In this paper, I want to challenge an assumption about forgiveness that is widespread in philosophical discussions of the topic, viz., that it is a voluntary phenomenon, initiated by the aggrieved party and accomplished under his or her control via his own resources (such as his will or judgment). I don't claim that forgiving is never up to me. Sometimes forgiving is just a matter of deciding to. However, if that is just one variety of forgiveness, then an exclusive focus on the voluntary mechanisms involved in forgiving can yield a one-sided picture of the practise. To offset this, I explore some (but surely not all) of the dynamics that contribute to forgiveness independently of our direction.

Camilla Kronqvist cautions that

Thus, I characterise, not define, forgiveness as a positive response toward a perceived wrongdoer in a context in which the impetus to a hostile response is salient.

In the first part, I examine two accounts of forgiveness, by Avishai Margalit and Pamela Hieronymi, respectively, to which the issue of control is central, arguing both are compromised by the assumption that forgiveness is voluntary. I then explain that control is relevant not only to discussions of forgiveness that address it explicitly, but also to the wide-ranging debate about the conditionality of forgiveness. I end by trying to take account of how some of the elements of forgiveness, including time, other persons, affective associations, and grace operate independently of our intentions.

I begin with Avishai Margalit's account of forgiveness, which tries to divorce forgetting from forgiving on the grounds that forgetting is incompatible with the allegedly voluntary character of forgiveness. I argue that an adequate conception of forgetting will indeed be a significant component of a satisfactory account of forgiveness. Both forgiving and forgetting, properly understood, may but needn't be under voluntary control.

Margalit's argument has a tripartite Hegelian structure of thesis, antithesis and synthesis. The thesis and antithesis oppose two conceptions of forgiveness: forgiveness as a policy of disregarding the offence in dealing with the offender and as overcoming feelings of resentment. These, in turn, are derived from two Biblical metaphors for forgiveness: blotting out versus crossing out (names, debts from a ledger or the Book of Life). One expunges the sin, the other leaves a record of it.

The issue of control is central to Margalit's argument for the thesis that ‘the blotting-out image is incoherent. If it is necessary to forget the sin totally in order to forgive, we are faced with a contradiction. It is like Philip Roth's injunction: “Remember to forget.”’ (Margalit, Reference Margalit2002, p. 200). While Margalit isn't the first to try to drive a wedge between forgiving and forgetting, his version is distinctive for resting centrally on the supposed mismatch between these voluntary and involuntary phenomena: ‘Just as I cannot voluntarily avoid thinking of a white elephant[,] I cannot decide to forget something just like that. And so if forgiving involves forgetting, it would seem that one could not decide to forgive. Forgiveness would not be a coherent concept’ (Ibid., p. 201. Emphasis added.). For Margalit, forgiveness as forgetting is conceptually impossible not only because the intention to forget X defeats itself by bringing X to mind, but because forgiving is assumed to occur under voluntary control. ‘[W]e are discussing… what constitutes forgiveness. And the answer to this question stresses that forgiveness is the product of a voluntary decision’ (Ibid., 202). By process of elimination, Margalit buttresses the view that forgiveness must consist in covering up or disregarding a debt, rather than forgetting. ‘To disregard is a decision, to forget is not. Therefore, forgiveness, which is voluntary, should not be tied to forgetting, which is involuntary’ (Ibid., p. 203).

The antithesis of Margalit's argument qualifies, but leaves in place, the assumption that forgiving occurs under voluntary control. Here, the kind of control involved shifts from direct to indirect – the kind one exercises over one's heart muscle, for example, when one speeds it up by vigorously engaging one's leg muscles (Ibid., p. 204). What gives way under this revised notion of control is the divorce between forgiving and forgetting. Because he thinks forgiveness can be brought about indirectly, Margalit revises his understanding of forgiveness to include and depend on forgetting: ‘The antithesis is that forgiveness is not a policy or decision but a change in the mental state of the one who was wronged (“a change of heart”). Forgetting the injury is part of what is required for this change of heart and for successful forgiveness’ (Idem.).

The self-defeating character Margalit first attributed to the intention to forget is dealt with in the antithesis by introducing a temporal gap between the resolution to forgive and its fulfilment: ‘The word forgiveness denotes both a process and an achievement, just as the word work denotes both the process of working and the work that is accomplished’ (Ibid., p. 205). That process begins from remembrance but eventuates in forgetfulness. However, beginning from an intention distinguishes forgiveness from mere forgetting and confers moral value upon it:

The synthesis of Margalit's argument, which presumably represents his considered view, ‘reconciles’ the conceptions of forgiveness put forth in the thesis and antithesis by exploiting the ambiguity of the description ‘overcoming resentment’: ‘forgiveness as a policy does not contradict the idea of forgiveness as overcoming resentment… because overcoming resentment does not require forgetting’ (Ibid. p. 206). Rather than taking it to denote a change of heart (which in the antithesis was said to require forgetting), the synthesis posits that restraining resentment from translating into action counts as ‘overcoming’ it: ‘The decision to forgive is an expression of a second-order desire not to act upon our first-order feelings of resentment or vengefulness. This does not mean that the… resentment… has disappeared, but only that the second-order desire has won’ (Idem.).

Not only is forgetting not essential to forgiving, but Margalit suggests it can undermine an agent's practical identity, ‘since who we are depends on our not forgetting things that happened and that are important in our lives… [T]he role of memory in constituting who we are… is in tension with the ideal of successful forgiveness as that which ends in forgetting the wrong done to us’ (Ibid. p. 208).

It isn't altogether clear where the synthesis of this argument stands on the voluntary character of forgiveness. It can be taken to reaffirm a crucial postulate of the antithesis, viz., that forgiving is a process that is only subject to voluntary control at its commencement. Margalit says ‘the end-result of a successful course of forgiveness… is not in our hands. Only its beginning is up to us’ (Ibid. p. 208&ff). The aggrieved adopts a policy of not taking his injury as a reason dictating his behaviour toward the offender and entertains a second-order desire to overcome his resentment. What happens as a result isn't up to him. In the antithesis, forgiveness began with an intention to forget. Forgetting was under indirect control and, when attained, led to a change of heart that constituted forgiveness. Here, forgiving begins with a policy and a desire, the consequences of which aren't even said to be under one's indirect control. However, Margalit can be read as taking that policy and desire to constitute forgiveness. On that reading, the claim that ‘the end-result of a successful course of forgiveness… is not in our hands’ would mean that we can forgive at will but not determine the consequences of doing so, rather than that forgiving successfully is outside our control.

If the difference between the antithesis and synthesis is merely terminological, if they offer much the same psychological picture but disagree on whether the voluntary beginning of the process should be called ‘forgiveness’ or not, the argument is deeply unsatisfying. As we saw, the voluntary character of forgiveness was a central premise of the thesis, the basis for divorcing forgiveness from forgetting. In the antithesis, that divorce was undermined by the introduction of indirect control, which would allow forgiving and forgetting to be part of the same process without rejecting the voluntary character of either. In the synthesis, control is attributed to the beginning of the process and apparently the involuntary later stages of the process do not matter because the early ones alone properly constitute forgiveness. No argument has been offered against the view put forth in the antithesis that the end stages of the process – say, the change of heart the antithesis emphasises – matter to whether forgiveness is consummated or not. The synthesis’ revisions rest entirely on the ambiguity of the expression ‘overcoming resentment’.

Margalit's antithesis generates aporias of its own. As we saw, it attempts to bypass the self-defeating character of an intention to forget by depicting forgetting as a process which only begins with the experience to be forgotten in conscious focus. But even if it isn't directly self-defeating, reconciling forgiving and forgetting on this basis has very counterintuitive implications, viz., that nobody remembers forgiving and that forgiveness can always be undone by a simple reminder. These seem indisputably false.

Since he concludes that forgetting is inessential to forgiveness and even undermines one's practical identity, Margalit might be sanguine at seeing the credibility of his antithesis undercut (fatally, in my view). But if the association between forgiving and forgetting is so obviously farfetched, why have authors from Kolnai to Levinas felt compelled to dispute it? Wherefore its perennial attraction? Before following all these thinkers in divorcing the two, it behoves us to scrutinise the strongest possible form of the argument for linking them.

I don't believe Margalit formulates and addresses the best possible case for connecting forgiveness to forgetting. To appreciate this, we need only ask ourselves what ‘forgetting’ means in the context of his argument. From his argument that the intention to forget is self-defeating, it seems clear that forgetting is taken to be a loss of awareness. It is, say, what I do when I go about my day exactly as if I didn't have an appointment or when the next stanza of the poem I am declaiming becomes inaccessible to me. But this isn't the only notion of forgetting in circulation, nor is it obviously the best suited to explaining forgiveness.

Let's consider some uses of ‘forgetting’ in diverse sources. Flaubert says ‘Time passes, water runs and the heart forgets’ (Quoted as the epigraph to Tablada, Reference Tablada1971, p. 48). Since the heart doesn't truck primarily with information, it seems this should be different from information being lost. The poet Alcaeus writes ‘But let him, married into the family of the Atridae, devour the city… until Ares is pleased to return us to arms; and may we forget this anger; and let us relax from the heart-eating strife and civil warring’ (Campbell, Reference Campbell1983, p. 104). In Sophocles’ Electra, the chorus admonish her to

Here, forgetting is the antipode to hating, not to knowing; the heroine is urged to find an Aristotelian median between them. The element common to these uses of forgetting seems to be intense emotions losing their hold on the subject. When we ‘forget our anger’, it's not that we don't know we felt it, but that we are no longer consumed or moved by it willy-nilly. Emerson writes ‘In the death of my son, now more than two years ago, I seem to have lost a beautiful estate, – no more. I cannot get it nearer to me… something which I fancied was a part of me, which could not be torn away without tearing me nor enlarged without enriching me, falls off from me and leaves no scar’ (Reference Emerson and Porte1983, p. 473); he doesn't mean he no longer knows what his son meant to him. Distance is the figure for this manner of forgetting, as time is its catalyst.

Stephanie Dowrick brings such a conception of forgetting to bear squarely on forgiveness:

What difference does it make to Margalit's argument if we replace forgiveness as loss of awareness with this more pertinent variant he acknowledges only in passing? When it comes to affective release, the intention to forget needn't be self-defeating (as it would be under Margalit's cognitive notion of forgetting), even without a time gap. To be sure, keeping an offence in mind can sometimes fan the flames of resentment, but this isn't inevitably the case; thinking through my resentment can also help me get over it. Because this forgetting can occur actively as well as passively, not only does tying it to forgiveness generate no paradox, it also harmonises with my stance that forgiveness can occur by either means. Nor is such forgetting open to Margalit's charge that it undermines an agent's practical identity, unless I am invested in defining myself as a victim.

As for the characterisations of forgiveness Margalit tries to choose between, the answer depends on context. In some cases, the policy he describes will count as forgiveness. If a casual acquaintance, say, won't let his dislike of me decide how he behaves toward me, I may count myself forgiven; if an intimate does likewise, that probably won't suffice. By contrast, a positive change of heart toward the offender will count as forgiveness even when punishment is administered.

On the other hand, when it comes to control, I believe Margalit's claims about forgetting apply equally to both varieties: we have at best indirect control over affective release. Yet forgetting as affective release seems to me indisputably connected to forgiving as a change of heart; it's hard to take a more favourable stance toward someone while one remains in the grip of resentment toward her. If the condition he introduces in the thesis that ‘forgiveness is the product of a voluntary decision’ stands, it opens up the possibility that whether a given stance toward an offender counts as forgiveness or not will depend on whether it was brought about under the agent's control. I doubt that is a legitimate condition. To expand on my stance that forgiveness can be voluntary or spontaneous, let me compare it to awaking. Most mornings, I simply awaken as a matter of course. However, every now and then, I become aware that I am dreaming and decide to awaken.Footnote ¹ Whether deliberately or spontaneously, however, no one would dispute that I've legitimately awakened. I hope my discussion in the second half of this essay makes plausible that forgiveness should be regarded the same way. Thus, using forgiveness’ ostensibly voluntary character to divorce it from forgetting turns out to be a red herring. Such determinations are more sensibly made based on what the comportment in question can contribute to forgiving, and I have argued forgetting has a role to play: the distantiation from my resentment I achieve through it clears a space within which my stance toward the offender can go on to evolve toward a more neutral or benevolent one.

Pamela Hieronymi offers a different, starkly rationalistic picture of voluntary forgiveness. She commends David Novitz for eschewing any notion that forgiveness is subject to direct voluntary control, but takes him to task for basing his account on strategies that might fit Margalit's analogy between forgetting and raising one's heartbeat by activating one's leg muscles; he ‘portrays [emotions] as objects of indirect manipulation (we can decide to do things to ourselves to make them go away)’ (Hieronymi, Reference Hieronymi2001, p. 535). Indirect control, in Hieronymi's view, isn't a viable alternative to direct control because forgiving isn't a matter of willing but of judging. ‘An articulate account must make use of the fact that emotions are subject to rational revision by articulating the revision in judgment or change in view that allows us to revise our resentment’ (Idem.). I will argue that her analysis – like, perhaps, any account that reduces forgiving to judging – suffers from significant explanatory lacuanae.

Like Margalit, Hieronymi tries to bring forgiveness into focus by comparing and contrasting it with another act – in this case, excusing:

For Hieronymi, forgiveness can't involve a revision in one's judgments about the wrongness of the offence, one's own standing or the offender's, since revising such judgments corresponds with other acts, like excusing. Rather, she thinks forgiveness acknowledges a retrospective change in the significance of an offence that initially constituted a threat:

When an offender apologises, this story goes, the offence is cut off from the source of its threatening meaning. The wrongness of the act is established, so there is no longer anything for me to protest and resentment loses its point.

I count the improbable juxtaposition of resentment's defensive aims and retrospective focus an important insight of Hieronymi's analysis. However, I don't think she accounts for it properly. First, it is not always the case that there is nothing but my resentment to mark the offence as wrong. For instance, it's possible for someone to injure me in a way that is clearly prohibited by law. On her account, there is no real reason in that case for a victim to resent such an injury, since its wrongness is already clearly established. But this seems counterintuitive. In fact, I don't think we are any less likely to resent manifest wrongs than less definite ones. If anything, the reverse seems reasonable, since a manifest wrong allows no room to give the offender the benefit of the doubt.

Second, Hieronymi bases resentment on a rather gratuitous mix of theoretical and practical concerns. What the victim is said to protest is a claim implicit in the act: that the injury is right and permissible. What grounds do I have to protest such a claim? In the domain of theory, we could ground that protest in its inaccuracy: what it presents as permissible is in fact unacceptable because it's wrong. But why can't its wrongness justify a protest directly? Why should the victim concern himself with what is claimed, rather than what was done? As it happens, Hieronymi doesn't justify the protest on theoretical grounds but on practical ones: the claim poses a threat. But this raises the same question the theoretical justification would: What threat does the claim pose? Apparently, the threat that I will be injured again, by those who accept it. But if those prospective harms can justify the protest, why can't the wrong I have already suffered do so itself?

Later, I will offer some thoughts about why resentment is defensive and retrospective. Here, I want to address Hieronymi's comments on the question of control. As we saw, judgment by itself is supposed to usher in the remission of resentment, at least in the ‘well-functioning person’ or ‘well-functioning psyche’ (Ibid., p. 535). Those outside that exalted category may resort to ‘manipulation’ to temper their unruly feelings, but she has little to say about that ‘non-ideal case’. I want to suggest that the ‘non-ideal case’ is the real case – or, at any rate, an ordinary one – and that the gaps in Hieronymi's ‘ideal’ case hinder it from shedding much light on some important cases of forgiveness.

Hieronymi's ‘well-functioning person’ forms judgments which limit or defuse his resentment. However, the judgments’ hegemony over his emotions can't be taken for granted. Hieronymi complains that Novitz fails to recognise resentment and anger ‘as attitudes sensitive to one's judgments, subject to rational revision’ (Idem.). My suspicion is that judgments are at least as sensitive to emotions as the other way around; for instance, being angered by a given act may incline me to judge it wrong. On Hieronymi's theory, forgiveness is grounded in the revision of my judgment that a given act poses a threat – revising it not to the view that I misperceived it as a threat, but to the judgment that it has ceased to pose one.

Hieronymi's account can and has been criticised by those who believe, like Margalit, that the kind of control involved in forgiveness is volitional, not merely rational. For example, Lucy Allais rejoins that ‘[r]ather than approaching… forgiveness in terms of… warranted epistemic judgment, we should approach it with ideas such as a change in affective focus, or a change in concern-based construal,’ which ‘can be more directly subject to the will’ (Reference Allais2013, p. 646). Although she believes such changes are circumscribed by judgments, the agent has some latitude and discretion to bring them about. I will suggest the judgments in question might have less sway over our affective responses than Hieronymi and even Allais allow.

Hieronymi's account brings to mind this exchange in Dostoevsky:

Similarly, I worry Hieronymi's theory makes forgiveness too simple.

Hume observed that for any given effect, the cause is typically joined to a plethora of concomitants which, while inessential in producing the effect, ‘have such an influence on the imagination, that… they carry us on to the conception of the usual effect, and give to that conception a force and vivacity’ independently (Reference Hume and Selby-Bigge1960, p. 148). For instance,

The upshot of Hume's analysis is that we can respond to something very similar to a threat as though it were a threat, even if we judge it to be distinct. Hieronymi would say this can't be the response of the ‘well-functioning person’, but it doesn't seem pathological or even unusual. On her account, forgiveness ensues when we judge that an offence that threatened us no longer does because it has been marked as wrong. The difference between past and present threat seems subtle enough that, if Hume is right, it would not be surprising if the aggrieved continued to respond to the offence as a threat. Therefore, if resentment indeed responds to a threat, we have reason to think that often a judgment will not suffice to defuse it. Thus, Hume helps us account within Hieronymi's framework for the uncontentious fact that forgiving can sometimes be difficult, and strongly suggests we need something to supplement the judgment.

Norvin Richards gives an account of elective forgiveness that shares a fundamental premise with Hieronymi but approximates a Humean view of the recalcitrance of emotions. Like Hieronymi, Richards thinks forgiveness can be a response to a change that removes a threat; he likens an offender who repents because his moral views have changed to a once-parlous stairwell that has been repaired. In that scenario, ‘[f]orgiveness… become[s]a permissible option with appeal to persons of… good character,’ yet ‘neither would there be anything amiss in the hard feelings continuing,’ since ‘it might be that the harm was too serious for the change to relieve one's other associations between this person (that face, those hands, that smirk) and what she did to you, despite those features having not even partly caused the harm’ (Reference Richards1988, p. 88). Clearly, these strong associations aren't elements of a judgment about a threat; they potentially occlude that judgment.

The way judgments bring about emotional changes according to Hieronymi's theory is reminiscent of the way they are commonly thought to provoke actions, only more direct, since the change effected isn't even mediated by an intention. I believe Hieronymi has good grounds to reject ‘any suggestion that our emotions are subject to our immediate volitional control’ (2001, p. 535). But I also think the presumption of such control is why we consider judgments decisive and explanatory for actions.Footnote ² (It's no accident that Plato represents Hieronymi's ‘well-functioning psyche’ as a charioteer pulling on his steeds’ reins.) Where that presumption doesn't apply, what is there to fill the gap, linking judgments to emotional shifts? In fact, there is nothing in her theory to account for the putative emotional efficacy of judgments except a picture of how an ideal psyche is supposed to operate.

In Hieronymi's ‘ideal case,’ forgiveness is a rather transparent process: apology marks the offence as wrong, thereby defusing the threat and making resentment otiose. Except for a contrast with the ‘well-functioning person,’ her account envisages no difficulty in forgiving. In particular, there is no difficulty in accurately judging that the threat has been neutralised or, having made that judgment, in relinquishing one's resentment. Accordingly, there is no indication of how forgiveness could become possible when difficulties do arise, beyond a reference to ‘indirect manipulations’ she associates with ‘non-ideal cases.’ Her comments on excusing suggest those ‘manipulations’ are of no real interest, since they ‘are simply our attempt to achieve self-consistency’ – that is, they operate in the service of judgments, bringing the emotions into line with them. This seems dubious. For example, Hieronymi accepts that love can contribute to forgiveness, but it is far from obvious that a judgment is in any way prior to that love, even in the cases of apology she deals with. Assuming that every aspect of forgiveness is governed and explained by its relation to a judgment could leave us with an impoverished understanding of forgiveness and ourselves.

Margalit and Hieronymi aren't alone in making explicit the assumption that forgiveness is voluntary. However, I don't believe it is confined to works that do so. There is a debate running through much of the philosophical literature on the topic about whether forgiveness can or should be free or conditional. This brings into relief the assumption that forgiveness is under voluntary control, since it is hard to imagine that debate arising about a spontaneous process. This is easier to appreciate in relation to conditional forgiveness. While most analyses of forgiveness approach it as a timeless practise, Martha Nussbaum offers a genealogical account that suggests this conception of forgiveness is rooted in an influential strand within Christianity:

If Nussbaum is right, this tyrannical process serves as a prototype for what is now a commonplace view that forgiveness becomes appropriate when the offender has met certain conditions like remorse and apology. Since it is modelled on a relationship to God, it would seem to carry a strong presumption of control on the part of the forgiver. But even if we disregard transactional forgiveness’ putative religious roots, it is clear that this transaction is a very poor deal for the offender if he cannot count on the aggrieved party to fulfil his end of the bargain, granting forgiveness. To consummate the transaction, it must be the case that once the offender has ‘paid the price’, the other can forgive. If that is in question, the wisdom of undergoing penance, etc., seems dubious.

Although unconditional forgiveness doesn't require this same investment from the recipient, Nussbaum's gloss on it makes it clear that it is at least compatible with control by the forgiver – control not only over the act of forgiving but over the one to be forgiven. ‘[U]nconditional forgiveness… is not free of moral danger. The minute one sets oneself up as morally superior to another, the minute one in effect asserts that payback was a legitimate aim - but one that I graciously waive – one courts the dangers of both the road of status (inflicting a status-lowering on the offender) and the road of payback [by mortifying him]’ (Ibid., p. 77). As an instrument to elevate myself above the offender and/or get back at him, unconditional forgiveness would seem to also be a controlled act. If forgiveness turns out not to always be a predominantly controlled act, it would be appropriate for future work to question how relevant the question of conditionality is to our understanding of forgiveness on the whole. Thus, this debate is part of the stakes in determining whether forgiveness is necessarily voluntary; to the extent that we cease to see it as such, the question of whether it should be conditional becomes less relevant.

Having made my case that volitional and rational control explain less about forgiveness than their proponents make out, I will now try to illuminate it by exploring what roles certain elements that aren't under our control can play. The first factor I want to consider is lived time. In my discussion of Margalit, I claimed that time catalyses forgetting, which, properly understood, is an element of forgiveness. Time brings about affective release because our emotions tend to return to a baseline state; it is not easy to sustain intense emotion, good or ill. To infer from this that time is an agent of forgiveness that works independently of our wills invites a familiar objection: the overcoming of resentment should not be conflated with its atrophy or exhaustion. Forgiveness is (the outcome of) a choice. This distinction is invidious: as Margalit put it, ‘“Natural” forgetting of an injury is not forgiveness and has no moral value’.

But how do we know these ostensibly valuable and valueless processes are distinct? Andrea Westlund reports that a couple's forgiveness of their daughter's murderer provoked outrage ‘given… how little time had passed’ (Reference Westlund2009, p. 507). Further, it is perfectly intelligible for someone to decline a request for forgiveness that is made too soon. Apparently when an offence is non-trivial, it is expected that time will be instrumental in dissolving the resentment it causes. Indeed, time is often more effective for this end than an arduous effort to forgive. Of course, those processes we do control are also consummated over time. But the question is whether the purportedly voluntary and valuable process of overcoming resentment takes place independently of the spontaneous, ‘valueless’ process of ‘natural’ forgetting that unfolds concurrently and undermines resentment. It is hard to see how the task of deliberately forgiving could be insulated from the effects of spontaneously forgetting. The involvement of ‘natural’ forgetting in forgiving may be the simplest explanation for why the latter takes time: the ostensibly valuable depends on the natural and ostensibly valueless.

I believe reflection on time's relation to resentment can expand our understanding of the role it plays in forgiveness, for resentment is opposed to the flow of time. Hieronymi argued that a past offence, not the offender, remained a present threat in virtue of an enduring claim, which resentment protests. Having critiqued that argument, I want to propose a different account of resentment's retrospective focus. I suggest that resentment can reduce the offender, in the victim's eyes, to a single moment, the moment of the offence.Footnote ³ For example, virtually every word Electra speaks to or about Clytemnestra in Sophocles’ play refers to the murder of Agamemnon. Even if the past standardly influences or forms part of our perception of another person, that perception normally remains a work in progress that can evolve as the Other does. But resentment can reify a past perception that constantly reinserts itself under its own inertia, like an unending flashback that occludes the present and forecloses the future. When the offender is wracked by guilt, his perception of himself can become similarly petrified, establishing an unhappy harmony between victim and victimiser.

Bernard Williams usefully elucidates how resentment aims at the past:

While deterrence and rehabilitation look to the future, retribution aims at the past, engaging in a bootless (Williams says ‘magical, fantasied’) exercise of willing backwards.

If resentment subverts the standard temporality of one's relationship to the wrongdoer, it would seem forgiveness should restore us to that ordinary temporality. I will not attempt a comprehensive account of how this could occur. For my purposes here, what is relevant is that the victim doesn't deliberately choose this relationship to the past and that the interruption of that relationship probably isn't always chosen either.

This brings me to a second basis for forgiveness: other persons. That includes the offender, who is commonly taken into account in analyses of forgiveness, but only as a part of the process that is very much under the forgiver's control: by meeting the conditions I dictate, such as repentance or atonement, he brings about the state of affairs in which I am willing to forgive.Footnote ⁴ Conditional forgiveness may well be a real phenomenon worth understanding. However, I want to suggest that sometimes other persons can contribute the most to the forgiveness process when they undercut or bypass the victim's control.

Various philosophers have described forgiveness as a reacceptance of the offender into the moral community that had expelled him for his trespasses. To my mind, however, it's just as likely for the victim to exclude himself from community through his resentment. To appreciate this, first consider Peter Strawson's account of what he calls ‘the objective attitude’:

Strawson says the objective attitude is the default toward immature or defective agents, but that we can sometimes also adopt it toward fully functional ones ‘as a refuge, say, from the strains of involvement’ (Idem.). For Strawson, resentment and the objective attitude are opposites but in fact, they seem to be points on a continuum. His account underscores that a restricted range of reactive feelings and closure to mutual give and take is destructive of relationship. He implies resentment is one of the strains we can adopt the objective attitude to avoid, and is probably right that a wholly objective attitude excludes resentment. Nevertheless, resentment isn't the only source of such strain that one risks in interpersonal relations; uncertainty, vulnerability and helplessness are others. I contend that resentment can foment a more restricted range of emotional responses that mitigates such risks, but also obstructs the ways in which the other can affect me positively. In this sense, resentment can be a more (though not fully) objective attitude than allows for true interpersonal relationship. The case in which one ‘can fight, but not quarrel with’ the person one resents is hardly unusual. Resentment then becomes a refuge but also a solitude.Footnote ⁵ Conditional forgiveness, as described earlier, is uncomfortably close in spirit to this.

My account of resentment so far points to two tasks for forgiveness: to open the victim back up to a future that doesn't reproduce the dynamics of the past when it comes to the offender; and to restore the possibility of relationship where resentment has blocked it, by broadening the victim's range of reactive responses.Footnote ⁶ Perhaps these goals can be pursued deliberately when the aggrieved attains a high degree of self-awareness. However, if I am right that resentment begets a need for control, there may be something to be said for the case where they are achieved not through my own agency. Let's explore that scenario.

Consider Agnes Callard's explanation of how anger is dispelled. For Callard, as for Hieronymi, anger or resentment fulfils a function that is ideally served by something else – in this case, by joint caring; one cares about the relationship by resenting the defection that undermines it (Reference Callard and Cherry2017, p. 130). An apology, on this analysis, defuses anger by offering a path back to joint caring: ‘Just as the transition from co-valuation “down” to anger involves a move from “we” to “I”, so too the transition “up” from anger involves moving from “I” to “we”. And that is not a move you can anticipate in the loneliness of your anger’(Ibid., p. 133). To begin with, the anger becomes less lonely when it encounters its counterpart, the offender's contrition:

Some of Callard's insights amplify the argument I have put forth. In particular, it illuminates the isolating effect I argued the increasing ‘objectivity’ of resentment brings about, as well as the importance of mutuality to overcoming it. Anger and contrition, she observes, are different ways of caring individually about the same act; that convergence can lead both parties to return to jointly caring about their relationship, rather than the act that marred it.

Crucially, Callard underscores that the aggrieved lacks the wherewithal to overcome his resentment on his own: it takes the discovery of another person with independent but complementary and convergent feelings to open up a way back to co-valuing. ‘When you try to [ignore me or]… avenge yourself on me, to “teach me a lesson,” you arrogate to yourself the resources to solve our problem on your own. Either way, you claim an independence and self-sufficiency that you do not have.’ (Idem.) The victim's will can't, by itself, lead him out of this trance. Forgiving, on this picture, is more a matter of receiving, than of exercising control. Even if one thinks there are ways a self-aware victim can overcome his resentment independently, Callard's analysis illuminates one path to forgiveness that leads through openness rather than control.

Callard's observation that the victim can't anticipate the offender's contrition is also significant. However, the argument that he cannot do so because any expectation to that effect leaves out the offender's reality sets the bar much too high: by that standard, no one can anticipate anything, since the expectation is never the reality. My analysis of the temporality of resentment suggests a different explanation: the resentful victim can't anticipate the offender's contrition or its concordance with his anger because what he perceives of the offender is a sedimented picture of her as offender. If the offender can make her contrition felt (not under the victim's control but in the face of his avoidance), I believe that can begin to disrupt that perception, since it is hard to interpret such care as an iteration of the remembered hostility or disregard. Thus, the past loosens its hold over their evolving present, opening their future to, for example, the concordance Callard describes. Again, this is likely not something the resentful victim wills to bring about.

We are used to thinking of forgiveness as unilateral, as something one person does to another. That picture, in turn, buttresses the assumption that it occurs under the victim's control. But if, at least in some cases, it involves two persons discerning the concordance of their respective pain and giving it up for a shared aspiration, then perhaps the grammar of ‘X forgives Y’ is misleading, and forgiveness should be viewed instead as a joint accomplishment.

I now want to consider a narrative by Dostoevsky that brings out the ways in which identification and affective associations can contribute to forgiveness. My discussion centres on a scene from The Brothers Karamazov that illuminates the character of the Elder Zossima. Zossima's principal teaching that each of us is personally responsible for all mankind and all creation leads him, for example, to instruct the faithful to ‘[b]e not angry if you are wronged. Forgive the dead man in your heart what wrong he did you’ (Dostoyevsky, Reference Dostoyevsky and Matlaw1976, p. 44) and ‘shun above all things’ the ‘desire for vengeance on… evildoers’ (Ibid., p. 300). However, that teaching isn't originally Zossima's, nor did he always live by it. It harks back to his boyhood, when his older brother Markel takes ill with consumption. Markel's terminal illness coincides with a transition from an irreligious attitude to a piety founded on a sense of his own sinfulness and a reverence for forgiveness.

‘Dear, kind people, why are you doing so much for me, do I deserve to be waited on?’ (Ibid., p. 268), Markel professes his unworthiness not only to friends and family, but to his servants and the birds:

Interestingly, he is unable to provide a rationale for these professions or to counter objections like ‘Robbers and murderers have done that, but what sin have you committed… that you hold yourself more guilty than all?’ (Idem.) ‘Though I can't explain it to you, I like to humble myself before them, for I don't know how to love them enough. If I have sinned against everyone, yet all forgive me, too, and that's heaven’ (Ibid., p. 268&ff). The doctor confirms his mother's admonishment that ‘it's your illness makes you talk like that,’ declaring that ‘[t]he disease is affecting his brain’ (Ibid., p. 268). Unlike Hieronymi's, Markel's forgiveness isn't based on a judgment – certainly not on one concerning the offender.

The incident that directly precedes his entry into the monastic life finds an adult Zossima enlisted as a cadet in a military academy. Infatuated with an aristocratic woman who spurns him for another man, he insults his successful rival to provoke him into accepting a duel. Zossima goes home to prepare for the duel, where he gratuitously beats and bloodies his servant, who bears his abuse stoically. The morning the duel is to take place, Zossima awakens to a beautiful sunrise and birds singing. He feels a deep unease, which he soon traces to his assault on the servant. When he recoils at that act, memories of dead Markel and his philosophy come flooding back. Mortified, he begs his servant's forgiveness. The servant's response, ‘Am I worth it?’, directly echoes Markel's words. In his new frame of mind, Zossima is appalled at the upcoming duel, resolving not to injure his opponent. When the latter misses his first shot, Zossima throws away his weapon and begs his forgiveness.

How are we to make sense of this sequence of events, including Zossima's forgiveness of his rival? I suggest what happens is that he comes to apprehend the situation through Markel's eyes – not because he intends to examine Markel's perspective, but more spontaneously as a response to the birds and servant, to whom Markel took responsibility. In Markel's case, that responsibility was something of an abstraction, whereas Zossima has sinned against his servant quite tangibly, making his brother's principle more salient. Further, by inadvertently echoing his words, the servant becomes a stand-in for Markel himself, an object for Zossima's care. Thus, factors unrelated to the man he is set to duel with operate revolutions in Zossima's outlook. Forgiving doesn't fulfil Zossima's intentions either – his intention was to get even.

Even if Zossima's background influences him to forgive, Markel's teaching is so unusual that one might wonder whether one can draw any general conclusions from this case. I believe one can. First, Markel's views are undoubtedly decisive for Zossima's surrender of his resentment. However, they are embedded in his affective relationship to his lost brother; it is doubtful they would exercise the same influence if expounded by a stranger. Most of us care about someone. Regardless of whether that someone advocates forgiveness, the narrative suggests that discovering affinities between him and the targets of our anger can check our resentful attitudes. This is why, for example, Shakespeare's Romeo tells Juliet's menacing kin ‘[T]he reason I have to love thee doth much excuse the appertaining rage to such a greeting’ (Romeo and Juliet, Act III, Scene I).

Moreover, affective associations are only one way in which the impetus to forgiveness can be situational. In Sophocles’ Philoctetes, they are complemented by others. Philoctetes agrees to accompany the Greeks he hated for stranding him on an island to Troy, not because they have changed or apologised but because it offers a way forward, the prospect of achievement in battle and his incipient friendship with the young Neptolemus, who is part of their party. As Julius Moravscik remarks, ‘the new friendship that is in the offing is based on a conception “Καλόѵ” that is far deeper than the links the resentful person can have towards others. In light of such a relation, resentment seems pointless. There are more important things to work out in life now, with the friendship gained’ (Reference Moravscik, Cleary and Gurtler1997, p. 273). One might rejoin that Philoctetes nevertheless chooses to overcome his resentment. However, due to circumstances beyond his control, he really has no alternative to forgiving.

Returning to Zossima, although regarding his actions through Markel's eyes may well have informed his bouleversement, there is reason to think that taking another person's perspective will generally bring the ugliness of anger into relief. Adam Smith writes:

Smith also thinks that, alternating between the role of actor and spectator, we are continually attuned to one another's perspectives because we seek the sympathy of others by modulating our responses to what is fitting from a generalised perspective. If so, a standpoint from which our anger appears offputting is normally readily available to us. Zossima's case may be unusually fraught, but Smith's account suggests a similar dynamic could commonly inhibit anger. This is a variant of the notion of forgetting I offered as an alternative to Margalit's: when the victim abstracts from his individual perspective, he acquires some distance from his resentment. Even if Smith is right that it serves a certain purpose, it is unlikely that this habitual alternation of perspectives is a deliberate act.

It might be objected that even if Zossima acted upon convictions about forgiveness, what he did isn't really forgiving, since his rival hadn't actually wronged him. Hieronymi, in particular, insists forgiveness requires that the judgment that I have been wronged hold firm. The flip side of this is that most theories of forgiveness simply take that judgment as a given and never examine the potentially very active role the aggrieved party plays in constructing what constitutes a wrong against him. I am sceptical that overcoming resentment follows incommensurable paths depending on how convincing the aggrieved party's argument that he has been wronged is from an outside perspective. In Zossima's case, the thought that he hasn't really been wronged only crosses his mind after his disposition has changed. An explanation that requires the aggrieved to have been unmistakably wronged may not shed much light on the common case where there are arguments – and even resentment – on both sides.

Zossima's case also contrasts with the scope of forgiveness in most recent theories. Those theories usually treat forgiveness as an episode concerning (a) particular offender(s), beginning with a specific offence and ending when resentment is overcome. That approach favours the presumption of control, since particular actions are more commonly taken deliberately than a way of life is explicitly chosen. But Zossima's forgiveness is a way of life rooted in his childhood that extends well beyond one particular offender, and we have seen it doesn't emerge from his intentions or judgment.

For Herbert Morris, forgiveness is ‘close to the divine, involving… some detachment from self in circumstances where the pull runs deep in precisely the opposite direction’ (1988, p. 19). Thus, it ‘reveals something we might fairly view as beyond the merely human in us is importantly beyond the operation of our will’ (Idem.). That beyond, Morris thinks, is illuminated by ‘[t]he concept of grace, something that makes for the good that transcends what is within our will's compass,’ specifically ‘the grace bestowed upon the forgiver. If those who have wronged another beg for forgiveness, those who have been wronged may pray so that they may be forgiving’ (Idem.).

To illustrate what Morris describes, consider Kerri Rawson's recollections in a memoir with the self-explanatory title A Serial Killer's Daughter: My Story of Faith, Love, and Overcoming. She describes the difficulty of forgiving, even as she immersed herself in religious teachings that advocated forgiveness: ‘I spoke of God's unending ability to forgive – to love. But I was stubbornly holding out on doing it myself. I didn't know if I could forgive my dad’ (Reference Rawson2019, p. 315). Eventually, however, she experiences a breakthrough:

Zossima's story helped us appreciate that an intention to forgive wasn't necessary for forgiveness; Rawson's, that it isn't sufficient.

Obviously I'm sympathetic to Morris’ claim that forgiveness reveals something beyond the operation of our wills. I also agree that ‘[i]t is both a mysterious and beautiful aspect of who we are that we are so constituted that forgiveness is both a possibility and sometimes… a reality for us’ (Morris, Reference Morris1988, p. 19). I am less convinced that ‘[t]he concept of grace… allows for grasping this mystery’(Idem.). It is wise to acknowledge that forgiveness is mysterious to no small extent. The idea of grace, as something that is received, not forged, may apply to many instances of forgiveness, but affords us only a partial understanding of what transpires in them. Whether they must remain as mysterious cannot be known prior to inquiry. (After all, the essence of mystery, according to Stephen Kaplan, is the ‘promise of more information’ (Reference Kaplan and Nasar1988, p. 50).) Morris observes that ‘if we till the soil and sow the seed for forgiveness, there is something beyond these actions, the sun and rain if you will, that must cooperate for forgiveness to come about’ (Ibid., p. 17&ff.). In Zossima's case, the sun and rain may have been his brother's teachings and his servant's humility; for Rawson, the scriptures she studied. Nothing in those cases suggests we can't identify the sun and the rain or come to understand how they nourish our soil. Future work can progress in this direction, and may do so more effectively without concentrating excessively on preconditions, judgments or the will. Although they have their place in a philosophical account of forgiveness, focusing on them too narrowly can blinker us to the wide array of elements that contribute to the rich phenomenon under examination. This discussion has assayed a few steps to enlarge our view of forgiveness beyond the voluntary.Footnote ⁷