1. Agent-Regret

How do we distinguish agent-regret from ordinary regret? Return to the example of the careful lorry driver. Williams noted that a bystander might experience ordinary regret over what happened; anyone who thinks such a thing was lamentable might regret it (B. Williams, Reference Williams1981, p. 27). We can even regret actions or agency without feeling agent-regret: the driver might regret the actions of the child who ran into the road, or the lorry driver's friend might regret that the lorry driver killed someone.Footnote ¹ There are forms of regret that are not agent-regret. Agent-regret has a far more specific object. What is central to agent-regret is that only the agent can think ‘I wish I hadn't done that’.Footnote ² Only the driver can lie awake at night and think ‘I regret killing the child’. Someone who feels agent-regret has the self-referential thought ‘I regret that I did it’ (Baron, Reference Baron1988, p. 261; Gaita, Reference Gaita2004, p. 53; Sussman, Reference Sussman2018, p. 794; B. Williams, Reference Williams1981, p. 27).

Guilt also takes one's own action as its object. Agent-regret is distinct from guilt insofar as guilt involves fault, whereas agent-regret need not (Williams, Reference Williams1981, p. 28). We can proceed with a rough understanding of what ‘fault’ means. The driver wasn't at fault because he was driving carefully; on the other hand, a speeding driver or a drunk driver would have been at fault. When you smashed the vase, it was just an accident; had you been juggling the vase or running through the corridor, you would have been at fault. Yet some accounts of agent-regret hold that to feel agent-regret one must think that one was, or might have been, at fault (Blackburn, Reference Blackburn2015, p. 222; Enoch & Marmor, Reference Enoch and Marmor2007, p. 419; Jacobson, Reference Jacobson and Shoemaker2013, p. 114; Rosebury, Reference Rosebury1995, pp. 512–17; Wallace, Reference Jay Wallace2013, p. 44). I deny that this is true.Footnote ³ The details of this need not detain us, but, on my understanding, someone who was not at fault but incorrectly thinks they were at fault feels guilt. It strikes me that the lorry driver can quickly affirm that he was not at fault and that he was driving safely; still, I take it that he can feel differently from a bystander and feel agent-regret. The characters I will introduce do not think they were at fault, so are firmly in the realm of agent-regret.

I want to draw our attention to one further feature of agent-regret. One feels agent-regret because of what happens. What happens affects what one has done. Let me explain this a little. The result of killing the child is the dead child.Footnote ⁴ Results are outcomes (roughly: things in the world)Footnote ⁵ that affect the constitution of actions: the death of the child affects what action the driver has performed. The death of the child means that we can say that not only did the driver drive, not only did he hit someone, but he killed someone. Had the outcome merely been that the child was injured by the driver, then the driver's action could not possibly be ‘killing’. This picture might be complicated in various ways, but for my purposes we just need to see that in many cases, outcomes affect what action has been performed. As Thomas Nagel (Reference Nagel1979, pp. 29–30) put it, ‘how things turn out determines what he has done’.

In these cases, one only feels agent-regret because of the result. The drunk driver can feel guilty merely for driving drunk.Footnote ⁶ But when it comes to our cases of agent-regret, the driver only has reason to feel anything if the bad result arises – he only has a reason to feel agent-regret because the child dies – and that's because he regrets his action (and not any associated fault). Without the result, there is no regrettable action; there is no smashing or killing, there's just walking down the corridor or going for a drive (Williams, Reference Williams1981, pp. 23–26).

I will suggest that although results are central to one performing a particular action, one does not have to regret the result itself to feel agent-regret. To feel agent-regret about smashing your vase I must regret smashing your vase, and, for me to smash your vase, your vase must smash; nonetheless, I need not regret that your vase smashed. The shards strewn over the floor might not bother me.

Before moving on to this, I need to address a complication. We seem to find agent-regret in two kinds of case (Baron, Reference Baron1988, sec. 3.2; de Wijze, Reference de Wijze2005). The cases I discuss involve bad luck: the driver is driving along and out of nowhere comes a child. They are the sort of case where we might not foresee anything bad happening. In the other sort of case, whatever one does is lamentable, and one knows it ahead of time. This is where dilemmas and hard choices arise. This is what happens in the case of Agamemnon, torn between sacrificing his daughter Iphigenia or abandoning his war (and risking the starvation of his troops) (Williams, Reference Williams1973b, p. 173). I will not discuss these cases in depth. For one, much of the literature on agent-regret focusses on cases like the lorry driver, and cases like Agamemnon's receive a different treatment. I suspect that is because cases like Agamemnon's are complicated by the presence of a choice to knowingly do something that one will regret, which introduces further issues, such as dirty hands. This introduces complications that distract from my main argument, so I will not explicitly discuss dilemma cases, though my analysis should (with the necessary changes) extend to them.

2. Purity

One way of incorporating the role of results into an analysis of agent-regret is to note that often agent-regret involves regretting the result of one's action. That is to say, the agent sees the result as a bad thing independently of her involvement. Were the lorry driver just a bystander then we imagine he would feel ordinary regret. The lorry driver doesn't just regret that he killed the child, he also regrets that the child died.

On this picture – the standard picture in discussions of agent-regret – regretting the result is a constitutive part of agent-regret. John Gardner (Reference Gardner2018, p. 139) presents the standard picture clearly: ‘I regret the injury to the child who ran out in front of my lorry, and on top of that I hold myself to have been responsible for it. Then, naturally enough, I also regret the fact of my responsibility. My pained thoughts include not just “if only the lad hadn't run out” but also “if only I had gone for a different route this morning” (or a smaller truck, earlier start, etc.)’. On Gardner's picture, we regret the result we are responsible for and that regret transmits across to become regret about what we have done.

The standard picture offers a plausible evaluative explanation of many cases of agent-regret: the agent regrets their action because the result is itself regrettable. Yet the standard picture builds this evaluative explanation into the nature of agent-regret. It says that to feel agent-regret always is to see one's action as bad because one sees the result as bad.

This is not part of the nature of agent-regret. I will argue that the correct account of the nature of agent-regret allows room for alternative evaluative explanations and the standard picture fails as an account of agent-regret because there are cases of agent-regret that do not involve lamenting the result of one's action. These cases are pure: one regrets what one has done yet this cannot be explained by one's independent regret (e.g. where one would feel ordinary regret were one a bystander) about the result.

This means that we need to adopt a new picture of agent-regret.

The simple picture just says that agent-regret involves regretting what you have done, it does not build in a particular evaluative explanation, so it allows for pure cases. More broadly, it pushes us to understand why agency matters, aside from our assessments of results.

I will now do two things. Firstly, I will sketch some pure cases; then in the next section, I will offer some evaluative explanations of pure cases, some explanations of why an agent might think what she has done is regrettable without thinking the result itself is regrettable.

The two cases I start with are fairly extreme and far from commonplace, yet they vividly illustrate agent-regret where an agent doesn't regret the result itself.Footnote ⁹ After sketching these, I will offer a few more mundane examples. In our first example, we encounter a rifleman in a firing squad. He is a keen marksman and is drafted via a fair lottery, that he willingly partakes in, to execute a notorious convict.Footnote ¹⁰ After the killing, he does not regret that the convict was killed by a firing squad – he believes this convict deserved to die at the hands of the state. He does not see the result itself as regrettable. Yet he feels agent-regret. Why? Because he regrets that he killed the convict. He sees something regrettable in his doing the deed.

This example is built out of a genuine practice from which it derives its plausibility: the dummy bullet. Firing squads often have one gun loaded with a blank cartridge or a wax bullet, but no rifleman knows who has the dummy. Thus, one rifleman (the one with the dummy) did not in fact contribute to the convict's death and each other rifleman might be able to deceive himself into thinking that he did not fire a fatal bullet and that he had the dummy.Footnote ¹¹ In theory, the riflemen can each absolve themselves of responsibility. This practice doesn't make sense if what the members of the firing squad might regret is the state-sanctioned execution of the convict. Including a blank makes sense only if they might regret what they each have done: that is, their own role in killing the convict. Unfortunately for him, our rifleman knows he fired a fatal bullet. He regrets killing the convict.

The rifleman case strikes me as both interesting and plausible. It helps to motivate the idea that there are cases of pure agent-regret. But if you are not convinced, a second example makes an even stronger case against the standard picture of agent-regret because it keeps the basic structure of the lorry driver's case. If you accept the basic cases of agent-regret, this case should pressure you into thinking that the evaluative story can be more complicated than as presented by the standard picture. Think about the celebrations of some people following Margaret Thatcher's death (Neild, Reference Neild2013). Or take Bob Dylan's (Reference Dylan1963) attitude towards warmongers in his ‘Masters of War’: ‘And I hope that you die/And your death'll come soon/ I will follow your casket/ In the pale afternoon/And I'll watch while you're lowered/Down to your deathbed/And I'll stand o'er your grave/’Til I'm sure that you're dead’. It should be clear that sometimes we do not regret, and actively welcome, the deaths of certain people. Certainly, some of these celebrating figures would still be filled with joy were they to discover that, say, Thatcher or a warmonger had been hit by a lorry. They might even take a lionizing attitude towards the driver, seeing him as some sort of inadvertent hero. Clearly, they won't necessarily regret the death of this hated figure just because someone (inadvertently) killed them.

Combine this attitude with the lorry driver and we get our case. We can easily imagine that were our lorry driver Bob Dylan, and were our victim a warmonger, Dylan would be wracked with agent-regret even though he would have been delighted had the warmonger died in some other way or with another driver at the wheel. Our fictionalized Bob Dylan feels no regret about the death of the warmonger - he positively relishes it - but feels agent-regret because he killed the warmonger.

What the rifleman and the hate-filled driver show us is that we can feel agent-regret despite not regretting the result: the death.

Although these cases might seem to be extreme, we can also imagine more mundane examples of pure agent-regret: I might not regret that you broke your nose (I don't much like you and find your vanity annoying) but I regret that it's because I ran into you; I might not regret that your secret was spilled (your behaviour towards your spouse was appalling) but I regret that I inadvertently divulged it; I might not regret that your vase smashed (it's an eyesore on our mantelpiece) but I regret that I tripped and knocked it over. These are not the most common cases of agent-regret, but they are the purest: they are pure because they are unmingled with other forms of regret and thus the object of agent-regret is most clear. One regrets what one has done, and this need not connect (evaluatively) to regretting the result.

I now want to address some challenges to these cases. One possibility is that the rifleman realises that he does care about the death of the convict, and his (causal, but also spatial) proximity to the convict's death helps reveal this to him. If this is true, then agent-regret would still depend upon regretting the result. I do not deny that sometimes agent-regret can reveal to us what we care about (Arpaly, Reference Arpaly2002, pp. 49–50; White, Reference White2017). But we need not suppose that what happens here is he changes his mind, coming to think that convict should not have died. He can still think that the convict deserved to die and that the convict's death was a good thing, he just wishes that he hadn't been a part of it. In the rifleman's case, there might be something revelatory: he might not know beforehand that he would come to regret killing the convict, and he might now also regret being part of the firing squad. Yet revelation is not the driving force of this example. What matters is that we can imagine he does feel agent-regret, and this is not explained by any new-found regret concerning the convict's death.

One might also object that, even if the agent does not regret the result, there will be plenty of regrettable occurrences independent of the agent's action that explain why the agent regrets what they have done. Perhaps I am being unduly strict in saying that it is results that carry the load for the standard picture, when instead what matters on the standard picture is that something regrettable concerning what happened explains agent-regret. For instance, we might say that the rifleman might regret that the convict acted in a way that deserved death, or the hate-filled driver might regret that the politician pursued such a bellicose and destructive policy that meant they deserved such hate; or we might say that death is always regrettable, no matter who dies. The proponent of the standard picture might say that such regrets explain the agent-regret each character feels.

But I think it would be a mistake to try to save the standard picture in this way. Think about how our characters would respond to the event if they were not involved. The rifleman would be happy to see the convict shot by another firing squad, the hate-filled driver would be content to hear the master of war is dead. Were they bystanders, neither would feel ordinary regret about the result. They might think there is nothing regrettable about these events. This suggests to me that they might not subscribe to the idea that every death is regrettable – and this strikes me as a plausible (if not pleasant) position to hold. So, such a thought cannot explain their feelings of agent-regret.

Even if these characters do recognise there are regrettable elements in the vicinity of the result, the mere fact that something regrettable is present need not give our agents any reason to regret either the result or their own action. The rifleman might regret that the convict acted in such a way, but that is not the same as regretting the convict's death. This applies to our agents when they consider their agency, too. The fact the convict acted in a horrifying way simply does not explain why the rifleman might regret shooting the convict; after all, such a consideration is more likely to weigh as a justification for such an action. The fact the convict behaved appallingly is a reason to regret that the convict acted in such a way, it is not a reason for the rifleman to regret what he himself has done. Expanding our horizons to include the past horrible behaviour of the convict or the master of war does not give rise to ordinary regret about these deaths, we do not explain our characters’ agent-regret by appealing to such possible attendant regrets.Footnote ¹²

Following from these thoughts, one might wonder whether it is accurate to say that the rifleman would feel no regrets were he to think about this case impersonally. In fact, this point deserves some nuancing, and it gives me the opportunity to make an important point. My argument does not rely on imagining characters who are entirely egotistical, thinking that their own involvement is all that matters; rather, my argument is that it is agency, rather than what happens, that matters to agent-regret. Imagine that our rifleman is not selected for the firing squad, but a friend of his (who will make similar judgments that the convict's death itself is not regrettable) is selected. Our rifleman might come to regret that his friend was implicated in a death in this way, or he might recognize (and appreciate) that his friend feels agent-regret. Still, if our rifleman thinks just about the convict's death, he feels no regret. The impersonal regret concerning this scenario arises only if he thinks about the agency of those involved. To explain our rifleman's regret that his friend was involved, we need to look to why (his friend's) agency matters, given the lack of regret about the convict's death.

My point is this: absent their involvement, we can easily imagine either the rifleman or the hate-filled driver feeling no regret (or perhaps they feel only a sympathetic regret for other agents involved). What brings regret is agential involvement. But given that they would have no ordinary regrets were they not involved (or if they feel any regret, they feel regret for those who are implicated as agents), we need to look at something other than these results to explain their agent-regret. This is what we turn to in the next section.

3. Evaluative Explanations

For pure agent-regret to be comprehensible we need to understand why one might regret what one has done without regretting the result – why one might regret killing without regretting the death. The rifleman does not regret that the convict died.Footnote ¹³ So, what is it that makes the rifleman regret killing the convict?Footnote ¹⁴ Before focussing on agent-regret, I want to appeal to other cases where it matters to a person that they are implicated regardless of their assessment of the broader situation independent of their involvement.

Take the example of a discovery.Footnote ¹⁵ This can be active or passive. By an active discovery, I have in mind someone like a scientist who wants to make a big discovery.Footnote ¹⁶ We hope that such a scientist will think that the discovery is important regardless of her involvement, but it is also possible that she just wants to make the discovery herself and is indifferent to the broader interests of science. Maybe she wants fame, maybe she is like the mountaineer who wants to be the first to scale a particular peak but who doesn't think there is any value (for her) in anybody else scaling it first. The mountaineer might think her own endeavours are important but not care about the endeavours of other mountaineers, our scientist might take just the same attitude towards science. She would be an odd figure, but a comprehensible one.

Discoveries can also be passive. One can overhear or be told something without seeking it out. Sometimes, it can be a very good thing that someone has discovered some fact, but it can be a great burden upon the person who discovers it, and they might wish that they hadn't been the one to find out. Take, for instance, a whistle-blower who overhears a politician cut a crooked deal. The whistle-blower might think that it is a very good thing that someone knows about this corruption and can start the process to remove this corrupt politician, but the burden upon the whistle-blower is immense. Or one might find out about a friend's extramarital affair and face the difficult situation of confronting the friend or potentially telling their spouse.

These examples involve discovering something, but the point does not rest upon anything particular about discovery. The point is simply that someone can be glad or dismayed that they are implicated in some event when their assessment of that event, regarded independently of their involvement, differs; they might not be glad if it occurs but glad if they are implicated, or they might be dismayed that they are implicated when they would be glad were the event to implicate somebody else instead. Our assessments can differ depending on whether we are implicated, so it should be no real surprise, then, if our evaluations of our actions sometimes come apart from our evaluations of the result regarded independently of our actions. I will now offer two evaluative explanations of why one might regret one's action that do not turn upon one's assessment of a result.

Firstly, we find a plausible evaluative explanation elsewhere in Williams's other work. What lurks behind much of Williams's work, and what has not been commented on in much of the discussion of agent-regret, is the importance of performing certain actions – where certain actions depend upon certain results, including unintended results – regardless of one's independent assessment of the results and regardless of one's assessment of one's own decision procedure. Part of what Williams wanted to show was that what we have done is important to our sense of ourselves as agents (Williams, Reference Williams1981, pp. 29–30). And surely something like this thought is what underpins Williams's (Reference Williams, Williams and Smart1973a, sec. 5) famous integrity objection. There is something important, not least to Jim, about whether it is Pedro or Jim who kills the captive in the jungle. And we might not make things much easier for Jim (so long as he is averse to violence) if we imagine that the captive is someone whom Jim hates or wishes dead.

We can better understand integrity cases by reflection on a parallel sort of case: revenge. Shylock might be wryly amused were Antonio to lose a pound of flesh; he might even be satisfied to some extent were someone else to cut it from Antonio; but for full satisfaction Shylock must cut the flesh himself because he seeks revenge due to how badly Antonio has treated him and, to redress this, he must inflict the injury himself.Footnote ¹⁷ Or take the example of Inigo Montoya in The Princess Bride. For half his life he has been waiting to say to his father's killer: ‘Hello. My name is Inigo Montoya. You killed my father. Prepare to die’. Montoya would not be content to learn that his father's killer has died. Montoya doesn't merely want him dead, he wants to kill him. Revenge seems to be achieved, at least in many cases, only if I – the person who seeks revenge – achieve it. And the evaluative judgments we make of the result and the action can differ: Montoya might regret his father's killer's death at someone else's hands, but he relishes killing him himself. What we do, and that we do it, can matter to us independently of the result.

Integrity cases suggest that who performs an action can be important aside from what result is realized. Revenge cases show us that sometimes it matters who brings about a desired result. The pure cases of agent-regret that I am exploring are similar insofar as they also turn around the idea that it isn't what happens that matters, it's that I did it.

The second sort of evaluative explanation comes from a point raised by John Gardner (Reference Gardner2018) and Tony Honoré (Reference Honoré1999): what we do – including what we do unintentionallly – affects our identities. Our actions, in part, make us who we are. This is a deep and complicated issue – just how do our actions affect our identities, which actions do this, and just why should we care about this? – and it is an issue I cannot fully explore here. But surely our actions do sometimes affect our identities, and this might provide a further explanation of why we care about what we do. Let me just try to sketch the idea.

Raimond Gaita discusses a Dutch woman who hates Hitler most of all for making ‘a murderess of her’: she was in a plot to assassinate him and had to turn three Jews out of her protection so as not to risk the plot. They were killed (Gaita, Reference Gaita2004, p. 43). Although she clearly regrets the death of those she was protecting, one can also recognise a self-focussed element.Footnote ¹⁸ It is not just what happens that matters, but what performing these actions does to somebody's own sense of who they are. She is now, in her own eyes, a murderess. To apply this thought to our examples: the rifleman and the hate-filled driver are now killers.Footnote ¹⁹ For the rifleman, what matters might be how killing the convict changes the rifleman himself. The rifleman does not regret the death of the convict in itself: he regrets killing the convict because in killing the convict he becomes a killer (Dan-Cohen, Reference Dan-Cohen1991, p. 984; Raz, Reference Raz2011, pp. 234–35). The same thought applies to the hate-filled driver.

Now, one might worry that to see being a killer as a bad thing, as something that might make one's action regrettable, one must see the results of one's killing as a bad thing. But I think that would be a mistake. To adapt an example from Les Misérables, Jean Valjean might think that being a thief is a shameful thing, even though he does not think there was anything shameful in his particular case where taking the loaf of bread was necessary to feed his starving sister. The rifleman might think that being a killer is a bad thing, and he is now a killer, although he does not think that the death of the convict was a bad thing. This is comprehensible if we grant that we might make rough-grained evaluations of having certain traits, or being certain ways, that do not necessarily track our evaluations of all the particulars involved. Being a killer might be bad because the results of a killing are usually bad, or because others regard killers (no matter the details of what went on) as somehow tainted; this does not mean that every such death must be seen as bad.Footnote ²⁰ Still, one is a killer and that is a bad thing.

So, we see that it is sometimes important who performs an action (irrespective of our evaluations of the result), and that what we do might affect our identities in important ways, and that these are plausible explanations of why the evaluations one makes of what one has done need not depend on one's evaluations of the result. Thus, we see two evaluative explanations of why we might regret what we have done without regretting the result itself. And this is not meant to be an exhaustive account of the various evaluative explanations that might explain why one regrets one's action; we can proceed with what we have whilst recognising that there might be other explanations that we can add to further bolster the plausibility of pure cases and to better understand how agency matters.

Perhaps a proponent of the standard picture of agent-regret might fight back at this point. They might say something like: ‘Ah! If what you do affects who you are, then we have a result: being a killer… and that is what explains the agent's agent-regret.’ But this is not a result in the sense used by the standard picture. The result to which the standard picture appeals to explain agent-regret is, say, the death. To move to the idea that the result which explains agent-regret is that the rifleman becomes a killer concedes my point. If we grant that the effect on their identity is what explains our characters’ agent-regret, we put evaluative explanations that focus on agency, rather than outcomes like the death, in the driving seat.

One might also worry that the evaluative explanations I have appealed to in sketching pure cases lack a certain explanatory force that the standard picture has. Namely, the standard picture explains variance: why we might regret some things but not others and why agent-regret can vary in intensity. For instance, you might not feel agent-regret over lightly brushing against someone in a crowded train carriage, but you might feel agent-regret if you break their nose. Further, this will be far weaker than the regret that the lorry driver feels. On the standard picture, we can say that some results are more important than others, so some results justify agent-regret, and the more important the result the deeper the regret. When it comes to cases of pure agent-regret, results do not play this explanatory role. Yet my account still can explain variance for pure cases. For instance, some identity-affecting factors are more important than others (anyone would rather be a nose-breaker than a killer, but better not to be a nose-breaker at all), some actions do not affect our identities, and sometimes it is more important than other times that a particular person performs a task. Why? Well, that will greatly depend on the details of the case; what matters is that we can see that these factors can vary in importance in a way that might explain variance.

Too often philosophers accept a misguided account of agent-regret: they recognise that there's something more than just regretting results and they see that Williams is on to something, but they focus on the bad result of ordinary regret. Rather than accepting the standard picture, we should accept the simple picture I have offered. The simple picture just says that agent-regret involves regretting what you have done, it does not hold that a particular evaluative explanation based on results is central to agent-regret. To explain agent-regret we need to explain why our actions matter in the first place, and why we might think that our actions are regrettable. Sometimes we cannot do this just by appealing to one's regret of the result, as the standard picture does. My hope is that we can reach a clearer understanding of the nature of agent-regret which might result in a more interesting exploration of the variety of explanations of why one might regret what one has done.

4. Care

To end, I want to explore what this tells us about those who feel agent-regret. There is a popular, and to my mind plausible, understanding of the emotions on which care and emotional reactions go hand-in-hand (Anderson, Reference Anderson1995; Helm, Reference Helm and Goldie2009; Jaworska, Reference Jaworska2007; Kolodny, Reference Kolodny2003; Scheffler, Reference Scheffler, Jay Wallace, Kumar and Freeman2011; Tannenbaum, Reference Tannenbaum2007; Wallace, Reference Jay Wallace2013, pp. 22–32). The fact I am ashamed, angry, or feel guilty shows that I care, in some way, about the object of my emotion; likewise, if I care about something then I will be angry if someone needlessly damages it or I will be happy if it flourishes. If this is right, then what we feel agent-regret about will reveal something about what we care about. The rifleman's regret is self-centred. He doesn't regret the result that the convict died. He regrets that he killed someone. This is what he cares about.

This leads into a criticism we sometimes find of agent-regret, namely that it is ‘improperly self-regarding’, given that ‘the agent is focused on himself rather than the child – as if what he regretted was that he was involved in the death of the child rather than the death of the child’ (Tannenbaum, Reference Tannenbaum2007, p. 53, my emphasis)?Footnote ²¹ Self-regard is not always problematic; take ‘narcissism’ to be an improper level of self-regard.Footnote ²² Is the rifleman narcissistic, and does my analysis – where elements concerning the agent's identity, or the agent's own role, can carry the load – mean that agent-regret, as an emotion, is fundamentally narcissistic?Footnote ²³

Whether or not an agent should ever feel agent-regret, we should not base any opposition to agent-regret on the charge that it is narcissistic. Think about what we ought to care about. It's true that in the purest case of agent-regret we care not about what happened, but about what we have done. But sometimes that is okay. It's at best arguable whether the rifleman should care about the convict, yet he clearly should care about his own role in the world. The worry about narcissism comes out more potently when we think that there is something else the agent should care about, like if the lorry driver didn't care about the dead child. It is worth pointing out here that in most cases of agent-regret there will be a variety of explanations of why an agent evaluates her action as bad. The Dutch woman from Gaita's example was concerned that she was made into a murderess by releasing the Jewish protectees, but this was not her only concern: she cared also about the victims. So, there is no reason why the lorry driver must regret his role rather than the death of the child; he might regret both things. As long as he regrets the death of the child sufficiently, it is not improper to feel agent-regret, nor is it improper to also be concerned with how he has become a killer. It would only be improper were he to relegate the child's importance too much and were he to demonstrate too much concern for himself. But the problem here would not be with my account of agent-regret, or our general proclivity to feel it, the problem would be with what the driver cares about.

Agent-regret does not have to be narcissistic. Yet my point in this paper has been to emphasise how agent-regret will greatly implicate the agent and will involve some deal of care about one's own agency. This is one of the fundamental lessons Williams tried to bring out with agent-regret and in his other work. This feature is brought out in lucid clarity in a haunting piece by Alice Gregory (Reference Gregory2017) in The New Yorker, that examines the lives of several people who have accidentally killed someone. In it, we find the story of ‘Patricia’, who hit a motorcyclist:

What is important to recognise – what this truck driver recognised – is that doing something as momentous as killing another person clearly affects the agent, and we must focus on the agent's life as well as the victim's. We care about what we have done: those who kill often care that they have killed. This is true for the rifleman, even though he does not care about the person he killed. And it is perhaps so much the worse for those like Williams's lorry driver or Patricia who care not just about being killers but also about the valuable life they have taken away from the world.Footnote ²⁴