P. S. Greenspan

Department of Philosophy

University of Maryland

College Park, Maryland 20742

(301) 405-5689, fax. (301) 405-5670

Click for abstract

Harry Frankfurt's general contribution to the contemporary discussion of moral responsibility seems to me to be the emphasis on self-reflective capacity. A similar notion comes up in different forms in essays by a number of recent authors working from Strawsonian sources, but I find it most clearly exemplified in Frankfurt's hierarchical account of free will in terms of higher order volition. (2)

On the other hand, Frankfurt maintains that freedom in the sense of free will is not really a condition of moral responsibility. (3) Moral responsibility depends on free action; but that just involves doing what one wants, acting on whatever desires constitute one's will. It is free will that involves action in accordance with higher-level reflective desires in Frankfurt's volitional hierarchy.

Frankfurt's distinction is of interest to me in the attempt to understand cases of criminal responsibility that intuitively seem to involve unfreedom. The central cases turn on "aggressive impulsivity," a trait that recent studies suggest may be partially attributable to genetic factors operating via serotonin deficit. (4) Perhaps such cases exhibit freedom of action but not free will.

My rereading of Frankfurt's essays with that application in mind brings into focus some differences from his approach, though on the basis of a substantially revised conception of it. I think that many people have misread Frankfurt on these issues, perhaps partly because he relies so much on distinctions phrased in terms of "the will," which is essentially a remnant of faculty psychology. I take "the will" just as short for general decision-making capacities and ordinarily use "free will" just to cover the whole tangle of questions in this area. However, in this paper I shall avoid uses of the term that clash with Frankfurt's.

One more particular source of the common misreading of Frankfurt is that he intends his own use of "will" to build in reference to action, in a way that ultimately blurs his distinction between freedom of action and will: One's "will" on a given matter is a desire that actually moves one to act; and Frankfurt refers to a desire "meant" to do so as a "volition" (p. 16). My earlier readings of Frankfurt's work have been thrown off by just this sort of terminological complexity, plus absorption in some complex and difficult cases. I want to work here with some cases that are at any rate more familiar, in an effort to get a better grip on Frankfurt's account, or the elements of it that might be applied to problematic criminal cases.

1. Frankfurt's treatment of free will complements a negative argument, against the standard requirement of alternatives to what one does. This requirement, which Frankfurt sums up as the "principle of alternate possibilities" (PAP)--"a person is morally responsible for what he has done only if he could have done otherwise" (p. 1)--is put forth in the first instance as a condition of moral responsibility. But Frankfurt's initial counterexamples to the principle involve acts an agent is assumed to do with full freedom--on the basis of a decision made for "reasons of his own"--even though, unbeknownst to him, there are causes in play that would have had the same effect anyway.

A nice case along these lines that Frankfurt brings up in a later essay, "What We Are Morally Responsible For" (p. 96), is that of a "dormant addiction": Someone who is unaware that he is addicted to a certain drug decides to take it just for its desired effects, though as a matter of fact his addiction would have caused him to take it in any case. The addiction amounts to a "backup factor" that plays no role in the actual sequence of events.

The point, in a nutshell, is that moral responsibility depends on what actually explains or motivates action in a given case--its connection to the agent's will--rather than on counterfactual causes and options. Free will, on the other hand, is characterized by Frankfurt in terms of the internal structure of the agent's will--or equivalently, its relation to the agent: whether he has the will he wants.

Since "will" or "volition" is defined in relation to action in the way I have indicated--and for other reasons I want to explore in what folows--Frankfurt's positive and negative arguments are intertwined. However, it is important to note that they are not put forth as arguments about the same thing. The negative argument focuses on moral responsibility, the positive argument--and his hierarchical account--on free will.

2. What Frankfurt discusses as the typical case of unfree will on his hierarchical account involves motivational conflict--between more immediate desires for action and a higher level of self-reflection, which he also conceives in terms of desires: desires about one's more immediate desires (pp. 16-22). One acts without free will insofar as the lower-level desires win the conflict.

A familiar example of unfreedom in this sense would be an agent who cannot manage to carry out the desire to give up smoking, interpreted as a desire not to be motivated effectively by the desire to smoke. As we shall see, this may or may not amount to a desire not to have the desire to smoke, as in one common misreading of Frankfurt's view. The typical addicted smoker is unfree because she acts in opposition to higher-level reflective desires: She does not want to be a smoker, but she smokes. She smokes because, on the level of immediate desire, she wants to.

My description of the case assumes that the agent in question does decisively commit herself on the upper level, rather than being in conflict, say, about how much her health really matters to her--and rather than simply pursuing her lower-level desires without reflection, on the other hand, in the manner of what Frankfurt calls a "wanton" (p. 16).

However, as these two assumptions--that the agent acts "wholeheartedly" and is not a "wanton"--are explained and expanded in Frankfurt's later essays, questions of moral responsibility resurface in connection with the first: what Frankfurt calls "identification," meaning identification of the self with certain of its desires. (5) Wholehearted self-identification, the resolution of any conflict on the upper level, seems to be understood as a requirement of moral responsibility--part of what it means to "take" responsibility for action on a lower-order desire like the impulse to smoke (see pp. 47-54 and pp. 53-54; cf. pp. 159-76).

For reasons that will emerge in what follows I think the assumption of wholeheartedness really works better as a requirement of free will than of responsibility in the sense Frankfurt has distingished from free will at the outset. The assumption comes up initially as a way of stopping appeals to higher levels of desire and hence an infinite regress in Frankfurt's volitional hierarchy. We might sum up his view as saying that free will requires clear-cut upper level commitments that do in fact yield action--so that we can attribute action unambiguously to the reflective agent.

An unfree smoker or other addict who is not a "wanton" on Frankfurt's account would view her addicted desires as external to herself. Frankfurt's examples typically involve impulses or excesses that the agent upon reflection disowns as shameful or degrading (see pp. 58-68 and pp. 159-76). However, from my own days as a smoker, I recall no conflict about being a smoker per se, except to the extent that this involved smoking, a habit with ill effects. If it had been just as easy to give up the behavioral habit without giving up the desire, that would have been adequate for my purposes.

On rereading Frankfurt, I think he would agree. He could grant me free will if I had managed simply to act on upper-level desires not to smoke, even while continuing to experience conflicting lower-level impulses. (6) Free will, as I now understand his view, does not require absence of motivational conflict, but just a clear-cut practical victory for the level of reflection. This is compatible with conflict on the lower level or between the two levels, though I think it presupposes clear commitment on the upper level.

3. On the other hand, lack of conflict among desires at the upper level--wholeheartedness--is not sufficient for freedom. As Frankfurt notes (pp. 24-25), a willing addict--someone who is quite content to be a smoker, say--also lacks free will insofar as her lower-level desires would have driven her to do the same thing in any case. (7) In that sense she would not be smoking particularly because of her upper-level desires and hence would not be free--contrary to another common misreading of Frankfurt's view--though she still may be responsible for smoking.

Frankfurt's view does demand that the upper-level desires be conceived as higher order: desires about one's desires, as effective in action. This is the element of self-reflection he stresses, in the first instance as what distinguishes human persons from children and nonhuman animals, who also may be deliberative agents but are "wantons" in the strong sense of lacking any higher order desires (pp. 11-12).

The term "wanton" has overtones of excess that do not quite fit all the relevant cases here, but this second assumption of Frankfurt's free will does supply an answer to another question about familiar sorts of cases, namely in what sense my cat is not free. A cat or other animal--one whose nature, training, and needs do not dictate a particular action in response to its circumstances--does seem to exercise a kind of unconstrained choice about what to do. However, this is not freedom of the will in Frankfurt's sense but at most free action.

In response to a call, say, my cat "Rambo" seems to have several options: Sometimes he comes at once; sometimes he responds just with eye and tail movements (and perhaps comes in a few moments, but often not); and in some cases he just ignores the call completely. It does not matter for my purposes whether there are unnoticed causes distinguishing these different cases but just that, at least in the intermediate cases, Rambo does decide what to do--as Frankfurt allows one to say here.

One might even say that Rambo decides on the basis of reasons: whether he feels like being petted or wants to avoid having his claws clipped or the like (combined with beliefs about what I am likely to do). What one cannot say--and despite the more imaginative reconstructions of feline thinking, I agree--is that Rambo's reasons include some notion of how a given response would accord with higher-order preferences: something on the order of a desire to be aloof or independent in a sense that entails wanting to act for reasons of his own, not just because I want him to.

So on Frankfurt's view Rambo lacks free will and (for the same reason) does not count as a person. He is a wanton--though no more or less so than a dog or other animal who reliably obeys my will, even in opposition to his own immediate impulses. What I would like to ask, though, is how his lack of free will--in contrast to my former lack of free will on the matter of smoking--is related to the question of moral responsibility.

I assume that Frankfurt also wants to say that along with other nonhuman animals Rambo lacks the requisites of moral responsibility. I, on the other hand, was responsible for smoking for many years, despite my lack of free will, because I did have self-reflective desires capable of influencing my behavior. But the explanation is not just that I am no wanton. Someone who was a wanton (and hence also lacked free will) on the particular question of smoking would still be said to be responsible for her behavior--in view of its effects on her long-term health prospects and medical expenses, say. She would count as a person, with a general capacity for reflective self-control--as opposed to being a wanton in Frankfurt's "strong" sense--even if she lacked any inclination toward self-control on the matter of smoking. (8)

It is easy enough to explain the relevant differences here--between Rambo, on the one hand, and both myself and the wanton smoker, on the other--in terms of intellectual development. Rambo presumably lacks the capacity for self-reflection, meaning reflection on his own desires and other mental contents. But in line with Frankfurt's aim of stressing will as opposed to intellect (see the preface to his collected papers, pp. vi-vii), I think we should say that Rambo lacks a kind of volitional capacity--the ability to modify his behavior on the basis of self-reflection--that the wanton smoker does have in general terms but fails to apply to smoking.

If so, responsibility would depend at least on some of the ingredients of free will. It would not be just a matter of free action in some completely independent sense, a sense that would seem to fit Rambo on Frankfurt's account as well as the wanton smoker. Perhaps the point may be put best in terms of Frankfurt's later comments on "taking" responsibility, by noting that Rambo is incapable of either identifying or not identifying with his lower-order desires, on the assumption that he lacks self-reflection.

To say all this involves extending Frankfurt's remarks on animals, as well as on moral responsibility in some of the trickier human cases. The main focus of his positive account is on free will. However, I think the result of extending his remarks would be intuitively satisfying, at least to the extent that it exhibits a connection between moral responsibility and free will. It would be odd if Rambo's lack of the one had nothing to do with his lack of the other.

4. Perhaps moral responsibility involves two things, then, one of them overlapping with the requirements of free will: first, freedom of action in the particular case in question, and secondly, a more general capacity for reflective self-governance, understood in Frankfurt's terms as the ability to determine which of one's desires motivates action (at least in the sense of an attempt to act--to allow for Frankfurt's counterexamples to the principle of alternate possibilities).

The second factor might be viewed as the potential for free will, but in combination the two factors amount to something actual in the cases under consideration: a failure by the agent in a particular case to exercise self-control; a free (though maybe not free-willed) omission. (9) I think this is important because Frankfurt's counterexamples to the principle of alternate possibilities turn on the agent's actual deliberative contribution to the causation of action (see p. 7). What makes the agent morally responsible for action in spite of the lack of alternatives is not just something about his abilities or character or even his reasons for action in the present case but the fact that his action was indeed motivated by those reasons.

Can something like the suggested account handle more complex cases? Let me refer to it as the "Frankfurtian" account of responsibility, by contrast to the Strawsonian account in terms of interpersonal reactive attitudes. It may not be what Frankfurt really intends, so I shall leave that as a question, rather than attempting to formulate clear-cut objections to a view clearly ascribed to him, in what follows.

The Strawsonian approach involves sorting out cases by the appropriateness of moral blame. But a two-part Frankfurtian account of the sort just sketched might attempt the same job in terms of freedom of action and will, or more fundamentally, in terms of desire--doing what one wants and having the potential for willing what one wants, or the capacity for second-order desire, as a substitute for more complex appeal to the moral sentiments. Complexity on the Frankfurtian account would be introduced by the hierarchical structure of desires rather than the variations in content and affective quality that differentiate desires and beliefs into a range of sentiments.

Let us consider a case from my days as an addicted smoker. Some of my early attempts to quit failed because I felt I had to smoke in order to make progress on a paper. First, Frankfurt's comments on self-identification allow us to distinguish at least two possibilities here: In one sort of case I find myself unable to concentrate while writing and give in to psychological pressure to smoke despite an earlier decision not to. Here I perform a free act when I smoke, though I do not act with free will. I do not have the will I want, in Frankfurt's terms, since my higher order desire to abstain from smoking loses out to the impulse to smoke in the struggle to influence my behavior.

In another sort of case, though, perhaps after one too many experiences of struggle and defeat, I determine my priorities and psychological weaknesses before starting on the paper and avoid conflict by deciding that I simply need to smoke while writing. Am I "taking responsibility" for smoking in Frankfurt's sense?

Coming to see myself as a smoker by temperament or as unable to "have just one," on the model drawn from Alcoholics Anonymous, might amount to shirking responsibility unless I retain an upper-level desire not to smoke. Otherwise, I would be resolving conflict either by becoming a wanton, without any upper-level desires on the subject, or by self-delusion, blocking out the negative aspects of my addiction (which in fact, we assume, are overriding).

If I do manage to delude myself, Frankfurt would grant that I act freely--"of my own free will"--though not that I act with free will, since my will still is not free (p. 24). (10) I am not able to have whatever will I want, that is--on the matter at hand, smoking--though in this case I am content with the will I happen to have. Essentially, this is a variant of the case of the willing addict--which Frankfurt does count as unfree, remember--but with willingness arranged for the occasion. Frankfurt suggests that such a case would fulfill the requirements for moral responsibility.

Presumably what I should do, rather than simply accept myself as a smoker, is make a decision to postpone my attempt to give up the habit until my other priorities permit. There would be opportunities for self-delusion here too, of course. But by "taking" responsibility in this or some related sense, would I become responsible, as Frankfurt's comments on his own cases of taking responsibility suggest?

5. Some recent authors maintain that in general Frankfurt means to be discussing a notion of responsibility distinct from the one linked to blameworthiness: autonomy rather than moral responsibility. (11) Indeed, in his preface (p. vii) Frankfurt disclaims primary interest in moral responsibility. However, he seems to mean by this only that morality is not the main or only source of the ideals he has in mind in discussing responsibility for action on our ideals.

In the sort of case I am now considering, in fact, we typically have a clash between nonmoral ideals: health versus professional productivity or the like. But moral responsibility in the sense linked to blameworthiness does seem to extend to such cases, to the extent that they involve harmful consequences or less than admirable character traits. We could put together a case of clearly moral significance, for that matter, if we assumed that part of what was in question was a risk of birth defects to a developing fetus as a result of smoking.

Many of Frankfurt's comments on cases, moreover, are explicitly (if sometimes tentatively) addressed to questions of moral responsibility, including his references to becoming responsible by "taking" responsibility. However, perhaps his remarks here might be better redirected toward another question. In the case where I make a deliberate choice to keep smoking in order to complete a paper, I take control of my behavior and in that sense act more autonomously than I would by giving in to the impulse to smoke in the middle of writing. But in both cases I would seem to be responsible for smoking and at least potentially blameworthy--for damage to a fetus, say, or if I had a heart condition that I knew could be worsened even by short-term smoking.

The object of responsibility here is presumably either a particular act of smoking or a succession of such acts, but I think we would say something similar about acquisition and maintenance of the habit, at least assuming it took place in adulthood. In cases of addiction that might turn out to have a genetic basis varying from person to person, such as alcoholism on some accounts, the impulse or tendency in question might not have been so clearly foreseeable as a consequence of the agent's earlier choices, but the point is just that the agent's level of responsibility would seem to be unaffected by coming to identify with his condition.

6. What seems to shift in the different cases of smoking in order to write a paper is the more precise locus of blame or excuses, with consequences for how likely I am to make excuses for myself. In the case where I make a deliberate choice to smoke despite the risks to myself or others I might be held blameworthy for my decision and my priorities. On the other hand, contrast the case where I allow myself to be overtaken by conflict--as we might say once I have had the experience enough times to predict it. That is, I have repeatedly resolved to give up smoking and then yielded to the urge to smoke when the pressure to concentrate on writing became too intense. I am presumably blameworthy for the later act, the act of yielding, whether or not I acknowledge it clearly then or thereafter as a moral failure--and whether or not it was completely within my control.

The case in some ways resembles one in which I am responsible for crashing my car while intoxicated, assuming I was sober enough earlier to anticipate the risk and postpone driving (or earlier still, to limit my drinking in light of it). However, I do have enough control over smoking for the act to count as voluntary even under pressure. What happens under pressure, let us say, is that I tend to minimize the reasons against smoking, including the difficulty of stopping again as soon as the paper is done--a variant of self-delusion that is likely to be shorter-term than simply accepting myself as a smoker. This is the sort of case that suggests that we might have moral responsibility without free will as applied to a given act.

Part of what makes me morally responsible for smoking even here, I should think, is the capacity for reflective self-control--not just as a general feature of persons but more specifically as something I might have exercised with respect to the act in question. (12) Actually exercising the capacity involves becoming responsible only in a "forward-looking" sense: assuming the task of managing my behavior by long-range planning in light of my addiction. Acknowledging my addiction in a clear-headed way can be part of this, though I am not sure it has to involve (or ideally involves) identifying myself with the impulse to smoke. It might be best for me to continue to see the impulse as external to myself, the product of an acquired habit, and hence as something I can shed without fundamental self-transformation.

In any case, self-identification would seem to be a step toward free will--as opposed to moral responsibility in the "backward-looking" sense at issue in judgments of blameworthiness. In terms that are central to Frankfurt's understanding of his account: Self-identification makes me "active" rather than "passive" with respect to my lower-level desires, essentially by removing my resistance to them at the upper level.

7. Let us continue to explore the case of conflict and unfree will: I am still morally responsible here for yielding to the impulse to smoke. It is not necessary to "take" responsibility in order to have it in that "backward-looking" sense. On the Frankfurtian account I have suggested, moral responsibility implies the capacity for free will, though it does not imply free will; it also implies free action with respect to whatever act or pattern of action is in question. In this case we can focus on a particular act: resorting to a cigarette in order to concentrate on writing.

In the minimal sense implied by voluntariness this is an act I do because I want to. On the other hand, the act seems to be something I do only because my options are limited--limited beyond what is normal, by the physiological and psychological effects of addiction. More precisely (to spell out an assumption hidden by "because"), I see my options as limited. I consider myself unable to resist the urge to smoke without distraction, too much distraction to write a well-focused paper in a reasonable length of time. So essentially, a combination of actions--writing successfully while abstaining--seems to be closed off to me.

Frankfurt's counterexamples to the principle of alternate possibilities do not really rule out this sort of appeal to alternate possibilities: options as perceived by the agent. My apparent inability to write without a cigarette is part of the explanation here of why I smoke. The case assumes that it is in one sense the reason why I "want" to smoke. One might think of cases of this sort as involving a kind of internal or psychological coercion. However, Frankfurt limits that term (implausibly, I think) to genuinely irresistible motivation; the term he would use is "duress" (p. 37; cf. p. 49).

The essential point is not that duress in this sense of psychological pressure necessarily makes smoking unfree, but just that it makes it less free--the act, that is; not just the person or her will. (13) It still may exceed a threshhold for freedom set by moral or other variable circumstantial considerations. If more were at stake than a risk to one's own health, say--something on the order of harm to a fetus--the difficulty of resisting the impulse to smoke, or the other things I would have to forgo if I did resist, might not count for much.

On the other hand, a more debilitating addiction might be enough to make an act unfree. Remember that I am treating the question of responsibility as distinct from that of freedom--meaning (for me, as opposed to Frankfurt) freedom of action as well as will. Degrees of freedom and responsibility presumably co-vary, but the threshholds for denying the two notions might be set differently. I shall not resolve those questions of detail but just want to extract my main differences from the Frankfurtian approach in what follows.

It seems to me that the addicted smoker has less than full freedom of action as well as of will, though she still may be morally responsible for her action and its consequences. The smoker's ability to do otherwise, not just to want otherwise, is limited beyond what is normal insofar as the physiological and psychological costs of failing to satisfy the addiction interfere with doing otherwise. Whether she is responsible in a moral sense for not doing otherwise will depend on how much could reasonably be expected of her, given the moral stakes.

I think the Frankfurtian view actually builds a kind of freedom of action into the notion of freedom of the will, despite Frankfurt's early distinction between the two notions, by building reference to action into "will" in the way I have indicated. Something is one's will--a desire amounts to a "volition" in his terms--only if it issues in action, at any rate some sort of attempted action. But the result is that there are two distinguishable ways in which the will may be unfree on his acount--because the agent cannot determine his desires, and because he cannot determine which of them issues in action--one of which intuitively seems to be a kind of freedom of action.

To see this in the present context, we might ask whether there is a sense in which someone suffering from a more debilitating addiction would be less free than a very addicted smoker. The two might be equally addicted insofar as they both act on impulses that they do not want but cannot manage to modify. However, does this also mean that they are equally unable to resist acting on the impulses?

With the lines drawn in the way Frankfurt proposes I do not think we can accommodate all we might like to say about these cases. A particular addict--possibly a smoker, though let us say an alcoholic--might not be so unresponsive to reasons that satisfying his addiction fails to count as a voluntary action, and hence a free action in Frankfurt's minimal sense; yet he might be psychologically incapable of getting through the day without the drug or of giving up the habit on his own. (14)

I, on the other hand, was eventually able to give up smoking, some years ago--while writing, in fact, though without my accustomed sharpness of focus. However, if I understand Frankfurt's view, my will might have been no less free than the addicted alcoholic's while still in the grips of the addiction. Both the alcholic and I while still smoking might have wished, equally fervently, to be freed of our respective impulses. However, neither of us acted on these second order desires during the time in question. So neither of us had the will he or she wanted. Any desires that did not lead to action would not count as our "will."

The difference between us surely had something to do with our ability to resist the impulses, or the differing degrees to which our second order desires (and therefore we ourselves) influenced action. Both long-range and more immediate control are relevant, moreover. I would say that I was able to resist smoking in a particular case even while still addicted--at a real cost, but a cost that might be bearable for certain purposes: to avoid serious health risks, or just to avoid giving annoyance to various people bothered by smoke in my environment. But my ability was impaired by my addiction.

8. I am not quite sure how Frankfurt's view would accommodate degrees, but I shall leave that as a question here. (15) Let me turn instead to some more fundamental worries, about the nature and role of self-reflection. The result of full freedom in some of the cases I have discussed here, and in the cases involving control of aggression that I also have in mind, seems intuitively to be "spontaneous" action, of a sort that gets beyond the need for self-reflection in the sense that involves conscious planning--beyond the stage where one has to ban cigarettes from the house or postpone writing deadlines to avoid smoking. The impulse eventually goes away, or at any rate one learns how to "let go" of it when it arises (as my old impulse to smoke occasionally does when I focus on it in writing this paper).

This is the stage that I think is missing in the current picture of aggressively impulsivity (or for that matter, depression--another case that is thought to involve serotonin deficit). Managing himself by advance reflection may still be possible for the impulsive agent--and if so, and the stakes are high enough, he is morally responsible--but he lacks a form or degree of freedom that we value for its own sake, independently of its cognitive presuppositions or its role in our moral practices: freedom as directly effective motivational self-influence.

What is in question here can be understood as a variant of self-identification: seeing oneself, not as puppetmaster with respect to oneself, a remote self-manipulator, but as the immediate source of action on a particular occasion of choice. It is this idea of freedom that I would want to detach from responsibility. But as I have suggested, it is a notion that does seem to be naturally understood in terms of alternate possibilities or options--in a form that Frankfurt's negative argument allows.

Many authors apparently take Frankfurt's counterexamples to the principle of alternate possibilities as having shown, or at any rate as purporting to show, that options are beside the point in this area. Frankfurt himself in his next paper shifts to his positive account of free will in terms of the hierarchy of desire. But his cases against the PAP turn on something that might be explored instead: the deliberative agent's conception of himself as lacking options, whether or not he really does. Why not take this, or moral sentiments based on it, as a candidate for explaining unfreedom? Its reference to what the agent thinks may make it seem superficial, but it is real enough in motivational terms. Where psychological options are in question, moreover, it seems to be self-fulfilling.

Frankfurt himself formulates a similar modified version of the PAP toward the end of the paper in which he refutes the standard version: "[A] person is not morally responsible for what he has done if he did it only because he could not have done otherwise" (p. 10). But his main concern is to show that in this form the principle would not pose a problem for determinism. In any case, it is stated in terms of responsibility rather than free will. Having already laid down a distinction between the two notions, Frankfurt treats free will in a separate paper, setting it up in quite distinct terms and dropping any explicit reference to the modified principle.

In the terms I prefer, to sum them up briefly, personal or motivational freedom (in a sense that applies to action as well as will) is undercut when an agent suffers shortages in the normal options for response--as in a case of addiction, where the agent anticipates a physiological price for abstaining. (16) He is unfree if the price is too much to expect him to bear for whatever purpose is in question.

This way of picking out what is relevant to addiction and similar cases is not meant to provide a full account; for that I think one would need to go beyond Frankfurt's essentially Cartesian emphasis on how we construct our individual selves and talk in more Strawsonian/Humean terms about constructing the notion of character as an object of social emotions. But in any case, the reference here to a "normal" range of options renders irrelevant traditional worries about free will versus universal determinism.

Although freedom still comes out as relevant to moral responsibility, moreover, it is not required by it--since there may be indirect modes of self-management that an agent can resort to besides simply abstaining from action at a given time. A smoker who wants to abstain from smoking, say, can avoid taking on demanding writing projects, banish cigarettes from the house, and so forth. Both questions, of responsibility and motivational freedom, depend (but in a variable mix) on matters of degree: what is at stake and at how much cost to the agent.

9. What Frankfurt means by self-reflection, of course, is not a mental activity on the order of self-management but something more naturally taken as dispositional. A second-order desire is a kind of self-evaluation--assessing the sort of person one is in light of one's other desires and their influence on action--though it functions as a cause of action by way of a second-order volition, an event. But when Frankfurt handles questions about particular cases--whether a willing addict is free or not, say--he has to cash out his talk of desires in terms of counterfactuals. Did the willing addict smoke (or whatever) because of her self-reflective desire, her endorsement of that aspect of her motivation; or would she have done so in any case, because of her addicted impulse? If one asks what these causal claims and their component terms come down to, I think one will soon get to claims of the sort implied by the modified PAP.

The apparent simplicity of Frankfurt's hierarchy of desires is quickly qualified, in this way and others. Though he represents his talk of "the will" as intended to correct an overemphasis on intellect in philosophy, his understanding of the higher level of desire in terms of self-evaluation seems really to blur the two categories. (17) Perhaps that is all to the good, but one is left wondering whether the lower level is not given short shrift. Where the impulse to smoke is my reason for smoking, I need not always be blindly driven toward action. Nor am I simply caused to act by a brute feeling. I have reasons for my reasons, or at any rate for acting on them, besides those determined by my self-ideal. If I look into them, they have something to do with my alternatives. I may not often look into them in acting; but that is true of the second order too.

Another question is whether there may not be some standards of importance applicable to the upper level, apart from an agent's decisions about the self she wants to be. Consider the following: In the middle of writing this paper I spent some time with smokers at a conference and ended up seriously considering the possibility that I still think more clearly with a bit of nicotine--though at this point I am firmly off the habit of taking it in first-hand. Suppose I decided that I would do a much better job in discussion of a paper (possibly also in writing this one) if I allowed myself to relapse. There is only so well I can do without smoking; my options are limited--if not beyond what is normal (though that is at least a hypothetical case one ought to consider), then beyond what I once came to take as the norm for my own performance. If I have by this point adjusted my priorities or self-conception so that I no longer care enough to start smoking, is that just a matter of my personal choice of ideals? There are those who would heatedly deny this.

To conclude, then, Frankfurt's approach promises a shift from intellect to will, but will is conceived so abstractly, in terms of desire, that I am not sure how he means this to make a difference. The rough idea appeals, though I would want to substitute "motivation" as suggesting a more concrete locus of freedom or unfreedom--and one with more obvious bearing on action--in the sense that needs to be disentangled from responsibility. The distinction between free will and responsibility in Frankfurt's own work seems really to function as a barrier to more basic reconfiguration of the categories in this area to the extent that it assigns the two notions to separate treatments.

1. This paper was published in revised form in The Journal of Ethics, 3 (1999), 325-40. Let me thank David DeGrazia for comments on an earlier version.

2. Cf. R. Jay Wallace, Responsibility and the Moral Sentiments (Cambridge, Mass.: Harvard University Press, 1994), esp. ch. 6, for a concept of reflective (or sometimes, normative) self-control (or self-governance, competence, or the like), explained in terms of the agent's capacity for removed reflection on his desires in light of moral reasons. Cf. also Paul Russell, Freedom and Moral Sentiment (New York: Oxford, 1995), esp. pp. 91-93; cf. p. 180) for a Humean version of the same sort of account.

3. See Harry G. Frankfurt, The Importance of What We Care About (Cambridge: Cambridge University Press, 1988), pp. 23-24; I shall use page-references in my text in what follows to refer to this volume of Frankfurt's collected works.

4. See my "Free Will and the Genome Project," Philosophy and Public Affairs, 22 (1993), 31-43, and "Genes, Electrotransmitters, and Free Will," in David Wasserman and Robert Wachbroit (eds.), Genetics and Criminal Behavior: Methods, Meanings, and Morals (New York: Cambridge University Press, forthcoming); cf. also "Humean Sentimentalist Free Will" (tentative title; unpublished). An earlier approach to similar cases but without specifically genetic explanation appears in "Unfreedom and Responsibility," pp. 63-80, in F. Schoeman (ed.), Responsibility, Character, and the Emotions (New York: Cambridge University Press, 1987).

5. I should note that I am reading Frankfurt as clarifying and defending essentially the same view throughout his collected papers. Some authors take him as changing his view significantly between his earlier and later papers; see esp. John Martin Fischer and Mark Ravizza, "Introduction," pp. 1-41, in Perspectives on Moral Responsibility (Ithaca, N.Y.: Cornell University Press, 1993), pp. 25-26--but

6. Cf. esp. pp. 14-15, where Frankfurt discusses a stronger case, in which the agent actually wants (rather than not caring about) the lower-level desire. In the case I have in mind, presumably Frankfurt would add some conflicting lower-level desires (the "wants" motivating a voluntary action) to those that move me to smoke.

7. Perhaps Frankfurt's view on this matter is misread as a result of confusion with Gerald Dworkin, esp. The Theory and Practice of Autonomy (Cambridge: Cambridge University Press, 1988), in which a view similar in overall structure is applied to some of the same sorts of cases.

8. Cf. Wallace, esp. ch. 6, for what amounts to an argument that the reference to general capacities does not beg the question of the principle of alternate possibilities.

9. This is in contrast to Wallace, esp. ch. 5, which apparently brackets all questions of freedom (though initially on the grounds that free will in a "strong" sense is contra-causal; see, e.g. pp. 2-16) and instead focuses directly in other terms on explaining excuses from moral responsibility.

10. Let me thank Richard Hanley for directing my attention to this distinction.

11. See esp. Wallace, pp. 52-53. Cf. David DeGrazia, "Autonomous Action and Autonomy-Subverting Psychiatric Conditions," The Journal of Medicine and Philosophy (1994), 279-297, which more closely follows the account defended by Dworkin.

12. Cf. Wallace, ch. 6, for what seems to be a weaker view. Simply having the general capacity would distinguish me from Rambo as a "responsible agent," but here what is in question is responsibility for a particular act.

13. See my argument in "Behavior Control and Freedom of Action," Philosophical Review (1978), 87 (1978), 225-40; reprinted in J. M. Fischer (ed.), Moral Responsibility (Ithaca, N.Y.: Cornell University Press, 1986), pp. 191-204. Note that Frankfurt's psychological sense of "duress" is not the sense in the law, where what distinguishes duress from coercion is just whether the threat in question is imposed by another agent.

14. Cf. the recent film, "Leaving Las Vegas."

15. Wright Neely offers an account of degrees of freedom in terms of priority of desire in "Freedom and Desire," Philosophical Review, 83 (1974), 32-54, esp. pp. 48ff. However, Neely's account is unsatisfying for my purposes, since like Frankfurt's it is framed in terms of desire. It appeals, I think implausibly, to irresistible desires to make sense of the cases that concern me (see p. 46; cf. the treatment of psychopaths in these terms on p. 47). Neely also takes his notion of freedom not to be the one that is relevant to moral responsibility (p. 50).

16. I should note that "anticipates," as I mean it, covers anticipation in feeling, of a sort that seems to mark off motivationally unfreeing addictions--those that involve an uncomfortable craving or other perceived need for the item in question--from those that are merely physiological.

17. On standard desire/belief accounts, an evaluative belief would indeed be classified as desire. It is not ruled defective for failing to fit the world; the world is supposed to fit it. But presumably these norms of "fit" are set by the agent. There is another level of self-assessment here, which might be made out a basic evaluative stance toward the elements of one's But in what sense is self-critical thinking on this level other than intellectual?