Supererogation and Blame

1. Introduction

Supererogatory acts are ‘beyond the call of duty.’⁠Almost all discussions of supererogation begin with this phrase. See, for example, (Urmson 1958; Chisholm and Sosa 1966; Peterfreund 1978; Heyd 1982; Kamm 1985; Mellema 1991; Horgan and Timmons 2010; Portmore 2011; Dorsey 2013; Ferry 2013; Cohen 2015; Archer 2018; Fritts and Miller 2020). They are morally optional and yet morally better than some other permissible act. Suppose, for example, that a child is about to be crushed by a falling building. You can save the child, but you know your arms will be destroyed in the process. Alternatively, you can stand by and do nothing. Saving the child is supererogatory. It would be morally best, but you are permitted to do nothing instead.

Or, at least, so say many philosophers.⁠Most philosophers who have written on the topic believe that some acts are supererogatory. Exceptions include (Moore 1903; New 1974; Pybus 1982; Feldman 1986; Hale 1991; Fritts and Miller 2020). But there’s a paradox here, dating back to at least Sidgwick (1907). If saving the child is morally best, why is it not morally required? What stops it from being morally required? Recently, two more paradoxes of supererogation have occupied philosophers’ attention: Horton’s All or Nothing Problem (2017) and Kamm’s Intransitivity Paradox (1985; 2001, chap. 12; 2007, 30–31). Muñoz (2020) suggests that we discuss the three puzzles together. I follow his lead here. But while Muñoz takes the lesson of the trio to be that permissibility depends on what he calls ‘comparative prerogatives,’ I come to a different conclusion. There are no supererogatory acts. Morality requires us to do what’s (morally) best.

This antisupererogationist view is counterintuitive, so it might seem to incur a large debt from the start. But, I argue, this debt is not as large as it appears. For to say that all morally suboptimal acts are wrong is not to say that all such acts make their agents blameworthy. Wrongness and blameworthiness can come apart.

I then argue that antisupererogationism can more than repay the debt that remains. If morality requires us to do what’s best, the three paradoxes of supererogation evaporate. The residual puzzles are about blameworthiness, and they turn out to be not very puzzling at all.

Before we reach that point, though, we must acquaint ourselves with the three paradoxes. Following Muñoz, I will explain the drawbacks of extant solutions as we go.

2. The Three Paradoxes of Supererogation

Sidgwick’s account of the original paradox is as follows:

Certainly we should agree that a truly moral man cannot say to himself, ‘This is the best thing on the whole for me to do, but yet it is not my duty to do it though it is in my power’: this would seem to common sense an immoral paradox. (1907, 220)

Why did Sidgwick consider this a paradox? It might be because he accepted the following claim:

The Good-Ought Tie-Up⁠This moniker comes from Heyd (1982, 4).

If an act is morally better than each available alternative, then we are morally required to perform that act.⁠It is important to note that ‘morally better’ does not mean the same thing as ‘has better consequences.’ On some deontological theories, moral betterness and betterness of consequences come apart. The paradox of supererogation is not just a problem for consequentialist theories. I often drop the ‘morally’ in what follows, but it should be remembered that my concern is moral bestness, rather than bestness of consequences.an act A can be morally better than an act B even though A does not have better consequences than B.It is important to note that what I mean by an act’s being morally better than each alternative is not the same as that act’s having better consequences than each alternative. A deontological theory might claim that your killing an innocent person would have best consequences without claiming that this act is morally best. I don’t have a theory-neutral and complete definition of ‘morally better’ in this sense. On deontological theories, the ‘morally better than’ relation over actions may be incomplete. But it seems to me that in my example above involving saving a child from a falling building, all theories should admit that saving the child is morally better than letting the child die. Similarly for my examples below. I’ll often drop the ‘morally’ in what follows, but it should be remembered that I’m talking about moral bestness, rather than the bestness of consequences.It is important to note that an act’s being morally better than each alternative is not the same as that act’s having better consequences than each alternative. On some deontological theories, these come apart. A theory might admit, for example, that in some situation killing an innocent person would have best consequences but not call this action morally best.Worth noting that morally better and moral bestness isn’t consequentialist. The morally best thing is not necessarily the thing with best consequences. A deontological theory might claim that your killing an innocent person would have best consequences wihtout claiming that this act is morally best. I’ll often drop the ‘morally’ in what follows, but it should be remembered that I’m talking about moral bestness, rather than the bestness of consequences.

This claim seems plausible, but it rules out the possibility of supererogation. That is because supererogatory acts are defined to be those that are morally better than some other permissible act. If some act besides the morally best is permissible, then the morally best cannot be required. The original paradox, then, is the conjunction of the Good-Ought Tie-Up and the claim that some acts are supererogatory (cf. Dreier 2004, 148; Portmore 2008, 378–81; Horgan and Timmons 2010, 36–38; Heyd 2019, sec. 3).

Since acts like saving a child at the cost of one’s arms certainly seem supererogatory, it is tempting to deny the Good-Ought Tie-Up. One way of doing that is to embrace a satisficing conception of morality (Slote 1985, chap. 3). On these views, we are not required to do what is (morally) best. We need only do what is (morally) good enough.⁠See Hurka (1990) for a distinction between absolute and comparative satisficing. My objections apply to both forms of the view. Returning to our falling building example, advocates of satisficing views can claim that both saving the child and doing nothing clear the bar, so that both acts are permissible. And since saving the child is morally better than doing nothing, it qualifies as supererogatory.

Satisficing views thus offer a thrifty account of supererogation. But there are problems. The first concerns the position of the line that separates permissible acts from impermissible acts. The Good-Ought Tie-Up gives us a principled place to draw the line: only the best is permissible. Satisficing views draw it somewhere lower down the scale. But where exactly? Suppose they draw it 70% of the way, measured by value, from the worst available act to the best available act. Why not 71% instead? No relevant reason favours the former position over the latter. In light of this dearth of reasons, any position between 0% and 100% will look unacceptably arbitrary.⁠Advocates of satisficing views might claim that some acts are indeterminately permissible. This move will not solve the problem since it will remain arbitrary where the range of indeterminacy begins and ends.One way of avoiding arbitrariness in these kinds of cases is to claim that the position of the line/range is determined by our linguistic conventions. That seems plausible for properties like baldness, but not for permissibility. We cannot discover what is permissible simply by consulting linguistic data (cf. Dougherty 2014, 370; Schoenfield 2016, 265).Another way of avoiding arbitrariness is to tie permissibility to blameworthiness (Chappell 2019). I argue below that permissibility and blameworthiness should be kept apart.A final option is to claim that it is a brute fact that — say — acts above the 70% line are permissible while those at or below the line are not. The existence of this kind of fact would be very mysterious.Many moral views — like certain kinds of deontological/non-consequentialist view — do not allow talk of an act being 70% as good as another act. A satisficing version of those views might instead draw the line at — say — the ninth-best act available to the agent. This line also seems arbitrary.

The second paradox of supererogation — Horton’s All or Nothing Problem (2017) — presents another difficulty.⁠Pummer (2016) presents a similar case. Sinclair (2018) discusses Pummer’s and Horton’s cases together.Pummer 2021 on cases ‘similar or relevant’ to Horton’s All-or-Nothing Problem: For similar or relevant cases, see Charles Fried and Derek Parfit, “Correspondence,” Philosophy and Public Affairs 8 (1979): 393–97; Derek Parfit, “Future Generations: Further Problems,” Philosophy and Public Affairs 11 (1982): 113–72, 131; Kagan, Limits of Morality, 16; Douglas Portmore, “Dual-Ranking Act-Consequentialism,” Philosophical Studies 138 (2008): 409–27, 420–21; Victor Tadros, The Ends of Harm: The Moral Foundations of Criminal Law (Oxford: Oxford University Press, 2011), 161–62; Theron Pummer, “Whether and Where to Give,” Philosophy and Public Affairs 44 (2016): 77–95; Joe Horton, “The All or Nothing Problem,” Journal of Philosophy 114 (2017): 94–104; Jeff McMahan, “Doing Good and Doing the Best,” in The Ethics of Giving, ed. Paul Woodruff (New York: Oxford University Press, 2018), 78–102; Thomas Sinclair, “Are We Conditionally Obligated to Be Effective Altruists?,” Philosophy and Public Affairs 46 (2018): 36–59; Portmore, Opting for the Best, sec. 6.4; Helen Frowe, “If You’ll Be My Bodyguard: Agreements to Save and the Duty to Minimize Harm,” Ethics 129 (2019): 204–29, 210; Ralf Bader, “Agent-Relative Prerogatives and Suboptimal Beneficence,” Oxford Studies in Normative Ethics 9 (2019): 223–50; Tina Rulli, “Conditional Obligations,” Social Theory and Practice 46 (2020): 365–90; Muñoz,“Three Paradoxes of Supererogation”; Christian Barry and Seth Lazar, “Supererogation and Optimisation” (unpublished); Kerah Gordon-Solmon, “Between All and Nothing: Or, Defending the Impermissible” (unpublished). Not all of these authors discuss cases with the same structure as those that are of interest to me here. In some of these cases, one faces a series of choices over time: first one can either incur a personal cost to help others or not, and if one does incur the cost, one can then at a later time choose whether to help more at no additional personal cost. In the cases that are of interest to me here, one faces a single choice in which one can incur a personal cost to help others, incur a personal cost that is no greater to help more, or incur no personal cost by helping no one at all. The dialectical importance of the distinction between these sorts of cases in part depends on whether it is possible to argue for conclusions about the single-choice cases on the basis of claims about the series-ofchoices cases (for discussion that bears on the extent to which it is, see McMahan, “Doing Good”; Bader, “Agent-Relative Prerogatives”). — Pummer, 2021 Suppose this time that two children are about to be crushed by the falling building. You can save both, save one, or do nothing. Whether you save both children or just one, your arms will be crushed. As before, doing nothing seems permissible. You are not required to sacrifice your arms. But the presence of the extra child seems to make saving just one child wrong. After all, you could save both at no extra cost. However, saving one child still seems morally better than saving none. The result is that the wrong act of saving one is morally better than the permissible act of saving none. But that seems absurd. Doing wrong cannot be morally better than doing right. We find ourselves in the grip of another paradox.⁠One might think that non-consequentialists can deny the principle that ‘doing wrong cannot be morally better than doing right.’think that doing wrong can sometimes lead to better consequences than doing right. ButNon-consequentialists might deny that saving one is morally better than saving none. Although saving one has better consequences, non-consequentialists say that moral value of actions is not just a function of consequences. But here it seems difficult to deny. Non-consequentialists usually value consequences, It’s just that that’s not all they value. If maximising good consequences violates some constraint, it ought not be done. But saving one doesn’t violate any constraint that saving none respects.Nonconsequentialists can happily say that, in some cases, doing wrong can have better consequences than doing right, But recall that I’m using morally better to mean ‘more choiceworthy than’ [MAYBE YOU COULD DENY THAT SAVING ONE IS MORE CHOICEWORTHY THAN SAVING NONE]. Put another way, the standard nonconsequentialist reason for thinking that doing what has better consequences can be wrong is if doing what has better consequences violates some deontological constraint and doing what has worse consequences does not. But if saving one violates a deontological constraint, then so does saving none.

This problem is especially acute for satisficing views. Saving one is higher on the scale of moral betterness than saving none, so there is no way of positioning the line such that doing nothing comes out permissible and saving one comes out wrong.⁠Therefore, if doing nothing is permissible, saving one is permissible. Satisficing views thus permit the gratuitous prevention of goodness (Bradley 2006). So long as we hold fast to the intuition that saving none is permissible, the lesson of Horton’s puzzle seems to be that permissibility depends on more than an act’s place along a scale.

Hence the move to two-dimensional theories of permissibility. On these views, supererogatory acts outstrip merely permissible acts along one dimension. But there is a second, independent, deontically-relevant dimension along which acts vary. And supererogatory acts are not required because their strength along the first dimension is outweighed by the strength of some merely permissible act once we take into account this second dimension. These dimensions have gone by different names in the literature — requiring versus justifying reasons (Gert 2007; 2012; Archer 2016b), impartial versus partial reasons (Parfit 2011), moral versus non-moral reasons (Slote 1991; Portmore 2011), and reasons versus prerogatives (Scheffler 1982; Muñoz 2020) — but the form of each solution is the same. For ease of exposition, I use the language of reasons and prerogatives, but the points below apply equally to other two-dimensional solutions.⁠Raz (1975), Dancy (1993), Bedke (2011), and Portmore (2019) also offer two-dimensional theories of supererogation.

To see how reasons and prerogatives help solve Horton’s problem, consider the following table:

	Reasons	Prerogatives
Save Two	10	0
Save One	5	0
Do Nothing	0	10

One act is morally better than another iff it is favoured by greater reasons, so saving two is better than saving one, and both are better than doing nothing. But an act’s deontic status depends on both reasons and prerogatives. An act is permissible iff the sum of reasons and prerogatives in its favour is at least as great as the reasons in favour of each other available act. Otherwise, it is impermissible. That implies that saving two and doing nothing are both permissible, while saving one is impermissible. And since saving two is morally better than doing nothing, saving two comes out supererogatory.

This is a neat solution, but it flounders on the third paradox of supererogation: Kamm’s Intransitivity Paradox (1985; 2001, chap. 12; 2007, 30–31). This paradox, too, can be rendered as a variant of our original case. Suppose that you promised to meet a friend for lunch, and you encounter the child about to be crushed by the falling building on your way to the restaurant.⁠In Kamm’s (1985) rendering, the supererogatory act is donating a kidney. I follow Muñoz’s (2020) presentation. If the choice is between saving the child at the cost of your arms and keeping your promise, saving the child seems supererogatory. It would be good to make the sacrifice — good enough to make breaking your promise permissible — but it is not required. Return now to our original case, in which you can save the child or do nothing. In that case, we decided it would be permissible to do nothing. You are not required to sacrifice your arms. But now suppose that there is no falling building. You are on your way to lunch, and you can either keep your promise or do nothing. In this case, you are required to keep your promise. Doing nothing would be wrong.⁠Hence why it has been called Kamm’s Intransitivity Paradox. It appears to show that permissibility is intransitive. When the other option is keeping your promise, saving one is permissible. When the other option is saving one, doing nothing is permissible. But when the other option is keeping your promise, doing nothing is impermissible.

These judgements seem plausible, but we cannot capture them with our system of reasons and prerogatives. If saving the child is supererogatory when you could keep your promise, then the reasons in favour of saving the child must outweigh the reasons in favour of keeping your promise. If doing nothing is permissible when you could save the child, then the combined reasons and prerogatives in favour of doing nothing must at least match the reasons in favour of saving the child. But then simple arithmetic dictates that the combined reasons and prerogatives in favour of doing nothing must outweigh the reasons in favour of keeping your promise. That would make doing nothing rather than keeping your promise permissible.

Kamm’s solution (2001, chap. 12) dispenses with reasons and prerogatives. Instead, she employs two different dimensions: efforts and precedence.⁠Lazar (2019, 246) presents a similar view. The efforts standard measures the effort that an act requires of us. The precedence standard dictates when it is permissible to let one act take precedence over another. We can assign scores along these dimensions as follows:

	Efforts	Precedence
Keep Promise	10	0
Save One	5	10
Do Nothing	0	5

We can say that an act is permissible iff its precedence score is at least as great as each other act’s effort score. An act is impermissible otherwise. That criterion gives us the deontic judgements we want: you are permitted to save one rather than keep your promise, and do nothing rather than save one, but you are not permitted to do nothing rather than keep your promise.

This too seems like a neat solution. But a problem arises when we consider Muñoz’s combined case (2020, 8). Suppose that you are on your way to your lunch-date, as before, when you encounter two children set to be crushed by the falling building. You can save both, save one, save none by keeping your promise, or save none by doing nothing. Whether you save both or save one, your arms will be crushed. That makes saving just one child seem impermissible, so we have to set the effort score of saving two higher than the precedence score of saving one. But this move sets us off on a long and ultimately contradictory path. An effort score that high makes it impermissible to do nothing when you could save two. To avoid that implication, we have to set the precedence score of doing nothing as high as the effort score of saving two. But a precedence score that high makes it permissible to do nothing rather than keep your promise. To avoid that implication, we have to set the effort score of keeping your promise higher than the precedence score of doing nothing. But an effort score that high makes it impermissible to save one rather than keep your promise. To avoid that implication, we have to set the precedence score of saving one higher than the effort score of keeping your promise. But a precedence score that high makes it permissible to save one child rather than two. To avoid that implication, we have to raise the effort score of saving two again, and then we are back where we started. There is simply no way of assigning two-dimensional scores that deliver our intuitive permissibility judgements.

How about three dimensions? Archer’s (2016b) model posits requiring, justifying, and favouring reasons.⁠Horgan and Timmons (2010, 54) endorse a similar view. Permissibility depends on requiring and justifying reasons: an act is permissible iff it is supported by justifying reasons at least as great as each other act’s requiring reasons. But moral betterness depends on favouring reasons: an act is better than another iff it is supported by greater favouring reasons. That allows us to capture the judgements in Kamm’s case as follows:

	Requiring	Justifying	Favouring
Keep Promise	8	8	8
Save One	5	10	10
Do Nothing	0	5	0

You are permitted to save one rather than keep your promise, do nothing rather than save one, but not to do nothing rather than keep your promise. And saving one comes out supererogatory because it is supported by the strongest favouring reasons.

However, even these three dimensions fail to give us what we want in Muñoz’s combined case. Introducing the option to save two children makes saving one impermissible, so the requiring reasons to save both children must exceed the justifying reasons to save one child. But this move sets us on the road to another contradiction. Setting the requiring reasons to save both children that high makes it impermissible to do nothing rather than save both children. To avoid that implication, we have to set the justifying reasons to do nothing at least as high, but that makes it permissible to do nothing rather than keep your promise. To avoid that implication, we have to set the requiring reasons to keep your promise higher, but that makes it impermissible to save one child rather than keep your promise. To avoid that implication, we have to set the justifying reasons to save one at least as high, but that makes it permissible to save one child rather than both, and the circle begins again. Even Archer’s three dimensions cannot account for our intuitive deontic judgements.

3. Comparative Prerogatives

Muñoz (2020, 8–10) spares us a jump to four dimensions. Instead, he argues that reasons and prerogatives can give us all that we want. The key difference is that Muñoz’s prerogatives are not absolute. They are comparative. The strength of your prerogative to perform an act depends on the act with which it is being compared.

Returning to the three paradoxes, Muñoz claims that you have a comparative prerogative to do nothing rather than save the children. That is because the two acts differ in their harm to you (2020, 9). Saving the children would crush your arms, whereas doing nothing would leave all your limbs intact. But, Muñoz claims, you have no such comparative prerogative to do nothing rather than keep your promise, because the harm to you is the same either way. Neither act would crush your arms. Similarly, you have no comparative prerogative to save one rather than save two, because the harm to you is the same. Both acts would crush your arms. So, Muñoz concludes, a comparative prerogative based on personal harm generates our intuitive deontic judgements in all three paradoxes and the combined case.

Muñoz’s solution is clever, but there are drawbacks. The first is a problem it shares with satisficing views: the strength of comparative prerogatives seems unavoidably arbitrary. The personal harm of losing both of your arms may render doing nothing permissible, but what if the harm were marginally less? Suppose that in a second case, the falling building would crush your arms 99% of the way up to your shoulders; in a third case, it would crush them 98% of the way; and so on. In the last case, the building would just scrape the nail of your middle finger. At some point in this sequence of cases, doing nothing must become impermissible. But Muñoz gives us no reason for drawing the line in one place rather than any other.⁠Claiming that permissibility is vague does not solve this problem. See footnote 5.

A second drawback is what comparative prerogatives commit us to saying. Along with other multi-dimensional views, it implies that — in some cases — performing an impermissible act is morally better than performing a permissible act. Horton’s problem is one such case. Saving just one child is wrong and yet morally better than the permissible act of doing nothing. That commits advocates of the comparative prerogatives view to the jarring claim that, sometimes, it is morally better to do the wrong thing than a right thing.⁠Muñoz (2020) recognises this implication and claims that we should embrace it. I do not take it to be fatal for multi-dimensional views, but it is certainly strange. As we will see in the next section, antisupererogationism commits us to no such oddness.

The ontological commitments of the comparative prerogatives view are a final drawback. Muñoz writes that his view offers a ‘unified and minimal’ solution to the paradoxes (2020, 10), and it certainly seems that way at first look. Rather than add a fourth dimension to defuse his combined case, Muñoz pares us back to reasons and comparative prerogatives. But the view is not as parsimonious as it might seem. Note first that the strength of a comparative prerogative must supervene on features of the acts being compared. There cannot be a difference in the strength of some comparative prerogative without a difference in some feature. Therefore, since you have a comparative prerogative to do nothing rather than save the children but have no comparative prerogative to do nothing rather than keep your promise, saving the children and keeping your promise must differ in some deontically-relevant way. As we noted above, the difference is in their personal harms. Keeping your promise leaves all your limbs intact. So, Muñoz’s view posits at least two deontically-relevant dimensions: the reasons in favour of an act, and the personal harm to the agent.

But we cannot cap the dimensions there. As Muñoz acknowledges, personal harm cannot be the only ground of comparative prerogatives. To see why, consider the following set of cases adapted from Dorsey (2013, 365–69). First, you can donate £30,000 of your own money to an effective charity, thereby saving ten lives, or you can buy a new car. In this case, buying a new car seems permissible. Second, you can steal £30,000 from rich Jerry as the only means to save ten lives, or you can respect Jerry’s rights and leave him alone. In this case, stealing from Jerry seems permissible. Third, you can steal £30,000 from rich Jerry and use the money to buy a new car, or you can respect Jerry’s rights and leave him alone. In this case, stealing from Jerry seems impermissible.

Suppose, for reductio, that reasons and personal harms are the only deontically-relevant dimensions. In the first case, the reasons to save ten lives outweigh the reasons to buy a new car. But missing out on a new car is a personal harm. And since buying a new car is permissible, the comparative prerogative generated by this personal harm (plus the reasons to buy a new car) must outweigh the reasons to save ten lives. That is the first component of the reductio. In the second case, there is no difference in personal harm. Stealing from Jerry to save ten lives and respecting Jerry’s rights both involve no harm to you. Therefore, since stealing from Jerry to save ten lives is permissible, the reasons to save ten lives must outweigh the reasons to respect Jerry’s rights. That is the second component of the reductio.

These two components have an undesirable implication. Saving ten lives in the first case and respecting Jerry’s rights in the third case both involve the same personal harm: you miss out on a new car. So, the comparative prerogative in favour of buying a new car in the first case and stealing from Jerry in the third case must be the same. But then, since the reasons and comparative prerogatives to buy a new car outweigh the reasons to save ten lives in the first case, and the reasons to save ten lives outweigh the reasons to respect Jerry’s rights in the second case, the reasons and comparative prerogatives to buy a new car must outweigh the reasons to respect Jerry’s rights in the third case. That permits you to steal from Jerry and buy yourself a new car.

To avoid this implication, Muñoz posits a third deontically-relevant dimension (2020, 10). Comparative prerogatives also depend on whether the compared acts violate someone’s rights (cf. Kamm 2007, 30–31; Hurka and Shubert 2012, 10). That forbids you from stealing from Jerry, but it comes at the cost of parsimony. It means Muñoz’s view requires at least as many deontically-relevant dimensions as Archer’s.

And even three dimensions cannot capture all of our common-sense intuitions about supererogation. Consider another Kamm-style example. First, you can save a child drowning in a shallow pond, thereby ruining your £3,000 watch, or you can stay dry.⁠This case is a variant of Singer’s (1972). In this case, staying dry seems impermissible. The comparative prerogative based on personal harm is not strong enough to justify letting the child die. Second, you can save a child by donating £3,000 to an effective charity, or you can keep the money. In this case, keeping the money seems permissible. But note that the reason to perform the altruistic action is the same in each case: a child will be saved. The personal harms are the same too: you have £3,000 less than you otherwise would. And, plausibly, the cases are also the same with respect to rights-violations: if the child in the shallow pond has a right to your help, so does the child who would benefit from your charity. So, on a three-dimensional view, either both altruistic acts are required or neither of them are. Therefore, if we want to preserve all of our common-sense intuitions about supererogation, we need to posit at least four deontically-relevant dimensions. The result is a view that no longer looks so unified and minimal.

Of course, we might not want to preserve the intuition that donating our spare cash to charity is supererogatory. Perhaps Singer is right that the world’s affluent are morally required to ‘give until it hurts.’⁠This phrase is from Noggle (2009). Singer’s own wording is that I must give until I ‘reach the level… at which, by giving more, I would cause as much suffering to myself or my dependents as I would relieve by my gift.’ (1972, 241). That would let us get by with three deontically-relevant dimensions. But it would also mean admitting that a large swathe of our common-sense intuitions about supererogation are mistaken. That admission would cast doubt on our other intuitions about supererogatory action, and open the door to the view that no acts are supererogatory. I defend this antisupererogationist view in the next section.

4. Blameworthiness

Suppose that there are no supererogatory acts. Morality requires us to do what’s best. That means that, when you come upon the falling building, you are required to sacrifice your arms to save the children. Doing nothing is impermissible.

This view is often taken to be counterintuitive. But what makes it so? A point that comes up again and again in the literature is that you would not be blameworthy for failing to save the children.⁠See, for example, McElwee (2010, 396):See also (Urmson 1958, 202–3; Raz 1975, 64–65; Jamieson and Elliot 2009, 243; Horgan and Timmons 2010, 48; Ferry 2013, 586; Heyd 2019, sec. 3.1; McElwee 2020, 200). Were you to stand by and do nothing, it would be inappropriate for others to blame you and inappropriate for you to blame yourself.

However, blamelessness is not the same as permissibility. To make the inference from your blamelessness in doing nothing to the permissibility of doing nothing, we need an extra premise: all impermissible acts render their agent blameworthy.⁠Arneson (2004, 51) recognises this point. As does Archer (2016a, 183). And, I argue, this premise is false. Wrongness and blameworthiness can come apart.⁠Fritts and Miller (2020, 16) point this out in their own defence of antisupererogationism. Driver (1992) argues that permissible acts can be blameworthy. Pummer (2021) argues that impermissible acts can be overall praiseworthy.

Before I present that argument, though, I need to say a little more about what it means to be blameworthy. This notion can be understood in at least two distinct ways. First, we can take a person to be blameworthy when it would be fitting to blame them. That is, when they deserve blame. We might analyse this idea in the standard way: blame is fitting when an agent does wrong voluntarily, under control, and knowingly.⁠See Shoemaker (2017, 498–99), who cites Brink and Nelkin (2013, 292–303) as advocates of this kind of view. See also McKenna (2012, 12–13) who writes of epistemic and control conditions for moral responsibility. Or we might offer a quality-of-will account: blame is fitting when an agent does wrong freely and knowingly, and in doing so reveals an objectionable lack of regard for others and for moral considerations (McKenna 2012, chap. 3). Alternatively, we might take blame to be fitting when an agent merits indignation or anger (Portmore 2011, 43; Shoemaker 2017, 508). Or we might claim that there is no simple analysis of fitting blame, and content ourselves with identifying certain common features (McElwee 2010, 401–5; Fricker 2016, 168).

Second, we can take a person to be blameworthy when it would be expedient to blame them. That is, when blaming them would have good consequences. This might be because blame would alert the agent to their wrongdoing and motivate them to do better in the future. It might also deter others from acting in the same way.⁠An anonymous reviewer suggests the following problem for this kind of account. Suppose Anwar is responsive to blame, while Bella is stubborn. Then there may be actions that it would be expedient to blame Anwar for and not expendient to blame Bella for. Thus, on the expedient blame account, Anwar is blameworthy for such actions while Beth is not. That seems counterintuitive. I take this case to show that the fitting blame account better captures what we mean by ‘blameworthy.’ Nevertheless, I discuss both fitting blame and expedient blame accounts bleow3, to cover my bases.

These two notions of blameworthiness differ. Sometimes, blame is expedient but not fitting. A coach might blame her best player after a defeat, knowing that she responds best to criticism. At other times, blame is fitting but not expedient. A debtor might decline to blame a short-tempered gangster for spilling red wine on his shirt. Thankfully, I need not pin my case to just one of these accounts. Whether blameworthiness is about fittingness, expediency, or a conjunction or disjunction of the two, wrongness and blameworthiness can come apart. The only account incompatible with my argument is that according to which an agent is blameworthy iff they act wrongly.⁠Strawson (1962) offers a view along these lines, as do (Wallace 1994; Watson 1996; 2004; Fischer and Ravizza 1998; Darwall 2006; McElwee 2010; Portmore 2011; Brink and Nelkin 2013; Schulte 2014; Pereboom 2014; Menges 2020). See Mason (2019) for recent work on the connection between wrongness and blameworthiness.Pummer 2021: It is commonly held that unexcused impermissible acts are necessarily blameworthy, not praiseworthy. [[See, e.g., Allan Gibbard, Wise Choices, Apt Feelings (Oxford: Clarendon, 1990), chap. 3; John Skorupski, The Domain of Reasons (Oxford: Oxford University Press, 2010), pt. 3; Stephen Darwall, Morality, Authority, and Law: Essays in Second-Personal Ethics I (Oxford: Oxford University Press, 2013), 21; Brian McElwee, “Demandingness Objections in Ethics,” Philosophical Quarterly 67 (2017): 84–105.]] On some views, what it means for an act to be impermissible is that it is blameworthy, absent excusing conditions. On other views, all unexcused impermissible acts are blameworthy, even if what it means for an act to be impermissible is not that it is blameworthy, but, for example, that it mustn’t be done. [[See Derek Parfit, On What Matters (Oxford: Oxford University Press, 2011), vol. 1, chap. 7.]] I now argue that we have good reason to reject this account.

Consider the following case, from Cohen (2015, 1006–7):

Your brother is suffering from a rapidly progressive lung disease, which is fatal without treatment. The only treatment is lung transplantation. Lung donation (unlike kidney donation) incurs, beyond the hardship and hazards of a major surgery, a significant impact on one’s prospective quality of life (e.g., no ability to play sports anymore). You are the sole matching donor. If you decide to donate your lung, you will be praised as having done an incredibly selfless deed—you made yourself handicapped out of caring! You will be rightly lauded as having acted especially nobly. If, on the other hand, you decide not to donate, you will be considered to have acted wrongly: you did not just refrain from doing what is noble; you rather miserably failed to do the right thing—you failed to act even though you were the only one who could help, thus facilitating the death of your own brother.⁠Cohen uses this case to argue for the existence of a new category of moral action: forced supererogation. I do not draw that conclusion. I use the case only to argue that wrongness and blameworthiness can come apart. Cohen (2015, 1007) also offers two more cases with a similar structure. I find them equally convincing as examples of wrongdoing that is not blameworthy, but leave them out for reasons of space. Driver (1992) presents a similar case and argues that the agent acts permissibly but is blameworthy!

Here is another case, from McKenna (2012, 19):

Consider the case of a woman whose clear moral obligation is to save a small group of people from certain death. As it happens, her child is in danger of a nontrivial harm, though not life-threatening. She cannot both save the group and aid her child, and in the absence of any “agential impediments,” she decisively settles upon protecting her child. She freely does so knowing that she is violating her obligation.

McKenna goes on to write:

Here, her failure to save the group counts as morally wrong, but there is nothing negative revealed in the quality of her will in terms of her regard for others’ morally relevant interests. Her reasons for acting showed no lack of moral concern for these others, and so on. In this sort of case I believe that we can make good sense of a person knowingly and freely doing morally wrong but yet not being blameworthy since the quality of her will does not show that in so acting she lacks due regard for others. She was morally responsible for what she did, but she was not blameworthy for it. She was, as one might put it, in a tight spot. (2012, 19)

I agree with McKenna’s appraisal, and submit that we should draw the same conclusion in Cohen’s case: refusing to donate a lung to your brother would be wrong, but you need not be blameworthy for your refusal.

We can flesh out the case for these judgements in terms of both fitting and expedient blame. Take fitting blame first. Neither the prospective lung donor nor the mother deserve blame for their acts. That might be because their choice was not under control. Although it was voluntary, in the sense that they were not constrained or coerced by external forces, they nevertheless could not have brought themselves to perform the morally best act. If we accept what I have called the standard analysis of fitting blame, which requires acts to be performed voluntarily, under control, and knowingly, this lack of control renders the lung donor and mother blameless. If we instead accept a quality-of-will account, we will come to the same conclusion: the donor’s and mother’s acts — although wrong — do not reveal an objectionable lack of regard for others. A response-dependent theory of blameworthiness gives us the same answer: neither the donor nor the mother merit anger for their choice, so blame is not fitting.

Blame is not expedient either. The prospective lung donor and mother already recognise their wrongdoing, so blame does not serve the purpose of moral education in these cases. It also seems extremely unlikely that blame would motivate these agents to do better in future situations. Even if we suppose that the agents are likely to encounter similar circumstances in the future, the costs of acting rightly are so great — a lung in the donor case, an injury to a child in the mother case — that any memory of past blame will almost certainly fail to tip the balance. That means blame would likely make no difference. To criticise the prospective lung donor and mother for their choices would be to add a pointless scrap of ill-will to an already heart-rending situation.

Examples like Cohen’s and McKenna’s are the core reason for thinking that wrongness and blameworthiness can come apart. But there are other reasons too. Wrongness concerns the appropriateness of acts, while blameworthiness concerns the appropriateness of responses to those acts. Claiming that the two always go together, then, means claiming that no factor besides wrongness can affect whether blame is an appropriate response. It also means claiming that we can immediately infer blameworthiness from wrongness. That seems drastically contrary to our actual practices of blame. Determining whether blame is appropriate is often taken to be a difficult further question, requiring us to consider various extenuating factors.⁠The above arguments draw on widely-shared and somewhat-uncontroversial premises. But there are further arguments that depend on more contentious premises. For example, many ethicists believe that the permissibility of an act does not depend on the agent’s intentions (e.g. most consequentialists). Blameworthiness, meanwhile, does seem to depend on the agent’s intentions.Second, many ethicists believe that moral obligations are objective: an act’s wrongness does not depend on the agent’s beliefs (Moore 1903; Ross 1930; Feldman 1986; Thomson 1986; Zimmerman 1996; Graham 2010). Blameworthiness, meanwhile, does seem to depend on the agent’s beliefs. Suppose, for example, that two doctors kill their respective patients by administering a certain treatment. All else equal, the doctor who believed that the treatment would kill their patient is more blameworthy than the doctor who believed that the treatment would cure their patient.Third, many ethicists believe that moral dilemmas — in which one acts wrongly no matter what one does — are possible (Sinnott-Armstrong 1988). Suppose, for example, that Sophie must choose one of her children to be killed or else see them both killed. Perhaps every choice available to Sophie is wrong. But, plausibly, not every choice makes Sophie blameworthy (cf. McKenna 2012, 20).

5. Paradoxes of Blameworthiness

We are not always blameworthy for doing wrong. That is the upshot of the previous section. This fact means that antisupererogationism need not be counterintuitive. For it allows us to say that, although you act wrongly when you fail to save the child in the falling building case, you are not blameworthy for your failure. It would be inappropriate for others to blame you and inappropriate for you to blame yourself. The reasons why are the same as those in Cohen’s and McKenna’s examples above. Blame is not fitting, because failing to save the children does not reveal an objectionable lack of regard for others and does not merit anger. And blame is not expedient, because the costs of acting rightly are so great. Blame would not spur you to do better in similar future situations, so any criticism for your act would be a pointless extra harm.

It might be objected that this antisupererogationist view is still counterintuitive. Some people might have strong intuitions that, not only is failing to save the children not blameworthy, it is not wrong. I find that my intuitions about wrongness in this case, once set apart from my intuitions about blameworthiness,⁠Which includes self-blame, remember! are faint. But even if antisupererogationism still comes at some intuitive cost, I argue that this cost is more than repaid by the view’s advantages.

The first advantage is the ease with which antisupererogationism solves the three paradoxes. In the original paradox, there is no mystery about how doing nothing can be permissible. It is not. In Horton’s problem, there is no mystery about how doing nothing can be permissible even though it is morally worse than the impermissible act of saving one child. Both acts are impermissible. In Kamm’s paradox, there is no mystery about how doing nothing rather than keeping your promise is impermissible, even though it is permissible to save the child rather than keep your promise, and permissible to do nothing rather than save the child. This last clause is false.

Those questions that are left are about blameworthiness. We can phrase them as follows. First, the original paradox: if saving the child is morally best, why are you not blameworthy for doing nothing? Second, Horton’s problem: if saving one child is morally better than doing nothing, why are you blameworthy for saving one child but not blameworthy for doing nothing? Third, Kamm’s paradox: if you are not blameworthy for saving one rather than keeping your promise, and not blameworthy for doing nothing rather than saving one, why are you blameworthy for doing nothing rather than keeping your promise?

These questions admit of easy answers. We need not posit any extra deontically-relevant dimensions: reasons that are requiring, justifying, or favouring, efforts standards or precedence standards, prerogatives that are absolute or comparative. We can use what we already have: ordinary reasons and blameworthiness. You have most reason to save the child in the original paradox, but you are not blameworthy for doing nothing. Blame is not fitting. The extreme difficulty of doing the right thing means that your wrong act does not reveal an objectionable lack of regard for others and does not merit anger. Nor is it expedient. Blame would almost certainly not induce you or others to act rightly in similar situations.

The same goes for doing nothing in Horton’s problem. If you stood by, blame would be neither fitting nor expedient. But if you saved just one child, blame would be fitting. Since you could bring yourself to sacrifice your arms, saving both children was under your control. Your refusal to save the second child at no extra cost thus reveals an objectionable lack of regard for others and merits anger. Blame is also more likely to be expedient in the case where you save just one child. It has a better chance of spurring you and others to do better in similar situations. That is because such blame need not overcome your fear of losing your arms, only whatever impulse prevented you from saving the second child.

Kamm’s paradox is also not much of a puzzle when interpreted in terms of blame. Blame is not fitting when you miss your promised lunch-date to save the child, because — most would agree — this is not an act of wrongdoing and does not merit anger. It is presumably not expedient either. It would be bad to discourage people from making altruistic sacrifices, even when those sacrifices involve blowing off lunch-dates. And blame is neither fitting nor expedient when you do nothing rather than save the child, for the reasons given above. But you are blameworthy when you do nothing rather than keep your promise. This is an act of wrongdoing performed voluntarily, under control, and knowingly. It reveals an objectionable lack of regard for others and merits anger. And blame would be expedient in motivating you to keep your promises in the future.

With these answers, all the mystery of the three paradoxes is dispelled. What’s more, the solution posits no theoretical machinery beyond reasons and standard notions of blameworthiness. Parsimony, then, is the first major advantage of the antisupererogationist view.

Here’s a second advantage. Recall that the comparative prerogatives view suffers from a problem of arbitrary line-drawing. The comparative prerogative based on personal harm permits you to let the child die when the cost is both your arms, but not when the cost is a bruised nail. The line separating permissible from impermissible acts must be drawn somewhere between these two poles. But we have no good reason to draw the line at — say — 42% of your arms rather than 41%. Any line we do draw will be unavoidably arbitrary.

The antisupererogationist view avoids this problem. ‘Only the best is permissible’ is a principled way of drawing the line. And there is no analogue of the problem in terms of blameworthiness because blameworthiness comes in degrees.⁠Sinhababu (2018) argues that rightness and wrongness are scalar properties that come in degrees (see also Norcross 2006; 2020). That view can avoid the problem of arbitrary line-drawing, but it does not vindicate supererogation as ordinarily defined. On the scalar view, morally better acts are also ‘righter,’ so there are no pairs of acts such that one is morally better than the other and both are permissible simpliciter. Blame can be more or less fitting, and more or less expedient. Take fitting blame first. In the first case in our sequence, failing to save the child does not reveal an objectionable lack of regard for others. But, plausibly, the acuteness of your lack of regard increases as we proceed down the sequence. Refusing to save the child to avoid a bruised nail would reveal a very acute lack of regard for others. As the acuteness increases, so too does the fittingness of blame. Similarly, you would not merit anger for failing to save the child in the first case, but anger becomes increasingly merited as your personal harm decreases. Were you to let the child die to avoid a crushed finger, you would merit anger to some degree. Were you to let the child die to avoid a bruised nail, you would merit anger to a greater degree.

Expedient blame also admits of degrees. In the first case, blame would not be expedient because it would not overturn your strong aversion to losing your arms. But as the personal cost decreases, the likelihood that blame would tip the balance and spur you to do better in similar situations increases, rendering blame more expedient. If it were just the prospect of a bruised nail holding you back from saving the child, then blame would be very expedient. So, whether we understand blameworthiness in terms of fittingness or expediency, blameworthiness comes in degrees. That means that antisupererogationism avoids the problem of arbitrary line-drawing.⁠There might seem to be some unavoidable arbitrariness in the point at which you move from entirely blameless to somewhat blameworthy. This problem can be addressed in (at least) two ways. First, can avoid arbitrariness by appealing to our view about the fittingness or expediency of blame. e.g. if we adopt a view on which an action is blameworthy if it reveals an objectionable lack of regard for others, we can say that an action becomes blameworthy when it first reveals an objectionable lack of regard. One might complain that this point is arbitrary. But here too there are two responses: less arbitrary than prerogatives view, and the problem is for the theory of blame, rather than the antisupererogationist view. Second, one could claim that you are blameworthy to at least some very small degree in all cases where you do less than the best. And hence slightly blameworthy even in the case where the building would crush all of your arms. Then there’s no line to draw as we move along the spectrum of cases. The only changes would be quantitative, rather than qualitative. I thank an anonymous reviewer for raising this problem, and for suggesting the second response.

A third advantage is one I hinted at above. Views on which permissibility depends on more than just reasons are committed to the strange-sounding claim that, in some cases, doing what is right is morally worse than doing what is wrong. Horton’s problem is one such case. Doing nothing is permissible, but it is morally worse than the impermissible act of saving one child. Antisupererogationism’s corresponding commitments are not nearly so odd. We need only claim that (1) in some cases, we are not blameworthy for doing what is wrong, and (2) in some cases, doing what is blameworthy can be morally better than doing what is not blameworthy. But the weirdness of (2) can be explained away by appeal to our theory of blame. If we adopt the standard view, for example, (2) translates as ‘in some cases, it can be better to do an action that reveals an objectionable lack of regard for others than an action that does not reveal an objectionable lack of regard for others.’ Horton’s All-or-Nothing Problem is one such case. It’s better to save the one child. But since you could have saved a second child at no extra cost, saving the one child reveals an objectionable lack of regard.

‘morally better actions can be more blameworthy than morally worse actions’. But that’s okay because the connection between moral betterness and blameworthiness is not nearly so tight as the connection between moral betterness and rightness/wrongness. Blameworthiness can depend on factors like intention, etc.

doing what is not blameworthy is morally worse than doing what is blameworthy. But the weirdness of (2) can be explained away by appeal to our theory of blame. If we adopt the standard view, for example, (2) translates as ‘in some cases, doing wrong in a way that doesn’t revela an objectionable lack of regard for others is morally worse than doing wrong in a way that does reveal an objectionable lack of regard for others.’

If we adopt a quality-of-will account, for example,

We might analyse this idea in the standard way: blame is fitting when an agent does wrong voluntarily, under control, and knowingly.⁠See Shoemaker (2017, 498–99), who cites Brink and Nelkin (2013, 292–303) as advocates of this kind of view. See also McKenna (2012, 12–13) who writes of epistemic and control conditions for moral responsibility. Or we might offer a quality-of-will account: blame is fitting when an agent does wrong freely and knowingly, and in doing so reveals an objectionable lack of regard for others and for moral considerations (McKenna 2012, chap. 3).

For example, if we adopt the ‘meriting anger’ view, (2) translates as: doing what doesn’t merit anger can be morally worse than doing what merits anger.

‘BLAMEWORTHY CAN BE MORALLY BETTER THAN BLAMELESS’ IS OKAY BECAUSE:

1. We have lots of examples besides ‘All-or-Nothing’

Examples of better acts that can be more blameworthy:

1. Someone more in control (e.g. not developmentally-stunted)

2. Someone with worse intentions

3. Someone more susceptible to blame

4. Someone who does something expected to be bad but turns out good

2. Relatedly, not such a tight connection between blameworthiness and moral betterness as there is between wrongness and moral betterness.

3. Blame can come in degrees. So claim could be weakened to ‘More blameworthy can be morally better than less blameworthy’.

4. We can appeal to our theory of blame to explain away.

e.g. not so confusing how a morally better action can be more blameworthy, given that the morally better action reveals an objectionable lack of regard for others and the morally worse action doesn’t. [BUT WHY CAN’T WE PULL THE SAME MOVE WITH WRONGNESS?]

5. Blameworthiness can be conditional on what we are willing to do, plausibly. That’s more implausible for wrongness. [See ‘Conditional blamelessness makes more sense than conditional permissibility’]

6. Demandingness

That completes my case for antisupererogationism. In sum, there are no supererogatory acts. Morality requires us to do what’s best. This view is less counterintuitive than it might first seem, because it does not imply that we are (always) blameworthy for doing less than the best. It also offers a principled and parsimonious solution to the three paradoxes of supererogation.

Nevertheless, one might well have objections. Here I will respond to just one: that antisupererogationism makes morality too demanding. Baier (1965, 109) writes that such views:

have the absurd result that we are doing wrong whenever we are relaxing, since on those occasions there will always be opportunities to produce greater good than we can by relaxing. For the relief of suffering is always a greater good than mere enjoyment. Yet it is quite plain that the worker who, after a tiring day, puts on his slippers and listens to the radio is not doing anything he ought not to, is not neglecting any of his duties, even though it may be perfectly true that there are things he might do which produce more good in the world, even for himself, than merely relaxing by the fireside.

However, we have already taken the sting out of this objection. If the worker could do better, then he is acting wrongly, but that is not to say he is blameworthy. He might even be praiseworthy. It might be fitting and expedient for others to praise him — and for him to praise himself — for doing as much as he does.⁠Pummer (2021) argues that impermissible actions can be overall praiseworthy.

Another response to Baier’s objection is related. As Ferry (2013, 587) writes, ‘Talk of morality’s ’demands’ involves a useful metaphor, but it can be misleading.’ That is because our ordinary notion of ‘demandingness’ implies some kind of sanction for failing to meet the operant demands, and morality does not impose sanctions. Consider an analogy. Suppose that your university keeps an internal record of exam scores. All scripts scoring 60 or above are classified as ■, while all other scripts are classified as □. This system is purely for record-keeping purposes. Students are not rewarded or sanctioned based on their ■/□ classification. Now suppose that the system is changed. In the new system, only those scripts scoring 100 are classified as ■. All others are classified as □. But there are no other changes in the system, and no change in the behaviour of the professors. Calling this new system ‘more demanding’ would strain the meaning of the word. Now suppose instead that the ■/□ classification remains the same, but the passing grade rises from 40 to 60. All students scoring less than 60 are forced to retake the course. This new system would be more demanding.

The move from supererogationism to antisupererogationism is like the first kind of change. More acts become impermissible, but that does not entail that more acts make their agents blameworthy. We can make the move without changing our views about which acts make blame appropriate, so antisupererogationism does not itself make morality more demanding. What would make morality more demanding is an increase in our blameworthiness standards.

But one might now worry that a notion of wrongness so divorced from the notion of blameworthiness expresses almost nothing, and that we are on our way towards an evaluative scalar view of morality in which notions of rightness and wrongness play no role.⁠Ferry (2013, 574–76) expresses this worry, as does Hooker (Ideal Code, Real World, p.75). Norcross (2006; 2020) embraces this kind of view. That may be. I am not entirely opposed to that conclusion, for it has been one of my aims in this paper to show that the notion of blameworthiness is more important in our everyday moral practice. But that said, I do think that there remains a place for notions of rightness and wrongness. An act is right if we have most reason to perform it, and wrong otherwise. That makes rightness a kind of lodestar in our everyday moral practice: not a bar to clear regularly, but an ideal to aim towards.

References

Archer, Alfred. 2016a. ‘The Supererogatory and How Not to Accommodate It: A Reply to Dorsey’. Utilitas 28 (2): 179–88.

———. 2016b. ‘Moral Obligation, Self-Interest and the Transitivity Problem’. Utilitas 28 (4): 441–64.

———. 2018. ‘Supererogation’. Philosophy Compass 13 (3).

Arneson, Richard J. 2004. ‘Moral Limits on the Demands of Beneficence?’ In The Ethics of Assistance: Morality and the Distant Needy, edited by Deen K. Chatterjee, 33–58. Cambridge: Cambridge University Press.

Baier, Kurt. 1965. The Moral Point of View. New York: Random House.

Bedke, Matt. 2011. ‘Passing the Deontic Buck’. In Oxford Studies in Metaethics: Volume 6, edited by Russ Shafer-Landau, 128–52. Oxford: Oxford University Press.

Bradley, Ben. 2006. ‘Against Satisficing Consequentialism’. Utilitas 18 (2): 97–108.

Brink, David O., and Dana K. Nelkin. 2013. ‘Fairness and the Architecture of Responsibility’. In Oxford Studies in Agency and Responsibility: Volume 1, edited by David Shoemaker, 1:284–313. Oxford: Oxford University Press.

Chappell, Richard Yetter. 2019. ‘Willpower Satisficing’. Noûs 53 (2): 251–65.

Chisholm, Roderick M., and Ernest Sosa. 1966. ‘Intrinsic Preferability and the Problem of Supererogation’. Synthese 16 (3/4): 321–31.

Cohen, Shlomo. 2015. ‘Forced Supererogation’. European Journal of Philosophy 23 (4): 1006–24.

Dancy, Jonathan. 1993. ‘Beyond the Call of Duty: Supererogation, Obligation and Offence by Gregory Mellema’. Philosophical Books 34 (1): 48–49.

Darwall, Stephen. 2006. The Second Person Standpoint: Morality, Respect, and Accountability. Cambridge, MA: Harvard University Press.

Dorsey, Dale. 2013. ‘The Supererogatory, and How to Accommodate It’. Utilitas 25 (3): 355–82.

Dougherty, Tom. 2014. ‘Vague Value’. Philosophy and Phenomenological Research 89 (2): 352–72.

Dreier, James. 2004. ‘Why Ethical Satisficing Makes Sense and Rational Satisficing Doesn’t’. In Satisficing and Maximizing: Moral Theorists on Practical Reason, edited by Michael Byron, 131–54. Cambridge: Cambridge University Press.

Driver, Julia. 1992. ‘The Suberogatory’. Australasian Journal of Philosophy 70 (3): 286–95.

Feldman, Fred. 1986. Doing the Best We Can: An Essay in Informal Deontic Logic. Dordrecht: D. Reidel Publishing Company.

Ferry, Michael. 2013. ‘Does Morality Demand Our Very Best? On Moral Prescriptions and the Line of Duty’. Philosophical Studies 165 (2): 573–89.

Fischer, John Martin, and Mark Ravizza. 1998. Responsibility and Control: A Theory of Moral Responsibility. New York: Cambridge University Press.

Fricker, Miranda. 2016. ‘What’s the Point of Blame? A Paradigm Based Explanation’. Noûs 50 (1): 165–83.

Fritts, Megan, and Calum Miller. 2020. ‘Must We Be Perfect?: A Case against Supererogation’. Inquiry, 1–30.

Gert, Joshua. 2007. ‘Normative Strength and the Balance of Reasons’. The Philosophical Review 116 (4): 533–62.

———. 2012. ‘Moral Worth, Supererogation, and the Justifying/Requiring Distinction’. The Philosophical Review 121 (4): 611–18.

Graham, Peter A. 2010. ‘In Defense of Objectivism about Moral Obligation’. Ethics 121 (1): 88–115.

Hale, Susan C. 1991. ‘Against Supererogation’. American Philosophical Quarterly 28 (4): 273–85.

Heyd, David. 1982. Supererogation: Its Status in Ethical Theory. Cambridge: Cambridge University Press.

———. 2019. ‘Supererogation’. In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, Winter 2019.

Horgan, Terry, and Mark Timmons. 2010. ‘Untying a Knot from the Inside Out: Reflections on the “Paradox” of Supererogation’. Social Philosophy and Policy 27 (2): 29–63.

Horton, Joe. 2017. ‘The All or Nothing Problem’. The Journal of Philosophy 114 (2): 94–104.

Hurka, Thomas. 1990. ‘Two Kinds of Satisficing’. Philosophical Studies 59 (1): 107–11.

Hurka, Thomas, and Esther Shubert. 2012. ‘Permissions To Do Less Than the Best: A Moving Band’. In Oxford Studies in Normative Ethics: Volume 2, edited by Mark Timmons, 1–27. Oxford: Oxford University Press.

Jamieson, Dale, and Robert Elliot. 2009. ‘Progressive Consequentialism’. Philosophical Perspectives 23 (1): 241–51.

Kamm, Frances Myrna. 1985. ‘Supererogation and Obligation’. The Journal of Philosophy 82 (3): 118–38.

———. 2001. Morality, Mortality: Volume II: Rights, Duties, and Status. New York: Oxford University Press.

———. 2007. Intricate Ethics: Rights, Responsibilities, and Permissible Harm. New York: Oxford University Press.

Lazar, Seth. 2019. ‘Accommodating Options’. Pacific Philosophical Quarterly 100 (1): 233–55.

Mason, Elinor. 2019. Ways to Be Blameworthy: Rightness, Wrongness, and Responsibility. Oxford: Oxford University Press.

McElwee, Brian. 2010. ‘The Rights and Wrongs of Consequentialism’. Philosophical Studies 151 (3): 393–412.

———. 2020. ‘The Ambitions of Consequentialism’. Journal of Ethics and Social Philosophy 17 (2): 198–218.

McKenna, Michael. 2012. Conversation and Responsibility. New York: Oxford University Press.

Mellema, Gregory. 1991. Beyond the Call of Duty: Supererogation, Obligation, and Offence. New York: State University of New York Press.

Menges, Leonhard. 2020. ‘Responsibility and Appropriate Blame: The No Difference View’. European Journal of Philosophy.

Moore, G.E. 1903. Principia Ethica. Edited by Thomas Baldwin. Revised Edition. Cambridge: Cambridge University Press. 1993.

Muñoz, Daniel. 2020. ‘Three Paradoxes of Supererogation’. Noûs, 1–18.

New, Christopher. 1974. ‘Saints, Heroes and Utilitarians’. Philosophy 49 (188): 179–89.

Noggle, Robert. 2009. ‘Give Till It Hurts? Beneficence, Imperfect Duties, and a Moderate Response to the Aid Question’. Journal of Social Philosophy 40 (1): 1–16.

Norcross, Alastair. 2006. ‘The Scalar Approach to Utilitarianism’. In The Blackwell Guide to Mill’s Utilitarianism, edited by Henry West, 217–32. Hoboken: Wiley-Blackwell.

———. 2020. Morality by Degrees: Reasons without Demands. Oxford: Oxford University Press.

Parfit, Derek. 2011. On What Matters: Volume One. Oxford: Oxford University Press.

Pereboom, Derk. 2014. Free Will, Agency, and Meaning in Life. Oxford: Oxford University Press.

Peterfreund, Sheldon P. 1978. ‘On the Relationship Between Supererogation and Basic Duty’. The Personalist 59 (1): 53–57.

Portmore, Douglas W. 2008. ‘Are Moral Reasons Morally Overriding?’ Ethical Theory and Moral Practice 11 (4): 369–88.

———. 2011. Commonsense Consequentialism: Wherein Morality Meets Rationality. Oxford Moral Theory. New York: Oxford University Press.

———. 2019. Opting for the Best: Oughts and Options. New York: Oxford University Press.

Pummer, Theron. 2016. ‘Whether and Where to Give’. Philosophy & Public Affairs 44 (1): 77–95.

———. 2021. ‘Impermissible yet Praiseworthy’. Ethics 131 (4): 697–726.

Pybus, Elizabeth M. 1982. ‘Saints and Heroes’. Philosophy 57 (220): 193–99.

Raz, Joseph. 1975. ‘Permissions and Supererogation’. American Philosophical Quarterly 12 (2): 161–68.

Ross, W. D. 1930. The Right and the Good. Oxford: Clarendon Press.

Scheffler, Samuel. 1982. The Rejection of Consequentialism. Oxford: Clarendon Press.

Schoenfield, Miriam. 2016. ‘Moral Vagueness Is Ontic Vagueness’. Ethics 126 (2): 257–82.

Schulte, Peter. 2014. ‘Beyond Verbal Disputes: The Compatibilism Debate Revisited’. Erkenntnis 79 (3): 669–85.

Shoemaker, David. 2017. ‘Response-Dependent Responsibility; or, A Funny Thing Happened on the Way to Blame’. Philosophical Review 126 (4): 481–527.

Sidgwick, Henry. 1907. The Methods of Ethics. Seventh Edition. London: Macmillan.

Sinclair, Thomas. 2018. ‘Are We Conditionally Obligated to Be Effective Altruists?’ Philosophy & Public Affairs 46 (1): 36–59.

Singer, Peter. 1972. ‘Famine, Affluence, and Morality’. Philosophy and Public Affairs 1 (3): 229–43.

Sinhababu, Neil. 2018. ‘Scalar Consequentialism the Right Way’. Philosophical Studies 175: 3131–44.

Sinnott-Armstrong, Walter. 1988. Moral Dilemmas. Oxford: Blackwell.

Slote, Michael. 1991. ‘Shelly Kagan’s The Limits of Morality’. Philosophy and Phenomenological Research 51 (4): 915.

Slote, Michael A. 1985. Common Sense Morality and Consequentialism. Routledge & Kegan.

Strawson, Peter. 1962. ‘Freedom and Resentment’. Proceedings of the British Academy 48: 1–25.

Thomson, Judith Jarvis. 1986. Rights, Restitution, and Risk: Essays in Moral Theory. Edited by William Parent. Harvard University Press.

Urmson, J. O. 1958. ‘Saints and Heroes’. In Essays in Moral Philosophy, edited by A. I. Melden. University of Washington Press.

Wallace, R. Jay. 1994. Responsibility and the Moral Sentiments. Cambridge, MA: Harvard University Press.

Watson, Gary. 1996. ‘Two Faces of Responsibility’. Philosophical Topics 24 (2): 227–48.

———. 2004. Agency and Answerability: Selected Essays. Oxford: Oxford University Press.

Zimmerman, Michael J. 1996. The Concept of Moral Obligation. Cambridge: Cambridge University Press.