Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

In normative ethics there has been a long-standing debate between consequentialists and deontologists. To settle this dispute, moral theorists have often used a selective approach. That is, they have homed in on specific aspects of our moral practice and have attempted to show that their side can offer a more plausible theory of the normative factors that pertain to this specific field. Many theorists have, e.g., focused on the issue of lying (e.g. Bok 1978; Hodgson 1967; Kant 1785; Kant 1799; Ross 1930/2002; Sidgwick 1907). Others have taken up special obligations and duties of friendship (e.g. Aristotle 2010; Cocking and Oakley 1995; Sidgwick 1907). Yet others have discussed duties to oneself (e.g. Donagan 1979; Hill 1973; Kant 1785; Parfit 1986). One of the problems which has always occupied centre stage is the morality of promising. In this paper I consider this issue from a general moral-philosophical perspective.Footnote 1 I review arguments that have been put forward by consequentialists and deontologists and examine how far it is possible to adjudicate their dispute. I focus, in particular, on a new argument by Walter Sinnott-Armstrong who has claimed, contrary to received opinion, that consequentialism has the upper hand. My overall conclusion is negative. Given the arguments on the table, neither consequentialists nor deontologists have successfully established that their account of promissory obligation is the more plausible one.

The remainder of the paper is structured as follows. First, I consider the deontological and the consequentialist accounts of promissory obligation. Then, I examine how the issue of promissory obligation has been used to make a case against consequentialism. I focus, in particular, on the influential and much discussed arguments by Hodgson (1967) and Ross (1930/2002). After that, I consider the aforementioned argument by Sinnott-Armstrong (2009) which seeks to establish that consequentialists have a better conception of promissory obligation. Having rejected both the case against consequentialism and the case against deontology, I conclude in the last section with a few methodological thoughts on how the debate about the morality of promising may be adjudicated, if any.

Two Theories of Promissory Obligations

A good way to start the investigation is, I think, by getting clear on the two rival views that are at issue. There are certainly various ways of delineating both consequentialism (Sinnott-Armstrong 2011) and deontology (Gaus 2001a, b) from other moral outlooks. The following broad characterization, however, seems to be fair. Consequentialism, we may say, is roughly the idea that the rightness of an act depends solely on the goodness of its consequences, utilitarianism being the most paradigmatic instance of such a view. Deontologists, it seems fair to say, are united in their rejection of consequentialism.Footnote 2 To be sure, most of them agree that consequences matter morally. However, they reject the consequentialist idea that only consequences count. They believe, rather, that there are further aspects which determine the moral status of an act. These include, in particular, the intrinsic nature of the act itself (Kagan 1992). This said, we can move on to a tentative characterization of the consequentialist and deontological accounts of promissory obligations. Let us consider the latter first.

Deontologists such as Kant (1785), Prichard (1940/2002) and Ross and Stratton-Lake (1930/2002) believe that promises are one of the many factors which determine the rightness or wrongness of an act. They believe that the fact that I have promised to do X ipso facto creates an obligation for me to do X. There are, however, two main differences between the various deontological views. One difference lies in the way in which deontologists conceptualize the relation between promissory obligation and obligation in general. As Kagan (1998) explains, deontologists can view a promissory obligation either as fundamental obligation or as derivative obligation. Those who opt for the latter view can explain the obligation to keep a promise, e.g., as a special case of the duty of veracity. This does not seem all too implausible. After all, if I break my promise to do X, my assertion “I promise to do X” might be taken to have been a lie. Another way to make sense of promissory obligations in terms of a more fundamental moral requirement is to regard them as special cases of our duty not to harm others. If I have promised you to meet you for lunch, you will probably show up. If I then break my promise, I plausibly harm you in the sense that I frustrate your expectation and waste your time. The second way in which deontologists differ in their attitude towards promissory obligation concerns the weight with which they invest the duty to keep a promise. Radical deontologists, e.g. Kant (1785), interpret the obligation to keep a promise as an uncompromising moral duty that allows of no exceptions. Moderate deontologists, e.g. Ross (1930/2002), take it to be a prima facie duty. They allow it to be overridden when there are other moral concerns which seem to be more important.

While deontologists agree that promissory obligations have intrinsic normative significance (and disagree only regarding their weight and place in the system of moral duties), the consequentialist moral outlook seems to imply that there is no such thing as an intrinsic moral duty to keep a promise. Promises, after all, are a thing of the past, while consequentialism focuses only factors that pertain to the future, viz. consequences (Rawls 1955). Nevertheless, it is generally acknowledged that consequentialists can account for the fact that we normally have an obligation to do what we have promised (Habib 2008). It seems plausible to assume that under normal circumstances keeping a promise will have better consequences than breaking it. There are at least two reasons for this. The first reason is the fact that breaking a promise may have a bad immediate effect. It may harm the promisee in the way I have described above. The second reason is that the breaking of a promise may have a bad mediated effect. When I break my promise to you, you may tend to disbelieve my future promises. And this may make it harder for us to coordinate our behaviour in mutually beneficial ways. Breaking a promise may, in other words, affect the stability of the practice of promising.

The bottom line seems, then, to be this. Deontologists believe that a promissory obligation possesses intrinsic normative significance. On their view, the fact that I have promised to do X always counts in favour of doing X, even though moderate deontologists may, on occasion, allow this obligation to be overridden by more important moral concerns. In contrast, consequentialists appear to be committed to the view that a promise is, in and of itself, never a weighty moral reason, though they can explain why we should normally keep our promises. As far as promissory obligation is concerned, the dividing line between deontologists and consequentialists seems to be this. Deontologists believe that even a promise whose breaking causes no harm at all can be a weighty moral reason, while consequentialists are apparently committed to denying this. This, at any rate, is received opinion.

The Case Against Consequentialism

Many ethicists believe that the consequentialist position on promissory obligations should be rejected (and with it the doctrine of consequentialism itself). There are, broadly speaking, two arguments to this effect. I shall briefly examine each of them in what follows. One argument claims that the consequentialist position on promissory obligation is radically at odds with our moral intuitions. The other maintains that it is essentially self-defeating. I shall start with the latter argument.

Why should the consequentialist position on promissory obligation be self-defeating? Influential arguments to this effect have been proposed by Hodgson (1967) and Nida-Rümelin (1995). The reasoning proceeds in two steps. In the first step, it is established that the practice of promising is desirable on consequentialist grounds. This is a rather plausible idea, as David Hume has persuasively argued. He describes the notion of a promise as a “human invention” that is “founded on the necessities and interests of society.” (Hume 1888/1960, 519) The reason why its existence is consequentially desirable is that it provides us with an effective means to facilitate mutually beneficial interactions. When two individuals, A and B, exchange goods and services there is sometimes a time delay, such that either A or B has to move first. But they will do that only if the other party credibly assures that they will keep their end of the deal. This is where the institution of promising comes in. In order to incentivize A to move first, B has to promote the expectation that he will, in fact, do her part, given that A has done hers. She can accomplish this by promising A to keep to her end of the deal, thus creating a moral obligation to in fact do so. It is clear that making the promise is not only in the interest of A. B has a like interest in the regulation of his conduct. For if B does not commit herself to doing her part, A will not do hers and both will miss out on the opportunity to engage in a mutually beneficial interaction. It is sensible to suppose, therefore, that the practice of promising is desirable on consequentialist grounds. It makes everyone better off! (Theorists of the order-ethical tradition in business ethics, e.g. Homann 2002; Lütge 2005; Mukerji and Schumacher 2008 as well as Petrick and Pies 2007, have re-emphasized that this is the point of moral rules generally.)

In the second step of the argument, it is shown that the practice of promising could not arise in a consequentialist society. This, it is claimed, holds even under ideal conditions, that is, where everybody is a consequentialist and everybody knows that everybody else is a consequentialist and everybody knows that everybody knows that everybody is a consequentialist and so on. To see this, consider A’s and B’s problem again. For the practice of promising to get off the ground it is essential that B’s promise to A promote A’s expectation that B will do what she has promised. But how is this possible? By assumption, B is a consequentialist and A knows this. So when B says to A “I promise you to do X”, A will assume that B will do X only on the condition that this will have better consequences than not doing X. Since B knows that A knows that B is a consequentialist, B will expect that A will expect her to follow through on the promise only if she, B, judges this to be the best course of action. For this reason, B will not even bother making the promise, since it would be a waste of time. It seems, then, that two consequentialist agents, A and B, cannot utilize a social institution which is morally desirable according to their own moral outlook – a result that can, of course, be generalized to society as a whole. The issue of promising, it seems therefore, illustrates the fact that consequentialism is a self-defeating moral view. At least some moral theorists have taken this to be a disturbing conclusion (Stocker 1990).

The other argument seeks to establish that the consequentialist position on promissory obligation, as we have tentatively circumscribed it above, is counter-intuitive. We said that the obvious interpretation of the consequentialist view is that there is no intrinsic moral duty to keep a promise. To be sure, the consequentialist seems to believe that we often do have the obligation to do what we have promised, but not in virtue of the fact that we have promised this. The reason is, rather, that under normal circumstances keeping a promise will have better consequences than breaking it. Now suppose we face an abnormal case, such that breaking a promise would, in fact, have slightly better consequences than keeping it. In that case the consequentialist view might appear intuitively problematic, as Sir David Ross has famously pointed out. He asks the following question.

Suppose (…) that the fulfilment of a promise to A would produce 1,000 units of good for him, but that by doing some other act I could produce 1,001 units of good for B, to whom I have made no promise, the other consequences of the two acts being of equal value; should we really think it self-evident that it was our duty to do the second act and not the first? (Ross 1930/2002, 34–35)

Ross, of course, answers this question in the negative. And this appears to be in line with common sense. The consequentialist position, however, appears to imply the opposite. It seems, therefore, that it is not only self-defeating, but also counter-intuitive.

What can consequentialists say in reply to these arguments? I believe that they can reply in two distinct ways. First, they can confront the two criticisms head on. They can argue that it is not at all clear whether the arguments go through, even when they are targeted at the paradigmatic form of consequentialism, viz. utilitarianism. A number of theorists have pursued this line of counter-argument and have formulated interesting and powerful replies (e.g. Hörster 1973; Lewis 1972; Norcross 2011; Singer 1972; Sumner 1969). I shall neglect them, however. Instead, I want to focus on a second line of defence which makes a more fundamental point.

As a number of scholars have emphasized, consequentialism is an extremely capacious view (Dreier 1993; Portmore 2011). Utilitarianism which is one specific variant of the idea may (or may not?) fail by the arguments that we have considered above. But it is far from clear that this is the case with every version of consequentialism. The crucial point is this. Both arguments presuppose the position that I have tentatively ascribed to consequentialism above. I said that consequentialists seem to be committed to the view that the keeping or breaking of a promise does not, in and of itself, have any normative significance. But is this characterization adequate? To be sure, the idea is fairly wide-spread. And it is prima facie plausible. Consequentialism, after all, seems to focus only on the future, viz. on what happens after the act. Promises, one may say, are a thing of the past and do not pertain to the future. Nevertheless, this reasoning can be challenged. To show this, I shall make two brief points before I connect the dots.

One reason why one may doubt that consequentialists can acknowledge the intrinsic normative significance of promissory obligations is the belief that the breaking of a promise is an act. Hence, one may reason, it cannot be described as an act’s consequence. This, however, would be necessary to make it accessible to consequentialist evaluation. The first point that consequentialists may make, then, is that such reasoning presupposes a sharp and clear distinction between acts and their consequences when, clearly, there is no such thing.Footnote 3 As Larry Sumner points out, what I do “may be described variously as making marks on a piece of paper, signing a cheque, paying a bribe, or ensuring the survival of my business.” (Sumner 1987, 166) In other words, the boundary between the act and its consequences can be pushed back and forth, depending on the chosen description of the events. For this reason consequentialists can claim that the act itself should routinely be included amongst its consequences (Scheffler 1982/1994).Footnote 4 This, in turn, means that consequentialists can, in principle, take the fact that an act constitutes the keeping or breaking of a promise into account. (Critics of consequentialism may find this interpretation overly charitable. But it is clear that any attempt to show that consequentialists cannot do this would require us to restore the distinction between the act and its consequences in an arbitrary place).

Critics of consequentialism, I presume, have to grant this point. They can, however, make a second objection. They can argue that consequentialists cannot get any mileage out of this first point. For even if they can take into account the act itself, they cannot evaluate the fact that a given act constitutes the breaking of a promise. This fact about an act pertains, after all, to an event in the past, viz. the promise. And, as we know, consequentialism cannot take such past-regarding factors into account, as it is essentially a forward-looking moral view (Rawls 1955). This objection, however, suffers from a confusion. Of course, consequentialists can only consider features of the act that pertain to the present and the future. But they can use an axiology that takes past factors into account. As Derek Parfit explains, they may say that it is

better, for example, if benefits went to people who had earlier been worse off, or if we kept our promises to those who are dead, or if people are punished only if they earlier committed some crime. (Parfit 2011, 373–374)

How can consequentialists theoretically underpin these ideas? As Vallentyne (1988) suggests, they can commit to a historical axiology. Such an axiology does not evaluate the act merely in light of its impact on the present and the future. It evaluates how it impacts a whole world history, that is, past, present and future. Of course, no act can change the past. But it can fit the past in a way that is more or less morally adequate. Hence, consequentialists can claim that the axiological value of an act depends at least partly on how well it fits into world history as a whole. This idea seems to sit rather well with G. E. Moore’s well-received notion of organic unity. According to Moore (1903/1959), the organic value of a whole is not necessarily equal to the value of its parts. It may be greater or smaller, depending how well the individual parts fit together. In the present context that means that the axiological value of an act may not only depend on the act itself and its consequences, but also on how the act and its consequences fit into a given world history as a whole. It is plausible to assume that when an act constitutes the keeping of a promise there is at least one respect in which it fits well with its history. And this aspect, one might say, should plausibly materialize in its evaluation. As it turns out, then, consequentialists can recognize the intrinsic normative significance of a promise.

This result can help consequentialists to rebut the two aforementioned arguments. Let us consider each in turn. The first, recall, says that the institution of promising could not exist in a consequentialist society. Would-be promisees would recognize that promisors will keep their promises only if this has the best consequences. Hence, they will not believe the promise. Since the latter know this, they will not even bother making a promise and the institution of promising collapses. Now consider what changes if we allow consequentialist agents to hold a historical axiology of the appropriate kind. In that case they can consider the breaking of a promise to be a bad consequence of the act (Broome 1991). Suppose, again, that B promises A to do X. By hypothesis, A knows that B is a consequentialist and that she considers the breaking of a promise to be a bad consequence. A will, then, judge that if B promises her to do X, she will have a good consequentialist reason to keep her promise. Hence, A will expect B to keep her promise. B, in turn, will know this. She will, hence, make her promise. Self-defeat is, thus, averted.

The second argument, recall, claims that it is counter-intuitive to suggest, as consequentialists do, that a promise may permissibly be broken, even if this has only slightly better consequences than keeping it. We considered David Ross’s example where keeping a promise confers upon the promisee a benefit of 1,000 units of good and breaking it confers upon somebody else who is not the promisee a marginally greater benefit of 1,001 units of good. Under these circumstances, thinks Ross, it is not reasonable to judge that the promisor should break the promise. But, as we have seen, consequentialists are not committed to this view. Those who accept the intrinsic normative significance of promising may judge that the badness of breaking the promise in this situation outweighs the marginal increase in overall welfare that is thereby effected. Their recommendation will therefore be not to do it. To be sure, I do not mean to suggest that any historical axiology will do this trick. All that I claim is that consequentialists seem to possess the theoretical resources to bring their moral outlook in line with our common-sense moral judgements about Ross’s example. This, however, should be enough to address the objection.

The Case Against Deontology

As we have seen, it seems that the deontologists’ case against consequentialism rests on a shaky foundation. Consequentialists, it appears, can account for the intrinsic wrongness of promise-breaking just as well as can deontologists. This prevents their view from being self-defeating and furthermore brings it into line with our intuitive moral convictions. As far as the issue of promissory obligation is concerned, the jury is still out on consequentialism and deontology. But perhaps this changes once we consider the consequentialists’ arguments against deontology. Contrary to received opinion, the consequentialist account of promissory obligations may, after all, have more to offer than the deontologist account. In a recent paper, Sinnott-Armstrong (2009) has argued towards precisely this conclusion. In what follows I shall consider and reject his reasoning.Footnote 5

Sinnott-Armstrong’s argument is based on a change in perspective. He notes that the question whether consequentialists or deontologists can give us a more plausible conception of promissory obligation is usually discussed under a particular presupposition. It is assumed that we are after an explanation and systematization of our intuitive conviction that it is somehow morally wrong to break a promise. But this, Sinnott-Armstrong thinks, is only part of what we should expect from a philosophical account of promissory obligation. We should also be interested to know how and why different promissory obligations differ in strength. Consequentialists, he claims, can do a better job than deontologists when it comes to explaining this. He suggests that the consequentialist doctrine can easily be extended for the purpose at hand. On consequentialism, it may be said, I am obligated to do an act A to the extent that this has better consequences than doing anything else. If the difference in goodness between doing A and doing the second best act, B, is small, I have a weak obligation to do A. If this difference is large, I have a strong obligation to do A. Therefore, it seems sensible to suppose that I have an obligation to keep a promise to the extent that this has better consequences than anything else I might do. Sinnott-Armstrong hypothesizes, therefore, that the strength of a promissory obligation should be strictly proportionate to the harm that breaking the promise causes (at least when there are no other morally relevant consequences). If this was correct, it would seriously embarrass deontologists who, thinks Sinnott-Armstrong, do not possess an explanation that is similarly congenial to their moral outlook. This, however, is not the case, as I shall go on to show. First, however, let me state Sinnott-Armstrong’s argument.

Sinnott-Armstrong’s argument builds on John Stuart Mill’s method of concomitant variation. This method is used to check whether two empirical phenomena, x and y, are causally connected. We may conclude that variations in x cause variations in y if x and y are correlated and if we have ruled out alternative explanations which may explain this correlation. In particular, we have to rule out that (i) the correlation is accidental, (ii) variations in y cause variations in x (rather than vice versa) and (iii) variations in some other factor, z, cause variations in both x and y.Footnote 6 To confirm his claim that the strength of a promissory obligation is strictly proportionate to the harm that breaking the promise causes, Sinnott-Armstrong proceeds, then, along the lines of Mill’s method. First, he observes that the harm which is done by the breaking of a promise is correlated with the strength of the associated obligation of the promisor to keep the promise. He does so by looking at two cases:

AIRPORT: I have promised you to drive you to the airport. If I break my promise, you will miss your flight and this will have quite bad consequences for you (e.g. you have to buy an expensive new ticket, you miss an important meeting and so on).

LUNCH: You have invited me and some other folks over for lunch. I have promised you to come. If I break my promise, the harm that is thereby done is very little. You will be a bit disappointed, but you will still have an enjoyable lunch with your other friends and we can still see each other some other time.

As it turns out, my promissory obligation in AIRPORT appears to be much stronger than in LUNCH. There seems, then, to be a correlation between the harm of promise-breaking and the strength of the respective promissory obligation.

The second step in the argument consists in ruling out the alternative explanations for this correlation: (i) The correlation is accidental. (ii) The strength of a promissory obligation explains the level of harm which is caused by the breaking of the promise. (iii) There is a third factor which explains both the harm of promise-breaking and the strength of the obligation. Sinnott-Armstrong rules out these possibilities with a few parsimonious remarks. (i) is off the table, he says, “because consequences are at least part of what matters in morality” (Sinnott-Armstrong 2009, 440). Hence, a correlation between the strength of an obligation and the consequences of violating it should not be accidental. (Indeed, only radical deontologists would deny this.) (ii) should be dismissed, because the strength of an obligation apparently cannot explain the harm of violating it. E.g., in AIRPORT the comparatively greater strength of the obligation “cannot explain (…) why it is so bad to miss this flight” (ibid., 440). Finally, he rejects (iii) with a rhetorical question: “what would that third factor be?” (ibid., 440).

In the remainder of this section, I shall focus on two aspects of the argument: the alleged correlation between strength of obligation and harm of violation and the dismissal of the alternative hypothesis (iii). I shall conclude that there is, indeed, a correlation between strength of obligation and harm of violation. But this correlation is not perfect. And I shall argue that Sinnott-Armstrong falsely dismisses (iii), since there is a third factor which explains both the harm of violating an obligation and its strength. The first point disproves his hypothesis. The second explains why it is wrong. After that, I shall propose a more plausible account of the strength of promissory obligations.

Let me start with the correlation. If the strength of an obligation to keep a promise is, indeed, fully explained by the harm that violating it causes, then there should be no two cases in which the levels of harm are different, while the promissory obligations are equally strong. As I shall argue, however, there are pairs of such cases. Consider the following modification of LUNCH.

LUNCH*: You have invited me and some other folks for lunch and I have promised you to come. As in LUNCH, if I break my promise, the harm to you is very little. You will be disappointed, but you will still have an enjoyable lunch with your other friends and we can still see each other some other time. But there is this one person, Walter, whom you have also invited. Walter is a philosopher like me. You have told him that I have found a mistake in a paper of his which he intends to present at a conference on the very next day. This conference is really important to him. So he has a very strong interest in seeing me about the apparent problem in his paper.

If I break my promise in LUNCH*, I disappoint you. The harm to you is the same as it is in LUNCH. But there is an additional harm, the harm to Walter, whose (much stronger) interest in seeing me is also disappointed, if I do not show. If Sinnott-Armstrong’s claim that the strength of an obligation and the harm of its violation are perfectly correlated was correct, I would have a stronger obligation to keep my promise in LUNCH* than in LUNCH. But the obligations are equally strong. To see this, we need to turn to the question how the strength of an obligation is measured.

Sinnott-Armstrong explains that the relevant strength of an obligation “is measured by how much is needed to override the obligation” (ibid., 439). In LUNCH, he says, a reason which may justify that I break my promise is that my teenage child is sick at home, even though the kid would be safe without me while I am away. It suffices that I say something like this: “Hey, I know I have promised you to come to your lunch and I know you would like to see me. But my son is sick and I feel that I should stay at home with him.” Now it seems to me that I could say the very same thing to excuse my promise-breaking in LUNCH*. If you then pointed out that Walter would be extremely disappointed if I did not come, I could say something like this: “Oh, that’s too bad. Please tell Walter that I am very sorry I can’t come, since my son is sick.”

Now this comparison, in and of itself, does not establish my claim that my promissory obligations in LUNCH and LUNCH* are equally strong. The reason why the same justification suffices to justifiably break the promise in both cases may not be that my obligations in LUNCH and LUNCH* are equally strong. Rather, it may be that my justification is strong enough to override even my stronger obligation in LUNCH*. So let us consider a third case which indicates that this is not so:

LUNCH**: You have invited me and some other folks for lunch and I have promised you to come. If I break my promise, you will be mildly disappointed, because you would have liked to see me (as in LUNCH and LUNCH*). More importantly though, you will not get my feedback on a paper of yours which you will present at a conference. Apparently, I have found a mistake in it and it is really important to you that I tell you about it before the conference.

In LUNCH** there is no Walter. The interest he had in seeing me is now your interest in seeing me. The overall amount of harm that I do if I break my promise is then the same as in LUNCH*. So if, as Sinnott-Armstrong believes, the harm done by breaking a promise determines the strength of the obligation, the justification that sufficed to break the promise in LUNCH* should suffice in LUNCH** as well. But it does not. At any rate, so it seems to me. In LUNCH**, I cannot say: “Listen, I can’t come to your lunch. I know I have promised to come. And I know you need my opinion on your paper before you go to that conference. But my son is sick and I feel I should stay with him.” If I said this, you would have every right to ask me: “Is your son so sick that he needs you twenty-four-seven? Can you not come over for an hour or two? You have promised that you would. And you know that I really need your opinion on my paper before I go to that conference.” I would have to concede: “No, I guess he can stay by himself for an hour or two. And you’re right. I have promised you to come and I know that you really need my comments. So I guess I should come over.”

Our intuitive responses to LUNCH, LUNCH* and LUNCH** establish, then, contrary to Sinnott-Armstrong’s claim, that the harm of promise-breaking does not explain the comparative strength of promissory obligations. Nevertheless, Sinnott-Armstrong is not entirely on the wrong track. In general, the harm of promise-breaking seems to go hand in hand with the strength of obligation, as his paper shows. This is an important insight which, however, Sinnott-Armstrong fails to adequately explain. In what follows, I shall draw out an account of promissory obligations which rectifies this. It can explain cases like LUNCH and LUNCH** in which the strength of obligation and the harm of violation are correlated and cases like LUNCH* in which this correlation breaks down.

Let me start with an important distinction. Sinnott-Armstrong talks about the harm of promise-breaking. I think, however, that we should discern two kinds of harm a promisor may do by breaking her promise. The first kind of harm is the harm that is done to the promisee, harm P . The second kind of harm is the overall harm which is done, harm O , that is the harm which is done to the collective of morally relevant subjects which includes not only the promisee. These two kinds of harm are, of course, closely connected. Usually the two are identical, since the harmful consequences of a broken promise are typically all borne by the promisee. This is the case in AIRPORT, LUNCH and LUNCH**. In all three cases, you, the promisee, are the only person who is harmed if I break my promise. In AIRPORT, the harm of my promise-breaking is the fact that you miss the flight (plus all bad consequences this may have for you). In LUNCH it is your slight disappointment that I do not come. And in LUNCH** it is your disappointment about not seeing me plus the fact that you do not get my comments on your paper. There are, however, cases in which harm P and harm O come apart, such as in LUNCH*. In LUNCH*, if I break my promise, the harm to you, the promissee, is as small as in LUNCH, while the overall harm which includes Walter’s disappointment is as great as in LUNCH**.

Now my account of the strength of promissory obligations turns on the distinction between harm P and harm O and goes as follows:

The promissory obligation in CASE A is at least as strong as the promissory obligation in CASE B if and only if harm P in CASE A is at least as great as harm P in CASE B.

If this account is correct, it means that Sinnott-Armstrong made a mistake by dismissing hypothesis (iii) in step two of his argument. There is, then, a third factor which can explain both the strength of the obligation to keep a promise and the harm of its violation. This third factor is harm P . Variations in harm P explain both variations in the strength of a promissory obligation, as my account says, and variations in harm O , since harm P is a part of harm O . Cases like LUNCH* in which the strength of the obligation does not correlate with the harm of its violation can be explained by the fact that variations in harm O may occur not only because of variations in harm P , but also for independent reasons.

Now let us check whether my account fits our intuitions about LUNCH, LUNCH* and LUNCH**. We observed that the strength of my promissory obligations in LUNCH and LUNCH* were equally strong, while the obligation in LUNCH** was stronger. According to my account, the levels of harm P must then be the same in LUNCH and LUNCH*, while the level of harm P in LUNCH** is greater. Indeed, this is the case. If I break my promise in LUNCH, the harm to you, the promisee, is your mild disappointment about not seeing me. The same goes for LUNCH*. In LUNCH**, though, the harm to you, the promisee, is your mild disappointment about not seeing me plus your more severe disappointment that you do not get the comments on your paper. So, in LUNCH**, harm P is greater than in LUNCH and LUNCH*. It seems, then, that Sinnott-Armstrong is partly right. The harm of promise-breaking can apparently explain the strength of an obligation. However, it is only the harm to the promisee that is relevant to the strength of the promissory obligation.

Now it may be objected that my view has a somewhat absurd consequence. Imagine a case in which the breaking of my promise involves very little harm P , but an enormous amount of harm O . Assume, e.g., that in LUNCH* Walter will not only be very disappointed if I do not come. Rather, he will die. Does this not increase the strength of my obligation to keep my promise and come to your lunch? No, it does not, since “[u]nlike paradigmatic moral duties, the duty not to harm for example, promissory obligations are not owed equally to everyone, but rather only those we have promised.” (Habib 2008; emphasis added) In LUNCH*, you are the promisee. Hence, I owe it to you, and you only, that I keep my promise. harm P is the only factor which determines the strength of my obligation to do that. Note, however, that there may be independent reasons to act in keeping with my obligation (or otherwise). The fact that Walter would otherwise die is one such reason. To be sure, then, I am not claiming that harm O is a morally irrelevant factor. I am only claiming that it is a factor which is irrelevant to the strength of a promissory obligation, since a promissory obligation is owed only to the promisee(s).

A few qualifying remarks seem to be in order. It is, I believe, important to stress that the account that I have presented as an alternative to Sinnott-Armstrong’s hypothesis pertains only to the question how the strengths of promissory obligations are to be compared relative to one another. It is silent on and, therefore, compatible with the view that promises have intrinsic normative significance which, as we saw above, can be held both by consequentialists and deontologists. There is, then, no reason to think that my account represents a stage victory for the one or the other side. But it may, perhaps, seem that my account is somehow more congenial with the consequentialist paradigm, because it proportions the strength of a promissory obligation to the bad consequences, viz. the harm, that breaking a promise causes for the promisee. But I do not see why this should be the case. Deontologists generally recognize the normative significance of doing harm (Kamm 2007). And, as I said above, some of them even conceptualize the obligation to keep a promise as subcategory of the duty not to harm. They of all moral theorists should be able to recognize that the strength of the obligation to keep a promise is proportional to the harm for the promisee that would result from breaking it. As it turns out then, on closer inspection there is nothing in Sinnott-Armstrong’s interesting and insightful paper which can show that the consequentialist position on promissory obligation is superior to the deontological one.

Conclusion

Let me review the dialectic. I have argued that when it comes to the morality of promising neither deontologists nor consequentialists can claim the upper hand. I substantiated this assertion by reviewing arguments on both sides. First, I reviewed the case against consequentialism. I addressed the concern that the consequentialist position on promissory obligation is self-defeating and the allegation that consequentialists cannot account for our intuitive judgements about the morality of promise-breaking. I argued that both arguments can be addressed once it is granted that consequentialists can, contrary to received opinion, account for the intrinsic normative significance of the obligation to keep a promise. After that, I considered whether the consequentialist position on promissory obligation may, in fact, have an edge over the deontological one. To this end, I looked at a new argument by Walter Sinnott-Armstrong. He claims that consequentialists can make more sense of the relative strengths of our promise-keeping obligations. I showed why and where he errs and introduced a more plausible account of obligational strength which appears to be compatible both with the consequentialist and the deontologist perspective. The bottom line of all of this is that there seems to be nothing in the debate about promissory obligation which makes the consequentialist or the deontological position appear preferable.

How, one may ask, should we then choose between the two views? How, that is, can we figure out whether we should think of our obligation to keep promises in consequentialist or deontological terms? In concluding, I would like to suggest an answer to this question which can, I think, teach us a valuable methodological lesson about the way we should approach practical matters. Some applied ethicists seem to believe that all we need to do in order to settle on a normative conclusion about a practical question is to check which of the available moral-philosophical answers to a given case appears to make most sense. As we have seen, however, this is not always possible. In the case of promissory obligation, at any rate, there is no clear-cut answer to the question whether the consequentialist or the deontological account is preferable. In such a situation, I think, we should back up and look at the bigger picture. We should consider the respective positions in light of their general merits. Should it turn out that consequentialism is more attractive than deontology on independent grounds (or vice versa), then this would give us a reason to prefer its conception of promissory obligation and to address practical issues of promissory obligation in its terms. The ways in which theory and practice are connected are not often obvious. Nevertheless, more often than not it may turn out that we have to climb up the greasy pole of theory before we can give a truly justified answer to a practical question. As Kurt Lewin rightly remarked, “Nothing is more practical than a good theory.” (Lewin 1951, 169)Footnote 7