Why Frankfurt-Style Cases Don t Help (Much) Neil Levy Contemporary debates about free will and moral responsibility frequently focus on arguments around Frankfurt-style cases (FSCs). Their centrality reflects the fact that they have widely been seen to constitute a decisive step forward in a debate that had seemed to be irredeemably bogged down. In Fischer s (1994) influential phrase, they break the dialectical stalemate that had come to be characteristic of the free will debate and they break it, moreover, in favor of compatibilism. By showing that alternative possibilities are not necessary for moral responsibility, they remove the main source of the intuitive attractiveness of libertarianism. In this paper, I will argue that this view is false: FSCs do not break the dialectical stalemate at all. Instead, they leave everything more or less as it was. This paper, unusually, therefore does not claim to advance the debate, but to take it backwards. How are FSCs supposed to have decisively advanced the debate over the compatibility of free will and causal determinism? Before these cases came onto the scene, debates about the compatibility problem often turned on the question whether causal determinism ruled out alternative possibilities (Fischer 2002). Both sides assumed the truth of the principle of alternative possibilities (PAP): PAP. An agent is morally responsible for his or her choice at time t only if he or she could have done otherwise at t. 1 Debate concerned, not the validity of PAP, but the question whether PAP was compatible with causal determinism. But this debate made comparatively little progress. Generally, it became bogged down in seemingly interminable disputes about the correct analysis of the notion of agents abilities: in an earlier era disputes about whether the ability to choose and to act otherwise is best understood as a conditional ability, compatible with compatibilism, or an unconditional ability incompatible with it; more latterly disputes about the socalled consequence argument and varieties of local miracle and backtracking accounts of ability. 2 FSCs do not help us to analyse the notion of ability; the question whether causal determinism is incompatible with the relevant ability remains as obscure as ever. But the beauty of the FSCs is that they enable us to finesse the question. FSCs allegedly show that PAP is false: we do not have to have alternative possibilities in order to be morally responsible for an action. Hence (on the near-universal assumption that free will is the property of agents that underwrites ascriptions of moral responsibility), we need not settle the question whether causal determinism rules out alternative possibilities. It simply doesn t matter either way. If causal determinism is incompatible with moral responsibility, it must be for some reason other than the supposed need for alternative possibilities. Incompatibilists ought to turn, therefore, to the search for this other reason; if they cannot provide it, they should concede defeat.
FSCs It is past time to introduce an FSC. I do not propose to enter into the dialectic which has grown up around such cases (the complexity and the inconclusiveness of which itself gives us reason to doubt its efficacy as a stalemate breaker); I shall therefore limit myself here to a relatively simple example: Jones is operated upon for a brain tumor by the brilliant but twisted neurosurgeon Black. Unbeknownst to Jones, Black installs a microchip in Jones s brain during the course of the operation. The device has two settings: monitor, in which it produces moment-by-moment images of Jones s brain, and intervene, in which it alters the activation of Jones s neural networks. Black intends to use his device to cause Jones to vote for Dick Cheney in the forthcoming election. If Jones shows some sign (a certain pattern of neural activation) which is associated with being in the process of deciding to vote for Hilary Clinton instead, Black will intervene and cause Jones to vote for Cheney. As it happens, however, Jones decides to vote for Cheney on his own; Black does nothing other than monitor Jones s neural states. Black s device does not alter Jones s deliberative processes or decision-making in any way. It merely monitors without intervening. Since Jones makes his decision entirely on his own, he seems to be responsible for it (absent any story about coercion or mental illness). But because Black waited in the wings to ensure that if Jones showed any sign of making any other choice, he would intervene to bring about precisely the choice that Jones actually made, he lacked alternative possibilities. Hence Jones is responsible despite a lack of alternative possibilities which demonstrates that PAP is false. Much of the debate over FSCs has centred around the question whether these cases must assume causal determinism. Incompatibilists present Frankfurt-style compatibilists (as Fischer (2002) calls the fraternity of which he is perhaps the most distinguished member) with a dilemma: if the relationship between the prior sign, which is Black s signal for intervention, and Jones s action is deterministic, then it begs the question against libertarianism to claim that Jones is morally responsible despite lacking alternative possibilities, but if the relationship between the prior sign and the action is indeterministic, it is false that he lacked alternative possibilities (Kane 1985, 1996; Widerker 1995). Replies to this line of argument have come in two major forms: first, many Frankfurtstyle compatibilists (and others) have attempted to construct cases which circumvent the Kane-Widerker objection, by explicitly assuming an indeterministic relationship between the prior sign and the subsequent action (Hunt 2000; Mele & Robb 1998), and second, by arguing that proponents of Frankfurt-style compatibilism need not assert that agents in FSCs are morally responsible, but instead that these cases show that if they are morally responsible, it cannot be simply in virtue of lacking alternative possibilities. In other words, we can concede that the relationship between the prior sign and the subsequent action is or might be deterministic, and that therefore the agent might not be morally responsible, but given that it is intuitively plausible that these agents do not lack moral responsibility because they lack alternative 2
possibilities, it is incumbent on the incompatibilist to show how determinism is supposed to rule out responsibility (Fischer 2002). The debate over the Kane-Widerker objection and the various possible replies to it has become immensely complex. I do not propose to add to that complexity by entering any further into it (for the latest reply on behalf of the incompatibilist, see Goetz 2005). Instead, I want to explore another tack, pursuing a path suggested by Michael Smith (2004a; 2004b). Finking Frankfurt Smith denies that agents in FSCs lack alternative possibilities. On his view, Jones can do otherwise because he would do otherwise, in a nearby possible world. In order to hone in on the relevant world(s), Smith has recourse to the idea of finkish dispositions. A disposition is finkish (ignoring various complications) if the very conditions which elicit it also and simultaneously prevent its manifestation. Consider, for instance, Mark Johnston s shy but powerfully intuitive chameleon (Johnston 1993: 119). Johnston s chameleon is green, but, because it is so shy, it blushes bright red whenever it intuits that it is about to be put in a viewing condition. The chameleon is therefore disposed to look green, in normal viewing conditions, unless and until it is actually about to be in normal viewing conditions. Now, despite the fact that the chameleon never actually looks green, it still seems right to say that it is disposed to look green. Dispositions are constituted by intrinsic properties of objects, and the chameleon never loses the intrinsic properties that constitute its disposition to look green. Instead, the disposition is merely masked by another property the chameleon possesses: its disposition to blush red whenever it intuits that it is about to be placed in normal viewing conditions. In order to test whether the chameleon actually possesses the disposition to look green, we must abstract away from its properties except the intrinsic properties of its surface. When we do that, we find that in fact the chameleon does indeed have the disposition to look green in normal viewing conditions; a disposition constituted by intrinsic properties of its surface. Similarly, Smith claims, we ought to assess whether Jones has the ability to do otherwise in FSCs by abstracting away from everything in the scenario other than those properties that constitute his power to choose freely and responsibly. We test whether Jones has the requisite ability by asking whether, in various counterfactual scenarios, he might have chosen and acted otherwise. The relevant counterfactuals are those in which Jones has the all the intrinsic properties that constitute his ability to choose, but in which the masking counterfactual intervener is absent. We ought to ignore the counterfactual intervener, in honing in on Jones abilities, since the counterfactual intervener is obviously not in any way constituted by any intrinsic property of his. When we do that, we find that though the counterfactual intervener masks Jones s ability to do otherwise, it does not remove it: Such an agent is thus free and responsible because, abstracting away from Black s presence, he could have done otherwise (Smith 2004a: 103). FSCs therefore fail to falsify PAP. 3 3
Finks and Powers Several thinkers have mounted arguments against Smith s proposal, as well as closely related defences of PAP. These criticisms share a common feature: they hold that the powers that Smith takes agents to retain in FSCs, and which he believes provide them with genuine alternative possibilities, can be finked so thoroughly that it is no longer true that we can bracket the fink when we determine which possible worlds are relevant to assessing the agent s abilities. Versions of this objection are due to Pereboom (2001) and Cohen and Handfield (under submission); they differ in the precise target of the objection. Cohen and Handfield target the claim that the agent has alternative possibilities available to her at t, whereas Pereboom targets the claim that the agent possesses a power of dual control that is necessary for moral responsibility. These objections are crucially different and must be considered separately. Cohen and Handfield concede that Smith s claim that agents in common-orgarden FSCs possess alternative possibilities. However, they believe that his victory is merely temporary: we can construct an FSC in which (a) the agent is morally responsible despite the presence of the counterfactual intervener but (b) the masking intervener entirely obliterates alternative possibilities. Consider a classic case from the free will literature, Frankfurt s willing addict (Frankfurt 2003). The willing addict has an irresistible desire to take her drug; a disposition which is constituted by her addiction (the picture of addiction assumed here is rather implausible (Levy 2005; 2006) but let that pass). However, the addict never actually enters into the circumstances that would elicit this disposition: she takes the drug willingly, and therefore never experiences the cravings that would impel her to consume. Cohen and Handfield argue that her addiction plays the same role, in her behavior, as a counterfactual intervener in an FSC. Willing addict acts on her own; that is to say, her addiction does not alter her deliberative process or causally impact upon her actions. For that reason, she is morally responsible for ingesting her drug. But does she have the ability to do otherwise? If not, PAP is false. Cohen and Handfield argue that, for all Smith can show, Willing Addict does not possess the ability to do otherwise. If Willing Addict s addiction is an intrinsic property of hers (which seems plausible, on the assumption that persons are at least partially constituted by their bodies), then she lacks the ability to do otherwise: intrinsic duplicates of Willing Addict will be forced to take the drug in all possible worlds. Counterfactual interveners are extrinsic to agents, and therefore we can rightly bracket them when we consider counterfactuals, but we cannot bracket an agent s own intrinsic properties. Willing Addict has had the ability to do otherwise not merely masked, but radically masked. But if Willing Addict can be (a) responsible for her behavior in spite of (b) lacking the ability to do otherwise, PAP is false. It seems, therefore, that if we are to show that Willing Addict has alternative possibilities, and thereby rescue PAP, we must deny that her addiction is an intrinsic property of hers. Cohen and Handfield argue that this move won t 4
succeed. Any psychological duplicate of Willing Addict would have the addiction; it is therefore intrinsic to her. Any attempt to bracket her addiction when attempting to hone in on her abilities would be ad hoc, Cohen and Handfield argue, and would beg the very question at issue. They drive home the point by introducing another character from Frankfurt (2003): the unwilling addict, who is forced to consume his drug against his own will. Unwilling Addict seems blameless for his consumption; surely that is because he lacks the ability to resist? But on what basis can we say that Unwilling Addict lacks this ability, while Willing Addict possesses it? If the addiction is intrinsic to Unwilling Addict, then it must be intrinsic to Willing Addict too in which case both lack the ability to refrain from taking the drug. But if Willing Addict has the ability to refrain from taking the drug, then so must Unwilling Addict possess this ability in which case we are left at a loss to explain our disposition to excuse him from moral responsibility. In other words, Smith is committed to treating Willing and Unwilling Addict on a par, morally speaking, but that conflicts with our settled intuitions with regard to their moral responsibility for their behavior. There are many possible lines of reply to Cohen and Handfield. We might (following Smith) distinguish between sets of intrinsic properties. Smith distinguishes between two properties that Johnston s chameleon possesses, which underlie conflicting dispositions. We can say that the chameleon possesses the disposition to look green in normal viewing conditions even if its disposition to blush red when it intuits it is about to be put in normal viewing conditions is also intrinsic. We can abstract away from one set of intrinsic properties in order to assess the dispositions constituted by another (Smith 2004b: 121). But if the question whether the agent is able to do otherwise concerns what possible worlds are actually accessible to her, right then, then this response is unattractive. We can bracket one set of intrinsic properties in order to discover what dispositions are underwritten by the other, but if the bracketed properties regularly trumps the non-bracketed, we do not thereby uncover alternative possibilities that are genuinely available to her. However, there are other lines of reply that might be developed. The most obvious is to resurrect a response which was often aimed at early versions of FSCs: the flicker of freedom strategy (Fischer 1994; 2003). In the debate over FSCs, it has come to be accepted by many that agents lack robust alternative possibilities because any alternative is too exiguous to ground moral responsibility. As Fischer (2003) puts it, it would be bizarre to claim that an agent is morally responsible for an action in virtue of the existence of an alternative pathway along which she does not act freely (for instance, the alternative pathway in which she exhibits a certain involuntary pattern of neural activation, or otherwise evinces the prior sign that is Black s signal for intervention). But on a plausible interpretation of Willing Addict, there is a great deal of room for robust alternative possibilities. Willing Addict takes her drug willingly (by hypothesis), and never experiences the cravings that would compel her to take it. The alternative possibility open to her is obvious: she can refrain from taking the drug long enough for her craving to kick in. If she decides to refrain from taking her drug and acts on her decision, on any plausible interpretation of what an 5
addiction is, she will succeed for some stretch of time. Willing Addict cannot refrain indefinitely from consuming her drug, but she can transform herself from a Willing Addict to an Unwilling Addict: an addict who consumes against her will. She can settle her will on resisting. In so doing, she can actualise an alternative possibility, an alternative possibility which passes Pereboom s plausible test for robustness ( she could have willed something different from what she actually willed such that she understood that by willing it she would thereby have been precluded from the moral responsibility she actually has for the action (Pereboom 2001: 26)). However, Cohen and Handfield deny that such a robust alternative possibility is open to Willing Addict. They stipulate (in correspondence) that there is no gap between Willing Addict s volition and her compulsion: had Willing Addict hesitated over taking her drug, at that precise moment her addiction would have kicked in, and forced her to take it. Hesitating over a decision hardly seems a robust alternative possibility; it seems too exiguous to ground Willing Addict s responsibility for her action. There are, however, problems with this response. The more convincing that Cohen and Handfield make the case that there is no robust alternative possibility available to Willing Addict, the less convincing is their case that her addiction is an intrinsic property of hers. The intuition that it is intrinsic depends upon the thought that it, unlike the counterfactual intervener, cannot be set aside. But when we endow the addiction with the strange powers that Cohen and Handfield attribute to it the power to intervene at precisely the right moment, just as though it were an agent or quasi-agent it no longer seems plausible to regard it as a part of her psychology. In order to continue to think of it as an intrinsic property of hers, we must assume that it has the following odd combination of properties: it is a disposition which (a) constitutes a part of who she really is, but (b) a disposition that is never manifested, while, however, (c) remaining poised to manifest instantly the agent hesitates. This combination of properties is, I submit, so unlikely, and so contradictory, that we cannot easily keep it all in mind, and any intuitions pumped by it should be regarded with a great deal of suspicion. It is likely that when we regard Willing Addict s addiction as extrinsic to her when we assess her as responsible for her behavior, and when we regard the disposition as intrinsic we no longer regard her as responsible for her actions. 4 In any case, we need not settle the question whether Willing Addict lacks alternative possibilities as a result of her addiction. However we decide on that issue, it is clear that Smith is wrong in claiming that agents in FSCs have alternative possibilities. As McKenna (1998) notes, philosophers who defend views like Smith s appear to run together two issues: what alternatives an agent in an FSC has available to her, and what her powers and capacities actually are. 5 Possible worlds in which the fink is absent are relevant for assessing the latter question, but not the former. Consider Campbell s most recent formulation of his view that agents in FSCs have alternative possibilities. Suppose we ask whether a certain person can play One O clock Jump now. We might be asking (at least) two quite different questions. We might be asking whether she is a competent musician who knows the standard jazz repertoire, or we might be asking whether she has the ability plus the opportunity the time, space and equipment 6
to play the song right now. In other words, we might be enquiring after her powers, or about the alternatives currently available to her (where the possession of the relevant powers is a necessary, but not sufficient, condition of having the relevant alternative currently available to her). Campbell claims that when we ask whether she has the alternative available to her, we bracket facts like the unavailability, right now, of a piano. If she has the ability to play the piano, then, on the relevant facts account he defends, she can play the piano right now, regardless of the lack of the piano. Similarly, on the relevant facts account of alternative possibilities, agents in FSCs can do otherwise right now. But the sense in which a pianist can play One O clock Jump right now, in spite of the current lack of an instrument, is what we might call the power-sense of can, not the occurrent-sense. She can play One O clock Jump inasmuch as she possesses the requisite talents and knowledge; she has the right powers to play it. But she cannot play it on request; she can t actually perform the song in the absence of a piano. And in that sense, agents in FSCs lack alternative possibilities. And it is that sense, the occurrent-power sense of can, that is the sense at issue in PAP. That principles states, recall, that agents are morally responsible for their choices at t iff they could have done otherwise at t. In that sense, Jones can t do otherwise: he can t at t, vote for Hilary Clinton. For all Smith and Campbell show, then, FSCs do falsify PAP. Alternative Possibilities and Powers This is not to say, however, that Smith is not looking in the right direction when he draws our attention to finkish dispositions in the context of FSCs. Smith takes his dispositional analysis of ability to show that the agent could have done otherwise, even in the presence of a counterfactual intervener. The claim fails, as we have seen, in the occurrent-power sense of can. But it might succeed in the power-sense, and that might be sufficient in the dialectical context. The fact that an agent lacks alternative possibilities, in the occurrent-power sense, does not entail that she lacks them in the power-sense. Return to our shy chameleon. The chameleon possesses the intrinsic disposition to look green in normal viewing conditions, but that disposition is not actually manifested in normal viewing conditions. It is masked by another disposition, a disposition which might also be intrinsic. It may be that the chameleon will never look green; nevertheless it will continue to possess this intrinsic disposition despite that fact. Similarly, we might claim, agents in FSCs retain the power to do otherwise, despite the fact that temporarily or permanently that power is masked. Before turning to the question of why the retention of this power might matter to judgments of moral responsibility, it is worth pausing to see whether this power might not itself be vulnerable to radical finking. We need, in this context, to reconsider Cohen and Handfield s objection, as well as an argument due to Pereboom, which is explicitly aimed at the power-sense of can. Pereboom (criticizing McKenna (1997)) asks us to consider the case of a mathematician suffering from a brain tumor, which puts pressure on his brain such that he is no longer able to perform cutting-edge mathematics. Perhaps we could still say that 7
he possessed the ability to do cutting-edge mathematics if the tumor was operable, but suppose that it is inoperable, in the sense that it cannot be removed from his brain without killing him. In that case, we cannot say that he retains the ability to do cutting-edge mathematics, Pereboom argues. Now imagine the FSCanalogue of this case: A mad neuroscientist implants a device in a Jones-like agent which prevents that agent from doing anything other than what he in fact does (and additional conditions on robustness are met). Pereboom claims that if the device cannot be removed without killing Jones, Jones has permanently lost the ability in question (2001: 28). In that case, the possible worlds which are relevant to assessing whether he had or lacked alternative possibilities are worlds in which the counterfactually intervening device remains in place: in the both the occurrent and the power-sense of can, he cannot do otherwise. Pereboom s argument is obviously closely parallel to Cohen and Handfield s. Though Cohen and Handfield target the occurrent-power sense of can, both arguments turn on the claim that powers can be radically finked: if the masking property which prevents the manifestation of the disposition is itself intrinsic, then the power that is not manifested has been lost, not merely masked. Here, however, we can rehabilitate Smith s claim that intrinsic properties can themselves be bracketed when we are considering the relevant counterfactuals (200bb: 121). Even if the shy chameleon s shyness is intrinsic, and always trumps its disposition to look green, it remains disposed to look green; we assess what its intrinsic dispositions are by bracketing everything that prevents their manifestation, without regard to whether the bracketed dispositions are themselves intrinsic. 6 If these considerations are correct, agents in FSCs cannot lose the ability, in the power-sense, to do otherwise not, at least, without also failing to be morally responsible for their action. If their actions are normal if they are product of the kinds of mechanisms that are sensitive to reasons in the right kind of way then they retain the relevant ability. Indeed, this claim follows more or less directly from a widely accepted definition of the relevant reasons-responsive mechanisms: they are mechanisms which are sufficiently sensitive to reasons such that they would recognize reasons for doing otherwise, and actually do otherwise in relevant possible worlds (Fischer and Ravizza 1998). But since the means of assessing reasons-responsiveness just are the means for assessing whether an agent possesses the relevant (power-sense) ability, agents who are morally responsible in FSCs possess the ability. Power-Sense Abilities and Moral Responsibility Why does establishing the fact that morally responsible agents in FSCs possess the ability, in the power-sense, to do otherwise, matter? Recall the dialectical situation. FSCs are appealed to in order to show that we ought not to be afraid of determinism. Libertarians, it is often claimed, think that determinism is incompatible with free will because if determinism is true, we lack the power to do otherwise. FSCs are wheeled in at this point in the debate, in order to show that this worry is misplaced: we do not need the power to do otherwise in order to be responsible for our actions. Thus, if determinism is incompatible with free 8
will (on the implicit assumption that free will is the property of agents that underlies the attribution of moral responsibility) it cannot be because it rules out alternative possibilities. But the argument from finkish dispositions gives incompatibilists a means of replying to the Frankfurt-style compatibilist. FSCs demonstrate that in the presence of counterfactual interveners, the power to choose otherwise is masked, and that agents can still be responsible when this power is masked. In other words, FSCs are counterexamples to PAP. But they do not show that we can be morally responsible in the absence of the ability, in the power-sense, to do otherwise. Why does that matter? Because it is open to the incompatibilist to claim that the power-sense ability to do otherwise is incompatible with determinism. The argument from finkish dispositions may come to the aid of defenders of (something like) PAP after all. Despite the fact that PAP is false, its power-sense analogue may be true. Call this principle PAP*. PAP*: An agent is morally responsible for his or her choice at time t only if he or she possessed the power-sense ability to do otherwise at t. In order to assess whether this power-sense ability is compatible with determinism, we need to know more about what, precisely, it consists in. How ought it to be analysed? Should it be understood to consist in guidance control the kind of control that agents have when their action issues from a moderately reasons-responsive mechanism of their own which Fischer and Ravizza (1998) claim is compatible with determinism? Or should it be understood to consist in something like plural voluntary control (Kane 1996) roughly, the ability deliberately to bring about incompatible states of affairs, holding the laws of nature and the past fixed? But this question seems to return us to the question of whether notions like ability could have done otherwise and related terms should be understood in compatibilist or an incompatibilist terms the very debate that FSCs were invoked by Frankfurt-style compatibilists to avoid. If the forgoing argument is correct, FSCs don t help us very much. They fail to break the dialectical stalemate that had developed between incompatibilists and compatibilists, as they argued over how notions like ability should be understood. FSCs establish that PAP is false, and that is an important finding. But, Frankfurt-style compatibilists to the contrary, they do not alter the landscape of the debate or shift the burden of proof to the incompatibilist. So far as the larger debate is concerned, they leave matters much as they were before they burst onto the scene. NOTES 1 This formulation ignores the possibility of tracking back accounts of moral responsibility, which hold that an agent can be responsible for φ-ing at t iff she was able to perform some action at a time prior to t, such that performing that action would have enabled her, or led to her, not φ- ing at t. I set such accounts aside for the sake of simplicity. Nothing much turns on this exclusion, 9
however, inasmuch as if tracking back accounts require alternative possibilities, they bottom out in a principle like PAP. 2 A simple conditional analysis runs as follows: an agent is able to φ or to refrain from φ-ing just in case, if she chose to φ (refrain from φ-ing) she would. Unconditional analyses of ability and related terms state, roughly, that a necessary condition of an agent s possessing the ability to φ or to refrain from φ-ing is that both φ-ing and refraining from φ-ing is causally open to her, holding the past and the laws of nature fixed. Local miracle accounts of ability claim that agents are able to perform some action A such that, were they to A, the laws of nature would be violated; backtracking accounts hold that agents are able to perform some action A such that, were they to A, the past would be different to the actual past. For a thorough discussion and defence of local miracle and backtracking accounts of ability, see Kapitan 2002. 3 Somewhat similar accounts of alternative possibilities in FSCs are defended by McKenna (1997) and Campbell (1997; 2005). 4 Cohen has also suggested another response (in correspondence). He asks us to suppose that the desire to consume overdetermines the behavior: the addict experiences the craving, but because she consumes willingly, it is her independent desire to consume the drug that is responsible for her action. In that case, she is responsible for taking the drug, even though were she to hesitate she would be compelled to take the drug and she therefore lacks alternative possibilities. The first thing to ask, with regard to this scenario, is whether the addict experiences her craving as irresistible. If she does, then it is far from clear that she is responsible for her action of taking the drug; she takes herself to have a constrained set of alternatives, between which to choose (willing or unwilling consumption), and when we face such options, we are responsible only for choosing between them, not for failing to choose the option in this case, refraining from consuming at all that we did not believe open to us (see Levy 2002). Suppose, on the other hand, she does not experience herself as compelled to choose. Then there is a robust alternative possibility available to her: she could have willed something resisting, or, as it actually would turn out, attempting to resist such that, as she understands, by willing that action, she would have been precluded from responsibility for the action. 5 In correspondence, Smith clarifies that his considered view concerns the powers of agents, and that there is a clear and obvious sense in which agents in FSCs do in fact lack alternative possibilities. Some of his formulations in his 2004a and 2004b, however, seem to warrant the interpretation of his view advanced by Cohen and Handfield. 6 It s also worth noting that Pereboom s thought experiment seems to me to be unreliable as an intuition pump, in ways analogous to Cohen and Handfield s. Pereboom seems to be trying to have it both ways: He wants it to be true that the agent acts in the normal way that the device plays no role in the actual sequence and yet also wants it to be true that the device is in some way intrinsic to the agent. To the extent to which he convinces us that Jones s action is normal, we cannot clearly conceive of the device as necessarily linked to the agent, and to the extent to which we are convinced that the device is in principle inseparable from Jones, we cannot clearly conceive of his action as normal. Pereboom can stipulate that action is caused in the normal way and that the dual power is radically masked, but it is far from clear that the described scenario is consistent. 10
References Campbell, J.K. 1997. A Compatibilist Theory of Alternative Possibilities. Philosophical Studies 88: 319-330. Campbell, J.K. 2005. Compatibilist Alternatives. forthcoming in Canadian Journal of Philosophy. Cohen, D. & Handfield, T. Under submission. Finking Frankfurt. Fischer, J.M. 1994. The Metaphysics of Free Will. Cambridge, MA: Blackwell. Fischer, J.M & Ravizza, M. 1998. Responsibility and Control: A Theory of Moral Responsibility. Cambridge: Cambridge University Press. Fischer, J.M. 2002. Frankfurt-Style Compatibilism. In Sarah Buss and Lee Overton (eds) Contours of Agency: Essays on Themes from Harry Frankfurt. Cambridge, MA: The MIT Press, 1-26. Fischer, J.M. 2003. Responsibility and Alternative Possibilities. In David Widerker and Michael McKenna (eds) Moral Responsibility and Alternative Possibilities. Burlington, VT: Ashgate, 27-52. Frankfurt, H.G. 2003. Freedom of the Will and the Concept of a Person. In Gary Watson (ed) Free Will. Oxford: Oxford University Press, 322-336 Goetz, S. 2005. Frankfurt-Style Counterexamples and Begging the Question. Midwest Studies in Philosophy XXIX: 83-105. Hunt, D.P. 2000. Moral Responsibility and Unavoidable Action. Philosophical Studies 97: 195-227. Johnston, M. 1993, Objectivity Refigured: Pragmatism Without Verificationism. In John Haldane and Crispin Wright (eds), Reality, Representation, and Projection. New York: Oxford University Press, 85-130. Kane, R. 1985. Free Will and Values. Albany: State University of New York Press. Kane, R. 1996. The Significance of Free Will. Oxford. Oxford University Press. Kapitan, T. 2002. A Master Argument for Incompatibilism? In Robert Kane (ed) The Oxford Handbook of Free Will. Oxford: Oxford University Press, 127-157. Levy, N. 2002. Excusing Responsibility for the Inevitable. Philosophical Studies 111: 43-52. Levy, N. 2005. Addiction, Autonomy and Ego-Depletion. Bioethics, forthcoming. 11
Levy, N. 2006. Autonomy and Addiction. Canadian Journal of Philosophy, forthcoming. McKenna, M. 1997. Alternative Possibilities and the Failure of the Counter- Example Strategy. Journal of Social Philosophy 28: 71-85. McKenna, M. 1998. Does Strong Compatibilism Survive Frankfurt Counter- Examples? Philosophical Studies 91: 259-264. Mele, A. R. & Robb, D. 1998. Rescuing Frankfurt-Style Cases. Philosophical Review 107: 97-112. Pereboom, D. 2001. Living Without Free Will. Cambridge: Cambridge University Press. Smith. M. 2004a. A Theory of Freedom and Responisibility. In Ethics and the A Priori: Selected Essays on Moral Psychology and Meta-Ethics. Cambridge, Cambridge University Press, 84-113. Smith. M. 2004b. Rational Capacities. In Ethics and the A Priori: Selected Essays on Moral Psychology and Meta-Ethics. Cambridge, Cambridge University Press, 114-135. Widerker, D. 1995. Libertarianism and Frankfurt s Attack on the Principle of Alternative Possibilities. Philosophical Review 104: 247-261. 12