5.1 Supererogatory Sets

I will argue for the superiority of (S) over (A), but let me first explain the idea of a supererogatory set of actions. We can start with two examples about other moral concepts:

Note that it was not good that Aaron failed to meet Benedict, and it was not wrong for Dimitri to give Enzo a birthday present. In both cases, we are saying something about both acts considered together. It's wrong to give one friend a birthday present but not another, but there is nothing wrong about either of these actions on their own.

So there is an intuitive sense in which we apply moral predicates such as ‘good’ and ‘wrong’ to sets of actions. There is no reason to think the case would be any different for supererogation, especially insofar as we define that notion in terms of these other predicates.Footnote ²⁰

Another observation is that the same behaviour can be described in different ways, sometimes obscuring hard boundaries between single actions and sets of actions. Consider the following two claims which both refer to the same conduct:

Sometimes we easily slide back and forth between talking about single actions and sets of actions. It would be strange if moral predicates properly only applied to the former. This should give some intuitive grip on how the notion of a supererogatory set of actions is possible.Footnote ²¹ Note that the idea of a ‘set’ only appears in our philosophical analysis; agents need not explicitly think about sets when they deliberate.

5.2 Concurrent Acts

Let me now turn to objections against the act-framework. First, there is a problem with concurrent acts. Imagine that there is a disjunctive duty to do x or y and assume that an actor is doing x and y at the same time. For example, Stannis has a duty to give to one of two charities. He uses online banking, fills in the details of both charities, and by pressing ‘Enter’ sends money to both simultaneously.

The defender of (A) has three choices. She can say that both acts of charitable giving are supererogatory, that only one of the two is, or that none of them is. The first option is not plausible. In giving to both charities, Stannis fulfils his disjunctive duty to give to either. So one of the two actions must count as fulfilling his disjunctive duty. Even if we claim that we do not know which of these two actions fulfils the disjunctive duty, we are committed to the conceptual claim that there are actions which fulfil a duty and are supererogatory. Given that the minimal content of the notion of ‘supererogation’ is to ‘go beyond duty’, this is conceptually implausible.

The claim that none of the two actions is supererogatory is equally troubling. There is something supererogatory about giving to two charities, and so an act framework should say that at least one action being done is supererogatory (or a candidate for supererogation). Supererogation has to be located somewhere in the example.

The act-theorist might deny that there are two actions being done here – instead, there is one action, which is Stannis giving to two charities. But this reply makes the situation worse: if this is one action, it is both required by the duty to be charitable and it is supererogatory, which again leaves us in conceptually awkward territory.Footnote ²²

Thus, the only feasible option for the act-theorist is to claim that one of the two actions is supererogatory, and the other one isn't. The main problem here is that any such determination will be arbitrary. Assume that Stannis’ two charitable givings are symmetrical in all morally relevant aspects: they do the same amount of good, are equally burdensome to Stannis, and so on.

A first line of reply for the act-theorist is to claim that perfectly symmetrical cases are rare. In most realistic cases, the actions being done differ in their non-temporal features, such as how burdensome they are. We can decide on the basis of these other properties which actions to classify as supererogatory in concurrent actions – e.g., the more burdensome action is the one which is supererogatory.

On what basis do we pick the properties which serve as tie-breakers in such cases? We could pick those properties which are associated with supererogatoriness – e.g., being burdensome. But if we do so, we move beyond theoretical resources internal to the duty-plus view. Rather, we invoke additional intuitions about the supererogatory.

The basis on which we design our tie-breaking criterion looks ad hoc. We are interested in explaining supererogation, so appealing to intuitions about what is supererogatory is dangerous at this point. Second, there is the comparative point: the set framework is preferable because it does not need to give a tie-breaking criterion in the first place. Thus, theoretical simplicity favours the set framework.

A second strategy is to embrace arbitrariness. In Stannis’ case, we should say that one charitable giving is supererogatory, and the other one isn't. Perhaps it is unknowable which one is which, but we know that one of them is, and the other isn't. One motivation for this strategy is that we ‘do not care’ precisely which of the two actions is supererogatory. We only care that the agent acted in a supererogatory manner.Footnote ²³

This reply, however, makes it hard to see why we should remain act-theorists. If we do not care which action is supererogatory, why insist on a theory that makes arbitrary stipulations? The set framework can account for the intuition that the agent acted in a supererogatory manner without incurring this theoretical cost. According to (S), {x,y} is (possibly) supererogatory. Both x and y are constitutive members of a supererogatory set, but the question whether x is supererogatory, or whether y is, does not arise. Thus, we need to make no arbitrary stipulations.

5.3 Accumulative Supererogation

Let's turn to a second class of cases. It is plausible that having and exercising certain virtues beyond some level is supererogatory.Footnote ²⁴ Most virtues express themselves in many different actions, most of which are comparatively insignificant. For example, kindness includes holding doors open for others. Call such cases accumulative supererogation: we go beyond the call of duty not in one big step, but through many small steps. Other examples of accumulative supererogation are being generous, charitable, friendly, helpful, and so on.

Let's start with a simplified case. Imagine that Renly has a duty to be kind. Consider the following two patterns,

Each of the two timelines shows a continuous pattern of kind behaviour, with the inset lines showing when a kind act has been done. All of these are comparatively insignificant actions – being friendly with one's barber, complimenting acquaintances, tipping the cab driver well, and so on.

Assume that pattern I describes (one instance of) a minimally duty-fulfilling pattern. If Renly does kind acts roughly this often and at roughly those intervals, he fulfils his duty of kindness. Assume that Renly acts as described by pattern II – that is, he is kind much more often than required of him.

A first objection is that each of Renly's actions is too morally insignificant to deserve the label ‘supererogatory’. Without trying to give a complete account, moral significance implies praiseworthiness, and is a fitting occasion for positive reactive attitudes; it goes hand in hand with merit, and moving beyond the socially expected; and significance is correlated with conferring benefits on others, or imposing burdens on the agent.

But each single act of kindness, considered on its own, is only a small nicety which does not have any of these features. Holding open a door for someone imposes a burden which will often not even be perceived, as will be the benefit to the other person. It seems odd to praise an agent for holding open a door. Some reactive attitudes might be fitting in response to such acts – e.g., gratitude – but the fitting attitudes will be very low-key, and unlike the stronger reactive attitudes associated with moral significance. Lastly, small acts of kindness are common and expected.

One reply is to severely weaken the predicate ‘supererogatory’, and reduce it to merely mean ‘beyond duty’. Under this definition, when we call holding open a door supererogatory, nothing is said about merit, praiseworthiness, sacrifice, or any other sense of moral significance. We allow for ‘trifling’ supererogatory actions.Footnote ²⁵

Severing the link between supererogation and moral significance would make the extent of the supererogatory both too large and uninteresting. For example, going for a long Sunday afternoon walk is ‘beyond duty’ in this merely technical sense. But we exclude it from the scope of the supererogatory because it is morally insignificant.

Even if you think that weakening the predicate ‘supererogatory’ is a plausible move, we should again note that an act-based theory incurs theoretical costs which the set-theory does not have. The actions done in pattern II, considered together, are highly morally significant. Doing all of these actions is meritorious and praiseworthy; it does much good for others; and so on. Thus, this set of actions is morally significant. The set theorist does not need to weaken the predicate ‘supererogatory’, because she can highlight the moral significance of the set of actions.

Another reply is to deny that kind actions, even considered collectively, are supererogatory.Footnote ²⁶ One might argue that kindness is too far away from paradigm cases of supererogation, such as Urmson's doctor volunteering to travel to the plague-ridden city. It is neither heroic nor saintly.

In reply, let me start with a point about philosophical historiography. While everyone remembers Urmson's examples of heroic supererogation, he also allows that ‘[i]t is possible to go just beyond one's duty by being a little more generous, forbearing, helpful, or forgiving than fair dealing demands’, and he goes on to explain that he drew ‘attention to the heroic and saintly deed […] merely in order to have conspicuous cases of a whole realm of actions that lie outside the [classic deontic] trichotomy’Footnote ²⁷ . So the article setting the terms of the debate on supererogation allows generosity and other virtues to count as supererogation.

I think this is correct. There are various ways how we can define supererogation – e.g., as what it is praiseworthy to do but blameless to omit, or what is good to do but not wrong not to do. Whichever definition you prefer, we are interested in a peculiar ‘realm’ of morality. Heroes and saints are paradigm cases in the sense that they first pointed us towards this realm. But once we have it firmly in view, there is no guarantee that the cases falling into this realm will look very much like our initial paradigm cases. Kindness and other virtues robustly fulfil various conceptions of the supererogatory. For example, exhibiting these virtues to a high degree is good but not required, and praiseworthy to do but not blameworthy to omit. Thus, we should classify them as supererogatory, despite their dissimilarities with Urmson's heroes and saints.

So much for moral significance. My second objection is again from arbitrariness. With regard to pattern II, the act-theorist needs to say which of Renly's actions are supererogatory and which aren't. But there is no non-arbitrary way to make this distinction. First, we might take temporal location as a criterion, and say that the actions done at a later time are (possibly) supererogatory, and the actions done earlier merely fulfil the relevant duty. For example, in pattern II x would fulfil a duty, while y and z would be candidates for supererogation.

But given that this is a continuous pattern of behaviour, there is no natural point at which the disjunctive duty starts to apply or ends. We might shift the time frame to the left, such that y and z are at the beginning of it; now they would cease to be candidates for supererogation. Because picking a starting point is arbitrary for many duties, determining supererogatoriness with reference to such a starting point is arbitrary as well.

A defender of (A) might again point to non-temporal features of the actions to determine whether they are supererogatory. Some actions are more morally significant than others. Assume pattern III describes Margaery's charitable giving, where duty is again described by pattern I.

The differing heights of the inset lines indicate the varying moral significance of Margaery's acts – e.g., Margaery's contributions to charity differ in how burdensome they are to her, or how much good they do. We might now claim that Margaery's big contributions are supererogatory – e.g., v and w in pattern III.

But it's not clear why we should pick out those actions and not others. Imagine the opposite view. We argue that Margaery's big contributions fulfil her duty, and it's the many small things she does which are supererogatory. I can see no principled reason why we should not choose this way of distinguishing the supererogatory from the non-supererogatory.

The set framework avoids the arbitrariness problem. (S) classifies the set of actions done in patterns II and III as (possibly) supererogatory. The set theorist does not embark on a difficult search for categorising some of Renly's and Margaery's actions as supererogatory, and some as duty-fulfilling. Rather, it is Renly's kind behaviour and Margaery's charitable giving, formally analysed as the sets of actions they do, which is supererogatory.

5.4 Other Advantages

There are three further advantages of a set-based over an act-based duty-plus approach to supererogation.

First, the set framework better allows us to account for standard ways of talking about supererogation. In reading the biography of a saint, we might single out no one particular action which we think was ‘the saintly one’. Rather, we say that ‘the saint's life’, or ‘the things the saint has done’, were supererogatory.Footnote ²⁸

Second, David Heyd has objected that any account which explains supererogation in terms of imperfect duty makes a category mistake, as the concept of imperfect duty only applies to sets of acts (or act-types), whereas supererogation is a quality of particular acts.Footnote ²⁹ Assuming there is some form of category mistake, (S) avoids it (while (A) commits it), as both disjunctive duty and supererogation are defined for the same category, i.e., sets of actions.

A third advantage is more speculative. Assume that Susie's house burns down. In response, she receives an overwhelmingly supportive response from her neighbours. All of them help her out in small ways; but her neighbours, collectively, help her so generously that her loss is significantly reduced. I find it tempting to think that this is a case of collective supererogation: while no particular neighbour went beyond the call of duty, the collective support which Susie received was supererogatory. In this case, a set of actions done by different agents is supererogatory.Footnote ³⁰ Whether collective supererogation exists is admittedly controversial, and the possibility has not been explored to my knowledge. However, a set framework is in a good position to account for such cases.

6.1 Individuating Duties

The defender of the duty-plus view is committed to the claim that all supererogatory sets are subsets of a set for which there is a disjunctive duty (conditions (i) and (ii) of (S)). In response, consider what Feinberg claims about Urmson's case of a doctor who volunteers to travel to a plague-ridden city:

We can outline the following supererogationist objection. There are certain actions that are supererogatory and which do not fall under any duty. These (sets of) acts are ‘in no way obligatory’Footnote ³² , i.e., not even covered by any disjunctive duty. Rephrased in the context of (S), there are some supererogatory sets of actions for which (i) is false.

There is an obvious reply: there are various disjunctive duties which cover the current case. First, there is the duty to help people in the plague-stricken city. If you think there is no such duty, consider the (role-based) duty the doctor has to help the sick and needy. And lastly, consider the generic duty to help others. If we just look widely enough, we will find a duty which covers the given case. Feinberg is right that there is no ‘duty to travel one step toward the plague-stricken city’, insofar as there is no perfect duty which requires this particular action. But it does fall under the scope of several disjunctive duties.

This strategy generalises. For any case the supererogationist presents, it is likely that we will find some general disjunctive duty which covers it.Footnote ³³ If this is the dialectical situation, it seems that the duty-plus theorist has the upper hand.

Feinberg might reply that this is a gimmicky way of describing the travelling doctor's situation. But what Feinberg needs is more than an appeal to intuition. Rather, he needs a full account which explains how duties are individuated, and how they apply to particular cases.

Consider now the following claims:

1. (1) The doctor has a duty to help others.

2. (2) Travelling to the plague-ridden city is an instance of helping others.

3. (3) Travelling to the plague-ridden city fulfils the doctor's duty to help.

The first two propositions are natural assumptions. Denying (1) commits the supererogationist to an implausibly minimalist view of what duties we have. (2) is obvious. If we denied (3), this would have the counter-intuitive implication that, after the doctor returns from his trip, he has not fulfilled his duty to help, and we can blame him for failing to fulfil his duties, whereas we should do the opposite and praise him. Moreover, (3) seems an implication of (1) and (2).

Here's a reply I can imagine on Feinberg's behalf. While we acknowledge (1)-(3), we might also claim that

1. (4) Travelling to the plague-ridden city is not required by the doctor's duty to help others.

Consider an analogy. You have a duty to support your country. If you become a soldier, you fulfil that duty. But the duty to support your country did not require you to become a soldier. Similarly, we might say that the heroic doctor fulfilled his duty to help others, but that he was not required by that duty to travel to the city. Thus, the supererogationist can say that supererogation can be duty-fulfilling, but it is not required by duty.

We should note that this reply already concedes some ground to the advocate of duty-plus. In this modified supererogationist position, we acknowledge that something can be supererogatory while ‘fulfilling’ a duty. It's not clear that this is compatible with the supererogationist intuition that there are some areas of morality which fully fall outside the realm of duty.

More importantly, it's hard to make the current distinction in deontic logic. If some action x ‘fulfils’ a disjunctive duty D this is naturally read as the claim that it was one of the disjuncts of D. But if x is one of D's disjuncts, then it is disjunctively required by D in the sense I defined. The distinction between ‘fulfilling a duty’ and ‘being required by a duty’ is thus likely to collapse; it might merely be a restatement of the supererogationist's intuitions in a question-begging way.

6.2 Individuating Actions

In a second set of counter-examples, an act is supererogatory, although not more than the number of acts necessary for fulfilment of a disjunctive duty have been done (condition (iii) of (S)).Footnote ³⁴ Consider Ayewa, the soldier who throws herself on a live hand grenade to save others. This act is supererogatory, but it seems that (S) cannot give it supererogatory status, as only one act is done.

We here encounter the problem of how to individuate actions. Remember Giulia, who helped Henrik get his life back on track. We could describe her behaviour as if it were one action (help Henrik) or as if it consisted of several actions (give him a place to stay, help him overcome addiction, and so on). Which description should we prefer? Which carves the situation at its metaphysical joints? Are both admissible descriptions of the situation?

These are extremely difficult questions, and I will make no attempt at an answer. Still, there are some responses the set-theorist can give. It seems to me that the advocate of (S) is not committed to the claim that on any possible way of individuating duties, and on any possible way of individuating actions, conditions (i)-(iii) of (S) are given. This would be much too demanding. Rather, she should claim that some behaviour is a candidate for supererogation if there is some plausible way to describe our duties and some plausible way to describe the behaviour such that conditions (i)-(iii) of (S) apply.

A corollary of this claim is the following. Some ways of describing our duties and actions come more naturally to us than others. The duty-plus theorist is not committed to the claim that under the best or most natural description of what is happening, the agent's acting fulfils (S). So it is no direct objection if you think that Giulia's behaviour should best be described as doing one action, as long as you agree that the alternative description is at least plausible. The same strategy can be applied to a wide range of cases.

Ayewa's example provides a hard case where individuating actions as several does not look very plausible. She throws herself on the grenade; that's it. To answer this case, let's start with a description of her duties: Ayewa has a duty to tolerate a small risk to make the survival of her comrades more probable. This duty is disjunctive, as there are many slightly risky ways in which she can fulfil it. So first of all, if Ayewa has done anything else covered by this duty aside from throwing herself on the grenade, we can count her case as supererogatory according to (S). This will be the case for most realistic scenarios.

Let's now assume this is not the case – Ayewa has just entered the army, say. As a second line of reply, we might re-describe the action itself. Throwing oneself on a live hand grenade is the acceptance of a very great risk to secure one's comrades’ survival. It is taking on the standard risk everyone has to accept, and in addition making that risk almost certain death. It is doing both of these things which is supererogatory, and so is covered by (S).

As a last resort, we might bite the bullet: no view can live without theoretical costs. If Ayewa has done nothing else to fulfil her duties as a soldier, and throwing herself on the grenade is the only one thing she has done, then perhaps we should reject that she acted in a supererogatory manner. She simply fulfilled her duty. We can still say that Ayewa acted in a very praiseworthy and heroic way – acting heroically is neither identical with supererogation, nor sufficient for it. Again, supererogation is a technical category: once we have discovered its existence, we should not expect that its contents are necessarily what we expect them to be.

6.3 Doing One's Supererogatory Share

Another set of objections concerns the compatibility of supererogation and thresholds. A first worry is this: According to the duty-plus view, if you do one more action than required by disjunctive duty, you may act supererogatorily. But one might object that this makes supererogation too easy to achieve. It's hard to see why doing barely more than duty requires could be supererogatory.

However, nothing in the duty-plus view conceptually requires that we identify two thresholds: (1) the threshold for fulfilling a duty, and (2) the threshold for supererogation. Thus, we can introduce two thresholds, m _D and m _S, such that 0 < m _D < m _S < n, where n is the number of disjuncts in the relevant duty. After doing m _D acts, the disjunctive duty is fulfilled; after doing m _S acts, the conjunction of all acts becomes a candidate for supererogation.

But perhaps there is a lingering deeper worry. If we view supererogation as fulfilling a threshold, it becomes possible to be ‘minimally supererogatory’. One might object that this is self-refuting, as supererogation excludes the idea of being moral in any minimal way. Imagine Peter ‘does his supererogatory share’, and stops afterwards, pointing towards his supererogatory achievements. Few would say that Peter acts in a supererogatory way.

The best response for the duty-plus view is to distinguish the motivations of the agent from the (deontic) classification of the action. If it was part of Peter's motivation that he won't do more than his supererogatory share, this introduces a motive which likely undermines the supererogatory status of his actions.

Consider instead Elo's case. She has also minimally gone beyond the supererogatory threshold. But she has been motivated to do her best, and it has been an accident that she was only minimally supererogatory – perhaps she has been quite busy, and despite her best efforts, had no further opportunities and resources to give to charity. In her case, it is much less counter-intuitive to think that she acted in a supererogatory manner.

6.4 Perfectible Duties

One last objection stems from the rigorist view that imperfect duties are open-ended or imperfectible: We can never fulfil them.Footnote ³⁵ Daniel Statman, for example, writes that ‘charity is the sort of duty we are never really freed from […]’Footnote ³⁶ . We are justified in ceasing to give to charity only if more stringent duties require us to do so.

As I have described disjunctive duties, they can be fulfilled, so the duty-plus advocate must reject rigorism. Whether we accept rigorism is likely to depend on the deeper view of morality we have, but let me offer some observations which soften the blow of rejecting rigorism.

First, the idea that duties can be fulfilled should not be mistaken for the idea that morality is undemanding. (S) can be used to describe a lenient ‘Yuppie ethics’Footnote ³⁷ approach to morality, but also a very demanding theory of morality. Everything depends on where one sets the relevant thresholds. Supererogation might be common, or rare, or contingently there might be no supererogation at all. In other words, the duty-plus view is a structural, not a substantive view of morality.

Second, I introduced the idea of disjunctive duties as a simplification. Real-life duties will be subject to indeterminacy and uncertainty. We will often be sceptical about claims that our duties have been fulfilled. Agents who claim to have fulfilled their duties might often be mistaken, and teaching children that duties have thresholds might be harmful. The duty-plus advocate can accept this. She only claims that as a matter of principle duties can be fulfilled. She need not take any position on the determinacy, knowability and so on of actual imperfect duties.

Third, we should again stress the difference motives can make. Agents who think of their duties as annoying hurdles which are in one's way are morally criticisable in various ways.Footnote ³⁸ The defender of the duty-plus view, however, is not committed to the claim that it's good if agents are motivated in such ways, and she might include proper motivation as a further necessary condition for supererogation.