It’s widely believed that maximising consequentialism implies that supererogatory action is impossible. And many philosophers, finding the idea of supererogation intuitively plausible, regard this as a reason to reject maximising consequentialism (and perhaps to adopt some form of “satisficing” consequentialism instead). As I shall argue, however, supererogation — properly uderstood — is perfectly compatible with maximising consequentialism.

Let’s begin with the following (rough) formulation of maximising consequentialism.

Maximising Consequentialism (MC): An action is permissible if and only if its outcome is at least as good as that of every alternative.

(A more complete formulation might have more bells and whistles, but this will do for our purposes.)

Now, in the usual idiom, a supererogatory action is one which goes “beyond the call of duty”; to engage in supererogation is to act so as to produce more good than was required for one to have acted permissibly. This rough characterisation suggests the following definition.

Definition 1: An action X is supererogatory if and only if there exists some alternative Y such that (a) X‘s outcome is better than Y‘s, and (b) X and Y are both permissible.

Clearly, MC implies that no action is supererogatory in this sense. Given MC, clauses (a) and (b) in our definition of supererogation are inconsistent: if X has a better outcome than Y, then Y is impermissible, according to MC.

However, there is a deficiency in our definition of supererogation, which the following example reveals. Suppose that you’re a college student with \$80 spare cash, and that you’re deciding what to do with it. Among your options are the following:

(A) Give \$40 to charity, and spend the rest on beer.
(B) Give \$40 to charity, and spend the rest on philosophy books that you need for a course you’re taking.

Now, suppose we judge that A is permissible but not supererogatory. In our view, A produces the minimum amount of permissible good; any less would be impermissible. Plausibly, however, the outcome of B is better than that of A, since you will be more greatly benefited by reading philosophy books than by drinking beer. Hence, it follows, on our definition of supererogation, that B is supererogatory. But that seems quite odd. Sure, B produces a surplus of good over the minimum permissible amount, but that surplus is all your own good. Yet is doesn’t seem that promoting your own good could make your action supererogatory — at least not as that term is ordinarily used.

Let us, then, revise our definition as follows:

Definition 2: An action X is supererogatory if and only if there exists some alternative Y such that (a) X‘s outcome is better for others than Y‘s, and (b) X and Y are both permissible.

On this revised definition, action B is no longer supererogatory, since, although its outcome is better than A’s, it’s not better for others (where by “others” is meant everyone besides the person whose actions are in question); B’s outcome is better, but only for you.

But notice that, on our new and improved definition, MC allows for the possibility of supererogatory acts. Consider, for example, utilitarianism — the paradigm exemplar of MC. And suppose you face a choice between two actions, C and D, whose outcomes are as follows:

(C) total of others’ utilities = 80; your utility = 20
(D) total of others’ utilities = 70; your utility = 30

In either case, the total of all utilities — both others’ and yours — is 100. Hence, utilitarianism says that both actions are permissible. But, although the outcomes of the two actions are equally good (by utilitarian lights), the outcome of C is better for others. Hence, utilitarianism implies that C is supererogatory.

## 23 Replies to “Supererogation for Maximisers”

1. I think you’re right that D1 is a bad definition of ‘supererogatory,’ or at least D2 better fits the sort of examples we usually think of. (Although I guess we might wonder whether ‘supererogatory’ has any pretheoretical sense, or is just a piece of jargon.) But something is funny about the dialectic here. You seem to be accepting that it’s better for the college student to buy the books, but permissible for him to buy the beer. Whether you call (B) supererogatory or not, you’ve still got a counterexample to MC. If the project is to defend MC, this isn’t helpful.
Also, I think the objectors to MC will be satisfied with a single case where an act we think of as supererogatory turns out obligatory according to MC. It’s not necessary to make the stronger claim that MC can’t account for *any* supererogatory acts.

2. Great post! It’s good to have someone else on the blog posting entries on consequentialism. I have one comment and one question.
First the comment: I think that Ben is right. Although you’ve shown that MC does not imply that supererogatory acts are impossible, you haven’t done much to make MC any more plausible than it already is. MC, as you define it, will only allow for supererogation in cases where two outcomes tie for first place. But, according to our considered moral judgments, there are many other cases of supererogation. For instance, consider:
(E) total of others’ utilities = 80; your utility = 20
(F) total of others’ utilities = 70; your utility = 31
Our considered moral judgment is that (E) is supererogatory, yet, according to MC, (F) is morally obligatory. But perhaps it was never your intention to argue that MC is significantly more plausible than we might have thought.
Now a question: would you consider the following theory to be an instance of MC: An action is permissible if and only if either (a) its outcome is at least as morally good as that of every alternative or (b) its outcome is at least as good, all things considered, as that of every alternative. Call this Portmore’s Consequentialism. The reason, I ask, is because I’ve argued in my paper that appears in _Ethics_ that PC not only makes supererogation possible, but also makes a wide range of supererogatory acts possible.

3. Ben, I think you make two good points.
I agree that there’s something funny about the dialectic in my original post. In order to discredit D1, I present a counterexample purporting to show that D1 is too broad; that is, it’s intedned an example of an action that is included as supererogatory by D1, but which, intuitively, shouldn’t be so included. But it’s hard to see how a proponent of MC could claim that D1 is too broad in this way. According to MC, the set of actions that are supererogatory (in the sense defined by D1) is just the empty set. Hence, proponents of MC couldn’t consistently claim that D1 includes too many actions, because, in their view, it doesn’t include any actions at all. So the funny thing is that, although I may have presented a counterexample to D1, it’s not a counterexample that a maximiser could endorse. (As you point out, insofar as it’s a counterexample to D1, it’s also a counterexample to MC!)
So I guess I’ll have to retract that part of my argument. What I need to do is offer a counterexample to D1 showing that it’s too narrow — that it fails to include some actions as supererogatory when, intuitively, they should be so included. I reckon that I probably could manage to find a fairly decent counterexample of that kind; but I won’t attempt to do so right now.
I also agree with your second point. However, I’m prepared to make the following bold conjecture: every plausible example of supererogation is consistent with MC; that is, for any action X that is plausibly regarded as supererogatory, maximisers may consistently hold that X is supererogatory (in the sense defined by D2). But, again, I won’t try to make good on that conjecture just yet.

You write:
according to MC, (F) is morally obligatory.
But that’s just false (as Michael Smith would say). According to utilitarianism (F) is morally obligatory. But MC doesn’t imply utilitarianism.

5. Doug Portmore says:

Dear Campbell,
I wasn’t assuming that MC implies utilitarianism. True there may be some suppressed premises that I’m relying on, but it’s not that MC implies utilitarianism. In fact, they all seem fairly obvious: (1) utility is good (you don’t have to be a utilitarian to accept that), (2) other things being equal, the more total utility in an outcome, the better it is, and (3) everything else is equal, axiologically speaking. If we accept (1)-(3), wouldn’t MC imply that (F) is morally obligatory? So I don’t see that you’ve given us any reason to think that my claim is just false. To show that my claim is false, you’ll have to do more than point out that MC doesn’t imply utilitarianism.
In any case, let me make the same point in a more general way.
Take some example of supererogation on MC, such as,
(C*) total good for others = 80; good for you = 20
(D*) total of good for others = 70; good for you = 30,
there will be a slightly different version of this example, such as,
(E*) total good for others = 80; good for you = 20
(F*) total of good for others = 70; good for you = 31,
where (E*) is morally wrong on MC, and yet our considered moral judgment is that (E*) is supererogatory.

6. Doug,
I may have interpreted you uncharitably. I understood you as claiming that MC by itself implies that F is obligatory. I think that claim is pretty clearly false. But, of course, the conjunction of MC and some other claims (e.g. your assumptions 1-3) might imply that F is obligatory. My apologies for the uncharitable interpretation, if that’s what it was.
Now, if I understand correctly, your objection is that, although MC allows for some cases of supererogation (on my proposed definition), it allows for too few cases — or, perhaps more accurately, it allows for supererogation in too narrow a range of cases. But I don’t see why that is so. As you point out, “MC … will only allow for supererogation in cases where two outcomes tie for first place.” Thus, proponents of MC may accommodate the view that supererogation occurs in a wide range of cases simply by adopting some theory of goodness according to which outcomes are tied for first place in a wide range of cases. Your claim, I take it, is that all such theories of goodness are implausible (perhaps because they violate your assumptions 1 – 3, which you regard as obviously true). But I’d like to see an argument for that claim before I accept it.

7. Campbell,
Rather than argue who has the burden of proof: you, to prove that at least one of (1)-(3) is implausible, or me, to prove that all of (1)-(3) are plausible. Let me try an argument that doesn’t rely on any axiological assumptions.
Given your definitions of both supererogation and MC, in order for an act C to be supererogatory on MC, there will have to be an alternative act D, such that
(C): total good for others = x; good for you = (n – x); total good for everyone = n
(D): total of good for others = y; good for you = (n – y); total good for everyone = n,
where n is greater than x, and x is greater than y.
Now my argument begins with the observation that, for any such case of C and D, there will be a slight variant on that case, such that
(E): total good for others = x; good for you = (n – x); total good for everyone = n
(F): total of good for others = y; good for you = [(n – y) + 1]; total good for everyone = (n + 1),
where n is greater than x, and x is greater than y.
And my claim is that our considered moral judgment will be that E is supererogatory, whereas MC implies that E is morally wrong. (The idea is this: you don’t turn an otherwise supererogatory act into one that is morally wrong merely by making the relevant alternative slightly better for you.) Thus, MC allows for too few cases of supererogation.

8. Doug,
I’m inclined to deny that E is supererogatory. But if your considered moral judgement is that E is supererogatory, then MC can accommodate this; you just need to hold that E and F are equally good.

9. Doug Portmore says:

Campbell,
You say, “But if your considered moral judgement is that E is supererogatory, then MC can accommodate this; you just need to hold that E and F are equally good.” Of course, whether MC can accommodate the judgment that E is supererogatory depends, not on what axiology I or anyone else adopts, but rather on what axiology is in fact correct. Thus, MC can accommodate the judgment that E is supererogatory only if E and F are, in fact, equally good. But E and F are NOT equally good. For, by hypothesis, we’re supposing that C is supererogatory on MC, and that means that C and D must be equally good. But how can E and F be equally good if C and D are equally good? For this to be the case, one would have to claim that an outcome’s having more than (n – y) good for you doesn’t, other things being equal, make the outcome better. But that is highly implausible. Why would an additional unit of good for you not make the outcome better except when the total good for you is less than (n – y)? Why is this true of you and not others? And why would an additional unit of good for you stop making an outcome better at the point where your total good equals (n – y) as opposed to [(n – y) + 1]?
You seem to think that the solution to MC’s putative, problematic implications is to just adopt whatever axiology makes MC’s putative, problematic implications go away. But whether MC has any problematic implications doesn’t depend on what axiology one adopts; it depends on what axiology is, in fact, true.
That said, there is a way out for you: you can, as you seem to be inclined to do, just dismiss our commonsense intuition that E is supererogatory if C is. In other words, our commonsense tells us that you can’t turn an otherwise supererogatory act (for you: C) into one that is morally wrong (for you: E) merely by making the relevant alternative (F rather than D) slightly better for you. But if our commonsense intuitions are to be simply dismissed, then showing that MC can accommodate our commonsense intuition that some acts are supererogatory is neither here nor there.
Or are you inclined to think that our commonsense intuition is that you CAN turn an otherwise supererogatory act into one that is morally wrong merely by making the relevant alternative slightly better for you? Could you give an example where this is intuitively plausible.

10. Jason says:

Hi Campbell,
I have a couple of questions / worries about your proposed definition of supererogation (Definition 2).
1) Consider the following options:
(G) Other’s total good = 50, your total good = 50.
(H) Other’s total good = 51, your total good = 70.
Of course, you’re working with MC. But it seems that taking Definition 2 by itself, if we allow for G and H as both being permissible, we’d end up counting (H) as supererogatory insofar as it creates a better outcome for others – despite the fact that you are the one who most benefits from (H). This seems to be a potential problem for this definition of supererogation [MC would not allow both (G) and (H) to be permissible, but it seems like you’d want your account of supererogation itself to rule out this case.] Would you want to add a clause to Definition 2 such that (roughly) the agent does not benefit more than others in supererogatory alternatives?
2) On the other hand, I myself don’t think that we should rule out the possibility of self-regarding supererogatory actions. For example, imagine a man working in a restaurant who earns a modest, but certainly adequate living. He has long wished to go to college, and now decides to do so at age 50. He needs to take on an extra part-time job to help cover the costs of tuition, books, and so on. He spends long hours studying after being away from school for over 30 years. This certainly seems to go beyond any basic self-regarding duties [if we wish to speak of such]; I think his actions are plausibly taken as supererogatory. [To the extent that there is a pre-theoretical notion of supererogation, I think a key element of it is that supererogatory actions are particularly demanding or difficult for the agent involved. I think in the present case the man returning to college is making a suitable range of costly, difficult sacrifices in order to obtain a good outcome.]

11. Jason says:

Sorry – meant to say “others’ total good”.

12. Doug Portmore says:

Jason makes an excellent point. Definition 2 needs revision. As Jason suggests, perhaps the correct definition is along the following lines:
Definition 3: An action X, performed by a person P, is supererogatory if and only if there is some available alternative Y such that (a) it is morally permissible for P to perform X but also morally permissible for P to perform Y, (b) P has more moral reason to perform X than to perform Y, and (c) P’s performing X is more demanding or self-sacrificing than P’s performing Y is.
Note that condition (c) would seem to imply, as I’ve been suggesting, that you can’t turn an otherwise supererogatory act into one that is morally wrong merely by making the relevant alternative slightly better for you. By making the relevant alternative better for you (for instance, by making the alternative to E, i.e., F, better for you than the relevant alternative to C, i.e., D, is), you make the act in question (e.g., E) even more self-sacrificing and hence, if anything, even further beyond the call of duty — and hence supererogatory.
I also think that it is better to formulate the definition of supererogation in terms of moral reasons as opposed to the goodness of outcomes. We should not presuppose in our definition of supererogation that the better an outcome the more moral reason there is to produce it. Consequentialists believe that moral reasons track the goodness of outcomes but we should not presuppose a consequentialist account of moral reasons in our definition of supererogation.

13. David says:

Couple of points. It looks to me like the advocate of MC has to do one of three things:
(i) Deny that E* is impermissible. (This the advocate cannot do without modifying the formulation of MC given in Campbell’s post.)
(ii) Deny that E* is supererogatory. (This the advocate cannot do without denying what seems to me a pressing intuition.)
(iii) Allow for the possibility that some action can be both supereroratory and impermissible, and assert that in fact E* is both of these. (This the advocate cannot do without modifying the definition of ‘supererogatory’ given in Campbell’s post. Also, it seems to me that generally speaking, for something to be both supererogatory and impermissible would be incoherent. If a supererogatory act is “above and beyond the call of duty,” and if an impermissible action is automatically “below” the call of duty, then a supererogatory and impermissible act would have to be, at the same time, both above and below the call of duty. This seems absurd. But maybe if a third dimension is added to this spatial metaphor (so that impermissible acts are somehow “beside” the call of duty, perhaps – or above but to the left of it, or something) then maybe something can be worked out here.)
I think (iii) is a potentially interesting route. But I think Campbell’s most direct route is to do (ii). I think cases like E* figure pretty centrally in many typical objections against views like MC, so there are many pre-existing attempts to accommodate them within an MC-esque framework.
[Corrected by Doug with David’s permission at 12:35 PM on 9/6/04]

14. Doug,
I agree that it’s implausible that E and F are equally good. But then I also think that it’s implausible that E is permissible. Hence, I would deny your claim that “you can’t turn an otherwise supererogatory act (for you: C) into one that is morally wrong (for you: E) merely by making the relevant alternative (F rather than D) slightly better for you.” I don’t know about “commonsense” intuitions, but my intuition tells me that your case is a counterexample to that claim. As you say, if we grant that C and D are equally good, then it’s implausible to claim that E and F are also equally good, since this implies that the difference between your having (n-y) good and your having (n+1-y) good makes no difference to the goodness of outcomes. But then it seems equally implausible to me to claim that this difference in your good makes no difference to permissibility; if we grant that C and D are both permissible, then it’s implausible to claim that E and F are also both permissible. In my view, the difference in your good is relevant both to goodness and permissibility: it makes E both worse and impermissible. Your position seems to be that the difference is relevant only to goodness (and not to permissibility). I find that position puzzling.
I suspect, however, that we’re not going to make much progress arguing about that case. Our differences seem to be at a more fundamental level. For example, you write (disapprovingly):
You seem to think that the solution to MC’s putative, problematic implications is to just adopt whatever axiology makes MC’s putative, problematic implications go away.
And, as it happens, that is what I think. You continue:
But whether MC has any problematic implications doesn’t depend on what axiology one adopts; it depends on what axiology is, in fact, true.
And I agree. But I think that the way to test the truth of an axiology is to plug it into MC and see whether it yields plausible judgements of permissibility. That’s why I think we should “adopt whatever axiology makes MC’s putative, problematic implications go away”; that’s how we adopt the true axiology. Of course this rests on a certain understanding of goodness: roughly, that goodness just is that which ought to be maximised. This strikes me as a fairly natural understanding, but I concede that it may be idiosyncratic among philosophers.

15. I’ve also worried about examples like the one Jason suggests. I’m not too happy with the idea of “self-regarding supererogation”; I tend to agree with Doug that supererogation must involve self-sacrifice. So I’d prefer to revise the definition of supererogation.
Doug’s proposal — Definition 3 — seems all right, except I’m worried that clauses (a) and (b) are inconsistent. If the balance of moral reasons supports doing X rather than Y, then I don’t see how Y could be permissible. (I assume here that we’re talking about moral permissibility.)

16. Jamie says:

Doug, you say:

Given your definitions of both supererogation and MC, in order for an act C to be supererogatory on MC, there will have to be an alternative act D, such that
(C): total good for others = x; good for you = (n – x); total good for everyone = n
(D): total of good for others = y; good for you = (n – y); total good for everyone = n,
where n is greater than x, and x is greater than y.

It seems to me that you must, in fact, be relying on unstated axiological assumptions, since the definitions MC and D2 don’t imply that each supererogatory act will have an alternative like you describe.
I think you are assuming that goodness is the sum of goodness for persons. MC and D2 do not together imply that, and admirers of MC (like myself) need not assume that goodness is even a function of goodness for persons, let alone an additive one.

17. Doug Portmore says:

Campbell: I think yours is a pretty loaded understanding of goodness, but I’m not entirely unsympathetic to it. And I agree that our differing intuitions about E leave us at an impasse. Lastly, you say, “If the balance of moral reasons supports doing X rather than Y, then I don’t see how Y could be [morally] permissible.” It may be, as I’ve argued elsewhere, that the moral permissibility of actions is affected by non-moral reasons as well as moral reasons.
And let me add that I’ve enjoyed the discussion. Thanks.
Jamie: You’re right. I should have written:
Given Campbell’s definitions of both supererogation (which should now be revised to include some clause about the self-sacrificing nature of supererogation) and MC, in order for an act C to be supererogatory on MC, there will have to be an alternative act D, such that
(C): goodness for others = x; goodness for you = p; goodness of the outcome brought about from C = n
(D): goodness for others = y; goodness for you = q; goodness of the outcome brought about by D = n,
where x is greater than y, and p is greater than q.
Now either (a) the goodness of an outcome is, other things being equal, equivalent to the aggregate sum of goodness for persons or (b) it isn’t.
If (a), I offer (E) and (F).
If (b), I offer (G) and (H) below:
(G): goodness for others = x; goodness for you = p; goodness of the outcome brought about from G = n
(H): goodness for others = y; goodness for you = q; goodness of the outcome brought about by H = (n + 1),
where x is greater than y, and p is greater than q.
Even here my intuition is that G is supererogatory, whereas MCers must hold that G is morally wrong. It seems to me that supererogation is a function of the relative goodness of outcomes *for others*, not a function of the goodness of outcomes simpliciter. I suspect, though, that neither you nor Campbell will endorse my intuition in this case.

18. Doug,
You write: “It may be, as I’ve argued elsewhere, that the moral permissibility of actions is affected by non-moral reasons as well as moral reasons.” This leads me to wonder what you mean by moral reasons. On one way of using that term, a moral reason is just any reason that pertains to (or affects) the moral permissibility of an action. But clearly that can’t be what you have in mind. (BTW, I was similarly puzzled by your use of moral goodness in your statement of Portmore’s Consequentialism. One might say that moral goodness is just the kind of goodness — whatever it is — that is relevant to moral permissibility.)

19. David says:

Let’s assume that supererogation is a “function of the relative goodness of outcomes for others, not a function of the goodness of outcomes simpliciter,” as Doug says. As has been made clear, MC makes permissibility and impermissibility a function of the goodness of outcomes simpliciter. This difference in inputs between supererogation and im/permissibility seems to me to be what has generated the whole problem here.
In order to solve this problem, we can try to remove the difference between the inputs of the function involved. To do this, we can either A. change the input of superogation, so it only depends on goodness simpliciter, or B. change the input of im/permissibility, so it only depends on goodness for others. Doing A. would, I think, quickly generate counterintuitive results. (These results are generated because there seem to be cases (such as that of E*) in which an action is supererogatory in virtue of its goodness for others, despite its relative badness simpliciter). Doing B. would force a modification of MC which would break its spirit. (This is because I take the spirit of MC to be the view that we ought only to take the action(s) with maximally good results.)
So, I think, we cannot remove the difference in inputs between supererogation and im/permissibility without either abandoning MC or generating deeply counterintuitive results. But maybe, as I’ve suggested above, we don’t need to solve this “problem of different inputs” at all.
If we preserve the difference in inputs between supererogation and im/permissibility, then we make four cases possible:
(a) An act which is supererogatory and permissible.
(b) An act which is supererogatory and impermissible.
(c) An act which is not supererogatory, yet is permissible.
(d) An act which is not supererogatory and is impermissible.
Only (b) is problematic. (In my last comment I mentioned some of the reasons for thinking (b) is problematic) But I think there are ways to make (b) less problematic.
One way is to say that supererogation is a notion pertaining only to the assignment of blame/praise. We sometimes want to refrain from blaming people even though we believe what they do is wrong. For instance, we recognize that a person can do the wrong thing with good intentions, and when this happens we (or some of us) want to refrain from blaming her. We may even praise her (though we might also want to correct her). Perhaps (b)-type acts fall into this general category. If a person does E* we say (if we are MC-ists) that she has done the wrong thing; nevertheless, we say that in doing E* she has done something praiseworthy in virtue of the willingness to self-sacrifice she has demonstrated. In that case, E*’s supererogation might consist in the fact that in doing E* a person has earned praise, even though in doing E* the porson’s action has been wrong.
This explanation of E*’s supererogation is strengthened, I think, by the fact that we may not want to call E* supererogatory if the person who does E* is mistaken about the outcome and believes she will, in doing E*, obtain a small benefit for herself at the cost of depriving others of a large benefit. If this is true, then supererogation, at least in this case, is likely a concept which pertains to the assignment of blame/praise to actors, and not directly to the assignment of right/wrong to actions. Specifically, in calling E* supererogatory we seem to be saying that the person who does E* has good intentions (if she understands what she is doing); and this does not rule out the possibility that she has nevertheless done the wrong thing.

20. Doug Portmore says:

Campbell,
You say, “On one way of using that term, a moral reason is just any reason that pertains to (or affects) the moral permissibility of an action.” Yes, that’s one way of characterizing moral reasons. But, on this characterization, we can’t even articulate what I take to be an interesting substantive question: namely, can other types of reasons/considerations besides moral reasons/considerations impact what it is morally permissible to do? Thus I prefer to characterize a moral reason as follows:
A moral reason to perform X is the type of reason that gives rise to the agent’s being morally required to perform X under the following conditions: (1) the agent has no reason (of any kind) not to perform X, and (2) the agent has no reason (of any kind) to perform some other available act that would in any way impede or preclude him or her from performing X.
It seems to me that one could argue (a) that the fact that your performing X would benefit you doesn’t give rise to your being morally required to perform X under conditions (1) and (2), (b) that the fact that your performing X would benefit someone other than yourself does give rise to your being morally required to perform X under conditions (1) and (2), and (c) facts of the kind described in (a) can affect whether facts of the kind described in (b) go on to generate a moral requirement to X. Perhaps, you would be morally obligated to do something that would benefit someone else but for the cost to you involved in your doing so.

21. Jamie says:

Doug, your test for moral reasons is hard to apply, because for lots of kinds of (intuitively) moral reasons it is hard to think of cases in which there would be no reason to refrain from the given act or perform an alternative act. For instance, that X would produce more equality in the world: it is hard to think of how that reason might bear on a situation without any other reason weighing in the (some) other direction.
It is also a little unclear to me whether some cases I can think of generate moral requirements, intuitively speaking. To clarify: suppose I can do a small favor for someone, which would help her out a little at no cost to me (or anyone else). Then surely I have some moral reason to do the favor, but it is not completely clear whether I am morally required to do it. (I am personally inclined to think that I am, but that’s because I am definitely convinced that I have a moral reason to do the favor, and on theoretic grounds convinced that if I have a moral reason to do something and no reason not to do it, then I am morally required to do it, so the problem is that applying the test seems to presuppose that which is supposed to be tested.)
Hmmm. I hope that clarified rather than obfuscating.

22. Jamie,
Thanks for the comment. Let me think about this, and then I’ll create a new post on how to best formulate the distinction between moral and non-moral reasons. This is really a whole new topic, and so we should probably move the discussion on this to a new post.

23. Campbell,
On reflection, it seems to me that you must reject Definition 3: An action X, performed by a person P, is supererogatory if and only if there is some available alternative Y such that (a) it is morally permissible for P to perform X but also morally permissible for P to perform Y, (b) P has more moral reason to perform X than to perform Y, and (c) P’s performing X is more demanding or self-sacrificing than P’s performing Y is. For if definition 3 is correct, and if MC is committed to the view that there’s more moral reason to perform X than to perform Y if and only if X’s outcome is better than Y’s outcome, then MC can’t allow for supererogatory acts. For the only way MC can meet condition (a) is if X’s outcome and Y’s outcome are equally good, but in that case condition (b) won’t be met. It also seems to me that Definition 3 is the one that best captures our notion that supererogatory acts are above and beyond the call of duty in the sense that we have even more moral reason to do the supererogatory act than the merely obligatory act. There’s something quite odd about MC holding that an act is supererogatory while denying that things would be morally better if one did the supererogatory act.