The Myth of Source Bernard Berofsky Columbia University

If determinism is a threat to freedom, that threat derives solely from its alleged eradication of power. The source incompatibilist mistakenly supposes that special views about the self are required to insure that we are the ultimate source of and in control of our decisions and actions. Source incompatibilism fails whether it takes the form of Robert Kane's event-causal libertarianism or the various agent-causal varieties defended by Derk Pereboom and Randolph Clarke. It is argued that the sort of control free agents need to possess and exercise can be secured without metaphysical excess. If there is a free will problem, it is the one G. E. Moore addressed in 1912. He concluded that persons can act otherwise in a deterministic world. We should continue to try to figure out whether he was right or wrong.

Keywords: freedom, determinism, compatibilism, libertarianism,

source incompati-

bilism, agent causation. Throughout most of the twentieth century, the problem of free will was regarded as the problem of understanding the relation of determinism to power--the power, once exercised, of being able to have acted otherwise. During the last generation or so, however, various concepts of self-determination (or autonomy) have nudged the concept of power--however the latter notion is understood--from center stage. Some of this is Harry Frankfurt's (1969, 1971) doing since he both forces a distinction between the two notions, power and self-determination, and challenges the identification of freedom with power, open options, or alternative possibilities. We now have a plethora of choices. There are: 1. Compatibilists who believe that freedom is essentially self-determination (Frankfurt, 1988); 2. Compatibilists (in the tradition of David Hume) who believe that freedom is essentially power (Ayer, 1954); 3. Compatibilists who believe that freedom requires both self-determination and power (Watson, 1987); 4. Incompatibilists who believe that freedom is essentially self-determination (Pereboom, 2001 );

4

The Myth of Source

5. Incompatibilists who believe that freedom is essentially power (van Inwagen, 1983); 6. Incompatibilists who believe that freedom requires both self-determination and power (Kane, 1996). This being philosophy, there are even more variants. We have not brought moral responsibility into the picture yet. John Fischer (1994) is a compatibilist about moral responsibility, but an incompatibilist about freedom. Bruce Waller (1990) is a compatibilist about freedom, but an incompatibilist about moral responsibility. Also, although we expect compatibilists and incompatibilists to differ about the nature of the self, a central rift exists within incompatibilism between conceptions of the self as transcendent (Chisholm, 1966; Campbell, 1967; Reid, 1969; O'Connor, 2000; Clarke, 2003) and conceptions of the self as resident in nature (Kane, 1996). We shall lump together under "libertarianism" (1) the standard view that freedom requires indeterminism plus agent causation (ACL), (2) the pessimism of those who agree with (1), but deny the existence of freedom on the grounds that agent causation is an illusion (for conceptual or empirical reasons), and (3) event-causal libertarianism (ECL), the view that freedom requires indeterminism plus self-determination understood naturalistically rather than in terms of agent causation. Although I have serious reservations about the identification of freedom with selfdetermination (Berofsky, 1995: ch. 9), most compatibilists do not find this identification problematic and, as a compatibilist, I will not challenge it here. Divergence between compatibilist and incompatibilist appears when conceptions of the self are adumbrated. But I will be neither defending compatibilism nor directly challenging the nonnaturalistic construal of self-determination--agent causation. Indeed, I am delighted to let the opposition fight my battles. All these internal libertarian squabbles have produced powerful arguments against the coherence and/or reality of agent causation. These arguments are found in defenders of ECL like Robert Kane (1996, 2002) and in pessimistic libertarians, like Randolph Clarke (2003), Derk Pereboom (1995, 2001), and Saul Smilansky (2000). I will argue that, if determinism does constitute a threat to freedom, that threat derives solely from its (alleged) threat to power. As a compatibilist, I believe that such a threat is bogus; but I want to allay the fear that special views about the self are required to insure that we are the ultimate source of our decisions and actions. In focusing specifically on this view--source incompatibilism--I will be able to steer clear of most of the objections to libertarianism already found in the literature. More specifically, I will evaluate the appeal to source by the leading proponent of ECL, Kane, and by two of the leading proponents of ACL, Randolph Clarke and Derk Pereboom.

Robert Kane

"Focusing on the power to do otherwise and alternative possibilities alone is just too thin a basis on which to rest the case for incompatibilism" (2002: 59). "The

Bernard Berofsky

5

concern for the origins of our will leads incompatibilists to UR (ultimate responsibility)--to the demand that the ultimate source of our wills lies in us and not in something outside of us" (2002: 71). Kane is certainly right. After all, I might be free to choose among alternatives that have been forced upon me. For this reason, compatibilists too generally insist on self-determination; so there will be agreement by all parties on certain requirements at a certain level of generality. It turns out that the agreement is surprisingly extensive. For it encompasses the very nature of self-determination. Kane's characterization is indistinguishable from the compatibilist's! "What makes the woman's choice her o w n . . , is that it results from her efforts and deliberation which in turn are causally influenced by her reasons and her intentions . . . . And what makes these efforts, deliberation, reasons, and intentions hers is that they are embedded in a larger motivational system realized in her brain in terms of which she defines herself as a practical reasoner and actor. A choice is the agent's when it is produced intentionally by efforts, deliberation, and reasons that are part of this self-defining motivational system and when, in addition, the agent endorses the new intention.., created by the choice" (2002: 234-235). Since this description (call this account of self-determination KSD) can be fulfilled in a deterministic world, it is from that perspective an acceptable description of the self to a compatibilist. The neutrality of KSD commits Kane to acknowledge as self-determined many agents whose entire lives---every decision and every action-are determined.1 So even if this woman never performs one of Kane's "self-forming acts of will," acts that are by their nature undetermined, and is, therefore, in his eyes, not ultimately responsible, she is, in his eyes, self-determined.2 UR turns out to be distinct from self-determination. Some self-determining acts are determined; but when they are self-forming, they must be undetermined. At one level, the picture sounds incoherent because the description of a selfforming act of will embodies a paradox. How can a self that is already there perform an act to form itself? One might try to remove the incoherence by treating the process of self-formation as a gradual one, one in which each self-forming act serves to determine the eventual character of the emerging self. We form our characters slowly through the cumulative force of the choices we make. But even if we ignore the special problem of the early acts when there is no formed self at all to serve as the internal source, the incompatibilist cannot anyway accommodate this familiar model. Each self-forming act must be thought of as a first cause just because it cannot be limited or constrained by past self-forming acts. To be sure, the past exerts its influence by (1) determining future acts of will that are not selfforming3 and by (2) narrowing the options and perhaps determining the desires and values that are being considered for expression; but past self-forming acts cannot determine the specific choice to express one or the other. So the paradox persists. But I want here to emphasize other concerns. Kane is acutely aware of the historical failure of incompatibilists to formulate conditions of free will given that indetermination of the will is clearly insufficient.

6

The Myth of Source

Free will requires more than the absence of an external source of the will. Indetermination might guarantee the absence of an external source, but it does not insure against a situation in which my will is formed in a completely random manner. So further conditions are required to assure that the will's actual source is characterized in the right way and it is no easy matter, as we have seen, to ascertain those conditions. If self-determination in a deterministic setting is not the answer, it is not clear how the severing of present from past that is indetermination is a better answer. It must be a better answer, Kane insists, because determinism insures that the source of the will is external, that the ultimate ground of our wills lies beyond us. Under determinism, the locus of agency falls outside the human agent, receding indefinitely into the past. So even without a clear conception of the self under indeterminism, we know that determinism will not do. Thus, Kane has more of an affinity with Carl Ginet (1990) and Hugh McCann (1998), who find freedom essentially in the absence of a sufficient cause, rather than agent causal theorists, who reject naturalistic accounts of the self. It is no surprise that Kane is uncomfortable with the label "libertarian." Before delving further into Kane's rationale for including indetermination as a requirement for UR and, therefore, free will, let us see just how much freedom we can secure without it. Suppose a particular choice meets the following two conditions: 1. The choice between A and B is the agent's (the self's) in Kane's sense (KSD) (a condition that, as we have seen, does not entail indetermination). 2. The agent possesses all the freedoms in the compatibilist repertoire--rationality, knowledge, including self-knowledge, reflectiveness, wholehearted identification with her will once she makes up her mind, emotional stability, mental health, and the absence of coercion, undue duress, subliminal conditioning, and other forms of manipulation. These conditions can be met in a deterministic world. Let us consider a third. 3. Although the agent chose A intentionally and voluntarily (and, perhaps rationally), he could have chosen B intentionally and voluntarily and acted on it. Kane would agree with non-Frankfurtian compatibilists upon the importance of (3) in an account of free will. They would, therefore, also agree that it is insufficient to characterize the power requirement in a weaker way as simply the demand that an alternative to the choice that the agent made is possible. In an indeterministic world, for example, an agent who chooses intentionally and voluntarily might not have done so. He might have just found himself not making that choice, yet without the intention to desist. This sort of alternative possibility is insufficiently robust, therefore. The sort of power possessed by a truly free agent requires a set of plurality conditions, to use Kane's phrase. That is, it must be that the agent chooses intentionally and voluntarily and that he could have chosen otherwise intentionally and voluntarily.

Bernard Berofsky

7

As an incompatibilist, Kane would insist that (3) entails that the choice is undetermined. But before examining that contention, it is worth recalling that Kane and source incompatibilists generally insist that self-determination is a special requirement that is missing from accounts that limit themselves to power or alternative possibilities. If, as Kane contends, determinism is a threat to freedom not just because it is a threat to power, we should be entitled to imagine what it would be like if the classical compatibilist is right and we can have the power to choose and act otherwise (in the robust sense that satisfies Kane's plurality conditions) in a deterministic world, 4 so that we can then see if we need to add a self-determination requirement. Since, in the actual world, we sometimes act voluntarily and intentionally under circumstances in which we do not know whether our actions (or choices) are determined, Kane must regard as intelligible the possibility that our actual voluntary and intentional actions are determined, and, therefore, the possibility that there are voluntary, intentional, yet determined actions. If I have performed such an action, Kane, as an incompatibilist, would deny that I acted freely. But if I am entitled to assume that plurality is satisfied, Kane's position is indefensible. For we have a self-determined agent choosing intentionally and voluntarily to do something such that she could have done something else intentionally and voluntarily, where the circumstances satisfy all the compatibilist freedoms. If something crucial is missing, then these conditions should not be sufficient for freedom. But they certainly seem to me to be so. Kane tries to undermine this conclusion by challenging the terms of the dialectic-my position that it is fair to insist upon robust power in a deterministic world as a heuristic device to test Kane's view that self-determination is a requirement over and above power. If we just look to Kane's insistence upon indetermination to fulfill a d e m a n d - for "ultimate origination"--that transcends the demand for self-determination, I would complain that I have no idea what indetermination brings to the mix. It sounds as if the allegedly distinct worry about sources is a disguised demand just for indetermination and the appeal to source cannot be coherently extracted as a genuinely separate concern. Source incompatibilism, at least for ECL, would then simply fail. But the picture is more complex when we probe into Kane's rationale for insisting upon indetermination. We know that it has something to do with the fact that deterministic backtracking takes us beyond the self. Of course, in one way, so does indeterministic backtracking. That is, in a self-forming act of will, it may well be that all the causal factors--the desires, values, and circumstances--are completely determined back to the Big Bang, even though the agent's act of will itself is not. For example, my desire to ski on my vacation, my (conflicting) desire to surf on my vacation, and my having a vacation this year in April may all be determined. But it is not determined that I choose to ski. What indetermination provides, according to Kane, is the possibility of the agent "setting" the will by making a decision he did not have to make. Thus, "wilt setting"

8

The Myth of Source

incorporates both features crucial to free will: indetermination and (robust) alternative possibilities (the plurality condition). Nothing external to me determines the choice and, in selecting skiing I am selecting it rather than surfing, which I could have (intentionally and voluntarily) selected. Thus, free will, which requires that I be ultimately responsible for (some) acts of will, is constituted basically by the performance of a will setting action, an action in which indetermination and plurality (the strong sort of alternative possibilities) are indissolubly linked. This is the reason Kane is not prepared to accept the dialectic as I have set it up. Robust power and indetermination are inextricably intertwined. It is unclear to me why one cannot sever the two components of will setting in order to test the source incompatibilist's contention that power is insufficient for free will. Why can I not suppose plurality without indetermination? What is Kane's argument for an indissoluble bond between indetermination and plurality in the act of setting the will? Kane's argument is that, in a deterministic world, in which the antecedent circumstances determine the outcome of an act of will, the antecedent circumstances themselves, factors distinct from the agent herself, determine the unique outcome and, therefore, set that outcome in that one direction. When the agent comes along ostensibly to select from a group of actions and set the outcome herself, it turns out that the selection has already been made. So there are no robust alternative possibilities just because there is no indetermination. Kane points out that there can be indetermination without plurality. One can stop a regress of sufficient conditions by supposing an undetermined event that happens at random, for example. But the sort of regress stopping one needs for UR is one that permits an agent's reasons for acting to be genuinely hers. Only the agent, not something else can, therefore, set the will. Setting is selecting one option to pursue voluntarily and intentionally such that if one selects A, one might have selected B intentionally and voluntarily. If B were merely possible in virtue of indeterministic randomness, will setting would not have taken place. See Kane (2000). I do not challenge the conclusion that setting entails plurality. But I do challenge the position that plurality in a context of will setting entails indetermination. First of all, in a causal chain in which each link is sufficient for the next and there is no overdetermination, each link is counterfactually and causally essential to the outcome. The batter struck out because of the pitcher's wicked curve ball and would not have done so had the pitch been different, even if the throwing of that curve ball is determined and its causes are in turn determined, etc. Nothing changes if the chain is a link of sufficient causal conditions. Here, too, the presence of a remote sufficient cause does not nullify the (determining) causal status of the proximate sufficient causes. Determinism is not fatalism. Suppose, then, that we are presented with a case of KSD that happens also to be one in which the woman's choice is determined. The reasons, intentions, deliberation, efforts, circumstances add up to a sufficient causal condition of the choice.

Bernard Berofsky

9

For Kane, then (assuming that we cannot trace this back to an earlier self-forming act), she is not ultimately responsible because she did not "set" her will. If "set" is just a synonym of"determines," then, as we just said, the fact that there are earlier determining causes does not entail that she did not really determine her choice (through her uncoerced, endorsed desires, etc.). If Kane intends "set" to mean something stronger, as might be suggested by examples of purposive global controllers, then the subject has changed from the putative threat of determinism. God may set my reasons in a way that renders any other "subsequent" event irrelevant; but then we are back to fatalism, not determinism. In spite of this rejoinder, Kane might still insist that even a conventional remote sufficient cause narrows the options to one, so that even if the proximate causes (desires, endorsement) are essential, they are like slaves rowing a boat. Even though their actions are necessary to the direction of the boat, they have no choice but to follow orders. This fear is that the remote cause violates the plurality condition by making it impossible for the agent to choose intentionally and voluntarily the option that is in fact not chosen (even though Kane is forced to concede that the remote cause does not preclude the agent's choosing intentionally and voluntarily the option that is chosen). The value of indetermination, therefore, is that it leaves open the possibility that the agent can act otherwise. Here is the wedge I need. I simply want to ask Kane if he would examine the implications of the assumption that it is a mistake to suppose that remote sufficient causes always make it impossible for an agent to act otherwise intentionally and voluntarily. He rejects that assumption and I accept it. But there is nothing incoherent about my strategy. He is free to believe that plurality is found only under indeterminism; but whichever of us is right, there is no use for more exotic requirements for free will. In other words, if Kane is prepared to leave open the fundamental issue concerning power that divides compatibilist from incompatibilist, he has no way of showing that power needs to be supplemented by some sort of indeterministic selfdetermination. Hence, a compatibilist victory on the issue of power would be complete; it is not to be compromised by some lingering worries about the ultimate source of action. The source incompatibilist appeals to the deepseated desire to be the genuine source of our decisions and actions. But once we suppose that the agent retains the power to choose, we cannot then worry that determinism reduces the agent to a passive conduit for external forces. Environment and heredity do not just sweep an agent off her feet on the assumptions we are making. F. H. Bradley (1927) worried that the construction of the self in a deterministic world would obliterate true origination or creativity. But if, as ECL says, a self is formed naturally, then the coherence of the very idea of ultimate origination, as we have seen, falls prey to a dilemma argument. If my self is already formed, then nothing, not even the self, can form it. And if my self is in the process of formation,

10

The Myth of Source

it is a conceptual necessity that its origins lie outside the sell This dilemma explains why some incompatibilists revert in desperation to rescuing the self from the world by depositing it outside that world. But it is hard to see what advantage in terms of originality is conferred on a person whose traits developed indeterministically prior to the time the self is sufficiently formed so as to exercise rational will. What difference does it make to freedom whether prerational development is determined or not? If a childhood trait is undesirable, the child who is a victim of chance is as much a victim as the person living in a deterministic world. Although these thoughts are very unoriginal, I find that incompatibilists need to be constantly reminded of them. That is not to say that it is unreasonable to demand for full freedom constraints on early development. But many of these worries can be assuaged by strengthening the nonmanipulation requirement. We turn now to ACL versions of source incompatibilism.

Agent Causation Theory Derk Pereboom contends there that the key worry about causal determinism is that it removes control from the agent (2005: 239). If that old friend, the demon neurologist, ignoring completely our current desires and values, implants two utterly alien desires that happen to conflict with one another, the power to decide between them is hardly sufficient to confer freedom upon us. If so, we should be equally unsettled by cases in which the agent does not control the circumstances that play a causal role in his behavior, even if those circumstances arise naturally, not as a result of manipulation. How might we eradicate this unsettled feeling? Before we answer this question, it must be noted that recent critics of Pereboom, such as Alfred Mele (2005), John Fischer (2004), Michael McKenna (2005), and Ishtiyaque Haji (1998) have highlighted the freedom (and responsibility) enhancing features of histories that may turn out to be deterministic. They believe that Pereboom fails to see the rich possibilities of deterministic settings for making crucial distinctions that ground differences between freedom and its absence. But let us just cut to the chase. For we are inevitably forced to confront the old refrain, "if all of our behavior was 'in the cards' before we were born, in the sense that things happened before we came to exist that, by way of a deterministic causal process, inevitably result in our behavior, then we cannot legitimately be judged blameworthy..." (Pereboom, 2005: 235). We need not, at this point, concede the day or even accept a stalemate. A preferable (and unoriginal) strategy for addressing that unsettled feeling is to ask for an elaboration of the libertarian alternative. We then learn that that bad feeling persists even when we suppose that the circumstances are undetermined! If a man contemplating murder would be swayed from his evil plan only by moral reasons recognized when (and only when) the stock market is rising, and it is just a matter of chance rather than agential manipulation that the stock market is declining,

Bernard Berofsky

11

then, as compatibilists like to say, the murderer is still a victim, but of chance rather than some devious manipulator. The stock market might have risen and he would then have been swayed by moral reasons to refrain. A suggestion that might attract a libertarian is to demand that circumstances themselves be determined by the agent in that special freedom-generating way. But, of course, this is a hopeless suggestion. Any action, free or otherwise, depends upon a large number of factors and to suppose that freedom requires that each of them be agent-caused is to limit freedom to God. Libertarians may disagree about the extent of our freedom. If they agree that freed o m requires agent causation, some see freedom as quite ubiquitous, extending along the entire range of human agency (Taylor, 1974), whereas others have doubts about its reality at all (Broad, 1952; Strawson, 1986; Pereboom, 2001; Clarke, 2003). But even though it is surely implausible to believe that all but a very small number of factors beating on our actions are possibly under our control, even if the scope of our freedom is very narrow, its nature is not diminished. Perhaps, as Descartes believed, where it is found, it is as great as God's. Randolph Clarke is another proponent of source incompatibilism who sees control as crucial: "Acting f r e e l y . . , requires the openness of alternatives, as well as an exercise of active control by the agent" (2003: 3). "Actually exercising active control (over which open alternatives becomes actual) is required for the origination, difference-making and attributability that we value; it is partly constitutive of the dignity that we think acting freely confers on us" (2003: 17). Up until now, we have been shifting freely among the expressions "power" "alternative possibilities," and "open options." The latter two expressions and sometimes the first connote a negative characteristic--the presence of an opportunity to be exploited. Clarke insists that something of value is lost if, in some way, the agent is unable to exercise causal influence to determine how this openness will be exploited. In demanding indetermination, event-causal libertarianism posits a leeway so that active control can be exercised in different ways; but does not acknowledge the significance of positive control over the action that is actually chosen. This posRive control confers upon the self the ultimate sort of freedom. Timothy O'Connor (2000: 27) registers a similar complaint. But in conceding that agent causation is an external relation between agent and action, both allow that an action's taking place implies nothing about its cause and, therefore, that an action description, say, "I determine the action of raising my arm" might just be, as Carl Ginet (1990: 12) recently and I (1971: 70) once upon a time argued, a pompous way of identifying the performer rather than an informative account of the cause of the action. Just because a libertarian's active control over action A requires that A itself be undetermined (by events or states), compatibilists often complain that, "due to this chance, agents exercise less active control than they could exercise in a deterministic world" (Clarke: 72). Clarke convincingly challenges this position, arguing that

12

The Myth of Source

chance here does not remove the action from the control of the agent, nor does it render it unintentional, lucky, or bizarre. 5 He also responds to the charge that, since, at a certain point prior to A, everything relevant to its occurrence is in place, yet it is still a matter of chance whether A occurs, "we can only wait and see" (100) whether the action occurs. He rightly notes that the agent does not wait and see; he acts (or not). Precisely. This response to the compatibilist does not invoke agent causation. In the presence of alternative possibilities, an agent intentionally decides to act in one way rather than another. The agent influences the outcome by deciding. I would argue that we have all the power we need. To defend this position, we begin by looking more closely at control. Incompatibilists worry about determinism because they suppose that it eliminates control. The account of positive control of the event-causal libertarian can avail itself of the agent's reflective capacities as he weighs the reasons for and against the various options. And when he acts, he will be acting (at least in part) because of some of those reasons. But without reference to the agent as cause, Clarke objects, we do not see how she actually exerted her influence. The agent "exercises real causal influence" (152) as a determinant distinct from reasons or desires. For Clarke, control by one's own reasons plus the presence of alternative possibilities, presumably of the robust variety that incorporates Kane's plurality conditions, still does not add up to genuine positive control. The conclusion is a non sequitur. Clarke needs more than a vague appeal on behalf of "positive" control. Sometimes a so-called "positive" state is constituted by a "negative" one. A qualified applicant may be one who does not satisfy any disqualifying condition. I will argue that, for an incompatibilist, the conditions advocated by Kane--indetermination plus a causal role for antecedent mental states-are enough. 6 A similar complaint is echoed by Pereboom (2005: 244). He claims that ACL does, whereas ECL does not, provide the agent with a causal or explanatory role. Agent causation, he insists, can explain decisions in an indeterministic setting, where antecedent conditions (like desires and beliefs) do not suffice. But, since "Ann" is not much of an explanation of"Ann decided to flip the coin at t," Pereboom is forced to cite Ann's causing the decision to flip the coin as the explanation of the fact that the decision was made. So we then have "Agent Ann caused the decision to flip the coin at t" as the explanation of "The decision of Ann to flip the coin at t took place," But we must then allow that an explanans may entail its explanandum, and there is then no way to challenge the competing view that the noncausal assertion that Ann decided to flip the coin at t explains the decision. If decisions are necessarily actions, then neither alternative is informative; the decision of Ann cannot take place unless Ann decides or, if ACT is right, Ann causes the decision. If we want to isolate an event component of the decision, then perhaps an analogy for the anti-agent causation view is the relation between "I raise my arm" (understood noncausally) and

Bernard Berofsky

13

"my arm rises" Although I cause neither the raising nor the rising, it is not entirely empty to explain the rising by the raising. Most parties accept that explanation in an indeterministic world is necessarily incomplete. But if an argument for agent causation is grounded on the position that reference to the agent succeeds in completing the explanation of the decision, the opponent is surely entitled to note that "the agent caused the decision" is no better in this respect than the simpler "the agent decided." The fear that the agent is left out of the story is bogus just because deciding is something agents do. Antecedent states on their own do not act. As I said, there seems to be agreement among libertarians that the very distinction between action and passion does not in itself implicate agent causation. So long as there is no effort to resuscitate the view of Richard Taylor (1974) that "I raise my arm" is true if and only if "I cause the rising of my arm," nothing more than the elementary distinction between action and passion is needed to provide a role for the agent. So, is anything really added by moving from "Sue decided to A" to "Sue determined that she would decide to A"? Presumably, the causal paraphernalia is needed to make sense of the idea that an agent is really controlling an outcome. I would challenge that view.

Control I have never been sympathetic with the compatibilist position that an undetermined decision cannot be one that is under the control of the agent (1971: 68-70). Consider the charge of arbitrariness. An arbitrary decision is one that the agent has no good reason to prefer over other options. The magician asks a person to pick a card and the person looks at and considers several, finally deciding on one. In what respect is the agent not in control of this arbitrary decision just because it is undetermined? I propose then to try to formulate an agent causation-free set of conditions sufficient, but not necessary, for control in an indeterministic world. 7 Since I want to undermine the need for agent causation, I only need a sufficient condition. (And since I am sympathetic to compatibilism, I would not want a necessary condition in which indetermination is essential.) Sue, an intelligent and rational agent, confronted with several options, is deliberating reflectively in order to decide to act on one of them. 1. When Sue decides to A, she A's intentionally, voluntarily, and deliberately because of the decision. 2. Sue decided to A because she judged that, in light of her desires, interests, and values, A is the best thing to do or at least as good as anything else. 8 3. All compatibilist criteria of freedom are in place. 4. The decision to A is not determined. Therefore, nothing external to Sue controls her decision.

14

The Myth of Source

5. The way in which the decision to A is not determined provides Sue with the opportunity to decide and to act on any of the options. Therefore, it is possible that Sue decides to A and it is possible that she decides otherwise. 6. Sue has all the pertinent abilities (understood in a type sense) to decide and to act on any option. Thus, given opportunity and ability, Sue has the power to decide and to act on any option, such that, if she had decided to do otherwise, she would have intentionally, voluntarily, and deliberately done otherwise (Kane's plurality condition). (This is not intended as an analysis of "power," but rather as a statement of sufficient conditions.) When Sue A's, there is no other sort of control pertinent to freewill that she lacks. Although perhaps not necessary, this agent causation-free set of conditions is sufficient for control. To dramatize this point, suppose that Sue has spent her entire life devoted to the elimination of smoking. There is nothing more important to her than her antismoking campaigns. She identifies completely with this aspect of her nature after fully informed, thorough, rational, nonneurotic reflection. She is only unhappy when she is blocked from the pursuit of this activity. She developed this passion completely naturally, without being coerced or having been indoctrinated by anyone. Fearful of her success, an unethical executive of a tobacco manufacturer secretly injects her with a nicotine-laden pill that induces in her a powerful desire for a cigarette and then leaves a pack of cigarettes at her bedside. When Sue awakens, she confronts a conflict between her aversion to smoking and her induced desire to smoke. (Perhaps Sue is not a perfect model of compatibilist freedom, but no matter.) Sue has a sufficient reason to refuse a cigarette and she does indeed refuse it. As Clarke and O'Connor have both recognized, it does not follow that Sue's action is determined. In cases of (potential) akrasia, sufficient reasons may not be sufficient causes--the person may experience akrasia. But if she does not, if she refuses to smoke, then, the action is caused by a sufficient reason, but not determined by it. Sue's action is a reflection of her true nature as much as any action can be. Given that Sue fulfills all the above conditions, is the addition to Sue's powers to smoke and to refrain of the powers to c a u s e smoking and c a u s e refusing really crucial? Let us look more closely at the nature of this special power. Powers, including causal powers, rest on the possession of certain properties. There are a variety of theories regarding the relation between properties and powers; but the point I wish to make does not depend on a proper understanding of this relation. According to Clarke, agent causation is a special type of causation in which the causal power is possessed and exercised by a substance, the agent. In ordinary event causation, on the other hand, the event (or state of affairs) carries the power. But in both cases the relation can be expressed as three-term: X confers on Y the power to cause Z

Bernard Berofsky

15

The difference between the two types of causation, according to Clarke, is only in terms of Y. For agent causation, we have X confers on S (a substance) the power to cause Z And for event causation, we have X confers on S's having property P (a state of affairs) the power to cause Z Although I found the interpretation of X to be difficult to make out, the best reading of Clarke, I believe, is as follows: Event causation: P (the property component of S's being P) confers on S's being P (the state of affairs of which P is the property component) the power to cause Z. Agent causation: P (the property component of S's being P) confers on S (the substance) the power to cause Z. Whether you view properties as particulars or universals, and whether you view the relation between properties and powers as a relationalist or a dispositionalist, both the substance and the state of affairs are dependent on identical properties and are possessed of identical powers. Yet a profound difference is supposed to reside here. Clarke characterizes this as a difference in directedness. The property confers a power either on the subject to exercise its power directly or on the subject's possession of the property, in which case the exercise is "indirect" or "via the property?' Is this a real difference? No. It is not as if the property is less essential to the existence or exercise of the power in the case of agent causation. Are we supposed to think that, in agent causation, the agent is able to bypass or skip over the property and get right to Z-production. Surely this is either a misplaced metaphor or obscurantist metaphysics. All parties agree that Sue has the power to A (smoke, refrain) when her being P is able to cause A-ing. Must we also insist for full freedom that Sue also have the power to cause her A-ing? I do not see why. Clarke claims that it is more plausible to suppose that, if there is agent causation, then substances other than human agents are also possessed of the power to cause directly. But, of course, then, only agent causation, causation by a thinking, rational substance, is relevant to the active control required for free will. Agents are not pure egos. We must also allow that the behavior of substances may be undetermined. So we can have substance causation with alternative possibilities, but no agency and, hence, no freedom. How important, then, is that directness Clarke makes so much of, when all the real work of agency is done by the properties of the agent-reflectiveness, intelligence, evaluation of norms, commitment to values?

16

The Myth of Source

Conclusion If a libertarian rejects the traditional complaint that the perniciousness of determinism rests upon its rendering agents powerless to resist forces merging upon the agent, he needs to tell us why self-determination, in a form that is incompatible with ordinary determination, sets us free. I have argued that this burden is met neither by the agent causal theorist nor his more naturalistic ally. The latter is willing to permit the self to step gingerly into the world. I invite it to come all the way in. If there is a free will problem, it is the one G. E. Moore addressed in 1912. He concluded that persons can act otherwise in a deterministic world. We should continue t o try t o f i g u r e o u t w h e t h e r h e w a s r i g h t o r w r o n g .

Notes 1.

2.

3. 4.

5.

6.

7.

8.

In fact, a strong case can be made that self-determination, as an E C L theorist sees it, is more readily obtained in a deterministic world. For indetermination can cut one off from one's self! Only in a deterministic world can we be assured that a perfectly rational agent would never stray from the dictates of her rational self. Kane's characterization of acts of self-forming will do not require that the acts be uncaused for he supposes that it is possible that an act can be caused, even if it not determined. Causes need not be or presuppose sufficient conditions. Kane allows that an agent can be ultimately responsible for a determined action A so long as there is an earlier self-forming action in the etiology of A. Classical compatibilism is the view that determinism is compatible with freedom as the power to do otherwise, Hierarchical compatibilists may or may not believe that determinism is compatible with the power to do otherwise; but they do not regard the latter as essential to freedom. This statement may seem to be inconsistent with the previous acceptance of the argument against Pereboom that the alternative to determination is chance and that does not enhance control. There are two crucial differences. First of all, Pereboom needs to suppose that indetermination enhances control, a far stronger claim than simply the point currently being made that indeterminism is compatible with control. Second, I will argue that control presupposes a formed self, one that has a rational will and a set of preferences, etc. Pereboom is looking for that impossible entity, a self without a history that controls its destiny in a freedomenhancing way. As a compatibilist, 1 do not believe that indetermination per se adds anything to the idea of control. But the point here is that, even if we add indetermination, there is no other "positive" condition that provides a deeper sort of control. Control may well be a matter of degree. But I do not wish to get into that level of complexity here. It is worth noting that Kane (2002: 235) concedes that there is a loss of control under indeterminism, although he provides an explanation of that fact consistent with his own position on free will. 1 mean here to accommodate incompatibilists who believe that free actions can be rational. Even O'Connor, who rejects any causal role for antecedent states in a free decision, would have no problem with this condition, although he might insist on a noncausal reading of "because."

Bernard Berofsky

17

References Ayer, A. J. 1954: 'Freedom and Necessity', Philosophical Essays, London: Macmillan & Co., 271-84• Berofsky, B, 2004: 'Autonomy Without Free Will', Personal Autonomy: New Essays on Personal Autonomy and its Role in Contemporary Moral Philosophy, Ed. by J. S. Taylor, Cambridge: Cambridge University Press, 58-86• • 1995: Liberation from Self, Cambridge: Cambridge University Press• • 1971: Determinism, Princeton: Princeton University Press. Bradley, E H. 1927: 'The Vulgar Notion of Responsibility in Connexion with the Theories of Free Will and Necessity', Ethical Studies, Oxford: Clarendon Press, 3-57. Broad, C. D. 1952: 'Determinism, Indeterminism, and Libertarianism', Ethics and the History of Philosophy, London: Routledge & Kegan Paul, 195-217. Campbell, C. A. 1967: In Defense of Free Will, London: Allen & Unwin. Chisholm, R. M. 1966: 'Freedom and Action', Freedom and Determinism, Ed. by K. Lehrer, New York: Random House, 11-44. Clarke, R. 2003: Libertarian Accounts of Free Will, Oxford: Oxford University Press. Fischer, J. M. 2004: 'Responsibility and Manipulation', The Journal of Ethics 8, 145-77. • 1994: The Metaphysics of Free VCill,Oxford: Blackwell. Fischer, J. M., and Ravizza, M. 1998: Responsibility and Control. A Theory of Moral Responsibility, Cambridge: Cambridge University Press, 28--41. Frankfurt, H. 1988: 'Three Concepts of Free Action', The Importance of What We Care About, Cambridge: Cambridge University Press, 47-57. - - . 1971: 'Freedom of the Will and the Concept of a Person', The Journal of Philosophy 68, 5-20. • 1969: 'Alternative Possibilities and Moral Responsibility', The Journal of Philosophy 66, 829-39. Ginet, C. 1990: On Action, Cambridge: Cambridge University Press. Haji, I. 1998: Moral Appraisability, New York: Oxford University Press• Kane, R. 2002: 'Free Will: New Directions for an Ancient Problem', Free Will, Ed. by R. Kane, Malden: Blackwell Publishers, 222--48. • 2000: 'The Dual Regress of Free Will and the Role of Alternative Possibilities', Philosophical Perspectives 14, 57-79. - - . 1996: The Significance of Free Will, New York: Oxford University Press• McCann, H. J• 1998: The Works of Agency: On Human Action, Will, and Freedom, Ithaca: Cornell University Press. McKenna, M. 2005: 'The Relationship Between Autonomous and Morally Responsible Agency', PersonalAutonomy, Ed. by J. S. Taylor, Cambridge: Cambridge University Press, 205-34• •. 2001: 'Source Incompatibilism, Ultimacy, and the Transfer of Non-Responsibility', American Philosophical Quarterly 38, 37-52• Mele, A. 2005: 'A Critique of Pereboom's "Four-Case" Argument for lncompatibilism', Analysis 65, 75-80. O'Connor, T. 2005: 'Freedom With a Human Face', Midwest Studies in Philosophy: Free Will and Moral Responsibility 29, 207-27. • 2000: Persons and Causes: The Metaphysics of Free Will, New York: Oxford University Press. Moore, G. E. 1912: 'Free Will', Ethics, Oxford: Oxford University Press, 84-95. Pereboom, D. 2005: 'Defending Hard Incompatibilism', Midwest Studies in Philosophy: Free Will and Moral Responsibility 29, 228-47. • 2001 : Living Without Free Will, Cambridge: Cambridge University Press•

18

The Myth of Source

• 1995: 'Determinism AI Dente', Nous 29, 21-45. Reid, T. I969: Essays on the Active Powers of the Human Mind, Cambridge, Mass.: MIT. Smilansky, S. 2000: Free Will and Illusion, Oxford: Clarendon Press. Strawson, G. 1986: Freedom and Belief, Oxford: Oxford University Press. Stump, E. 2003: 'Moral Responsibility without Alternative Possibilities', Moral Responsibility and Alternative Possibilities, Hams: Ashgate, 139-58. - - . 1996: 'Libertarian Freedom and the Principle of Alternative Possibilities', Faith, Freedom, and Rationality. Ed. by J. Jordan & D. Howard-Snyder, London: Rowman & Littlefield, 73-88. Taylor, R. 1974: Metaphysics, Englewood Cliffs, N.J.: Prentice-Hall. ~ . 1966: Action and Purpose, Englewood Cliffs, N.J.: Prentice-Hall. van Inwagen, R 1983: An Essay on Free Will, Oxford: Clarendon Press. - - . 1975: 'The Incompatibility of Freedom and Determinism', Philosophical Studies 27, 185-99. Waller, B. 1990: Freedom Without Responsibility, Philadelphia: Temple University Press• Watson, G. 1987: 'Free Action and Free Will', Mind 94, 145-72. ~ . 1982: 'Free Agency', Free Will. Ed. by G. Watson, Oxford: Oxford University Press, 96--110.

Received: July 2006 Revised: August 2006

Bernard Berofsky Department of Philosophy 708 Philosophy Hall Columbia University 1150 Amsterdam Avenue New York, NY 10027 U.S.A.