It is uncontroversial that the blameworthiness of agents for what they have done or failed to do is importantly connected to our practices of blaming—a connection that in one way or another is reflected in criminal punishment, and perhaps in other institutions of liability and responsibility as well. It is far from clear, however, how the relationship between blameworthiness and blame emotions like resentment and indignation is to be spelled out. One natural idea—what we can think of as the standard view—is that the fact that someone is blameworthy for what she has done or failed to do is a reason or a ground that explains why blaming her or otherwise holding her to account would be justified.

However natural the standard view may be, it is also controversial. Many philosophers resist the idea that the aptness of blame is to be explained by blameworthiness in this way. Instead, they reverse the order of explanation and argue that a person's level of blameworthiness for an act or omissionFootnote ¹ is determined by the blame responses that it would be correct (in some sense to be specified) to have toward that person for this act or omission. Taking this view of blameworthiness shifts the focus to two other concepts that require analysis: one is the relevant notion of a blame response, and the other is what it is for such a response to be correct. Within these accounts, a blame response is understood in terms of the so-called reactive attitudes or emotionsFootnote ²—guilt, resentment, and indignation. Because of the prominent role that the reactive emotions or attitudes play in such accounts, we refer to them as “reactive emotion accounts”—or, equivalently, “reactive attitude accounts.”Footnote ³

Views of this sort are becoming increasingly popular. Allan Gibbard, for instance, argues that “[t]o say that he would be to blame is to say that it would be rational for him to feel guilty and for others to resent him.”Footnote ⁴ Jay Wallace has defended the view that one is morally responsible for some action just in case it would be appropriate to hold that person responsible for it, where “holding responsible” is understood in terms of the reactive emotions.Footnote ⁵ While others have endorsed such views as well,Footnote ⁶ these accounts all can claim as authority P.F. Strawson's seminal paper Freedom and Resentment. Strawson is widely credited as the originator of the reactive emotion accounts,Footnote ⁷ although, given the sweeping nature of his discussion, it is controversial whether he actually advanced any particular version of the view. In the end, whether Strawson advanced a version of the reactive emotion account of blameworthiness is of little importance. His work remains the inspiration for the vast majority of the accounts of interest to us here.

The differences between the standard view and a Strawsonian reactive attitude account are subtle, but not small. On the standard view, one is blameworthy for an act or omission in virtue of certain facts about one's agential relationship to the doing or omitting—for example, the fact that it was the product of a defective character, wicked intentions, a bad will, or some other kind of moral failing of the agent. One's blameworthiness in a sense supervenes on such facts, and so the project is to identify what facts about persons and their actions ground blameworthiness. In contrast, on reactive attitude accounts, one is blameworthy in virtue of facts about the responses that it would be appropriate for others to have toward one. The burden of such accounts is to identify the conditions under which blaming responses are appropriate or correct. So on the former view, blameworthiness is what explains the justifiability or rationality of the blame responses, while on the reactive attitude account, blameworthiness is to be explained by the blame responses being correctly held, felt, or otherwise directed toward one.Footnote ⁸

What might lead one to prefer a reactive emotion account over the standard view? For one thing, reactive emotion accounts promise to explain a relatively mysterious normative concept—blameworthiness—in terms of something of which we have a firmer grasp: namely, feelings of guilt, resentment, and indignation, all of which are familiar features of our lives. As well-socialized adults, we are likely to have a clear sense of the conditions under which our blaming practices permit responses like resentment and indignation, and so it might seem a promising strategy to appeal to them in order to understand the arguably more obscure theoretical notion of blameworthiness.

Second, one might hold that “blameworthy” is like other normative predicates—for example, “valuable” or “desirable”—in being apt for a so-called “buck-passing” analysis.Footnote ⁹ According to buck-passing accounts, to say that something is valuable is just to say that placing a certain kind of value on it is appropriate. Similarly, to say that something is desirable is just to say that desiring it is correct (apt or fit). On these accounts, in saying that something is valuable or desirable, one is saying only that valuing and desiring is an apt, fitting, or more generally correct response to have toward it. It is, in effect, a way of “passing the buck” to the underlying reasons that exist for valuing or desiring while not itself a reason for doing either.

Reactive attitude accounts treat blameworthiness analogously. As we saw, they hold that saying that a person is blameworthy just is a way of expressing the judgment that there are reasons for blaming him—that is, the judgment that blaming him is correct or warranted by reasons. Accordingly, for reasons of theoretical unity, those who are drawn to a buck-passing analysis of value and desirableness might be inclined to adopt the analogous view of blameworthiness, which is no more (and no less) than what the reactive attitude account amounts to. (Of course, one does not have to believe that desirableness and value are apt for a buck-passer account in order to believe that blameworthiness is.)

We now have in place a characterization of the reactive attitude accounts of blameworthiness, a rough idea of some of the important ways in which they differ from the standard view, and some of the reasons why those who pursue reactive attitude accounts are inclined to do so. Interestingly, whereas such accounts are much in vogue among philosophers, they have gone largely unnoticed by legal theorists—even those who are primarily interested in the moral foundations of the institutions of criminal punishment. This is unusual given the strong interest among theorists of the criminal law more generally in exploring the implications of philosophical disputes for criminal law theory—as evidenced by legal theorists’ interest in what philosophers have had to say about concepts such as intention, control, voluntariness, and the very idea of action, to name but a few. Reactive attitude accounts are naturally of concern to the theory of criminal law in that they promise to illuminate the concept of the blameworthy that lies at the heart of retributivist theories of punishment. Indeed, part of the attraction of such views to those who hold them is that punishment is an institutional expression of aptly held blaming attitudes—attitudes that aptly target the blameworthy and not others.Footnote ¹⁰

Though there is much to recommend reactive attitude accounts of blameworthiness, we remain unpersuaded that blameworthiness is a matter of correctly held or felt reactive emotions.Footnote ¹¹ Several objections to reactive attitude accounts are familiar and the focus of considerable discussion among philosophers. Our focus, however, will be on a distinctive objection that has gone virtually unnoticed, having to do with the ways in which the passage of time affects both the moral significance of what we have done and the correctness of our responses to those doings.

We suggest that the passage of time drives a wedge between blameworthiness and the reactive attitudes in a way that both calls into question the viability of the reactive emotion account and opens up new avenues in legal theory generally. In particular, the considerations we draw attention to provide a novel way of thinking about the appropriateness of statutes of limitations, that is, constraints on the period of time in which the state can legitimately prosecute wrongdoers and private parties may press claims against those who have wronged them.

The conventional view is that these time-based limitations on pressing legal claims are justified on epistemic or instrumental grounds. A common thought is that because the passage of time can lead physical proof to decay, witnesses to become unavailable and evidence generally to become less reliable, it can become inappropriate to prosecute certain criminal offenses or to press certain civil claims after enough time has elapsed.Footnote ¹² Moreover, the practical reality is that states do not have unlimited resources to devote to punishing wrongs or settling disputes from long ago. To be sure, these epistemic and pragmatic considerations help explain why there are limits on the state's power to press criminal charges and on its willingness to permit others to press civil claims. Considerations of these sorts do not suggest that it would be wrong or inappropriate to press such claims; only that it would be too difficult or wasteful to do so. In what follows, however, we argue that the passage of time can diminish the appropriateness of having or expressing the reactive attitudes (even if it does not similarly impact the blameworthiness of agents for what they have done). Thus there is an argument for statutes of limitations that grounds them in our blaming practices and goes far beyond the epistemic and instrumental considerations that are generally thought to underwrite such limitations.

We begin, however, by setting forth our basic objection to the reactive emotion account of blameworthiness: namely, that the appropriateness of the blaming emotions is sensitive to the passage of time in ways that the blameworthiness of agents for what they have done is not. If we are right, any effort to analyze the latter in terms of the former will face problems that have to this point gone largely unnoticed.

I.

If one is blameworthy now for performing some morally bad action, then one will generally still be just as blameworthy for having performed that action at any later point in one's life. Time, in other words, does not appear to erode one's culpability for what one has done or failed to do.Footnote ¹³

Of course, it is a good question whether one can diminish one's blameworthiness for a particular action by apologizing for it, atoning, making things right, and so on. It is quite natural to think that I would be less blameworthy today for what I did several years ago if I have sincerely apologized for it, undertaken actions to make it right and so on. However, others might object that apology and atonement do not diminish one's blameworthiness for particular actions but instead only reduce one's net level of blameworthiness (e.g., by balancing one's blameworthy act against another act that improves one's overall moral condition).

We need not settle the matter here. For whichever view one takes on the effect that apology, atonement, and redress have on one's blameworthiness for what one has done, it is clear that in the absence of taking any such steps, one would be as blameworthy today for what one has done wrong as one was when that misdeed was originally performed. Nothing would have changed in the meantime that diminishes one's culpability or blameworthiness. Thus the mere passage of time—the mere unfolding of one's life as time passes—does not alter the extent of one's blameworthiness for a particular action or omission.

Matters are different for the reactive attitudes, however. While it may be entirely appropriate today, say, to resent someone for a particular misdeed, it will not necessarily be appropriate to resent that person for that misdeed five, ten, or fifty years on; and even when it is still appropriate to feel some resentment or indignation, the extent or depth of the emotions is often appropriately diminished as time goes by—quite apart from the steps the wrongdoer has or has not taken to atone for his mischief.

Thus, if blameworthiness is not affected by the passage of time, while the appropriateness of the reactive emotions intuitively is, then considerations about the passage of time drive a wedge between blameworthiness and appropriately felt blaming attitudes, which threatens to undermine any attempt to analyze the former in terms of the latter.

These points are illustrated by the following simple example. Suppose that Megan, a ten-year-old girl, is playing with a ball in her front yard one day, when suddenly the ball rolls over into the flowerbed belonging to the neighbor, Mr. Anderson. No damage is done, but Mr. Anderson rushes over to the girl and lashes out at her, berating her excessively and inappropriately for what was only a harmless error. Megan leaves distraught and in tears. Anderson's action is uncalled-for and warrants the full panoply of the reactive sentiments. Megan would be justified in resenting him; observers would be justified in feeling indignation; and Anderson himself, upon even modest reflection, should be pained by suitable feelings of guilt, even remorse.

Thirty years on, however, (and probably long before then) we should expect matters to have changed substantially. Even supposing that Anderson never apologized, atoned, or sought to make matters right with Megan, it would nevertheless be unreasonable for Megan, now forty years old and (let us imagine) with a family of her own, to feel anything like the same resentment toward him that would have been appropriate when the wounds were still fresh. By this time it is reasonable to expect that Megan would have gotten over, or largely over, what in her youth was no doubt a painful and possibly even traumatic experience.

The crucial point here is not that Megan is likely to be better off this way. Presumably she will be, but whether she is or not plays no role in our argument (for as we explain below, we are prepared to allow that considerations of prudence or personal health and well-being are not reasons that properly bear on the appropriateness of emotions like resentment).Footnote ¹⁴ Rather, our central claim is that it would be inappropriate for Megan to harbor the same resentments she did thirty years ago, because there is something perverse, unreasonable, or confused about dwelling on the past in this way. One would somehow fail to understand the place of the reactive attitudes in our lives were one to dwell on certain forms of wrongdoing in this way (though not all wrongdoing, of course).Footnote ¹⁵ After all, this kind of childhood experience is something we generally think one should be able to move beyond and not remain hung up on indefinitely—however unpleasant, even traumatic, it might have been at the time. As memories fade and life proceeds, it is appropriate—at least for many sorts of wrongs—for resentment to diminish as well.Footnote ¹⁶

Emotional responses, therefore, seem properly to diminish in intensity over time in many different kinds of cases. It is natural and correct that they do so. Resentment of a certain intensity, i, felt by Megan toward Mr. Anderson right after he mistreated her seems to be inherently called for in a way that an episode of resentment of the same intensity felt by her toward him thirty years after the fact would not be.

To deny that time can make a difference to the appropriateness of episodes of reactive emotion in this way would be to insist on a view of what it is for these emotions to be apt that bears little resemblance to the way in which blame figures in our moral and social lives. We do not generally regard those whose levels of resentment or indignation toward wrongdoers diminish as time passes as being guilty of any conceptual confusion about the relationship between blameworthiness and resentment and indignation. Nor are we likely to conclude that such people are overly lax in their moral responses. That is, we do not regard people who cease to resent or feel indignant toward wrongdoers as time passes as letting these wrongdoers off the hook too easily—as if they were mistakenly foregoing some additional amount of resentment or indignation that they would have been permitted to feel, but which they are passing up merely for practical or extraneous reasons. Thus our experience seems to make room for the possibility that the blame emotions can become less appropriate as time goes on—and not because of reasons of prudence or fairness, or for other intuitively collateral considerations.

The level of reactive emotion it is appropriate to feel in response to a particular action can diminish merely because of the passage of time, whereas the degree to which wrongdoers are blameworthy for a misdeed cannot diminish (at least not in the absence of the blameworthy person taking steps to atone or otherwise make matters right). This difference is reflected in the idioms of common parlance. We are prone to say that “time heals all wounds” but say neither that “time renders a culpable person innocent” nor that “time rights wrongs.”

II.

In emphasizing the way in which the passage of time impacts the appropriateness of the reactive attitudes, we should not be read as claiming that there is something magical about time as such. Rather, our point is that as time passes, it is inevitable that events will continue to unfold, and in some instances these events will bear on the appropriateness of the reactive emotions while not bearing on the agent's blameworthiness. For instance, it is not the passage of time in itself that explains the diminishing appropriateness of resentment and indignation toward Mr. Anderson for his mistreatment of Megan. Rather, what explains it are the things that happen as time passes: memories fade; Megan becomes an adult and grows up both mentally and emotionally; she moves away from the neighborhood; she takes on the responsibilities of parenthood herself; and so on.

If one's degree of culpability or blameworthiness for a particular action can be diminished at all, this could only be in virtue of certain distinctive events—e.g., atonement, apology as well as other ways of making things right, or a breach in personal identity (“I am literally not the same person now as before”). And while the events that represent the normal course of life as time marches on can bear on appropriately felt reactive emotion, they do not have a similar bearing on blameworthiness.

So the more precise way to make our point is this. Rather than saying that the blaming emotions are “sensitive to the passage of time” and that blameworthiness is not, we might say that the kinds of events (if any) that can occur over time that defeat or diminish blameworthiness for a particular action—atonement, rectification, apology, and such—are quite different from those that render the blaming attitudes inappropriate—like moving beyond the events of the past, memories fading, or constructing a narrative of one's life in which the event's significance is reduced, and so on.Footnote ¹⁷

In other words, as time goes by, events unfold in the lives of those who are the victims of culpable mischief, the history of the moral community as a whole, and indeed in the lives of wrongdoers themselves that can impact the appropriateness of the feelings of resentment, indignation, and guilt (respectively) that each might have. The claim is not that this is always so; only that the emotions are such that they are appropriately responsive to these features of the passage of time.

The same does not appear to be true of a wrongdoer's blameworthiness for what he or she has done or failed to do. Life unfolding does not in the normal course of events create reasons that diminish or eliminate blameworthiness for a given act or omission. Reductions in blameworthiness for an action (if indeed this is possible at all) are the result of distinctive kinds of events: making repair, atonement, apology, or changing one's identity altogether. Thus blameworthiness and the blaming attitudes are differentially responsive to the passage of time in ways that pose a serious objection to reactive attitude accounts.

This, then, is our basic objection.Footnote ¹⁸ Is it one that can be met? That is the question we are concerned to investigate in the remainder of this paper.

III.

According to the simplest versions of the reactive emotion account, correctness consists in appropriateness, broadly construed. For an emotion to be appropriate in this broad sense is just for one to have sufficient reason (of one kind or another) to feel or otherwise experience it. The trouble, as a number of philosophers have pointed out, is that not all of the considerations that at first might seem to bear on whether to feel one of the blaming emotions are reasons of the “right sort.” After all, some considerations might seem to count in favor of or against feeling a reactive emotion in a particular case, even though they do not seem relevant to whether the target of those emotions is blameworthy for what she has done or failed to do. They are instead what Stephen Darwall calls “reasons of the wrong kind.”Footnote ¹⁹

Take a straightforward example. Suppose your sibling has wronged you in a manner that renders her both blameworthy and a proper target of resentment. It is not hard to imagine circumstances in which displaying or even experiencing resentment toward her would create tension in your relationship, in some cases resulting in complete fracture. Intuitively, the threat of fracture, however real and unfortunate, does not constitute the right sort of reason for determining the appropriateness of experiencing anger or resentment toward her for what she has done. After all, it has no obvious bearing on her degree of blameworthiness. This suggests that it is not enough that we have good reasons for not having or experiencing the reactive attitudes. We need good reasons of “the right sort.”

Some considerations are clearly reasons of the wrong sort—those of health, prudence, and etiquette, for example—but not just these. To this list Justin D'Arms and Daniel Jacobson have added a range of moral considerations—including those that bear on one's own or others’ well-being, and perhaps fairness as well.Footnote ²⁰ Thus there is a wide range of reasons for feeling or withholding the reactive emotions that intuitively have little to do with the blameworthiness of agents for what they have done or failed to do. If such reasons of the wrong kind are allowed to bear on the appropriateness of the reactive emotions, then a reactive emotion account of blameworthiness is doomed from the outset. For allowing them to count would give rise to cases in which the reactive emotions are appropriately felt toward someone who is not blameworthy, or cases in which these emotions are not appropriately felt toward someone who is.

It is plain, therefore, that the reactive emotion account must be reformulated so as to take account of the distinction between reasons of the right kind and reasons of the wrong kind. In particular, the account should be put roughly as follows:

Does the distinction between reasons of the right and wrong kinds help to answer the objection from the passage of time? That will depend, of course, on whether one can make out the distinction in a principled way when it comes to the relevant emotions. It is one thing, after all, to assert that the passage of time is the wrong kind of reason for the blaming sentiments and quite another to explain why that should be so. Unless some account of the distinction between reasons of the right and wrong kinds can be given that explains why time—just like prudence, health, etiquette, and so on—deserves to be included among the “wrong reasons,” the present response would appear to be plainly ad hoc.

Pause for a moment to take a look at the laundry list of reasons of the wrong kind compiled so far: prudence, self-interest, etiquette, welfare, morality, fairness, and now considerations that arise in virtue of the passage of time. If a reactive attitude account of blameworthiness is to be sustained, all of these considerations must turn out to fall on the “wrong reasons” side of the ledger. But what, beyond the desire to save a reactive emotion account of blameworthiness, could explain why they are to count as the wrong kind of reasons for the blaming attitudes? Perhaps one does not need a theory in order to explain why a few isolated considerations here and there do not bear on the appropriateness of the reactive attitudes. But once the list of reasons that are ruled out as “the wrong kind” grows as long and diverse as it now has, the plausibility of REA hangs on offering an explanation of why some considerations bear on the appropriateness of the reactive attitudes and others do not.

Given the difficulty of spelling out the distinction between considerations of the right and wrong sort, we would do well first to explore other approaches to meeting the objection from the passage of time. For if the objection can be met without having to characterize what is no doubt a controversial distinction between reasons of the right and wrong sort, then so much the better. As it happens there are several options available, some more promising than others. None, however, are ultimately persuasive—or so we argue.

IV.

Faced with the objection that blameworthiness and appropriately felt reactive emotion can come apart because of considerations involving the passage of time, one response would be simply to bite the bullet. That is, the defender of a reactive emotion account could argue that it is simply a consequence of his position that blameworthiness is also sensitive to the passage of time in the same way that resentment, indignation, and guilt are.Footnote ²¹

Without more, this response simply begs the question. After all, if one simply insists that it is a consequence of one's position that blameworthiness must be sensitive to the passage of time in the same way that the blaming attitudes are, then it is one's theory that drives the attitude one is taking toward one's intuitions, and not the other way around. To avoid begging the question, one needs some additional consideration that would justify biting the bullet. As it happens, we can think of at least one such consideration.

Recall that the reactive emotion account of blameworthiness is an instance of the buck-passer account of normative predicates more generally. One could argue that we must rethink our ordinary intuitions about the kinds of considerations that bear on blameworthiness in order to preserve the theoretical advantages of treating blameworthiness in the same way that we treat other normative concepts such as value and desirability. That is, maintaining a certain kind of theoretical unity in the treatment of normative predicates might speak in favor of absorbing the costs involved in abandoning our intuitions about the factors that can bear on ascriptions of blameworthiness for wrongdoing.

In principle, theoretical unity is the kind of consideration that could warrant biting the bullet about whether one's blameworthiness is responsive to the passage of time. Here, however, theoretical unity hangs on the aptness of blameworthiness for a buck-passer account. It is important to recall that a buck-passer account of blameworthiness just is what a reactive attitude account amounts to. And as we saw in the previous section, reactive attitude accounts of blameworthiness are defensible only if the distinction between right and wrong kinds of reasons can be drawn in a principled way.

This does not make buck-passer accounts of blameworthiness unique or otherwise special. A buck-passer account of any normative concept—not just blameworthiness but also value, desirability, and so on—will need to rule out some considerations (e.g., prudence, health, etiquette, and fairness) as reasons of the wrong kind in order to avoid a range of intuitive counterexamples.Footnote ²² The difference is that it may be considerably easier to draw that distinction in the case of the desirable and the valuable than it is in the case of the blameworthy. After all, it is not clear that time impacts valuing and desiring in anything like the way it does for resentment, indignation and guilt.

The theoretical unity we would secure by treating blameworthiness in the same way as other evaluative concepts obtains only if the distinction between reasons of the right and wrong sort can be made out in a principled way. Whether this is possible still remains to be seen. Thus, on closer inspection, the strategy of biting the bullet turns out not to be much of an alternative solution to the objection from the passage of time. Instead, it is just another way of raising the concern that any adequate response will require making out the controversial distinction between considerations of the right and wrong sort.

V.

A different sort of answer to the objection from time is this. As formulated in the literature, reactive emotion accounts typically make no explicit mention of time, and so in expressing our concerns about them, we have been satisfied to treat them as they are invariably characterized. One might think, however, that the best way to meet our objection that blameworthiness and blame come apart due to the passage of time is to formulate the reactive emotion accounts in ways that introduce time indexing in a particular way. Specifically, one might think that the objection can be defused if one takes it that a wrongdoer's level of blameworthiness is to be tied to the level of reactive emotion that it is appropriate to feel toward him at the time of his wrongdoing.Footnote ²³ More precisely formulated, the revised view can be stated as follows:

One might think that adopting REA_t would allow one to accommodate the case of Megan and avoid the problems it brings up. Recall that it would have been entirely appropriate for Megan to resent Mr. Anderson a large amount at the time he harshly and unfairly berated her for upsetting his flowers. But the argument from time insisted that after many years had elapsed, it would be inappropriate for Megan to resent Anderson as much as she was entitled to do on the day he wronged her. However, the defender of REA_t can accept all this, it seems. After all, what REA_t entails is that Mr. Anderson's level of blameworthiness is fixed by the level of reactive emotion it would be appropriate to feel toward him at the time of his wrongdoing. And once this level is fixed, there is nothing in REA_t to suggest that Anderson's blameworthiness for his mistreatment of Megan would change, even if the level of reactive emotion that would be aptly felt toward him were to subsequently diminish. Thus, on REA_t, blameworthiness is tethered once and for all to the amount of reactive emotion that the wrongdoer is appropriately subject to at the time of wrongdoing. So it is no problem for the view if blameworthiness and appropriately felt reactive emotion then proceed to part ways as time passes.

As initially promising as the time-indexing response might seem, it faces two distinct problems. First, difficulties seem to arise when considering acts of wrongdoing that do not take place at any identifiable instant but rather extend over a prolonged time period. In particular, we encounter the problem when none of the actions that together constitute culpable wrongdoing would be objectionable if taken on their own. For example, suppose there is an agent—call him Bernie—who engages in an extensive plot to defraud investors. Suppose he performs a range of facially innocuous actions over the course of many years, each of which would be morally permissible considered in isolation. (He slightly downplays cash-flow problems, gets management to release a cookie-jar reserve at the ideal moment, makes use of complex derivative instruments to make risk less apparent to the investor, and so on—all within seemingly normal limits.) However, suppose that the net effect of Bernie's actions is to mislead investors and induce them to part with their money in ways they would not have consented to had they fully understood the investment opportunities offered to them. Suppose further that the investors discover the fraud only after Bernie's plans have already come to a close.

It seems plausible in this case to regard Bernie as extremely blameworthy for his wrongdoing. However, it is not clear how REA_t is able to capture this result. The issue is not that in a case of an extended wrong we cannot make sense of the claim that blameworthiness is to be settled at the time of the wrongdoing. For REA_t can just take the relevant time to be the interval during which the extended wrong occurred. Thus blameworthiness during an interval would equal the sum of (or really the integral over) all the momentary levels of appropriate reactive emotion during that interval.

Rather, the problem arises because Bernie's fraud consisted of a string of individual component actions, none of which was morally problematic considered in isolation. Accordingly, the level of reactive emotion it would have been be appropriate to feel toward Bernie at the time of any of these component actions was zero or something close to it. Thus it is not clear how REA_t can generate the result that Bernie is highly blameworthy for the entire string of the individually innocuous actions that taken together constitute the fraud. Because wrongdoing can be extended across time and emerge from a string of individually unproblematic actions, it seems mistaken to take one's blameworthiness for some wrongful act to be equivalent to the level of reactive emotion that is appropriate at the time of that wrongdoing, as does REA_t.Footnote ²⁵

The second problem arises because the harm caused by a particular act of wrongdoing might itself materialize only after a great deal of time has elapsed since the initial wrong. To illustrate, suppose you work in a small pharmaceuticals company and on the first day of January 2010, you did your job negligently, with the result that you created a 10 percent risk that a particular person would contract a painful disease ten years later. Suppose that your possible victim turns out to be unlucky and does in fact end up contracting the disease in 2020, thanks to the risk you exposed her to a decade earlier. When the harm actually befalls your victim, it seems she would then be warranted in resenting you a great deal. By contrast, if she were told of your negligence at the time it occurred, in 2010, your then-merely-potential victim would seem warranted in resenting you somewhat—but presumably much less than she would in 2020, when the risk materializes. After all, in 2010, resentment is called for only by the risk that you exposed your potential victim to, while in 2020 resentment would also be called for by the actual pain and suffering she endures from the disease. In general, it seems plausible that directly causing a harm of some magnitude, m, merits a greater degree of resentment than merely exposing someone to a risk of experiencing a harm of that same magnitude m. So in the instant case, your victim would seem to be entitled to resent you more in 2020 than in 2010.Footnote ²⁶

What, then, is the level of resentment that your blameworthiness for this negligent act should be set as equal to—the heightened level that would be appropriate in 2020 or the lower level that would be appropriate in 2010? REA_t entails that it would have to be the former. After all, REA_t takes your blameworthiness for an action to correspond to the level of reactive emotion that it would be appropriate to feel toward you at the time of that action. However, this invites the worry that REA_tundervalues the amount of blameworthiness that accrues in virtue of your negligent behavior. After all, your blameworthiness would be higher if pegged to the level of resentment warranted in 2020 instead of the level that is warranted in 2010. And one wonders what principled reason there could be for not pegging your blameworthiness to the reactive emotion that is warranted at this later time. It thus seems arbitrary to privilege the time of your wrongdoing over the time of the resultant harm as the point at which we must look at how much reactive emotion is appropriate in order to determine your level of blameworthiness.

Arguably, one could say that the blame is appropriate to actions under particular descriptions. In that case, the action can be described in ways that do not include its causal upshots or in ways that do; and the level of blame appropriate would depend on the description in question. Setting aside a range of problems that are familiar in action theory for such an approach, it is enough to note here that any such strategy runs the risk of begging the question. After all, one cannot choose which description is apt by looking at its effects on the reactive attitude account, since the very plausibility of that account hangs in the balance. Someone pursuing this line of argument needs to formulate independent criteria of aptness for descriptions of events and then see whether this would yield results about the present case that would save the reactive attitude account.

VI.

Perhaps these difficulties can be avoided if one moves to an ideal observer version of the reactive emotion account instead. The reactive emotion account can be reformulated as follows:

In the above case, a fully informed observer would know, even at the time of your negligent action (i.e., in 2010), that the end result is that your victim will in fact become ill. Accordingly, the level of reactive emotion it would be appropriate for such an ideal observer to feel at the time of the wrongdoing (i.e., in 2010) would be the same as the level it would be appropriate for her to feel at the time of the resultant harm (i.e., in 2020). Thus REA_i does not face the worry about undervaluing your blameworthiness or the charge of arbitrariness that afflicted REA_t.

Nonetheless, REA_i faces other problems. For one thing, it is unclear what the value is of determining what it would be appropriate for an ideal observer to feel, because no actual person would ever be relevantly like the ideal person. Actual people labor under epistemic limitations about the consequences of their own and others’ actions, and this fact of life is deeply embedded in the emotional responses of actual people. But one of the motivations of the reactive emotion account is that it promised to explain a relatively mysterious normative concept in terms of something we have a firmer grasp on—namely, a set of attitudes that we are all familiar with from our own emotional lives. But once we revise the reactive emotion account by introducing the notion of a fully informed idealized agent and appeal to her emotional reactions to determine a person's blameworthiness, then this central motivation for the account is lost. After all, the emotional reactions of a fictitious idealized agent are at least as obscure as the underlying normative concept to be analyzed.

Perhaps even more worrisome, REA_i seems to face the underlying objection based on the passage of time with which this article began. Insofar as we are able to get a firm enough grasp on the notion of a fully informed observer to predict what his emotional responses would be, it seems to us that any resentment or indignation that such an idealized person might feel in response to wrongdoing would be sensitive to the passage of time in just the same way as the emotional responses of a real person. Just as it seems appropriate for Megan's resentment toward Mr. Anderson to have diminished somewhat thirty years after her being wronged, so, too, do we suspect that the ideal, fully informed observer would be similarly inclined to bear less reactive emotion toward Mr. Anderson thirty years on. By contrast, as we argue above, Mr. Anderson's blameworthiness for this action would not diminish merely in virtue of the passage of time. Thus we think it is possible for blameworthiness and the level of reactive emotion that an idealized, fully informed observer would feel to come apart as time passes. As a result, REA_i seems to suffer from the initial problem concerning the passage of time that we have been seeking an answer to all along.

Someone drawn to the ideal observer formulation might seek to insulate it from this objection in either of two ways. First, one might hold that the level and intensity of the reactive attitudes that the ideal observer experiences toward wrongdoing in fact would not change with time.Footnote ²⁷ After all, while Megan's resentment might understandably diminish with time because her life conditions change and she acquires other more pressing concerns, one might insist that the ideal observer does not change in any such way.

The trouble with this response, however, is that insisting on this threatens to render the ideal observer view too abstract to preserve the purported benefits of the reactive emotion account. In fact, we do not seem to remain equally concerned with every experienced wrong until the end of our days. While some traumatic events might stick with us, a common feature of human psychology is that we become less concerned with many sorts of events as they recede into the past. To abstract away from this feature of human psychology threatens to transform the ideal observer into a demi-god whose emotional reactions would be dramatically different from our own and therefore difficult to predict. This would be to deny the reactive emotion account one of its main attractions—namely, the ability to explain blameworthiness in terms of attitudes and practices that we are all intimately acquainted with.

Thus another way in which we might steer clear of these problems is by combining the time-indexing strategy with the ideal observer version of the reactive emotion account.Footnote ²⁸ This would yield a version of the view according to which P is blameworthy to degree d for an action a, performed at time t, if and only if it would be appropriate for a fully informed observer, at t, to feel reactive emotion to degree d toward P in response to a. Call this view REA_it.

There are several problems with this alternative. First, if the ideal observer's emotional response to a given action can change as time passes (as we suggested above must be supposed to avoid severing the tie to our actual blaming practices), then it seems ad hoc to insist that the ideal observer's reactions at the time of wrongdoing are what determine the agent's blameworthiness, rather than any later reactions the observer might have. Why shouldn't blameworthiness be pegged to the ideal observer's later emotional responses? Is the idea just to preserve the theory by insulating the reactive emotions at issue from any possible relevance that time might have? This would be to beg the question. On the other hand, if one wants to avoid this problem by claiming that the ideal observer's reactions do not change over time, then her responses again become divorced from the realities of our actual emotional lives.

What is more, arbitrariness threatens the time-indexed ideal observer view in cases involving genuinely indeterministic processes (just as risk initially made problems for REA_t). Consider your favorite genuinely indeterministic process—a quantum system, say, or a roulette wheel. Suppose it is involves two possible outcomes, A or B. One or the other will occur at t₁, but before that point it is undetermined which one it will be. All it is possible to know in advance is that there is a 10 percent chance of A at t₁ and a 90 percent chance of B at t₁. Suppose further that at some earlier time, t₀, you hook me up to this indeterministic system such that I will be harmed if A occurs at t₁, but not if B occurs. Clearly you are blameworthy for exposing me to this risk. But how blameworthy? Since at the time of your wrongdoing, t₀, it is undetermined whether A or B will later occur, any reactive emotion the ideal observer would feel toward you at t₀ would have to be based just on the risk of harm you exposed me to.

However, if A ends up resulting at t₁ and I get harmed, then a more intense reactive emotion would seem appropriate at t₁ (one that reflects the harm done) than the ideal observer would have felt at t₀ (one that reflects only the risk of harm being done). The time-indexed ideal observer view, REA_it, entails that in such a case your blameworthiness must be pegged to the lesser amount of reactive emotion the ideal observer would feel at t₀ rather than the greater amount she would feel at t₁ when she sees me get harmed. But why is this the right answer, one wonders? REA_it thus seems arbitrary.

Again, one might respond that to be truly ideal, the observer must be given not just all the facts that are available at the time of your wrongdoing (t₀) but all the facts period—even those pertaining to undetermined future events. This means granting her something approaching omniscience, however, and it is far from clear what to make of the emotional responses such a being would have. Would her emotional responses be responsive to all the downstream effects of our actions until the end of time, even those that creatures like us haven't the slightest chance of foreseeing? Thus the cost of going this route, again, is to divorce the ideal observer's responses from our actual blaming practices and strip the reactive emotion account of its ability to shed light on the notion of blameworthiness by grounding it in practices we are familiar with.

Thus the general problem we keep running up against for the ideal observer versions of the reactive emotion account is that they continually face pressure to resort to a mysterious concept in order to explain the notion of blameworthiness: namely, the emotional responses of a creature with mental capacities and affective dispositions that are drastically different from ours. In other words, the worry is that we can preserve the ideal observer strategy but only at the cost of robbing it of its explanatory value. Accordingly, we remain skeptical even of this temporally indexed version of the ideal observer reactive emotion account.Footnote ²⁹

The upshot, then, is that neither the time-indexing strategy nor the ideal observer version of the theory provides a satisfactory response to the basic temporal challenge that reactive emotion accounts of blameworthiness must grapple with. Thus we are left to turn elsewhere for a solution.

VII.

Recall that all forms of the reactive attitude accounts of blameworthiness rely on a notion of the emotions being correctly held, felt, or experienced. The burden of such accounts is to specify the standard of correctness for the reactive attitudes. Spelling out the notion of correctness in terms of “appropriateness” does not remove the problem to which we are calling the reader's attention. It provides little illumination to say that the attitudes are correctly held if held appropriately, and appropriately held if held for the right kinds of reasons. The ultimate upshot is that there is significant pressure to distinguish between reasons of the right and wrong sort in a principled (non-question-begging) way. And while we see the intuitive appeal of counting health, prudence, and well-being among the class of considerations of the wrong sort (since they are often not related to the moral significance of what we have done and thus to the appropriateness of responding to it in various ways), it is considerably more difficult to see why the passage of time should not bear on what responses are appropriate to what has been done.

We can put the point briskly. The standard of appropriateness calls for a distinction between considerations of the right and wrong sort but itself provides no guidance as to how that distinction is to be drawn. Perhaps an alternative way of characterizing the standard of correctness will offer more guidance.

Just such an alternative has been suggested by D'Arms and Jacobson, who in a series of papers have advanced the view that for emotions, correctness is a matter of what they call “fittingness.”Footnote ³⁰ Their notion of fitting emotions relies on an analogy with the propositional attitude of belief. Beliefs have objects—the propositions believed—and they represent their objects as being true. Thus beliefs fit their objects, or are fitting, insofar as their objects actually have the property they are represented as having, namely truth. In other words, a belief that p is fit to its object (or fitting) if and only if p.

Similarly, on this view, emotions have objects and represent their objects as having certain properties.Footnote ³¹ Just as beliefs represent their objects as true, emotional responses such as envy or fear represent their objects as having certain properties—namely, being enviable and being fearsome, respectively (or something of that sort). An emotion, then, is said to be fitting, or fit to its object if and only if its object actually has the properties that the emotion represents it as having. Envy, for example, would be a fitting emotional response whenever its object is worthy of envy, and fear would be fitting whenever the object of the emotion in question is worthy of being feared (e.g., because it is dangerous). When it comes to the reactive emotions—namely, resentment, indignation, and guilt—the thought is that they are fitting responses only if their objects actually have a certain property, presumably blameworthiness.Footnote ³²

With this notion of fittingness in hand, we can now reformulate the reactive emotion account of blameworthiness as follows:

Substituting fittingness for appropriateness in formulations of reactive attitude accounts of blameworthiness is potentially a significant advance, both in understanding our emotional lives and in providing a defensible version of the reactive attitude accounts of blameworthiness. The great advantage of fittingness accounts is that they obviate the need to provide an independent account of the distinction between reasons of the right and wrong kinds for feeling the emotions.

While it is not obvious, on the appropriateness approach, why considerations involving the passage of time must be regarded as being of the wrong sort for resentment, indignation, and guilt, no similar problem need arise when the reactive emotion account is formulated in terms of fittingness. For on such an approach, emotions are taken to represent their objects as having certain properties and thus provide their own standards of correctness. This implies that if the blame emotions represent their objects as having a property that is not sensitive to the passage of time, then considerations involving the passage of time would have no bearing on whether these emotions are fittingly felt.

VIII.

For all its promise, the concept of an emotion being fit to its object raises problems for the reactive emotion account. We raise two main objections to the view—one concerning the passage of time, and the other involving a more general worry about the structure of this view.Footnote ³⁴

The problems begin to emerge when we look more closely at the question of what the property is that the reactive emotions are supposed to represent their objects as having. One preliminary concern is that the blame attitudes seem to differ importantly from one another. Resentment is fittingly experienced or expressed only by those with standing to resent: namely, victims of culpable wrongdoing. Guilt may be felt only toward oneself for one's own misconduct. And indignation may be felt by all members of the moral community against those who have failed to measure up to the norms of behavior required of them. These facts about the reactive attitudes suggest that they may differ with respect to the property that they represent their objects as having.

Nonetheless, the blaming emotions are obviously related in an important way in that they are felt in response to some kind of culpable misconduct. This suggests that despite their differences, there is indeed some common property that all three blame emotions represent their objects as having. And so, for the fittingness version of the reactive emotion account to be defensible, it likely will have to be cashed out in terms of whatever common property it is that the blame emotions represent their objects as having. More specifically, the account would be reformulated to say that a person is blameworthy if and only if she possesses the common property, whatever it is, that all three reactive emotions represent their objects as having (even if each also represents its object as having other properties besides).Footnote ³⁵

But now we encounter the first major difficulty with the proposal. Assuming there is some common property that the reactive emotions represent their objects as having, why think that it is not sensitive to the passage of time? This, after all, is precisely what the fittingness version of the reactive emotion account has to suppose in order to avoid the objection. Recall that the original problem was that the reactive emotions seem to be sensitive to the passage of time in a way that blameworthiness is not. To avoid the ensuing problems, the fittingness version of the theory would have to take the reactive emotions to represent their objects as possessing a property that is not sensitive to the passage of time. For only in this way can it ensure that fitting reactive emotion does not part ways with blameworthiness as events unfold with time.

But besides one's desire to save the fittingness version of the theory, what reason could there be for supposing that the reactive emotions represent their objects as possessing a property that is time-insensitive? Why should the reactive emotions not be thought to represent their objects as having a property that can vary in degree as time passes and life goes on? To stipulate that the reactive emotions ascribe a time-independent property to their objects simply because that is what it would take to save the account from its temporal problems would be to beg the question.

Above we argue that one cannot rule out temporal considerations as reasons of the wrong kind without some principled explanation of what makes them belong on the wrong-reasons side of the ledger. And now we see that the defender of REA_f faces a similar challenge. To answer the temporal problem in a non-question-begging way, she would have to provide (i) a substantive proposal about what property the reactive emotions actually represent their objects as possessing, and (ii) an explanation of why this property is time-insensitive in the relevant sense. The burden, therefore, is on the defenders of REA_f to explain why the reactive emotions ascribe to their objects a time-insensitive property of the sort that would save their view from its temporal problems.

Even if we suppose that such an argument is in the offing, other problems remain, the most pressing of which is that fittingness versions of REA face an inescapable trilemma: either they are circular or false or they render the reactive emotions otiose.

IX.

With the hopes that the notion of fittingness would provide a defensible version of the reactive emotion account apparently dashed, we are left to reassess our initial skepticism about the prospects for success of accounts on which blameworthiness is a matter of being the target of reactive emotion that is supported by the balance of reasons of the right kind. As we saw above, any such account is sustainable only if it is supplemented by a principled way of distinguishing reasons that are of the right kind from those that are not. Accordingly, we now turn our attention to four of the leading attempts to systematically draw this distinction. We argue that they are all inadequate to the task of saving the reactive emotion account of blameworthiness.Footnote ⁴² We begin with Derek Parfit's familiar account of the distinction between reasons of the right and wrong sort.

X.

The fact that none of these ways of marking the difference between right and wrong reasons for or against the reactive attitudes succeeds does not establish that there is no way of doing so that would be adequate to the task of saving the reactive attitude account of blameworthiness. Establishing an “impossibility theorem” in this context is beyond our grasp. It is, moreover, unnecessary. If the arguments we have adduced to this point are sound, there is little reason to be confident that a sustainable version of reactive attitude account is in the offing. That is the posture we endorse. The reactive emotion account—despite its many attractions—seems to be ultimately unable to provide a defensible solution to our initial objection concerning the passage of time.

Hence our skeptical tone; and in this case a skeptical tone should be followed with a reminder of what the stakes are. There is an intuitive distinction between finding someone guilty of a crime and expressing condemnation, resentment, and indignation toward that person through punishment. For instance, if the relevant statute of limitations has run such that punishment would be inappropriate, we might still have a reason to ask whether the person in question had acted culpably. Is this a distinction that is grounded in our commonsense moral practices or is it a function of pragmatic considerations alone?

If we are right in thinking that blameworthiness and appropriately felt reactive emotion can come apart, then we have a natural way to ground this distinction in our familiar blaming practices. In particular, we can say that to reach a judgment that someone is guilty of a criminal offense is to express a view about the person's blameworthiness, whereas carrying out a punishment centrally involves an expression of reactive attitudes toward the wrongdoer. Since the reactive attitudes are sensitive to the passage of time in a way that blameworthiness is not, we can easily understand why we might find someone guilty even though it has become inappropriate to express resentment and indignation toward the person through punishment.

But if one instead wants to insist that blameworthiness is tightly tethered to appropriate reactive emotion in the way that the reactive emotion account suggests, then it becomes harder to understand the case where we want to judge someone guilty even though the statute of limitations has made it inappropriate to express blame toward the person through punishment. This is not to say that the task is an impossible one. One might appeal to a range of practical considerations to explain it—say, the fact that the evidence might have gone stale or the limited resources of the state to punish crimes long past. But we think there is some value in attempting to view core features of our legal system, such as statutes of limitations, as expressions of our moral practices, where possible. Thus the impact that the passage of time has on the reactive emotions not only poses problems for a well-known theory of blameworthiness but it also promises to open up new theoretical possibilities in the criminal law.