GENES, ELECTROTRANSMITTERS, AND FREE WILL¹

There seems to be evidence of a genetic component in criminal behavior. It is widely agreed not to be "deterministic"--by which discussions outside philosophy seem to mean that by itself it is not sufficient to determine behavior. Environmental factors make a decisive difference--for that matter, there are nongenetic biological factors--in whether and how genetic endowment manifests itself phenotypically.² Moreover, even if its manifestation were inevitable, its bearing on criminal behavior apparently turns on general personality traits on the order of "impulsivity" that under different conditions of life could take very different forms. My concern in this paper is not with "genetic determinism"; it should be obvious that the notion makes sense only if determinism admits of more limited (and less worrisome) forms than the one involving universal causal necessitation that philosophers have in mind by the term. Instead I want to ask whether current accounts of the link between genes and criminal behavior would manage to undermine free will anyway, even apart from worries about determinism.³

I shall eventually suggest that some of the other terms in this area should also be pried apart. In particular, the genetic accounts in question may well be thought to undermine freedom and yet to allow for responsibility in the sense of blameworthiness. Omitting reference to issues raised by the philosophic literature (but inserting some parenthetical cautions against misinterpretation), my discussion can be summed up in capsule form as follows:

The implications of genetic research for issues of responsibility depend on what sorts of causes of behavior it yields evidence for (assuming it yields evidence for any). One distinction commonly made is that between causes of normal agency and "interfering" causes. A simple sort of psychological model in terms of interfering causes is provided by the usual construal of kleptomania and similar (putative) cases as involving "irresistible impulse." I make these out as cases of psychological compulsion insofar as they involve a kind of internal interference: a threat of continuing mental disturbance that is sufficient to interfere with the ability to do otherwise.

However, there is an alternative picture guiding current research on criminal behavior (among other things): of inadequate resources of behavioral control. This substitutes for interfering causes an absence of "enabling" causes of normal control such as adequate supply of serotonin and other electrotransmitters. The effects of serotonin shortage on criminal behavior according to the current accounts would seem to involve a kind of localized learning disability. But even if this undermines freedom, the relation of the incapacity in question to norms of character may still allow for an element of moral responsibility. Blame as a reactive attitude may still be warranted (even if it also ought to be offset by compassion) toward an agent whose unresponsiveness to social learning manifests itself in patterns of voluntary harm to others.

My argument divides into three main parts: In Section I, I discuss standard philosophic approaches to free will issues, including some work of my own. In II, I attempt to make a different kind of sense of current genetic (and other biochemical) accounts of "aggressive impulsivity" as problems for freedom. Finally, in III, I indicate how we might defend an element of responsibility in the cases in question even without free will.

According to most philosophers causal determination of behavior or behavioral traits does not pose any particular problem for freedom as long as it works within normal motivational channels of the sort involved in moral agency. This is the dominant view of free will in modern philosophy, known as "compatibilism" (since it understands free will as compatible with determinism), or sometimes "soft determinism" (since it effectively cushions our ordinary notions of moral responsibility against the effects of determinism). As a general approach to free will issues the view rests on a distinction between two possible ways of influencing an agent's behavior: First, one that allows the agent the usual causal role in generating behavior--via deliberation, choice, and other exercises of what we call "the will," even if these are ultimately traceable to external factors--and secondly, another mode of influence that essentially "shortcuts" the will, bypasses normal agency, by linking behavior more directly to external causes.

An easy illustration of this distinction might be provided by contrasting, on the one hand, a typical sort of physicalist account of action in terms of desires, taken as equivalent to brain states and traced back to physiological and environmental causes, and on the other hand, action produced by experimental intervention in the normal course of desire formation, say by electrical stimulation of the brain. On a compatibilist account only the latter sort of influence would undermine the agent's responsibility--by attributing responsibility to something else instead of the agent, a cause that interferes with normal agency; not just by explaining normal agency in terms of something further that allows it the same pivotal role in the chain of events.

Whether or not the compatibilist account yields a plausible approach to free will issues generally, the intuitive distinction it rests on does seem to ease worries about genetic explanation in particular. Current advances in genetic research--the prospect of coming up with more and more genetic explanations of human malfunction, possibly including behavioral malfunction--are treated in the popular and scientific press as if something special were at issue beyond the usual worries about any sort of scientific explanation. But it is unclear why genetic causes would be more of a problem in principle than any other causes we might discover-- social and environmental causes for criminal behavior, say--unless they in some way "shortcut" normal agency in the manner of the brain experimenter. If they leave our usual picture of moral decision-making intact but simply enlarge it with an account of the sources of some particular patterns of decision-making--behavioral traits like those we sum up as individual moral character--then there are two possibilities: Either compatibilism is true, and our normal responsibility ascriptions still apply. Or compatibilism is false, and causal explanation would undermine our normal ascriptions; but the same would also have to be said for nongenetic causal explanation--in terms of social patterns, events in the environment, and similar factors. So news of advances in genetic explanation is at any rate no worse news for free will than potential advances in the other forms of scientific explanation that it replaces or supplements.

Perhaps it is worse news in something like a "public relations" sense: It looks worse, just to the extent that genetic explanation combines so simply and readily with nonscientific views of personality and personal causal influence of the sort summed up by the notion of moral character. A common philosophical alternative to determinism, a form of incompatibilism known as agent- causation, understands "the agent" as a distinct type of cause, to be distinguished from putative "event-causes" of action.⁴ But genetic explanation of personality traits would seem to plug the agent--or his nature as an agent, his moral personality or character--right into a deterministic chain of events, by way of a causal account of personality formation. The appeal to genetic causes of character may look worse for free will, then, to the extent that it s capable of co-opting --taking over and using for its own purposes--the explanatory model assumed by a standard incompatibilist defense of free will. However, there also are more selective ways of understanding free will--without treating our ordinary talk of agency and decision-making as unanalyzable into simpler causal terms, so that at most only formative causes can get a grip.

What I have in mind as a more selective account would be one that allows for our intuitive distinction between acts that are within an agent's control and others that may not be--on the model of the distinction I outlined earlier between a normal act and one resulting from electrical brain stimulation by an experimenter. The distinction depends on the particular causal mechanism appealed to in explaining action rather than on causal explanation per se. The case of experimental brain stimulation is a relatively easy one, because it picks out another agent, the experimenter, as a cause acting earlier that can serve as an alternative locus of responsibility. But there are more problematic cases, where the causal mechanism explaining agency seems to shortcut normal self- control in favor of something impersonal--on the order of a gene.

However, on a more selective account free will depends on how a gene (or other putative cause beyond the agent) gives rise to its behavioral effect. If it works through normal channels, as it were--allows for normal exercise of choice or will in the agent--then genetic causation apparently falls on the unproblematic side of our intuitive distinction and leaves free will intact. It is in this area, I want to say, that genetic explanation indeed may pose a threat to our ordinary view of freedom, particularly as illustrated by current research on criminal behavior. But the threat is not genetic causation simpliciter but rather the combination of that (or some other) appeal to impersonal causes of character or temperament with a particular understanding of the causal mechanism as one that essentially takes self-control out of an agent's hands.

Compatibilist and incompatibilist defenses of free will agree in locating the causes of free action within the agent in some sense, in contrast to causes operating on the agent from without. The distinction follows Aristotle's treatment of compulsion as a species of involuntary action--though Aristotle was not concerned particularly with the possibility of causal chains leading back beyond the present situation or with psychological variants of compulsion.⁵ The more problematic cases arise when we allow for internal (in the sense of "mental") causes that still might seem to be external to what philosophers think of as "the agent," or the bearer of moral responsibility: the agent's core self, or "will" as an active principle, distinct from at least some of the various mental states an agent might exhibit. There are impulses, say, on the order of the kleptomaniac's urge to steal (to take a familiar sort of problematic case), that might be seen as "coming over" an agent, characterizing him only in some passive sense--even if they do so regularly enough to be said to be "in character."

An account along Aristotelian lines might treat some such cases as voluntary, to the extent that the will may be seen as actively yielding to impulse, on the model of a robbery victim coerced to surrender his money. He does so because he wants to (and reasonably so) under the circumstances--circumstances of external threat--though he does not want to be in those circumstances in the first place or to have to make the choice he does. I have proposed a different application of the model to psychological compulsion, understood as involving internal constraint, presumably by some sort of threat of emotional discomfort.⁶ The thought is that the dividing line between free and unfree action is a matter of degree: whether a threat is disturbing enough to interfere with the ability to do otherwise.

The model may initially seem strained in application to kleptomania--assuming there really is such a thing, so that we do want to say that the agent's action in such cases is unfree. The only threat a kleptomaniac faces if he resists the urge to steal would seem to be the continued agitation of that unsatisfied impulse. Of course frustration is unsettling; but is it disturbing enough in these (putative) cases to be said to make the agent unable to refrain? Does the kleptomaniac really act on an "irresistible" impulse, in short--an impulse that, if not strictly impossible to resist, is at any rate hard enough, in light of the threat, that resistance is unreasonable for us to expect of the agent under those circumstances--on the model of a spy who gives up information under torture? He may have overriding reason to resist, that is, but on this account he is overcome by a threat of continuing distress that undermines his ability to act on the balance of reasons.

The account does not really turn, then, on how bad a threat is (or even how bad it seems in prospect) but rather on how upsetting it is, how much it tends to disrupt deliberation, or more specifically self-control, the ability to resist acting on an impulse. So the model of psychological compulsion, or internal interference, might be extended to fit kleptomania and like cases-- along with more usual instances of cravings and addictions that similarly might not involve intense discomfort. But there is another possible approach, distinct from psychological compulsion, that is exemplified by the current research on criminal behavior. Instead of looking at the element of internal threat--the urges that arguably overcome the will to resist and hence make action less than fully voluntary--we might shift our explanatory focus to the will itself, the psychological resources that enable an agent to resist: what is commonly called "strength of will." It is also commonly thought of as something that is up to us, but that is what a genetic basis seems to call into question.

To sum things up very simply: The current research attempts to locate causes of criminal behavior in a genetic abnormality in the supply of electrotransmitters and other biochemical factors that regulate self-control. The usual appeal is to serotonin shortage-- with "impulsivity" as the associated trait, manifesting itself in aggressive behavior where other factors are in play, but also at issue in other forms of behavioral or mood malfunction such as depressive tendencies.⁷ In the case of criminal behavior, then, the explanation apparently offers impersonal causes (genes and their biochemical effects) for processes that limit the ability to exert behavioral control.

This is an alternative to representing self-control as overcome by some sort of internal threat. The agent lacks the resources to withstand certain normal kinds of mental upset--the frustration of an impulse or other desire--rather than facing an abnormal kind or degree of upset, as suggested by the phrase "irresistible impulse" in the kleptomania example. So an impulsive agent's will is undermined from inside, as it were, by the absence of an enabling factor presupposed by normal agency, rather than being defeated or counteracted by some sort of powerful interfering factor (even one that is internal to the agent in some sense) on the order of an impulse. His will or his character is defective, in the sense of just not being up to the task imposed on all of us, the task of normal behavioral self-control. But if the defect is inherited, one might ask, who is to blame?

The point is not that genetic explanation of criminal behavior is itself problematic for free will but rather that it may pose problems in combination with a shift away from our intuitive view of the causal mechanism involved in failures of self-control. One might say that the current research model takes the "mania" out of kleptomania--the suggestion of "irresistible impulse" as a stirred up counterforce to self-control, something on the order of a demon fighting oneself--and substitutes personal inadequacy, a shortage of the means of self-control, with an impersonal cause. The cause on this account, rather than "shortcutting" the will, shortchanges it: It denies the agent the very stuff of self-directed agency--on the model of a spy under torture who just becomes too exhausted to resist after a long period of sleep- deprivation. The cause of uncontrol here is something more like depletion than disturbance or discomfort.

I do not mean to suggest that such a simple account of the postulated causal mechanism is adequate to capture it. Nor am I in a position to assess it fully. It may or may not replace our intuitive view of these cases. But for purposes of argument I want to keep it in simple terms and just suppose that it is essentially on target. It does seem to have had some successes, as indicated by practical criteria, most notably the effects of medication such as Prozac in controlling depression and related forms of impulsivity.⁸ I think we can see that, if this should turn out indeed to be the appropriate model for impulsive (or even just some impulsive) criminal behavior, it would essentially set up the relevant trait as an inherited inability. By virtue of the genetic bases of personality, according to the model, certain agents simply lack the biochemical equipment for normal self-control.

The "impulsive" criminal has the same impulses as the rest of us, at least initially, on this account; he simply lacks our capacity to suppress them, or to let them pass unsatisfied. He is no more disturbed at suppressing them but just cannot manage it, or cannot manage to learn to do it, because he lacks the requisite supply of serotonin. What serotonin does is to facilitate transmission of impulses (in the electrical sense: nerve impulses) across synapses. This might be seen as a graphic depiction in neural terms of failure to "let go." It amounts to a defect of thought-processing. In application to criminal behavior, according to the current accounts, it apparently involves an inability to learn from punishment.⁹ But this sort of failure is no more within control, assuming it is genetic, than other forms of learning disability.

To take a case we can understand from the inside, consider the effects of memory loss due to age. Once I have lost a certain number of brain cells, let us suppose, I no longer have the resources (on a comparably primitive picture of mental storage) to memorize this paper for oral presentation--or rather, perhaps, to do so without undue distraction, in the way required by a reasonable performance. So there is a clear sense in which I cannot manage the task: I cannot manage it easily enough to make it reasonable to expect of me, even on the assumption that it is possible, strictly speaking. We might imagine a situation where I am asked to give a paper extemporaneously. If my brain cells really are just not up to it at this point, then the demand is unreasonable.

This is compatible with the acknowledgment that there may be various indirect or extraordinary measures, beyond the normal means of deliberative self-control, that I might take, or might have taken, to build up my mental storage capacity. Similarly, one might say, for someone with an inherited tendency to impulsivity who commits a crime because he does not have the thought-processing capacity needed for normal resistance to impulse. His will is impaired, or the cognitive underpinnings of will or decision-making are impaired, just as we are supposing my memory is. We both fall short of a general human norm, though my deficiency counts as a normal effect of age, a normal process, whose unwelcome effects we might like to change but have more or less learned to live with, whereas his is traced back to a genetic abnormality, with behavioral results we cordon off as unacceptable because of their moral significance.

Besides this reference to norms there are other likely differences between the two cases. I shall bring out what I take to be the crucial difference in my next section. But first, one might want to say that the cases differ just insofar as memory defects tend to frustrate an agent's ability to act in his own interests, so that they mainly are unwelcome to us. However, this may be true of the aggressively impulsive agent too. He should not be pictured as a psychopath, that is; the usual category is a broader one, "anti-social personality," that allows for guilt or other negative reactions to his deeds, based on empathy with their victims--at any rate, after-the-fact, when the agent has managed to reflect on them. If aggression is something he just cannot help, though, in the sense I have tried to capture, I think we would count his behavior as unfree, even if he engages in it without regret or other motivational conflict.¹⁰

Another thing to consider is the role of differences in social placement of the agents in the two cases, as exemplified by the likelihood of my knowing about various self-training practices and memory aids. This means that short-range unfreedom in a case like mine--the genetically based inability to do otherwise, at least in immediate terms, as things stand--does not imply that my situation cannot be changed by my own efforts. By contrast, the impulsively aggressive agent is likely to be someone who by virtue of upbringing and current situation just would not know how to change his habits of thought-processing, or even that he has reason to do just that. Nor would he find out soon enough and with enough external resources to manage the requisite behavioral change.

If society does know these things and does have the relevant resources and is partly responsible for the situation where the agent does not, it might be said to bear responsibility for such cases--contrary to the moral often drawn from popularized genetic accounts, which tend to be politically conservative. However, to say that society is responsible need not be to deny all moral responsibility to the agent, even supposing that an agent is not free if his action results from an inherited deficit of the electrotransmitters needed for normal self-control. The notions of freedom and responsibility, along with other elements of the debate on these issues, need to be pulled apart to yield an expanded conception of the alternatives (and with it, a less polarized picture of the political options) than the debate usually assumes

Even if we make all the assumptions I have allowed so far about the success and free-will implications of current causal models of criminal behavior, it is possible to detach the question of freedom from the backward-looking or "reactive" component of responsibility.¹¹ The will may be subject to a disability--internal limitation as opposed to interference--but some form of blame may still be warranted to the extent that the act in question does depend on the agent s will and hence counts as voluntary.¹²

In Aristotelian terms, the immediate cause of action in these cases is within the agent, though it undermines his ability to choose, in the sense that implies deliberation.¹³ I would call these cases of weak character, with impulsivity seen as a relatively stable trait of an agent--a trait of temperament liable to moral blame; in philosophers' terms it yields weakness of will--whether or not it is ultimately caused by some shortage in the prerequisites of normal agency.¹⁴ Unlike some contemporary proponents of virtue ethics, Aristotle of course has no intention of banishing blame from ethical discourse; he distinguishes virtue and vice, as states of character, from passions or feelings by the fact that we do consider them fit subjects of praise and blame.¹⁵ What he has in mind by blame is itself something other than a feeling; an alternative translation is "censure." But it is worth noting that Aristotle s definition of anger in the Rhetoric- -in terms of pain at an unjustified slight to ourselves or our friends--would allow for a corresponding moral sentiment in cases like those we are concerned with, supposing that the aggressively impulsive agent, with whatever incapacities we have allowed him, is at any rate capable of insult.¹⁶ I think there is something right about this, as a point about emotional blame--and ultimately about warrant for a corresponding judgment of responsibility.

With aggressively impulsive behavior taken as unfree but voluntary, we can pinpoint the main way such cases differ from genetic impairment of memory--whether due to age, as in my earlier analogy, or to some sort of lifelong mental handicap. Volitional defects seem to be internal to an agent in a sense in which purely cognitive deficiencies of the usual sort are not. The distinction between them makes sense if we think, along more Humean lines, of the notion of an agent's "character" as set up to provide a proper target of emotion in its role in enforcing moral norms.¹⁷ Retributive emotions like anger and resentment are justified in general terms, counting as appropriate in the sorts of cases where they normally serve as ways of modifying behavior, such as cases of volitional failure, even if the behavior turns out not to be modifiable in the particular case at hand.¹⁸

The question of responsibility often takes the form in contemporary discussion of a question about the justification of retributive emotions because of Strawson's influential treatment of free will as a question about our "reactive attitudes": whether they (and the social practices built on them and supporting moral judgment) could survive a serious belief in determinism.¹⁹ Though my concern here is not with determinism (in the philosophers' sense of universal causal necessitation), Strawson's appeal to emotional and other reactive attitudes helps in addressing a separable worry: that at most we would be "going through the motions" of responsibility attribution if we imposed punishment just to influence behavior.

That is, though society has to protect itself from aggressive crime, typically by imposing legal penalties, whether the penalties amount to anything like justified punishment would seem to depend on what sort of attitude toward the lawbreaker is warranted.²⁰ The thought that inflicting punishment on someone would avert future harm is notoriously inadequate. Punishment is supposed to express blame. This is not meant to say that we necessarily ought to feel blame when we punish or that expressing blame is the prime purpose or justification of punishment, or even a consideration that ought to affect the penalties we inflict, but just that the penalties are supposed to be justified as expressions of blame. That would seem to be the test of whether it is punishment we are justifying, or rather something else--some use of the agent for others good or her own that is not in the same way directed toward her as a person.

Strawson thinks of blame as involving a "participant" attitude toward its object: treating the object of blame as a fellow participant in the moral institution. The alternative is the "objective" attitude, which is essentially impersonal. Though his account allows for a continuum between these two poles of reaction, thinking of some agent as abnormal is supposed at least to incline us toward emotional distance.²¹ However, I think that a fuller picture of our emotional alternatives in such cases would still permit a form of blame for aggressively impulsive agents, as cases of volitional impairment that involve defective character.

Consider again at the contrasting case of impaired memory. On plausible versions the problem is not localized in the way that aggressive impulsivity is. We would indeed be inclined to blame someone with a comparably selective form of memory failure-- for instance, someone who could not seem to keep track of any commitments to other people while absorbed in his own pursuits. What would be subject to moral criticism in such cases is something like the agent's personal priorities--a basic fact about his character, even if not itself (or currently) open to control. Similarly, what is blameworthy about the aggressively impulsive agent is the way his impulsive behavior targets others, rather than being channeled into harmless, or at any rate self-destructive, avenues like so many of the rest of us. If he were impulsive in all ways at once, he would not represent a coherent object of blame; but the fact that his character developed in the particular direction it did requires a further explanation that may be extenuating but not exonerating, depending on details.

There are different possible ways of filling out the cases in question, that is. Some may rest on a degree of delusion--not immediately a problem of control, but rather of knowledge. In Aristotle's terms the excusing condition might not be external causation, or compulsion, but instead a form of ignorance. One possibility mentioned in the literature as linked to serotonin deficit is a pattern of attributing hostile intent to others. An agent whose aggressive behavior stems from this limited sort of delusion might not be fully responsible in the usual terms to the extent that he miscontrues the situation in which he acts--assuming that his delusion is not willfully self-induced, on the model of cases of harm done while intoxicated.

Some other possibilities that come up often in discussion seem to involve an appeal to childhood suffering on the part of the agent. If the cause of serotonin deficit were some sort of childhood trauma--as it could be in other cases, such as cases of abusive punishment during childhood--we might want to withhold a punitive reaction on the grounds that the agent has already suffered enough. But a genetically induced imperviousness to punishment may not always lead parents and other caretakers to pile on more. There may be other cases where adults just give up on punishing a child who is unresponsive to it, so that he actually manages to escape some of the normal forms of childhood emotional suffering.

More to the point would be a simpler appeal to the age of character formation. The agent's priorities presumably were set at a stage of development too early for intelligent choice.²² They might be treated as basic beliefs about what is important: what matters and at what cost; when other agents' alleged hostility deserves a violent response, say. But note that these are evaluative beliefs, encoding moral error rather than ignorance of fact. In Aristotle's framework this does allow for responsibility--at least of the sort that goes with weakness of will, where the belief encoding an agent's priorities conflicts with his "better judgment."

Sympathy may still be warranted toward such an agent, in the way that we sometimes have in mind when we attribute someone s immature adult behavior to the fact that he was "spoiled" as a child--even if blame is warranted too. We feel sorry for him in part just because he is unfree: "He can't help himself." But blame can also be seen as a legitimate response to such an agent in virtue of his character. This need not mean that we hold him responsible for his character--certainly not in a sense that entails control at some earlier stage over its formation. Rather, we hold him responsible for action--not for his character per se but for a consequence or expression of it--insofar as we attribute what he does to something basic about who he is: a self-defining pattern of response that at this point sets his priorities.

Familiar cases--in philosophy and in everyday life--indicate that we do blame people for inadequate emotional resources of the sort that affect moral agency. One of the advantages often cited of an Aristotelian or Humean over a Kantian approach to ethics is the way the former allow for our intuitive view of the moral status of character traits involving emotional response. The image of an unsympathetic person--a moral Scrooge--awakens feelings of censure, which may be softened but are not stilled by the acknowledgment that Scrooge was shaped by his background and in current terms (under more ordinary circumstances) is not open to change. The same applies to the sort of lawbreaker that is commonly called "incorrigible"--a word that in common parlance can itself be used to express blame.

As to our own responses to such a person: Note that anger is not the only available emotion that might be proposed as a form of blame. If the impulsively aggressive agent has a defective character, more passive emotions on the order of contempt--scorn, repugnance, horror, and so on; the kind of disgust that goes with "He's incorrigible!"--would provide a version of blame that is sufficiently punitive for most purposes. A feeling of contempt on the part of others (or for that matter, oneself) is aversive for most agents and hence can be used to control behavior, assuming a modicum of control over emotions themselves via acts of attention. Whether or not retributive, then, contempt is in that sense a punitive reaction. However, it would not be ruled out--if anything, it might be reinforced--by the admission that its object is unfree. He is more to be pitied on that account, perhaps, but in the sense of pity that is not at all incompatible with looking down on someone.

In any case, anger itself (including as a variant Strawson s resentment) is not incompatible with pity--or for that matter, with various more charitable emotions like compassion--even if the content of these opposing reactions might be said to pull in opposite directions, one pressing for retribution, the other for some form of aid or personal support. Ambivalence would seem to provide an answer to the worry that our usual range of moral reactive attitudes may be undermined by unfreedom.²³ Instead, the range may be expanded to include various opposing emotions. This broader perspective is not the same as an "objective" stance--even on the assumption that the conflict ultimately should settle into a more balanced attitude, of "mitigated" blame, as the proper "all-things- considered" backdrop to legal punishment, which of course has to involve a blend rather than simple juxtaposition of opposing considerations.

What emotional or "reactive" blame registers on this account is something like the salience of the agent as a personal cause of wrong, which is not erased by the acknowledgement that he was unfree to do otherwise. In terms of Aristotle's definition of anger, the agent who cannot control his aggression deserves anger to the extent that he "slights" others. If we cannot expect change of him, then instead of forgiveness in the sense that entails social acceptance or reconciliation, blame fades into a tempered attitude that resembles objectivity in its lack of personal involvement but still involves a negative moral evaluation. My conclusion is that we do have a foundation here for a form of moral blame.

I also ignore some other fine points of Aristotle's discussion, e.g., his distinction in Bk. VII, Sec. 7, between weakness and impetuosity, as a kind of incontinence involving hasty action (prior to deliberation rather than ignoring its results). The term "impulsivity" suggests this, though its application to depressive tendencies and the like argues for a broader interpretation.

15. See ibid., Bk. II, Sec. 6, 1105b28- 106a2; cf. esp. Bernard Williams, Ethics and the Limits of Philosophy (Cambridge, Mass: Harvard University Press, 1985).

16. See Rhetorica, Bk. II, ch. 2.

17. For reference to blame as a "feeling or sentiment" see Hume, Treatise of Human Nature, p. 469. I do not take emotional blame to amount to a particular emotion, as will become evident below. (Nor do I take all blame to be emotional. The most basic form would seem to be linguistic, if we judge by etymology: The term comes from a Late Latin contraction for "blaspheme.") I stress variants of anger rather than Hume's hatred in this discussion in order to support the relation to legal punishment (for acts rather than character, at any rate in the first instance) that I go on to outline. Where I may differ more seriously from Hume is in thinking of "character" as essentially constructed from our emotional responses--or perhaps better, from the attempt to achieve "reflective equilibrium" between our responses and the objects we identify as grounds for them--in a way that need not simply sum up the agent's characteristic act-tendencies.

18. For my general account of these issues see Emotions and Reasons (New York: Routledge, Chapman and Hall, 1988). Its fuller application to ethics--a realist alternative to moral "sentimentalism," with moral judgment seen as based on emotion but as having a distinct content--is given in Practical Guilt (New York: Oxford University Press, 1995).

19. See P. F. Strawson, "Freedom and Resentment," Proceedings of the British Academy, 48 (1962), pp. 1-25; reprinted in Studies in the Philosophy of Thought and Action (London: Oxford University Press, 1968).

20. Cf. Joel Feinberg's discussion in Doing and Deserving (Princeton, N.J.: Princeton University Press, 1970), pp. 95-118.

21. See ibid., esp. p. 11; cf. p. 9 and p. 19. On p. 17 Strawson briefly suggests a different line (though not quite the one I defend here) on specifically moral reactive attitudes. However, note that what he has in mind at that point are essentially forward-looking attitudes--holding an agent to moral demands--as opposed to responsibility in the backward-looking sense I attempt to isolate here (though cf. p. 19). At any rate, Strawson's concern with universal determinism makes his discussion at points inapplicable (as he frequently acknowledges) to the more complex sorts of cases at issue here.

22. Otherwise, we could appeal to a notion like Scanlon's "demonstrative" value of choice to explain why localized volitional defects should be picked out for blame. See T. M. Scanlon, Jr., "The Significance of Choice," in S. McMurrin (ed.), The Tanner Lectures on Human Values, VIII (Salt Lake City, Utah: University of Utah Press, 1988), pp. 179-80. However, the notion apparently assumes that a self or character and its contents (Scanlon speaks mainly of "preferences" here and elsewhere; cf. esp. p. 209) exists prior to choice as something to be manifested in it.

23. See my "A Case of Mixed Feelings: Ambivalence and the Logic of Emotion," in A. O. Rorty (ed.), Explaining Emotions (Berkeley: University of California Press, 1980), pp. 191-204. Cf. Gary Watson, Responsibility and the Limits of Evil, in Schoeman (ed.), pp. 256-86, esp. pp. 275- 76.