II. The Case for the Noninstrumental Goodness of Guilt: The Caring Argument

We find what I think is the most compelling argument for Good-Guilt in a variety of different sources that are united by an appeal to the idea that caring about the right things is manifested by guilt under the right circumstances. Consider the following account of the role of guilt by Hilary Bok:

And

Bok is not defending Desert here, or even Good-Guilt explicitly. But she provides the ingredients for an argument that is taken up more directly by Randolph Clarke:

The basic idea is that if we care about the right things—other people and doing right by them—then we will (typically) feel guilty when we’ve wronged them.^{Footnote 11} To feel guilt in such circumstances is a manifestation of caring, which is clearly a good thing, and the manifestation itself is a closely related good thing.

And Michael McKenna points to an instructive analogy between guilt and grief, arguing that while both are painful they are both essential manifestations of care, which is itself a noninstrumental good.^{Footnote 12}

While the theorists arguing for this connection between guilt and caring do not explicitly take up accounts of caring, they gain support from the most prominent among these accounts. Consider, for example, what I here call the Emotional Vulnerability Account of caring:^{Footnote 14}

The central idea in the reasoning about guilt is that if we care about the right things, then suffering as a response to harm that befalls the object of our care is simply a manifestation of such care; and, arguably, if we are the agent of said harm by way of our own blameworthy wrongdoing, guilt would seem to be a particular form of suffering that manifests caring of an important sort. As such a manifestation, it is itself noninstrumentally good.

It is notable here that there are several different objects of care mentioned in the relevant passages: another individual, a relationship, one’s own standards, justifying oneself to another. But in each case it seems that there is a worthy object of care, and guilt or a painful feeling of some similar sort is the manifestation of such care when one violates the relevant standards.^{Footnote 16}

Does it follow that actual suffering of guilt associated with caring is noninstrumentally good independent of the specific circumstances of wrongdoing? It appears that the answer is no, since the premise that it would be good to be disposed to X if one were in circumstances C does not entail that it would be good to X in circumstances C. But it might still be the case that in particular circumstances, such as when one has wronged another, the actual suffering of guilt is noninstrumentally good.^{Footnote 17} In other words, one might argue in the following way:

Does the Caring Argument succeed? The answer I will defend is that it either fails in an interesting way that could prompt us to revisit the nature of caring, or it succeeds, but too well for the ultimate aim of vindicating Desert-Guilt. But first, turn to the case against Good-Guilt.

III. The Case Against the Noninstrumental Goodness of Guilt

A natural starting point is to present candidates for counterexamples. Gilbert Harman presents himself as one such case, along with others whom he knows:

And he goes on to argue that

Now, we would need more detail to see the case as a counterexample to the emotional vulnerability account of caring as applied to guilt. (Harman says he is a moral person, but doesn’t refer to caring in particular.) Further, Harman’s own focus—at least in the quoted passage above—seems to be about instrumental good, so, again, we don’t have enough detail to assess the case as a potential direct counterexample to the noninstrumental goodness of guilt. Is some noninstrumental good lost if we could be as Harman claims himself to be? The implicit suggestion seems to be “no.” But again, it would be helpful to have more detail about the case.

Consider, then, a thought experiment, which gives us the ability to stipulate some of the conditions that are left incomplete in Harman’s self-description. Compare our world to a world in which otherwise very similar creatures (call them the “No Guilt Creatures”) do not feel guilt, but possess a syndrome of dispositions, desires, and cares similar to ours. So, when they freely and culpably act wrongly, they wish that they had not acted in that way, resolve not to act that way in the future, invest time and effort working on making the resolution meaningful, and so on. Is this world better than ours? Worse? Equally good? Before we can answer these questions, we need to make sure the thought experiment is coherent.^{Footnote 20}

We could have a number of different reactions to the thought experiment’s claim to coherence:

1. (1) The thought experiment isn’t coherent. If you don’t feel guilt or suffer in some closely related way in such circumstances, it isn’t true that you care in the right way.

2. (2) The thought experiment is coherent, but couldn’t be true of (most) human beings. So the disposition to guilt and suffering is necessary after all for the right kind of caring insofar as we are human. (Or a variant: it is coherent if we replace “caring” with “creature-caring” and take it that creature-caring is no less valuable than caring.)

3. (3) The thought experiment is coherent, but couldn’t be true of human beings as we are, with the histories we have, and so on. So the disposition to guilt and suffering is necessary after all for the right kind of caring insofar as we are human beings with our actual histories.^{Footnote 21}

4. (4) The thought experiment is perfectly coherent and humans could shed their dispositions to guilt without shedding the related dispositions.

If either (2), (3), or (4) is correct, then it is plausible that the hypothetical world is better for including less suffering, and retaining the goods that the suffering of guilt was initially supposed to be essential for. If (2) or (3), then at least guilt would seem to be a perhaps unfortunate aspect of being human or of being human in our circumstances, but not essential to what is good about caring about the victims of one’s wrongdoing, or about doing the right thing. I confess that in some moments I think that the thought experiment is coherent, and in some not. If it is, then it suggests a revisiting of the Emotional Vulnerability Account of caring described earlier. In that case, emotional vulnerability to painful experiences is merely characteristic of, but neither constitutive of nor necessary to, caring.

Consistent with any of (2), (3), or (4) is the idea that guilt is excellent evidence of caring about the right things. In fact, it is such good evidence that we might confuse it with an essential manifestation of it. And this might explain why (1) might appear true. I believe that the thought experiment, if coherent, provides support for the idea that the disposition to guilt is not itself an essential aspect of caring about the right things as it seems that the argument requires. And it goes some way to undermining the idea that guilt is noninstrumentally good.

It does not go all the way because the idea that the world of the No Guilt Creatures is a better world than ours, all things considered, is consistent with the idea that it is nevertheless missing a noninstrumental good that we have in our world. Perhaps guilt is noninstrumentally good in one way, even though overall the world in which guilt does not exist is a better one. At the same time, even if this were the case, the significance of Good-Guilt would be reduced. The reason is that any claimed noninstrumental goodness of guilt would seem to be outweighed by the overall badness of guilt, which it presumably possesses in virtue of its painfulness. Thus, even if we recognize that something good is missing in the No Guilt Creatures world, as long as we find their world better overall, we must concede that there can be no net positive amount of goodness added by promoting guilt feelings to a world that was otherwise the same as ours in all relevant respects, including caring and manifestations of it in action.

In the end, while I take there to be good reason to reject Good-Guilt on the strength of this thought experiment, I won’t assume that it is false in the next two sections. The question for us now is whether it follows from Good-Guilt that we have thereby supported Desert-Guilt. Not necessarily. I turn to this question in the next section. And here I think the answer is “no.” Of course, Good-Guilt is not intended as an analysis of Desert-Guilt in the first place. But an axiological claim of this sort is often used to explicate a claim about desert, is sometimes itself taken to help us understand a “desert thesis,” and is thought to support something like Desert.^{Footnote 22} In the next section, I will argue that Good-Guilt—particularly when supported (exclusively) by the Caring Argument—not only does not explicate Desert-Guilt; it goes some way to undermining it.

IV. Goodness, Fittingness, and Desert: An Argument Against Good-Guilt as a Defense of Desert-Guilt

Suppose, then, that emotional vulnerability is essential to caring. Even in that case, it is essential to note that the argument that guilt is a noninstrumental good depends ultimately on the fact that it is related (noninstrumentally) to the good of caring. But guilt is good, then, not because it is painful or negatively valenced, but rather because of its connection to caring about the right things. The fact that guilt is painful or negatively valenced does not then explain why it is on the candidate list of things deserved in virtue of one’s being blameworthy. And this casts some doubt on the idea that guilt is deserved even if there is something good about it.

Analogously, grief is good insofar as it is a manifestation of caring, but not in virtue of its being painful. And the problem for the project of supporting Desert-Guilt by means of the Caring Argument for Good-Guilt is that the value of guilt is too much like that of grief in this case. If its value does not lie in the blameworthy getting something negative in response to wrongdoing, but rather in its connection to caring, then it seems not to capture the idea of desert, as opposed to a more general kind of fittingness that applies equally to grief and guilt. To see this in another way, we wouldn’t typically say that grief is deserved, and though I don’t want to put too much stock in ordinary ways of talking, it seems that the concept of desert doesn’t apply to grief in this case. Derk Pereboom argues that a painful kind of regret can be appropriate as a response to harming others, but that this, too, is not deserved.^{Footnote 23} Perhaps the same can be said of guilt even in response to culpable wrongdoing. We might say, then, that Good-Guilt better supports

Now I think that the defender of Desert-Guilt has three kinds of response. The first is to deny the claim that when it comes to guilt—and grief for that matter—it is possible to separate out the painfulness and the relationship to caring. The second is to find a way of distinguishing guilt from grief so that the former but not the latter is deserved in a robust way that sets it aside from mere fittingness. (Another way to put this is to say that desert is a special kind of fittingness that applies to guilt but not to grief.) The third kind of response is more concessionary, and it is to say that even if guilt is merely a fitting response to blameworthiness, fittingness can do the important work of desert.

The first response denies the starting point of the reasoning, and suggests that there is really no sense to be made of the claim that guilt is good in virtue of its connection to caring, while it is not good in virtue of its painfulness. McKenna, for example, argues that both guilt and grief are best thought of as “an organic unity whose whole has a value that is not analytically decomposable . . .”^{Footnote 24} There is something appealing in this idea. But I think I can make sense of the idea that even in cases in which painfulness is a constitutive part of an experience that is noninstrumentally good, it is not so in virtue of the painfulness. Take, for example, an empathetic response. It may be a noninstrumentally good thing for a doctor to have an empathetic response to the pain of her patient, but its goodness is arguably not in virtue of its painfulness, but in virtue of her richer understanding of his experience and as a manifestation of her willingness to treat him as a person and not as a symptom. Further, even if this turns out to be the wrong way to think about these things, and one cannot separate out the painfulness and the other features that make the experience good in the case of guilt, it is still the case that the guilt seems structurally quite similar to grief on the assumptions of the Caring Argument. Both are good insofar as they are manifestations of caring about the right things. So there remains a question of what makes guilt deserved while grief is fitting, but not deserved. Thus, the proponent of the first response might in the end need to appeal to additional resources that can be found in either the second or third response.

The second response is to distinguish guilt from grief in a relevant way. A natural first idea is to say that while it is not the case that grief is good in virtue of its being painful (but perhaps despite it), guilt is good in virtue of its being painful. But this is precisely what the Caring Argument places in doubt. If the good of guilt is really its connection to caring, then it appears to be in exactly the same situation as grief.

An alternative attempt to distinguish guilt from grief is to say, as Andreas Carlsson does, that guilt is “morally good” in contrast to grief, in that it gives others reasons to act so as to induce it, while grief does not.^{Footnote 25} And it might be argued that what I have called Reason-Guilt has conceptual priority, and itself supports Good-Guilt.^{Footnote 26} If Reason-Guilt is true, and there is no analogue of Reason-Guilt for grief, then this would be a way of distinguishing between guilt and grief. It is worth noting that to do so, however, would be to offer a very different route to the defense of Good-Guilt than that of the Caring Argument. Still, it will be important to take up the question of whether Reason-Guilt is true, and I will turn to this task in the next section.

Before doing that, it is important to consider the more concessionary response to the challenge that guilt and grief are too similar in respect of their noninstrumental goodness for Good-Guilt to support Desert-Guilt. According to that response, we don’t need a special kind of fittingness to do the work in question; we just need it that guilt is fitting. Fittingness is sufficient to give us reason to treat people in guilt-inducing ways, say, and to still allow for a non-skeptical position about blameworthiness if it is true that it is (sometimes) fitting to induce guilt. Things get quite subtle here, and I will not be able to do this issue justice. I will just make one observation now, and leave somewhat open how much, if anything, is lost by moving from Desert-Guilt to Fittingness-Guilt.

There is what might seem to be a reason to resist the concession to abandon Desert-Guilt in favor of Fittingness-Guilt. It seems even less plausible that mere fittingness by itself yields reasons for action than it does that desert yields such reasons. As I have argued elsewhere, fittingness alone does not yield such reasons.^{Footnote 27} For example, the fact that a racist joke is funny seems to entail a certain fittingness of laughing. But such fittingness needn’t provide any reason at all, it seems to me—not even a pro tanto one that is overridden. Perhaps one could claim that it is not the overt response, namely, laughing, that is fitting, but the emotion or feeling of amusement, and, thus, that one has reason to feel amusement insofar as it contains a true appraisal of the situation. I am not sure whether this is correct, for it might be that one simply has reason to judge the joke funny without having reason to feel amusement. But even if it were correct, this move would sever reasons for feeling from the reasons for acting in ways the feeling disposes one to act. This seems a potentially unstable position.

Perhaps another case is an even clearer one for casting some doubt on Fittingness-Guilt. It might be fitting to respond to something with surprise, but it isn’t clear what it would mean for one to have reason to bring about one’s own response of surprise. For this reason, it might seem that a special kind of fittingness is important in order to obtain a direct connection to reasons.

To sum up this section, then, we have arrived at the conclusion that feelings of guilt might be fitting, and even noninstrumentally good, but the very argument that we have seen support its noninstrumental goodness either fails to support, or counts against, the claim that it is deserved in a sense distinct from fittingness.

V. Desert, Goodness, and Reasons

Now let us turn briefly to Reason-Guilt. It is important in its own right, but also, because, as we saw earlier, at least some argue that Reason-Guilt is the more fundamental claim, which in turn supports Good-Guilt. This seems to me even less intuitively plausible than Good-Guilt, particularly if we are not able to appeal to Good-Guilt in order to support it. A thought experiment might best bring this out.

Imagine that you have a special power (call it “the Look”). By looking at another person in the right way, you can bring about feelings of guilt. The other person culpably wrongs another—it is not a trivial offense, but neither is it the worst possible. Imagine that she betrays the confidence of a friend and as a result the friend has a bad day. You now have the chance, by looking at the offender in that way you have mastered so well, to bring about guilt feelings in her. It would be the easiest thing, requiring nothing in the way of effort or sacrifice. But now also imagine that there is no further good to come from your exercising this power you have. The offender has already resolved not to do the same sort of thing again, no one else is around to experience the results, the relationship is either already irreparably damaged no matter what, or all has been forgiven, so that there is no benefit to the relationship to be had, and so on. Would you be making a mistake, or leaving a reason on the table, so to speak, by taking a pass on inducing this painful feeling? Would you have failed to promote a (noninstrumental) moral good that would have been costless to achieve?

I do not have the intuition here that you would be making this kind of mistake. That suggests to me that there is not a pro tanto reason to induce guilt that stems from blameworthy action alone. For by hypothesis, there were no overriding reasons in favor of passing (no cost, no effort, and so on). So, if there is indeed such a pro tanto reason (and of a sort that involves justice and that isn’t overridden), it seems that you would be making a kind of mistake. Thus, I tentatively conclude that Reason-Guilt is false.^{Footnote 28}

Of course, in many actual cases, guilt plays a valuable instrumental role (and we can see this from the human development literature). And, again, guilt is an excellent piece of evidence for the kind of genuine resolve I stipulated in the thought experiment. But the idea behind the thought experiment is to strip away the existence of typical instrumental goods and to provide alternative sorts of evidence by stipulation.

Now one might object to the use of the case to draw a general conclusion about reasons. One might do so by appealing to a distinction between two roles for reasons: requiring and justifying.^{Footnote 29} According to Joshua Gert, a reason requires the performance of an action when it would be irrational not to perform it. In contrast, a reason justifies an action by making it not irrational not to perform it. In turn, corresponding to each role is a distinct measure of strength: justifying strength and requiring strength. As Gert elaborates, given two reasons, R1 and R2, R1 has more requiring strength if and only if R1 would make it irrational to do anything R2 would make it irrational to do, and R1 would make some things irrational to do that R2 would not.^{Footnote 30} On the other hand, given two reasons R1 and R2, R1 has more justifying strength if and only if R1 would make it rationally permissible to do anything that R2 would make it rationally permissible to do, and R1 would make some things rationally permissible to do that R2 would not make it rationally permissible to do. An illustration will help bring out the distinction. A reason might have high justifying strength without having high requiring strength. To set out an adapted example of Gert’s, the fact that you could save forty children from a burning orphanage might have high justifying strength (justifying even your risking your life to do so), while it might have relatively lower requiring strength, so that you are not rationally required to do it, even if the cost would be substantially lower than risking your life, such as risking an index finger.^{Footnote 31} Applying the distinction to “the Look,” one might say that while there is a reason with justifying strength for giving the Look, giving the Look lacks a reason of requiring strength and this fact entails that there is nothing irrational about not giving the Look even though the opportunity to give the Look is reason-giving in the sense of offering justification.

While the proposed distinction is an interesting one, I think that the objection can be answered without engaging in an evaluation of the view according to which there are two kinds of strength that come apart. For it is notable that even if dimensions of requiring and justifying come apart in cases such as the orphanage, nothing is implied about the special case in which one has (by stipulation) only one reason on one side and none at all on the other. In this case, to claim that there is nothing irrational about the choice not to give the Look, despite the existence of a reason with justifying strength on the side of giving the Look, the following claim would also have to be true: the requiring strength of the opportunity to give the Look is not simply small, but, rather, it is zero; that is, it plays absolutely no role in the calculation of what one ought to do. The claim that there is justifying but not requiring strength in the reason at hand is doubtful, however, and, were it true that giving the Look has zero requiring strength, this fact would considerably weaken Reason-Guilt. To explain: first, we would have to have reason to think that the requiring strength of reasons can ever be zero, let alone zero in this case. In the motivating examples, the requiring strength is non-zero. If one could save the forty orphans at no cost at all, for example, it is hard to see that one would not be rationally required to do so.^{Footnote 32} It is plausible that if there is no reason against a course of action, and taking the action is supported by reasons, one would indeed be irrational in not taking that action. At the least, without a rationale for thinking that the requiring strength of reasons can be zero, it seems that there is no reason to doubt the lesson of this thought experiment.

A second important concern about deploying this strategy is that even if the claim could be defended, it would considerably weaken Reason-Guilt insofar as it would no longer offer guidance about what we ought to do. While it is true that Reason-Guilt could still play a justifying role in our treatment of others on this view, it could not play a guiding role of any kind, and this seems not only to countenance an odd (if not incoherent) kind of reason, but also to detract from at least one significant aspect of the thesis.

One additional consideration might be thought to allow the deployment of the distinction between justifying and requiring reasons without committing to a null requiring strength for giving the Look. It is to point to the fact that guilt itself is painful, and so there is, contrary to the explicit supposition of the thought experiment, a reason against giving the Look. Insofar as the experience of guilt is painful and all suffering is bad, the experience of guilt is itself bad and offers reason not to bring it about; insofar as it is a manifestation of caring, it is good and offers reason to bring it about. There is certainly something to the idea that suffering is, in general, bad. But it is important that for the thought experiment to allow for rational permission to pass on giving the Look, these reasons—the suffering of guilt and the rightness of guilt—would seem to have to precisely cancel each other out on the dimension of requiring strength. And if they always cancel each other out, while the painfulness of guilt is inextricable from guilt, then bringing about guilt in wrongdoers would never be net-reason requiring. Thus, this move does not ultimately help in undermining the conclusion from the thought experiment that bringing about guilt is not reason-giving in a noninstrumental way.

To this point, I have offered a negative argument, suggesting that Reason-Guilt is false, and that there is not even a pro tanto reason to promote the experience of guilt in the blameworthy simply because they are blameworthy. Here I want to highlight one important implication of this conclusion, and one important limitation of it. First, the implication. If we accept the general thesis that the noninstrumental goodness of a state of affairs itself provides a pro tanto reason to promote the experience of guilt, then the absence of such a pro tanto reason would entail the lack of (net) noninstrumental goodness of such an experience. In that case, the thought experiment would serve to cast doubt on the claim of noninstrumental goodness (or at least the idea that there is a net amount of such goodness), independently of the details of positive arguments, such as the Caring Argument, in favor of Good-Guilt. Now the general entailment between goodness and pro tanto reasons might be questioned. And it is important to note that Good-Guilt does not include an explicit claim to net-positive goodness; so, in principle, one could still accept the Caring Argument with its conclusion of Good-Guilt while rejecting the idea that guilt ever provides a net-positive good. Still, it seems that at least some of the force of Good-Guilt would be undermined if this were the case. Thus, the conclusion of the argument based on the Look is significant, casting doubt on the idea that there can ever be a net noninstrumental good in experiencing guilt, as well as on Reason-Guilt.

Now for the limitation of the argument based on the Look. It does not follow that there is no connection at all between desert—even desert of guilt—and reasons. In fact, it is plausible that desert offers a certain kind of conditional reason for bringing it about that someone gets what she deserves. Under certain conditions, the fact that one deserves something on account of one’s blameworthy wrongdoing can make it the case that it would be better than the alternatives for one to get what one deserves. And under certain conditions, it can also give one, as well as others, reason to bring about that state of affairs.^{Footnote 33} Let us abstract for a moment from the idea that what the blameworthy deserve is to feel guilty. The basic idea can be captured as follows:

Suppose, for example, that you are in a position in which you have no choice but to promote someone’s being harmed, and you can harm someone deserving of it or harm someone else. This may give you a reason to promote the deserving person getting what she deserves. In such circumstances, it is not unfair to impose harm on the deserving whereas it would be unfair to do so to the undeserving. Nevertheless, in other circumstances, the fact that someone is deserving may provide no reason at all. Thus, there is a systematic connection between desert and liability to harm, even if the connection is not in place in all circumstances in which one is deserving of harm.^{Footnote 34}

Now let us offer a specific version of the thesis in terms of guilt as what is deserved:

Conditional Reason-Guilt is consistent with the rejection of both Good-Guilt and Reason-Guilt. Importantly, if it is true, it does not gain support from Good-Guilt (or the Caring Argument in favor of it). It is also not clear that it is more plausible than the more general Conditional Reason-Desert thesis, and a full defense would require a special reason for restricting the thesis to a systematic relationship between blameworthiness and liability to the promotion of guilt in particular. But it is not undermined by the argument based on the Look, and in my view remains a thesis well worth further consideration.

VI. Conclusion

In this essay, I have explored two particular claims that have been thought to explicate and support Desert-Guilt. I have presented some arguments against Good-Guilt that seem to me to be strong; but I have also conceded that it might be true, while maintaining that even in that case the best argument for Good-Guilt—the Caring Argument—is one that undermines its ability to support Desert-Guilt in particular. Some have argued instead that Desert-Guilt is best supported by Reason-Guilt, but it seems to me that this gets things the wrong way around at best, and in any case I have offered some reason to reject Reason-Guilt, at least as a fundamental principle. At the same time, I have argued that for all of the concerns about Good-Guilt and Reason-Guilt, Conditional Reason-Guilt could still be true and revealing of an important aspect of desert. Finally, there is no doubt much more to say about these theses, their relation to Desert-Guilt, and about the more general claim of Desert which is itself untouched by anything I have argued here.