Blog Coordinator

« VBW (Bonus) Episode 14: "Snitches, Tattletales, and Whistleblowers" | Main | Final CfA: NOWAR 2 (Due March 1, 2013) »



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

Is this a problem for skeptics? Or is it, rather, a problem for those who accept that Frankfurt-style cases show that absence of alternatives is irrelevant to moral responsibility? In other words, why shouldn't the skeptic hold that the agent could have and ought to have refrained, but deny that the agent is morally responsible?

Sure Derk articulate all the AR 'oughts and ought nots' you like; no telling when you might bring about a better world. But it would be simply horrible if there were never any justification for uttering backward looking, 'ought to have' claims. I want to be able to stare the criminal down as I pronounce sentence believing with all my heart that he ought not to have robbed the bank. My most fervent hope, as we enter the penitential season of Lent, is that I one day hear from the Lord the words "Well done good and faithful servant" (Matthew 25:23) meaning I did my duty when I could have shirked it.

That is a great question, Derk. I’m trying to get a handle on what the axiological recommendation comes to. I think I see how the evaluative “ought” can apply to actions. In the same sense that we say “Larry ought to win the lottery,” we can also say, “Harry ought to refrain from making the hurtful remark.” In both cases, the idea seems to be something like that “the world would be better if…” Is it the particular use of the latter utterance that makes it a recommendation? Saying that “the world would be better if Harry refrained” might be used with the specific intention of having an effect on Harry’s behavior and in that sense be a recommendation, and yet the content would not be different in kind from the evaluative ought in the Larry case. If that is how we should understand it, then at least for now I am thinking that we lose an important distinction. Relatedly, I wonder if the distinction made by some moral theories between what would be good or wonderful to do and what we ought to do collapses, and if so, whether you would see this as a problem?

I have not read the relevant background literature from Humberstone and Schroeder, so I apologize in advance if what I say here is blatantly naïve. That being said…

Suppose Becky, a libertarian, reads your post and agrees with the general distinction between oughts of specific agent demand and oughts of axiological recommendation. Suppose the next day, Becky’s friend Leo tells her that his boss at work has been engaging in unethical behavior that could have bad consequences for Leo if he remains at his job. Becky forms the belief that Leo ought (in both senses of the word, let’s say) to quit his job. Becky tells Leo, “You ought not to go back to work.” Perhaps Leo has read your post, too, and Becky even informs Leo that she means to be using both senses of “ought.”

If I understand you correctly, and if determinism is true, then Becky cannot “legitimately invoke” the ought of specific agent demand when she informs Leo that he ought not go back to work. I find it interesting that, in such a scenario, Becky can form parallel judgments about both kinds of oughts but fail legitimately to invoke one of them. Why is that? Is it that when Becky says “You ought not go back to work” (in the specific agent demand sense), she speaks falsely? Or is it that she says something nonsensical? I presume you would say the latter, and that Becky’s action would be akin to someone trying to invoke a non-existent amendment in court. But does this mean that Becky’s *judgment* that Leo ought (in the specific agent demand sense) not return to work is also nonsensical? It isn’t clear to me that it is nonsensical, and here’s why that worries me:

Suppose Esther forms the sincere judgment that the present King of France ought (in the axiological recommendation sense) be praised. Esther says, “The present King of France ought to be praised.” It isn’t clear to me that Esther said something nonsensical. Instead, it seems to me that she said something false. It is not the case that the present King of France ought to be praised. (This doesn’t entail that the present King of France exists.) If Becky likewise says something false rather than something nonsensical, how do the actual, determined, future actions of Leo bear on that truth value? If determinism entails that Leo does not return to work, then is it accurate to say now that Leo actually *can* heed Becky’s advice? And if he can, does that mean that Becky *does* legitimately invoke the ought of specific agent demand? Alternatively, if Becky says something nonsensical rather than false, why is it nonsensical? If Leo is determined to heed Becky’s advice, then what is nonsensical about telling him (in effect) that he can?

(I wrote this in haste. I hope it is intelligible!)

Hi Derk, This is another great post. I just have a simple question for you here. Why is it that, in your estimation, those convinced by (something like) Frankfurt's argument should remain committed to accepting 'ought implies can'? If a relevant Frankfurt case can be used to challenge a principle that blameworthiness requires the ability to do otherwise, why can't a similar example be used to challenge the pertinent version of a principle that ought to do otherwise requires the (agent specific) ability to do otherwise?

I myself have no considered view on this. I confess, I struggle with this puzzle. But as one who does accept Frankfurt's argument regarding blameworthiness, it seems to me, at least initially, appealing to take the same line with respect to the relevant sense of ought.

I don't see what specificity has to do with the specific agent demand (other than the name).

Suppose I am watching someone about to commit a crime and I yell, 'You should not do that!' 'That' refers to the particular action in the near future I predict she will perform if nothing interrupts her activity. Sounds like a specific demand leveled at a particular agent. But if axiological recommendations don't imply the ability to do what is recommended, then I don't see why this 'ought'-claim would, because I think it follows from axiological recommendations (AR).

Take (1) 'The world would be better without that action than it would be with that action' (where the 'that's refers to the same action that the earlier one did)

Now take the AR (2) 'You ought to do what makes the world better, all things being equal' where this just says 'The world would be a better place if you performed the actions that make it a better place, all things being equal.'

It seems to me that (1) and (2) imply the original 'ought'-claim that I directed at the soon to be criminal, or at least do with an extra claim about the situation that none of the ceteris paribus clauses are violated.

But if (1) and (2) (along with the 'no violation of ceteris paribus' claim) don't imply that the agent can do the thing recommended, then presumably any proposition they do imply does not itself imply the ability to do the thing it recommends.

So if a specific agent demand is defined as an 'ought'-claim from which a 'can'-claim can be derived, then I don't see what specificity has to do with it (despite the name). It seems to me that we can get very specific, particular axiological recommendations. Isn't that just what consequentialism is? A theory that derives particular duties from claims about what would make the world better? And should we be surprised that such a thouroughly Kantian principle as 'Ought implies Can' finds no home in consequentialist theory?

In a predeterministic world, when agent A tells agent B that B ought to do something, and B follows A’s advice, the interaction may be modeled as A having an effect on B. But the FW skeptic would label everything in that sequence as being predetermined, including A’s desire to influence B. The skeptic would claim that A's influence on B is only in the “weak sense”, and there isn't truly any agent causation involved. So in the weak sense, the use of “ought to” is legitimate – I don’t think there’s much argument about that.

In our real world (not predeterministic in my opinion), I think we can go a step further and claim that the use of “ought to” is also legitimate in the “strong sense”. The reason I say that, is because when agent A conveys an opinion to agent B, multiple ideas interact in B’s brain in a manner that isn’t controlled *solely* by the four fundamental forces of physics. The *idea* that A gives to B interacts with other *ideas* in B’s brain, and the intelligence associated with that interaction isn’t innate to the 4FFOP.

One last comment… I agree that predeterminism precludes alternatives, but I don’t believe determinism precludes alternatives. The theory of determinism doesn’t have anything fundamentally against the idea that new forces may emerge (i.e., forces which aren’t simply a direct sum of preexisting forces), and those new forces (i.e., life) may then add into the mix in real time thereby becoming part of what “determines” the path of reality.

On Neil’s question, what gives rise to the problem is the conjunction of: (1) determinism (or causal determination of action by factors beyond the agent’s control), (2) ought-implies-can (supposing the ‘ought’ of specific agent demand), and (3) determinism’s ruling out the relevant sense of ‘can.’ This is a potential problem for any view, whether compatibilist or incompatibilist, that affirms determinism, and thus this will be an issue for hard and soft determinists.

But as Neil indicates, there are ways out. Supposing we hold determinism fixed, one might deny ought-implies-can (e.g., John Fischer, Peter Graham); or one might, emulating the classical compatibilist, deny that determinism rules out the relevant sense of ‘can’ (Dana Nelkin, and as you suggest). However, these moves are controversial, and there are counterarguments. My sense is that many advocates of the consequence argument would claim that their argument also shows that the relevant sense of ‘can’ for OIC is ruled out by determinism.

As Michael suggests, one might think, as John Fischer has argued, that Frankfurt cases successfully challenge both PAP and OIC. This is the line for rescuing the ‘ought’ of specific agent demand I am closest to accepting. But reading Dana (2011) on OIC makes me doubt this move at least a little. An ‘ought’ of specific agent demand in ‘you ought not to do A at t’ seems agent directive, in the sense that it directs the agent to refraining from A at t when the ‘ought’ is prospective, and when it’s retrospective it presupposes that the agent was directed to refraining from A at t. If the agent can’t or couldn’t refrain from A at t, then it seems that a condition for the ‘ought’ claim correctly applying is not met. My intuitions about Frankfurt cases make me think that blameworthiness (in the basic desert sense) is not similarly agent directive, and so the agent’s being blameworthy does not require that the agent could have refrained from the action. Unlike John, Ish Haji accepts Frankfurt cases but also OIC and that ‘ought’ judgments are undercut by determinism.

I feel the force of Robert’s comment. I agree that the ‘ought’ of specific agent demand is ingrained in moral practice, as Dana also intimates, and she is right to suggest that we stand to lose an important distinction between true ‘oughts’ of specific agent demand and ‘oughts’ of axiological recommendation. Maybe the distinction between what would be good or wonderful to do and what we ought to do collapses, but I wonder how bad this would be for moral practice. On Robert’s point about divine judgment, what if the Lord said to you “Well done good and faithful servant,” meaning that you brought about much of the good in the world he presented to us as an ideal, by revealing to us ‘this… is how things ought to be”?

On Benjamin’s thoughts, I would opt for the ‘false’ option for ‘oughts’ of specific agent demand, if they are indeed ruled out by determinism, by contrast with the ‘nonsensical’ option. The idea is that a condition of the truth of the ‘ought’ claim, that the agent can or could have complied with it, is not satisfied. As Benjamin also indicates, if the agent is causally determined to act in accord with an ‘ought’ claim of this sort, then the condition is satisfied. There would then be an asymmetry: the ‘ought’ judgments of this kind to which the agent complies are true, the others are false. Sean Stapleton (dissertation 2010) argues that given determinism, this is what turns out to be the case, which would raise a challenge to the prospective use of these ‘ought’ claims – we wouldn’t know in advance which ones were true and which ones false.

"What if the Lord said to you 'Well done good and faithful servant,' meaning that you brought about much of the good in the world he presented to us as an ideal, by revealing to us ‘this… is how things ought to be'?"

Exactly. But that is only half of it. I also want to be given to know that I had other infinitely less worthwhile, but seemingly more attractive, options that I could have pursued: worldly concerns that I might have gotten caught up in but for the grace that flows from Calvary and exercises of my own free will. I spread the Gospel- a largely thankless task- to the best of my ability while I could have been seeking fame and fortune or just watching ESPN.

Michael: 'ought' implies 'can' is theoretically motivated in addition to whatever intuitive plausibility it posseses. If the arguments in favor of the princple are congent, that would give you reason to resist the suggestion that Frankfurt-style cases are counterexaples to ought-implies-can as well as PAP. Of course, in order to maintain that the examples still refute PAP even if not ought-implies-can, you'd have to reject the thesis that blameworthiness requires wrongdoing, but you should reject that premise anyway!

One more thought on the collapsing of ought and good and moral theory. If we don’t have specific agent demands, neither consequentialist or non-consequentialist moral theories are correct, because there is nothing that anyone ought to do in the relevant sense. But it might be possible to capture some analogue of a moral theory, so that we could distinguish between, say, rights violations (or possibly analogues thereof) and failures to go above and beyond what we might previously have thought of as the call of duty. In Living Without Free Will, if I’ve got this right, there is the suggestion that we might correctly use “wrong” even if not “ought.” I think you have reasons for rejecting this suggestion, but if this move were available, then these further distinctions could still be drawn; the view would be consistent with analogues of either consequentialist or non-consequentialist moral theories. But if the replacement of “ought” in a specific agent demand sense is only “ought” in the axiological sense, then there would seem fewer resources to capture these moral distinctions. That may be why it seems more like a move toward consequentialism, as I think Patrick mentioned in his comment. I suppose we could still make distinctions like that between “strongly recommended” vs. “recommended”. But these would be distinctions of degree rather than of kind.

So, if determinism is true, there are still legitimate (as in true and accurate) oughts of specific agent demand, it is just that we are never fully justified in invoking them. Does that sound right? Is it fair, then, to say that specific agent demands to X are merely conjunctive claims (a) that one can X, and (b) that is it axiologically recommended that one X? The specific agent demand is thus false whenever determinism precludes one from X-ing, yet true if determinism entails X-ing (and the axiological recommendation is also accurate).

I worry that this picture faces some strong intuitive resistance. I think many will feel as though specific agent demands go beyond, and are indeed grounded in, the type of conjunctive claims I just described. Returning to my previous example, Becky (the libertarian) might think of her friend Leo, “He really can refrain from returning to work, and it really would be better for him so to refrain. For these reasons, he really should refrain from returning to work.” I would think that even free will skeptics can make sense of Becky’s reasoning and can feel the appeal of thinking that Becky says something new when she concludes that Leo “really should refrain from returning to work.” If Becky herself thinks so, is she wrong? Must we accept that Becky's thought process is akin to, "I have an apple. I have an orange. Therefore, I really do have an apple and an orange"?


In response to the questions raised in your original post, and with apologies for potential redundancy given the other replies, it seems that this would be very bad indeed.

As far as I’ve understood your remarks above, given causal determinism and given that such determinism precludes alternatives, we are imagining a situation in which the actual sequence of events is the one and only sequence of events, so that whatever happens is what must happen. That is, there are no alternatives, there are no possibilities, so there is nothing other than what was, is, and will soon be the case.

In such a situation, there seems to be no intelligible sense in which any agent ought to do anything at all, since nothing transpires other than the actual sequence of events. Thus, in a situation like this, saying that the agent "ought" to refrain from performing any particular action introduces a state of affairs that is *impossible*. Such an axiological recommendation seems about as good as saying "You ought to make it the case that q and not-q", which is at best an ineffective expression of hope.

Benjamin, I think you’re correct to resist the idea that an ‘ought’ of specific agent demand is just a conjunction of a ‘can’ claim and an ‘ought’ of axiological recommendation, and that in your example Becky says something new and significant. Dana, I need to rethink what I said about wrongness in Living without Free Will. It would be good to be able to capture the distinctions that you (and also Benjamin and Robert) have in mind, and maybe I can revive that route. And my sense is that you’re right to say that both consequentialist and non-consequentialist moral theories have been in effect framed in terms of the ‘ought’ of agent demand. Patrick’s take on what consequentialism is committed to on ‘ought’ claims is interesting; Alastair Norcross (in “Reasons without Demands: Rethinking Rightness,” in Blackwell Contemporary Debates in Moral Theory, J. Dreier, ed., Oxford: Blackwell) proposes a consequentialism that fits Patrick’s conception as I’m understanding it, and it’s a view I find attractive.

Brent, I’m thinking that in a deterministic setting prospective ‘oughts’ of axiological recommendation can be expressions of hope, but they need not be ineffective. It may be causally determined that your axiological recommendation be causally efficacious in getting someone to act in a way in which she would not have acted had you not made the recommendation.

Robert, I like very much your emphasis on grace, but on my conception of grace, there’s a tension with your also citing free will in this context.

I agree with James that the distinctions we’d like would be available on agent-causal libertarianism, but my sense is that we can at best merely hope that this view is true. I’m doubtful that the notion of emergence James mentions is credible enough to sustain significant components of moral practice.

I couldn't help noting this passage from Michael Brent:

"As far as I’ve understood your remarks above, given causal determinism and given that such determinism precludes alternatives, we are imagining a situation in which the actual sequence of events is the one and only sequence of events, so that whatever happens is what must happen. That is, there are no alternatives, there are no possibilities, so there is nothing other than what was, is, and will soon be the case.

In such a situation, there seems to be no intelligible sense in which any agent ought to do anything at all, since nothing transpires other than the actual sequence of events."

Hmmm... I should have thought that no matter what kind of scenario we are envisaging, "nothing trnaspires other than the actual sequence of events". !!!!

I've offered what I've modestly described as a "revolutionary" way of thinking about these matters. Well, revolutions don't occur overnight, and, when Mao was asked what he thought of the French Revolution, he replied, "It is too early to tell." So I don't expect everyone to agree (yet). But I've argued that we should give up both PAP and OIC (Ought-implies-can). If we give up these maxims, that's a big change in our paradigm for freedom and moral responsibility. But doing so taps into a deep and important set of intuitions, or so I would argue, and it allows us to make sense of robust moral responsibility in a causally deterministic world--or (again), so I would argue.

I think this is a very interesting discussion and I also think, as Derk says, incompatibilist proponents of the Consequence Argument should think determinism is just as threatening to the truth of many of our moral 'ought' claims if they also accept OIC as those of them who accept PAP think they are to all of our blameworthiness claims (especially those who take PAP to be derived via OIC (Copp, Widerker, and (I believe) van Inwagen)!)

[An important difference, though, of course (and here, I know, I'm just pointing out the obvious), is that whereas incompatibilism and PAP threatens the truth of ALL claims to the effect that X is blameworthy for phi-ing, incompatibilism and OIC only threatens the truth of those claims of the form that X ought (not) to phi for which X does not (not) phi (where the 'ought' is that of specific agent demand). So even if the world is deterministic (and incompatibilism and OIC are both true as well) there may well be many many true "X ought to phi" and "X ought not to phi" claims (those such claims about those Xs who do phi in the former case and those who don't in the latter.) [Now, all 'ought' (of specific agent demand) claims would indeed go out the window if one also accepted the Ought Implies Can Refrain (OICR) as well as OIC. Surprisingly (to me at least), a number of people do seem to accept this principle. I don't feel the force (at all!) of the reasons in support of it that they offer (that "morality would be "pointless" if it were false") and I think a very strong case can be made against it. But that's an entirely different issue...]]

I take it that the proposal is that though it is correct that many of our 'ought'-of-specific-agent-demand claims will turn out to be false, we can have as a replacement the 'ought'-of-axiological-recommendation (in those cases). I find it a little less clear though what is meant by having the 'ought'-of-axiological-recommendation as a replacement. I think I'd be rather disappointed in losing the 'ought'-of-specific-agent-demand, but, hey, if it's gotta go because certain arguments show that it has to, then I gotta let it go. But I'm not sure what settling for the 'ought'-of-axiological-recommendation would amount to. One thing one might be wondering about in asking the settling claim is whether this less problematic 'ought' can "do some of the work" that the old 'ought' that we have to let go of did. But though some people talk in this way, I'm not really sure I understand it--that is, I'm not sure I understand talk of such and such concept's doing this work and such and such other concept's doing that other work (unless this is just another way of describing the conceptual relations that the one concept stands in to other concepts (but if that's what it means, I don't see how one concept could do any of the work of another--it would have to be the other concept to do its work)). Though I hope (very much so!) that the hard determinists (by which, just to be clear, I mean those who accept the truth both of incompatibilism and determinism) are wrong, I do think the project of what might remain even if it is true is very important and can even go some way toward making me less frightened by the prospect of the truth of hard determinism. But I get a little wary when we start talking of "replacements" (see my confusion about such talk above) because the project then can start to take on the aspect of a shell game (and to be clear I am not convicting anyone of actually doing this). If 'ought'-of-specific-demand has to go, then, I think it should go. But I'd counsel against then going on to say, "but it's okay, we can still keep all our 'ought' claims just as before but we'll reinterpret them using another 'ought' that's lying around" (it is indeed a real 'ought' (the axiological one), I don't deny that) because that can seem like an underhanded attempt to hide the cost of our hard determinism, and even if that isn't the goal (which I suspect it isn't), it may nonetheless in fact hide the true costs of hard determinism from the incautious.


Thanks for the reply. If I understand you correctly, then I’m not sure it works. Given a causal determinism that precludes alternatives and possibilities, my axiological recommendation cannot cause my interlocutor to act in a way that she would not have acted had I not made that recommendation, for two reasons. First, there is only one course of action for me to take, so there is no alternative or possible state of affairs in which I do not make that recommendation. Second, there is only one course of action for her to take, so whether I make that recommendation or not, there is no alternative or possible state of affairs in which she acts otherwise. Hence, it seems that my axiological recommendation cannot be effective. Perhaps I’m missing something here?

John Fischer,

Thanks for noticing the remark, even if you could not help yourself in doing so!

If I understand your point here, then I agree: given the sort of causal determinism that I took to be at issue, since there are no alternatives and no possibilities, nothing transpires other than the actual sequence of events. That is, the kind of causal determinism that I assumed to be at work precludes the kind of modal realism that permits the existence of alternative possibilities (or “possible worlds”, on one interpretation thereof). On a causal determinism that precludes the existence of alternative possibilities, the actual world and its actual sequence of events is the only such sequence.

"Robert, I like very much your emphasis on grace, but on my conception of grace, there’s a tension with your also citing free will in this context."

There sure is, Derk, and my philosophical goal in life is to eliminate it. How's this for starters:

When discussing the efficaciousness of grace, St. Anselm does not make Susan Wolf’s mistake of bypassing the role of the will. For him, neither grace nor free choice alone suffices to bring about an individual’s salvation. Rather, grace “harmonizes” with free choices to achieve sanctification. Neither one on its own can bring one closer to the Lord; but are together necessary and sufficient to achieve this end, with each having its own distinctive role to play.

Grace restores rectitude to the will- if it is but accepted by a free choice, that is a self-determined yet divinely prompted exercise of the will to justice, now able to realize its purpose, the retention of the will’s rectitude for its own sake. The will’s self-control, on the other hand, allows grace to function as a gift, not something forced upon one, so that its acceptance, being free, can redound to one’s credit.

Should grace and the rectitude it entails be (freely) rejected, it will be withdrawn (but not the ability to preserve it should it be possessed, the will to justice). From its free acceptance, though, will flow additional grace, needed to preserve rectitude in the face of future trials: occasions on which one’s affection for happiness inclines one to act against God’s will, which requires instead an exercise of the will for justice. Doing one’s part on these occasions, freely choosing yet again to exercise the will to justice, re-affirming one’s commitment to God’s will, allows grace to once more preserve the rectitude of one’s will.

There is not a hint of Pelagianism here, no attempt to isolate some sanctifying act for which the fallen man alone would be due credit. One cannot be praised for accepting a gift, especially one that is infinitely precious: should we laud a drowning man for snatching and clinging with all his might to the life preserver thrown his way?

Free will, according to Anselm, is cooperating with God. Without grace there could neither be a turning to the Lord, one not having been prompted by Him, nor preservation of the will’s rectitude, entirely lacking but for His sacrifice on Calvary, only a self-aggrandizing manifestation of the will to justice. Sans self-determination, grace could not be what it is- a gift with which one is free to act in concert. In its place would be something automatic, even coercive: responsible for the will’s rectitude in the same credit nullifying way that Goodness and Truth would be were freedom asymmetrical.

Without Him we can do nothing, yes; but without freedom we cannot have Him: Love cannot be received any way but freely. The irresistibility of grace must, then, be taken as conditional: it will ‘have its way’ with one, do its salvific work, provided that it is freely accepted- retained through an act of self-determination. That is to say, should it be freely chosen as part of a commitment to the primacy of the will to justice-- God’s authority, there is no temptation that cannot be resisted, no sin that one cannot avoid. By its very nature grace can be rejected, but by that same nature it cannot fail to conquer sin once accepted.

Michael Brent writes: "Given a causal determinism that precludes alternatives and possibilities, my axiological recommendation cannot cause my interlocutor to act in a way that she would not have acted had I not made that recommendation, for two reasons. First, there is only one course of action for me to take, so there is no alternative or possible state of affairs in which I do not make that recommendation. Second, there is only one course of action for her to take, so whether I make that recommendation or not, there is no alternative or possible state of affairs in which she acts otherwise."

This is a mistake. The bit after "First" is irrelevant to whether my recommendation causes someone to act in a way other than they would have had I not made my recommendation. Even if determinism is true the relevant counterfactual might be true. (Confusions of determinism and strong inevitibilism might be at the root here.) The bit after "Second" is also irrelevant to the causation question. Again it depends on the truth of the relevant counterfactuals and they may indeed be true even if determinism is true.

Pete Graham,

Thanks for the comment. I would like to avoid corrupting this thread with a potentially unrelated tangent, so I limit myself to saying this: the claim that counterfactual statements possess (determinate? gap-laden?) truth-value is controversial, especially given the notion of determinism that I assumed to be at issue here, namely, one where there exists but one sequence of events and no alternatives or possible states of affairs. If I have misunderstood the relevant notion of determinism at issue, I would be happy to be corrected.

I took it that the notion of determinism at issue in the traditional free will debate--or, more carefully, the notion at issue in, say, van Inwagen's _An Essay on Free Will_--is one according to which determinism is true just in case the laws of nature of the actual world are deterministic. And the laws of nature, L, of the actual world's being deterministic was just a matter of its being the case that for any two (metaphysically) possible worlds w1 and w2, if the laws of nature of w1 and w2 are L, then either w1 and w2 match exactly at every point in time or there is no time t such that the complete state of the world in w1 at t matches exactly the complete state of the world in w2 at t. As determinism here so understood says nothing about whether there are or there are not any (metaphysically) possible worlds (be they with the same laws of nature as the actual world or be they with laws of nature different from those of the actual world) other than the actual world, I took it that determinism as understood in the traditional free will debate is compatible with there being metaphysically possible worlds (and even physically possible worlds) other than the actual world.

[Here I've been talking in terms of what you might call forward-and-backwards determinism. There are of course other properties of laws of nature that one might call "deterministic"--namely, forward-deterministic laws, backward-deterministic laws, etc. But the point about determinism's being silent on the existence of other metaphysically possible worlds holds for these versions of determinism too.]

One other quick note. Though it might seem that the evaluative 'ought' is immune to threats from hard determinism, on certain views many such 'ought's might also go out the window. (Here I may just be fleshing out the parenthetical "at least not directly" in Derk's comment in his original post: "an evaluative ‘ought’ claim does not (at least directly) entail a ‘can’ claim".)

On some views (certain buck-passing views (I believe), for instance), claims about goodness, the kinds of claims I take it axiological 'ought's are fundamentally about, are reducible to claims about reasons and, perhaps, reasons for action. And if, as seems plausible to me, many of the considerations which support OIC also support Reasons Imply 'Can' (RIC)--i.e., X has a reason to phi only if X can phi, then there might be an indirect threat to certain axiological 'ought's as well. (Again, only certain axiological 'ought's and not all axiological 'ought's because hard determinism won't threaten the truth of many X-has-a-reason-to-phi claims, namely those where X does in fact phi. (Unless, of course, one also endorses Reasons Implies Can Refrain (RICR). Which, again, seems silly to me; but I think a number of people seem committed to it from the things they say....)

So, in advance of settling the question whether goodness claims are reducible to reasons claims, and if so how (and whether RIC is true), we shouldn't be too sanguine about there being axiological 'ought's lying around to do the work we want them to do if hard determinism is true. (These worries, of course, are orthogonal to my worries about replacement concepts, which I expressed earlier.)

Pete, I really like your contributions to this thread.

As someone who doesn't think determinism rules out the relevant sense of ‘can,’ I can only answer one of your questions usefully, the "how bad would it be?" question. My answer is "very" - but not on account of how it directly affects interpersonal ethics. Rather, if specific agent demands are out, then specific agent demands I make of myself are out - even merely prudential ones.

And that seems to make nonsense of much ordinary deliberation about what to do. Because if I were to verbalize the conclusion of my deliberation, a fair statement of it would be "OK Paul, let's do *this*." Maybe I am an outlier, and no one else deliberates this way, but I doubt it.

I agree with Pete’s view that it’s important to be wary of “replacement” talk. My general line on living without free will is that there are significant aspects of our practices that we would lose, and some of those losses are bad – for example the loss of basically deserved praise and gratitude – but we’ll be able do without. I’m open to this being so for ‘oughts’ as well. But here I don’t have a settled view about what we’d lose, and how bad the losses would be, and thus the request in the initial post.

Would we incur significant losses for deliberation, as Paul suggests, or with respect to reasons, as Pete mentions? We had a discussion on the potential loss for deliberation at the Arizona event, and the idea came out in discussion of my paper (I think with Gunnar Björnsson) that the last stages of many deliberations often amount to acceptances of axiological recommendations -- for example: it would realize a significant amount of value for me to give to Oxfam, so I will do that. In response to Paul, the specificity of the conclusion wouldn’t appear to show that an ‘ought’ of specific agent demand was at play in the deliberative process. The last stage of the deliberative process might instead be: I ought (agent-demand) to give to Oxfam, so I will do that. But either sort of deliberative reasoning seems possible. I’d like to know how much we’d lose if we were restricted to the first type.

On reasons, Ish Haji has a new book, Reason’s Debt to Freedom (OUP 2012) that defends the concern Pete raises (you can look for my review of it at NDPR in the near future). I’m thinking that there’s a robust notion of reason linked to an epistemic ‘can.’ Suppose it’s epistemically open to me that I will give to Oxfam, even though I believe that because determinism is true it’s also epistemically open to me that I will be causally determined to refrain from giving to Oxfam whereupon it would be the case that I cannot (in the pertinent metaphysical sense) give to Oxfam. It seems that in this epistemic situation, its being epistemically open to me that I will give to Oxfam is enough to allow me to rationally consider the good that Oxfam does as a reason for me to give to Oxfam.

Robert, I'll have to wait to respond to your proposal about grace -- I need to run to a meeting.

Fun discussion - I'm submitting a paper to Philosophical Studies on this topic this very day.

So here's the argument, very briefly:

Derk has suggested earlier in response to Ish that 'ought implies can' might not hold for a certain action-guiding use of 'ought', where we use 'ought' in deliberation and advice. For this kind of 'ought', epistemic openness is sufficient, not what the agent actually CAN do. This is a bit problematic, since if I advise someone to take her car somewhere, and it turns out that she could not do that because her car had been stolen, we naturally think that this advice was, in an important sense, mistaken.

Rather, I want to argue that the 'ought implies can' principle is MOTIVATED by the action-guiding function of 'ought'. When we use 'ought' in deliberation and advice, we try to arrive at an action, so we try to arrive at a decision to do something we CAN do. However, this is a compatibilist 'can'. All kinds of information can be relevant for deliberation and advice, but what the agent is actually determined by the past and the laws of nature to do is not one of them (a laplacean demon would not be a particularly good deliberator or adviser - it would not be a deliberator or adviser at all).

There is also, as has already been pointed out in this thread, a kind of 'ought' merely used to point out that something is desirable, ideal and so on, and precisely because this 'ought' isn't action-guiding, it's not bound by the 'ought implies can'-principle.

Ish argues that he's not interested in some action-guiding use of 'ought', but in an objective moral 'ought'. Okay. I think, however, that there are two ways to understand this objective moral 'ought'. Either it merely expresses an ideal, altough an ideal pertaining to action rather than states of the world. In that case, there is no reason to assume that it is bound by 'ought implies can'. Or it expresses a highly idealised action-guiding 'ought' (this seems to be, for instance, what Judith Thomson means when she discusses objective 'ought' and says that we TRY to find out what we OBJECTIVELY ought to do when we deliberate - what I objectively ought to do is something I would conclude that I ought to do if I were perfectly well-informed). In that case, it is bound by 'ought implies can', but although this 'ought' might take into account all kinds of information that the agent does not in fact know about, it is still a compatibilist 'ought'.

I don't think it's plausible that there is any kind of 'ought' that is BOTH bound by 'ought implies can' AND where the 'can' is incompatibilist.

That, in extreme brevity, is the argument.

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan