Alexandra Couto has a well-known paper,
"Reactive Attitudes, Forgiveness, and the Second-Personal Standpoint", in which she gives an argument that elective forgiveness is only warranted and virtuous in cases where the wrongdoer has repented and made amends:
(1) Forgiveness involves overcoming the reactive attitudes stemming from a wrongdoing.
(2) The reactive attitudes issue an implicit second-personal demand.
(3) To know when forgiveness is warranted, we need to know what that demand is and how it may be answered.
(4) Wrongdoing entails a violation of the recognition respect owed to the victim.
(5) The demand implicit in the reactive attitudes experienced by the victim is a demand for the re-establishment of recognition respect.
(6) The demand for the re-establishment of recognition respect is a fundamental demand.
(7) Not insisting on the demand to re-establish recognition respect would be tantamount to showing lack of self-respect.
(8) It isn't virtuous to forgive when the demand for the re-establishment of recognition respect hasn't been answered.
Therefore
(9) The view that unconditional forgiveness is virtuous (ceteris paribus) is not justified.
We have to be careful to some extent since 'virtuous' may mean either 'appropriate to virtue' (something a virtuous person could and perhaps would generally do) or 'required by virtue' (something a person would have to do to be acting as a virtuous person would). The background account here is a sentiment-based theory of obligation, deriving from Darwall: a moral obligation is such that its violation would warrant a 'reactive attitude' like blame or resentment. All reactive attitudes in some way call for action, a sort of expectation that what the reactive attitude concerns will be appropriately addressed. This is what gives premises (2) and (3). 'Recognition respect' is respect in the sense that you treat someone or something as worth taking into account appropriately in your deliberations. The claim in (4) requires a particular restriction; Moriarty and Holmes both render each other recognition respect (among other kinds of respect) in the basic sense that they treat the other seriously worth considering in deliberation and action, but I take it that the assumption is that we aren't dealing with recognition respect in quite this sense. There is a narrower sense of recognition respect in which we are talking about specifically moral cases, in which failure to treat someone as worth taking into account appropriately in your deliberations would be regarded as a moral wrong. (4) clearly assumes something like this. It's unclear to me why one would think that wrongdoing
entails a violation of such respect; I suspect that Couto is taking the 'appropriately' very seriously here. (5) would more or less follow from the rest, although one could argue that the demand is for re-establishing recognition respect or something that can be treated as equivalent or better (more on this in a moment). (6) sets up for (7) and (7) for (8). As to (1) itself, I am not at all convinced of reactive attitude accounts of most major moral concepts (they typically confused indicators of things with the things themselves), but the Darwallian emphasis on what is
warranted makes this less serious for this particular case than it might otherwise be, so let's assume the account correct for the moment. I take it that the "ceteris paribus" in the conclusion is to take into account cases where the action might be virtuous for completely independent reasons.
Getting (6) is tricky. What Couto wants to argue is that the demand for re-establishment of recognition respect that warranted resentment against wrongdoing establishes is fundamental in the sense that it is not superable or defeasible. The point I previously noted, that you could argue for a more flexible claim than (5), is directly relevant here: if there are other things one can do, even if only occasionally, it would massively complicate the argument. Couto's attempt to argue that the demand for re-establishment of recognition respect really is fundamental is somewhat complicated, and I am not sure I fully understand it. If we take the point to be "a relationship of equal accountability", and consider what would be experienced by hypothetical members of an ideal moral community, and as a violation of recognition respect is such as would warrant blame or resentment, this is how members of an ideal moral community would react. As I said, I'm not sure I fully understand this line of thought, but I think the point is that, since we are, Darwall-like, tying moral obligation to what is warranted, it follows that if the blame or resentment is warranted, it is obligated, so an ideal moral community would act accordingly. (Usually we would say that 'X is warranted' does not imply 'X is required', but this cannot be the sense relevant here.) What's unclear to me is why this would be taken as suggesting that the ideal moral community's blame/resentment must be narrow (specifically, one must re-establish recognition respect specifically) rather than broad, beyond the fact that Couto seems to assume there is no substitute for re-establishing recognition respect in particular.
Setting aside for a moment cases in which forgiveness is given knowing that the forgiven will never repent, the big kind of case that needs to be considered is forgiveness in advance of repentance and/or amends. There are a number of different kinds of this forgiveness-in-advance, but two obvious cases could be called
anticipatory (you forgive expecting that repentance and amends will probably be made) and
optative (you forgive hoping that your forgiveness will spur the forgiven to repent and make amends). Couto would rule out both cases on grounds of self-respect. Self-respect requires the belief that one is owed respect; such a belief requires that one thereby have 'appropriate attitudes'. This argument is utterly baffling to me. Self-respect is naturally read as requiring the belief that one should treat
oneself as worthy of respect; 'worthy' is a weaker term than 'owed', and 'oneself' is a narrower extent than one that would include anyone who might have wronged one. And how one conceives respect to work is clearly relevant. Consider a Stoic conception of self-respect, in which you treat yourself as worthy of respect (because of your participation in Reason), and that primarily consists in holding yourself to moral action and not caring all that much about how other people treat you. The proper response to wrongdoing would be to do well oneself, not go about demanding that other people make it up to you. The Stoics are interesting, because I think Stoics would usually agree with Couto that you should only forgive those who have shown a willingness to correct the matter (following a strict interpretation of some Socratic comments), but this does not follow from the Stoic conception of self-respect. And one has to consider too that one common reason people give for anticipatory and optative forgiveness is that they came to recognize that by carrying their resentment or blame they
weren't respecting themselves, but rather holding themselves back, letting the wrongdoing imprison them. People not uncommonly justify their anticipatory or optative forgiveness as a way of rising to what they should be; and this, if true, cannot be inconsistent with self-respect. But whether self-respect can include such a thing seems already to depend on whether you think it is the sort of thing that could be regarded as appropriate to someone worthy of respect; and thus Couto's self-respect argument seems to beg the question.
Couto has another argument, based on the notion, previously mentioned, that the point of it all is "a relationship of equal accountability". On a Darwallian account, holding people accountable requires that we do so entirely on factors internal to the practice of holding people accountable; this excludes acting simply on what would be desirable. So Couto suggests that forgiveness for a reason other than repentance and amends would be the "wrong kind of reason". Now, I think the Darwallian account of accountability is gravely wrong -- I think most accountability is based on goods in the appropriate larger context of the practice rather than goods internal to the practice -- but let's assume that it is right. Optative forgiveness would work as a target of the "wrong kind of reasons" argument, because it is in some sense based on desirability. But anticipatory forgiveness is not so clear a matter. How does this really affect the anticipatory case, where there is an expectation that probably repentance and amends will be made? In particular, how is anticipatory forgiveness different from ordinary forgiveness on probable grounds, in which you don't know with absolute certainty that repentance and amends were made, but you have good reason to think that they probably did happen? They both regard the same reason, and the only difference is in the note of time. Thus either anticipatory forgiveness is fine for the same reason ordinary probability-based forgiveness is, or we have to deny that forgiving on only probable grounds is acceptable (which drastically reduces the allowable cases of forgiveness, since most forgiveness is based not on certainty, even moral certainty, but on signs of repentance and amends-making), or we have to hold that holding-accountable is an intrinsically backwards-looking practice (which is implausible and seems arbitrary).
There are a number of cases of forgiveness that Couto's strictures rule out as capable of being virtuous. It is worth going through them a moment, and seeing how they work. I take it that there are three major families of cases, at least.
(1) We have already raised the first family,
forgiveness in advance of repentance and/or amends, and the two main kinds,
anticipatory forgiveness and
optative forgiveness. These are often done, contrary to what Couto's argument suggests, out of a concern for self-respect, and in particular out of a sense people have that they need to be the better person, or that they need to hold themselves to a higher standard than others do, or that they need to avoid being passive about moving on.
(2) The second major family of cases is
forgiveness in circumstances of impossible repentance and/or amends. The obvious impossibility here is simple impossibility (the most common case is forgiving people who have already died); perhaps there are cases of relative impossibility, impossibility for incidental reasons, but let's focus on a recurring case where amends are simply impossible: forgiving the dead. Why do people sometimes come to the decision that they have to forgive people who are already dead and can't make amends? A common reason is that they have to do it to move on with their life. They aren't getting amends, period. So what do you do? Do you just dwell on the wrong done to you, which will never be righted, never even given symbolic recognition by the wrongdoer? Human beings can't live like this. This is perhaps a problem with the Darwallian tendency to see everything as a matter of accountability. Is accountability really the primary thing demanded by blame and resentment? Could not one argue that healing in some sense is what is demanded instead? Then one could say that accountability is sometimes the best way for healing to proceed, but perhaps not always. Another reason people give is self-respect, as with the case of forgiveness in advance of amends. A third reason, perhaps, that people do it is that once death has come, it's absurd to demand anything more. Neither Darwall nor Couto really consider whether human accountability has limits that are sometimes in fact reached -- limits like death itself -- so that once they are reached, there is nothing to do but count it a loss and move on, or else find a way to move on constructively.
(3) The third major family is a particularly interesting one, which I call cases of
forgiveness by deemed amends. What counts as amends? The category has to be able to include things that are symbolic, as well as things that are purely conventional. So if that's the case, why cannot some things merely be deemed as amends? OK, so let's assume you can't virtuously forgive without there being amends, but what if you often have the power simply to count something as amends? There are three cases that would sharply cut into the thrust of Couto's. Sometimes we treat the bare fact of repentance as amends enough, even though it could only possibly be so symbolically and because we treat it as being amends enough. Sometimes we don't even require definite repentance; just something repentance-ish, repentance-like, gets treated as amends enough. And sometimes, even where there has been no repentance, we count other actions as taking the place of amends -- for instance, we might take someone's good deed to someone we love to work as well as if they had repented and made amends to us, even if that wasn't the intent. Of these, the first could technically be given a place within the letter of Couto's account, but perhaps not the spirit; the second, if allowed, would make it impossible to eliminate most cases of anticipatory forgiveness; and the third seems entirely inconsistent with it. But all three of them are plausible given a very common assumption about forgiveness: forgiveness is not a passive response but an exercise of power on the part of those who have been wronged. Minimally, you have the power to accept repentance and amends. But it also seems that you have at least some power to decide what you will count as amends. (Who else would draw the line for you?) So how far does this go? Is it very restricted, or (as actual practices of forgiveness suggest) does it extend quite far?
There is an oddity in all of this kind of discussion. Forgiveness has been a topic that has picked up interest recently, so more people are discussing it; it's not surprising that they discuss things from their own ethical perspectives; given the make-up of the academy, and its conventions, it's not surprising that these perspectives are often not particularly religious. But on a matter like forgiveness, it is very strange how little anyone considers the religious aspect to it. If we are talking about unconditional forgiveness in general, why do most people who consider it a good option do so? Because they think it fits closely with Christian requirements about forgiveness. Even that aside, people regularly forgive under conditions that do not fit Couto's account and do so for religious reasons, direct or indirect, whether the religion be Christian or another. Part of the argument was about hypothetical people belonging to an ideal moral community. But what about actual people who converge on our moral ideals in many ways? What have they done? Even if you regarded all these as non-definitive, surely they are relevant to understanding the topic? But there is a curious disconnect in certain parts of ethics, as if ethics were a thing entirely done in the head of the ethicist, rather than something done in the real world by real people, some of whose advice is worth hearing and some of whose practices are worth taking seriously. And while I think you can find sages who would agree with something like Couto's basic idea that amends is necessary, I think even in those cases you would often find that they work with a very different conception of amends; and in other cases, I think you would find that the advice and examples we see in our best suggest that forgiveness is possible in many cases where there are no amends in at least an ordinary sense of the term.