A Critique of Scanlon’s Contractualism

Abstract

Part of T. M. Scanlon’s project in What We Owe to Each Other (1998) is to explain the importance and priority of moral reasons. But Scanlon also argues that this priority of moral reasons is compatible with the pursuit of other things we value, such as friendship. To this end, Scanlon claims that contractualist moral reasons internally accommodate our interests in such values. In this paper, I argue that Scanlon is unsuccessful in showing the compatibility of morality and the pursuit of our other values. The contractualist may not be able to be a good friend.

1. Introduction

Many moral theorists believe that if there is moral reason to act in some given circumstances, that reason is conclusive. Morality enjoys a privileged place of importance and priority among our reasons for action. In What We Owe to Each Other (1998), T. M. Scanlon aims to defend (i) this privileged status of his contractualist moral reasons and (ii) the compatibility of these conclusive moral reasons with the pursuit of other values.¹ I will call this latter claim ‘Compatibility.’ Delivering on Compatibility is important: if adherence to moral reasons precludes realizing other important nonmoral values, then agents will have good reason to question why they should act morally at all.

Scanlon’s argument for Compatibility involves showing that his contractualist moral reasons are internally responsive to other things we have reason to value. Because morality is determined by what agents seeking to live on justifiable terms with others could reasonably reject, morality just is an arbiter of agents’ interests. It follows that if agents have reasonable interest in, e.g., friendship, then it is plausible that sufficient space for friendship can be carved within morality. If these conclusive moral reasons are responsive to other values, then the fact that they have a unique priority among our reasons for action does not threaten to swamp an agent’s life.

I argue that Scanlon is unsuccessful in securing Compatibility. His moral reasons are insufficiently responsive to what we have good reason to value.² I present two objections to Scanlon’s argument for Compatibility. My weaker objection claims that while Scanlon purports to capture our ordinary intuitions concerning the moral permissions generated by friendship, he offers no explanation for such lines drawn by conventional morality. I call this the explanatoriness objection. In this case, while contractualism is compatible with the pursuit of nonmoral values, it is unclear how. My stronger objection claims that friendship fails to generate sufficient permissions to act in ways that would otherwise be impermissible. Whatever permissions are generated fail to protect space for good friendship; they make possible only okay friendship. I call this the structural objection. In this case, Scanlon fails to accommodate Compatibility.

In this paper, the nonmoral value I focus on is friendship. This is because Scanlon explicitly defends the compatibility of friendship with his version of morality and intends the treatment of friendship to generalize to other nonmoral values.³ This attention is not ill-placed: friendship is a particularly important value, given its necessary role in a meaningful human life. Additionally, we commonly recognize ourselves to have obligations with respect to our friends; to fail to act as friendship requires is no small matter. So, if contractualist moral reasons conflict with reasons of friendship, it is not obvious that agents should act morally. Friendship therefore serves as an exemplary value for which morality should make space. Thus, my argument is that the contractualist is not guaranteed to be able to be a good friend; these values are not necessarily compatible. Any value that is structurally similar suffers the same fate.

2. Friendship

I begin by saying a bit more about friendship. Friendship, like many values, can be realized to different degrees. We recognize okay friends, good friends, and best friends without positing different kinds of relationships or values.⁴ Colleagues can be friends, rival athletes can be friends, parent and child can be friends. The relationships realized between these individuals may be importantly different, but they share the following necessary features which unify them as one kind of thing: they act on reasons for each other, adopt each other’s ends, provide each other support, and share affection. This should be an uncontroversial and familiar conception of friendship.

Each of these features can be realized to a greater or lesser degree. Best friends presumably further many of each other’s interests (thereby acting on many reasons for the other), share many ends, are each other’s first line of support, and share strong affection. But an individual who shares some of our ends, feels some affection toward us, and so on, is still a friend. Your colleague who would not read a draft of your paper but would sit with you rather than someone else at lunch may be an okay friend. Your colleague who would read a draft of your paper and give you feedback but would not pick you up from the airport at 2 a.m. may be a good friend. And your colleague who would read a draft of your paper and give you feedback and pick you up from the airport at 2 a.m. may be a best friend. They are all friends, and one metric of picking out how good of a friend they are is by the reasons they act on with respect to you—presumably both the volume and import of the reasons.

Scanlon claims to protect a meaningful form of friendship we have reason to value, and not merely a watered-down form of friendship. I suggest that to make good on this claim, space not only for okay friendship but also for good friendship must be protected. I cannot spell out in precise detail what this amounts to, and we need not accept the morally contentious adage that a friend will help you move while a good friend will help you move a body. But it does gesture toward a line we are familiar with—vague though it may be—that separates good from okay friends. This line seems primarily concerned with the reasons your friend will act on for you. For Scanlon to make good on Compatibility, then, I claim that he will have to show that friends can act on the right sorts of reasons in the right kinds of cases such that they can be good friends.

3. Scanlon’s view

Scanlon’s first line of defense for Compatibility is to introduce a version of friendship that is built upon a moral relationship. If friendship is defined around moral reasons, then reasons of friendship will not recommend that you do something immoral.⁵ There should be no case in which acting on a moral reason is incompatible with acting on some reason of friendship. Therefore, this version of friendship is guaranteed to be compatible with morality.

What does this version of friendship look like? Scanlon thinks that it is one that most of us (“us” being human agents in the late twentieth century) intuitively recognize. He uses an example to demonstrate our familiarity with it. Consider a friend who would steal a kidney from some stranger for you because you need one. Scanlon thinks that you will be unnerved by this, and that you do not think that there is a reason of friendship for your friend to steal a kidney for you. It is unnerving, Scanlon claims, “because of what it implies about [your] ‘friend’s’ view of your right to your own body parts: he wouldn’t steal them, but that is only because he happens to like you” (1998, 165). There is something objectionable going on here: your friend is willing to violate another human’s bodily rights because he likes you. He recognizes the moral claims only of friends; if you are not his friend, he does not see you as a person with moral standing. This is why you might be unnerved, then: it is merely contingent that your friend isn’t stealing your kidney—luckily for you, he likes you. But if he didn’t, then your kidney would be less secure.

What this is intended to demonstrate is that we expect our friends to respect the moral claims of ourselves and others. The kidney case, that is, generalizes to all moral claims. As Scanlon explains it, if “the conception of friendship that we understand and have reason to value involves recognizing the moral claims of friends qua persons, hence the moral claims of nonfriends as well, then no sacrifice of friendship is involved when I refuse to violate the rights of strangers in order to help my friend” (165). Limits are placed on reasons of friendship by what is morally required; the good friend would never violate the rights of others in order to help a friend because friendship itself does not require it. As Scanlon writes: “Compatibility with the demands of interpersonal morality is built into the value of friendship itself” (165; emphasis added). As such, there would be no reason of friendship to do something immoral.

This will help Scanlon a bit; it ensures that reasons of friendship and reasons of morality will not conflict. But that alone is insufficient to make good on Compatibility. Paring back the claims of friendship so that they are compatible with morality might ensure that the moral agent and “friend” is not conflicted, but this does not amount to rendering compatible morality and the intrinsically valuable form of friendship that is necessary for a good life. It could, instead, protect only okay friendship, or a “watered-down” form of friendship.⁶ Consider that the form of friendship left could be one in which it is permissible to act on reasons of friendship only when one, say, chooses to volunteer at a charity in a time slot with their friend rather than with a stranger. This would be excessively restrictive of the scope of reasons of friendship. But, as far as we can currently see, this may be the only form of friendship Scanlon can render compatible with his morality. So, we’ll want an assurance that morality is sensitive to friendship just as friendship is sensitive to morality. Friendship, that is, should be capable of generating permissions to act in an otherwise impermissible manner. This would allow us to recognize and protect the intrinsic goods of friendship. And, ideally, that would guarantee the compatibility of good friendship and morality.

Let’s now look at Scanlon’s contractualist account of morality to assess whether he makes morality sensitive to friendship. Scanlon’s contractualism holds that an act is wrong if and only if its performance under the circumstances would be forbidden by a set of principles for the general regulation of behavior that no one suitably motivated could reasonably reject (153).⁷ Reasonable rejection itself is comparative.⁸ What this means is that an individual might be burdened by the acceptance of p ₁ and the rejection of p ₂, but if another individual bears a greater burden by the acceptance of p ₂ and the rejection of p ₁ and there are no alternate principles, then p ₂ can reasonably be rejected and p ₁ cannot. Note also that only generic reasons can generate reasonable rejections. Generic reasons are based on information about what individuals have reason to want in virtue of their situation; no particular information about individual subjects can factor into reasonable rejection. This serves the dual purposes of limiting the demandingness of acting morally—one need not calculate the effects on each individual given their particular preferences—and picking out what is morally salient—what people have reason to want, not what they possibly irrationally or indefensibly do in fact want.⁹

What qualifies as a burden, or what can be a ground for reasonable rejection? Scanlon thinks that grounds for reasonable rejection (nonexhaustively) include loss of well-being, unfairness, inadequate recognition of one’s entitlement to something, and impossibility of recognizing other values one has good reason to recognize (218–19).¹⁰ In justifying the inclusion of the last sort of potential objection, Scanlon explains that “insofar as these are things that people have reason to pursue and to value, these reasons will be among those that can make it reasonable to reject some principles. Therefore, there will be pressure within the morality of right and wrong to make room for these values” (166). This is where Compatibility comes in. Scanlon thinks, correctly, that people have good reason to want to recognize the value of friendship.¹¹ So, if a potential principle made it impossible to recognize the value of friendship, it could reasonably be rejected (depending on what the alternate principles and their costs are).¹² This would help guarantee the compatibility of friendship (which, recall, is our representative nonmoral value) and morality.

It should be clear that for Scanlon, morality is something like an adjudicator of agents’ interests. Morality defines how we can act justifiably with respect to one another, and that is determined by weighing our interests and seeing which are sufficiently important to protect. Friendship is one such interest that will ideally be protected in the moral sphere.

Let me note that accidental, actual conformity with some value is insufficient for protecting the benefits of that value, and therefore would be insufficient to defend Compatibility. Consider Scanlon’s treatment of privacy to see why this is so. Scanlon notes that in order for us to have the benefits of privacy, it cannot merely be the case that people happen to not go through our stuff and listen in on our phone calls. Rather, we need an assurance that people will not do this. Only then can I plan around, e.g., making a call concerning sensitive information. One way of securing this assurance is achieved by the general acceptance of a principle that protects privacy (203). We should expect a similar protection within the principles of contractualist morality for friendship if good friendship is to be compatible with morality. Only then can I plan around, e.g., being there for my friend when she needs me.

Let’s consider a case to test Compatibility: call the case ‘Loan.’ Your friend is strapped for cash and has asked to borrow $50 so that she can pay to run her air conditioning at a comfortable temperature during next week’s heat wave. There are many people starving and in need of potable water, and there are effective charities that can provide them these needed basic resources as they are given donations. You have $50 to spare and could either loan it to your friend or donate it to one of the effective charities. What could Scanlon’s contractualism say about this case?

The main principle operational here seems to be Scanlon’s ‘Rescue Principle.’ The Rescue Principle states that, “[i]f you are presented with a situation in which you can prevent something very bad from happening, or alleviate someone’s dire plight, by making only a slight (or even moderate) sacrifice, then it would be wrong not to do so” (224). On the one hand, since you have $50 to spare, it would presumably involve little sacrifice to donate it to an effective charity that can help prevent malnourishment and dehydration. On the other hand, it is possible that it comes at a greater than moderate sacrifice to deny your friend this loan, even though you know the money would be going toward something comparatively luxurious. If so, it seems plausible that the reason of friendship will prevail here. Whether it does will depend on how important being able to aid your friend is, how much aid you have given to effective charities in the past, and so on.

This is good for Scanlon. In the important cases, it seems likely that these reasons of friendship generate new moral permissions. So, it seems that Scanlon can defend Compatibility.

4. Analysis of Scanlon’s view

4.a The explanatoriness objection

However, there is reason to be skeptical of Scanlon’s ability to accommodate and explain the breadth of cases we want.¹³ For Scanlon to deliver on Compatibility, it must not only be the case that he can generate the correct range of moral permissions for friendship; it should also be the case that his moral theory can explain in a principled manner why and when such permissions are generated. Scanlon himself embraces this explanatory burden. He notes that unlike aesthetic judgments, moral judgments cannot reduce simply to “this act is wrong”; we must be able to explain, through “reasons and principles,” why the act is wrong:

The emphasis that contractualism places on justification, hence on reasons and principles, captures a central feature of everyday judgments of right and wrong. Typically, our intuitive judgments about the wrongness of actions are not simply judgments that an act is wrong but that it is wrong for some reason, or in virtue of some general characteristic. Judgments of right and wrong are in this respect quite different from many other types of evaluative judgment such as judgments that something is beautiful, or ugly, or funny. In the latter cases the evaluative judgment comes first—we “see” that the thing is beautiful or funny—and the explanation comes later, if in fact we can supply it at all. But we rarely, if ever, “see” that an action is wrong without having some idea why it is wrong. There may be cases in which some action “just seems wrong,” even though one cannot say what the objection is. But these reactions have the status of “hunches” or suspicions which need to be made good: there is pressure to come up with an explanation or else withdraw the judgment if we cannot explain what our objection is. (Scanlon 1998, 197–98)

In this section, I argue that Scanlon’s contractualism cannot always explain the permissions it grants. My strategy in this section is to argue in what Scanlon calls an “upward” manner: I begin with cases in which it seems clear that a principle can (or cannot) reasonably be rejected and then analyze the cases to see on what grounds those principles are (or are not) reasonably rejected.¹⁴ This strategy reveals that the contractualist’s means of generating moral permissions may be ad hoc. Rather than a principled basis for the new moral permissions that render morality compatible with friendship (and, by extension, like values), we find a patchwork of intuitions recalcitrant to explanation.

Let me begin by considering a new case: call it ‘Stranger Aid.’ Abby and Bane are each poised to suffer some pain in the immediate future. Abby will suffer the destruction of one of her kidneys; Bane will suffer the lesser harm of a broken leg. Abby and Bane are both strangers. You can prevent only one of their horrible accidents, and either intervention would come at little cost to yourself. Is it permissible for you to intervene on Bane’s behalf rather than Abby’s?

Recall that the Rescue Principle demands that an agent who can prevent something very bad from happening at little or moderate cost to herself do so. Abby’s succumbing to an accident that would destroy one of her kidneys qualifies as something very bad, and it would come at little cost for you to intervene on her behalf. Bane’s potential harm is less bad than Abby’s, and that is morally salient here. If you can prevent only one of their horrible accidents, mustn’t you intervene to prevent the worse accident? Scanlon (plausibly) thinks so (1998, 227–28). So, if both Abby and Bane are strangers, and Abby faces more dire prospects than Bane, she could reasonably reject a principle that did not require you to rescue her on account of her worse scenario. So, you owe Abby aid under the Rescue Principle.

Now let’s introduce a new case: call it ‘Friend Aid.’ Friend Aid changes one feature of Stranger Aid: rather than Bane being at risk of breaking his leg, your friend Bruce is. The other features of these cases are the same. Abby stands to lose one of her kidneys and remains a stranger to you, and you can save only one person. The question is whether the fact that Bruce is your friend makes a difference to what you ought to do. Are you morally permitted to save Bruce rather than Abby despite her worse scenario?

Scanlon will almost certainly want to say that you could aid Bruce rather than Abby in Friend Aid. Consider the Rescue Principle again. You are required to prevent something very bad from happening to someone when it would impose a slight or moderate sacrifice. But it looks like the introduction of your friend changes another feature of the case: it would require a greater than moderate sacrifice from you to save Abby, since you would have to eschew a reason of friendship and allow your friend to break his leg. It is one of the quintessential reasons of friendship to be able to prefer your friends in situations like this—to be able to aid them over strangers, plausibly even if the stranger is in more need.¹⁵ So, it is plausible that you can (and should) save Bruce instead of Abby.¹⁶ For these reasons, let’s accept that your objection to the principle that required you to save the stranger (Abby) instead of your friend (Bruce) is stronger than Abby’s objection to being left to suffer greater misfortune under the principle that permitted you to aid Bruce.

But why is this the case? Why does Friend Aid generate a permission to aid Bruce instead of Abby, while in Stranger Aid you must aid Abby? There is surprising asymmetry between the verdict in Friend Aid and that in the bodily rights case (call it ‘Kidney’). Recall Scanlon’s treatment of Kidney, in which your friend would steal a kidney, or violate another’s bodily rights, for you. Scanlon claimed that friendship would not give you a reason to take the kidney because friendship would never require you to do something that violated the rights of others. So, in Kidney, friendship does not generate a permission to act in a way that would otherwise be impermissible. But in Friend Aid—in which your friend’s survival is not on the line—a reason of friendship does generate a permission to act in a way that would otherwise be impermissible. Why does friendship generate new permissions in Friend Aid, but not in Kidney?

The reason for this asymmetry is not immediately evident. Why does the objection “I would be unable to prevent my friend’s pain” qualify as a reasonable rejection to the principle that would require you to prevent the destruction of a stranger’s kidney, but the objection “I would be unable to prevent my friend’s death” fail to qualify as a reasonable rejection to the principle that would forbid you from taking a stranger’s kidney? We will want an account of why reasons of friendship count as the right sort of reason in Friend Aid but not in Kidney. Scanlon will not have given us a satisfying account of Compatibility if the reason why morality and friendship are compatible is simply that we think it makes sense in the former case but not the latter. This reports our intuitions but does not explain them. So, let’s consider some potential explanations of this asymmetrical generation of moral permissions between the two cases.

For a first possibility, one could say that there is a salient difference between causing harm and withholding aid. In Kidney, you are required not to cause harm; you are required not to violate a person’s bodily autonomy and take their kidney.¹⁷ In Friend Aid, you are permitted only to withhold aid; you are permitted to aid Bruce rather than Abby despite the fact that in other circumstances you would be required to help her due to her worse predicament. The suggestion is that reasons of friendship can never give us permissions to cause harm but can give us permissions to withhold aid. So, one could say on Scanlon’s behalf, the principled distinction for when reasons of friendship can generate permissions in our contractualist calculus is only when we would not directly cause harm to a stranger.¹⁸ That would explain why you cannot steal a kidney for your friend but can help Bruce rather than Abby. Stealing a kidney would cause harm to the stranger whose kidney you’ve taken, but helping Bruce rather than Abby doesn’t cause her any harm—you merely withhold aid.

But this is not right. This distinction doesn’t adequately capture the cases we want. Consider a new case: call it ‘Alligator.’ In Alligator, you can either save a stranger, Chris, from excruciating death by pulling him out of the closing jaws of an alligator or wave to your friend Dolly across the lagoon (she’ll see you only if you wave right now). Surely you must save Chris. Neither you nor Dolly could reasonably reject the principle that required you to save the stranger from impending death on the grounds that you would be unable to act on a reason of friendship. But Chris certainly could reasonably reject the principle that permitted you to act on a reason of friendship instead of saving a stranger from impending death. So, you must act to save Chris at the cost of greeting your friend Dolly.

However, Alligator is the same sort of case as Friend Aid, if not a bit ramped up—what is at issue in both is only withholding aid. In both cases, you either provide aid to a stranger or you act on a reason of friendship. But in Alligator, unlike in Friend Aid, friendship does not generate a permission to do what is otherwise impermissible. So, it is not the case that friendship can generate permissions whenever only withholding aid is at issue. We have a new asymmetry and understanding when friendship can generate permissions to withhold aid seems to bottom out in a prior understanding of the costs at issue. The distinction between causing harm and withholding aid is insufficient to explain our asymmetries.

Here one might respond that even if when withholding aid is permissible bottoms out in a prior understanding of the costs at issue, that is fine so long as we can give a principled account of these costs. For instance, one might suggest that we should be able to provide our friends only with serious aid over strangers even if the stranger is in more need. This would explain the asymmetry between Friend Aid and Alligator—you would be providing Bruce with serious aid, but not Dolly. In this case, we might roughly state the principle of permissions of withholding aid as follows: if one’s friend stands to suffer serious harm and a stranger stands to suffer very serious harm, it is permissible to aid one’s friend rather than the stranger; otherwise, one must help the person who stands to suffer the greater harm. Put another way, if one’s friend stands to suffer harm n and a stranger harm n+1, friendship is sufficiently important to offset that +1 to generate a permission. But when the gap between the harms is greater—such as in Alligator, where it would perhaps be n and n+10—friendship is insufficiently important to offset the disparity of harm to generate a permission.

There are two problems with this suggestion. First, I am skeptical that this can provide a principled explanatory basis for when withholding aid is permissible, for how are we to judge these comparative costs of withholding aid? What renders the needed aid serious or merely trivial? In Kidney, the harm one stands to suffer from not receiving a kidney is very serious indeed, and without an assumption of bodily rights, the harm the stranger stands to suffer from losing a kidney is less severe given that one can function perfectly well with a single kidney. But providing such aid to your friend is impermissible. To make these judgments, we need either a prebaked understanding of the moral costs at issue—which would make the contractualist moral theory circular, as Nicholas Southwood argues¹⁹—or our particular unfounded intuitions about such costs—which would be ad hoc, as I argue. In either case, we lack the principled kind of explanation that Scanlon himself seeks.²⁰

But second, even if we can adequately explain this cost assessment, the suggestion is extensionally inadequate: it fails to generate a permission in Loan. It is implausible that the harm your friend will suffer in not receiving a loan from you to run her air conditioning at a comfortable temperature is serious.²¹ On the other hand, the burden of starvation and dehydration faced by the global poor is very serious. If we must weigh the severity of potential harm done to our friends against that done to the global poor, it would be impermissible to aid your friend in Loan. But failing to generate a permission in this case would be very demanding indeed.²²

One might point out that I provided a counterexample only to the withholding aid bit, and so think that there is still something informative the distinction gives us: it may still be true that reasons of friendship cannot generate permissions to cause harm where it would otherwise be impermissible. But there’s a plausible counterexample to this as well. Consider a new case: call it ‘Gouty Toes.’ While walking, I must either step on a stranger’s gouty toes or my friend’s gouty toes. Surely I’m permitted to step on the stranger’s toes rather than my friend’s, even if the stranger has a worse case of gouty toes. But that causes the stranger harm it would otherwise have been impermissible to inflict. The “causing harm” and “withholding aid” distinction faces counterexamples on both sides. It is insufficient to explain our asymmetries.

A second option for Scanlon would be to posit a ranking between the different grounds of reasonable rejection. Maybe security of bodily integrity comes first, then some other interests, then friendship, then relief from horrific suffering, and so on. If there is such a ranking, then we can appeal to it when deciding which objection is the strongest with respect to a certain set of potential moral principles, and we would have a principled way of doing so.

However, this would not work since the ranking could not be rigid: friendship comes before relief from horrific suffering in Friend Aid, but after it in Alligator. We would have to appeal to something further to explain when the ranking shifts or when there are special exceptions. But it isn’t clear what that would be, and that’s what we’re looking for. So, this doesn’t seem like a promising explanation of our asymmetries.

I do not have the space to assess all the available options, but I have canvassed what I take to be the most plausible ones and found them lacking. It seems that what the contractualist relies on are just our unexplained intuitions about each particular case. If this is correct, then Scanlon faces a serious problem: the way values generate moral permissions in his account is mysterious. Until we have a principle that explains our intuitions about these new moral permissions, we must simply take Scanlon’s word that his account can capture Compatibility. He hasn’t offered us a defense of the claim. As far as we know, it may be true, or contractualist morality might require more than commonsense morality, and it may be false.²³

4.b. The structural objection

It is, however, unclear that friendship can generate permissions to act in ways that would otherwise be impermissible. In this section, I argue that the contractualist procedure doesn’t sufficiently allow for friendship to be a ground of reasonable rejection. Friendship would provide a ground for reasonable rejection were its realization made impossible by a potential principle (depending on alternative costs), but as it turns out, friendship will nearly always be possible given the multiple realizability of the reasons of friendship. The problem is that the version of friendship that is nearly always possible is merely okay friendship. And when good friendship is measured against other interests, it is comparatively too weak to ground a reasonable rejection. The structures of Scanlonian friendship and contractualism combine infelicitously to make it hard to see how Scanlon could make good on Compatibility.

First, recall that Scanlon insists that the version of friendship he seeks to protect is one that has built-in sensitivity to the demands of morality. If so, then there’s something strange in thinking that friendship could give you a basis for objection to a potential moral principle. The good friend wouldn’t expect you to steal a kidney for her when she needs one, since friendship is incompatible with such immoral acts. So, if he is a good friend, would Bruce want you to save him instead of Abby, if he is also (in virtue of being a good friend) a moral agent? It’s not clear why he should, or how he should. There is a structural difficulty in thinking through how Scanlon can permit the value of friendship—or any value that already accepts the claims of morality—to shape morality at all.²⁴

Let me elaborate on this. Scanlon is concerned to protect only a morally constrained version of friendship. As we saw in Kidney, which is intended to generalize, the good friend does not have reasons of friendship to do anything immoral. For this version of friendship to generate an objection to any putative moral principle, it would have to have some content that could be curtailed by such a principle. But friendship is structured around the demands of morality. Scanlon noted that the value of friendship has the demands of interpersonal morality built into it. So, how could a potential moral principle constrain the content of friendship if friendship already accepts whatever moral principles there are? It seems that it could not.

This is problematic for contractualism. Scanlon introduced the possibility of friendship constraining morality to show the compatibility of friendship and morality. But if he cannot license friendship to provide an objection to any potential principle, then this compatibility is not secured: friendship cannot generate a new permission unless it can generate a sufficiently strong objection to a possible principle.

One might object that I have shortchanged what sort of content friendship can have and, thus, how it could provide an objection to a potential moral principle. Friendship is not simply a moral relationship even if Scanlon is right that friendship is sensitive to moral demands. There are some goods of friendship, external to morality, that could figure in as bases for reasonable rejection. For instance, friendship requires that you further your friend’s interests in certain ways, that you be there to aid them when they require it, and so on. But you don’t owe these things to strangers; they are unique goods of friendship. So, wouldn’t a friend be able to object to some principle that made these things impossible? If so, friendship could influence the principles of morality.

I think it is correct that there are nonmoral goods of friendship and that this could be true of Scanlon’s morally informed friendship. But this fact is insufficient to guarantee that the value of friendship can serve as a basis for reasonable rejection to some potential moral principle. Whatever other content there is to friendship, it is sufficiently abstract that it can be rendered compatible with almost anything morality requires. So, even though this other content of friendship could potentially provide an objection to a moral principle, it will not; morality comes first and defines how exactly we understand these other goods of friendship.

To see that this determining relation exists between morality and the other goods of friendship, consider the nonmoral requirement of friendship to promote your friend’s interests. Clearly, promoting your friend’s interests is limited in Scanlon’s moral version of friendship—you can pursue only her morally sanctioned interests.²⁵ After all, as we saw in Kidney, you cannot further your friend’s interest in life by stealing a kidney for her. While living is a real interest of your friend’s—and while it is true that you, as her friend, have reason to promote her interests—Scanlon does not think you have a reason of friendship to steal a kidney for her even though having a kidney is necessary to secure her interest in living. Morality forbids this, and so it is not part of friendship.

But there are still some interests of your friend’s that you can act to realize. You can, for instance, donate to the causes that are dearest to her, or help her move to a new neighborhood (but only if she has bought or rented the new property and not stolen it). Importantly, the only interests of your friend’s that you can act on are the ones that are not morally forbidden. This sort of moral determination plausibly extends to any other nonmoral requirement of friendship. These requirements of friendship are sufficiently abstract that they can be constrained in this way—that helps explain how so many different forms of friendship are familiar to us, and yet, nonetheless, are classified as friendship. So, it is clear that for Scanlon, the goods of friendship are already morally tinged. Thus, even though there is nonmoral content to friendship, it does not seem to generate objections to any possible moral principles, much less generate new permissions.

As an example, let’s again consider the requirement of friendship that you pursue your friend’s interests. Could a friend object to the principle that requires she steal no kidneys on the basis that this makes it impossible to pursue her friend’s interests? No, she couldn’t. There are still many other ways in which she could pursue her friend’s interests (she can donate to her friend’s favored charity, send out emails requesting kidney donations, etc.). Her objection that it is impossible to pursue her friend’s interest is false. It is still quite possible. In sum, the abstractness or multiple realizability of the requirements of friendship makes it the case that they cannot offer any genuine objection to a potential moral principle. You can always do something else—something that is morally permissible—to meet your friendship duties.

Let’s see how this finding bears on Friend Aid. I said above that Scanlon could probably permit you to save Bruce rather than Abby. But can he? Consider the Rescue Principle, which requires you to prevent great harm to another when it comes at little or moderate cost to yourself. Does it in fact come at greater than moderate cost to you to save Abby rather than Bruce given that you can further Bruce’s interests in ways other than just sparing him a broken leg? After all, you can save Abby’s kidney, then drive Bruce to the hospital, take him to physical therapy, etc. Acting on any of these reasons is still acting on a reason of friendship; you still further Bruce’s interests. So, saving Abby rather than Bruce would not prevent you from acting on a reason of friendship, and it would not make the realization of the value of friendship impossible. Wouldn’t you have to save Abby then? According to Scanlon’s Rescue Principle, I suspect you would.

This presents a problem for securing Compatibility. For morality to be compatible with friendship, it should be sensitive to friendship. But we see that it is not; even in important cases, friendship cannot generate permissions to do what is otherwise impermissible.

One might push back on my claim that because the goods of friendship are multiply realizable and there is always something else you can do, friendship cannot generate new permissions. After all, those other actions may be unsatisfactory. Driving Bruce to the hospital, taking him to physical therapy, and so on, are certainly ways of being Bruce’s friend. However, they are poor replacements for preventing Bruce’s leg from breaking in the first place. If all the other options are unsatisfactory, then wouldn’t friendship generate a permission not to save Abby in Friend Aid on the grounds that you couldn’t be a sufficiently good friend? Isn’t that a greater than moderate cost?

Well, what would your objection to the Rescue Principle qua good friend be? I will argue that whatever objection you could pose is comparatively weak and that it is unlikely it would amount to a reasonable rejection in the face of other burdens. This structural difficulty arises from the comparative nature of contractualist reasonable rejection. While good friendship might be capable of posing an objection to a possible principle, it is unlikely to generate any new permissions. Furthermore, because accidental, actual conformity with the value of friendship is insufficient for giving us the benefits of friendship, as we saw in the case of privacy, this inability to generate new permissions shows that being a good friend is not compatible with morality.

To show this, first consider that a comparatively strong objection to a potential principle would be that it would make the realization of a value impossible. Scanlon explicitly uses this language of impossibility when demonstrating that morality is sensitive to nonmoral values in his argument for Compatibility; he claims that we can be sure the demands of morality are sensitive to nonmoral values because principles can be reasonably rejected if they left no room for or made impossible acting on these values (1998, 160, 204). There is, therefore, reason to think this strong form of objection is the only sort of objection Scanlon thinks nonmoral values like friendship could issue to potential moral principles. It is perhaps telling to consider Scanlon’s example for showing how morality is sensitive to nonmoral values: he claims that a principle requiring strict neutrality between friends and strangers could reasonably be rejected because it makes acting on the value of friendship impossible (219). That is a weak assurance for friendship; so far, all it assures us is that you can, e.g., volunteer with a friend rather than a stranger, or that you can be an okay friend.

If impossibility of realizing a value is a necessary condition on posing an objection to a possible principle, then assurance for good friendship is all but lost. Recall that friendship is a value that comes in degrees, and good friendship is a higher degree of friendship. But it remains a realization of friendship; one could fail to be a good friend and still be a friend. And it is false, evidently, that you cannot act on a reason of friendship at all in Friend Aid if you must save Abby. Saving Abby does not make friendship impossible since it does not preclude driving Bruce to the hospital. Furthermore, it is implausible to claim that acting on an unsatisfactory reason of friendship amounts to the impossibility of friendship. So, this objection doesn’t work.

Let’s assume that Scanlon can permit weaker objections than the impossibility of realizing a value. In that case, it seems your objection would be that you are unable to best act on your friend’s interests. How else could we parse the objection that you cannot act on the satisfactory reason of friendship? The most natural objection would be that you cannot be a good friend to Bruce. But this intuitive way of capturing the objection reduces to being unable to best realize a value or to realize the value to a greater degree; that would be the generic reason at play.

Now, this is a comparatively weak objection to a principle. Best realizing anything rings of luxury when compared to plausibly more basic interests like prevention or reduction of physical pain and protection of bodily health. (Note also that there is something absurd about claiming that best maintaining your health is luxurious, in a way in which claiming that best realizing the value of friendship with another individual is not, even though it is also a necessary part of a good life.) As noted, impossibility of realizing a value is not comparatively luxurious; it, reduction of pain, and protection of bodily health all seem relatively basic and comparatively weighty. But that does nothing to protect good friendship. And this is what we need to deliver on Compatibility.

Let me say a bit more to motivate why the objection that we cannot be good friends would be comparatively weak. I’ve claimed that good friendship is a necessary part of a good life. Accordingly, each of us has a strong interest in being able to realize this value. But, according to contractualism, this strong interest must be balanced against the competing interests of others. I find it highly plausible that an individual’s interest in food and water (as in Loan) or protection of bodily organs (as in Friend Aid) is comparatively weightier than the interest in being a good friend (even with the friend’s own interests in bodily integrity and comfort factored in). If we may appeal to a hierarchy of needs, we can see this more clearly. Good friendship is essential to a good life, but it becomes possible only after more basic needs are secured. (To be there for your friend, you must have reliable access to food and shelter, protection of bodily security, etc.) If so, then it seems clear that if my higher-level needs are compared to your more basic needs, yours win out. In a nutshell: the root problem for contractualism is that without a presupposition of either a right to good friendship or some antecedently protected space for an agent’s own projects, relationships, and so on, we can derive protections for such interests only against the competing claims of other agents. If so, a guarantee for good friendship seems possible only in a world more idyllic than our own, one in which everyone’s more basic interests are already secured.²⁶

Consider, for a final time, Friend Aid. Does being unable to best act on your friend’s interests amount to a greater than moderate sacrifice in the face of the destruction of Abby’s kidney? I think not: it seems that Abby’s well-being generates a stronger objection than having to settle for the best way of realizing a value. It also seems likely that this would generalize across cases, though not to the point at which one is barred from acting on any reason of friendship, since that would amount to making friendship impossible.²⁷ The contending objections would have to be quite weak for the objection that you are unable to best realize a value to win out.

So, given that reasonable rejection is comparative, if the objection that a potential moral principle makes you unable to best realize a value is comparatively weak, it follows that it is unlikely that this objection will win out in cases of conflict. But it would have to, in order to guarantee space for good friendship. So, Scanlon has not shown that morality is compatible with good friendship.

Note that morality is compatible with what I identified above as okay friendship. This is protected since Scanlon noted that the principle of strict neutrality could reasonably be rejected insofar as it would make friendship impossible; furthermore, any principle that entailed strict neutrality might likewise reasonably be rejected (depending on alternate costs at issue). But that is the extent of permissions that friendship is likely to generate. Recall that friendship is a value that comes in degrees, and that part of how we track good friendships and okay friendships is on the basis of the reasons the friends may (and do) act on with respect to each other. From the rejection of strict neutrality, we are guaranteed the permissibility to act on some of the “lesser” reasons of friendship. You are permitted to do something to help Bruce in Friend Aid, such as taking him to the hospital after aiding Abby. You are permitted to do something to help your friend in Loan, such as sending her an article on how to stay cool in the summer months. But, as noted above, this meager compatibility is insufficient to allay the worries that motivate accommodating Compatibility in the first place. Being permitted to act on only these reasons is insufficient for good friendship; this is not, intuitively or familiarly, the sort of thing we expect our good friends to do for us. Yet, since okay friendship and good friendship are two sides of one and the same value, only the former is protected. When compared to the other pressing interests of strangers considered in the cases above, the better realization of a value is a luxury.

I acknowledge that I walk a fine line here. I have claimed that friendship can generate new moral permissions, but that these permissions are both rare and inadequate, since only okay friendship can generate satisfyingly strong objections. I suggested that okay friendships are insufficient for Compatibility, and that only good friendship will do. This is indeed a fine line, but I believe it is defensible. Okay friends—whom I would intuitively describe as people with whom you are friendly, with whom you spend time and know quite well, but in whom you would not necessarily confide—are good. But they are not the good friends that we cherish, in whom we find unflagging support, and with whom we, in important ways, share life. I suggested at least part of the way we can differentiate these forms of friendship is based on the sorts of reasons of friendship the friends act upon, and I have argued that only the reasons befitting okay friends are protected. Furthermore, I now claim that no number of okay friendships is capable of providing the meaning found in the best of friendships.²⁸ Nonetheless, these friendships are still of the same kind as okay friendships. So, while okay friendship is compatible with morality, good friendship is not.²⁹

Let me now connect this finding back with Compatibility. Scanlon’s claim there is that the priority and importance of contractualist moral reasons are compatible with the pursuit of other values. There is a weak and unsatisfying sense in which this is true. For any value, it will be the case that we can pose comparatively strong objections to potential principles that would make the realization of that value impossible. So, for any value, we are guaranteed room to realize its “okay” form. But surely this is an unsatisfying guarantee. Dabbling in values is far from delivering us the source of meaning and importance in our lives available through full-fledged commitment to such values.³⁰ So, in so far as we care about Compatibility because we want assurance that we can act morally without sacrificing our own lives, projects, commitments, and so on, we have not gotten sufficient proof that contractualism can deliver on Compatibility. Instead, we have good reason to be skeptical that it can.