours for pragmatic reasons, so we can't expect them to be reasons-responsive', which is vaguely Prinz-like.

Hi Joe, I've often thought about these parallels (I think inspired by discussions with you or your past posts!). Personally, I love Chalmers' piece "Matrix as Metaphysics," from which I draw the conclusion that this claim is false: "If you knew that you had a hand, then you could rule out that you were a BIV, for BIVs don’t have hands." I think BIVs (and persons in a Matrix) do have hands and know it just as well as we do; their hands are composed of different foundational components as ours (e.g., programming language rather than quarks or whatever--it's not as if we know what our hands are ultimately composed of). Notice that if Neo comes out of the Matrix as a blog of ectoplasm, he's not going to conclude, "Oh, so this is reality and this is the *real* me and now I see that I never actually had hands." He's going to say, "Put me back in reality! Give me my hands back!"

This sort of move then suggests a similar anti-skeptical move about free will. We know we have free will just as much as we know we have hands--where free will is the sort of control over actions that make us causal difference-makers and morally responsible agents--but we don't know "what our free will is made of." It could be a product of our complex brains, or perhaps a non-physical soul, or even a complex computer program in a Matrix. It could be instantiated in a deterministic system or an indeterministic one, etc.

Lest someone object that this move is too easy and makes us have free will necessarily, there are satisfaction conditions, just as there are for our (or BIV's) knowledge about hands (or colors or dogs or reasons). I will leave it as an exercise for the reader to consider what they are.

Our mutual former student, Jason Turner (soon to be back stateside at SLU), has a great paper that I read as making similar sorts of moves (though I'm not sure he agrees), "(Metasemantically) Securing Free Will", which can be found here, along with a great comic book cover of the paper:

Just a couple points, Joe.

Something to be mindful of when drawing this parallel is that skeptical arguments can be ambitious/global or localized. The skeptical cases you point to are instances of the former. Now there can be instances of the latter that even someone with Moorean sympathies may find compelling enough to withhold judgment about. (Ex. I know I am driving through fake barn country. Is that a real barn? All I know is that it might be.) One might think that skeptical arguments about free will are more localized, or that they can be cast that way.

There can be skeptical arguments for the claim that we do not have free will, and there can be arguments for the claim that we do not know we have free will but which do not depend on making the former claim. Kearns has an argument like this (well, the end of his Nous paper suggests there is a collection of such arguments).


I share Neil's concern about the distinction between a priori and contingent truths. I understand how I might not have a hand, if I am a brain-in-a-vat (Eddy outlines an interesting counter-argument). But I don't understand how 2+2 might not equal 4. Or that the law of non-contradiction might be false. Those things are always true, as far as I understand, whether I am a BIV or not.

This brings us back to free will skepticism. As I understand it, G. Strawson bases his argument on a priori truths, and not contingent ones. His argument is basically that A. free will requires self-creation (by definition, he alleges), and B. self-creation is impossible. For (B), self-creation is logically impossible, always, forever. Self-creation is impossible whether we are normal human beings, or brains-in-vats. Self-creation is impossible in the same way that 2+2 = 4.

Not all skeptics agree with this point. Notably, Pereboom grants that free will is logically possible, but simply not actual. But, as much as I love Pereboom, and as attractive at it might be to give free will a chance (so to speak), Strawson and others have persuaded me that this view is mistaken. Free will never had a chance. I think what ultimately motivates both Strawson and Pereboom (and Levy, or earlier Sommers), is constitutive luck, which is just a concern that self-creation is impossible. It's always impossible, even if we are brains-in-vats.

Interesting stuff. For what it is worth, I also explored the parallels between the arguments for epistemological skepticism (employing the closure principle) and the Consequence Argument in my 1994 book, The Metaphysics of Free Will. I just thought I'd mention it because it might be of interest to you or other readers.

Also, Garrett Pendergraft's very fine dissertation at UCR explored these issues insightfully.


I agree that self-creation is impossible, but I don’t believe free will requires self-creation. I’m thinking that free will (i.e., new emergent forces) may be associated with a type of creation that isn’t self-creation. FW may be associated with a new higher-level life form that’s created by lower-level life, wherein the resulting higher-level life isn’t predeterministic in nature. It’s simply new life. Here’s an example: A large set of lower-level living entities (e.g., billions of neurons within your brain) cause a higher-level entity to emerge (i.e., a thought), while at the same time those neurons aren’t the sole source of the control when that thought interacts with another thought. We *know* that must be true, because the intelligence associated with the interaction of our thoughts isn’t innate to the laws of physics (sorry to repeat myself, but that’s a crucial point). The intelligence is a new emergent property, and it exerts forces which aren’t predetermined.

Does that line of reasoning make sense, and isn’t it therefore reasonable to believe that FW doesn’t require self-creation?

Neil: There are skeptical arguments against a priori claims. In the first Meditation, Descartes suggests that while mathematical claims escape the worries of the dream argument they don't escape the worries of the evil genius argument. Also, Hume gives rather extended skeptical criticisms of reason in general in the Treatise. And you can find similar criticisms, of course, in Sextus Empiricus.

I seem to remember arguments for epistemological skepticism based on luck but I can't recall any authors. Let me keep looking. I did find this rather interesting quote from Mylan Engle:

"Epistemic luck, then, is ubiquitous and unavoidable. If all forms of epistemic luck are incompatible with knowledge, as the incompatibility thesis maintains, skepticism is correct and the knowledge thesis is false. And yet, we remain convinced that we possess lots of knowledge. The task facing the anti-skeptical epistemologist is to reconcile the rather strong intuition that epistemic luck is not compatible with knowledge with the equally evident observation that it must be."

Luck isn't just discussed in Gettier. It was discussed by Plato, as well. And used to rule out the claim that knowledge is just true belief. Historically, something like the incompatibility thesis was accepted and this was one measuring stick against which adequate analyses were measured.

James: I don't think you can PROVE that we have free will. I think the best you can do is show that there is no reason to doubt we have free will. That's what I'm trying to do at least.

CJ: Interesting points. I can't address all of them now but I certainly do think that the parallels between the various skeptical arguments gives us parallel potential solutions. Thus, you could be a contextualist about "freedom" or "moral responsibility." You could adopt something like the relevant alternatives view (but it would be called a relevant facts view). Both of these possible solutions are discussed in more detail my "Compatibilist Alternatives" (Canadian JP, 2005).

There are also comparable potential solutions to Strawson's argument, once you compare it to the ancient trilemma (which I'll talk about in more detail tomorrow morning). For instance, agent causal views are comparable to foundationalism; Lehrer discusses the possibility of infinite causal chains that go back to some specific point of time, which might allow for emergent actions ("'Can' in Theory and in Practice: A Possible Worlds Analysis" (1976)).

I gave a talk recently at the University of North Florida and Paul Carelli suggested that my second Adam example ("Incompatibilism and Fatalism: Reply to Loss," Analysis 70.1, 2010) is similar to a kind of coherentist response to the ancient trilemma. I'll have more to say tomorrow and Friday.

Thanks John!

Great post, Eddy! I love the Chalmers piece too and I like the application to the problem of free will skepticism. I lean toward a kind of commons sense reply in both the free will and knowledge skeptical problems but this puts more oomph in that kind of approach. I'll have to think about this more.

Thanks for the Turner reference. I talked with Jason a bit about this once but never read the paper.

Thanks James (Gibson). Interesting points. I'll have to think more about them. Do you have a title for the Kearns paper?

Kip: I like James Laird's response! Still, I'm not sure that Strawson's argument is as strong as you think it is. More to the point, I don't see how it is more powerful than the ancient trilemma. On the face of it, it seems like knowledge (or justification) is impossible given that there are only three ways in which a justificatory story could go and none of them seem plausible. I don't understand why the Strawson argument is more compelling than this. Yet in the end we dismiss the epistemic skeptic rather readily.

Here is another interesting observation. Our conception of knowledge has changed over history from an infallibilist concept to a fallible one (and perhaps from an internalist one to an externalist one). We seem to be willing to concede that perhaps infallible internalist knowledge is impossible but that doesn't mean we don't have anything. Why haven't we made a similar transition in the case of free will? We could look at the Strawson argument and conclude that it shows that a kind of unbelievable God-like freedom -- self-creation -- is impossible. Thus, whatever free will is it isn't the God-like power we might have originally thought that it was. One might think of compatibilism as analogous to fallibilism. That seems to be what James is doing.

CJ, your analysis of argument (II) in terms of foundationalism, coherentism and infinitism is awesome! But it's also wrong, I think. There *is* a parallel to coherentism available for FW, precisely because of the linear - rather than vectorial - nature of time.

Remember from geometry that a line extends in both directions, while a vector extends in one (it only has one arrow on it). In most historically important deterministic physical theories, the logical implications (from a description of one time-slice plus the laws of nature) go in both directions. It is just as easy to compute the present from a specification of the future, as from a specification of the past. This logical bi-directionality means that neither past nor future can be labeled "master" and the other "slave". Rather, they are simply interdependent. Much like the justificational relationships among beliefs in coherentism.

Carl Hoefer explains all this much better than I could in his SEP entry on Causal Determinism - - especially sections 2.3 and 6.

Joe and James:

Yes, I agree that it is reasonable to think free will does not require self-creation. I also think that it's reasonable to think that free will does require self-creation. Basically, I believe that the definition of free will is under-defined and that people (including me) waste a lot of time arguing about it, without using a clear definition. This was essentially Double's view.

Mathematicians would never make this kind of mistake - in fact, it's fascinating to contrast how mathematicians handle problems with under-defined terms, with how philosophers handle them. You can find classic math problems that have the solution "unsolvable, because key term is not fully defined." Does that ever happen in philosophy? After thousands and thousands of years of arguing?

I'm still not convinced about arguments for skepticism about self-creation being impossible.

For Joe: couldn't the impossibilist just backpedal to this claim:

Given that the world is not a giant illusion, and is more or less as it seems, then we don't have free will? I think most impossibilists would not consider this to be much of a concession at all, considering that we take the antecedent for granted in our day-to-day lives. I don't think impossibilists have a huge vested interest in proving that free will doesn't exist even if an evil genius is distorting my entire concept of reality.

There are at least two forms of FW skepticism available basically. One is exemplified by hard-incompatibilism based on opposition to some stable concept of controlled indeterministic choice that is untrue by virtue of the claimed truth of determinism, and mirrors coherent atheism in that any atheist should be able to state what a concept of god is that does not in fact exist. No such FW skeptic thus can claim ground to that kind of skepticism unless she can describe the concept that she is existentially skeptical about. The law of excluded middle only applies to intelligible concepts after all, and so whereas the claim that minds are either deterministic or indeterministic is true, the claim that minds are geshesmerfberg (conceptually vacuous term) or non-geshesmerfberg is senseless. The hard incompatibilist skeptic knows perfectly well (enough) what she is skeptical about, and needs to know that for the pose of oppositional skepticism.

But another form of skepticism is about the coherence of a proposed concept itself, and thus may claim that something supposedly answering to the phrase "free will" turns out to be (on analysis) another case of geshesmerfberg. Now while Richard Double's position is nuanced and complex metaphilosophically and epistemically (especially as regards the subjectivism of values as that relates to FW metaphysical issues), ultimately Double's view is that "free will" is a form of non-sense, non-objective geshesmerfberg. There is no stable concept answering to the term "free will".

Thus there is existential skepticism of a stable concept, and skepticism that a proposed concept is itself stable.

Not exactly Double's point--but an attempt to capture his spirit of argument.

Again--great posts Joe!

I am reluctant to mention quantum mechanics and free will in the same sentence, but...there is a relationship between free will skepticism and epistemological skepticism in the literature on Bell's inequality (Conway and Kocken, Renner and Colbeck, Lloyd), which it seems to me even extends to the classical regime. That is, if an experimenter cannot assume her freedom of choice, she can never be sure of the interpretation of an experiment. In Bell's inequality experiments, for example, the options are "super-determinism" and local causality, or quantum weirdness/nonlocality.

Sure here are skeptical arguments targeting a priorir claims. My point is only that it is harder to get such arguments off the ground: the parity claim in premise 1 might therefore be more plausible with regard to the former than the latter. Not everyone who says they are talking about luck are actually talking about luck: more commonly, when they are talking about luck they are gesturing vaguely at as mess of events and there is luck thereabouts, but it's not doing the work they want it to (something else is). It is to avoid that they I think we need an account of luck.

David: That is an interesting point. Here is a related point. In an attempt to motivate time travel paradoxes, David Deutsch and Michael Lockwood write the following:

"The real core of the grandfather paradox is not the violation of free will but of a fundamental principle that underlies both science and everyday reasoning; we call this the autonomy principle. According to this principle, it is possible to create in our immediate environment any configuration of matter that the laws of physics permit locally, without reference to what the rest of the universe may be doing. When we strike a match, we do not have to worry that we might be thwarted because the configuration of the planets, say, might be inconsistent with the match being lit. Autonomy is a logical property that is highly desirable for the laws of physics to possess. For it underpins all experimental science: we typically take for granted that we can set up our apparatus in any configuration allowed by physical law and that the rest of the universe will take care of itself." (“The Quantum Physics of Time Travel, Scientific American, March 1994, p. 71)

As Deutsch and Lockwood note earlier, classical free will skepticism (no one is or ever was able to do otherwise) seems to undermine the autonomy principle. This is another way in which free will skepticism might have an impact on epistemological skepticism.

Kip: But here is my point: There is a kind of dismissive attitude in your reply by the impossibilist to the problem of skepticism. Yet (as I hope to show further in a post later in the day) there are important similarities between say Strawson's argument for impossibilism and the ancient trilemma. These arguments show that the appropriate kind of grounding is impossible since, once we exhaust the possible ways in which such a grounding might hold, we realize that none of them is feasible. Thus, what is allegedly grounded is also impossible. Where they differ is in what is allegedly grounded, whether it is free action or knowledge.

I'm not suggesting that thus far I've shown more than a superficial comparison between skeptical arguments. On the other hand, the comparison is strong enough that anyone promoting a Strawson-type argument for free will skepticism should be able to at least provide an explanation for the differences between the two arguments.

Alternatively, if illusionists can say "I'm not really worried about your (epistemically) skeptical possibilities" why can't I similarly dismiss their argument? Saying that you're not worried about wild epistemically skeptical hypotheses is not really responding to the paradox; it is merely dismissing it. But if skeptical paradoxes don't worry you when the subject is knowledge, why should they be more worrisome when we change the subject to free will?

Al: I might have misunderstood Double's argument, so let me think about this further and get back in week 3 when I've had a chance to look closer at the argument. If he thinks that "free will" is nonsense, then clearly his argument is much different than, say, Unger's. I thought the claim was more that free will contains a bundle of slightly related concepts that, when taken together, are inconsistent. In any event, thanks this is helpful.

Neil: It seems that the concept of luck is not much better than the concept of free will in terms of our understanding and agreement of the concept! Still, I appreciate the point that we need to get clearer on the concept of luck and it might be that once we flesh the concept out that these differences suggest comparable differences between the two kinds of skeptical problems. In any event, I'm glad and you and others are looking into this more closely! This is another case in which I might have to read more and report back during week 3.

It is not for nothing that Descartes established the existence of God before proving the existence of the external world. For unless one has a good reason for thinking that one has been created by a benevolent being, one is hard pressed to defend the notion that one is not a BIV or PEG (pawn of the evil genius). So, yeah, I can see FWS and ES at least standing together. How am I so sure that I'm not a BIV? God would not allow such a thing. Ditto for me being a PEG. But there is an additional, philosophical, cogito related reason for rejecting FWS, should my theology appear unconvincing. When I assert the existence of freedom of choice, I am speaking of myself, my own mind, about which cannot err, unlike the external world. Could FW be an illusion? I would have to convince myself of it, for FWS is certainly not obvious, nay it violates my own sense of myself. But then I would have to exercise my intellect, perform various mental acts, direct my thoughts, in a word, make free choices. (One is not thinking rationally if one is merely the passive recipient of a series of even the most intricately woven ideas.) Like the notion that one does not exist, the attempt to prove FWS must end in failure, proving its antithesis. Reflectively one would see oneself doing the very thing one had set out to prove was impossible.


I agree that people frequently use different meanings for the term “free will”, and those different meanings cause lots of miscommunication during debates regarding the existence of FW. Perhaps one of the FOF contributors will open a thread on that subject sometime, so we could cooperatively develop standard meanings for use here on FOF.

In your post above (at 7:06 PM on 11/4/13), you appear to be discounting the comment that I posted at 2:42 PM on 11/4/13, because you believe I’m using a “weak sense” of the term FW. I believe I’m using the strongest possible sense for the term FW in my comment, which is FW: the existence of a new emergent force within a physical human brain that doesn’t result solely from a direct sum of preexisting forces, thereby allowing a person to make a decision that isn’t controlled *solely* by the laws of physics.

I would think that definition of FW would be satisfactory for even the most hard-core incompatibilists.


I understand that, mathematically, the physical facts at any time can determine those in the past and future. But we need more than just this kind of mutual determination for the MR-analogue of coherentism. We'd want two acts x and y such that I am responsible for x because [I was responsible for y and y brought about x], and also I am responsible for y because [I was responsible for x and x brought about y]. (The epistemological analogue is two beliefs x and y such that x is justified because [y is justified and x is supported by y], but y is justified because [x is justified and y is supported by x].) It seems to me that the relevant sense of 'brings about' has to be more than 'determines according to the laws of physics'; it probably has to mean 'causes', and causation runs in one direction.

Having said that, I remember reading an article that argued that the consequence argument fails because, if present time slices really do mathematically determine past ones, then we can be responsible for events in the past if we are responsible for events in the present. For all I know, you actually wrote it, but it seems a bit of a stretch to me.

(NB: I think a better MR-analogue for coherentism might be something like 'x and y are contexts for action such that I am responsible for what I do in x because [I am responsible in y and act in y to bring about x], but also vice versa'. I.e. the variables range over contexts for action rather than acts or events. But it doesn't improve the plausibility of coherentism for MR at all.)


The epistemic coherentism you're thinking of has a more demanding conception of "coherence" than the one I'm thinking of. For example, "x is justified because [y is justified and x is supported by y]" could be replaced with "x is justified because x is supported by y".

You may also be asking for a stronger analogy between the two coherentisms than the one I'm offering. In epistemic coherentism, most (but not all!) philosophers feel that beliefs need some positive merit, else they must fall. The support relation provides that merit, in a grand and virtuous circle. In MR coherentism as I conceive it, putatively free action already has some positive merit, like moderate reasons-responsiveness. But this merit is threatened by a potential negative: an act's being caused by something outside your control. Normally we think of causes as masters and effects as slaves, which makes the thought of being caused a troubling one. The bi-directionality of determinism blocks the threat by breaking this habitual thought pattern and taking the sting out of being caused. Perhaps my "coherentism" is best compared to those few philosophers who deny that beliefs must fall, as if beset by gravity, unless supported. Or perhaps, given the role of reasons responsiveness, it's like a hybrid of foundationalism and coherentism.


I did not mean to sound too dismissive. If I do, I think it's because I *agree* with you. Let me explain a little.

First of all, I agree that there are essential and important similarities between arguments for knowledge-skepticism and arguments for free will skepticism.

Because I feel the force of the knowledge skeptical argument, I am willing (or interested) in conceding ground, to save what is left of the free will argument, assuming that the knowledge argument works. That's why I proposed a compromise conclusion: P. given that the world is essentially the way it seems to be (absent massive deception), then we lack free will. My compromise thesis P is essentially an agreement with your argument that the knowledge and fw will arguments conflict. I'm not sure what more you might expect me to say? (I'm still unsure whether massive deception could undermine my knowledge of logical truths, but I'll put that to the side for now.)

The comments to this entry are closed.