Blog Coordinator

« What It's Like to Be a Mele | Main | Call for Papers: Oxford Studies in Experimental Philosophy »

07/05/2015

Comments

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

Back from the Joint Session and I see that an awful lot has been happening on the blog in my absence! Thanks, everyone, for contributing such terrific comments. Once again, I find myself a bit spoilt for choice about what precisely to respond to.

Maybe it's time for me to say something about the Consequence Argument. It's true (as John points out) that ED on its own will produce certain versions of the Consequence Argument - PvI's original version, and presumably other close relatives. But the original version seems vulnerable to the well-known Lewisian move, if we're only prepared to be Lewisian about our laws. The laws *aren't* things no one has any choice about (on the Lewisian view), because they are merely descriptive and so what they are will depend on what we choose (and not the other way around). I think this is what Justin said. On that interpretation, then, a crucial premise of the original argument is false. To get a version of the CA that does the trick for the incompatibilist, and establishes that we have no choice about anything if determinism is true, it looks like we need either something stronger than ED, which builds in the governing conception of law somehow (Justin's 'I've got to be missing something' suggestion, above, I think), hence moving to something a bit more like my original MD; or else one's got to hope that there's something about the actual laws that in fact exist that will take us beyond what ED alone is able to establish (this, I take it, is Paul's suggestion).

I have to confess that I've not sat down properly to figure out what exactly I think about the Pruss proof. But my inclination is to think that the Pruss interpretation of 'N' effectively shifts us over to a non-Lewisian conception of the laws and past according to which it follows that we are *constrained* by them (something I think does not follow from ED on the Lewisian view). So it's no accident that the Consequence Argument looks to be in business again, once we offer this sort of interpretation of 'N'.

But Joe's question about what the source of the constraint might then be is a good one. Before we can say whether the 'oomphiness' of causation might be involved, I guess we need to be a bit clearer about how 'oomphy' causation might relate to necessitating laws - and indeed what on earth oomphy causation and necessitating laws might be! One question is whether it might be legitimate just to suggest that there is a very basic notion at work in our ordinary thought about causation, law, etc. according to which we allow that certain things are just *made to happen* by others(though that way of putting it seems to encode a past to future directionality which perhaps we'd want to eschew - on the other hand, perhaps not, given that we're trying to encode fairly non-technical worries here). My own feeling is that such a notion is indeed part of what gives rise to the ancient worries about the compatibility of liberty and necessity - and that this notion just vanishes from the picture when all we have to encode the thesis of determinism is ED.

Joe,

The reason we can believe that “things aren’t settled or fixed in advance by things over which we have no control”, is because there’s something more than simply the fundamental forces of physics controlling what happens in reality. Here’s why I say that, and I welcome constructive criticism:

I believe that it’s reasonable to say that the interaction between any two neurons in a physical brain isn’t one and the same entity as a thought, just as two individual pieces of a 1000-piece picture puzzle that are initially put together aren’t one and the same entity as the emergent picture resulting from the completed 1000-piece puzzle.

With that in mind, a thought may be modeled as billions of neurons interacting with one another in a coordinated wavelike manner. So if we use that as the definition of a “thought”, I claim the following is true: Since I experience my thoughts interacting with one another (i.e., control is exerted from the neural wave level), and since said control results in *physical* changes within my brain during said interaction, and since forces are required in order for physical changes to occur, I therefore believe my thoughts exert new emergent forces.

If a person believes that my definition of a “thought” (the model I mentioned above) is reasonable, but they deny that their thoughts interact with one another, (i.e., they simply believe that all of the action within their brain is controlled solely from the four fundamental forces of physics (4FFOP) in a bottom-up predeterministic manner), they’re believing in something which contradicts their daily experience.

Okay, it seems to me that science generally denies that thoughts emerge at the neural wave level, and instead science believes that a thought is simply a direct sum of all the individual neural actions taken together as a composite. In other words, science believes that waves of neural activity are associated with thoughts, but science doesn’t believe that any new control emerges at the neural wave level. By holding that belief, scientists run into an issue: Where does the intelligence originate from that’s associated with the interaction of our thoughts? Is the intelligence somehow directly sourced from the 4FFOP (e.g., gravity)? I don’t think science is making that assertion. Instead, science claims that human intelligence is sourced solely from the current neural net wiring configuration of a brain. But that claim leads to a different conundrum: How do people learn – how do they acquire said intelligence? In other words, if the 4FFOP are controlling all of the physical changes that occur within the neural nets of a brain, then where does the intelligence come from that’s associated with those changes?

If we believe that our thoughts exert physical forces which cause our neural wiring to change, thereby enabling humans to learn, we eliminate a long-standing conundrum and we develop a basis for ambitious free will. In summary, the fact that humans are capable of learning from their thinking supports the hypothesis that thoughts exert new emergent forces, and it supports the idea that “everything isn’t settled or fixed in advance by things over which we have no control”.

Thanks very much, Helen.
I'm not convinced that one needs more that ED to generate at least a plausible version of the CA. I myself find the CA convincing, but also I do not find it a knockdown argument; reasonable people can (and do) disagree. Recall that even Lewis thinks that the past has to be different from what it actually was, in the worlds in which the agent does otherwise; that is, even on his approach, a deviation from an actual natural law is not all that is required; some change in the near past is also required. And so, no matter what your view of laws, if you find it plausible that the past is fixed, then the CA is still plausible, even given ED. And I believe that if one accepts the fixity of the remote past, one should also accept the fixity of the near past.

To clarify. I meant that ED is all in the way of a definition of causal determinism that's required. Of course, one needs other ingredients expressing the fixity of the past and natural laws, and perhaps a modal transfer principle (but that is contentious).

My point is that even on Lewis's approach presented in "Are We Free to Break the Laws?", *some* deviation from the actual past is required. And I think that if one believes that the remote past is fixed, one should also believe that (say) yesterday is fixed.

Helen,

I think you're right that "the Pruss interpretation of 'N' effectively shifts us over to a non-Lewisian conception of the laws and past." But it's even stronger than that. It shifts us over, I suggest, to an asymmetric, hence non-scientific conception of the laws and past. There's something about the actual laws that, with or without ED, leads to a natural compatibilism, albeit, in the without-ED case, not by Lewis's route.

I don't get the connection between MD and Pruss' (2013) interpretation of 'N,' which is (as Pruss notes) Humer's (2000) “counterfactual sufficiency” interpretation of 'N.'

The formal definition of 'N' is given in Pruss (2013, 433), so I'll spare those details. Here is Pruss’ argument for N(L & P), where P is “a statement of the complete state of the world in the pre-human past” (430): "it is implausible that there is something we can do such that were we to do it, the conjunction of laws and pre-human past wouldn’t hold" (433). I guess I don't see how the difference between MD and ED matter to Pruss' results.

OK – so to respond to some of this. Firstly, and to raise an issue I’ve not mentioned before in this thread, I agree with the spirit, if not the letter, of James’s remark that there’s something more than the fundamental forces of physics controlling what’s happening in reality. The reason I can’t agree with the letter is that I don’t think it makes sense to think of the fundamental forces of physics as things which control anything at all! Fundamental forces *constrain* what happens, but that’s not the same thing as control. (Literal) controlling (as I see it) has to be done by things of which it makes sense to say that they might be controllers – and that means: persons, animals, certain sorts of machines, perhaps certain sorts of computer programmes that are hooked up in the right way to other systems. I suppose that one might allow that the values of a variable on which the values of certain other variables depend might be in an extended sense a ‘controller’ of those variables – and insofar as the magnitude of a force can be represented by such a variable, then perhaps forces could be thought of as controllers in this sense? But in view of the fact that we’re already much too eager to think of forces as quasi-agents, I think it’s a rather dangerous usage, probably better avoided.
But James’s idea that free will requires that we recognise some sort of top-down, or emergent causation seems right to me. The whole animal needs powers over its parts, as I’d put it – if everything that happens at the psychological level is just a consequence of things happening in or to the small parts of the body, as a result of an entirely bottom-up form of determination, then I don’t see how we (and our psychology) could fail to amount to mere epiphenomena. The final chapter of A Metaphysics for Freedom tries to explain how a place might be found for top-down causation within a basically naturalistic (though not physicalistic) picture.
The other thing I’d like to comment on is John’s suggestion that as long as we believe in the fixity of the past, we won’t find Lewis’s response to the Consequence Argument plausible, even if it isn’t completely watertight. But I’m not sure about this. I’ve always been convinced by what Helen Beebee says about this in her ‘Reply to Huemer on the Consequence Argument’ (Phil Review 2002). Beebee notes that Huemer focuses on defending the claim that no matter what S does, she will be unable to falsify any statement about the distant past, as though *that* were the premise of the Consequence Argument that Lewis means to deny. But it isn’t the premise that Lewis means to deny; as he makes clear in ‘Are we Free to Break the Laws?’, the premise he means to deny (on one of its salient interpretations, anyway) is the premise that no matter what S does, she’ll be unable to falsify any law of nature. For Lewis claims that there are some acts that S is able to perform such that, were she to perform any of them, L would be false. It’s the law S can falsify, not the past.
How does this square, though, one might ask (and perhaps John implicitly asks) with Lewis’s insistence that for S to perform such a law-falsifying act, a ‘local miracle’ would have to occur prior to that act? Well, it’s true that Lewis does indeed claim that I am (now) able to perform some act A such that if I were to perform A (when in fact I don’t do so), the past *relative to the time of the act* would be different – a miracle will have to occur prior to the time at which A occurs. But that doesn’t entail that I’m now able to perform any act A such that if I were to perform A the past relative to *now* would be different. Here’s what Helen says, and I think she’s right:
“To see the distinction clearly, we need to get our time-indexing straight. We can distinguish two theses, each of which entails (WT) [ = I am able to do something such that, were I to do it, a law of nature would be broken] – the thesis to which Lewis explicitly subscribes:
(WT1) For some times t1 and t2, where t2 ≥ t1, and for some non-actual act A: I am able, at t1, to do A at t2, and, were I to do A at t2, a law would have been broken prior to t2 (and hence the past-relative to-t2 would have been different).
(WT2) For some times, t1 and t2, where t2 ≥ t1, and for some non-actual act A: I am able, at t1, to do A at t2, and, were I to do A at t2, a law would have been broken prior to t1 (and hence the past relative-to-t1 would have been different).
Lewis holds (WT1) … But nothing he says suggests that he holds (WT2)”.
So in my view, even if one is totally convinced of the fixity of the past (and I am!), this can’t be a reason to reject Lewis’s move. There might, of course, be other reasons, but I don't think the fixity of the past is one of them. This is why (in my view) it all ends up turning on questions about laws. More to say on this, but this post is already long enough!

Helen,

Perhaps the following is a reasonable way of looking at “control”.

Whenever an entity determines an action that occurs in reality, the entity may be thought of as exerting control. There are many different kinds of entities which determine actions, and therefore, there are many different kinds of entities which exert control.

Since there are many different kinds of entities which exert control, reality ends up being comprised of a sea full of different controllers, all of which end up having their control add together to form a “net sum of control” if you will, which thereby determines the overall path forward. In that sense, “determinism” is true.

The main idea that I think science has been missing, is that control is an emergent property, and control isn’t determined solely from the bottom-up by the fundamental forces postulated by physics. If we believe that control is an emergent property, and we believe that said control is caused by subcomponents while at the same time said control isn’t determined solely by said subcomponents (i.e., emergent life is fundamentally indeterministic in nature), then it becomes reasonable to believe that the “animal system level” exerts control that isn’t predetermined (i.e., animals are alive and free).

Predeterminism is false, while determinism is true.

It’s nice to read that you’re thinking about “Persons as Animals” (stated in your initial post). There’s a whole spectrum of life that’s relevant.

Hi Helen--

I love your close analysis of the claims about fixity of the past--thanks for that. Now I can't claim I'm any expert here, but I have a few remarks about Lewis, laws, and fixity of the past.

Caveat: much of what I'm going to say is based on BeeBee's and Mele's analysis of Lewis through the lens of their Humean Compatibilism paper. (Joe will no doubt be smiling broadly throughout all of this.)

First, in that paper B&M declare that while Lewis endorses WT (and WT1, not WT2 I take it), he repudiates ST--the strong thesis that one can break a law. I take it this is all based on the crucial claim about ability here. Namely, that while WT maintains that one has the ability to do a non-actual A under counterfactual law-breaking conditions, one has no ability to break a law in the actual world. (So the quote by Lewis by B&M: "Can you bend spoons?") So the ability claimed to do non-actual A is akin to a counterfactually-analyzed dispositional property, which B&M analogize to a rocket that was never fired but had the ability to clear a certain building had it been fired. I take it that such ability claims are basic physical ones which when unexercised in the actual world require counterfactual context to show their potential as powers. But I think Lewis never wished to claim that, while we have basic physical powers that can be counterfactually explained by employing local miracle-lawbreaking that tweaks context for the exercise of those powers, we might also have some magical ability to break the actual laws of nature.

Now B&M claim that if we assume a strict Humeanism about laws, then we can in fact actually have ST--we have the actual-world power to break the laws. But by-the-by, if Humeanism of that stripe is true, then anything that is held to have some dispositional power also has that ST property, and doesn't single us out as some special target of the CA, or for avoiding the premise about our supposed inability for influencing laws.

(I hope Joe is listening!)

Thanks again Helen, and I hope I've made some sense.

Hi Helen,

Thanks for engaging with my somewhat informal remarks about Lewis and the fixity of the past. I am traveling and away from my office, and I can now only go by memory, but I do recall that, on his view of counterfactuals, it will turn out in most cases that the nonrelational past (i.e., facts about the past) would have had to be different in order to allow for the "miracle". So it is not just that if the relevant agent were to act differently, a law would have been broken, but there would also have been *some* differences in the past facts to "allow for" the "local miracle".

But let's put that aside. As I wrote above, I'm just going on memory here. Consider the quote from Beebe:

“To see the distinction clearly, we need to get our time-indexing straight. We can distinguish two theses, each of which entails (WT) [ = I am able to do something such that, were I to do it, a law of nature would be broken] – the thesis to which Lewis explicitly subscribes:
(WT1) For some times t1 and t2, where t2 ≥ t1, and for some non-actual act A: I am able, at t1, to do A at t2, and, were I to do A at t2, a law would have been broken prior to t2 (and hence the past-relative to-t2 would have been different).
(WT2) For some times, t1 and t2, where t2 ≥ t1, and for some non-actual act A: I am able, at t1, to do A at t2, and, were I to do A at t2, a law would have been broken prior to t1 (and hence the past relative-to-t1 would have been different).
Lewis holds (WT1) … But nothing he says suggests that he holds (WT2)”.

Consider (W2). Doesn't the agent also have it in her power at t2 to do A at t2? Then wouldn't it follow that she has it in her power at a time (i.e., t2) to perform an act, A, such that, if she were to perform it, the past relative to that time (i.e., t2) would have been different from what it actually was?

That is, if we take it that a law's being broken in the past is the relevant kind of fact about the past, then (W2) would seem to commit one to a violation of the fixity of the past.

But, again, I don't have the opportunity really to think this through here, and the issues are highly nuanced. Thanks for your patience with these off-the-cuff ruminations.

There's a typo in my previous comment--SORRY.
It should have been, "Consider (W1)."

These issues about Lewis's view of counterfactuals, and how that interacts with ability claims, and thus, the fixity of the past (and laws), are incredibly complicated. Perhaps I screwed up in my above comments.

In my book, The Metaphysics of Free Will (an oldie but [I hope] goodie), I distinguished THREE views, the last of which I think is Lewis's: Multiple-Pasts Compatibilism, Local Miracle Compatibilism, and Mixed Compatibilism. The latter view says that when an agent is free to do otherwise in a causally deterministic world, she is free to perform an act such that, if she were to perform it, an actual law would not have obtained, AND some actual facts about the proximal past would not have obtained. I *think* Lewis defends a version of Mixed Compatbilism.

//Doesn't the agent also have it in her power at t2 to do A at t2?//

I'd say not (for non-actual A). If you're not actually doing A at time t2, then you can't do A at time t2. It's too late. Even if you were already on your way to doing A, it couldn't start until after t2, because if it started at t2, you'd be doing A at t2, contrary to the hypothetical.

To clarify. Sorry, in my previous post, I wrote "W2" instead of "W1". I think I was bringing up essentially the same point as Mark Young, i.e., "Doesn't the agent also have it in her power at t2 to do A at t2?"

I just don't see any easy way for the Lewisian to avoid fixity-of-the past problems. If the remote past is fixed, then so is the proximal past, I think. But the issues are highly nuanced.

John,

I *think* I*'m arguing against you -- but it's entirely possible I'm extremely confused. I do know that your correction of Tuesday evening didn't arrive at my computer until after I posted on Wednesday morning, so I was responding based on the flawed text....

But now I don't know what "(W1)" is -- did you mean "(WT1)"? If so, then I think your point is that (WT1) allows t1 and t2 to be equal, and so any time before t2 is also a time before t1 -- meaning that there is no fixity-of-the-past.

The point in my earlier post is that (WT1) should be replaced by (WT1'), in which the "t2 ≥ t1" is replaced by "t2 > t1". That is, I hold that, whenever it is true [that I am not doing A at t2], the claim [that I am able, at t2, to do A at t2] is false.

(I'd also add that, while (WT1) *allows* t1 to be equal to t2, it doesn't *require* it, so it doesn't require that fixity-of-the-past be violated.)

With such a replacement, any putative "miracle" occurs between times t1 and t2. The past of t1 does not need to be different than it is, because the counter-factual future of t1 is different from what the actual future will be.

I hope that's clear....

Of course it's only *after* I click Post that I notice something unclear.

Please interpret the previous to say that you're right about WT1 (if that's what you were talking about) -- it *does* allow violations of fixity-of-the-past, and those occur when t1 is equal to t2.

Also, I seem to has revised my past :-) *When I posted my first comment*, my point was not that WT1 should be replaced with WT1'. But *now* that would be a resaonable way to interpret what I wrote back then.

OK – so I’ve been prompted by John’s remarks to go back and read some parts of ‘The Metaphysics of Free Will’ again. It certainly is a goodie. I feel a bit alarmed by how little of the detail I’d retained from the first time I read it (but then again, that was probably about 20 years ago now!). The principle John calls ‘The Fixity of the Past’ goes as follows:
(FP) For any action Y, agent S and time t, if it is true that if S were to do Y at t, some fact about the past relative to t would not have been a fact, then S cannot at t do Y at t.
The time index here in John’s formulation is the same for the power ascription as it is for the action (“S cannot at t do Y at t”). This does seem, on the face of it, to raise Mark’s question – viz., if one isn’t already doing Y at t, then shouldn’t we say that one can’t do Y at t, because it’s too late? But there are some exceedingly tricky complications here, I think. I don’t see any reason to suppose that in general, the time index for the power ascription and for the relevant action can’t be the same. I can at 6pm precisely tonight shout (or perhaps more exactly, initiate the shouting of) ‘Hallelujah!’ at 6pm precisely tonight, for example (6pm is in the future, by the way!). What’s crucial to Mark’s point is that he specifies that ‘one isn’t already doing Y at t’. And this is relevant only in cases where we’re considering what we can do at t in a context in which it’s already t, as it were. ‘What can I do now?’ is a silly question in a way because I can’t do anything now that I’m not already doing! But one isn’t always considering the question whether S can at t do Y at t from t itself. And it has started to seem to me as though this might really be quite crucial – and so that there are not just two time indices that matter, but three – time that power is possessed, time of action and time of assertion.
To see why it might matter, return to (FP). John’s phrase “If S were to do Y at t …” is the kind of phrase we use, I suggest, when t is assumed still to be in the future relative to the time of assertion. But in that case, ‘the past relative to t’ will include some times which are also in the future, relative to the time of assertion. There is then a question about the sense in which there are ‘facts’ about this period (which has not yet taken place). Suppose we allow that there are facts about what will happen between the time of assertion and t – and that some of these would not have been facts if S were to do Y at t. Say, for example, that if S were to mow the lawn at t, he would have to acquire a desire to mow the lawn between the time of assertion and t – and that in fact he will not acquire any such desire. According to FP, in this situation, S cannot at t mow the lawn at t. But surely if I said at the time of assertion (now, say) ‘S can at t mow the lawn at t’ (and assuming no special obstacles) I’d be right. Even if S will in fact not acquire the necessary desire between now and the time of mowing, it doesn’t follow that she couldn’t acquire it. And so even though, if S were to mow the lawn at t, she’d have had to have a desire at t-1 that she doesn’t in fact acquire, I still think that S can at t mow the lawn at t, because we haven’t yet said that she *can’t* acquire this desire, only that she doesn’t. So I would want to deny (FP) as formulated here, I think. (This point is actually inspired by one made by John himself against Van Inwagen’s restrictivism!).
I might well be confused about this – it’s all exceedingly tricky. But at the moment, I think we’d need to formulate (FP) slightly differently to turn it into something I’d want to sign up to.

Hey Al: I just got around to commenting on your post about our project. There are a lot of interesting things going on here and it is hard for me to keep them straight. So let me say a few things about my (new) thoughts on Lewis, some that I haven’t shared with you yet but will help the project.

I agree with you that Lewis does not deny the fixity of the past but I would slightly rephrase your point and note that Lewis sees an ambiguity with ‘render’ as in ‘render p false,’ not an ambiguity with ‘ability.’ Van Inwagen contends that his argument works no matter what analysis of ‘ability’ one holds, and Lewis (and many other philosophers) accept this claim and go from there. I’m happy to discuss this in more detail but here are some initial thoughts.

Roughly, Lewis’ strong thesis about ‘render’ (ST) suggests that rendering a proposition false is a causal power (it entails that one can break a law) whereas Lewis’ weak thesis (WT) suggests that it is not a causal power (it entails that I can do it, and if I do the past and laws would be different, but it doesn’t entail that I can change the past or laws).

Here’s why this can’t be an account of ‘ability’: Because Lewis favors WT over ST but ability is (plausibly) a causal power. We lack the ability to change the past or laws.

Lehrer made this same point previously. Let S be a person, A an action, and t a time. Lehrer writes: “If S had done A at t, then, of course, either the laws of nature would have been different or the state of the universe would have been different. But that is not to say that the person could have brought about these conditions” (1980, 199).

Thus, Al is correct that Lewis does not deny the fixity of the past. Nor (in his reply to the consequence argument) does Lewis deny the fixty of the laws. Compare Lewis with Hume. Both were Humeans about laws of nature but neither (I contend) use their Humeanism to respond to CA. Bold claims but I'm sticking with them!

The other thing that interests me in this thread is whether or not one can deny the fixity of the past without also denying the fixity of the laws, and vice versa. I don't think one can. Helen's introduction of time indices complicates matters (though I favor this approach, and these are helpful analyses). To simplify, I focus on the time of ability.

Suppose I am a no-fixed-past compatitiblist. I respond to the CA by denying the fixity of the past and claim:

I am able at t to do A, and if I were to do A, some past proposition that was true is now false. Of course, I won't do A in the actual world, @ (we're assuming determinism). I do A in some other possible world W.

Does W share the same laws of nature (= L) as @? It is doubtful. Compare @ and W. I'm not sure I understand MD but if ED holds in @, then any other possible world that shares the past of @ yet has a false proposition where @ has a true one, will not share the laws of nature of @. According to ED, any past proposition about the way @ is at a point in time t' entails every other true proposition. Thus, if determinism is true, there cannot be two possible worlds alike in all respects (including laws) save for the truth of one past proposition. Changes in the past have a ripple effect, one that affects the laws. And it doesn’t matter what account of laws you hold.

On the other hand, if I am a no-fixed-laws compatibilist and respond to the CA by denying the fixity of the laws, I claim:

I am able at t to do A, and if I do A, some law of nature would be false. Of course, I don't do A in @ (we're assuming determinism). I do A in some other possible world W.

Does W share the same past as @? No. The past includes the laws (which are propositions), so if the laws change, the past changes. QED.

Helen,

Thanks for your very helpful, and very gracious, reply.

Right, as you point out, my statement (in The Metaphysics of Free Will) of the *conditional* version of the fixity of the past is: (FP) For any action Y, agent S and time t, if it is true that if S were to do Y at t, some fact about the past relative to t would not have been a fact, then S cannot at t do Y at t.

Would this be preferable:

(FP) For any action Y, agent S and time t, if it is true that if S were to do Y at t, some fact about the past relative to t would not have been a fact, then S cannot at t (or just prior to t) do Y at t.

Joe: the way I see it, Lewis certainly commits himself to a denial of both the fixity of the past and laws, where "fixity" is understood properly (in terms of the "conditional version" or the "basic" or "possible worlds version" of the relevant fixity principles.

Hi Joe--and thanks so much--you always do such careful work.

I wish to say something both to you and Helen with respect to her claim:

"Say, for example, that if S were to mow the lawn at t, he would have to acquire a desire to mow the lawn between the time of assertion and t – and that in fact he will not acquire any such desire. According to FP, in this situation, S cannot at t mow the lawn at t. But surely if I said at the time of assertion (now, say) ‘S can at t mow the lawn at t’ (and assuming no special obstacles) I’d be right."

I agree with this, and think this is what Lewis was trying to get at with his WT reply to the CA. We very ordinarily say we have assorted abilities to do this or that, and though we don't usually try and provide sufficient conditions for action in terms of ability, we do try and state necessary conditions in terms of "can". "S can at t mow the lawn at t" because if I assert this at an earlier time t*, then a Lewis WT counterfactual assessment of S's ability at t (and even at the time of assertion t* in @) stands: if S did that at t, then at some point in the counterfactual situation W something was sufficient to produce mowing at t. But--and here is what I take Lewis to focus on--nothing in W is *significantly different* than my actual world in terms of our understanding of what constitutes a power or ability--even causal powers. If when at t* at @ I say that S can mow at t, I'm not stating a sufficient-condition set analyzing that ability--I'm saying that, under a counterfactual understanding of how S may behave in the details of psychology and physiology, the big picture account of S's physical ability to mow--she up to t* in @ is physically capable, no one is barring her from the yard, etc.--is intact. "S can mow the yard at t" is true stated at earlier t* at @, specifically understood as a necessary-condition assertion of physical capability analyzed (I think dispositionally) at the earlier t*.

Now because this analysis of an ability claim at t in @ is one in terms of necessary conditions asserted at t*, there can't be anything entailed about ability as a specific causal power in this world except to parse its exercise under sufficient conditions of counterfactual W. To that extent Joe you are right (I think).

I hope I've made some sense here. Thanks for your comments and your patience.

//But surely if I said at the time of assertion (now, say) ‘S can at t mow the lawn at t’ (and assuming no special obstacles) I’d be right.//

I disagree, but perhaps it's because we are interpreting "can at t" differently. I don't know how you are interpreting it, but I am interpreting it as a statement of the time at which a possibility is open. I think this interpretation is consistent with the way "mows the lawn at t" is being used:

[S is mowing the lawn at t] = "S is mowing the lawn" is true at time t.

[S can at t X] = "S can X" is true at time t.

Thus the "at t" in each case is specifying a time at which the related assertion can truthfully be made. "S is mowing the lawn" can truthfully be asserted at time t: [S mows the lawn at t]. "S can walk" can truthfully be asserted at time t: [S can at t walk].

Basically, each verb needs to be given a time to pin down the meaning of the assertion. In English, tenses restrict the range of times available.

So if you are asserting *now* that S can mow the lawn at time t, what you are asserting would be written [S can at now mow the lawn at time t]. But [S can at t mow the lawn at t] says that you can truthfully assert *at time t* that S can be mowing the lawn at time t -- when it's given that S is not mowing the lawn at time t.

"S isn't mowing the lawn, but S can be mowing the lawn." There is no time at which that statement is true. Thus [S isn't mowing the lawn at t and S can at t mow the lawn at t] is also always false.

On the other hand, [S can at t1 mow the lawn at t2] may be true when t1 < t2, even if [S is mowing the lawn at t2] is false. Thus "S can mow the lawn" should be interpreted as saying that, at this time, it's an open possibility that S will mow the lawn at some time in the (near) future.

Joe,

One can deny the fixity of the past without also denying the fixity of the laws. You just have to go whole-hog with the alternate past, i.e. world W has to differ at all times from @. This needn't be as bad as it looks, though; most of the differences in the past could be microscopic. The differences in their futures could be vast, but presumably that's OK. (See David Albert, *Time and Chance* pp 128-130, for an explanation of this asymmetry in the way changes propagate to the past and future.)

Lewis thinks such a world W would be "further" from @ than would another with slightly different laws. I don't see it.

John: Right, I was commenting on Lewis' response to van Inwagen's first argument, and in that reply he claims there is an ambiguity about 'render;' once you flesh that out his response, it does not line up in the right way with his denying the past and the laws, though this is a long story.

If we shift to some other version of CA, things get complicated. For instance, in the Basic Argument (Fischer 1994), the fixity of the past and laws gets wrapped up with other claims. In that argument all you need is the thesis of determinism and an extension principle: free actions are an extension of the past and the laws (cf. Fischer 1994; Haji 2009). If this is correct, any classical compatibilist would have to reject the extension principle. It is hard to deduce a commitment to the rejection of the fixity of the past or laws from the fact that one rejects the extension principle.

But you, John, are talking about the conditional argument and I'll have to look at that more closely before I say anything about it in particular. Still, Lewis' (1981) main points are (a) the past and the laws matter, so when considering what we can and can’t do, we need to hold the past and the laws fixed as much as we can; but (b) we can’t do this perfectly given determinism (thus the need for a miracle), so every accessible world will differ in past and law. However, (c) this doesn’t mean that someone or something was a cause of these changes in past or law. The worlds serve merely as models.

There are some complications when considering claims about the fixity of the past and laws and Lewis’ views. We might think that claims about the fixity of the past or laws are just claims about whether the past or laws could be different, meaning, that they are different in accessible worlds. But as we saw, all classical compatibilists must admit to this difference.

On the other hand, there are versions of the consequence argument where claims about the fixity of the past or laws play a broader role. For instance, van Inwagen’s first argument, where the assumption is not just that the past is fixed in the not-accessible sense but that no one can do anything such that, were she to do it, the past and the laws would remain the same. Say what you want but this is a much more complicated claim, involving modal operators and other fancy stuff. How Lewis’ (1981) response plays out in the first argument is clear. How it plays out in other versions of CA, with other modal apparati, is less clear.

I was just looking back at the Taylor and Dennett article "Who's Still Afraid of Determinism" in Kane's Oxford Handbook on FW, 2nd edition, and it made me want to ask anyone who is reading here whether anyone has written a response to this article (or their first version of it in 1st edition) or to Perry's "Compatibilist Options" in the (Joe!) Campbell et al. volume. I can't decide how effective I think these responses to CA are (or how different they are from others, including Lewis'), but I find them more interesting than many others, and I was curious what others thought.

Gosh - this stuff is hard. Great comments, everyone; and as usual, there’s too much to respond to. But here are some remarks about Joe’s very fascinating post.

Joe - having reread 'Are we Free to Break the Laws?', I think I'm inclined (newly inclined!) to agree with you that Lewis doesn't really seem to rely on his Humeanism about laws in his reply to Van Inwagen – though I have to confess that it’s a bit of a mystery to me why he doesn’t. It seems to me like the quickest and simplest route to a response to the Consequence Argument that really draws its sting (hence my original concern about ED). The focus is all very much on the distinction between ‘I am able to do something such that, if I did it, a law would be broken’ and ‘I am able to break a law’. And I also agree, I think, that the ambiguity he points to is an ambiguity, effectively, in ‘render false’. But surely an analysis of ability must be at least implied in order for that ambiguity to exist in the first place? Lewis seems to be trading on the idea that I can render things false either by exercising some causal power of mine directly (e.g. I can render it false that a particular leaf will remain on the mulberry bush outside my window at 12 noon today, by going and picking it off); or alternatively, I can render something false if it is within my power to do something (else) which, were I to do it, would imply (by Lewis’s theory of counterfactuals) that the first thing was false too. Only in this latter sense can I render a law of nature false. But isn’t it the case that one’s only going to think it *is* in my power to do things which (were I to do them) would render some actual law false if one has already implicitly accepted some compatibilist-type analysis of ability? If I already think that e.g. to be able to A I only need to have the *general* power to A; or that if I were to choose to A, I’d A; or that I just need the ability and the opportunity to A (or something similar … I don’t mean to suggest that these are exhaustive of the compatibilist options) then it becomes obvious why one might think that Lewis’s weak and strong theses differ. But doesn’t the claim that there is such a difference depend on some such view? – so that even though there’s no explicit analysis of ability given by Lewis, a compatibilist account of some sort is required in order to make the whole thing work?
The other thing I wanted to ask about was your argument that one can’t believe in the fixity of the past without also believing in the fixity of the laws. The argument seems to be that “the past includes the laws”. But presumably the particular facts about W@ up until t1 might be best regimented by way of laws L (which achieve the best combination of simplicity and strength for that portion of universe) while the particular facts about W@ until t2 (t1< t2) require laws L* instead. So in at least one good sense, the past *doesn’t* include the laws – for if we look only at the past, L* (the correct account of the laws for the whole of W@) is not delivered up as the correct set. But I may have misunderstood the argument here. On the face of it, though, I feel inclined to think that one *can* deny the fixity of laws while not denying the fixity of the past.

Eddy,

I've written a reply to Dennett's reply to the CA. You might find it of interest: "Dennett on the Basic Argument," part of a book symposium on Daniel Dennett, Freedom Evolves, Metaphilosophy Vol. 36, No. 4 (July 2005): 427-435.

The comments to this entry are closed.

Books about Agency


3QD Prize 2014: Marcus Arvan