PAP: A person is morally responsible for what he has done only if he could have done otherwise.

In what follows I will assume that PAP has been refuted. The object of this paper is to investigate the prospects for libertarianism in a post-PAP world. The results of my investigation are surprising. Contrary to received opinion, it will turn out to be rather a good thing for libertarianism if PAP goes.

I will argue that the kind of libertarian positions which survive post PAP are ones which have the resources to make sense of moral luck in a way unavailable to the compatibilist. This, I claim, provides a non-question begging reason to favour libertarianism over compatibilism. If I am right then the refutation of PAP, far from heralding a new renaissance for compatibilism, might in fact the beginning of the end for that position.

Historically PAP has provided the main support for libertarian positions on freedom and moral responsibility. Although compatibilists could argue for a conditional analysis of PAP according to which it can be met even if determinism is true, this was never a particularly natural reading of the principle. So, PAP gave the libertarians an advantage over compatibilists.

However, if PAP is false then things are different, or so it is assumed. Although there is logical room for libertarian positions post PAP, the advantage seems to lie with the compatibilist, if it lies with anyone. This is a picture I want to question, but for the time being we need to get clear over why this picture has come to be received wisdom.

First we need to get clear that the refutation of PAP does not directly refute libertarianism. With PAP out of the picture the question becomes whether causal determinism in the actual sequence rules out moral responsibility. There is room here to insist that it does, and hence there is room for a variety of what Fischer has referred to as hyper libertarian positions (Fischer 1999, pp. 129-130fn).Foot note 1_1 The term is meant to be disparaging, but I rather like it and will use it hereafter. A hyper libertarian then, is simply a libertarian who thinks that determinism undermines moral responsibility for reasons that are not to do with the availability of alternative possibilities -- not, in other words, to do with an agent's ability to do otherwise.

There are a number of reasons why the hyper libertarian might claim that determinism rules out moral responsibility. They might claim that causal determinism in the actual sequence is incompatible with our being the initiators or originators of our decisions and choices (where this is not a requirement of having a certain kind of control, but a separate requirement in its own right), or that it might rule out an essential type of creativity. As a hyper libertarian myself, what I claim is that determinism deprives one of ownership over one's decisions. We need to be the ultimate or, to use a phrase of Kane's «buck-stopping» explanation of our decisions in order for them to be truly ours (see Kane 1989 p. 254). This we cannot be if determinism is true, for the explanation of why we made one decision rather than another will always trace back to factors external to our deliberative process. Kane refers to this requirement as a requirement for ultimacy (Kane 1989, p. 254). Again, it is important to recognise that this concern is not about control. As far as I am concerned the only type of control worth wanting (or indeed, the only kind of control that is intelligible) is of a sort compatible with determinism. But important though control is, it is not all that there is to freedom: in addition to controlling our decisions we need also to own them in the sense that I have just outlined. According to my position then, it is ownership that is threatened by determinism.Foot note 1_2

But a compatibilist is simply going to object that it begs the question against the compatibilist to insist that indeterminism is needed for ownership, creativity, importance, significance and so forth. (see Fischer 2003 pp. 198-210). After all, there are perfectly good compatibilist interpretations of ownership, creativity, importance, significance and any other plausible notion that the libertarian might care to maintain is threatened by determinism. The compatibilist could say something like this: your decisions and choices are significant in as full a sense as anyone could want because the world had to go through you to get that to happen. So, where the libertarian stresses independence from the world, the compatibilist could stress the agent's indispensability from the world and so on (on this see Fischer 2003, p. 208). More could be said here, (Fischer, for example, has his own, compatibilist, account of ownership) but the basic point is clear. Some reason needs to be given why we should favour libertarian interpretations of ownership, independence, creativity and so forth over compatibilist ones. As Fischer says «once the debate is shifted away from the relationship between causal determinism and alternative possibilities, it is difficult to present a non-question begging reason why causal determinism rules out moral responsibility» (Fischer 2003, p. 201). Indeed, compatibilism should be favoured «insofar as our basic views about ourselves -- our views of ourselves as persons and as morally responsible -- should not be held hostage to the discoveries of a consortium of scientists about the precise nature of the equations that describe the universe» (Fischer 2003, p. 211).

So, it looks as if the libertarian is indeed disadvantaged by the passing of PAP. My own hyper libertarian position survives the falsity of PAP, but according to Fischer and others it does not have anything to recommend it to any who are not already converted to libertarianism, and may in fact look like the less attractive option when we bear in mind that compatibilism does not hold certain of our views «hostage to the discoveries of a consortium of scientists». If you like, we have nothing to loose and everything to gain by being compatibilists in a post PAP world.

In fact, there is a very good reason to favour the kind of hyper libertarianism I have outlined. To see why, we need first to understand just why the old-style libertarian was so concerned about an ability to do otherwise. To answer this question we need to get clear about just what compatibilist control amounts to.

Consider that, according to a popular compatibilist account of control what it is for a decision or choice to have been controlled is simply for it to have been the output of a certain type of mechanism: one sensitive or responsive to an appropriate range of inputs. The kind of control mechanism that is relevant to moral responsibility is one that is reason-responsive (the term `reason-responsiveness' was coined by Fischer in his Metaphysics of Free Will and he and Ravizza have further developed and refined this conception of control in their book Responsibility and Control: A Theory of Moral Responsibility 1998). In other words, the mechanism that issues in the agent's decisions and choices must be one that is sensitive or responsive to a certain degree (for otherwise it will fail to be a control mechanism) and what it is sensitive or responsive to must be reasons. Other things will plausibly be important too, for instance the history of the mechanism in question, and the way in which the mechanism came to issue in the decision or choice that it did -- but strictly speaking these kinds of concern are not to do with control. So, having control of a kind relevant to moral responsibility means simply having a mechanism of a certain sort, sufficient to give you the «power to be moved by reasons» (Dennett 1984, pp. 18-19; p. 25; pp. 50-51; p. 98). Essentially then, what having control amounts to is having capacities and dispositions to respond in certain ways to a relevant range of inputs.

This account of compatibilist control is far too brief, but it would be impossible to say more within the scope of this paper, suffice it to say that this paper is addressed to those compatibilists who accept something like the above account of control.

Libertarians have concerns about this type of control. It is not that they deny that it is an account of a type of control. After all, no libertarian should deny that we can distinguish between controlled and uncontrolled events even if determinism turns out to be true. Rather, the concern is over the adequacy of this type of control when it comes to moral responsibility. This concern has been nicely captured recently by Paul Russell:

On the face of it, therefore, the agent is liable to blame and retribution, on the moderate reasons-responsive account, merely for possessing capacities that he is not able to exercise control over. [à] in the case of [wrongdoing] the agent appears to be simply unlucky enough to be moderately reasons-responsive and placed in circumstances where the mechanism fails to track the reasons that were present (2002, p. 595)

I believe that it is plausible that it is this type of concern that lay behind the commitment to PAP. The thought was that if an agent has only compatibilist control over their decision and choices, then they nevertheless lack control over how their mechanism operates in the actual sequence. Given that they are, by hypothesis, morally responsible for the decisions that they make, this means that the agent will be exposed to a certain kind of moral luck. It is their bad luck, for instance, that they possess a mechanism which, in these exact circumstances, will issue in this, morally reprehensible decision. For in different, but relevantly similar circumstances, it would have issued in a different, blameless, or even praiseworthy decision.

By relevantly similar circumstances I mean any possible world in which both the mechanism and the morally relevant reasons are held fixed. Unless the mechanism in question is, to use Fischer's term, strongly reason responsive, then there will always be such possible worlds -- possible worlds in which the mechanism and the morally relevant reasons are held fixed, and where the mechanism delivers a different output for which the agent is, by hypothesis, morally responsible (Fischer and Ravizza 1998, pp. 41-42)

In response to this concern one move would be to insist that strong reason-responsiveness is what is required for moral responsibility. If an agent is strongly reason responsive then they are not exposed to the kind of moral luck in question. Whatever decision the strongly responsive mechanism delivers, it would have delivered the same decision in all scenarios in which both it and the morally relevant reasons are held fixed. It is therefore no matter of luck that it delivered the decision that it did in the actual sequence. However, strong reason responsiveness is far too demanding a control requirement which would in practice all but rule out moral responsibility. Indeed, to avoid exposure to this kind of moral luck, the reasons to which the mechanism would have to be strongly sensitive would have to be the moral reasons present, which would have the upshot that no agent could ever do wrong culpably.

To see how it is this kind of concern about moral luck that lay behind PAP consider that if the agent is not to be exposed to moral luck of the kind in question (and assuming that strong reason responsiveness is an implausible demand for moral responsibility) she will need some additional type of control -- a type of control that is not captured by talk of reason-responsive mechanisms of various kinds. Unlike mechanism accounts, this type of control will require genuinely alternative possibilities. In the actual circumstances in which the mechanism operated and failed to track reason, it needs to have been genuinely possible for it to have succeeded in tracking reason. Only if such a genuine possibility exists, can there be room for the agent to have had additional control over how the mechanism operated in the actual sequence. For only if there are genuinely available alternative possibilities does the agent have the ability to track, or fail to track reason in the actual sequence. So, it is a concern about moral luck which gives rise to a requirement for a kind of executive control which can only be accommodated if there are genuinely available alternative possibilities.

To reiterate then, if we only have compatibilist control, then it seems fundamentally unfair to hold us morally responsible for the decisions and choices that we make in the actual sequence, for we would have made different decisions in alternative, but relevantly identical sequences. The kind of executive control associated with PAP would have ruled out this kind of moral luck.

With PAP gone, then the kind of executive control in question is not available or not actually required for moral responsibility. But just because PAP is false, this does not mean that the concerns about moral luck were not legitimate, and nor does it mean that they magically disappear. This concern about moral luck does not seem to me to be question-begging in any interesting way. I would say that the kind of moral luck in question is prima facie problematic, and in the absence of an adequate explanation it has to count as a serious weakness in the compatibilist's position that they must affirm its existence. But of course this same concern now applies to my own hyper libertarian position, for I acknowledge that the kind of control outlined above is all the control required for moral responsibility.

So the situation is this, post-PAP both the hyper-libertarian and the compatibilist should agree about control. The only kind of control relevant to moral responsibility is the mechanism kind outlined above. However, the original concerns about the adequacy of compatibilist control are still with us. Both hyper-libertarian and compatibilist must affirm the reality of the kind of moral luck outlined. Yet, the reality of this kind of moral luck is prima facie problematic.

It is now that my hyper libertarian position can be seen to have an advantage over compatibilism. For my position can make sense of moral luck in a way unavailable to the compatibilist. Consider that the kind of moral luck in question can, we have seen, be ruled out if the agent is strongly reason responsive. If, in other words, they had full compatibilist control. It would, as I have already said, be implausible to insist upon strong reason responsiveness as a condition of moral responsibility. But nevertheless what we can notice is that for a compatibilist the strongly reason-responsive agent could, in principle, be morally responsible for the decisions and choices that they make. The same is not true for anyone holding the type of hyper libertarian position that I have outlined. For the only way in which an agent could be strongly-reason responsive is if their mechanism is internally deterministic. Only if the mechanism is internally deterministic will we be able to say of it that it would have delivered the same decision in all relevant possible worlds -- worlds in which the same morally relevant reasons are present and the mechanism is held fixed. If the mechanism were internally indeterministic then there would always be some possible worlds in which the same reasons are present and the mechanism is held fixed, yet a different decision issues. Internal indeterminism therefore entails something less than strong reason responsiveness. If internal indeterminism is required, as I maintain that it is, then something less than strong reason responsiveness is actually a requirement of moral responsibility. This means that the kind of freedom needed for moral responsibility actually exposes the agent to the moral luck in question. Exposure to moral luck is a condition of having the kind of freedom necessary for moral responsibility.

Here we can see the truth in something that the compatibilists have often claimed. Namely, that indeterminism does nothing to enhance control and will in fact diminish it ( Kane 2003, p. 318). The libertarians used to charge that this was question begging. But now that, post PAP, hyper libertarian and compatibilist agree about control, they should agree to this claim. In the absence of any special kind of libertarian control, making an agent's reason responsive mechanism internally indeterministic will diminish the degree of control it can be said to deliver, for it will invariably make it less responsive than it would otherwise have been.

It is important to recognise here that if the agent's reason-responsive mechanism is internally indeterministic, that does not preclude its being sufficiently reason-responsive for moral responsibility. My target audience here are those compatibilists who allow that indeterminism does not rule out moral responsibility. I take it that the appeal of the reason-responsive account of control is precisely that it has this upshot. So long as only some degree of compatibilist control is required -- so, just some degree of reason responsiveness - then an internally indeterministic mechanism can, in principle, be sufficiently responsive to satisfy the control requirement to any of my target compatibilist's satisfaction. Note also, that any compatibilist who held otherwise would then loose the compatibilist their supposed advantage over hyper libertarianism -- it would be to place our «views of ourselves as persons and as morally responsible [in] hostage to the discoveries of a consortium of scientists» (Fischer 2003, p. 211).

So, moral luck begins to make sense when we take the hyper libertarian's perspective. No such account is available to the compatibilist. They cannot insist upon internal indeterminism as a condition of moral responsibility without giving up their compatibilism. So they are stuck with having to simply insist that moral luck is a brute fact unamenable to explanation. By contrast, the hyper libertarian such as myself can offer a principled explanation of the phenomena in question. The compatibilist is challenged to present a systematic account of moral luck, or else concede that there is a powerful, non-question begging reason to prefer libertarianism to their own position.

So, the concern about moral luck stems from focussing only on the control condition associated with moral responsibility. But once we understand that control is not all that there is to freedom there opens up the possibility that some other condition of freedom might be in tension with (though not contradict) the control condition. Ownership, according to the libertarian position I have outlined, is in tension with the control condition insofar as it can only be satisfied if the agent has less than the kind of full control that would rule out moral luck. Hence moral luck is explicable as being the cost of freedom. Freedom, when properly understood, involves being exposed to the kind of moral luck in question, for it requires that we have less than the kind of full control that would rule out such exposure.

The unfairness of moral luck remains. But we have seen that to rule out unfairness completely would require a level of control incompatible with being free. And fundamentally an agent is morally responsible not because it would be fair to hold them morally responsible, but because they made their decision freely. It is an agent's freedom which explains and makes comprehensible their moral responsibility: in short, moral responsibility is about freedom, and not about fairness.

I shall now try to address what I take to be some of the more pressing objections to some of the claims I have made above. I should add however, that what I have said is supposed only to show how one kind of moral luck could be intelligible given our received ideas about moral responsibility. It is not supposed to be a comprehensive solution to the larger problem presented by other kinds of moral luck. I believe it can go some way towards such an account because I believe that the kind of moral luck we have been talking about is the most fundamental, but that is a matter for another occasion. On that front what I have said is merely suggestive.

To some criticisms, then. Central to my argument is the claim that indeterminism internal to the operation of the agent's reason responsive mechanism would rule out strong reason responsiveness. In other words, introducing indeterminism in the way needed to secure ownership would diminish control, and would rule out complete or full control. But this claim might be challenged. Mele has suggested a way in which the libertarian's demand for internal indeterminism could be met without sacrificing any compatibilist control.

[I]f it is causally undetermined whether a certain belief will enter into Jones's deliberation, then Jones lacks [compatibilist control] over whether the belief enters his deliberation. But this need not be an impediment to Jones's àcontrol over how he deliberates in light of the beliefs that do enter his deliberation. (Mele 1995, p. 215)

But here I need to make clear in what sense I meant that indeterminism needs to be internal to the deliberative process. For I would claim that an agent's beliefs form part of their circumstances, and are as such external, not internal to the deliberative process. They are, if you like, at the input end, whereas the indeterminism needs to be at the output end. It needs to be undetermined what decision the process will issue in at the point of decision making. Following Kane, the indeterminism needs to be at the point of decision making. The reason is that if the explanation of the agent's decision traces back to factors over which the agent did not exercise control then the agent is not morally responsible for that decision. For if this is the case then the agent is controlled by the past in a way that precludes ultimacy and ownership. Mele's proposal violates the ownership condition, for by hypothesis it is not under Jones's control what beliefs enter his deliberation. Things are different if it is indeterministic what decision will issue at the point of decision making. For by hypothesis the explanation cannot now trace back to factors not under the agent's control, because the decision is the output of the control mechanism.

So, I would claim that Mele's proposal does not introduce indeterminism in a way that would meet the requirements of ownership. It is only if the indeterminism is internal to the control mechanism in such a way that it must diminish control, that the ownership condition is met. However, I am willing to concede that there may be variations of hyper libertarian position according to which indeterminism need only enter in the way outlined by Mele. The precise strategic location for indeterminism can be a matter of debate amongst libertarians. But whilst such positions may be coherent, they would be incapable of offering an account of moral luck and thereby would offer no advantage over compatibilism. Indeed, the compatibilist position would, for reasons given earlier, actually have a slight advantage over such positions. So, the claim that the indeterminism needed for ownership will diminish compatibilist-control is, I hold, defendable, and recommends itself over libertarian alternatives, precisely because it facilitates an account of moral luck that would otherwise be unavailable.

A second kind of objection can now be raised. What I have said above about the decision being literally undetermined until the moment of choice might lend itself to being interpreted as the claim, associated with some libertarians, that the decision in question must be uncaused by prior events. This in turn will then lead to the objection that the resulting decision will be insufficiently connected to the agent's reasons for action.

However, from the fact that it needs to have been indeterministic, up to the moment of decision, does not mean that the decision in question will not have been antecedently caused. My claim is that ownership or ultimacy is achieved only if the ultimate explanation of why one decision was made rather than another stops with the agent. But this is not the same as the claim that the explanation of why the decision made was made needs to stop with the agent. The difference here is between a plain and a contrastive explanation. A plain explanation is one that answers the question «why P». A contrastive explanation answers the question «why P rather than Q». If universal indeterminism obtains so that every event is indeterministically caused then we will be unable to give contrastive explanations. If it is undetermined whether T will cause P to occur or Q to occur, then we cannot explain why P rather than Q occurred in the actual sequence. Citing T will be of no help. However, it does not follow that we will be unable to give plain explanations. We will still be able to explain why P occurred, for we can cite T. T plainly explains P. But T does not contrastively explain why P rather than Q. Consider that if determinism obtained, then it is in principle possible that T would both plainly explain P, and contrastively explain why P rather than Q. So determinism obtaining will mean that what previously would only plainly explain, can now contrastively explain. It is the absence of the relevant contrastive explanation that is required by ownership, not the absence of a relevant plain explanation.

So, if the agent's decision making process is indeterministic up to the moment of decision, that does not mean that we cannot give a plain explanation of the agent's decision in terms of their antecedent causes. So there is no reason to think that we will not be able to link the agent's decision with their reasons. What it does mean however, is that we cannot give a contrastive explanation of why the decision making process resulted in that decision rather than a different one.

Now this gives rise to another, third objection. If, in the actual sequence, the agent makes decision P, then we can cite the agent's prior reasons in explanation of why this particular decision was made. But if we roll back time to just before the decision was made, and run things through again, then the decision making process might issue in a different decision -- decision Q. Given that the prior reasons will be the same, how can Q be explained in terms of the agent's prior reasons?

This means that, if Jane is deliberating about whether to vacation in Hawaii or Colorado, and gradually comes to favour and choose Hawaii, she might have chosen otherwise (chosen Colorado), given exactly the same deliberation up to the moment of choice that in fact led her to favour and choose Hawaii (exactly the same thoughts, reasonings, beliefs, desires, dispositions, and other characteristics -- not a sliver of difference). (Kane 2003, p. 302).

Here I follow Kane in claiming that the relevant deliberative processes are ones where we have opposing reasons -- we have reasons in favour of making one decision and reasons in favour of making an alternative decision.xlfcFoot note 1_3 So, along with Kane, I claim that the kind of indeterminism needed for ownership can arise only on those occasions where the agent's deliberations take the form of trying to sort out which amongst competing reasons to act from or on. These occasions provide us with examples where the deliberative process can be indeterministic insofar as it is undetermined how this deliberative process will turn out, but at the same time, however it turns out, the decision in question will have been made on the basis of reasons. I will here let Kane, to whose account I am clearly heavily indebted, offer some clarification.

Imagine that the businesswoman is trying or making an effort to solve two cognitive problems at once, or to complete two competing (deliberative) tasks at once -- to make a moral choice and to make a choice for her ambitions àWith respect to each task à she is being thwarted in her attempt to do what she is trying to do by indeterminism. But in her case, the indeterminism does not have a mere external source; it is coming from her own will, from her desire to do the opposite. Recall that the two crossing neural networks involved are connected, so that the indeterminism which is making it uncertain that she will do the moral thing is coming from her desire to do the opposite, and vice versa. She may therefore fail to do what she is trying to do à[b]ut I argue that, if she nevertheless succeeds, then she can be held responsible because, like them, she will have succeeded in doing what she was trying to do. And the interesting thing is that this will be true of her, whichever choice is made, because she was trying to make both choices and one is going to succeed. (2003, pp. 312-313)

This account of internal indeterminism is quite consistent with the mechanism in question being sufficiently reason responsive to satisfy any plausible compatibilist-control condition. Obviously, a mechanism that is responsive or sensitive to competing reasons such that it is indeterminate how it will respond, cannot deliver full compatibilist control. But that was always the point. So a reason-responsive mechanism that is responsive to competing reasons, can nevertheless deliver a high enough degree of responsiveness to qualify for moral responsibility where control is concerned, and it will also deliver the kind of internal indeterminism needed to ensure ownership of the decisions that issue. For we will be unable to explain why the agent acted for the reasons that they did, rather that the competing reasons. The agent here terminates the explanation and thereby achieves the kind of significance and ultimacy required for ownership according to my account.

Finally, a fourth objection I can anticipate regards PAP. I have been assuming that PAP is false, yet the account of hyper libertarian freedom that I have given is one that says that, in fact, at the point of decision making it was undetermined what decision the agent would make. This was required in order to rule out the relevant contrastive explanation and thereby satisfy the ultimacy or ownership condition. But then it would seem that ultimacy or ownership does require that there are alternative decisions that, in the actual sequence, it was possible for the agent to have made. It was just such a possibility that Frankfurt-style cases were designed to rule out. If we are accepting that PAP is false on the basis of Frankfurt-style cases, then we cannot at the same time make it a condition of moral responsibility that there exist these kinds of alternative possibilities.

Two responses. Firstly, many compatibilists accept that Frankfurt-style cases cannot be constructed in which all alternative possibilities are ruled out, for to do this would require assuming determinism, which just begs the question against the libertarian (see Fischer 1999 p 122). Rather, what Frankfurt-style cases do, and why they refute PAP, is they rule out relevant alternative possibilities, which is to say alternative possibilities that the agent can «access». I have argued elsewhere that even if there is an alternative possibility in which the agent responsibly decides otherwise, Frankfurt-style cases can still show that the agent did not have the ability to decide otherwise, because the agent did not have the right kind of access to the alternative in question (its obtaining depending upon something lucky or improbable happening). To make this argument in any detail would require a separate paper, so I will say no more about this point apart from to reiterate that I believe (in line with many others) that Frankfurt-style cases refute PAP, but not by ruling out the possibility of deciding otherwise, only by ruling out the ability to decide otherwise. Since it was securing the ability to do otherwise that alternative possibilities were originally thought necessary, nothing I have argued here implies that the refutation of PAP is unstable.

But, just in case this does not satisfy those who might make this criticism, let us accept, for the sake of argument, that Frankfurt-style cases do rule out alternative possibilities in which the agent responsibly decides otherwise (after all, some have tried to design Frankfurt-style cases in which all relevant alternative possibilities are rule out -- see Hunt, D. 2000; Stump 1999). The point about utlimacy and ownership is that these are conditions that concern the origin, or way that the decision came about in the actual sequence. The agent is still the ultimate contrastive explanation where they are the indeterministic cause of their making the decision that they did, even if in the actual sequence there were external constraints meaning that no other decision would have been possible, in those precise circumstances. For the point is that whilst prior conditions might fully explain why the decision in question occurred (because of the presence of some suitably placed counterfactual intervention device) they cannot fully explain why the agent made the decision in question for the reasons that they did. So, even on the hypothesis that such scenarios are constructable without begging any questions (and I believe that they are not) , it is still the case that indeterminism, strategically located, can secure the kind of ultimacy needed for ownership. The point would be that ultimacy does not actually demand that no contrastive explanation can be given of why the decision was made, but rather that no relevant contrastive explanation can be given -- one making reference to the agent's reasons.

The contemporary debate over the compatibility of determinism and moral responsibility has paid insufficient attention to the problem of moral luck. This is a major oversight if, as I believe, much of the concern over the threat from determinism is traceable to concerns over moral luck. I have argued here that once we focus on this issue, we will see that in a post-PAP world both libertarian and compatibilist will find themselves committed to affirming the reality of moral luck. However, hyper libertarianism has the advantage here, as it has the resources to make sense of the reality of moral luck in a way unavailable to the compatibilist.

In the introduction to the latest edition of his collection on Free Will Watson says, «[s]uppose that we are causal systems whose operations are highly probable, but not certain, given their antecedents ..[t]his supposition satisfies the incompatibilist requirement, but it hardly gives us what we are after» (2003, p. 9). On the contrary, what I have argued is that it does give us what we are after. Once PAP is out of the way, we can start to see how the space opened up by an indeterministic physical process is a space needed for ownership, not for control. This requirement is one that should recommend itself to the compatibilist too, because it provides something that both sides need, namely a systematic explanation of the reality and comprehensibility of moral luck. So, by focussing our attention on the real problems associated with compatibilist control, the refutation of PAP helps us to see that we have a very powerful, non-question begging reason to favour a hyper libertarian position over compatibilist alternatives.

Clarke, R. 1995: «Indeterminism and Control» American Philosophical Quarterly 32 pp. 125-138.

Fischer, J. 2003: «Frankfurt-style Compatibilism» in Free Will ed. Watson, G. New York: Oxford University Press.

Fischer, J. 1999: «Recent Work on Moral Responsibility» Ethics 110 pp. 93-139.

Fischer, J. & Ravizza, M. 1998: Responsibility and Control: A Theory of Moral Responsibility. Cambridge: Cambridge University Press.

Haji, I. 1998: Moral Appraisability. New York: Oxford University Press.

Hunt, D. 2000: «Moral Responsibility and Unavoidable Action», Philosophical Studies 97.2, pp. 195-227.

Kane, R. 2003: «Responsibility, Luck, and Chance: Reflections on Free Will and Indeterminism» in Free Will ed. Watson, G. New York: Oxford University Press.

Kane, R. 1996: The Significance of Free Will. New York: Oxford University Press.

Kane, R. 1989: «Two Kinds of Incompatibilism», Philosophy and Phenomenological Research 50 pp. 219-254.

Mele, A. 1995: Autonomous Agents: From Self-Control to Autonomy. New York: Oxford University Press.

Stump, E. 1996: «Libertarian Freedom and the Principle of Alternative Possibilities,» in Faith, Freedom, and Rationality, ed. Jordan, J. and Howard-Snyder, D. Lanham, MD: Rowman & Littlefield.

Pereboom, D. 1995: «Determinism al Dente,» Noûs 29 pp. 21-45. York: Oxford University Press.

Watson, G. 2003: «Introduction» in Free Will ed. Watson, G. New York: Oxford University Press.

Russell, P. 2002: Review of Responsibility and Control: A Theory of Moral Responsibility in Canadian Journal of Philosophy, vol 32 no. 4 pp. 587-606.

[Foot Note 1_1]

Amongst contemporary hyper-libertarians can be numbered Randolph Clarke; Eleanor Stump and David Hunt.

[Foot Note 1_2]

Of course, there are, amongst hyper libertarians, those who would want to supplement this kind of account with an account of agent-causation. I think that this is unnecessary and unhelpful and returns the issue to being one of control. This was the problem with the old debate. PAP led to all the focus being on control. Now, I think that if we can possibly help it, it would be better if we can sell libertarianism without having to make recourse to agent-causation, simply because the resulting theory will not require any previous commitment to libertarianism.

[Foot Note 1_3]

This will obviously not always be the case. But then sometimes, maybe even quite often, it will be the case that it is not indeterministic prior to decision what decision the agent will make. In such cases the resulting decision will either be one for which the agent cannot bear moral responsibility, or will be one for which they can bear responsibility because what has made the decision inevitable is the result of earlier, genuinely free choices on the part of the agent.