II. GUILT AS WARRANTING RETRIBUTION

In this section, I sketch a version of Moore'sFootnote ⁶ argument in favor of positive retributivism as epistemically justified by the emotion of guilt. For the sake of simplicity, I reproduce the argument in seven steps, aware that many more steps could be added by a close reading of the text.Footnote ⁷ I trust, however, that omission of these further steps will not compromise the critical discussion that follows. Here is the argument:

1. 1. Emotions are heuristic guides to (or have epistemic import for) our particular moral judgments.

2. 2. Virtue (or lack of it) in the possession of an emotion is a fallible guide to the epistemic import an emotion may possess. In particular, “If the possession of an emotion makes us more virtuous, then that emotion is a good heuristic for coming to moral judgments that are true; if the possession of an emotion makes us less virtuous, then that emotion is a good heuristic for coming to moral judgments that are false.”Footnote ⁸

3. 3. The possession of guilt makes one more virtuous.

4. 4. Feelings of guilt typically cause one to judge (a) that one is guilty (in the sense of “morally culpable”) and (b) that one deserves (in the sense of “ought”) to be punished.

5. 5. There are no valid reasons that the first-personal judgment that one ought to be punished should not be transferred to third persons.

6. 6. The emotion of guilt validates particular judgments to the effect that someone deserves to be punished.

7. 7. By abductive reasoning we can generalize to the conclusion that moral culpability is sufficient to justify punishment.

In the end, we question the soundness of steps 3 and 4. This is in part because these premises are the ones that really bring emotional guilt into the picture and in part because steps 1, 5, and 7 have already received sustained critical attention.Footnote ⁹ Before focusing on our critical steps, however, it may be advisable briefly to run through the argument as a whole.

At the most general level, Moore's argument is grounded on a specific epistemological approach to ethical judgments as characterized by step 1. The idea here is that although first principles, such as the claim that retribution is intrinsically morally good, cannot be deductively proved, we can avail ourselves of evidence in their favor, and emotions may partly constitute such evidential support. As Moore notes, the idea is not new and was already present in J.S. Mill. Moore himself defends this view at length,Footnote ¹⁰ while a number of philosophersFootnote ¹¹ have afforded it further plausibility. Clearly a successful critique of this view would have repercussions on Moore's retributivist conclusions. Yet its critique would be an exercise in metaethics, a quick review of which does not indicate difficulties for this view that are any more insurmountable than those faced by its main contenders. What is more, an argument's critique is most effective when conducted as much as possible on its very terms. I suggest, then, that we grant this premise and focus on questions that are closer to the topic of guilt and punishment, which appear later in the argument.

Suppose, then, that some emotions play the epistemic role of supporting particular moral judgments. If only some emotions can play this role, however, the question we need to ask next must concern the criterion that tells emotions apart with respect to their epistemic credentials. This criterion is what Moore offers in step 2, where virtue and truth, on the one hand, and vice and falsehood, on the other, are connected. Moore himself is not entirely happy with this criterion. He finds counterexamples for the very case of guilt, whereby it would be virtuous to feel guilt in the case of moral dilemmas even though the judgment that one is guilty is false, and the inverse case, whereby correct moral judgments seem to spring from not especially virtuous emotions (e.g., selfish emotions and correct Lockean judgments about labor and property). Yet he concludes that though virtue (or lack of it) in the possession of an emotion is not an infallible guide, it is certainly a guide, albeit a fallible one, to the epistemic import such an emotion may possess.Footnote ¹²

The question now is whether “possessing” the emotion of guilt in particular is virtuous as Moore claims in step 3. Moore's argument goes as follows. Consider the case of Richard Herrin, a man who “intentionally smashed open the skull of a 23-year-old woman with a claw hammer while she was asleep, a woman [Bonnie] whose fatal defect was a desire to free herself from [Herrin's] clinging embrace.”Footnote ¹³ Then imagine what you would feel if you had acted like Herrin. As for Moore, he would himself “feel guilty unto death [and] couldn't imagine any suffering that could be imposed upon [him] that would be unfair because it exceeded what [he] deserved.”Footnote ¹⁴

Next Moore asks whether it would be virtuous to feel guilt in such circumstances and concludes that though “virtue” is perhaps an odd word to use in the context of extreme culpability, guilt seems at least very appropriate.Footnote ¹⁵ The shift from “virtue” to “appropriateness” is, as shown in Section IV, a crucial one.

At this point in the argument, the virtuous credentials of guilt have supposedly been established. The question, then, is: What exactly does this feeling validate? This is the topic broached by step 4. The crucial point here is the connection between guilt feelings and the judgment that one ought or deserves to be punished. Note that Moore specifically claims that the judgment here is to the effect that one ought to be punished rather than that one ought to repair the damage as best as one can. If only the latter were true, guilt feelings would at best justify corrective justice theories of punishment and not retributive ones. He argues, however, that that would be to put too nice a face on our guilt feelings:

Finally, with step 5 we move from the first-person judgment to the effect that in such circumstances we would deserve to be punished to the third-person judgment that Richard Herrin deserves to be punished. Here Moore argues that we should resist any temptation to withhold transferring the guilt to Herrin: doing so would be to count him as less of a person, as someone incapable of choice and responsibility. Such a move would be a form of elitism masquerading as egalitarianism.Footnote ¹⁷

If this and the other steps are indeed correct, guilt feelings do validate particular judgments to the effect that someone ought to be punished as in step 6. From there, and through abductive reasoning, we find that what these guilt-validated particular judgments have in common is the moral culpability of the person who deserves the punishment. We can then take such judgments to be evidence in favor of positive retributivism, that is, the (first) principle to the effect that moral culpability is sufficient to justify punishment (or again, that inflicting punishment on the morally culpable is an intrinsic good). With the general shape of this argument in mind, I offer a critique of step 4 in the next section and of step 3 in Section IV.

III. GUILT AND PUNISHMENT DESERVINGNESS

The first part of step 4 is a defense of the claim that guilt feelings typically cause the judgment that one is guilty in the sense that one is morally culpable. I understand this claim as follows:Footnote ¹⁸ when feeling guilt, we typically judge that we are morally responsible for a norm-violating act.Footnote ¹⁹ Though this claim is sometimes challenged,Footnote ²⁰ we side with Moore and, in the rest of the section, challenge the second part of step 4, that is, the alleged connection between the emotion of guilt and punishment deservingness. Clearly, if this connection cannot be established, the retributivist project will not take off. The question, then, is: Does guilt typically cause the judgment that one deserves to be punished? As shown below, the answer is “no.” In what follows, I first offer positive grounds in favor of this conclusion and then attempt to explain the origin of Moore's mistaken claim by looking at its Freudian and Nietzschean roots.

The claim that guilt involves the judgment that one ought to be punished (whether guilt partly consists in this judgment or typically causes this judgment) seems plainly false. Imagine any case of guilt less severe than the Herrin-the-axe-murderer case in which guilt would nonetheless be an appropriate emotion to feel. Suppose that you are in the context of an intimate relationship, the type of context, that is, in which, according to recent empirical research,Footnote ²¹ guilt is most likely to arise. You have said something quite offensive to your partner, something you did not really mean. She took it pretty badly and broke into tears. You think to yourself “Oh no, what have I done!” and generally feel bad about your behavior. You experience the situation as one in which you are responsible for harming someone you care about, and, clearly, avoiding harm to the people you love is something you care about. Suppose also that your emotional experience is accompanied by the disposition to make amends, maybe by way of an apology. Most of us think that a case such as this involves whatever is sufficient to be in the presence of guilt, with no mention of punishment made.Footnote ²²

That most of us actually think this way and hence that guilt does not typically involve (belief of) punishment deservingness is not simply my own private intuition but a fact that enjoys strong evidential support from empirical psychology.Footnote ²³ Perhaps, however, the evidence collected in these studies is not sufficient to discard Moore's view. It may very well be, the objection would go, that the experiments in question failed to mention punishment deservingness because their respective experimental designs were not aimed at finding out about punishment deservingness. This, however, is not the case. At least some of the methodologies employed in such studies, which include qualitative case-study analyses, content analyses of guilt narratives, participants' quantitative ratings of shame and guilt experiences, personality and emotional correlates, prototype “matching” analyses, and others,Footnote ²⁴ were open enough to have let punishment deservingness arise as a dimension if it were indeed an important dimension of guilt.Footnote ²⁵

What is more, a study of the empirical literature affords direct negative evidence against Moore's view or something akin to it. I would speculate that Moore's ideas about guilt and punishment deservingness originate in either Friedrich Nietzsche (who is extensively referred to by Moore)Footnote ²⁶ or Sigmund Freud, whose influential views associate guilt and punishment. In Civilization and Its Discontents, for example, Freud distinguishes fear of outer authority and fear of inner authority and argues that guilt is a kind of internal punishment due to one's internalization of figures giving rise to the superego mechanism.Footnote ²⁷ Nietzsche's conception of guilt is structurally the same: guilt is the symptom of bad conscience caused or motivated by a negative stance toward vital energies and values imposed by alien forces; guilt arises out of the progressive internalization and moralization of the fear of punishment attending those who fail to obey the socially defined dictates of morality. For Nietzsche, as for Freud, guilt is an arbitrary internal punishment.

Although there are important differences between Freud's and Nietzsche's accounts, the crucial thesis, whose assessment will be enough for our present purposes, is that moral, guilt-eliciting norms are internalized through fear of punishment. This is the substantial psychological assumption that motivates their criticism of guilt. But recent empirical findings on the issue of internalization as well as on that of the developmental path of shame and guiltFootnote ²⁸ show that it is fundamentally misguided. G. Kochanska and colleagues, for example, show that parental warmth and affection (rather than rejection and punishment) are positively linked to guilt development in the child.Footnote ²⁹ These findings raise serious difficulties for the view of guilt as the expectation of punishment. They also reported that love withdrawal is linked with high guilt. When parents instead respond by asserting their power, as in direct punishment, guilt is less extensive. Within this developmental literature, evidence has also been found for the claim that guilt does not originate from fear of punishment but from empathic concern with others.Footnote ³⁰ If, as the evidence suggests, this developmental story is correct, it is hard to see how punishment, as feared or as something deserved, should be part of the description of guilt.Footnote ³¹

Some may view the reliance of my argument on empirical support with suspicion. A large part of the argument, the objection would go, relies on the intuitions and experiences of nonexperts, that is, the subjects participating to the experiments crafted by various psychologists. Now, these intuitions, even if numerous, should not be allowed to override the intuitions or experiences of experts. Let us assume that this point is true. The question remains, however, whether there is reason to think that Moore's intuitions on this subject are any more authoritative than those of the average subject participating in the experiments cited above. However, if the claim that Moore's intuition derives from Freudian and Nietzschean theories is correct, there is no reason to think that they are any more reliable than those intuitions that are not corrupted by incorrect theoretical accounts of guilt.Footnote ³²

As a final objection, consider the following thought from Herbert Morris:

If Morris is correct, when feeling guilt we are sometimes disposed to self-punishment. There are two points to note in conjunction with this claim. First, this is once again a claim susceptible to empirical confirmation. If we consult the relevant literature, we find that guilt is strongly associated with reparative action tendencies. Those who experience guilt tend to want to repair the wrong, make amends, and apologize. Self-punishment as described in the last quote, then, might be a form of reparative behavior but is not the only form, nor is there evidence to the effect that it is the preponderant form. The connection between self-punishment and guilt, then, is not as strong as it ought to be to carry Moore's retributivist argument through. Second, even if the connection were very strong, the thesis advanced by Morris is that the action tendencies are reparative or restorative in nature.Footnote ³⁴ Those who are guilty and tend to seek punishment do so for reparative purposes and not simply because they think they deserve punishment. It would be more appropriate to take such action tendencies as the beginning of a justification for corrective rather than retributive justice. I take these considerations to be enough for blocking the retributivist argument in its fourth step. For those who are yet unconvinced, however, I now offer some grounds to reject step 3.

IV. GUILT AND VIRTUE

Having argued that the virtuous nature of an emotion is what qualifies its moral epistemic credentials, Moore assesses the virtuous nature of guilt by considering the case of axe murderer Richard Herrin and concludes that though it would perhaps be odd to use the word “virtue” in the context of extreme culpability, it would be very appropriate to feel guilt for anyone in the position of Richard Herrin. Note the shift from “virtue” to “appropriateness.” The normative assessment of emotions can focus on distinct dimensions. In particular, judgments about the virtue or the moral nature of any emotional occurrence or disposition are distinct from judgments about the fittingness or appropriateness of those emotions to their object. An emotion's formal object is what regulates fittingness or appropriateness judgments with regard to specific emotional occurrences.Footnote ³⁵ Hence, for example, fear is the appropriate emotional reaction to dangerous objects. It is a distinct task, however, to assess the moral or prudential nature of any emotional occurrence or disposition. Hence, though fear would clearly be fitting or appropriate when confronted with an angry bear, we can imagine circumstances in which it would be imprudent or even immoral to give in to one's appropriate fear.Footnote ³⁶ Similarly, a certain occurrence of envy may perfectly fit its object, and yet many would argue that envy is a systematically immoral emotion: in no case would feeling it add virtue.

Moore claims that it would be “virtuous” or perhaps “very appropriate” to feel guilt in the imagined circumstances. But which one is it? If the distinction above is correct, the claim that guilt is an appropriate reaction to a brutal homicide is not tantamount to the claim that feeling guilt “makes the person more virtuous.” Perhaps most of us internalize a norm against murder and would hence think of guilt as fitting its object in cases such as Richard Herrin's. What Moore needs to show, however, is that feeling guilt would also add virtue or be morally better than not feeling guilt, that is, that Herrin would be more virtuous or perhaps less vicious if, after the fact, he were guilt-ridden rather guiltless.

Friends of Moore's view could accept in principle that at the conceptual level the appropriateness and morality of an emotion come apart while still maintaining that at the substantive level the following thesis is correct:

Some will perhaps find this thesis intuitive. If you and I both committed a homicide and you felt guilty about it and I did not, in that respect you must be a morally better, more virtuous, or less vicious person than I am.Footnote ³⁷ Perhaps the idea that affords plausibility to this claim is that the feelings of guilt signal the presence of a disposition that we take to be morally good in a person's character. Substantive theses such as this one, however, are in need of an explanation (unless we intend to put them forward as intuitions incapable of any backing beside the very fact that they occur to us with the content they happen to carry).

A number of facts can serve as the basis for such explanation(s). On one important psychological account, guilt is understood as an interpersonal emotion that strengthens social bonds and attachments in many ways:

Roy Baumeister and colleagues support these claims about the interpersonal nature of guilt with empirical evidence. The central idea of the interpersonal analysis is that guilt in all relationships arises primarily from the communal component and is proportional to it.Footnote ³⁹ In particular, guilt serves three broad functions for relationships and strengthens social bonds and attachments in three ways. First, guilt motivates relationship-enhancing patterns of behavior by facilitating the enforcement of communal norms that prescribe mutual concern, respect, and positive treatment in the absence of self-interested return. Second, guilt operates as an interpersonal influence technique that allows even a relatively powerless person to get his or her way.Footnote ⁴⁰ Third, guilt redistributes emotional distress within the dyad. After a transgression, the victim is presumably suffering while the transgressor has benefited. If the transgressor feels guilty, however, his or her enjoyment is diminished, and the transgressor's guilt may make the victim feel better. In this way, emotional equity is restored (because bad feelings are in effect restored to the person who caused them).Footnote ⁴¹

What is more, by bringing the two partners' emotional states into closer harmony, guilt may be beneficial for the relationship because people who are in similar affective states tend to be able to have mutually satisfactory interactions with a positive outcome, whereas people in mismatched affective states tend to enter into a vicious cycle of poor communication, mutual dissatisfaction, and increasing withdrawal.Footnote ⁴²

Similarly, according to Morris though we may think that there is something suspicious about guilt insofar as it involves pain and suffering for what is over and done with, we do think it essential to human life insofar as it implicates “a concerned involvement with others.”Footnote ⁴³ A person without a conscience, dominated exclusively by pain and pleasure for himself alone, is not as we wish to be, and surely we would not be at ease with him around.

The fact that guilt strengthens attachments and social bonds by showing our concerned involvement with others is value-free as such. Yet we can easily imagine at least two moral accounts that could coherently place normative value on such facts. Many eudaimonist accounts of virtue, for example, recognize the inherently social nature of human beings and include the goods stemming from personal relations as parts or constituents of the virtuous or eudaimon life. The fact that wrongdoers feel guilt is a sign that standing in certain relations to others is something they care about, and for the eudaemonist, these relations are constitutive of the virtuous life.

Similarly, other types of account can use the support provided by these facts about guilt. Various types of consequentialism, for example, take on board the idea that strong social bonds are a constituent of or instrumental to well-being and take some positive function of each individual's well-being to constitute impartial well-being, the ultimate principle of practical reason. Occurrences of guilt would then be instrumental to the realization of strong social bonds and hence impartial well-being. Perhaps it is also better for most of us to live among beings who show “concerned involvement with others,” and if guilt is part of what it takes to show this concern, then greater impartial well-being may be realized when those for whom guilt is appropriate feel guilt. Finally, as claimed in the previous section, through its reparative action tendencies guilt fosters prosocial behavior, and once again, prosocial behavior may be instrumental to a number of prudential goods.

There are, then, a number of facts concerning guilt that are relevant to its moral status: it strengthens social bonds and attachments and is associated to prosocial behavior. As such, these facts do not talk about the moral goodness of guilt or the relative virtuous nature of guilty wrongdoers. What get us there are the normative theories that account for these facts, and in particular, theories such as eudaimonism and consequentialism. When accounted for by such theories and their normative premises, the facts relevant to guilt can be seen to acquire positive moral value. This conclusion may seem to constitute all the support needed to provide independent support to our initial intuition. Unfortunately, however, this is not so.

These facts and the normative theories that impart positive moral value to them should help explain why the guilt-ridden wrongdoer is morally better or less vicious than the guiltless wrongdoer. Note, however, that strictly speaking, the only difference between a guilt-ridden and a guiltless wrongdoer is the fact that the former but not the latter is subject to feelings such as pangs of conscience, the self-blame and remorse that go with guilt. In other respects, guiltless and guilt-ridden wrongdoers may be exactly the same. In particular, they may both judge that what they have done is wrong and regret having done so, may both be at least equally disposed to repair and amend as well as they can, and in general may be at least as determined as each other to do better in the future.

None of the empirical facts mentioned above excludes that in this sense of guiltlessness, guiltless wrongdoers may be as morally good as if not better than guilt-ridden ones. All the evidence shows is that guilt strengthens social bonds. This, however, is not tantamount to showing that guiltless people are less motivated to strengthen social bonds than guilt-ridden ones, nor that they are less effective in doing so. The evidence we would need for the latter claims would have to show that guilt-prone people (or those who are in the grip of an occurrence of guilt) score generally better with regard to these prosocial behaviors than those who are not guilt-prone (or not in the grip of guilt). All the evidence we have, however, is that those particularly prone to guilt are also particularly prone to shame and that shame-proneness has a number of unwelcome antisocial associations.Footnote ⁴⁴

V. EXTENDING THE ARGUMENT

In light of the above, we should agree that the emotion of guilt cannot validate particular moral judgments about punishment and hence retributivism because: (1) the connection between guilt and punishment deservingness has not been established; and (2) there is insufficient support for the claim that guilt is a virtue. Friends of retributivism, however, may still use the type of argument discussed above while appealing to emotions other than guilt. The question, then, is whether there are any emotions other than guilt that are epistemically useful in coming to particular judgments about punishment deservingness. Candidate emotions would have to “add” virtue while enjoying a relevant connection to punishment. At first sight, it would seem that anger in one or another of its forms (indignation, outrage, moral outrage, or hatred) is a plausible—if not the only plausible—candidate capable of replacing guilt in such a type of argument. As shown below, its connection to punishment is well established and has historically been thought to be a virtuous expression of the moral understanding of crime and its implications.Footnote ⁴⁵

There seems to be agreement about the connection between at least some forms of anger and punishment. Aristotle, for example, characterizes anger as an impulse, accompanied by pain, to a conspicuous revenge for a conspicuous slight directed without justification toward what concerns oneself or one's friends.Footnote ⁴⁶ More recently, Robert Roberts takes it that in anger we construe the situation in these terms:Footnote ⁴⁷ S has culpably offended in the important matter of X (action or omission) and is bad; I am in a moral position to condemn; S deserves (ought) to be hurt for X; may S be hurt for X. Unlike Moore's discussion of guilt and what beliefs it causes, these are claims about what anger is, that is, a specific motivation to inflict pain on an offender or a specific construal of a situation such that a culpable offender deserves to be hurt and one is motivated to inflict pain on him or her. These characterizations, however, should serve the retributivist's purposes quite well because they establish an evident connection to punishment: to say that someone deserves to be hurt for a culpable wrongdoing is to say that he or she deserves punishment. We henceforth grant retributivism with whatever parts of these claims it needs to get the required connection to punishment.Footnote ⁴⁸

Next we need to consider the virtuous nature of anger. The substantial thesis at issue here is as follows:

Perhaps some find this intuitive. But we should be particular wary of such intuitions, for as shown above, anger is intimately connected to the infliction of harm on others, and this affords at least some reason for not thinking it a virtuous emotion. One quick way of dismissing the substantial thesis would parallel the argument I propose for guilt in Section IV. Consider an individual who feels appropriate anger and another who, in the same situation, fails to do so. Is there any reason to think that the former is any more virtuous than the latter, provided that the two otherwise think the same thoughts and display the same action tendencies? What is it about the feeling of anger that makes its subject gain in virtue? Though I think this line of argument promising, in what follows, I set it aside an offer a new challenge.

In providing explanatory support for the substantial thesis about anger, we must keep in mind the following two points. First, in the context of an argument in favor of retributivism, any substantial explanation of why correct anger is a virtue cannot assume the truth of retributivism. Hence retributivist explanations such as “correct anger adds virtue because it provides incentives to give people their just deserts or to right the wrong” will not do. Second, as in the case of guilt above, we should expect such an explanation to account for the known empirical facts about anger.

The problem for the retributivist argument is that the best and most direct explanation of the known empirical facts about anger is itself retributivist and hence cannot appear in such argument on pain of begging the question. Strong empirical evidence shows that people punish primarily on the basis of retributive motivation; they want wrongdoers to get their just deserts.Footnote ⁴⁹ If, as granted above, in anger we construe the situation as justifying punishment and are motivated to inflict it and when punishing we do so primarily out of retributive motives, the motives involved by anger must themselves be primarily retributive. Any good explanation of the thesis that appropriate anger is virtuous must take this fact into account. Though perhaps a theory such as (indirect) consequentialism could explain in forward-looking terms why it is a good thing that agents display such retributive motives, the most direct explanation must itself be retributivist: correct anger adds virtue because it provides incentives to give people their just deserts or to right the wrong. But this explanation should not be allowed to appear in an argument aimed at establishing retributivism.