Blaming friends

The aim of this paper is to shed light on the complex relations between friendship and blame. In the first part, I show that to be friends is to have certain evaluative, emotional and behavioral dispositions toward each other, and distinguish between two kinds of norms of friendship, namely friendship-based obligations and friendship-constituting rules. Friendship-based obligations tag actions of friends as obligatory, permissible or wrong, whereas friendship-constituting rules specify conditions that, if met, make it so that two persons stand in a particular type of relationship defined by various friendship-based obligations. I argue that whereas friendship-based obligations apply to actions under direct voluntary control, friendship-constituting rules apply to emotional and evaluative attitudes. The second part develops an account of friendship blame by comparing Scanlon’s account of blame with Wallace’s Strawsonian account of blame. I demonstrate that Scanlon’s account picks out responses that become appropriate when friends’ attitudes are not in agreement with friendship-constituting rules, whereas Wallace’s account picks out responses that become appropriate when friends violate friendship-based obligations. Arguing that the responses picked out by Scanlon’s account do not amount to blame, I show that, when combined, the views give an illuminating picture of possible reactions to friends who fall short of the standards of friendship.


Introduction
While we occasionally blame the passer-by on the street or the stranger on the subway, our understanding of blame is largely shaped by interactions in the context of personal relationships: first and foremost, we blame our partners, relatives and friends. In view of this, it is remarkable that the protagonists of the examples from the literature on blame tend to be stripped-down anonymous ''agents'' who violate general moral obligations. This paper focuses on blame in the context of friendship. 1 In the paper, I first give a phenomenologically and psychologically rich picture of the nature and norms of friendship and then develop an account of friendship blame by examining and evaluating two influential accounts of blame, Wallace's Strawsonian account on the one hand and the account developed by T.M. Scanlon in Moral Dimensions on the other. 2 The paper consists of two parts. The first part provides an account of the nature and norms of friendship. In Sect. 2, I illustrate that to be friends is to have certain evaluative, emotional and behavioral dispositions toward each other. Section 3 concentrates on what I will call ''friendship-based obligations.'' These are rules that tag potential actions of friends as obligatory, permissible or wrong. I show that noncompliance with a friendship-based obligation constitutes a 'wrong' or a 'violation' and argue that friendship-based obligations apply to actions under direct voluntary control. Section 4 concentrates on what I will call ''friendship-constituting rules.'' These are rules that specify conditions that, if met, make it so that two persons stand in a particular type of relationship defined by various friendship-based obligations. I show that non-compliance with a friendship-constituting rule renders a friendship of a specific type 'null' and 'void,' and argue that these rules apply to attitudes and dispositions.
The second part of the paper draws out the implications of this account of friendship for our understanding of blame. In Sect. 5, I introduce Scanlon's account of blame by analyzing his central example of the disloyal friend Joe. Section 6 compares Scanlon's account with Wallace's account of blame. Based on my analysis of the nature and norms of friendship, I demonstrate that the responses picked out by Scanlon's account are best characterized as reactions to attitudes that are not in compliance with friendship-constituting rules, whereas the responses 1 For ease of exposition, I will use the term ''friendship blame'' throughout the paper to denote blame within the context of friendship. The term should not be understood as introducing a qualitative distinction between friendship blame and what might be called ''moral blame'' (i.e., the kind of blame normally under discussion in philosophical debates). Rather, it refers to moral blame in the context of friendship; that is, it refers to what we do when we morally blame a friend as friend. 2 In this paper, I set aside several questions. First, I will not address justificatory questions about the value of friendship or about the reasons one may have to start or sustain particular friendships. For a discussion of these questions, see Frankfurt (2004), Velleman (1999) and Kolodny (2003). Second, while on my account of friendship, friends are typically open to experiencing second-personal reactive emotions with regard to each other, I will remain noncommittal about the question whether our involvement in personal relationships in general and friendships in particular entails our susceptibility to the reactive emotions. Shabo (2012aShabo ( , 2012b and Coates (2013) have argued that our involvement in personal relationships entails our susceptibility to the reactive emotions, while Pereboom (2001, 199-204), Sommers (2007) and Milam (2016) deny this. picked out by Wallace's account are best characterized as reactions to violations of friendship-based obligations. Arguing that the responses picked out by Scanlon's account do not amount to blame, I show that both accounts capture essential aspects of our reactions to friends who fall short of the standards of friendship.

The nature of friendship
Part of what is missing in the available interpretations of blame is that they are not embedded in a phenomenologically rich account of the nature and norms of personal relationships. Such an account must be spelled out in dispositional as well as normative terms: to count as friends, two persons must have certain evaluative, emotional and motivational dispositions with regard to each other and be normatively bound to each other in a special way. Deferring the discussion of the normative aspects of friendship to the following two sections, I focus on its dispositional aspects here.
Building on the work of Helm (2009), it can be shown that the dispositions that agents must have in order to count as friends can be expounded on three dimensions, namely the dimensions of mutual care, intimacy and shared activity. 3 Let us start with mutual care. To care about something is to have a rational pattern of evaluative, emotional and motivational dispositions with regard to that thing (Helm, 2009, 75). Thus, to care about your car means, among other things and everything else being equal, to judge it a bad thing for it to be dirty, to be disappointed at learning that your partner took it off road, and to be motivated to bring it to the car wash on the next day. Since agents are the kind of beings who care about things, to care about someone as an agent is, other things being equal, to care about her cares. Caring about a friend's concerns, one need not care about the objects of a friend's concerns for their own sake; rather, all other things being equal, I must care about my friend's concerns for her sake (Helm, 2009, 84-86). Thus, if my friend cares about minimal music and I care about my friend, I may not care about the album release of her favorite artist for its own sake, but I should, all else being equal, be happy for my friend that the album is released.
To be able to see the broader spectrum, think of a friend and note that your friendship is partly constituted by your friend's evaluative, emotional and motivational dispositions with regard to the things you care about. To mention but a few examples, your friend will judge it a good thing that your paper was accepted for publication and a bad thing that it was rejected. She will hope that you will be awarded the prestigious scholarship that you applied for, just as she will fear that the committee might decide otherwise. And if you move house and ask your friend to help you out, she will be motivated to lend a hand. You and your friend care about each other if you have comparable evaluative, emotional and motivational dispositions with regard to your friend's cares. Importantly, friendships can typically endure occasional lapses of appropriate responses as long as a general pattern of such mutual dispositions is in place; as these lapses increase, however, friendships tend to destabilize and break down.
Besides mutual care, friendship requires some level of intimacy. Thomas (1987Thomas ( , 1989Thomas ( , 1993 emphasizes that friends create a sphere of intimacy by revealing their deepest secrets to each other. By telling each other what they would never dare to tell to others, friends make themselves vulnerable to each other, and thereby implicitly communicate their trust in the other's good will. But the scope of intimacy is broader than this (Cocking & Kennett, 1998, 514-518). Besides their deepest secrets, friends also share their interests and values. These may merely overlap in a somewhat contingent way, as with tennis buddies and the like, but in close friendships the interests and values will often be deeply interwoven. Our friends influence our interests and values, whether it is by playing us a record of an artist they have just discovered or by sharing with us their conception of the good life. Sometimes, the connection is so strong that it makes sense to say that friends share an evaluative perspective. Friends share an evaluative perspective if each friend takes the fact that the other friend values something to provide herself with a pro tanto reason to value that thing as well (Helm, 2009, 282-290;Friedman, 1989, 4). This may account for the particular way friends engage in shared activity, and it is to that dimension that I will turn now.
Friends do things together. They pursue projects together, as when they jointly start a band or a reading group, they play together, whether it be cards or baseball, and experience things together, as when they jointly go on a trip. Shared activity among friends is not mere social cooperation. Two agents may choose to cooperate because, given the situation, it is the best way to pursue their own interests. It is different with friends. For them, sharing the activity is part of the fun: part of the reason why they want to engage in some shared activity at all is the mere fact that the activity is shared. This can be explained by reference to the notion of a shared evaluative perspective. If I do not have time to inform myself about the plays that are running at the moment, it seems perfectly reasonable for me to delegate the decision about which show to see to my close friend. Unlike partners in social cooperation, we share an evaluative perspective: I trust my friend on her taste and suppose she trusts me on mine, and the fact that she chooses to see a certain performance provides me with a pro tanto reason to see it as well. And should the performance turn out to be a disappointment, then at least we are in it together. Though neither of us may value the performance, sharing the experience can still be valuable to us.

Friendship-based obligations
Now turn to the normative aspects of friendship. I will assume that we have special obligations to friends and that these special obligations are friendship-based. Special obligations are obligations the content of which cannot be specified without reference to the parties involved; that is, they are obligations that cannot be derived from, or reduced to, natural moral duties. To explain why we have a special obligation to a friend, we usually cite the nature of our relationship to her. Thus, I might say that I ought to help Julie move to her new house ''because she is my friend.'' To say that special obligations to friends are friendship-based is to take such expressions at face value; it is to say that these special obligations are based on the relation of friendship and cannot be derived from other obligation generating mechanisms, such as promising, consent or the acceptance of benefits. 4 Friendship-based obligations have two characteristics that follow analytically from the fact that they are obligations. First, friendship-based obligations define which types of behavior are obligatory, permitted or prohibited in the context of a friendship. 5 Second, once friendship-based obligations are in place, failure to comply with those obligations acquires a specific normative status and constitutes a 'wrong' or a 'violation.' The scope of friendship-based obligations does not encompass all aspects of friendship. It seems odd to say, for instance, that a friend is under an obligation to have an interest in certain things. For one thing, the fact that one ought to have an interest in, say, Italian cinema is not the right kind of reason to be interested in Italian cinema. For another, it would seem that ''ought'' implies ''can'' in the sense that the scope of obligation is restricted to actions that are under our direct voluntary control. An action x is under our direct voluntary control just when it is typically the case that if one decides to x, one will x; and if one decides not to x, one will not x. Whether we are interested in certain things is normally not under our direct voluntary control: you might want me to be interested in free jazz and I might want myself to be interested in free jazz as well, but-I must confess-I will probably not be able to bring myself to do what I want in this case.
The same holds for emotions. Since by a mere act of will I cannot produce an emotional state, let alone acquire an emotional disposition, it would be implausible to say that I have an obligation to feel a particular emotion or to be disposed to feel particular emotions in particular circumstances. By saying this, I do not mean to deny that our interests and emotions are under voluntary control. They are, at least to a considerable extent. Yet the kind of control we have over them is not direct. We exercise indirect control over our interests and emotions by having direct voluntary control over actions that promote or impede the development of certain interests or emotional dispositions. Accordingly, it makes perfect sense to say that it would be desirable that I have a certain interest or emotional disposition and therefore have an obligation to perform those actions that under normal circumstances contribute to the development of that interest or disposition.
If the scope of friendship-based obligations is restricted to actions under direct voluntary control, friendship involves mutual obligations like the following: on the 4 The classical statement of a non-reductivist view of relationship-based obligations is Scheffler (1997). For two recent defenses of this view, see Owens (2012b) and Wallace (2012). 5 Friendship-based obligations are not always overriding. Suppose, for example, that a friend confides in you that she is responsible for the recent surge of fires in the neighborhood and asks you not to tell anyone. In this case, it seems that you have a friendship-based obligation to keep the secret and a moral obligation to report the crimes to the police, where the moral obligation seems to override the friendshipbased obligation. On these assumptions, you would act morally by reporting the crimes to the police, even if it would still make sense to say that by doing so you wrong the friend as friend. See Cocking and Kennett (2000) for a broader discussion of these issues. dimensions of mutual care, the obligation to help a friend move house upon her request and the obligation to visit a friend in hospital when she is ill; on the dimension of intimacy, the obligation not to reveal things a friend confided in you; and on the dimension of shared activity, the obligation not to turn down an invitation to a birthday party for reasons of convenience. Of course, friendship also entails special permissions. To cite a few prominent examples, friends may be permitted to ask each other intimate questions, to call each other at 11.30 pm, or to come by for dinner uninvited.

Friendship-constituting rules
How do we come to have obligations to friends that we do not have to others? On the assumption that the special obligations of friendship are friendship-based, two parties acquire friendship-based obligations to each other simply by becoming friends. Our question is thus as follows: how do people become friends? Providing an answer to this question is not as easy as it may seem. On the one hand, friendships are different from familial relationships in the sense that people do not simply find themselves in them. On the other hand, people typically do not become friends intentionally either. 6 That is, for a friendship to develop, it is neither necessary nor sufficient that the parties involved have the intention to become friends. It is not necessary because people can become friends by regularly meeting each other, even if none of the parties intends, over and above that, to become friends. It is not sufficient because people do not become friends simply by communicating the intention to become friends: one does not make a friend in the way one makes a promise.
The existence of what I call ''friendship-constituting rules'' can explain how two persons become friends. To get our discussion on a concrete footing, consider how David Owens became friends with a colleague (Owens, 2012a, 102). At some point, Owens found himself taking the bus home from work together with a colleague. Though he did not especially like the colleague nor intended to become friends with him, in the course of their daily conversations and exchanges of small favors nonetheless a friendship grew between them and they both came to have the attitudes and emotional dispositions characteristic of it. From then on, they recognized such conversations and exchanges as obligatory.
According to Owens, this example shows that friendship is a ''socially defined package'' of ''norm, attitude and behavior'' and that ''to behave in certain ways […] is to buy into the package, is to make it the case that further attitudes are apt or inapt and new forms of behavior permissible or obligatory'' (2012a, 102). Although Owens certainly points in the right direction here, I think this is not exactly right. It is more accurate to say that I become your friend by performing those actions that would justify you in inferring certain emotional and evaluative attitudes on my part.
These actions are the kind of actions that people would typically not perform were they to lack those attitudes.
An example will illustrate the point. Berenice is an exchange student who meets up regularly with a couple of other exchange students, not because she likes them, but because she prefers hanging around with them to staying home alone. Basically, she sees them as a means of passing the time. If she meets the other exchange students regularly, converses with them, comes to their parties and so forth, it should not surprise her that the others come to hold her to special obligations. Note, however, that the friendly attitudes and dispositions that Berenice's actions seem to manifest do the work rather than her actions as such. To see why, imagine an alternative scenario in which the other exchange students would feel like Berenice and in which they would jointly decide always to make explicit that their actions do not manifest the attitudes they seemingly manifest (''Sure I come to your party, but mind you, that doesn't mean I care about you.''). Were they to do so, they would become cooperation partners rather than friends. What matters are thus not so much the actions as the attitudes and dispositions that actions manifest.
The mutual acceptance of friendship-constituting rules can explain our intuitions about these examples. Friendship-constituting rules specify conditions that, if met, make it the case that two people stand in a specific relationship defined by certain friendship-based obligations. They thus have the following form: Person A and person B stand in a friendship relation of the type R, defined by friendship-based obligations O 1-n if, and only if, person A and person B have attitudes and dispositions D 1-n toward each other and this is common knowledge among them For the sake of brevity, I will use the placeholder ''non-compliance with friendshipconstituting rules'' to describe situations in which the right-hand side of the biconditional is false. Thereby I do not mean to imply, however, that one complies, or fails to comply, with friendship-constituting rules as one does with friendshipbased obligations. We shall see shortly that non-compliance with friendshipconstituting rules is more akin to non-compliance with the constitutive rules of a game.
This formal structure of friendship-constituting rules can be fleshed out in various ways. For D 1-n we can fill in any of the attitudes or dispositions set out in Sect. 2 along the dimensions of mutual care, intimacy and shared activity. Examples are the disposition to be happy if things go well for the other person, the disposition to confide one's secrets in the other person and the disposition to take pleasure in the other person's company. Equally along the three dimensions of friendship, we can insert for O 1-n any of the friendship-based obligations expounded in Sect. 3. Examples are the obligation to visit a friend in the hospital, the obligation not to reveal a friend's secrets and the obligation not to turn down an invitation to a birthday party for reasons of convenience. Finally, for R we can substitute any type of relationship along the spectrum of friendship, such as soul mates, close friends, friends, acquaintance friends or colleague friends.
Friendship-constituting rules have some further characteristics. First, although they are mostly implicit and hence hard to spell out, we have great proficiency in applying the rules in practice and know very well what interacting with someone in a certain way entails. Second, friendship-constituting rules are socially accepted; that is, since we value friendship and enjoy having friends, we endorse the rules that make friendship possible and do not experience these rules as being imposed on us. Third, friendship-constituting rules are fine-grained; that is, there are multiple rules and each of the rules specifies specific conditions for a specific type of friendship (i.e., close friendship, acquaintance friendship and so on), where each type of friendship entails specific obligations and permissions. Fourth, friendship-constituting rules are culturally dependent. There may be cultures in which one specific form of friendship is not available; the special obligations entailed by a specific form of friendship may vary from culture to culture; and conditions that need to be satisfied for roughly the same form of friendship may vary across cultures as well. Finally, the specific content of the obligations to which friendship-constituting rules give rise is not already included in the social package but is largely determined by what is custom between individual friends. Thus, calling each other once a week can be an obligation and a minimal expression of mutual concern in one friendship but not in another. 7 Friendship-constituting rules can explain how we come to be bound by friendship-based obligations. Assume that two persons implicitly accept a set of friendship-constituting rules. Assume, furthermore, that in the course of their interactions, these persons acquire those attitudes and dispositions toward each other which feature in a commonly accepted friendship-constituting rule and that this is common knowledge among them. If so, it follows that they will take themselves to stand in the type of friendship specified by that rule (e.g., acquaintance friendship) and that they will take themselves to be bound by the obligations entailed by it.
This reveals a first important contrast between friendship-constituting rules and friendship-based obligations. Whereas friendship-based obligations label actions as obligatory, permitted or prohibited, friendship-constituting rules pick out attitudes and dispositions that give rise to new relationships defined by new friendship-based obligations.
Friendship-constituting rules also account for the ways in which friendships can end. Just as one does not begin a friendship by communicating the intention to hereby become friends, one usually does not end a friendship in that way either. To understand how friendships end, assume that two close friends both accept a friendship-constituting rule and that their acceptance of that rule partly explains the development of their close friendship. If at least one of the parties for whatever reason loses the relevant attitudes and dispositions, it follows that they are no longer close friends and hence no longer bound by the friendship-based obligations entailed by that particular type of friendship.
This makes clear a second contrast between friendship-based obligations and friendship-constituting rules. Whereas non-compliance with a friendship-based obligation constitutes a 'wrong' or a 'violation,' non-compliance with a friendshipconstituting rule makes it so that the parties do not stand in the type of relationship to which the rule gives rise. That is, non-compliance with a friendship-constituting rule renders a friendship of a particular type 'null' and 'void.' This need not mean, however, that the parties will no longer be friends at all. The reason is that the attitudes and dispositions of the one party may still be compatible with another type of friendship along the spectrum such that, say, a close friendship turns into an acquaintance friendship.
Owens disagrees with my claim that friendships ends because at least one of the parties lacks the relevant attitudes and dispositions. If a friendship ends, he claims, this is ''because one does what one knows will loosen the bonds of friendship'' (2012a, 103). Here, too, I think that attitudes matter more than actions. In keeping with the dimensions of mutual care, intimacy and shared activity, we can distinguish between three ways in which friendships can end without one of the parties being at fault: friends can stop caring about each other, they can 'drift apart' and they can 'lose touch.' Clearly, when two friends stop caring about each other or drift apart, their friendship ends because the friends lack the attitudes of mutual care and intimacy constitutive of friendship. The reasons for which friendships end are thus clearly of an attitudinal nature in these cases. However, at first face it may seem as if friends can 'lose touch' simply by failing to perform certain actions. Yet on closer examination this does not seem right: we can and do regard people whom we have not seen or talked to for a long time as 'good old friends,' which suggests that friends do not 'lose touch' simply because they fail to engage in shared activity. What matters are not actions or omissions as such, but the attitudes that actions or omissions express.

Scanlon on blame
In the second part of the paper, I will consider the implications of this account of friendship for our understanding of friendship blame. I start by delineating Scanlon's account of the nature of blame. This is the proposal: To claim that a person is blameworthy for an action is to claim that the action shows something about the agent's attitudes toward others that impairs the relations that others can have with him or her. To blame a person is to judge him or her to be blameworthy and to take your relationship with him or her to be modified in a way that this judgment of impaired relations holds to be appropriate (Scanlon 2008, 128-129).
To be able to understand this, consider Scanlon's example of the disloyal friend Joe. Suppose some acquaintances of mine talked about me at a party last week and someone started to make fun of me. Now I learn that my friend Joe, instead of coming to my defense or at least ignoring those killingly funny stories, only made it worse by revealing some embarrassing facts about me which I told him in confidence.
First turn to Scanlon's definition of blameworthiness. In the example, it is stipulated that Joe's behavior is impermissible, and although Scanlon is silent on the issue, most plausibly it is impermissible by reference to a friendship-based obligation of mutual trust. The fact that Joe acted impermissibly does not yet make him blameworthy. This will depend on what Scanlon calls the ''meaning'' of his action. This meaning is determined by the agent's reasons, intentions and attitudes as well as the nature of the relationship between the agent and others concerned (52-56). What does Joe's blabbing tell me about his intentions and attitudes, and what do these intentions and attitudes tell me about our relationship? Interestingly, this question takes the following form for Scanlon: ''Should I still consider Joe to be my friend?'' (129). We are thus asked to consider whether the attitudes and intentions that the action expresses make it the case that the friendship eased to exist. Scanlon puts one of the possible answers to the question as follows: ''At the extreme, I might conclude that Joe was not really a friend after all'' (136).
One need not draw this extreme conclusion. Indeed, the fact that Scanlon uses the locution ''at the extreme'' suggests that other, milder judgments could be more appropriate, depending on the nature of Joe's attitudes and his relationship to me. It also suggests that the various judgments can be located along a spectrum, where milder judgments differ from the extreme judgment in degree rather than in kind. The most natural interpretation of this is to say that if the extreme conclusion that Joe is not a friend of mine is unjustified, I might still conclude that Joe's behavior at the party shows him to be a less good friend than I thought he was. In that case, I judge that Joe's behavior shows him to fall short of the ideal of the friendship and adjust my relation to him along the spectrum. Thus, I might judge that Joe is not my soul mate but still a close friend, or that he is not a close friend but still an acquaintance friend, and so on. In this way, the case of Joe provides us with a blueprint of the judgments of blameworthiness that become appropriate when friends violate friendship-based obligations in less serious ways. Now turn to Scanlon's definition of blame. Once I have settled the question as to whether I should still consider Joe my friend, I have reason to modify my relationship with him in the way that my judgment of blameworthiness renders appropriate. If I am justified in concluding that Joe's behavior shows him to have intentions and attitudes toward me which make it the case that we are not friends at all, the appropriate response for me would be to write Joe off as a friend. To write Joe off as a friend means, among other things, to stop relying on or confiding in Joe, to stop valuing his company and to stop having special concern for his feelings and well-being (129-130, 136). For Scanlon, this adjustment of one's attitudes toward another is precisely what blame consists in: ''To revise my intentions and expectations with regard to Joe in this way, or in some less extreme way, is to blame him'' (136). Scanlon emphasizes that the so-called reactive attitudes are not essential to blame. If I blame someone, he notes, ''I might just feel sad'' (136).
Just like judgments of blameworthiness, blaming responses can be located along a spectrum where milder responses differ from the extreme response in degree rather than in kind. As we have seen in the previous section, Scanlon holds that ''a relationship is constituted by certain attitudes and dispositions'' (131-132). If so, revising one's attitudes toward a friend entails modifying the relationship that one has with him or her. In view of this, milder responses that involve revising some but not all friendly attitudes are most plausibly construed as ways of downgrading the relationship along the spectrum of friendship. Thus, a milder response to a violation of a friendship-based obligation can involve adopting an attitude toward a friend similar to one's attitude toward one's not so close friends or acquaintance friends, or anywhere in between.
Importantly, the story need not end with a modified relationship. After all, I may choose to forgive Joe, where ''forgiveness involves the restoration of an impaired relationship'' (160). Given that blame typically precedes forgiveness, however, this only confirms that blame involves a modification of the relationship in the way described.
This account of the nature of blame has important implication for the conditions of blame. In particular, it enables Scanlon to keep the bar for moral responsibility low in the face of skeptical challenges. 8 It does so in two ways. First, Scanlonian blame does not require control or voluntariness. Scanlon expresses this by saying that on his construal, blame does not entail the requirement of ''adequate opportunity to avoid'' but only that of ''psychological accuracy '' (179-198). 9 To see why it entails only the weaker requirement, suppose that due to factors beyond his control (say, a severely deprived childhood) Joe just cannot help blabbing at parties. Even so, on Scanlon's account it would still be appropriate for me to blame Joe for betraying my trust. Joe may not have had adequate opportunity to avoid acting as he did, but it is nonetheless psychologically accurate to say that his blabbing shows him to be disloyal and untrustworthy, and this gives me ample reason to stop relying on or confiding in him. ''If a person sees no reason to give any weight to the interest of others,'' Scanlon claims, ''the fact that he is like that because he was terribly abused as a child does not make it inappropriate to refuse to consider him as a candidate for trust, cooperation, or friendly relations'' (197-198, see also 178).
Second, Scanlonian blame does not require normative competence. Someone could argue that a severely deprived childhood undermines blameworthiness by rendering the person incapable of appreciating the moral reasons that count against acting in morally impermissible ways. Scanlon grants that if blame were a way of expressing a moral demand, it would not make sense to blame agents who are unable to understand such demands (188n.). 10 On his own account, however, blame is not undermined ''by the fact that [a person] is incapable of understanding the reasons against acting the way he does'' (178). Indeed, if your friend turns out to be, say, a psychopath, you have good reason to modify your relationship with her even if she lacks the capacity to appreciate moral reasons. Stronger even, you have good reason not to trust her precisely because she lacks that capacity.
Interestingly, Scanlon summarizes the two points as follows: ''Like refusals of friendship, blame is justified simply by what a person is like'' (188). This reveals 8 One worry is that while interpretation should be aimed at getting things right, Scanlon's interpretative claims may be motivated by a desire to counter skeptical challenges. In light of this, it is worth noting that a skeptic of blame like Pereboom is willing to embrace Scanlon's account (Pereboom 2013, 2014. 9 The implicit reference is to Watson (2004, 276), who holds that ''reasonable opportunity to avoid'' is a condition for responsibility as accountability. 10 I expound the Strawsonian view that blaming an agent involves holding her to a moral demand in the following section. that Scanlon is able to keep the bar for moral responsibility low by construing blame as a way of modifying a relationship. Consequently, Scanlon's arguments for less stringent conditions for moral responsibility succeed only if his account of friendship blame is plausible in its own right.

Blame and the norms of friendship
To be able to develop a plausible account of friendship blame, I will examine how Scanlon's and Wallace's accounts of blame relate to the picture of friendship laid out in the first part of this paper. Scanlon's account of friendship blame, I shall argue, picks out responses that are licensed when a friend's attitudes are not in keeping with a friendship-constituting rule. Return briefly to the example of Joe. Joe violated a friendship-based obligation of mutual trust. This makes his action impermissible, but it does not yet make Joe blameworthy. Blameworthiness depends on the meaning of the action, and that is why I must now determine what Joe's action tells me about his attitudes toward me. For my part, I would interpret Joe's behavior at the party as a sign that he lacks the attitudes that I expect a friend to have toward me: his behavior shows that he does not care about me.
If we assume the existence of a friendship-constituting rule saying that two people are friends on the condition that they have the appropriate attitude of mutual care, one conclusion I could draw is that Joe's failure to care about me makes it so that Joe and I are not really friends at all: Joe's attitudes render our friendship void. If this judgment is correct, it would be appropriate for me to write Joe off as a friend and, by implication, to stop holding him to a friendship-based obligation of mutual trust. Given that friendship-constituting rules are fine-grained, however, I need not conclude that Joe's behavior shows him to have attitudes that make it so that we are not friends at all. But a similar logic holds for milder judgments and responses in which I interpret Joe's attitudes as falling short of the ideal of friendship but not as incompatible with friendship as such. Responding in a milder fashion, I might for instance conclude that Joe's attitudes make it the case that he is not a soul mate but still a close friend, not a close friend but still a friend, not a real friend but merely an acquaintance friend, or anywhere in between, and modify my relationship with him accordingly. 11 11 One could object to my interpretation here by claiming that Scanlon need not accept my account of the nature and norms of friendship. Although Scanlon's remarks on friendship are sparse (see Scanlon 1998, 88-90;2008, 131-138), I think it can be shown that Scanlon accepts the key claims of my account of friendship. First, Scanlon endorses the claim that obligations apply to actions under voluntary control rather than emotional states or dispositions. ''The question of permissibility,'' he notes, ''applies only to something that can be the object of an agent's decision, something that, if one were to decide to do it, would be done intentionally'' (Scanlon 2008, 9). Later, he observes that being a friend involves not merely being disposed ''to fulfill what might be called the obligations of friendship'' but also ''being disposed to certain feelings,'' immediately adding the crucial caveat that ''a friend is not obligated to have such hopes and feelings'' (132). Second, Scanlon endorses the claim that friendship-constituting rules primarily apply to attitudes. Reflecting on the nature of personal relationships, he observes that ''a relationship is constituted by certain attitudes and dispositions'' (131) and proceeds to explain that, besides various behavioral dispositions, friendship involves ''being disposed to certain feelings: to take Now turn to Wallace's Strawsonian account of blame. No doubt the most widely discussed feature of Strawsonian accounts is their interpretation of blame in terms of the reactive attitudes of resentment, guilt and moral indignation. The feature of Wallace's account which I want to highlight here, however, is the claim that blaming a person essentially involves holding that person to a moral demand. 12 Strawson introduced this claim in his landmark essay Freedom and Resentment: ''These attitudes of disapprobation and indignation are precisely the correlates of the moral demand in the case where the demand is felt to be disregarded,'' he claims and adds, ''the making of the demand is the proneness to such attitudes' ' (2008, 23). The claim that blaming a person involves holding her to a moral demand is accepted by authors as diverse as Watson (1987), Wallace (1994), Fischer and Ravizza (1998), Darwall (2006), McKenna (2011) andShoemaker (2015). Yet the claim must obviously be fleshed out: what exactly does it mean to hold a person to a moral demand? I will give substance to this notion based on Wallace's work.
Although Wallace does not develop a systematic account of friendship blame, we can reconstruct such an account based on his more general interpretation of blame. The reconstruction is facilitated by the fact that Wallace situates his theory within a contractualist framework in which moral obligations are construed as mutually accepted norms. For Wallace, moral responsibility and blame are conceptually tied to the notion of holding someone to a moral obligation. ''To hold someone morally responsible,'' he notes, ''is to hold the person to moral expectations that one accepts,'' where ''the set of moral expectations that one both accepts and holds people to is basically the class of […] moral obligations'' (Wallace 1994, 63). Wallace defines the notion of holding a person to a moral obligation in terms of the reactive attitudes: holding a person to a moral obligation essentially involves ''a susceptibility to the reactive emotions if the person breaches moral obligations we accept, or the belief that it would be appropriate for us to feel those emotions'' (66).
The first disjunct of this proposition contains a definition of blame; the second contains a definition of blameworthiness. ''To blame a person,'' Wallace claims, ''is to be subject to one of these reactive emotions [i.e. resentment, guilt or moral indignation], because of what the person has done'' (75). What sets the reactive emotions apart from other moral emotions, he then points out, is that they ''are explained exclusively by beliefs about the violation of moral obligations (construed as strict prohibitions or requirements)'' (38). It would seem, however, that one need not feel resentment toward a person in order to hold her morally responsible. Sometimes, we may for whatever reason simply be unable to work up resentment toward a person who violated a moral obligation. Wallace uses an example of an especially charming colleague who cheated and lied to illustrate the point. Though Footnote 11 continued pleasure in the friend's company, to hope that things go well for the friend and to take pleasure in their going well when they do'' (132). ''A friend is not obligated to have such hopes and feelings,'' Scanlon explains, ''but a person who fails to have them, if a friend at all, is a deficient one'' (132). 12 In response to Scanlon's work, Wallace (2011) and Wolf (2011) thoroughly discuss the issue of whether the reactive emotions are essential to blame. For that reason, I will not explore this topic further here.
Blaming friends 1557 one may not be able to bring up resentment as a reactive attitudes in this case, it would seem that one nevertheless holds the person blameworthy inasmuch as one believes that being subject to this reactive emotion would be appropriate (76)(77). In such cases, one holds a person to be blameworthy without blaming her. An account of friendship blame can be derived from this in a straightforward way. If blaming a person qua moral agent involves holding her to a moral obligation, then blaming a person qua friend involves holding her to a friendshipbased obligation. When applied to the context of friendship, Wallace's account thus picks out responses that presuppose a violation of a friendship-based obligation. This is in stark contrast with the Scanlonian account of friendship blame. If blaming a friend qua friend involves holding her to a friendship-based obligation, friendship blame involves neither a modification of the relationship nor a judgment that the relationship is impaired. Since the friendship-based obligation to which one holds the friend is entailed by one's relationship with the friend, holding a friend to a friendship-based obligation essentially involves reaffirming (rather than modifying) this relationship. By blaming a friend, on Wallace's account, one expresses one's commitment to the friendship and one's willingness to continue it on the same, or at least comparable terms.
The notion of friendship-constituting rules helps to make the contrast more explicit. Whereas on Scanlon's view friendship blame involves judging a friend to have attitudes that are incompatible with the friendship and revising one's attitudes toward the friend accordingly, on Wallace's view it involves holding her to a friendship-based obligation entailed by the friendship. That is, on Scanlon's view blaming a friend amounts to denying the right-hand side of the relevant friendshipconstituting rule, whereas on Wallace's view it amounts to affirming the left-hand side of that rule.

Interpreting blame
In view of these results, we can say that our repertoire of responding to friends who fall short of the standards of friendship fundamentally divides into responses of two different kinds, one picked out by Scanlon's account of friendship blame, the other by Wallace's account. Both accounts thus capture vital aspects of our responsive repertoire. The question remains which account picks out the responses that amount to blame. I will argue that Scanlon's account of friendship fails to do so.
In the example of Joe, the responses picked out by the accounts may seem appropriate at different stages: I might first hold Joe to a friendship-based obligation and then realize that I have reason to modify my relationship with him. To be able to assess which type of response amounts to blame, we must thus look for cases in which the two types of responses clearly come apart. Such cases are possible because the two types of response are logically independent. To see why, note that we respond to a violation of a friendship-based obligation by modifying the friendship only if we interpret the violation as a sign of a relationship-impairing attitude on the friend's part. But one need not do that. As we have seen in Sect. 2, the dispositions that are constitutive of friendship can endure occasional lapses and only tend to destabilize as these lapses increase. 13 A friend can thus violate a friendship-based obligation without necessarily showing herself to be a bad friend and, conversely, comply with a friendship-based obligation without necessarily showing herself to be a good friend. Consequently, the type of response picked out by the account of friendship blame derived from Wallace's work does not entail the type of response picked out by Scanlon: blaming a friend as friend on this account essentially involves believing that she violated a friendship-based obligation, but not necessarily believing that she holds attitudes that impair the friendship.
Scanlon fails to give adequate conditions for friendship blame, or so I shall argue. First, to be adequately described as blaming a friend it is not necessary that one believes that the friendship is impaired and modifies the relationship accordingly. Correct as it may be to believe that a friendship is impaired when a good friend betrays your trust in the way Joe did, friends can violate friendship-based obligations in many other ways. For example, sometimes we believe that a friend's impermissible action is 'out of character'; that is, we believe that the action does not stem from a stable attitude or general disposition on her part. Suppose that a friend, whom you know as otherwise caring and punctual, for no good reason and without excuse fails to show up at a party to which you invited her. Assume, too, that in the years you have come to hold each other to the friendship-based obligation not to skip out on each other's parties for reasons of convenience. I assume that people will blame the friend for skipping out on the party and that they will take this blame to be appropriate. Accordingly, we blame a friend for violating a friendship-based obligation even if we do not believe that her action stems from an attitude that is incompatible with the friendship.
On Scanlon's account, to believe that a friend is blameworthy for violating a friendship-based obligation is to believe that the action expresses attitudes incompatible with the friendship. Scanlon is thus committed to the counterintuitive conclusion that the friend is not blameworthy for skipping out on the party.
Second, to believe a friendship to be impaired and to modify the relationship accordingly is not sufficient for blame either. It may happen that a person believes one of her close friends is 'holding her back' and therefore decides to stop treating her as a soulmate. Such a person does not blame her friend; it is just that she wants to explore new things in life and believes that keeping her friend at a distance is a necessary means to that end. This suggests that a modification of the relationship is not sufficient for blame. Scanlon could object to this counterexample by saying that on his construal, blame does not consist merely in a modification of the relationship, but in a modification as rendered appropriate by the judgment of blameworthinessand in cases in which a person believes that a friend is holding her back there is no such judgment. That is a valid point. Nevertheless, we can easily come up with a case in which a person judges that her friend holds attitudes that impair their relationship and modifies the relationship in a way that reflects this judgment, but does not do so for the reason that the other party holds these attitudes. A person may, for instance, think that she would be lenient with her friend if it were just the attitudes; but since the person goes through hard times and knows that by holding her friend to friendship-based obligations she will make herself vulnerable to being wronged again, she stops treating the other as a friend by way of self-protection. Even if this response involves a modification of the relationship that reflects a judgment about the attitudes held by the other party, it would be a stretch to say that it amounts to blame.
In contrast to Scanlon's account, the account of friendship blame derived from Wallace's work can accommodate our intuitions about these two examples. Since friendship blame involves holding a friend to a friendship-based obligation on the account derived from Wallace's work, proponents of this view can reject the counterintuitive conclusions that follow from Scanlon's account. Firstly, they are not committed to denying that the friend who fails to show up at a party is blameworthy. Secondly, they are not committed to affirming that a person who modifies her relationship with a friend by way of self-protection is blaming her friend. Regarding the first example, the account of friendship blame derived from Wallace's work yields that the friend is blameworthy for skipping out on the party because she violated a friendship-based obligation while lacking an adequate excuse. Regarding the second example, it yields that the self-protective person is not blaming her friend because she does not hold the friend to friendship-based obligation.
Although the account of friendship blame derived from Wallace's work can accomodate our intuitions about such cases, I think it is not the full story. To be able to fully capture our responsive repertoire in the context of friendship, the account of friendship blame derived from Wallace's work must be supplemented by an account of the responses that become appropriate when friends show attitudes that are not in compliance with friendship-constituting rules. Of course, Strawsonians like Wallace need not deny that responses other than blame might be appropriate when friends violate friendship-based obligations. It is just that these responses are not the object of their investigation. Because of their focus on blame, however, Strawsonians such as Wallace might not have an eye for other types of responses.
This can be problematic in two ways. On the one hand, a friend's voluntary violation of a friendship-based obligation may signal a stable attitude that is incompatible with the friendship. If we fail to recognize a plurality of possible responses, we might persist in blaming the person, where the proper response would be to write her off as a friend, or at least to level down our relationship with her along the spectrum of friendship. On the other hand, a friend can have attitudes that are incompatible with the friendship even if she does not violate a friendship-based obligation (or even if she did violate such an obligation but this violation cannot be traced back to any action under her voluntary control). If we fail to consider a plurality of possible responses, we could be inclined to think that in such cases no critical response toward a friend is licensed. Inappropriate as blame may be in such cases, it may nonetheless be appropriate to modify our relationship to the friend in a way that reflects her attitudes. Only the recognition of both types of responses will enable us to maintain healthy personal relationships. 14

Conclusion
The aim of this paper was to shed light on the relations between the nature and norms of friendship on the one hand and our understanding of blame on the other. I have given a phenomenologically rich picture of the nature and norms of friendship and demonstrated that Scanlon's and Wallace's account of blame both capture a vital aspect of the way we respond to non-compliance with the norms of friendship. Scanlon's account blame, I have argued, fails to pick out the responses that are properly described as instances of blame. If my argument is correct, friendship blame may be subject to stronger requirements than Scanlon suggests. In particular, it may be subject to the requirements of control and normative competence.