J'Accuse!

Actually, Richard Double accuses--I'm just the messenger.  But I've always wondered how libertarians might reply.   In his essay The Moral Hardness of Libertarianism Double writes:

I propose four premises to argue that most libertarians are hard-hearted (unsympathetic, not morally conscientious). (1) Libertarians believe that we may hold persons morally responsible only if they exercise libertarian free will. (2) Libertarians believe that we should hold persons morally responsible. (3) Most libertarians believe that we have scant epistemic justification that persons have such free will. I believe, but shall not argue in this paper, that libertarians who believe they have epistemic justification for libertarian free will are mistaken.  I add a fourth premise describing how soft-hearted persons would respond to accepting the first three premises, and conclude that those libertarians who do not believe they have epistemic justification that persons make free choices are hard-hearted.

Here's my version of the argument:

1. It is hard-hearted to think that someone should suffer punishment if we have little epistemic justification for the belief that he deserves to suffer. 

2. If an agent is not morally responsible for committing a bad act, he does not deserve to suffer punishment.

3. If an agent does not possess libertarian free will then he cannot be morally responsible for any act

4. We have little epistemic justification (EJ) for the belief that agents posses libertarian free will.

5. Therefore we have little EJ for the belief that agents are morally responsible.

6. Therefore we have little EJ for the belief that any agent deserves to suffer.

7. Libertarians believe that certain agents should suffer punishment. (Those who commit terrible crimes, for example).

8. Therefore libertarians are hard-hearted.

Double's point is this: Even if libertarians, both of the event-causal and agent-causal varieties, have come up with coherent theories of free will, they still have not provided any evidence that their accounts are actual.  (Or if they have, the evidence is of a highly questionable nature.) That brings the epistemic justification for their belief in free will well below what would be necessary to confidently deem someone deserving of punishment.

I have two questions, really.  First, what say you, libertarians?  And second, can this argument be revised to apply to compatibilists too?

(To ward off one objection: yes, I know that if there is no free will then libertarians would not be morally responsible for being hard-hearted.  But they would still be hard-hearted, responsible or not.)       

Comments

[Fritz Warfield]

This libertarian would take the argument more seriously if he was presented with some support for the key premises of the argument. In your formulation, #4 could use some support couldn't it?

And in the Double quote, what *kind* of claim is his claim 3? Has he surveyed the libertarians on this issue? What were the survey results? I wasn't included in the surveyed sample, and I'm guessing few other professional philosophers who are libertarians were in the survey either. But perhaps his survey data will disconfirm my anecdotally based speculation?

[Tamler Sommers]

That little section after my formulation was meant to be in support of premise (4) but obviously it needs some fleshing out.

Consider a libertarian theory of free will like Kane's. Set aside concerns about whether his account really would justify attributions of robust moral responsibility, and focus only on the evidence (or lack thereof) that the indeterministic events in his account really occur--and at those crucial junctures. If they do not occur, then by Kane's own lights no one could be ultimately responsible for their actions. So now we have to determine the probability of them occurring. According to Double, if the probability is below, say, .5, then we should certainly behave as though people are not morally responsible, at least with respect to things like retributive punishment. Otherwise, there is a greater than 50% chance that we are punishing someone who does not deserve to be punished.

Is the probability below .5? Well, first of all, for all we know determinism is true. And even if it is not, Kane's account requires the occurence of indeterministic events at crucial junctures in the decision-making process, and the amplification of these events so that the choice (or SFA) itself is genuinely indetermined. Double would claim, and I guess I agree, that we could not assign anything close to a .5 probability for that.

It seems like this same line of reasoning could apply to agent-causal libertarian theories as well, even more so maybe.

(By the way, You're right that Double's remark about libertarians "believing" that they have scant epistemic justification is a little odd. Interpreted charitably, maybe he means that when libertarians provide an account of free will, but fail to provide reason to believe that their account is actual, they are implicitly admiting that they lack epistemic justification.)

[Fritz Warfield]

A libertarian might, it seems, have justification for the belief that some libertarian theory is correct, without having justification for the belief that, eg, Kane's particular theory is correct, or O'Connor's, or Nozick's or.... . Compare: I strong reasons for thinking that some externalist theory of knowledge is correct, though I'm not confident that it's Plantinga's and not confident that it's Goldman's and not ....

I can assign a probability of way lower than .5 to each particular libertarian (or externalist) theory but go way higher on the disjunction.

Different point: Some libertarians are surely willing to modus tollens here on determinism. We know we're responsible, responsibility entails indeterminism, so... but that's a different discussion.

[William A. Edmundson]

My impression is that libertarians (and the libertarian that dwells in each of us) would balk at Premiss 4 ("We have little epistemic justification (EJ) for the belief that agents have free will"). The thought that we have direct knowledge of the efficacy of our own willings dies hard (though Wegner's book should help).

But, aside from that, anyone (libertarian or not) would be wise not grant Premiss 4 without noting that the "little EJ" there might go a longer way if transposed into the context of Premiss 1. "Little," in other words, invites equivocations of the "small NBA forward therefore a small guy" variety. This suspicion would be reinforced by contextualist reflections. EJ in 1 pertains to forensic contexts, in which the putative standard is a very high one (e.g., "beyond a reasonable doubt"--but influenced by commensense suppositions about action and will) but not as high as in 4, which is a baldly philosophical context (in which those commonsense suppositions are themselves justificanda).

[Derk Pereboom]

Punishment -- in particular, punishment designed to satisfy the retributivist goals -- harms people. If one aims to harm another, the justification must meet a high epistemic standard. Suppose that if retributivism is true, the execution of some particular criminal is justified, but if it not, the most that is justified is a long prison term. My sense is that for us to be justified in executing the criminal our degree of confidence that retributivism is true must be at least 90% – in fact, I would say that it must be beyond reasonable doubt – and for this our degree of confidence that we have free will must be that high as well. So what degree of confidence that we have libertarian free will is epistemically rational? Even if the luck objection against event-causal libertarianism can be answered, we would need evidence of the right sort of indeterminacy in the brain to have a significant degree of confidence that this view is true. But we don’t have this evidence. To have a significant degree of confidence that agent-causal libertarianism is true, we would need to have a significant degree of confidence that the philosophical arguments against this view can be answered, and (I’ve argued) we would need a significant degree of confidence that what happens in the brain when free decisions are made is at odds with our best physical theories. We have no such evidence. Do we have positive, perhaps phenomenological evidence that we have libertarian freedom? Many form the belief that at various points of time they have more than one course of action available to them. But this would also be predicted if determinism were true, and we were unaware of how our actions are caused. As Spinoza observed, “men think themselves free, because they are conscious of their volitions and their appetite, and do not think, even in their dreams, of the causes by which they are disposed to wanting and willing, because they are ignorant of [those causes].” I think that it pretty clear that even given Fritz’s point, it would be epistemically irrational for us to have the very high degree of confidence that we have libertarian free will required to justify execution of the criminal. Given the evidential situation, there’s no way to get the rational degree of confidence up above 90%.

[Chris Franklin]

I think Warfield's response is of the right type. Moreover, were the argument to be more broadly applied to include compatibilists as targets, then the response could be strengthened. So one need not maintain that Fischer, Haji, Dennett, Ginet, Kane, or O'Conner's theory is correct, but merely that there is a rather high likelyhood that one is (roughly) correct.

I think the argument may also miss the philosophical prgression of many philosopher's thought. Many come to defend some form of libertarian agency becuase they are convinced that people are morally responsibile for their actions (thus deserving to suffer for their evil actions). Given that one could demonstrate the libertarian agency is necessary for moral responsibility, then one could conclude that we have rather strong EJ for holding people accountable for evil actions.

[christian]

"That brings the epistemic justification for their belief in free will well below what would be necessary to confidently deem someone deserving of punishment."

That doesn't seem to crazy to me. If I had to assign a probability to the proposition that humans have libertarian free will, I would say .5, or better, I just don't know.

So, it might be epistemically irrational to believe that some person deserves some punishment. It might still be practically rational though. I guess that will depend upon how valuable one thinks putting a deserving person behind bars is and how disvaluable giving him a sentence that is less than he deserves. Of course, punishment might just be justfied on other grounds, deterrence say...or something like this. If so, punishment can be justified even if it is epistemically irrational to believe the punishment is deserved. Is that so bad?

[christian]

"That brings the epistemic justification for their belief in free will well below what would be necessary to confidently deem someone deserving of punishment."

That doesn't seem to crazy to me. If I had to assign a probability to the proposition that humans have libertarian free will, I would say .5, or better, I just don't know.

So, it might be epistemically irrational to believe that some person deserves some punishment. It might still be practically rational though. I guess that will depend upon how valuable one thinks putting a deserving person behind bars is and how disvaluable giving him a sentence that is less than he deserves. Of course, punishment might just be justfied on other grounds, deterrence say...or something like this. If so, punishment can be justified even if it is epistemically irrational to believe the punishment is deserved. Is that so bad?

[Tamler Sommers]

Chris,

"I think the argument may also miss the philosophical prgression of many philosopher's thought. Many come to defend some form of libertarian agency becuase they are convinced that people are morally responsibile for their actions (thus deserving to suffer for their evil actions). Given that one could demonstrate the libertarian agency is necessary for moral responsibility, then one could conclude that we have rather strong EJ for holding people accountable for evil actions."

As a sociological observation about philosophers, I think you're right. (I'm almost sure this is the case for someone like van Inwagen.) But the question is whether this kind of reasoning is justified. What basis do these philosophers have for being convinced that people are morally responsible? If there is no basis, if it's just dogmatism, then the hard-heartedness charge could still stand. If there is some basis for that belief, then let's hear it.

Christian,

"Of course, punishment might just be justfied on other grounds, deterrence say...or something like this. If so, punishment can be justified even if it is epistemically irrational to believe the punishment is deserved. Is that so bad?"

That's a good point, which is why the charge is hard-heartedness rather than something stronger, like wrongful imprisonment. If punishment can be justified by consequentialist means (and most varieties of it can, I think), then the complaint is that libertarians are taking a certain satisfaction in seeing criminals punished. Someone who did not see the criminal as morally responsible would see the punishment as a necessary evil.

Derk and Fritz,

What is interesting about your exchange, it seems to me, is that it makes the debate something that is (in principle) resolvable. Fritz can line up the evidence for the belief that we make indeterministic choices, and we can evaluate whether or not he makes his case. I think Derk raises an important point, though: our INTUITION that we have more than one option available to us is of little use to the libertarian since that might very well be the case even if determinism were true. I recall reading a section of Laura Ekstrom's "Free Will" in which she describes a paradise island where we can have anything we want, but find out later that we had no "real" choice--that there was only one dinner option, one attractive person that you would want to ask on a date...etc. It was all determined, unbeknownst to us. Ekstrom's point was that we would feel unsatisfied once we discovered this, but there is another lesson to be drawn. The people on paradise island, according to Ekstrom, believed that they were making free choices, just like we do. So clearly, at least according to Ekstrom (a libertarian), we could easily believe we are making free indeterministic choices in a fully determined world.

[Tamler Sommers]

William,

I'm not sure I understand your last point--is it similar to Christian's? That the kind of epistemic justification we would need to think that someone should be punished is not as strong because of the pragmatic reasons we have to lock up criminals? If so, as I say in my previous post, I think you're right. But again, the question revolves around how we regard the criminals. Do we look at wicked people, as Darwin suggested, like sickly ones, to be pitied and controlled rather than hated. Or do we resent them and want them to be punished regardless of any pragmatic benefit. (In a case here in NC, a father of a murdered daughter said he was "counting the days" until the murderer was executed. While I can relate to to the sentiment, it does not fit within a view that denies robust moral responsibility.)

[Derk Pereboom]

Christian: So suppose that non-retributivist theories would justify long-term confinement, and retributivism would justify execution, and that the epistemically rational degree of confidence that we are free in the sense required for moral responsibility is not high enough to meet the standard for justifying execution. Could practical reasons for believing that we are free in this sense fill in what’s required? Imagine the criminal were to demand a justification for execution given that the epistemically rational degree of confidence that we are free is as low as it is. One might answer: “Still, our practical reasons to execute people like you are strong enough to justify our having a very high degree of confidence that we are free.” But given that we have long-term confinement as an option, it doesn’t seem that there are any practical reasons with the required strength.

Tamler asked whether compatibilists also face this sort of problem. Maybe everyone does. The epistemically rational degree of confidence that we are free in the sense required for moral responsibility would have to be very high – beyond reasonable doubt, it would seem – to justify executing the criminal. As philosophers, it is clearly reasonable to defend, with passion, the view that we have this sort of freedom. But at the same time our epistemically rational degree of confidence that we are free doesn’t put it beyond reasonable doubt. So due to our epistemic situation regarding free will, it looks like it’s wrong to execute the criminal on the basis of retributivism, and it seems wrong to advocate doing so as well.

[Chris Franklin]

Derk,

I was wondering whether you would be willing to flush out "beyond a reasonable doubt" as you are using it. Normally, what is beyond a reasonable doubt differs relative to the context it was uttered in. For instance, I would not take it to be reasonable to exempt someone from punishment since it is possible that a really clever alien performed that crime and planted evidence that pointed towards the guilt of the accused. However, many people do take this type of argument to be a defeater against the internalist theory of knowledge.

Tamler,

I agree insofar as such a philosophical progression needs to be justifed. But if justified, your argument fails to demonstrate that libertarians are hard-hearted.

[Manuel Vargas]

Fritz-

I'm wondering about a couple of the things you said earlier in the comments thread:

"I can assign a probability of way lower than .5 to each particular libertarian (or externalist) theory but go way higher on the disjunction."

That seems right, though now I'm wondering what the evidence is that funds the "way higher" probability that some form of libertarianism is going to work out. I hope it isn't this:

"Some libertarians are surely willing to modus tollens here on determinism. We know we're responsible, responsibility entails indeterminism, so..."

This always looked to me like a terrible way to argue for a claim with implications in physics and maybe other sciences (and I say this knowing full well that van Inwagen, whom I find presuasive on many things, runs this sort of argument). Even if we did somehow know that we were responsible, I don't see why we should think that this is the kind of knowledge that would tell us anything about physics, or why even this in conjunction with arguments about the meanings of CDO, free will, etc., should tell us anything about physics.

But you didn't really advance that argument, so maybe you aren't the person to answer those questions. But it sounds like you may have a view about whether or why we *know* we have moral responsibility, independent of debates about free will. What do you have in mind? I'm just wondering (and for once, not trying to pick an argument), because I've always wondered what libertarians have in mind when they say these sorts of things and the literature seems to be a bit thin on this point.

[Derk Pereboom]

Chris: I had in mind the legal standard, and the legal standard doesn't require ruling out the alien hypothesis. But the view that we lack free will of the libertarian sort, as well as the view that we lack the sort of free will required for moral responsibility whether compatibilism or incompatibilism is true, is a lot more plausible than the alien hypothesis. Even my compatibilist-leaning colleagues think that the no-free-will position is at least 20% probable.

[Fritz Warfield]

Manuel -- right,the two points were independent. The sort of transcendental move sketched is a different way some might argue for indeterminism.

There are many other ways the other sort of position could obtain (the one where I'm well below 50% on each member of a set of particular libertarian theses but way way higher on their disjunction). One would be where I have an argument that generates great confidence that we have free will and that compatibilism is false. I assume I'm not being asked to give the argument here? If I am being asked to do that my quick reply is -- I don't have a favorite argument at all and certainly not one I can post in the 5 minutes I've got to spare right now.

On the issue of whether we could or do know that we're responsible independent of "debates about free will", recall that whether responsibility requires *freedom* is nearly as controversial a thesis as whether responsibility requires indeterminism. Assuming the controversies make sense, this suggests that at least a conceptual separation of these issues is workable. I think that responsibility does require freedom though, so the dependence is pretty plausible.
***Note, however, that even if freedom is necessary for responsibility, we might not need to know we are free in order to know we're responsible: knowledge that P doesn't always require knowledge that Q even where Q is strictly necessary for P.

sorry if I typed too quickly here to be my usual clear self.... more later when I've got more time.

[christian]

Hi Derk,

A response to [One might answer: “Still, our practical reasons to execute people like you are strong enough to justify our having a very high degree of confidence that we are free.”]

I want to say our practical reasons provide us with no epistemic justification for having a very high degree of confidence that we are free. They do provide us with practical justification for believing we are free, though. Simply because it feels better to believe that loved ones freely love us, consequences of our actions are consequences of free actions, etc. But, they just might provide us with moral justification for long term imprisonment. This last part would have to be spelled out clearly though.

I also think I agree with Tamler's point that if libertarians are taking a certain satisfaction in seeing criminals punished, then they are worthy of a certain kind of criticism (If this is the point). Their satisfaction, although it might be practically rational, i.e. it feels good to think justice has been done, it is still an epistemically bad attitude to have. Perhaps satisfaction ought to be coupled with an attitude of pity, pity for an individual who will spend her life in prison for something which might have been beyond their control. If both of these attitudes are reasonable attitudes to have, then I suspect that as a matter of contingent psychological fact, people that have these attitudes together will feel conflicted about feeling both towards the criminal. Something seems right to me about feeling this conflict, that it is both the epistemically and practically rational conflict to endure. It shows a kind of sensitivity that is good on all grounds.

[Bill Edmundson]

Tamler,

I could have made my point clearer by saying that the argument is invalid because of a fallacy of equivocation. The key stops in the argument are these:

D1. It is [cruel] to think that someone should suffer punishment if we have little epistemic justification (EJ) for the belief that he deserves to suffer.

***

D4. We have little epistemic justification (EJ) for the belief that agents possess libertarian free will.

***

D6. Therefore we have little EJ for the belief that any agent deserves to suffer.

***

D8. Therefore, libertarians are [cruel].

I say “cruel” instead of hard-hearted to emphasize that the argument is an accusation of vice, not of mere insensitivity, and to connect the conclusion with legal/political/moral strictures against cruelty (e.g. the eighth amendment).

The term “little EJ” has a meaning in premiss 1 that differs from its meaning in premisses 4 through 6. This is the equivocation. Epistemological contextualism is the basis for the claim that “little EJ’ is equivocal in a way that vitiates the Double argument. The argument is this:

1. Epistemological justification (EJ) is a context-relative concept (as Keith DeRose has argued).

2. The degree of EJ an utterance enjoys is relative to the epistemic standard (ES) pertinent to the context in which the utterance is embedded.

3. Philosophical contexts impose the highest standard of EJ; conversational contexts impose a low standard of EJ; punitive/forensic contexts impose a high but not the highest standard of EJ.

4. In Premiss D1, the phrase “little EJ” occurs in a punitive/forensic context, and thus implicitly refers to ES:high.

5. In Premiss D4, the phrase “little EJ” occurs in a philosophical context, and thus implicitly refers to ES:highest.

6. What is little relative to ES:highest need not be little relative to ES:high.

7. D8 cannot validly be detached from the conjunction of D1, D6, and D7.

Derk Pereboom makes the case that the ES in D1 should be ES:highest. I sympathize, but that is not how “proof beyond a reasonable doubt” is understood in the law. Courts have consistently rebuffed any attempt to give a statistical gloss to the concept. A typical paraphrase is “to a moral certainty.” Proof beyond a reasonable doubt has never been understood to displace the “folk” presumption that defendants are normally capable of voluntary action. In other words, the ES pertinent to forensic/punitive contexts is high but not the highest, in part because it does not question the folk presumption that we are typically free.

Where capital punishment is a possibility, I am inclined to agree with Derk Pereboom: the ES should be ES:highest. But what about lesser punishments? Isn’t it cruel to lock someone up for term of years or even months for something they had no control over? It is–unless the relevant ES is less than the highest. And I think there is some reason to think it is. Consider praise. Suppose you are appointed trustee for a foundation that awards prizes to the most praiseworthy achievement in some field of endeavor. Suppose also that it would be frivolous (“soft-headed”) to give awards to the undeserving. A Double-style argument could be constructed to show that libertarians are frivolous. But in that case, I think, we would readily accept the contextualist point about the first premise in the argument: “It is frivolous to think that someone deserves a prize if we have little EJ for the belief that he deserves to be praised.” The term “little EJ” here implicitly refers to ES:low. So the libertarians (and the rest of us who think people sometimes deserve prizes) aren’t proven to be frivolous. When prizes are being doled out, we don’t fuss as much about EJ as we do when punishment is involved–rightly so. And so we insist upon a higher and higher ES as we move from praising to blaming to jailing to hanging. But, in light of the contextualist’s point, it takes further argument to establish that there is an invariant ES standard of ES:highest that governs across all contexts.

[Derk Pereboom]

Bill: Your contention about different epistemic standards for praise and for blame seems right to me, as does most of the rest of what you say. One issue, on: “Derk Pereboom makes the case that the ES in D1 should be ES: highest. I sympathize, but that is not how proof beyond a reasonable doubt” is understood in the law. Courts have consistently rebuffed any attempt to give a statistical gloss to the concept.” I don’t want to claim that “proof beyond reasonable doubt” can be given a statistical gloss, or a statistical analysis, and I don’t want to deny that this notion is context-sensitive. But what I said does imply that below a certain degree of confidence, which can be described with a percentage, that standard hasn’t been achieved – and I think that this is true no matter what the context. We can all agree that no matter what the context, if our degree of confidence that a statement is true is only 50%, the standard hasn’t been achieved. My sense is that if our degree of confidence is only 90%, it hasn’t been achieved, no matter what the context. But now we’re just haggling over the numbers.

[Tamler Sommers]

Regarding the question of whether this argument could apply to compatibilists--Derk, you said perhaps it could apply to everyone. The problem is that premise 4 would not apply to most varieties of compatibilist freedom. We have plenty of epistemic justification for the belief that we can exercise higher order volitions, or that we act according to reasons. What we lack (in my view, of course) is justification for the belief that these varieties of freedom can ground strong accountability--the kind of accountability required to be deserving of blame or punishment. So for a paper called "The Moral Hardness of Compatibilism" one would need to show that compatibilists have failed to provide not evidence, but some sort of conceptual link between compatibilist freedom and the deserving of punishment. And if they fail to provide this, and at the same time endorse retributive punishment or blame of any sort, then maybe the hard-heartedness charge could stand.

[Chris Franklin]

Manuel,

I had a question about the following comment: "This always looked to me like a terrible way to argue for a claim with implications in physics and maybe other sciences (and I say this knowing full well that van Inwagen, whom I find presuasive on many things, runs this sort of argument). Even if we did somehow know that we were responsible, I don't see why we should think that this is the kind of knowledge that would tell us anything about physics..."

Let's assume that the following argument is true.

P1 If some agents are morally responsibile, then indeteriminism must obtain at some level in the world
P2 Some agents are morally responsible
C1 Therefore, indeterminism must obtain at some level in the world.

This argument appears to have implications for physcis and other sciences. Further, I cannot ascertain why it would be unattractive for an argument that employs knowledge about physcis to be thought to have implications about physics. Could you point out exactly why this is a terrible way of arguing? I might be missing something, but it seems that if the premises of the above argument are true, then it does (even though it might be a terrible wat of arguing) have implcations for the physicist. Thanks

[Tom Clark]

Tamler wrote:

"So for a paper called "The Moral Hardness of Compatibilism" one would need to show that compatibilists have failed to provide not evidence, but some sort of conceptual link between compatibilist freedom and the deserving of punishment. And if they fail to provide this, and at the same time endorse retributive punishment or blame of any sort, then maybe the hard-heartedness charge could stand."

I couldn’t discover any such link in Michael Moore’s lengthy book Placing Blame, in which he tries to defend retributivism while eschewing libertarian freedom. For what I see as Moore’s intransigent refusal to soften his retributivism in the face of what he admits is the full causal determination of the agent (including reasons-responsiveness and higher-order volitions) I called him a “hard compatibilist,” where “hard” pretty much means hard-hearted. See http://human-nature.com/nibbs/03/twclark.html .

Since moral desert, according to the retributivist, doesn’t pertain to any consequential consideration, it can only be triggered by a quality or characteristic of the individual which must be addressed simply for what it is, not in order to change it. This characteristic, I think, can only be that of being a self-caused, ultimate originator that’s done wrong. This characteristic won’t change, nor is it intended to change, when supposedly libertarian agents are punished – suffering is to be undergone because its deserved, not because of any consequences. But for Moore and other legal compatibilists espousing retributivism, libertarian agents don’t exist. So how do they get to retribution? Beats me.

[Manuel]

Fritz-

Thanks for the reply. It would be fun to chat about this at length, but I suspect that short blog posts aren't the way to go on this sort of thing. Maybe I'll see you at a conference this year and we can chat?

Chris-

Regarding your argument, it seems to me that the problem is with the confidence one has to put in P1 & P2 (especially). I don't see how these sorts of claims (presumably arrived by lots of intuition pumping, linguistic analysis, philosophical argument, and the like) have (1) the right evidentiary basis for telling us anything about physics and (2) could be justified enough to tell us truths about physics from the armchair. It looks like I'm supposed to think that my intuitions and analysis of language is enough to tell me about how the physical world is organized. But why think that? Why not just think that they tell me (1) how I happen to think about responsibility and (2) the way my concepts presume certain things about the world? But that they presume certain things doesn't yet give me evidence for thinking the world is that way.

Maybe here is another way to get at the point, though relying more on my naively scientistic inclinations: Suppose that tomorrow the best arguments in favor of your P1 and P2 are published in the Phil Review in a co-authored paper of the greatest incompatibilist minds alive. Let us suppose that they also go on to endorse C1. Now let us suppose that tomorrow a consortium of physicists from Caltech, Berkeley, and Stanford announced that they had determined that the world was deterministic- Let us suppose that they have both theoretical and experimental results that are, by their lights, conclusive. What group should we believe about whether or not the world is deterministic?

Maybe there is reasonable disagreement about this point, but I can't help but think that more philosophers would believe the physicists than physicists would believe the philosophers. And that seems right. Why? Well, I could point to the history of bad a priori speculation by philosophers about empirical matters. I could point to various epistemic virtues of what physicsts are up to compared to what we are up to. But for the moment, I'm content to say that I know of no reason to suppose that the evidence we have for (P1) and (P2) outstrip the epistemic standing of experimental and theoretical work of the sciences, in this context.

So, if there was some way to secure (P1) and (P2) in a way that engendered considerably more confidence than most controversial philosophical theses (say, something on the level of or better than the experimental and theoretical evidence mustered by physicists), then yeah, I'd probably buy the conclusion. But I haven't seen anything like that yet.

I don't think this is unfair to the incompatibilist or philosophers in general- even van Inwagen has indicated that he'd give up his incompatibilism (P1) if he was convinced the world was deterministic- though (interestingly) he takes the our moral responsibility (P2) to be better grounded than the claims of physics, if I remember correctly. Van Inwagen experts and incompatibilists are urged to chip in with corrections and complaints!

[Manuel]

Oh yeah, I should note that the physicist thought experiment is borrowed from some of Fischer and Ravizza's work (though they put the argument to use in a slightly different context, and one that I've even argued against!).

[James Gibson]

Manuel,

I'm not sure why we need to give so much stock to what the physicists at these institutions say about determinism. It is one thing for them to say that we observe regularities that give us a very high degree of confidence in that if one event occurs, another probably will. But how could they ever justify on *empirical* grounds that one event *entails* another?

If you're as doubtful as I am about our ability to know if determinism is true from mere empirical investigation, the fact that many philosophers would be persuaded by the physicists doesn't have much force, with me anyway. Maybe that says more about me than the argument. And even though many physicists would be suspicious of philosophers who make claims about the natural world from a priori reasoning, philosophers can equally be suspicious of the grand claims about the natural world by those who do the empirical work (I have in mind the many developments in the history of cosmology). So, I don't see why the libertarians who argue from P1 and P2 do not have just as good (if not better) arguments than those who try doing physics to determine if we have libertarian freedom.

[Mark Smeltzer]

I think James' criticism of the idea of scientific/empirical endeavors to provide justification for the claim that determinism is true (or false) works only if we reject inductive logic as a valid form of reasoning.

James said that he is unsure "how could they ever justify on *empirical* grounds that one event *entails* another?"

Transform this question into the form: how could P be justified on empirical grounds if P entails that all things of type S are B?

If P is an empirical claim and the set of all S type objects is epistemically inaccessible to us, then how could we be justified in believing that all objects of type S are B?

This is just another way of positing Hume's rejection of inductive knowledge. (The revised form of James’ worry fits both the raven problem and James’ question regarding justification of belief in determinism.) Consequently anyone who rejects Hume's rejection and believes that inductive logic is a valid form of reasoning will have to accept that it is at least logically possible to inductively justify the belief that determinism is true (or false), and thus reject James' criticism.

So, those who accept the validity of inductive reasoning will respond that, “It is indeed impossible to deductively argue that determinism is true/false on empirical grounds, since that conclusion requires attributing a description to the totality of an epistemically inaccessible set. However, that is not what scientists are trying to do. Rather, their arguments are inductive in nature, and they are trying to provide justification for the belief that determinism is true/false in an inductive manner, which is a perfectly reasonable goal (even if it should prove practically impossible).”

[Derk Pereboom]

Tamler: You say: “So for a paper called "The Moral Hardness of Compatibilism" one would need to show that compatibilists have failed to provide not evidence, but some sort of conceptual link between compatibilist freedom and the deserving of punishment.” You’re clearly right that what’s required of the libertarian here is different from what’s required of the compatibilist. But maybe what the compatibilist needs to provide isn’t a conceptual connection. Perhaps, (more generally) for compatibilism to be true, a metaphysical or constitutive connection between compatibilist freedom and moral responsibility is sufficient. So maybe the link need not be the sort that makes ‘bachelors are unmarried’ true, but rather the kind that makes ‘water = H2O’ true (given Kripke’s view). Against this, it may seem as if information about the way the world is, as opposed to information about the meanings of concepts or words, isn’t playing a role in the compatibilist’s investigation. But I think that Quine’s point still stands: there’s no clear distinction between statements justified solely by the meanings of words and those justified partly by the meanings of words and partly by information about the way the world is. And when it comes to investigations as substantive as the compatibilist’s, we’re at least into territory in which we don’t have a clear distinction. (It’s sometimes assumed that Galen Strawson’s argument for the impossibility of moral responsibility is a conceptual argument – see the discussion of Randy’s book on this site. The same point can be made there.)

[Tamler Sommers]

Derk,

Right, "conceptual" probably wasn't the best word, since I don't want to imply that the debate between the compatibilist and the skeptic is only one about words or meanings. I think we're definitely on that Quinean border (or territory) here. Maybe "intuitive" is better. The compatibilist needs to provide an intuitive link between compatibilist freedom and the deserving of punishment; failing that, he or she might be accused of hard-heartedness.

But unlike "water is H2O", it's a little unclear (to me, anyhow) how we might resolve conflicting intuitions on this subject. One of the advantages of your approach is that it guides our intuitions gradually to what most consider to be a counterintuitive conclusion. But for those who are not persuaded (and I'm not one of them), how is the matter to be decided? Susan Hurley, for example, in her new book, urges her readers to reject what she called "the regression requirement" for responsibility--the requirement that for one to be responsible for an action, one has to be responsible for at least some of the determining causes of that action. Now I could never reject that requirement. To me, it makes no sense to think that someone is deserving of blame for an action if he is not responsible for any of the determining causes of that action. But what kind of evidence could I present to show that I'm right about this? Maybe empirical investigation, of the kind that Eddy Nahmias is doing, might show that more people agree with one side or the other. But even if 66% of people thought it made sense to reject the regression requirement for responsibility, I would still find it hard to do myself. (Incidentally, I do not think the survey would turn out that way.)

[Eddy Nahmias]

I would love to see someone come up with a good test of ordinary people's intuitions about the regression requirement (I've tried a bit without success). One very indirect type of evidence that people are not wed to such a requirement is the fact that we tend to believe that an agent can go from acting with no moral responsibility to some to lots--that is, people grow up. It seems a bit counterintuitive to believe that at some point a child makes a first choice for which she is robustly morally responsible and then she is responsible for all the choices and action that spring from that choice. More intuitive perhaps is the idea that responsibility comes in degrees such that children become more and more responsible as they develop their abilities to self-reflectively consider their own goals and values and at some point we take this level of responsibility to be high enough to say the agent deserves praise and blame (though the praise and blame attributions likewise come in degrees).

[Tom Clark]

In choosing between Tamler’s and Eddy’s intuitions, one thing to notice is that they may not be all that different. They just describe the situation differently. Neither thinks that anyone is self-caused and so neither supposes that the regression requirement for responsibility is fulfillable. For Tamler, this means no one is blameworthy, while for Eddy this means that blame attaches to agents that are sufficiently self-reflective. But what practical difference does this make; that is, what difference does this make in our responsibility practices?

Tamler might recommend abandoning certain sorts of punishment that have traditionally been justified on retributive grounds, e.g., execution. But no doubt he’d still advocate responsibility practices that help to shape good behavior, perhaps including sanctions of various sorts. What about Eddy’s practices?

Seems to me that the self-reflective criterion motivates responsibility ascriptions not because the agent is ultimately self-caused (obviously), but because the agent is capable of conforming conduct to our expectations, as backed up by rewards and sanctions. Blame and punishment function as sanctions, and the anticipation of blame shapes behavior among the sufficiently self-reflective. So my read is that Eddy’s responsibility practices, including blaming, would have to be essentially forward looking and consequentialist, not backward looking in the way that justifies retributive punishment. So although he and Tamler have superficially quite different intuitions about responsibility, wrapped up in how they talk about it, they’re perhaps not so different in their practical ramifications. And it’s in these ramifications that real differences in intuitions, if any, get expressed.

If Eddy says that no, my agents *just deserve* retributive punishment for being self-reflective, irrespective of any consequences such punishment might have in shaping behavior, then I’ll ask why. What’s the intuitive link between blame and being self-reflective, if it isn’t the way the prospect of blame can modify the behavior of self-reflective agents?

The upshot, I think, is that compatibilists can’t blame deeply the way libertarians can, and so they should abjure retribution. Compatibilist moral responsibility doesn’t leave traditional responsibility practices untouched; significant, progressive changes are in order. So compatibilists that continue to support retributive attitudes and practices seem to me hard-hearted.

[Tamler Sommers]

Tom,

You raise an interesting question: Do compatibilists abjure retribution? Or do they believe that retributive punishment or blame of some kind is justifiable when agents exercize compatibilist freedom?

If they did not believe that retribution is justifiable under any conditions, then I would agree that the two positions--FW skepticism and compatibilism--are quite close; in fact, the disagreement may be terminological. But I'm almost certain that a good number of compatibilists do not abjure retribution entirely (hence the substantive disagreement).

Am I right about this, compatibilists?

(Response that I fear: "it depends on what you mean by retribution.")

[Mark Smeltzer]

"If they did not believe that retribution is justifiable under any conditions, then I would agree that the two positions--FW skepticism and compatibilism--are quite close; in fact, the disagreement may be terminological."

As I've argued here in the past, I think the two things that separate the free will skeptics from the Compatibilists are: (1) the denial of moral realism and (2) the subsequent denial that we are moral agents. I believe that the compatibilist must affirm these two propositions to maintain the distinction between the compatibilist position and that of the free will skeptic. (In other words, I don't think the dividing line between the two positions involves one's stance on retribution. I think that is a secondary issue.)

For a "Compatibilist" who denines one of those propositions I believe the "distinction" between their position and that of the free will skeptic *is* reducible to terminological usage.

[Richard Double]

Thanks to Tamler Sommers and the commentators for prompting me to think more deeply about my THE MORAL HARDNESS OF LIBERTARIANISM (available in PHILO, 2002, VOL 5, 226-34 and on Ted Honderich’s website). I see now that I had one main point, which I obscured with a lesser, conjectural point.
A title more descriptive of my main point would have been THE MORAL HARDNESS OF ANY LIBERTARIAN WHO HOLDS THESE THREE THESES: [[1] We may hold persons morally responsible [MR] only if they make libertarian choices, [2] We should hold persons MR, and [3] There is scant evidence that persons make libertarian choices.] My argument is conditional: Anyone who believes these three statements is hard-hearted. My reasoning is analogical. It would be hard-hearted to punish S for A if we lacked strong evidence that S did A. Likewise, it would be hard-hearted to blame S for A if we lacked strong evidence that S did A from free will.
The conjectural part of the paper suggests that most libertarians actually hold or should hold on the grounds of theoretical consistency beliefs 1-3. Here is where the critics are at their strongest. Libertarians may reject any of the three steps, thereby evading the accusation. I am happy to grant, as some of the commentators have noted, that if libertarians make certain moves they would not be vulnerable to my accusation.
One way to avoid my charge would be to think that one has strong evidence that persons make libertarian choices. So, my argument does not apply to C.A. Campbell and the Peter van Inwagen of 1983. [I think all the reasons I’ve seen given by libertarians fail to produce minimal evidence, but this is one of those parenthetical remarks that is irrelevant to the main argument.] Libertarians who believe they have sufficient evidence that persons make libertarian choices thereby avoid my charge. [Still, I would not want to call such libertarians ‘soft-hearted’: An executioner who said, ‘You’ve misused your libertarian free will. So, off with your head’ isn’t really a softie by my lights.] At any rate, this objection denies [3] and, thus, side-steps my main argument.
A second objection would be to emphasize the consequentialist benefits of the practice of holding persons MR, whether we make libertarian or determined choices. If someone thought that this practice was SO morally beneficial that it outweighed the potential moral deficit of blaming persons who don’t make libertarian choices (Kant, Kane), and thought the practice required believing we make libertarian choices [large assumption], perhaps one could avoid the charge of moral hardness. [In the article I argue that such cost-benefit approach to MR runs contrary to at least Kant’s ideas about using persons as mere means to achieve goals, but libertarians certainly could adopt this line.] As with the first objection, taking this position would reject [1], thereby not challenging my main argument.
A third way for a libertarian to resist the charge of being hard hearted would be to disallow HUMANS from holding others MR, but grant this prerogative to God, who would know who is responsible. This riposte works against the conjectural argument, but at the cost of rejecting [2], thereby leaving the main argument intact.
Finally, Tamler’s asking whether compatibilism is a hard hearted theory, which strikes me as a question about how much severity compatibilist free will can justify, raises a far more substantive question than either of mine. [My position on this question is that it is unanswerable.] I have the impression that a great many compatibilists hold that free determined persons would warrant a milder degree of blame and punishment than most libertarians think free libertarian agents merit. That is, the compatibilists sympathize with the incompatibilists on this issue. This makes me wonder: “Are compatibilists incompatibilists who have strayed?”

Richard Double

[Tom Clark]

Mark,

I’m wondering if you could give us your take on retribution. As a moral realist, what do you think is the characteristic of moral agents, if any, that makes them deserving of punishment independent of any consequential effect, and why?

I’m also curious to hear Eddy’s and other compatibilists’ take on this, or on the variation posed by Tamler:

“Do compatibilists abjure retribution? Or do they believe that retributive punishment or blame of some kind is justifiable when agents exercise compatibilist freedom?”

or by Richard:

“I have the impression that a great many compatibilists hold that free determined persons would warrant a milder degree of blame and punishment than most libertarians think free libertarian agents merit. That is, the compatibilists sympathize with the incompatibilists on this issue. This makes me wonder: ‘Are compatibilists incompatibilists who have strayed?’”

[Joseph Keim Campbell]

Sorry for jumping in late. I've neglected reading the blog of late (so I also apologize if anyone has already made this point).

There is at least one thing wrong with your inference, Richard. One might accept [1]–[3] yet still adopt restrictivism. One might think that very few persons are really morally responsible for their actions since--as Socrates presumably thought--nearly every immoral action is done out of ignorance. In that case one might hold that very few persons are morally responsible for their actions yet still adopt a 'libertarian' view of free will and accept [1]–[3].

Or are even restrictivists hard of heart, too?

All the best, Joe

[Mark Smeltzer]

Tom,

I can't provide you with a very substantive answer at this time as I am still exploring this issue.

I believe desert is primarily based upon character, and that the responsibility an agent exhibits bridges the gap from our judgements regarding an agent's character as it manifests to us through an agent's actions to the agent's actual character.

I believe we *may* justly punish an agent (I am hesistant to say that we *ought* to punish) in virtue of the correctness of our judgements regarding an agent's character, and that it is only by analyzing the moral quality an agent's (responsibly conducted) actions that we can correctly judge the agent's character.

This general schematic seems to hold regardless of, and perhaps in spite of, consequential concerns.

On one reading of this view, it seems that the general notions of retribution can be supported according to this understanding of desert. In other words, one who shares my beliefs regarding desert could use them to rationalize retribution in many situations. However, I am unsure whether that is a justified application of the view.

[Eddy Nahmias]

Speaking as one compatibilist, I abjure extreme versions of retribution (in some cases for reasons other than issues of free will) but I do not abjure retribution (or desert) entirely. I take free will and moral responsibility (including the sort that involves desert and retribution and its praise counterpart) to be matters of degree (see my point above about children developing into moral agents). Those agents that possess the requisite abilities to be free are appropriate subjects of moral attribution because they are loci of a unique (and uniquely important) type of control and self-control in the (known) universe. This control is the sort we in fact tie to desert and retribution and, in many cases at least, we do so properly. That's all I have time for right now.

[Richard Double]

Joe,

I would be inclined to give the restrictivists [as you portray them] some softness points for seeing ignorance where other libertarians see malice. But I do not see that this concession hurts my argument. The restrictivists still would be hard-hearted to the extent that they held that [1]-[3] are beliefs held by any libertarians. My argument would conclude that restrictivists are somewhat less hard.

Best wishes,

[Tom Clark]

Eddy wrote:

“Those agents that possess the requisite abilities to be free are appropriate subjects of moral attribution because they are loci of a unique (and uniquely important) type of control and self-control in the (known) universe. This control is the sort we in fact tie to desert and retribution and, in many cases at least, we do so properly. That's all I have time for right now.”

Eddy, for when you have time:

If desert and retribution are properly tied to our unique capacities for control and self-control, that suggests that your justification for retribution is essentially forward-looking, since it’s the anticipation of punishment that such capacities make possible. But of course retribution and desert are supposed to be deontological and non-consequentialist, so this can’t be right.

So I’ll reiterate my earlier question in the current context: If you say that agents *just deserve* retributive punishment for having a unique capacity for control, irrespective of any consequences such punishment might have in shaping behavior, then I’ll ask why. What’s the link between blame and this sort of control, if it isn’t the way the prospect of blame can modify the behavior of agents that have it?

In short, you haven’t shown there’s a non-consequentialist rationale for ascriptions of moral responsibility, and that’s what’s necessary to justify retribution.