Brandeis University, Philosophy Department
Spring 2006
Brandeis University Web Stite

Philosophy 22B

Philosophy of Law

Professor Andreas Teuber
Prof. Teuber

Copyright (c) 1995 Arizona Board of Regents
Arizona Law Review

Spring, 1995

37 Ariz. L. Rev. 117

LENGTH: 8880 words


NAME: Joel Feinberg *


* Regents Professor Emeritus of Philosophy and Law, The University of Arizona, Ph.D., University of Michigan (1957). This is a slightly revised version of a paper presented at the Symposium on Issues in the Philosophy of Law at the University of Arizona, October 1, 1994. Announcements of the talk used the author's original title: The Role of Luck in Law and Life. He hopes to use that broader title for a book-length work to appear at a later date.

  ... Once one gets into the thick of it, however, and undertakes a critical examination of the arguments on both sides, one finds that many of the deepest issues of criminal law theory are implicated. ... Feelings of shame are triggered by the realization that one is not the kind of person one thought one was; feelings of guilt by acts that rupture our relationships with other persons. ... Does the retentionist really wish to supplement all of that with a simple appeal to the amount of harm actually caused? Is he willing to let the final judgment of moral blameworthiness be swayed by a luck factor reflecting nothing about the actor's character? If the incredibly complex test of blameworthiness now used pulls in the direction of extreme moral culpability would he be willing to let the fortuitous absence of a serious harm pull it back in the direction of moral innocence? If so, how far back? I find no intuitive plausibility at all in basing criminal liability on moral blameworthiness, thus securing for one's position a certain prima facie respectability, but then basing moral blameworthiness upon, as luck would have it, the actual harm or absence of harm caused. ...  


Every bona fide philosopher of law tries his hand at least once at the ancient problem n1 of punishing failed attempts. This may seem surprising, given that one's first impression of the problem is that it is an amusing riddle of no crucial theoretical importance. Once one gets into the thick of it, however, and undertakes a critical examination of the arguments on both sides, one finds that many of the deepest issues of criminal law theory are implicated.

The puzzle I have in mind is ultimately a problem for the sentencing judge or the authors of sentencing guidelines. Should there be a "gap" n2 between the penalty for a completed crime and that for an unsuccessful attempt to produce the same forbidden result, other things being equal? n3 The standard way of illustrating the problem is to describe, through the method of hypothetical  [*118]  contrast, a matching pair of examples. In the first of these we are introduced to an angry man with a gun. He bears the simple but appealing name of A[1]. In the story, A[1], with the conscious objective of killing his enemy B[1], gets B[1] in the sight of his rifle, and carefully squeezes the trigger. Except for his later arrest and conviction for murder, everything works out as A[1] planned. The bullet emerges from the barrel of his rifle at high velocity and strikes B[1]'s body in a vital place that is the exact spot at which A[1] had aimed, and as A[1] intended, B[1] dies as a result.

Now consider another narrative with only a slight variation in the facts. This involves A[2], a man of similar disposition to A[1], but who despite his similar name, is unrelated to the killer in our first example. He too has an enemy, a person named B[2], and he has precisely the same hatred for him that A[1] had for B[1] in the first story. He too sets out to kill his enemy. He has precisely the same morally relevant past experiences with B[2] that A[1] had with B[1], the same motivating beliefs and emotions, the same fully formed intention to kill another human being. I mention all of these factors, because I want it to seem plausible that A[1] and A[2] were equally blameworthy morally, and blameworthiness is normally compounded out of such factors as beliefs, emotions, motives, objectives, and intentions, in respect to all of which A[1] and A[2] are exactly equal.

But now we inject into the stories the first significant differences between them. A[2]'s enemy B[2] escapes alive, while A[1]'s enemy B[1] is killed according to plan. In different versions of this much told tale, B[2] is wearing a bullet proof vest and B[1] is not, or at just the moment A[2] starts to pull the trigger a mosquito lands on his nose, throwing him off target, or a speck of dust causes him to sneeze, or his target coincidentally moves at the last second. In all versions of the story A[2] misses his shot, and therefore escapes guilt for murder, since there can be no murder without someone dying as its consequence. On these facts, A[2] will be guilty only of attempted murder and typically subject to a term of five years or less in prison. n4 His unfortunate counterpart A[1], on the other hand, could be convicted of first degree murder, and in most states he would be condemned either to life imprisonment or the death penalty.

The difference in the sentences inflicted on two persons whose criminal wrongdoing was the same and whose degree of moral blameworthiness was identical seems to indicate that the legal system which countenanced it is not committed to the principle of proportionality, which requires that the severity of the punishment be proportional to the moral blameworthiness of the offense. In these examples the moral blameworthiness of criminals is identical, yet the punishment is much more severe in the one case than in the other. Unless there is some reasonable explanation for this discrepancy, the sentences seem to be more arbitrary than rational, the difference between the fates of A[1] and A[2] being determined not by their deserts but by luck, plain and simple. n5 The full  [*119]  explanation of how identical intentions produced importantly different results in these cases must appeal to factors that were beyond the control of one or another of the would-be assassins. Surely A[2] earns no credit for the antecedently unlikely survival of his enemy B[2]. He did not provide B[2] with his bullet proof vest, or arrange for a mosquito to land on his own nose.

When such momentous matters as a choice between the death penalty and a minor term of imprisonment ride in the balance, we want the decision to be as free of arbitrariness as possible. Reliance on good or bad luck, or assigning weight to factors beyond the foresight or control of the central players, introduces an element of arbitrariness into court proceedings. Arbitrariness is the absence of rule, as in the bare will of an authority who can exert his power free of accountability, in a manner without rhyme or reason, which in turn makes predictability and security from abuse difficult, and fairness an inapplicable notion (except perhaps in reference to the honesty with which a randomizing procedure, like a roulette wheel, is used). Arbitrariness is to a legal system what corrosive rust is to machinery. n6 For this simple reason I line up with those theorists who would treat completed crimes and unsuccessful attempts essentially the same, other things being equal. Since most legal practice throughout the world treats failed and successful attempts quite differently, the position I have taken here is a reformist position, and its logical denial can be called the traditionalist or retentionist position. I am a reformist on this issue because I believe that if the law is arbitrary in some respect, then provided we can improve it in that respect at a reasonable cost in other values, we should improve it.

What I would propose, first of all, is that we eliminate the causal condition in the definition of all so-called completed crimes. Thus, in respect to a crime of killing, there would be no necessity that a death actually result from the criminal acts in order for the actor to be found criminally liable. Of course this change in practice would require a change in terminology. If a resultant death is no longer covered by the definition of "murder," for example, then participants in the criminal process would start saying some very odd things such as "Jones murdered Smith although Smith is still alive." n7 I would prefer to have a more comprehensive crime with a new technical sounding name, like "Wrongful Homicidal Behavior." Then we could say that Jones is guilty of "Wrongful Homicidal Behavior" (WHB) toward Smith without saying something that sounds silly or even contradictory like "Jones murdered Smith but did not succeed in killing him." The definition of "Wrongful Homicidal  [*120]  Behavior" would look very much like our current definition of "murder," except that the definition would have no component clause requiring that the victim actually die. The word "murder" then would become functionless and drop out of the law altogether. Moreover, no effort would be made to reinstate something like the old distinction between murder and attempted murder. Every act of endangering, threatening, or taking life that is judgable within the system would be either an instance of WHB or else legally innocent.

Perhaps legislative draftsmen would wish to distinguish WHB in the first degree from WHB in the second degree, because they believe, plausibly enough, that some accompanying motives are worse than others; or that some culpability conditions, like acting on purpose, are worse than others, like inadvertent negligence; or that premeditated crimes are ceteris paribus worse than impulsive ones, and so on. On first impression, at least, I see no reason why these judgments of comparative blameworthiness need be arbitrary, n8 and I would have no objection to them so long as they did not rely upon luck factors which, being beyond the criminal's foresight and control, are neither his fault nor to his credit. The important advantage of this change would be that luck factors would be purged from the sentencing guidelines. Some defendants convicted of WHB could receive more severe penalties than others, but only when their conduct seemed more blameworthy than the others, not simply more lucky.

The obvious defect in this proposal is that the drastic changes it requires in our terminology will cause confusion. Perhaps then we should seek a way to downplay the moral significance of the distinction between murder and attempted murder without such radical departures from ordinary as well as technical language. We could do that by continuing to use the terms "murder" and "attempted murder," while letting nothing substantive hinge on them. Our penal code, for example, could forbid WHB on pain of severe penalty, defining WHB as "any act of murder or attempted murder which...etc." The rest of the definition would specify mens rea, actus reus, and other definitional elements, except for the causal condition, which is left out because it has no relevant bearing on blameworthiness, and that being the case, the most blameworthy criminal actions may or may not satisfy it. In other words when we look at the whole motley collection of lawbreaking actions, and isolate those we would want to call the very most blameworthy (or morally culpable) ones, some of them will actually cause some victim's death, and some will not. In short, there are two distinctions here that can overlap: that between murder and attempted murder (expressions in ordinary language which could still be useful but which would have no legal importance), on the one hand, and the relatively high or low degree of blameworthiness manifested in one's criminal actions, whether they be killings or only attempted killings, on the other hand. The severity of punishment would depend on the latter distinction (high or low degree of  [*121]  blameworthiness), not on the former (attempt versus completed crime). Assessing blameworthiness, of course, is the hard part, but no harder than in any other penal code that is committed to proportionality.

These refinements should help disperse confusion, but they are not yet sufficient. The genus in which our definition places WHB must be made more inclusive. We could include in it, for example, in addition to failed and successful attempts, and degrees and types of blameworthiness, conscious and unconscious risk creation, so as to capture all the types of WHB--all instances of endangering, threatening, attempting to take, taking, and risking another person's life. So the definition of the crime of WHB might be something like this: "any act of killing or attempted killing or the faulty or blameworthy creating of an unreasonable risk of killing, whether or not the actor was aware of that risk, is an act of WHB." Then a further discussion of an actus reus condition, various mens rea conditions, circumstantial conditions, and so on, would make the definition more precise and permit us to distinguish the various categories of WHB. (If the word "homicidal" is confusing in this usage, perhaps "survival-affecting" would be better.)

I will belabor no further the point that reliance on luck in the law is a kind of arbitrariness, that arbitrariness is a bad thing in a legal system, and that it is inconsistent with the principle of proportionality to which most of us give regular lip service. What puzzles me is the fact that among those who have considered this matter, and among penal codes here and abroad, now and in the past, almost all of them advocate treating attempts more leniently than their corresponding completed crimes, the few exceptions being almost all recent. Legal theorists and philosophers, however, are found on both sides in this controversy. Perhaps even a small majority line up with me on the reformist side. That reassures me that I am not abnormally insensitive to intuitions that others find irresistibly compelling, but beyond that, I am surprised at the number and high quality of my opponents. It is sobering to learn that among those who disagree with me on this question, including only recent writers, are: Michael Davis, George Fletcher, the team of LaFave and Scott, Michael Moore, and Judith Thomson. Among the lesser allies of these luminaries, there is often a perplexing tone in their prose, perplexing to me, which suggests that they feel they must defend the status quo on this question as though the survival of Western Civilization itself depended on it.

Before considering some of these arguments, I should clarify one presupposition of the whole debate. We reformists are arguing primarily for a kind of equality. We insist that equal degrees of blameworthiness should be punished to an equal extent. But to what extent is that? When we begin with a sentencing code that decrees unequal punishments for crimes that we think ought to be treated equally, there are two ways that equality can be achieved. Either we can "level up" or we can "level down." If murderers are sentenced to forty years imprisonment and attempted murderers to ten years, we can make the prison terms equal either by raising the penalty for attempted murder, say from ten years to forty years, thus achieving equality in the sentences by "getting tough on criminals;" or we can level down, and achieve equality by reducing the penalty for murder say from forty to ten years; or we can do both at once: raise the imprisonment terms for attempted murder from ten to twenty-five years and reduce the term for successful murder from forty to  [*122]  twenty-five. Which option we should choose depends on a diversity of things that make it impossible to answer in a perfectly general way. We would have to look carefully at the conditions peculiar to given communities. Moreover, our decision would depend not only on our reaching agreement on the problem of failed attempts; it would also require a choice among the major theories of the justification of punishment generally. The position of this paper would put only one constraint on how those larger issues are settled. Unless impossible for practical reasons, punishments for what we now call attempts and corresponding completed crimes would have to be equal. One consequence of this point is that the reformist thesis is neither one of greater harshness nor one of greater leniency. What this position demands is greater consistency.


The argument I present in Part I is simply the standard case for reform recognized even by retentionist writers as having some prima facie cogency. It is not complex, subtle, or ingenious. All it does is begin the debate by moving the burden, ever so gently on to the retentionist's shoulders. The retentionist then offers his arguments ("proof" is much too strong a term) for maintaining the status quo. Only a small handful of these multifarious arguments are represented here. Those arguments are subtle, and often ingenious, probably because the assigned burden is so difficult. What puzzles me is how bad many of them are. It might be especially instructive, therefore, to examine the bad arguments of some good thinkers against it.

Bad Argument Number One is the work of Professors W. R. LaFave and A. W. Scott, authors of one of the best-regarded recent treatises on criminal law. n9 These genuinely distinguished authors appear to argue not only that unsuccessful attempts should be punished less than already accomplished crimes, but that mere attempts should not be crimes at all. Their case is put succinctly: "Since criminal law aims to prevent harm to the public, there can be no crime without harm." n10 This pithy statement the authors call a "basic premise of the criminal law." n11 I can only guess at their reasoning, but perhaps it could be reconstructed as follows. They may think that lawmakers should criminalize only behavior the predominant tendency of which is to cause harm to people other than the actor. This statement formulates the liberal theory of the proper limits of the criminal law, and is close to a principle that I have defended, n12 which has been called the "harm to others principle," or "the harm principle" for short. Anyone who approves of the harm principle will disapprove of socalled victimless crimes (criminalized activities of a kind that do not tend to do harm). Liberalism on this question, in effect, recommends to legislators that if an action is of a generally harmless kind, then it does not matter that they sincerely disapprove of acts of its sort; it is not their business to interfere with the liberty of those citizens who do not share their judgments of disapproval. So far, so good. But it does not follow that unsuccessful attempts to perform an act of a type that is very harmful (like murder, for example) should not be  [*123]  criminalized! That is a well-intentioned but illogical conclusion drawn from a principle (the harm principle) that gives it no support.

The second Bad Argument will tempt those who tend to confuse the functions of criminal and tort law. That would seem to be an easy kind of mistake to avoid, but language sets traps for the unwary at every turn. The advocate of this approach is likely to share with the other writers discussed in this section a fidelity to the principle of proportionality, together with a tenacious adherence to the retentionist position about attempts. She too effects a marriage between these unlikely partners by trying to show that the murderer is more blameworthy than the unsuccessful attempter even though they are, in all the respects usually thought to be relevant to blameworthiness, exactly alike. Why then does this person believe that blameworthiness is a function of actual harm caused? The answer might be that such a person has before her mind an inappropriate model for criminal law, inclining her to accept uncritically the following chain of inferences:
(1) If Sadie is more blameworthy for her conduct than Sally is for hers, then Sadie is to blame for more than Sally is to blame for.
(2) That in turn implies that Sadie produced more harmful results than Sally did, perhaps killing two victims instead of one, or a married person instead of a solitary one.

The point is this: if you are responsible for more harm, then you pay for more, or alternatively, the more harm you cause, the more harm you must pay for. But this approaches self-evidence only in the law of torts where we are adding up amounts so that we can present a bill to the harm-causer, not trying to assess the exact character of her moral desert.

If our problem is to explain why so many people identify wrongdoing with harm-causing, we should look at tort law as an example of the inextricable linkage of the two and criminal law as an example of a much looser connection. Then we need only look at the different aims of these two branches of law for an explanation of why criminal law can get along without responding directly and exclusively to harm actually caused, whereas tort law's function does not permit it to respond to anything unless there has been actual harm caused. Here too we may find the explanation of why some people find it natural, even in a criminal law context, to assume that "more blameworthy" is entailed by "blamable for more."

The reformist on the problem of punishing attempts, as we have seen, proposes the removal of the causal requirement from the definitions of crimes, for example by eliminating the requirement in the law of homicide that the actor's victim die as a consequence of the actor's attempt to kill him. If we took that step, then we would never have to decide whether the defendant was guilty in the sense of responsible for the death, but rather only if he was guilty of breaking the law.

Perhaps a part of the explanation of the persistence of the causal requirement in criminal law then is that there is still a confusion between the aims of criminal and tort law, and that a variety of English prepositions, such as "for," "in" and "of" exert their subtly different effects on meaning. n13  [*124]  Another reason is the American penchant for commercial metaphors. n14 A terrible harm happens to someone, and an angry cry goes up--"Someone must be made to pay for this!" Then a search is undertaken to locate a wrongdoer  [*125]  whose criminal act cum criminal mind was "the cause" of the harm, and who, once apprehended and convicted, must "pay" for it. The major, if not the entire, point of tort law is to compensate relatively innocent parties for the harm they suffered as a consequence of the defective actions of others. If there is no harm to begin with there is nothing more for tort law to do, since there is no loss to make compensation for. Here then is a context, namely tort law, where LaFave's and Scott's maxim that there can be no liability without harm is more at home. The maxim comes closer to the truth when the liability it refers to is tortious, not criminal.

Even where there is nothing for tort law to do, however, there may still be something for the criminal law to do, for its function is to punish the parties who perform criminal acts, whether or not harmful consequences in the individual case actually ensue. To be sure, legislators should enact prohibitory statutes only to prevent types of conduct that tend to cause harm to others. But it does not follow that such statutes should be enforced only when their violation actually harms someone. On the whole, the rationale of the criminal law is that it reduces the net amount of harm all around by discouraging all conduct that is in fact dangerous to others, even in particular instances when it does not issue in harm. But it does that job more efficiently when it punishes disobedience and neglects its confused mission of making people "pay" an additional surcharge for actual harms. A traditional usage worth reinstating would be to say that tortfeasors are civilly responsible for the harmful consequences of their acts and omissions, whereas criminals, as such, are criminally responsible not for subsequent states of affairs, but for their own criminal actions in disobedience to law--whatever the consequences.


We come now to a more respectable argument for the traditionalist position on the punishment of unsuccessful attempts. In one of its versions it is associated with the name of Oliver Wendell Holmes, Jr., n15 and in more recent times with that of George Fletcher n16 and Judith Thomson. n17 The argument begins by surveying facts it deems relevant to its argumentative purposes, namely that almost every developed legal system in the world distinguishes between attempted and completed crimes. Also, the overwhelming majority of sentencing codes and guidelines punish completed crimes more severely than mere attempts that fail to cause harm. There is also evidence that a large majority of the adult population favor the status quo in this respect. These are significant facts in a democracy. If we do not respect the sensibilities of common people on a matter that they take very seriously, we endorse by implication a government in which only an intellectual elite matters, and the genuine convictions of a majority consensus are dismissed as moral superstitions. In a democracy, Holmes insisted, those who make, enforce, and interpret the law cannot move too far out of harmony with the common people, else the law will seem tyrannical at worst, obscure at best.

 [*126]  George Fletcher, arguing for the traditionalist position on this question, gives his own voice to the consensus. In his recent book, A Crime of Self Defense, he concludes that, "We cannot adequately explain why harm matters, but matter it does." n18 Some elitist professors n19 may hold to the contrary, but Fletcher is firm: legal decisions should not go beyond "the...sensibilities of common people." n20 Whether that is because of superior wisdom among the uneducated or political wisdom among those who rightly defer to their greater numbers, or both, Fletcher does not make clear, but I suspect that he would say both. That the bulk of the people believe that a particular proposition is true is a good reason, I agree, for tolerance and respect. But it is not a good reason, even in a democracy, for believing that proposition to be true.

There is a tendency, I believe, when discussing issues that have both a moral and a legal dimension, to conflate three independent questions:
1. Which rule, x or y, is the more rational (or otherwise superior?)
2. Which rule, x or y, is more widely accepted in this political community?
3. Which rule shall we adopt? Shall we retain x or reform the status quo by adopting y? (This is a question for a legislature or maybe a court, or any group with the de facto power to give shape to legal rules.)

There is, in addition to these questions, a principle that has been used by one writer or another to generate answers to all three of the questions. We can call this, with all respect, the conservative principle, and formulate it as follows: People with the power to change legal rules ought not to do so unless some substantial percentage of ordinary people want them changed, or at least would approve the change were others to bring it about. At the least, it ought not to be imposed on a resentful, dissenting majority.

What is the status of the conservative principle? Clearly, it suggests an answer to question # 3, not to question # 1. It is, therefore, a fallacy to say that because "x is more rational (or more fair or more useful) than y, we should adopt x even in a community whose traditions and sentiments are strongly in favor of y." Many severe critics of the "fortuitous gap" between failed and successful attempts understand this well. Sanford Kadish, for example, concedes that "there are limits to how far the law can or should be bent by reformers  [*127]  [like me] n21 to express a moral outlook different from [those]...common to our culture, irrational though they seem to me." n22 Kadish's combination of views go well together, it seems to me. He is a conservative in the manner and timing of his public advocacy, but he holds on to liberalism and common sense in his private judgment.

Professor Fletcher is powerfully impressed by the depth of our "cultural attachment" to the idea that actual harm makes a significant difference in the proper assessment of wrongdoing. He is prepared to accept the probability of fire given all that smoke. Successful attempts, which actually harm people, he suggests, are more culpable morally, other things being equal, than equally genuine efforts that fail through lucky accident. If that is the case, then unequal punishment of the two crimes does not really violate the principle of proportionality after all. Fletcher, so interpreted, urges that we punish according to blameworthiness (ill-deserts, moral culpability) alone, as the proportionality principle seems to mandate, but one of the chief determinants of blameworthiness, in turn, is actual harm caused. So Fletcher can have his cake and eat it too, only now instead of an odd theory of the grounds for criminal liability, he has an odd theory of the grounds of moral blameworthiness.


The remainder of Professor Fletcher's discussion of the problem of punishing failed attempts is taken up with an analysis of "moral emotions" like guilt feelings, and feelings of remorse and shame. The attention he pays to these feelings has the effect, I think, of converting his reasoning into a kind of argument from public opinion, at least as public opinion is revealed in such emotions as guilt and shame. If the failed attempter feels no guilt, for example, and most others behave as if they do not expect him to feel guilt, then he and we are entitled to infer that this public, at least, believes that the harm caused by criminal acts is an independent determinant of moral blameworthiness. If we do not feel guilt, this suggests that we do not believe that unsuccessful attempts can be as morally culpable as successful ones. This kind of argument, common in ethics, is designed to convince the reader that he already holds the belief that "is to be proved," so that further demonstrative argument is hardly necessary.

To make his view more familiar and appealing, Fletcher has us consider what emotions are thought appropriate (in our culture) first to one's own "near miss," say, in a murder attempt, and second to one's successful intentional killing. His answer is that feelings of guilt and remorse are appropriate when the killing attempt was successful but not when it fails after a near miss; it is shame that we should expect the bungling assassin to feel, but not guilt. Moreover, if the would-be killer feels guilt at his near miss anyway, the guilt is probably morbid or perhaps neurotic:
Feelings of guilt and remorse are appropriate where harm is done, but if all is the same after as before the act, there would be nothing to be remorseful about, and the actor's feelings of guilt would make us wonder why he wanted to suffer inappropriate anguish. n23

 [*128]  Fletcher apparently treats "remorse" and "guilt feelings" as synonyms, which I think is a mistake. On the other hand, what he says about remorse, quite apart from its relation to guilt, seems right, though the advantage he gets from that insight are more than offset by the disadvantages he incurs through his mistaken account of guilt feelings. So I shall put aside, for the moment, the subject of remorse, n24 and focus on Fletcher's theory of guilt.

One of the more common motives for murder attempts is passionate jealousy. If a modern day Othello, while consumed with that passion, attempts to shoot Desdemona, but through lucky accident just barely misses her, he might very well feel, appropriately, other emotions a couple of hours later--renewed tenderness and affection toward his wife, self-hatred and loathing, maybe remorse (I am not sure) but certainly guilt. That is to say that if Othello has the normal sort of conscience that internalizes the teachings of one's parents, teachers, and ministers of religion, imprinting on one's character the precepts that fraud, theft, cruelty, rape, torture and murder are intolerable and inexcusable sins, and he nonetheless acts against his own conscientious convictions, he will not feel very happy with himself. We might even expect, appropriately, that he will beg forgiveness, demand punishment or release torrents of self-abuse. Except for the "conscienceless killers" who are often up for hire, attempting murder is no frivolous or routine thing.

Would it be some obvious moral category-mistake to add to this description of expectable or "appropriate" responses, the word "guilt"? Is it some kind of conceptual solecism to suggest that a rational person in Othello's position, aware that only a lucky accident for which he should get no credit at all, prevented his "near miss" from being a brutal obliteration of a fellow human being, that such a person might feel some guilt over what he intended to do? Professor Fletcher's account of guilt seems to me to be an accurate description only of the conscience of a simple sort of utilitarian. But surely people who are committed to moral codes that are at least partly nonutilitarian can feel guilty whenever they act, feel, or desire contrary to their own superegos. Adolescents express their guilt to priests for having enjoyed lascivious thoughts, with no obvious harm to others. Indeed self-blame for such fantasies is sometimes thought to be the very paradigm of a guilt feeling. And it is not only sexual taboos whose infractions can produce such feelings. The same is true, though usually to a lesser degree of intensity, of sabbatarian and dietary taboos. To be sure these are examples of what Fletcher would call "nothing to feel guilty about." If he means by this remark that he rejects the substantive moral beliefs embodied in the prohibitions, then I have no disagreement with  [*129]  him. But there is no denying, it seems to me, that there is conceptual coherence in using the language of guilt even for acts and thoughts that are morally innocent in the judgment of others. The mistake is moral, in those cases, not linguistic or conceptual. But there is no comparable mistake in judging that a failed attempt to murder is morally guilty. Such language is both conceptually and morally appropriate. And if the unsuccessful killer has no guilt feelings, there is something wrong with him, not with the language. In other words, it is not a case of "right feelings, wrong words."

This is not the place to analyze the concept of moral guilt, but let me suggest that the element in the concept whose absence is especially conspicuous in Fletcher's argument, is the idea of disobedience. At any rate, to whatever extent moral guilt is a kind of legalistic notion transferred to an extra-legal world, disobedience of authority is a quite essential feature. Consider the following weak effort at moral phenomenology, a picture of what one of my guilt feelings might be like. Actually, I think there are two distinct models of moral guilt. n25 The following feelings would fit the first model. I succumb to temptation to do something that has been forbidden by some rule or person whose authority I unquestionably accept (mommy, daddy, God, or the Municipal Penal Code). The authority is internalized. When I act He is watching. He knows! I squirm with the inescapable awareness of the insolence in my rule-infraction. My pain comes not primarily from the hurt or harm I may have caused incidentally and not from having embarrassed myself in public. (There may have been no witnesses.) I am upset simply because I have disobeyed thereby alienating myself from the source of moral authority, and also perhaps for fear of the punishment I know I deserve. n26

The moral appropriateness of the quite different feelings that satisfy the second model of guilt is more controversial. But there is little controversy over the linguistic appropriateness of labeling them "guilt feelings." Consider some standard examples. Through a fluke I survive the concentration camp while all the other prisoners perish, or I survive all of the battles, while my best friends among my fellow soldiers are all killed or maimed. Why me? I was not the most talented or virtuous, or the most deserving on balance. Now I feel unworthy of being alive. In another example, for some unknown reason, teacher gave me an A in the course, whereas Tom, Dick, and Harry, who are plainly more talented than I, got Bs. I can hardly look them in the face. I am so burdened by my unworthiness that I slink away whenever I see them coming.

The first description of guilt feelings follows the older established concept of guilt (disobedience of moral authority) and sin (disobedience of the sort that implies rejection of a rule-maker's authority). The essence of the first  [*130]  conception is disobedience or rejection, whereas the essence of the second is unworthiness. Remorse characteristically (that is, when dissociated from moral guilt) involves intense regret, a painful wishing that the episode could be called back, that the hurt or harm to others or embarrassment to one's self had never happened. I suspect that remorse is more often worthy of respect than guilt is in either of its senses, and is more directly linked to contrition.

The first of Fletcher's contentions, then--that it is appropriate to have moral guilt feelings only for completed crimes--seems mistaken. The second--that shame rather than guilt is a the appropriate feeling to have in reaction to a luckily failed attempt--does no better. Unlike guilt in Fletcher's account, shame at one's own conduct does not require that actual harm to any one need be caused. Thus, he writes:
If I contemplate killing the Pope and never do anything about it, I might feel ashamed of myself, but it would be out of place to feel guilty about my intentions. Feelings of shame are triggered by the realization that one is not the kind of person one thought one was; feelings of guilt by acts that rupture our relationships with other persons. n27

I find much to quarrel with in this brief passage. n28 First, however, we should look for common ground. All parties to the dispute should agree that it is possible to feel guilt and shame both, to some degree and in some fixed proportion, as a response to the same act or thought. I might feel shame in reaction to my bare consideration of the possibility of shooting the Pope, though it is more likely that I would visit a psychiatrist for aid in dealing with such a downright crazy thought. So it is much more likely, that I (for one) might experience a kind of moral alarm that I could even come as close to doing wrong as only to think about it. This, in my case, would be more like highly sensitive guilt than like shame. I feel guilty when I sin against moral authority (as I conceive it); I feel shame when I sin against my own ideals. Now part of my ego-ideal, part of what I feel proud of when I feel good about myself, is my impeccably moral negative conduct (the part of morality that is relatively easy but also the most important): I never kill people, beat them, rape them, cheat them. I am honest and trustworthy. I even have some of the easier positive virtues, like friendliness for example. All of that is of great importance, but it is also rather bare minimal. There is much more in my ego-ideal than just not being a moral monster, and I imagine that is true of most people. Some people pack into their ideal self-aspiration some moral virtues that are more difficult to achieve than the bare minimal ones--heroism and saintliness for example. When they fail to do something saintly, it is not necessarily because they do something immoral. So their consciences are left undisturbed, even though their self-image is tarnished in their own eyes and they feel shame.

Some shame has nothing to do with moral virtues of any kind. Part of my ego-ideal may be to be a baseball star. I may then feel shame if I make several errors in one game. Of course what is called shame, in this sort of public arena,  [*131]  may in reality be little more than embarrassment. To have failed in front of all those people is what really hurts. Genuine shame is a kind of embarrassment in front of myself only. The aspiring side of my nature feels something akin to embarrassment when it is alone, "face to face" with that part of my nature that formulates and assigns ideals for me to live up to. When I do well in approximating one of my self-ideals, I feel proud; when I "embarrass" myself in my own eyes, I feel shame. When I make an ass of myself before others whose admiration is important to me, I feel embarrassed in the strict sense.

What do I feel when my deliberate effort to shoot another human being is foiled by a random sudden movement of my prey or a mosquito in my face? Surely not embarrassment; there may have been no one watching. Certainly not pride. Probably not a predominant shame, though there may be a trace of that element in the complex mix. (If I am proud of my ability as a marksman, I may feel shame at having missed an easy shot.) But unless I am a satanic immoralist or an amoral psychopath, what I feel, if I feel anything, will be guilt.


The reformist approach to the sentencing of unsuccessful criminal attempts seems highly plausible initially because it conforms to the principle of proportionality which decrees that the just punishment for any convicted criminal is the one whose severity is proportionate to the degree of blameworthiness of that actor for doing what he did. The traditional approach, on the other hand, attaches more significance to the amount of actual harm caused than to the blameworthiness of the actor in contributing his share to the production of the harm, and does not exclude good and bad luck as major determinants of the severity of the penalty. To grant so large a role to luck is to invite arbitrariness to govern, and inevitably to corrupt the criminal process. The principle of proportionality, after all, does not decree that the severity of the punishment be proportionate to the offender's good or bad luck, but rather to his good or bad deserts, or blameworthiness.

But when we looked at some arguments for retention we discovered that none of them claim that the principle of proportionality is wrong in giving such an important role to moral blameworthiness. In fact, they all endorse the principle of proportionality but then go on to argue that the blameworthiness that punitive severity is to be proportionate to is determined by something else, namely the amount of actual harm produced by the offender's contribution in combination with the contribution of parts of the natural world over which the offender had little normal control. Giving good reasons for that proposition--that moral blameworthiness is itself grounded in actual harm caused--is such a heavy burden that it is no wonder the arguments we examined were so bad.

But to back up a bit, how is blameworthiness ordinarily understood, outside of the attempts problem? To begin with, the criminal's blameworthiness is not to be confused with the wrongness of what he did simpliciter, for it is possible for a generally virtuous person to do the wrong thing in the circumstances and yet not be properly subject to much blame for doing it, or for a generally evil person to do what is right but not deserve any credit for it. But what, apart from the wrongness of his act, determines the degree of blameworthiness of the actor?

 [*132]  A great number and variety of factors go into the determination, whether we are talking now of criminal sentencing guides or the moral judgments they partially incorporate. A sound if blurred insight is that the harm intended is much more important an indicator of an offender's desert than the harm actually caused. Far more useful, however, than the concept of intentionality, are the four "culpability conditions" first proclaimed in the Model Penal Code--acting purposely, knowingly, recklessly, or negligently in regard to some harmful result. Then, of course, the concept of a motive, ruthlessly kept out of the original criminal trial, forces its way back at the sentencing stage, and contributes its flavor to the emerging blameworthiness stew. Did the offender act cruelly? spitefully? from mercenary motives? out of greed? in an emotional explosion provoked by a sexual rejection? or by sexual jealousy? or through political conviction? out of mercy or compassion for another person's suffering? after forethought and deliberation? out of conscientious conviction or the determination to do at any cost to oneself what one sincerely believes to be one's moral duty? out of sudden violent impulse as mysterious to the actor as to anyone else? And after all those questions have been answered, and provocation considered, and other types of mitigation, and diminished responsibility, and the questions they raise, now at last the more traditional justifications and excuses enter the arena with their talk of subtle coercion and mistakes, mistakes of fact and mistakes of law, defenses that undermine and defenses that affirm, defense of self and defense of others, with duress and necessity, involuntary and voluntary intoxication, insanity, and short of insanity a host of neuroses and psychoses, compulsions and obsessions, and the great parade of syndromes, and on and on. There is surprisingly little disagreement about the factors that belong on this list, but much disagreement about the weight to be given different factors when they conflict.

Does the retentionist really wish to supplement all of that with a simple appeal to the amount of harm actually caused? Is he willing to let the final judgment of moral blameworthiness be swayed by a luck factor reflecting nothing about the actor's character? If the incredibly complex test of blameworthiness now used pulls in the direction of extreme moral culpability would he be willing to let the fortuitous absence of a serious harm pull it back in the direction of moral innocence? If so, how far back? I find no intuitive plausibility at all in basing criminal liability on moral blameworthiness, thus securing for one's position a certain prima facie respectability, but then basing moral blameworthiness upon, as luck would have it, the actual harm or absence of harm caused. The ancient view, that liability should be based not upon blameworthiness at all but instead directly upon the amount of harm caused, seems to me more honest, though no more plausible. It seems almost as if the retentionist is so fixated on actual harm that he keeps searching for the question to which it is the right answer. Not the question: What is the basis of criminal liability? Moral blameworthiness is that. Not the question: What is the basis of moral blameworthiness? The traditional multiplicity of culpability conditions, motives, mitigations, aggravations, etc. is that. How about the question: What is a necessary condition for tort liability? Now, that is more like it.

Why are good philosophers and legal theorists so determined to find a central place for actual harm caused (if any) that they are led into such bad arguments? I don't know, but I suspect that it may be partly because of a discomfort many persons have with moral concepts, which in order to be  [*133]  applied correctly, require our access to a human being's inner states, his emotions, passions, opinions and distortions. Not only is the invasion of another's privacy unsettling; being required to make fine subtle appraisals of what we find there is even worse. How much less slippery is our task when we are asked only to take measurements, conduct tests, read laboratory reports and the like. And it is much less daunting still to measure dollar losses, to subtract insurance coverage, and apply the other, relatively objective, measures of harm and loss.

And, finally, it is not as difficult for good thinkers to tumble into bad arguments as one might think. The subtle influence of compensation concepts on criminal punishment concepts has tricked many a complacent theorist. And it is an easy equivocation to think that one is praising the harm principle as a standard for the moral legitimacy of criminal legislation by urging the decriminalization of all unsuccessful attempts to inflict harms, even the harm of death. Professor Fletcher's mistaken analyses of guilt and shame, I realize, were animated by an unusually powerful (and commendable) motive to avoid elitism. Even though, on some scale of "cognitive blameworthiness," these mistakes are not very blameworthy, they can nonetheless cause considerable mischief, intellectual and other, and in actual cases, make the worse appear the better cause.


n1 The earliest reference to the problem is in Plato's last dialogue, THE LAWS IX, 876. Plato was so concerned with the various combinations of perpetrators and victims, particularly the possible combinations of family statuses--father-son, son-father, husband-wife, sister-brother, slave master-his own slave, householder-neighbor's slave, etc., etc.--and also with the combinations of intended and actually resultant harm involving that cast of players--intended killing-actual maiming, intended minor physical injury-actual killing, intended injury-actual absence of harm altogether, and so on, that he has little time for a detailed discussion of the combination that concerns us here, namely intended killing by anyone at all of anyone at all, resulting by a fluke in a failure to do any harm at all. Plato devotes only a couple of sentences to that kind of case, but the tone of his remarks suggests that he would concur with the position taken in this paper. He writes of the unsuccessful homicidal attempter that: "He is not to be pitied. He deserves no consideration but should be regarded as a murderer and tried for murder." 4 THE DIALOGUES OF PLATO 876, 445 (B. Jowett trans., 4th ed. 1953). Plato's model penal code has only one other requirement pertaining to this kind of case, and clearly it is a response to the intended victim's good luck: The luckily escaped would-be victim, lest his apparent ingratitude should tax the patience of the deity, must make an offering of thanksgiving.

n2 Yoram Shachar uses an especially felicitous name for our problem: "the problem of the fortuitous gap." See his interesting study in moral psychology--Yoram Shachar, The Fortuitous Gap in Law and Morality, 6 CRIMINAL JUSTICE ETHICS 2, 12-36 (1987).

n3 In California, wherever a sentence to a term of imprisonment with a fixed upper limit is specified in the sentencing code, it is stipulated that the penalty for the corresponding unsuccessful attempt will be exactly half as long as the specified maximum. CAL. PENAL CODE 664 (West 1994). In Great Britain, some jurisdictions permit the sentencing judge to fix her own discount rate. See e.g., R. v. Clfaver, 13 Cr. App. R(S) 449 (1991).

n4 A combination of factors including the traditional Anglo-American reluctance to be impressed by the moral seriousness claimed for crimes that cause no harm, (in the common law mere attempts were classified as misdemeanors only) and our current crisis of prison overcrowding explains why in some states it is now common for prisoners who have been convicted of attempted murder to be released on parole in only a few months.

n5 To those who respond to this point by saying that "all life is an uncertain risk" and "luck is ineliminable from human affairs," Richard Parker has the right answer: "Fortune may make us healthy, wealthy, or wise, but it ought not to determine whether we go to prison." Richard Parker, Blame, Punishment and the Role of Result, 21 AM. PHIL. Q. 269, 273 (1984).

n6 There is at least one kind of exception to this sweeping denunciation. When there is a reason against resorting to reasons, then random decisions and selections, ordinarily called "arbitrary," are perfectly appropriate. Thus random procedures for conducting lotteries to raise funds, or to raise troops in a wartime conscription, are meant to assure impartiality by excluding bias and favoritism. One can defend these techniques in context either by saying that they are not really arbitrary since they have a reason, or by saying that they are instances of justified arbitrariness. Which terms are preferable for putting this point is itself an arbitrary question with no clear answer, and no importance.

n7 It is interesting to note, incidentally, that this "very odd thing" is precisely what Plato says about the unsuccessful attempted killer: "he should be tried for murder," and his failure actually to kill his enemy is no reason for giving him any special "consideration." 4 THE DIALOGUES OF PLATO supra, note 1, at 876, 445.

n8 In this opinion I seem to differ from Chief Justice Rehnquist in Rummel v. Estelle, 445 U.S. 263, 281-84 (1980), who there writes: "[Yet] rational people could disagree as to which criminal merits harsher punishment...Once the death penalty and other punishments different in kind from fine or imprisonment have been put to one side there remains little in the way of objective standards for judging...." Two questions, at least, are involved here: (1) whether judgments of the comparative moral gravity of offenses and the comparative severity of different criminal penalties are subjective (as Rehnquist says) or objective (say in the manner of scientific reports) and (2) whether these judgments if subjective must therefore be "arbitrary" or beyond reasoning altogether.


n10 Id. at 7.

n11 Id.


n13 A philosopher should tread with caution, however, if she uses the phrases "for their actions" or "for the consequences." Quite apart from their moderate infelicity in some contexts, especially when they are used with the word "guilty," these phrases are conceptual booby traps. Perhaps they are used most commonly with the word "responsible" since that term has a legitimate role in moral discourse as well as in all branches of the law. Even the idiom "responsible for," however tends to cause mischief when used in a criminal context, and "guilty for" is even worse. The danger in complacent use of the phrase "responsible for" in criminal law, when what follows the preposition is some state of affairs--a presumably harmful state of affairs--is that it suggests a relation between a harm causer and his harm, with the implied question of whether he should be made to compensate a harmed person for her loss. That, of course is the model of the tort situation. It is no better when we speak of a person being responsible not for a state of affairs that is the result of her conduct, but for her own actions themselves. That is rather clearly another sense of "responsible for," functioning in most of its uses simply to identify the agent of a given action, and in other uses may serve to identify the party who is answerable.

The preposition "for" brings to mind the standard case of producing results, and suggests to the unwary that in ordinary action we stand in a relation to our own actions analogous to the relation between an action and its consequences, or between those immediate consequences and more remote results or effects. The preposition "for" suggests a causal relation, leading philosophers in earlier times to say that when we are responsible for our acts, our contribution to their coming about must have been to do something in our heads that caused our body to move in a certain manner. But in that case, either we should use a different prepositional construction, and speak say of being guilty of an act of disobedience, or being guilty by doing or in doing something prohibited, or else we should stipulate that we are using the preposition "for" in a sense different from its customary (causal) one in which it stands for some linkage between act and consequence. When you say that you are guilty for some state of affairs you suggest that criminal guilt is determined as tort liability for harm sometimes is, by measuring the extent of the harms to be compensated, so that the more the harm, the more the liability.

When people talk of being guilty for their actions they inadvertently invite even greater confusion. Since the preposition "for" often stands for a causal relation, it seems that when we are responsible for a result, we must have done something to produce it. Thus, when a person is said to be guilty for his actions, the implication is that she did something first, (something prohibited) in her head to produce a "volition," or in her body to produce a "muscle contraction" in her trigger finger, in order to cause her action to come about. After that her contribution to the whole sequence blends with contributions from the outer world to produce the effects of her minimal action, some harmful, some not. (Sometimes the whole sequence of linked phases is itself called an action--her action--perhaps because it began in her head and was initially under her control. In that case no distinction is made explicitly between acting, doing, and causing, and instead of saying she pulled the trigger and thereby propelled a bullet irretrievably in the direction of the victim, people say simply "she killed him" or "she did it."

But if a person cannot properly be said to be guilty either for her action or for a state of affairs, then how can we mention the ground for holding her guilty at all? It would be much less misleading to leave the preposition "for" over in the land of torts, and speak of a person's being guilty of murder, or say that in killing the victim she violated the law, or by killing him, she incurred great guilt. And finally there is still one clear and legitimate use of the phrase "responsible for some action," namely "being properly identifiable as the doer of the deed." See Joel Feinberg, Action and Responsibility in DOING AND DESERVING 119-51 (Joel Feinberg ed., 1970).

n14 Everyone is familiar with the contention that criminals ought to "pay their debts to society," and that wrongdoers are made to "pay for their sins" when they are punished, as if one could buy a license to sin, and then pay first and sin afterward. Kadish and Schulhofer tell us that there is a very simple theory "that it is perfectly grade punishment according to the amount of harm actually done, whether this was intended or not." SANFORD KADISH & STEPHEN SCHULHOFER, CRIMINAL LAW AND ITS PROCESSES 624 (5th ed., 1989). They go on then to nominate the following sentence for our list of commercial metaphors: "If he has done the harm he must pay for it, but if he has not done it, he should pay less." From this kind of mixed metaphor one gets a picture of a clumsy chap accidentally breaking a dish in a china shop and then taking out his wallet to pay the debt thereby incurred. There is very little properly called "punitive" about that, especially if the owner of the shop remains polite and sympathetic.

n15 Oliver Wendell Holmes, Jr., The Path of the Law, in COLLECTED LEGAL PAPERS 167, 170 (1920).


n17 Judith J. Thomson, The Decline of Cause, 76 GEO. L.J. 137 (1987).

n18 FLETCHER, supra note 16, at 83.

n19 It is interesting how many philosophers have resorted to terms like "elitism" in their attempts to explain either why the arguments of philosophers often fail to convince ordinary people or vice-versa. Michael Moore, who defends what he called (in private conversation, with a twinkle in his eye) "the blue-collar view" refers to the view that is more popular with philosophers (like me) as "the standard educated view," thus associating it with the educated class, even though he, a leading member of that class, rejects the view. Michael Moore, The Independent Moral Significance of Wrongdoing, 5 J. CONTEMP. LEGAL ISSUES 237 (1994). Jeffrie Murphy refers to the preferred positions of the majority being condemned by "a refined and condescending elite." Jeffrie G. Murphy, Getting Even: The Role of the Victim, 7 SOCIAL PHIL. & POLICY 209, 212 (1990) (Murphy is speaking here of the attitude of "retributive hatred," common among ordinary folk, but ill-regarded by professors). Judith Thomson in her article about the causal requirement in the definition of torts and some crimes, refers to those who, like the present writer, would downplay actually caused harms, as "moral sophisticates" who regard the majority view as a "vulgar error." Judith J. Thomson, The Decline of Cause 76 GEO. L.J. 137, 139 (1987).

n20 FLETCHER, supra note 16, at 83.

n21 The reference in brackets is to Joel Feinberg.

n22 Kadish here refers to himself.

n23 FLETCHER, supra note 16, at 482.

n24 The following example I take to be fairly typical of remorse. I say something snappish and irritable to my wife at a public gathering that wounds her feelings and causes her public embarrassment. Instantly I regret what I am saying and could "bite my tongue," but of course it is too late to call back the words once they are out. The damage is done. For days after, I cannot recall the episode without pain. It "bites me back" as the French etymology of the word suggests. The sources of my pain in the example are dual. I keenly regret the hurt I caused my wife and I keenly regret that I made an ass of myself before that audience by revealing an ugly character flaw. Perhaps another source of my anguish is the hard truth about myself that I have such a flaw in the first place, but that needn't be part of the example, and it might diminish its purity as an example of remorse. In any case my reaction is primarily to the occasioning event: would that it had not happened! Why did I let myself do such a stupid thing? The attitude of remorse may be supervenient on guilt feelings, shame, or embarrassment, but intense regret is its essence.

n25 See Herbert Morris's latest helpful essay in this area: Herbert Morris, Nonmoral Guilt in RESPONSIBILITY, CHARACTER AND THE EMOTIONS 220-40 (F. Schoeman ed., 1987). Morris there distinguishes two or more basic types of guilt--moral guilt (the most familiar kind, based on legal models) and nonmoral guilt. The latter category includes as guilt feelings "anomalous cases" that are likely to seem confused, morbid, or neurotic to those who have been excessively concerned with "moral guilt" on the legal model. The anomalies include guilt for states of mind only, for unjust enrichment, survivor's guilt and vicarious guilt.

n26 I include this element of intense anxiety over being caught if only to bring into the boundaries of relevance Dostoevsky's Raskolnikov, who is also considered a paradigmatic study. In Dostoevsky's novel Crime and Punishment, Raskolnikov lived in near constant terror of being caught. It seemed that the only thoughts capable of deflecting this anxiety were those that drowned him in moral guilt.

n27 FLETCHER, supra note 16, at 483.

n28 For example, I don't understand why Fletcher, supra note 16, believes that a near miss (or any miss for that matter) does nothing to "rupture" (his word) our relationships with other persons. Suppose you shoot at my wife and miss. And a good thing too that you missed. Otherwise you would have ruptured our relationship (yours and mine)! This would hold true, I should think, only in a community of excessive forgivers.



Send comments to: Andreas Teuber
Last Modified: 03/26/05
Instructor's Toolkit
Copyright © The President and Fellows of Harvard College