Blog Coordinator

« The collective noun for libertarians | Main | Totally awesome conference at USF »



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


Just a quick thought about the "p and no matter what anyone decides, p" interpretation: isn't the following a counterexample to beta-box so interpreted?

Let p be a proposition describing some consequence of a decision that I make. Then it will be true that NPL and, given determinism, it will be true that □(PL → p), but clearly false that Np, since the truth of p clearly does depend on what I decide.

But maybe you're thinking of a different understanding of the phrase "no matter what anyone decides"? I was just thinking of it in counterfactual terms, so that in this case it would have the false implication that the consequence of my decision would still have occurred even if I hadn't made that decision.

Neal, thanks for playing along! OK, let’s try a specific example. You're at the ice cream store and you decide to have chocolate, not vanilla. Let p = the ice cream clerk hands you a chocolate ice cream (a consequence of your decision). As you suggest, Np seems false, since p is true but it seems p would not have been true if you had decided to have vanilla. NPL is true. And if determinism is true, then Box(PL --> p). Good.

But doesn’t Np seem to come out false on the original interpretation too? p is true (the clerk hands you chocolate) but it seems you are partially responsible for the fact that p. After all, if you had decided to have vanilla, then p would have been false (he would have handed you vanilla). What am I missing?

Hi Eddy,

Great stuff to think about here. I’m worried a little bit about the need for temporal ordering and indexing along the way. So, here’s a purported counterexample to beta-box on the “p and no matter what anyone decides, p” interpretation of N.

Suppose I’ve just tossed a coin and it has come up tails at t. If p = the coin comes up tails at t, then now (after t) Np is true. Let q= the coin was tossed prior to t. I think this means that box (p→q) is true. But I feel no pull to accept Nq.

I gots lots more… but I’ll be patient.

Eddy, I don't know why you say about your ice cream example reply to Neal that NPL is true. It's clearly not true IF determinism is true. Had you decided to have vanilla, then either the laws or the past would have been different, so PL would have been false. Thus it's not true that the past and the laws would have been the same no matter what anyone decides.

So, though the "no matter what anyone decides" understanding of N may not make beta subject to counterexamples, it makes NPL clearly false in deterministic worlds.

Stephen, I'd be happy if I'm suggesting an argument that suggests we should give up NP or NL (or NPL)--I'm a compatibilist who thinks we need the ability to do otherwise and cashes that out in terms of backtracking counterfactuals.

So, if the ice cream example suggests that NPL is false on the 'no matter what anyone decides' reading, I'd ask why it doesn't suggest it is false on the 'no one is even partially responsible' reading. As you say, had I decided to have vanilla (in a deterministic universe), then PL would have been false. If that is the case, then isn't it also true that I am partially responsible for the fact that PL?

Dan, Fritz does not allow the truth-makers of q to occur before the truth-makers of p, so I think your counterexample betrays that stipulation. If we allow it, then wouldn't Nq also be false on the 'no one is responsible' reading?

I'm puzzled how you are using 'decides' here, Eddy. Perhaps you mean something along the lines of (1) does X as the result of a deliberative process comparing X and Y. In that case, isn't the relevant version of the CA involving a 'decision'-version of Beta-box clearly false?

Assume deternimism. The way the world was in the distant past, plus the laws of nature--neither of which were the result of my doing anything at the end of a deliberative process--entail that I ordered the pizza, rather than the salad, at lunch today. But clearly my ordering the pizza, rather than the salad, at lunch was something I did at the end of a deliberative process. No?

If, on the other hand, you mean something much more robust by 'decides' (such as deserving of praise or blame for what is done at the end of a deliberative process), then I think we haven't moved so far from the original discussion of beta-box.

I think the same is also true of your third example, depending on how we understand the 'should' and perhaps what you mean by 'pride' as well.

"Dan, Fritz does not allow the truth-makers of q to occur before the truth-makers of p, so I think your counterexample betrays that stipulation. If we allow it, then wouldn't Nq also be false on the 'no one is responsible' reading?"

Yup. That looks right.

OK, I was hoping to get some uptake with the more general framing of the issue (the proves-too-much strategy), and I'd be happy to hear what people think of that [I hear crickets chirping] ... But in the meantime, the cases being used to question my alternate readings of Np are making me wonder why anyone should have accepted Beta-box to begin with. [Warfield says, “I have no argument for the validity of Rule Beta Box. I can think of no example which demonstrates that it is not valid nor have I found anyone who can provide such an example” (1996:216).]

Let q = the waiter brings me a pizza at time T [or for moral responsibility cases, let q = CARE receives $500 for famine relief at time T; or let q = my son breaks his arm at T].

Suppose just a bit before time T I ordered a pizza [or clicked on a box to give $500 to CARE or failed to keep hold of my son’s bike as long as I should have when he was learning how to ride].

Suppose determinism is true (surely that’s allowed if we’re testing Beta-box, right?).
Let Np = p and no one is even partly (morally) responsible for p (Warfield’s reading).

NPL seems true enough (unless, perhaps, one uses a counterfactual reading).

Box(PL-->q) is true (since determinism is true).

But Nq simply seems false.
I *am* partly responsible for the waiter’s bringing me a pizza at T.
I *am* partly (morally) responsible for CARE’s receiving $500 at T.
I *am* partly (morally) responsible for my son’s breaking his arm at T.

If I am begging the question, why (and how would I avoid doing so)?
If I am not, what’s wrong with these examples? (Fritz?)

Hi Eddy--

I'll bite. I take it that with this and your other examples that incompatibilists favoring N would say that while you think that you are responsible--and that itself is determined of course--since the conditions you contributed themselves were determined *for* you by determinism and not *by* you in any way but the prevalent psychological sense that determinism has assigned us (through socialization and the like), then our resultant being-out-of-the-loop-as-agents-in-any-final-and-full-sense means we aren't responsible for anything despite thinking (as we must) that we are. And I suppose then that the incompatibilist could say that ignoring the force of N and taking our everyday sense of agency as real (that is itself determined by hypothesis) is question-begging.

But I have real doubts about N too, and I like your way of putting it as perhaps proving too much. Since N as the basis for a transfer of any property is mounted on Box(PL-->q), then the latter seems to eliminate any other supposed powers metaphysically because deterministic/causal power drives everything. If I took Np to be "no one is even partly accountable for any aesthetic features of p", and aesthetic features--good or ill--are attributed to all Ls and Ps (as is plausible), then we could show that a brilliant artist is not accountable for the aesthetic features of a terrific work of art despite the fact that she created it. That tells me (at least) that the CA builds incompatibilism on the funeral pyre of a helluva lot of attributions of properties and powers that interest us above causal power.

"1. If Beta-box is valid, then determinism entails lazy fatalism.

2. So, incompatibilists either (a) have to give up Beta-box, or (b) argue that determinism entails lazy fatalism."

If by "beta-box" you mean resentment-beta-box (for "p and and I'm able, in the sense required for resentment to be deserved, to do something that might render p false"), then I reject premise 1. If by "beta-box" you mean badness-beta-box (for "p and I'm able, in the sense required for badness-judgments to be true of me, to do something that might render p false"), then I reject premise 2.

I would reject badness-beta-box by citing ordinary cases. Orindary cases don't convince me that resentment-beta-box is invalid.

Sorry, if you mean "badness-beta-box" by "beta-box", then I would give up beta-box.

Is there a reason it would be unfair to offer the following interpretation of Np?

Np = p and p regardless of the existence of any human being

This interpretation seems to me to yield undeniably false conclusions when plugged into the argument originally provided by Eddy, or at least much less controversially false conclusions than we've seen so far. At the same time, this version of Np avoids reference to psychological factors and/or normative issues that seem to be muddling the debate -- e.g. what precisely do we mean by "decides", etc.

Is my view askew?

I should amend my comment. Clearly, the past would not be the same regardless of whether or not humans existed. The following interpretation better captures my line of thought:

Np = p and p without any humans doing anything whatsoever

Forgive my earlier sloppiness.

Alan (and others), keep in mind that the examples I was using in my 7/31 post were not meant to offer a compatibilist response to the *consequence argument*. Rather, they are meant to offer counterexamples to Beta-box. So, they should be assessed simply based on the way the relevant terms are best analyzed--in this case, the commonsensical idea of being "partially morally responsible".

To test the principle, we can assume determinism, such that Box[PL-->q] is true [or if you wish we could say we have no choice about whether determinism is true?]. We can assert NPL (if we can't then the CA will fail). And we then ask whether Nq is true--am I even partially responsible for CARE's getting $500 or my child's breaking his arm? [I think we could ask similar questions using a 'choice' reading of Nq, or a 'resentment' or 'badness' reading, as Nate proposes.]

I don't see how Nq comes out true without simply assuming incompatibilism (i.e., without already thinking that there's something about the truth of NPL and determinism together that should lead us to say something quite remarkable--that no one is even partially responsible for anything). Unless we're already thinking in terms of incompatibilist arguments, why else would we possibly want to say that, just because something benign like NPL is true and just because determinism happens to be true, it follows that I haven't done a praiseworthy thing for donating money for famine relief or that I haven't done a blameworthy thing for letting my child's bike go to look at my cellphone??

And just to stoke the flames of this burning conversation ;-)
I will assert that I think the same moves are going to work against *any* Beta (or Transfer) principle. It's the incompatibilist who begs the question by saying these principles cannot be countered by using examples of ordinary human choices in a deterministic universe.

This is an interesting argument. One thing to note is that there are a family of "proves too much" strategies. In several papers (most unpublished), I've tried to show that the same stuff used to prove incompatibilism via the CA can be used to prove free will skepticism (or denialism). See my 2010 Analysis paper, for instance.

Just a few other comments. First, Warfield 1996 is NOT a version of the CA. It is a version of the Direct argument. The difference is in the use of the operator (which unfortunately uses the same letter 'N'). Warfield's 1996 definition (which follows van Inwagen's original 1980 presentation of the Direct Argument) is in terms of responsibility whereas the CA is strictly an argument about the incompatibility of determinism and freedom.

Here's where the point matters: Responsibility-transfer principles are suspect in ways that freedom-transfer principles are not, especially if freedom is understood in the classical sense as the ability to do otherwise. It is one thing to say that the inability to do otherwise transfers (there is no known counterexample to Beta-box, IF the N-operator is understood as the no-choice operator (van Inwagen 1983; Widerker 1987)); it is another to say that non-responsibility transfers. Fischer is being generous when he says that the latter debate leads to a dialectical stalemate.

Lastly, lots of incompatibilists (Pereboom, for instance) are compatibilists about deliberation. So it isn't surprising to see them reject transfer principles when "Np" stands for "p and no matter what anyone decides, p," as Neal demonstrates above.

Joe -

Doesn't Van Inwagen explicitly say (both in EOFW and "How to think about the problem of free will") that he means to be talking about the ability to do otherwise "in the sense required for moral responsibility"? If inability of that sort transfers, then non-responsibility transfers also, right?

Eddy -

First, a side point...I find it difficult to respond to your Nq because I don't know how you're using the expression "morally responsible". If I count as being morally responsible, on your usage, if, say, it would be appropriate to stop trusting me on account of my wrongdoing, then I agree that ordinary cases are counterexamples to the relevant beta-box principle. If you mean that I'm MR only if (a certain hostile species of) resentment is appropriate, then I accept the relevant version of beta-box. (If you mean "what it means in English" then I have no opinion.)

I'll presume that we're dealing with resentment-beta-box (RBB), the beta-box principle for "p and nobody is able, in the sense required for resentment to be appropriate, to do something that might render p false". You ask why we would want to hold that Nq is true if we aren't already thinking in terms of incompatibilist arguments. What's wrong with just accepting this point? Sure, in the absence of incompatibilist arguments, there's no reason to accept Nq. But the incompatibilist needn't hold that Nq is obvious or that one should accept it without an argument...that's what she provides with the CA, on the basis of RBB, or with an informal version thereof. I'm not sure what it is to "beg the question", but if it's an objection to the fact that somebody argues on the basis of a principle that simply encodes her basic (alleged) insight, then I'd be untroubled by the charge.

The non-lazy-fatalist-incompatibilist need not hold that ordinary cases cannot, in principle, convince us that certain beta-box principles are invalid. Such an incompatibilist must reject the lazy-beta-box principles (whichever ones would lead to lazy fatalism if they were valid)...and I would think she would explain why she rejects them by offering up ordinary cases, recognizing that the lazy fatalist won't be convinced. So I don't think she should hold that you're making some dialectically inappropriate move by using ordinary cases in explaining why you reject RBB. I just think she should hold that you're mistaken in thinking that the cases are counter-examples.

I think the general “proves too much” strategy is very interesting (and, in general, represents a cool philosophical move).

But in order for the reductio (presumably against beta-box on the standard interpretation of N) to be compelling, I think, the lazy fatalist claim has to go beyond the standard conclusion of the consequence argument (that no one could have done anything other than what they’ve done) without being entailed by that conclusion. Otherwise, the defender of the consequence argument will just say that your complaint isn’t an objection—it’s just the argument. So, does determinism + beta-box entail something that is worse than “no alternatives” and not already entailed by the CA under the traditional N?

When Np = p and it’s useless for anyone to try to do anything to prevent p, then I’m not sure the conclusion goes beyond the conclusion of the standard CA. That is, it is reasonable to think that we can infer “it is useless for anyone to try to prevent p” from “no matter what anyone does, p”.

Now, when Np = p and no matter what anyone decides, p, then (in the spirit of Neal and Kevin above), I don’t think Nq follows from the premises. That is, beta-box does not license a valid inference on this version of the N operator. For beta-box to be valid on this interpretation of N, it would have to be true that determinism functions by way of what someone has called “by-passing”. But that particular someone has fairly persuasively argued that we have no reason to think that determinism entails this by-passing.

I don’t know what to say about Np = p and no one should take any pride in the fact that p. I could see the defender of the CA going either way with this one, depending on how much is built into “taking pride” in something, i.e. how morally loaded this concept is supposed to be.

To reiterate, then, I’m not yet convinced that too much has been proven either because what’s proven isn’t too much or because what would be too much hasn’t been proven.


Surprisingly, the answer is "no."

For a related explanation, see Fritz Warfield's "When Epistemic Closure Does and Does Not Fail," Analysis 64 (2004): 35-41. Warfield disproves the claim that "if a necessary condition on knowledge fails to be closed under entailment, then knowledge is not closed under entailment" (36).

You are making about a related claim: "If inability ... transfers, then non-responsibility transfers" = If non-responsibility (necessary for inability) fails to be closed under entailment, then inability is not closed under entailment.

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan