Blog Coordinator

« The Direct Argument | Main | Iffy Abilities »

11/06/2014

Comments

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

Skeptics about free will are not (necessarily) skeptics about control. Galen Strawson, with his skeptic hat on, can grant that a person has control over X. What worries the skeptic is not whether the person has control over X. Instead, the worry is that the person has no control over how the person exercises that underlying control. In other words, the worry is about meta-control. Or meta-meta-control, etc. Turtles all the way down.

An analogy is helpful. A thermostat has control over the temperature. This is true even in deterministic worlds. But the thermostat typically has no control over how the thermostat exercises that control. No meta-control.

We human beings don't want to be glorified thermostats. The example of thermostats show that lower levels of control are relatively meaningless without higher levels of control - or ultimate control.

You could imagine that a person has n-levels of control over their actions. The person has control, and meta-control, and meta-meta control, for example - n=3. But the person lacks the level n+1 of control. Al Mele's Diana comes along and exercises meta-meta-meta control over the agent's life. The person has control, and meta-control, and meta-meta-control - but the person exercises all of those perfectly according to Diana's plan. This is what worries the skeptic.

Now ultimate control, like ultimate responsibility, might be worthless. It might be incoherent. It might be many things. But, on the surface, it is what people often want, or think that they want. And it is incompatible with determinism, because it is logically impossible.

Kip,
I agree, there are certain sorts of control that we can't have. The sort of control I think we can have is the ability to do otherwise in a deterministic world. I'm a classical compatibilist. The big argument against my position is the Consequence Argument. While there have been many tactics trying to show where the argument goes wrong, it is my understanding that there are no alleged counterexamples to Rule Beta box. It's time for a first.

As far as free will skeptics, they need more than you mention here. On this account of free will skepticism, I'm a skeptic. We can't have ULTIMATE control and that may be what some people want. Yet, I accept the arguments both for semi-compatibilism and for classical compatibilism. I'm no skeptic. It seems what else is needed is that ULTIMATE control is required for moral responsibility.

It seems, though, that there are two distinct questions that both go under the name of 'free will'. One question is whether we can be morally responsible in a deterministic world. For that question the powerhouse argument for incompatibilism is the Direct Argument. In my last post, I have shown why it is invalid. The second question is whether we can have the ability to do otherwise in a deterministic world. For that question, the powerhouse argument for incompatibilism is the Consequence Argument. If my argument here is at all plausible, there is a radically new way to respond to the argument.

My target here is libertarians, skeptics, and the majority of people who call themselves compatibilists today the semi-compatibilists. While many semi-compatibilists are officially agnostic about the ability to do otherwise in a deterministic world so aren't direct targets it is time for them to return to their roots with classical compatibilists.

My position is that while the ability to do otherwise is not necessary for moral responsibility we have the ability to do otherwise in a deterministic world. Certainly, many incompatibilists will retain their intuitions. But, if the main arguments supporting their position fail, it may be time for them to reconsider. Certainly, they will find future recruitment drive more challenging.

I think I may have missed Kips concern. Notice, Bob produced a truthmaker for the disjunction and could have done otherwise. When we think of the ability to do otherwise as requiring the ability to render a proposition false it becomes quite challenging. Yet, when we think about producing truthmakers for a disjunction there is an alternative way to do otherwise. Someone who produces a truthmaker for a proposition could do otherwise by not producing a truthmaker for that proposition. While Bob both produced a truthmaker for this disjunction and could have done otherwise, he lacks the capacity to render the proposition false.

When I was composing this post, my two year old took off a dirty diaper and flung it at the screen. While the crap smeared down the screen he pointed at it and yelled "POOP". Since this post comes from a longer paper that is currently under review http://philpapers.org/rec/HERTAT-4, I appreciate people tearing into it. But, please be kinder than Alex (who got his name because being one of Mele's students I can think of nothing more honorable to be called than Al.)

Hi Charles,

(1) You said in your 2:58 comment above: "...when we think about producing truthmakers for a disjunction there is an alternative way to do otherwise. Someone who produces a truthmaker for a proposition could do otherwise by not producing a truthmaker for that proposition. While Bob both produced a truthmaker for this disjunction and could have done otherwise, he lacks the capacity to render the proposition false."

My response is: I agree. And, so what? All the CA needs is a definition of 'able to render P false' which is such that S is able to do otherwise if and only if S is able to render some true P false. Let P be the true proposition that Joan of Arc was burned at the stake. P entails (P v Q), for all Q. Let Q be the true proposition that I raised my arm. Was I able to render (P v Q) false? No, not on any of the usual definitions (and, in my view, this is a good thing because I'm intuitively unable to do anything that might falsify the disjunction, i.e. unable to do anything X such that the disjunction might be false partially in virtue of the fact that I Xed). Does it follow that I'm unable to do otherwise? No. I'm able to do otherwise if I'm able to render any true proposition false. And if I'm able to refrain from raising my arm (i.e. able to refrain from producing a truthmaker for the disjunction, as you put it), then I'm able to render Q false. What's the problem?

(2) I'm not following the argument that BETA-box is invalid. Can you tell me what P and Q are (and which story they are true of)?

Nate,
Thanks for the post. It looks like you follow the argument perfectly and in your last post agreed that Beta Box is invalid. Bob produced a truthmaker for the disjunction and could have done otherwise. Rule Beta Box, however, entails that Bob could not have done otherwise. After all, he had no control over the fact that Sue robbed the bank. Further, 'Sue robbed the bank' entails that 'someone robbed the bank'. So, Beta Box entails that Bob has no control over the states of affairs described by 'someone robbed the bank' (and remember the sort of control at issue here is the ability to do otherwise). Yet, you agree that he could have done otherwise. So, an ontological reading of Beta Box renders it invalid.

The problem with a semantic reading of the Consequence Argument is that it leaves us with only two options. Either a proposition is true or it is false. so, what would it require to do otherwise than make a proposition true? Defenders of the Consequence Argument wrongfully think, it must require the ability to render that proposition false.

This is the mistake of the Consequence Argument. It rests upon a false dilemma. Doing otherwise than rendering a proposition true is not rendering a proposition false. Instead, it is not rendering that proposition true. Yet, there are two manners in which someone can not render a proposition true. One is to render it false. Another is to not render it true even though some other truthmaker renders it true (as in the case of Bob).

What the Consequence Argument needs isn't a definition of 'able to render P false' that fits its purposes. It needs an account of the ability to do otherwise that entails that Bob could not have done other than produce a truthmaker for the proposition that 'someone robbed the bank.' Since you agree that Bob could have done otherwise, then it seems to me that you agree that any such account will be highly implausible.

Nate,
The argument form is merely to show that Rule Beta Box entails that if a disjunction has two true disjuncts then an agent cannot do otherwise then render that proposition true unless she has control over both disjuncts. It is intended as a general argument form but is easily applied to the case of Bob and Sue. Clearly, when read semantically, this is a valid argument. When read ontologically, it leads to the result that Bob cannot do otherwise then render the proposition "Bob or Sue robbed the bank" unless he has control over both disjuncts. He doesn't have control over both disjuncts. So, he cannot do otherwise then render the proposition "Bob or Sue robbed the bank" true. Yet, as you agreed in your last post, Beta Box is providing the wrong result. So, it seems to me that you actually accept the counterexample.

Hi Charles,

Thanks.

I don't agree that the cases show that Beta Box is invalid. A counterexample must have a true N P, a true necessarily (if P, then Q), and a false N Q. I do not agree that N Q is false in your cases. If Sue robbed the bank before Bob, then Bob is not able to perform any action X such that were he to X, his Xing would be or cause an event e which is such that the fact that e occurs partially explains the falsity of Q (i.e. the falsity of the proposition that either Sue or Bob robbed the bank or the falsity of the proposition that somebody robbed the bank). So it’s not a counterexample as I see it. Yes, I agree that Bob was able to do otherwise---it’s stipulated in the example that he was able to refrain from robbing the bank. But this does not require that he is able to render false the proposition (i) that Sue or Bob robbed or (ii) that somebody robbed. It requires that he is able to render false some true proposition. And he is, namely the true proposition that Bob robbed the bank.

Nate,
Rejecting Beta Box does not require showing that an agent can render the proposition false. Thinking that it does is merely to fall for the false dilemma. Why do we talk about rendering a proposition false in the Consequence Argument? The thought is that agents render certain propositions true. For the agent to be able to do otherwise, therefore, seems to require that the agent can also render that proposition false. It is that line of reasoning that causes defenders of the Consequence Argument to talk about the ability to render a proposition false. So, at the very core of the Consequence Argument is a false dilemma. To show that Beta Box is invalid I need to provide a case where an agent who renders a proposition true could have done otherwise even though Beta Box entails that he cannot. I claim that Bob rendered the proposition true, had the capacity to refrain from rendering the proposition true, but lacks the capacity to render the proposition false. Therefore, he can do otherwise than render that proposition true. Yet, Beta Box entails that he cannot do otherwise than render that proposition true. You reply that Bob cannot render the proposition false and I need to be able to show that. This is strange. I claim that the argument rests on a false dilemma. You agree that there is a third option. Then, however, you insist that to show that the argument is flawed I must employ only the two options in the initial false dilemma. I don't understand your concern.

As far as the timing issue, I think it might matter. In the long paper I have them simultaneously robbing the bank but missed the detail when condensing

Hi Charles,

Thanks. I’m still not following. Maybe this will help:

Here is my definition of the quasi-technical "able to render false" idiom: S is able to render P false if and only if S is able to do something X such that one of the nearest worlds w at which S X's is a world where (i) an event e occurs, (ii) e is S's Xing or an effect of his Xing, and (iii) e renders P false at w. This definition involves a more basic idiom--"event e renders P false at w" which I understand as the claim that, at w, the fact that e occurs explains the falsity of P (or, equivalently, as the claim that, at w, the falsity of P is grounded in the fact that e occurred). As I see it, this is enough for the CA just in case the ability to do otherwise requires the ability render to render some true proposition false. And I think that the ability to do otherwise does require this. (I don’t think you disagree with that.)

I stipulate that Beta-Box is to be interpreted in terms of the N-operator which is itself understood in terms of the "able to render false" idiom, as I just defined it. I claim that the ability to do otherwise requires the ability to render a true proposition false and, if I put my incompatibilist hat on, I will say that Beta-Box governs N, thus defined. You want to deny Beta-Box. A counterexample to Beta-Box must employ the "able to render false" idiom. So I want to see a counterexample that uses those words. Maybe you can translate your purported counterexample into the language of my version of the argument. But I don't see how.

Here is why I don't think the cases counterexample my Beta-Box, defined in terms of the "able to render false" idiom above:

In the Sue-Bob case, Bob is able to render some true proposition false--namely, the proposition that Bob robs the bank. So if 'Q' is the proposition that Bob robs the bank, there is no counterexample (where the conclusion is 'N Q') because Bob is able to render that proposition false.

Let 'R' be the proposition that either Sue or Bob robs the bank. Let 'T' be the proposition that Sue robs the bank. Since Bob can't render T false and since necessarily (if T, then R), Beta-Box entails that Bob is unable to render R false. I agree with all of this. But it does not follow that Bob is unable to rob the bank. I have not articulated any principle entailing that: Bob is unable to refrain from robbing the bank if he is unable to render R false. I only said: Bob's ability to refrain from robbing the bank requires that he be able to render some true proposition false. (He is: namely proposition Q above.)

Nate,
As I claimed in the post purely semantic versions of the Consequence Argument may be sound....but they are irrelevant to the ability to do otherwise. Beta Box is employed in the Direct Argument to reach the conclusion that I have no control over the proposition 'I raised my hand'. From there, it derives that I raised my hand and could not have done otherwise. At some point, the argument needs to move from semantics to the world. The manner in which it does so is to claim that having no control over the truth value of a sentence entails having no control over the states of affairs that the sentence is about. Interpret N in any way that you like, if it is relevant to free will this move back to the world needs to be allowed. Otherwise, you are creating an artificial language that has no bearing on the world. If this move back to the world is allowed, then we can move from Nb 'Bob robbed the bank or Sue robbed the bank' to Bob has no control over the states of affairs this proposition is about. Yet, you agree that this conclusion is false. He produced a truthmaker for this claim and he could have done otherwise. However we interpret N, if we are allowed to drop it and derive from it that the agent has no control over the states of affairs that the proposition is about (i.e. have no control over raising my hand)then it also leads to the problematic conclusion for Bob.

What we have learned in the truthmakers literature is that the inference rules that work perfectly well for semantics do not hold for the truthmakers for the semantics. It is for this reason that we derive problems whenever we try to preserve ontological concepts like control through semantic relations like classical entailment. You keep insisting that I remain on the semantic plane and that I cannot drop that N and think about what it says about the world. If that is right, the Consequence Argument is doomed because it can never tell us anything about the world. If we can think about what these semantics are saying about the world, then they are telling us that Bob has no control over the states of affairs that the disjunction is about. But, that is also false. Either way, the Consequence Argument is in trouble.

Hi Charles,

Good. I think we're making progress. But I still disagree.

You say: "At some point, the argument needs to move from semantics to the world. The manner in which it does so is to claim that having no control over the truth value of a sentence entails having no control over the states of affairs that the sentence is about."

I disagree with the second sentence. As I understand it, this is not how the CA in fact moves (or should move) "from semantics to the world" or, as I would put it, how it moves from a certain sort of quasi-technical claim about the ability to render propositions false to claims about what we are able to do. (And, I should note, the quasi-technical claim is defined in terms of relatively ordinary notions like the ability to act, the ability to cause something, 'might'-counterfactuals, and the notion of a fact grounding another fact. So I don't really understand the charge that it's a "semantic" notion, but that's just a terminological quibble.) I did not claim--as you seem to think that the CA defender must claim--that if I am unable to render a proposition false, then I "have no control over the states of affairs that the proposition is about." I think the CA defender should deny that claim. Instead, to repeat, what I think the CA defender can and should say is this: if I am unable to render any true propositions false, in something like the sense I've defined, then I am unable to do otherwise. Do you think that claim is false? Because it is a claim connecting "semantics and the world"; it articulates the intended and required connection between the "able to render false idiom" and the ability to perform actions.

Charles,

Thanks for taking the time to reply.

Let me clarify: the upshot of your argument above is that an agent can have control over X even in a deterministic world. This remains true, you concede, even if you grant that the agent cannot render false a true proposition in that world. For example, the agent can have control over whether he robs the bank, even if he cannot render it false that he robs the bank (if he does).

First, I think you're splitting some mighty fine semantic hairs. Also, I'm not sure that the English language is precise enough to support the kinds of distinctions you're making.

Second, the more important point is: skeptics can grant your conclusion without giving up skepticism. If you're saying "well, Kip, I'm attacking the Direct Argument, and it relies on X, so I'm still making progress, even if I haven't defeated skepticism." That's fine. I'm happy to give up the Direct Argument.

I consider myself a post-determinist, in the sense that I think that all talk about determinism is potentially misleading, because determinism and indeterminism are equally threatening to free will (if they are). The problem is self-creation, not determinism. And self-creation is impossible in both deterministic and indeterministic worlds. In this spirit, I follow Neil Levy, Galen Strawson, and Kristin Demetriou, etc. Depending on how broadly or narrowly you define compatibilist, I might also be a compatibilist like you, but one who is skeptical about the existence of free will (we discussed this semantic point earlier in Prof. Demetriou's threads).

Charles, I have a basic worry about equivocation here.

Re:

"Read (2000) employs a case where the only horses in a race are Valentine and Epitaph. The race's occurrence guarantees that either Valentine or Epitaph wins. Yet, the race's occurrence neither guarantees that Valentine wins nor guarantees that Epitaph wins. So, while the race's occurrence makes it true that either Epitaph or Valentine wins it does not make either disjunct true. The same point could be made with the coin example. Flipping the coin makes it true that the coin lands on either heads or tails. Yet, since the coin flip does not guarantee what direction the coin would land, flipping the coin neither makes it true that the coin lands on heads nor makes it true that the coin lands on tails. The reason the same arguments can be employed when considering free will and truthmaking is that they share the same logic."

My worry is that semantic-style language and logic--which seems essentially linked to expression about states of affairs as types--is interwoven in the above passage with truthmaker language and logic--which seems essentially linked to expressions about tokened events or the like (to convey specific instances of what makes a proposition true). The horse-race as described as a type entails propositionally that one horse wins of a disjunctive two. The race as a tokened run entails that exactly one horse wins (unless there is a photofinish tie), though epistemically we do not know which until the race is finished. So type and token-entailments might be placed together with involved epistemic issues of uncertainty that result with lots of confusion as to what is what.

Nate,
Thanks for the posts they are greatly appreciated. I've been really interested in the move you mentioned in the last post as it was appearing earlier so am delighted that you brought it up again.

I want to start my response to your last post with a defense of my argument that is less in the spirit of the learning environment that makes Flickers so fun and instead aimed at why I don't think the comment is too concerning for my paper (after all it is currently under review). I don't think a defender of the Consequence Argument can say what you do...but a defender of the Consequence Argument* can. Notice, the Consequence Argument concludes NsP. From this I am supposed to derive that I am unable to do otherwise than type this response. Certainly, we can iterate the Consequence Argument to obtain the same result for every proposition. Yet, the argument concludes before doing so. So, if the iterated Consequence Argument is superior to the single case it doesn't affect the argument of my paper. Possibly, however, it gives a way for the incompatibilist to respond to it.

I find your suggestion very intriguing. If the idea is that we end the argument with the conclusion that nobody can render any proposition false and then move from there to nobody can ever do anything other than what they do and deny that we can ever make from the inability to render a proposition false to the inability to do otherwise in that one instance my current argument won't work. The only problem I have with the view is what makes it appropriate to move from rendering false to doing otherwise inappropriate prior to that point from the perspective of the incompatibilist. What would motivate an incompatibilist to abandon the single case Consequence Argument and endorse the iterated version? Seriously, I find this move really neat.

Kip,
"For example, the agent can have control over whether he robs the bank, even if he cannot render it false that he robs the bank (if he does)." is something I don't think I'd ever commit myself to. Instead, it seems to me that if there are counterexamples to Beta Box when it is used to derive that nobody can do otherwise then the argument is also probably flawed for showing that nobody can render a proposition false. Since I think it is flawed for the ability to do otherwise there's strong evidence that it is also flawed for the ability to render a proposition false.

As far as your skeptic intuitions...I feel them as well at times. I see the goal of doing philosophy as trying to figure out what the costs of different positions are. By confronting arguments we are tying to discover what those costs are.

Alan,
I'm not quite sure I'm following your worry. My point with Read is that paying attention to the logic of truthmakers seems to make excellent sense because the most successful counterexamples to the early versions of both the Direct Argument and the Consequence Argument fit perfectly together. The exact same counterexamples that convinced people to abandon Beta and move to Beta Box are counterexamples to the principle that something is a truthmaker for a disjunction if and only if it is a truthmaker for both disjuncts.

Charles -

I didn't mean to mischaracterize your position. It seems highly technical to me, and I have trouble following it in the blog format.

In your last comment, you resist committing to my restatement of your position. But it still seems to me to follow pretty naturally from your original post, where you wrote:

"Yet, having at least some control over the states of affairs that make a proposition true is not the same thing as having the capacity to render a proposition false. An agent who produces a truthmaker for a proposition and has the capacity to refrain from producing a truthmaker for that proposition has a significant amount of control over the states of affairs that the proposition is about even if she lacks the capacity to render that proposition false."

That is essentially the same thing as I wrote earlier:

"For example, the agent can have control over whether he robs the bank, even if he cannot render it false that he robs the bank (if he does)."

My version is just written in terms of a concrete example. If I'm still mischaracterizing your position, perhaps you can clarify how my interpretation is inconsistent with what you wrote in the original post.

More importantly, I don't want to get bogged down in the weeds. At a higher level of abstraction, I do want to try to clarify my main point. So let me rephrase one more time.

In *this* post, you are attacking the Consequence Argument (rather than the Direct Argument). The Consequence Argument has many forms, as you know. As you describe it, the CA concludes "that nobody has any control over anything in a deterministic world."

The problem here is that this conclusion is too strong. Skeptics don't need it. Skeptics about free will are skeptics about... free will, not control. Skeptics can grant that agents have (some kind of) control, without granting that agents have free will. Of course, everything depends on the definition of free will, and Van InWagen defined free will (rather dogmatically) as the ability to do otherwise. But, skeptics can agree that people and thermostats exercise control, even if they never do otherwise. Fischer et al. show this, in distinguishing between regulative and guidance control. And Dennett shows it in discussing how agents evolve to exercise control even in deterministic worlds where they never do otherwise.

So if the Consequence Argument is flawed, whether because beta is false or otherwise, skeptics can just let it go. I know it's a popular and famous argument. And perhaps another version of it is sound. But I think skeptics (for one) can agree with you here that is false.

Instead of focusing on the Consequence Argument, and framing the debate in terms of determinism, I would focus on the issues of self-creation and constitutive luck, per Neil Levy, Galen Strawson, and Kristin Demetriou, among others. I think that will be more productive, because determinism itself is a red herring, at best a clearer lens to reveal a moral luck problem that occurs regardless of whether determinism is true.

Kip,
I think the claim that someone can have no control over the truth value of a proposition without having control over the truth value of that proposition leads to the result you and I both find counterintuitive. Instead, it seems clear that in cases of overdetermination this is precisely what occurs. Bob has significant control over the states of affairs that 'someone robbed the bank' is about even though he cannot render that proposition false. He provided a truth maker for it, and could have done otherwise.

Also, I agree that the Consequence Argument does not need to show that we have no control in a deterministic world. What it does need to show is that we lack the 'ability to do otherwise' sort of control. Fischer's control is certainly consistent with the Consequence Argument as well are many other forms.

I also agree that there is no reason that a skeptic can't let the Consequence Argument go. Accepting the Consequence Argument is not necessary for denying that we can do otherwise in a deterministic world. Nevertheless, it is one of the main reasons to do so.

I sort of agree with your comment about determinism not being the real threat to free will....but I don't agree about what the main threats are. Issues like self-creation are a real threat if one thinks that free will requires it. If I were convinced that creating oneself from scratch were required for free will, I would be a skeptic. Yet, this isn't what raises my pro-skeptic intuitions.

What generates my pro-skeptic intuitions the most is thinking of the relationship we bear to the particles that compose us and how our behavior is dictated by them. Regardless of whether the world is deterministic or indeterministic what I do seems to be grounded in the causal powers of the particles that compose me. From there, we can generate problems like the issue of self-creation and constitutive luck, but I think there are other issues that also stem from that same source.

Hi Charles,

As I see it, I am just parroting van Inwagen, not introducing a novel conception of the CA.

We can, I think, run a "single instance" version of the argument as long as we're careful about the Ps and Qs. The recipe is as follows: suppose we want to know whether or not S, at a deterministic world, is able to perform some arbitrary action X, an action that S did not perform. Let 'P' be the conjunction of the law-stating proposition and the remote-past proposition. Let 'Q' be the proposition that S did not X. We conclude, with Beta-Box, that S is unable to render Q false. It does follow from S's inability to render Q false that S is unable to X. (My claim is NOT this: there is no single proposition such that if the agent can't render that proposition false, then she is unable to do otherwise. Q above, in my view, is such a proposition.)

Now, the conclusion of the modal version, as I recall, is: no agent at a DET-world is able to render any true proposition false since the argument is run by saying "let Q be an arbitrary true proposition" and deriving the result that nobody can render Q false. By 'showing' that no agent is able to render any true proposition false, we have thereby shown that no agent is able to do otherwise, and for the following reason: for any action X that the agent doesn't perform, there is a true proposition stating that he didn't X, a proposition that (i) he must be able to render false if he's able to X, but that (ii) we have concluded he is unable to render false.

It is really nice to see you here, Charles! Just some random notes, and so far I haven’t read the other comments but I’m about to do so and hope to comment again later.

There are sound versions of the consequence argument! Are we not compatibilists?

You’re right that there are valid transfer principles used in the consequence argument (the render-false versions seem valid) and that they are not subject to the kinds of counterexamples you consider (which suggests that the consequence argument is a stronger, better argument than the direct argument).

Maybe this relates to Nate’s point (OK I peeked a bit). Here’s what I would say.

The consequence argument comes in many varieties. Some versions beg the question, some versions use a faulty transfer principle (you discuss many of those versions in this post), some use (according to you, and I would agree) valid transfer principles. Yet these latter are faulty in some other way, for instance they make faulty assumptions about the necessity of propositions about the past and/or the laws of nature.

On some versions of the consequence argument, valid inference rules are used and on other versions assumptions about the past and the laws of nature are true, but there are no (non-question begging) versions of the consequence argument where both valid inference rules are used and the key assumptions are true.

What do you think about that?

Joe,
I agree. My concern is that too frequently we assume that to do otherwise than render a proposition true requires the capacity to render that proposition false. This is a false dilemma that is at the heart of many versions of the Consequence Argument.

Nate,
I don't know what to say at this point that won't get us back in a circle. I greatly appreciate your comments they have significantly helped me think through these matters.

Thanks, Charles. This sounds right. (I tried to post something last night but I think it failed to send; I hope this isn’t a double post.)

If we agree that some CA transfer principles are valid (Alexander Pruss 2013, for instance) and we want to remain compatibilists, we’d have to find trouble with one of the two key assumptions, roughly: (1) no one is able to render any law of nature false; (2) no one is able to render any proposition about the past false. Both of these assumptions are questionable, though. On the one hand, Humean compatibilists question (1); on the other hand, the possibility of time travel seems to refute (2).

My Analysis paper provides a response to (2), as well. In that paper, I argue that formal versions of the consequence argument depend on assumptions about the remote past – the past prior to the existence of human beings. This is, for instance, explicitly conceded by Pruss, for his intuitive gloss of this assumption is characterized as “there is no choice about the pre-human past” (2013, 430).

Even critics of my argument seem to admit this much. Weak incompatibilists, who note that my criticisms don’t remove the threat of determinism for human beings since human beings have a remote past, can accept this much. For instance, in her August 22, 2013 post on this issue, Carolina Sartorio writes:

“My reaction to Joe’s objection to the consequence argument (a reaction that I share with others, although not with everyone, by all means) was, and still is, that we shouldn’t conclude that this is a problem for incompatibilism (or for incompatibilists that rely on the consequence argument) but, rather, that incompatibilism is a weaker thesis than we have typically taken it to be. And this is because the threat to our freedom is not determinism per se, or not just determinism, but determinism plus something else. Let’s call this claim ‘the demotion claim’ about the problem of determinism and free will.”

But if you accept the demotion thesis, then the past was never a threat to free will; it was the remoteness of our past that was the threat all along. In a recent paper (2015; available on her website), Carolina argues “that the threat to our free will isn’t determinism but determination by factors beyond our causal reach.”

If this is the case, the worry is not and never was about the past. The worry is more like the worry you might have after missing an important faculty meeting where certain policies were passed. You are destined to live under those policies, even though you had no choice about the matter. This isn’t a worry about the past. Nor is it clear how determinism is relevant since in the case I described you’d worry just the same whether or not determinism were true.

Charles,

At the risk of going off-topic from your intriguing strategy, I'd like to register my objection to the premise "Nobody can render ... propositions about the distant past false." I don't think it's true, even supposing that time travel is impossible and there are no closed time-like curves. I think the best understanding of physics we have implies we can and routinely do render (some select) propositions about the past false. For ironclad physical reasons, we can never *know* which propositions we rendered false: http://journals.aps.org/pre/abstract/10.1103/PhysRevE.89.052102 ; and for marginal relevance and lots of fun see http://phys.org/news170586562.html. But that's a different story.

The comments to this entry are closed.

Books about Agency


3QD Prize 2014: Marcus Arvan