University of Birmingham Learning from Friends and Terminating Friendships: Retrieving Friendship as a Moral Educational Concept

. In contrast to ancient times, friendship is rarely discussed nowadays as a resource in moral education. Even within Aristotle-inspired character education, where it could naturally claim pride of place, its coverage is miniscule compared, say, to that of the emulation of moral exemplars. The aim of the present article is to retrieve friendship as a moral educational concept: to explain how moral educational goals define and sustain deep friendships, and how the thorny issue of when friendships should be terminated is best understood in terms of considerations as to whether they have exhausted their educational potential. By arguing that education is the raison d’être of deep friendship, Kristján Kristjánsson shows how friendship is developmentally constituted and, in its most complete form as “character friendship,” educationally executed. There is no such thing as friendship per se , but rather friendship at a certain developmental niveau (or level), with its specific developmental assets and liabilities: qualitatively differentiated according to its educational affordances. While operating within a broad Aristotelian framework, Kristjánsson devotes two sections to charting the moral educational liabilities that may dissipate even the most complete friendships, a topic mostly overlooked by Aristotle himself.

The ancient Greeks considered philosophy to begin with wonder. Wonder is one of the motivations behind the present article, but the other is disillusionment with the current academic literatures on friendship and how those relate (or, in most cases, fail to relate) to education in general and moral education in particular.
It is almost a platitude that educational institutions provide the context within which some of the most enduring friendships are initiated and forged, often for a lifetime: among students or between students and teachers. Those contexts and the friendships they engender form a recurring theme in world literature, both "serious" (such as Brideshead Revisited) and "light" (such as the Harry Potter series). A veritable mountain of academic literature exists on the correlations -as causal links are more difficult to establish -between school-or university-formed friendships and various positive psychological variables, most specifically between friendship and (student) well-being. The potential psychological benefits are typically understood subjectively as happiness but in some cases objectively as flourishing. 1 Moreover, in some studies, friendship is evaluated in light of its effect on understanding of the notion of intrinsic (as opposed to extrinsic or even just instrumental) value. The standard ploy in friendship research in psychology is to offer some generic characterization of friendship, such as that friendship is "a reciprocal relationship where there is mutual liking and enjoyment spent in each other's company." 5 People's own specifications of their deep or best friendships, then, tend to be taken at face value and the question becomes not what doeslet alone should ideally -ground them morally, but rather what "provisions" such friendships between self-described kindred spirits offer (for example, companionship, help, intimacy, reliable assistance, self-validation, emotional security) and how those provisions are correlated with significant psycho-social variables, most notably either subjective well-being or a positive self-concept (understood as socially constructed "identity" rather than morally grounded character 6 ). The standard methodological approach to friendship research in psychology is therefore characterized by a unidimensional, instrumentalist, and amoral understanding of friendship, where friendship will vary quantitatively with respect to a number of variables but is not taken to assume normatively different types.
Contrast this tradition with the classical Greek conception of friendship as "a categorical repository for the hope of a mutually edifying moral covenant voluntarily negotiated between people" 7 and the difference could hardly be starker. The Greeks in general, and Aristotle in particular, were well-nigh obsessed with the moral and educational value of friendship. Two of the ten books of the Nicomachean Ethics and nearly a third of the Eudemian Ethics are devoted to friendship: its recognizably ethical nature, its educational ramifications, and the proper or improper reasons for its termination. Friendship, in its ideal instantiation, is seen there, along with the other virtues, as nothing less than the "perfection of man's [sic] potential as a rational being." 8 This fact illuminates an even more surprising aspect of the friendship lacuna in current accounts of moral education: namely, its absence in most of the recent character-education literature. Typically understood as the practical application of virtue ethics, which in turn is largely driven by Aristotelian or quasi-Aristotelian considerations, 9 contemporary character educators are in tireless pursuit of ways in which to apply Aristotelian conceptualizations to, and to craft Aristotle-based interventions for, school practices, including, to name just a few examples, by using habituation (via service learning), stories and music, or role modeling. discourse already exists on the last-mentioned strategy, also known (in Aristotelian language) as the emulation of moral exemplars. 10 Notably, however, Aristotle himself devotes much more space to friendship as a source of character education than to the emulation of glorified exemplars. Why does no one seem to pick up that fact and run with it? The only sustained discussion of this issue that I have come across is in an article by Diana Hoyos-Valdés, 11 who asks in exasperation why Aristotelian character educators prioritize role-modeling so much over friendships. However, tellingly, that article was published neither in an education journal nor in a pure philosophy one. Mainstream philosophers have written insightful papers on Aristotelian friendship, even foregrounding its educational element, 12 but those papers tend to be written with exegetical or moral theoretical aims in mind rather than as contributions to an essentially educational discourse.
To cut a long story short, the aim of the present article is to retrieve friendship as a moral educational concept: to explain how moral educational goals ground and sustain deep friendships, and how the thorny issue of when friendships should be terminated is best understood in terms of considerations regarding when friendships have exhausted their educational potential. By arguing that education is the raison d'être of deep friendship, I want to show how friendship is developmentally constituted and, in its most complete form at least, educationally oriented. While Aristotle idealized the most complete type of friendship overly and overlooked various potential reasons for friendship terminations, 13 I still consider a broadly Aristotelian framework best suited to the task of carrying out the current aim.
There are three main reasons for using Aristotle's account of friendship to anchor this analysis. Historically, Aristotle's discussion has served as an unsurpassable source of inspiration. Even those friendship theorists who ultimately reject Aristotle's moralized view of friendship -in favor of an aestheticized view, for instance 14 -still tend to use Aristotle as the springboard of their argumentation. Logically, I am in broad if uneasy agreement with Aristotle's taxonomy of friendships and consider it to further the aim of teasing out the educational nature of friendship. Finally, expositorily, the stringent conditions that Aristotle places upon the most complete type of friendship help develop my argument, for if I can show that even friendships so parsimoniously construed are liable to perversions that may reasonably lead to their dissolution, it will be more plausible to suppose that the more permissive types may also, and more easily, suffer the same fate.
The aim invoked above calls potentially for two distinct articles. One would set out the positive developmental conditions for becoming friends (by practicing friendship), how this process unfolds educationally, and how it can be stimulated in moral educational contexts through various classroom strategies. This article would also need to distinguish between the respective pros and cons of Aristotelian role-modeling and Aristotelian friendship. The second article, on the other hand, would need to explore the negative side of the story: how viewing friendship as an educational concept helps make sense of some of the shortcomings of friendship -specifically, cases in which it cannot hit its golden mean as a virtue and is liable to rupture. I have decided to begin here with the second proposed article, because focusing on liabilities first and getting those out of the way is often a good way of identifying assets. My analysis proceeds as follows: In the next section, I set out various moral and educational considerations relating to friendships, especially of the Aristotelian kind, and then, in the two sections that follow that discussion, I explore moral educational liabilities of associations between friends of equal and unequal social standing, respectively. In closing, the final section offers some concluding thoughts.
Readers should be forewarned that this article is very much exploratory. As a means of rehabilitating friendship qua moral educational concept, the best way to proceed is to conduct an exercise in talking about friendship in educational terms and see where that leads us, rather than taking for granted from the outset that friendship is unquestionably an educational resource, ripe for curricular application. This article serves the purpose of such an initial exercise.

Background: From Aristotle to Contemporary Social Science
In order to set the context of the argumentation, I apply a fairly wide lens in this section and present a brisk overview of Aristotelian theory. I also need, however, to stick my head above the Aristotelian parapet and consider some recent social scientific findings. Given Aristotle's own naturalistic methodology, according to which all moral theorizing is answerable to facts about what makes human beings flourish or wilt, deferential Aristotelianism -overlooking fresh empirical findings -counts ex hypothesi as misbegotten Aristotelianism.
Friendship for Aristotle is conscious "reciprocated goodwill." 15 It assumes three main types, where the first two (friendships for pleasure and utility) are "incomplete" -because of their essentially extrinsically valuable and transitory natures -but the most developed type (character friendships) is "complete" because of its unique intrinsic and enduring nature (NE, 211ff [1156a6ff]). Notably, pleasure and utility friendships have clear uses and are necessary for smooth human association, and they will continue even among those who have formed character friendships. While not entirely consistent on this point, Aristotle makes it a necessary conceptual condition of all the three (genuine) types of friendship that to be a friend to someone "you must wish [the person] good for his own sake" (NE, 210 [1155b31-32]). Nevertheless, once the raison d'être of the two incomplete types (of being mutually advantageous or mutually pleasurable) is lost, they dissolve fairly easily (in contrast with character friendships) despite the intrinsic goodwill that sustains them while they last.
That said, it would be easy to reconstruct Aristotelian theory about at least some instances of utility friendships as having significant moral educational value. However, Aristotle himself did not have the broad contemporary concept of "moral" at his disposal, so when he speaks of the educational value of friendship, it is always in the service of the "ethical": referring (in ancient Greek) to the cultivation of virtue-grounded character. As only a privileged group of people, "brought up in good habits" (NE, 6 [1095b4-5]) and already (being) conditioned into virtue from an early age, stand a reasonable chance of developing virtue-based character, Aristotle's references to what we would nowadays categorize as the moral educational value of friendship are limited to this narrowly defined group. I have explained above why it serves the purpose of this article to be parsimonious in exploring the thesis about moral education as the raison d'être of friendship, so I will resist the urge to "correct" Aristotle here.
Apart from the features they share with the incomplete types, character friendships have some features that are unique to them. For example, they involve (a) spending time together in shared activities, (b) sharing joys and sorrows, (c) loving the friend for her own sake in the form of loving her character (as her set of virtues), (d) "soulmateship" in the strong sense of being related to friends as they are to themselves, "since the friend is another himself [sic]" (NE, 246 [1166a30-33]; see also 260 [1170b6-7] and 265 [1172a32-34]), and (e) viewing the friendship as intrinsically valuable to the extent of seeing the friend as irreplaceable, not only painfully replaceable as in some developed forms of utility friendships.
When Aristotle elaborates upon those features as the defining criteria of character friendships, some of his discussion smacks of idealizations -and similar idealizations mar his account of when character friendships are or should be terminated, as we see presently. Take, for instance, the stringent conditions that Aristotle places upon character friends' willingness to expose themselves (including their foibles and weaknesses) to one another and their ability to do this through accurate self-knowledge; or the romanticized view of the unproblematic concord between friends as having "one spirit." 16 John Cooper explains the penchant for idealization as part of a teleological bias in Aristotle's thinking that induces him 16 to define things with respect to their most fully realized instances. 17 So despite his thinly veiled digs at the cardboard idealist Plato and the latter's vantage point of secure distance from real life, Aristotle cannot always avoid falling into the same trap. Lorraine Pangle, on the other hand, sees Aristotle's idealizations as a pedagogical ploy that aims to avoid breeding any whiff of cynicism among his readers as young budding phronimoi. 18 Whatever the reason, Aristotle ends up underplaying the fiery impulsiveness and precarious happenstance that characterize all friendships, even of the most complete type. Incidentally, many popular (literary and other) accounts of what people nowadays would simply refer to as "deep" or "best" friendship share this aura of romantic invincibility, so for those readers without a full command of the Aristotelian concept of "character friendship," it will suffice in what follows to think of the contemporary concepts of "deep friendship" or "soulmateship" instead.
One thing to note before proceeding further is that not only does Aristotle fail to bow to the prohibition on value judgments imposed by many current social scientists, the Humean bifurcation of facts and values is entirely foreign to him. He does not understand evaluative language as evaluating an independent world of description, but rather as describing a world of evaluation, which is as "real" as -and indeed inseparably mingled with -the factual world. This explains why he chooses to take the verdicts of the many and the wise as starting points about what we would call "value judgments," in just the same way as he does in the case of other "factual judgments." Those initial verdicts may well need to be rejected if good reasons appear to support their nonvalidity. However, prevailing views can never be fully consigned into inconsequence. We must engage with them, and the onus is on us to show that they are wrong. So to establish, for example, as Aristotle wants to do, the educational nature of character friendship, there is no need to elicit a logical connection, or even an empirical one, between the concepts in question, based on large samples; it suffices to enlist the views of the many and the wise and then fail to come up with plausible evidence to disconfirm them. Analogously, the question of when friendships are terminated (by wise people) and when they should be terminated is basically the same question for Aristotle.
So what is character friendship really for according to Aristotle? Although that question allows for a number of possible answers -and the sharing-joys-and-sorrows condition shows him, for example, not to be averse to contemporary concerns with the effects of friendship on mental well-beingthe core answer seems to be that moral education constitutes its raison d'être. Character friends become "better from their activities and their mutual correction" as "each moulds the other," and through this mutual molding they become "more capable of understanding and acting" (NE, 266 [1172a11-14] and 208 [1155a15-16]). Friendship of this kind educates by being knowledge-enhancing, virtue-enhancing, and life-enhancing in general terms, because the friends act as each other's procreators on the trajectory toward full phronesis. 19 The dynamics of how exactly this happens, and how it can be made to happen more effectively through educational interventions, is a topic for another paper, as already noted. In the remainder of this article, I will focus on the potential perils that the educational nature of character friendship poses. To do so, however, I need to say something about what we could call, to paraphrase Lev Vygotsky, the zone of proximal development of character friendships: the zone in which moral (character) education can take place through scaffolding by the friend -in between the two zones where I can, and cannot, cultivate my character on my own.
Simply put, the zone in question comprises people brought up with guidance in good habits who are neither too good nor too bad. A human does not befriend a god, although the god could figure as a role model; similarly, two phronimoi (perfectly virtuous people) do not engage in what we would normally consider to be moral education, although it could be considered "education" of a kind. At this extreme top end of the zone in question, then, the friendship relation involves an evolving bond within which "friends draw each other out and participate in the fine-toothed articulation of each other's character." They affirm each other's evaluative outlooks "unreservedly and unconditionally" and engage in a "running appreciation" of their "jointly produced sensibilities." 20 The idea here is that just as the phronimoi no longer need emulousness as a virtue (because they have nothing left to learn from role models), they no longer need to learn from each other either, but can simply relish each other's virtuousness and affirm it. One could argue that Aristotle is here, once again, trading in undue idealizations and that he should have been more mindful of his own examples of how even the highly virtuous can go wrong. 21 So, on a plausible Aristotelian account (although not Aristotle's own), the phronimoi would still need lifelong moral education in the ordinary sense (see, for example, Talbot Brewer who talks about Aristotelian character friendships as "lifelong sources of … ethical education" 22 ). However, even if we take Aristotle's own description of friendship between fully fledged phronimoi at face value, it does not subvert the thesis that moral education is the raison d'être of character friendship, for the phronimoi would not have got to where they are without having traversed the relevant moral developmental trajectory with the help of character friends. At the present time, they are simply enjoying the fruits of their hard work and exercising their virtues in the company of friends, making sure they do not lie fallow, much in the same way that some people might go to the gym with mates to stay at their ideal fitness level, without the motivation or need to improve. To be sure, this is a slightly unusual understanding of "education," but it is not a counterexample to the main thesis of this article. More relevant for present purposes are, however, three cohorts of peopleall budding rather than fully developed phronimoi -who engage in character friendships that are essentially moral educational in nature according to the prevailing current understanding. Those are people of (a) equal social status who are either (a1) equal in virtue or (a2) unequal in virtue; and (b) people of unequal social status who are unequal in virtue. Notably, Aristotle presupposes that inequality of social status implies inequality in virtue although equality of social status does not guarantee virtue equality. This is obviously a highly contentious assumption (as I have discussed elsewhere 23 ), but because I want to avoid scratching where it does not itch, I grant it here for the sake of argument. Consider, for (a), two undergraduates who become friends as freshmen and help each other grow in virtue, either (a1) because, being equal in virtue at the outset, they share experiences on parallel tracks; or (a2) because the one who is more virtuous at the beginning helps the other catch up while also developing herself. For (b), consider a professor and a student who both grow morally from their mutual interactions but where the more virtuous professor has more to give in terms of moral knowledge, yet more to take back in terms of gratitude and admiration. I return to those friendship dyads in the following two sections, respectively.
Notice that in order to honor the intrinsic value of (character) friendship and of loving the friend for her own sake, the friendship relation must not be entered into instrumentally, with the explicit aim of incrementally enhancing one's character through interacting with, and thereby learning from, the friend. However, again, this poses no serious threat to the thesis of moral education as the raison d'être of character friendship. Consider as an analogue Millian utilitarianism, according to which happiness is the sole intrinsic value. This does not mean that one's life is best led in direct pursuit of happiness where one asks oneself at every turn what will be most pleasing here and now. What Millians tend to recommend, rather, is engaging in worthwhile activities and then enjoying happiness as it supervenes upon those activities. Similarly here, one chooses friends and enjoys their company without asking oneself what they can contribute to one's character development. Yet the ultimate point of the institution of character friendships is mutual self-cultivation of virtue. This also explains why people will hardly ever offer the educational raison d'être as the explicit motivation for initiating or engaging in friendship. Education acts here as a self-effacing "regulative ideal," an implicit guiding background condition of our motivation to pursue character friendship, rather than as a direct source of motivation, just as in the case of the utilitarian who is best advised to forget the raison d'être of his ethical theory (except when engaging in a philosophical debate about it) in order not to instrumentalize it. 24 23. Kristjánsson, "Ten Un-Aristotelian Reasons for the Instability of Aristotelian Character Friendships." 24. Dean Cocking and Justin Oakley, "Indirect Consequentialism, Friendship, and the Problem of Alienation," Ethics 106, no. 1 (1995): 88-91.
The whole tenor of Aristotle's discussion of all types of friendship is that they are developmentally constituted and, in their most complete form as character friendship, educationally executed. There is no such thing as friendship per se, but rather friendship exists at a certain developmental niveau (or level), with its specific developmental assets and liabilities. Friendship is thus qualitatively differentiated according to its educational affordances qua resources or barriers. Aristotle's exploration bespeaks sensitivity to the liabilities and barriers in the case of the lower types of virtues, but he brushes most of those cavalierly aside in the case of character friendships. Yet, greater sensitivity to what may imperil character friendships will add ammunition to the thesis about their essential educational nature: namely, if it can be shown that what typically thwarts these friendships is, precisely, the exhaustion of their educational potential.
Aristotle claims that character friendship is stable and enduring, once established, as long as the friends do not live too far apart for too long (NE, 213 [1156b17-19], 223 [1159b3-4], and 238 [1164a12-13]). The only exception to this rule that is discussed in any detail is if the friend becomes "incurably vicious." That the friend regresses morally, by degrees, is not a valid reason for the termination of the friendship unless the friend's virtuous character has disappeared completely or is beyond repair: "If someone can be set right, we should try harder to rescue his character than his property" (NE, 244 [1165b15-35]). Aristotle does not tell us explicitly why the friend's characterological shift into incurable vice warrants the termination of friendship. Yet, seen from the perspective of deep friendship as an educational concept, two reasons immediately suggest themselves. First, I risk debasing myself rather than growing morally from continued interactions with the incurably bad. Such interactions would represent "an objectionable sort of moral complacency" about my own development toward full phronesis and would constitute a risky slope to slide down. 25 Second, from the point of view of the vicious friend, she also cannot gain anything from the friendship in terms of education because complete vice precludes self-knowledge, which in turn precludes self-love; and as lack of self-love prevents her from loving others in the philia sense (NE, 246-247 [1166b1-25]), the very foundations of any edifying character friendship have been shattered.
I have pointed out elsewhere that Aristotle may be setting the bar too high, however, and that if the two friends stop pursuing similar virtues in similar enough contexts, it is likely that the mutual interest in each other's destiny will gradually fade away long before one of them becomes incurably vicious. 26 From an educational perspective, this point can be bolstered by arguing that continuing to pursue soulmateship with a person who is on a slow but steady trajectory 26. Kristjánsson, "Ten Un-Aristotelian Reasons for the Instability of Aristotelian Character Friendships." toward moral decline may do irreparable damage to my own virtuous makeup. The respect and commitment to the friend notwithstanding, we seem at least entitled to modify Aristotle's thesis to say that one has good reasons for stamping out friendship with a friend who is on such an unstoppable trajectory, even long before the friend has wound up in a state of complete wickedness.
Given the sensitivity of Aristotle's naturalistic moral theory to empirical evidence, it is in order, at this juncture, to say something about why deep friendships do in fact end. Although current psychological research is not confined to character friendships and does not preselect only those friends whom Aristotle would count as budding phronimoi, something would be amiss with Aristotle's restrictive thesis about friendship dissolutions if it did not have a solid basis in reality.
In a nutshell, empirical research elicits lots of facts that do not seem to tally with Aristotelian theory. A common cause for the dissolution of best friendships is that the friends simply grow apart as their interactions are no longer psychologically or educationally rewarding. Moreover, best friendships do not either continue as such or come to an end; there are other possibilities as well: they can be "downgraded," temporarily or permanently, and then upgraded again. Some people seem to prefer settled, long-lasting deep friendships while others are more comfortable with "serial friendships" that change as their life changes. Only about 50 percent of best friendships among adolescents remain stable over one school year. About 84 percent of teenage girls report at least one complete dissolution or a downgrade of a best friendship. In general, the trajectory of deep friendships is much more fluid, more relative to personality styles and psycho-social preferences, and more subject to the wear of time and subtle changes in external circumstances than Aristotle's discussion suggests. 27 Admittedly, some of the empirical findings can be explained away as applying only to a younger and more eclectic group of people than Aristotle had in mind as his presumed readership of budding phronimoi. Others can be related to social circumstances of greater geographical mobility in the modern world than in ancient Greece (although some of those changes should be offset by people's current technological means to keep friendships going across geographical distances). All that said, one cannot shake the impression that, given Aristotle's own universalism about human nature, many of these empirical findings would have been the same if Aristotle had conducted rigorous social scientific research in ancient Athens. One must conclude that he may perhaps have been blinded by the lights of his fiery mission to promote (character) friendship as a moral virtue of enormous endurance, vitality, and strength. Unfortunately, understanding its raison d'être as educational comes with certain downsides that Aristotle conveniently ignores.

How Socially Equal Friendships May Lose Their Educational Value and Become Dissolved
Although Aristotle claims to be using his standard method of working through difficulties and contradictions in received wisdoms in his account of friendship, 28 his focus on the frailties of friendships and their terminations is almost entirely confined to the "incomplete" types (of pleasure and utility), implying that once having become elevated to the level of character friendship, the relationship is immune to most internal and external threats. I cannot, therefore, endorse Marina McCoy's contention that "Aristotle exhibits a tremendous sensitivity to human moral weakness in his account of friendship." 29 However, that is not all bad news. Having explored his educational rationale of character friendships, we can now use Aristotle's defining characteristics to hoist him with his own petard in terms of the potential causes of character-friendship termination that he overlooks.
In a previous paper, I suggested ten different reasons, beyond those identified by Aristotle himself, for the dissolution of character friendships. 30 I called those reasons "un-Aristotelian" in the sense of not being invoked by Aristotle, although many of them seem to be potentially compatible with his general theory of friendship. By drawing on the thesis proposed in the two previous sections about education as the raison d'être of deep (character) friendships, it is now possible to probe many of those reasons in more detail. In the present section, I explore some educational reasons for the dissolution of socially equal character friendships (beyond the reason of incurable vice), and in the following section, I scrutinize the demise of socially unequal character friendships in a similar way.
Aristotle does admit, albeit briefly, that in addition to drastic departures from the path of virtue, circumstances may intervene through which friends become separated -including increased disparities in wealth (NE, 221 [1159a33-34]). However, he remains reticent about the specific problems that can beset character friendships between people of an equal social standing that exhibit inequalities in virtue. I will focus here on divergent developmental paths among people who begin as social and virtue equals but who become increasingly distanced from each other as one of the pair turns in a direction that may ultimately induce the other friend to pull up the friendship drawbridge. I am not thinking here of Aristotle's example of perversion into vice, but of something less radical. Let me invoke as a case in point a perspicuous novel by the best-selling author Meg Wolitzer, The Interestings. 31 This work has all the ingredients of a blockbuster novel, but it also contains unusually penetrating observations regarding the complexities of relationships between friends, as their fortunes tilt precipitously during the course of their life journeys. Depicting the trials and tribulations of a group of six Americans who become friends during a teenage summer camp in 1974, and most of whom harbor (varyingly realistic) artistic aspirations, the book offers glimpses of lives that -to keep readers absorbedare more tempestuous than the lives ordinary people typically lead, but that are not far-fetched enough for readers to cease identifying with them.
With various narrative detours and a troupe of protagonists too numerous to explore here, the three most interesting characters, for present purposes, are Ethan and Jules, as well as Jules's best friend, Ash, whom she also met at the summer camp and who ends up marrying Ethan. The dynamics of the Ethan-Ash-Jules triad, where Jules tries to preserve her friendship with Ethan while avoiding making her best friend Ash jealous, is juicy enough to merit a discussion of potential eros-philia conflicts (which Aristotle, for one, surprisingly ignores). The more relevant dimension here of the (ultimately) fraught and tenuous friendship between Jules and Ethan is how Ethan's life takes a developmental turn that strains the friendship with Jules almost beyond the breaking point. To be sure, in the novel the friendship does endure -for otherwise the author would have had to bring the narrative to a premature end. However, it is easy to envisage an alternative trajectory where what happens to Ethan would have sufficed to dissolve the friendship completely by exhausting the friendship's educational potential.
Notice first that both Jules and Ethan are kind and caring people, "brought up in good habits," in Aristotle's sense. While neither are full phronimoi (Ethan is, for example, badly lacking in the virtue of courage), both would fall securely into the category of budding phronimoi. The very rationale of their friendship, which nourishes and sustains it, is their unremitting care for each other and the eagerness of each to help, advise, and strengthen the other for the other's sake, along with the desire of both to learn from the one another's experiences. In this sense, they are typical character friends, and there is not a whiff of charlatanism about their soulmateship. However, while Jules has to give up her artistic dreams for a modestly rewarding career as a therapist, Ethan becomes shockingly successful as an artist, "gilded with specialness and privilege" (TI, 347). Because of his strong moral self-identity, Ethan channels his riches into philanthropy and other charitable causes, turning himself into what Aristotle called a megalopsychos: one who, because of largesse-enabling external fortunes, can afford to be magnificent and grandiose in giving and in assisting others, not only mundanely generous.
Although being a megalopsychos is, by Aristotle's lights, a mixed blessing (because of the strain being a busy public benefactor places on one's intellectual resources 32 ), the moral developmental path of the megalopsychos is ethically more advanced than that of a (budding) phronimos and involves the possibility of blessedness in life rather than simple flourishing. However, here is the moral of the story: Jules becomes increasingly disengaged from Ethan because she can no longer identify with him and learn from him -nor give him advice relevant to his new life circumstances. The issue is not so much Ethan's raised status in life or his wealth per se -which by itself might just have subtly changed the character friendship with Jules into one of social inequality (rather than equality) while still retaining its characteristic ethical quality. The issue, rather, is that Ethan gradually grows into a "thicker, finalized adult self" (TI, 4) that does not allow for mutual correction and reinvention with Jules's self because their capacity for moral action has taken such different turns that the educational basis of their friendship becomes eroded. Whenever Ash and Ethan "cooked a chicken, it would feed a subcontinent" (TI, 50); whereas Jules's attempts at good deeds remain limited to offering her close friends and her own ailing husband (a chronic depressive) some modest comfort. She feels she wants to be "as good as" Ethan is (TI, 122), but because of lack of personal and financial resources, she falls desperately short in this regard and that discrepancy gradually creates a wall between the two of them.
The friendship between Ethan and Jules changes "into something so different from what it had originally been as to be unrecognizable" (TI, 135). The imbalance between them is "suddenly, jarringly evident" (TI, 236). The almost "telepathic mutuality" between them is lost (TI, 171), not because Ethan becomes irredeemably bad but rather because his goodness achieves heights that Jules cannot share with him as a soulmate. If they met now, "they would never become friends" (TI, 241). The lesson is that lives can divide, and even the best of friendships can become threatened, for a reason that is the diametrical opposite of the one that Aristotle foregrounded. The possibility for mutual education and self-cultivation in character friendships is parasitic on there being a shared area of experiences and activities that can inform the sensibilities of both parties and help them to correct each other. If one person becomes "too good," it can create a barrier that is just as insurmountable as the one that results from the person becoming "too bad" because, in either circumstance, the shared zone of proximal development that serves as the relevant educational crucible disappears.
Another instructive example has been elicited by Alexander Nehamas from the play Art by Yasmina Reza. 33 Notably, Nehamas uses this example to illustrate his theory about the grounding of deep friendships in the aesthetic qualities of friends rather than, as in Aristotle's theory, in their moral character. However, the example can easily be reinterpreted in the service of a moralized-cum-educational 32. See Kristján Kristjánsson, Flourishing as the Aim of Education: A Neo-Aristotelian View (London: Routledge, 2020), chap. 4. On Friendship, account of what actually grounds deep friendships. In the play, three friends (Marc, Serge, and Yvan), who had previously shared a mutual appreciation of art, fall out when Serge pays a fortune to buy a painting that the other two find aesthetically worthless. Marc accuses Serge of pretentiousness, maintaining that he has bought the painting simply because it is in vogue; Serge, in turn, takes Marc to have become a paternalistic besserwisser. Nehamas sees this example as indicative of the precariousness of best friendships insofar as they are based on essentially fleeting phenomena, such as one's taste in art. Lack of aesthetic harmony can thus easily undermine friendships, as friendship is more about a common characteristic "style" than a common moral "character."

Nehamas,
As an Aristotelian, I read the lesson of the story quite differently. If I were Marc, what would worry me is not the sudden incompatibility between my aesthetic tastes and those of Serge. I would rather see Serge's gesture as indicative of a more profound change of character profile: from authenticity to pretentiousness. While such a transformation falls short of a debilitating descent into vice, I still would worry that I had nothing more to learn from Serge and that we would no longer be able to hone each other's moral sensibilities. In other words, the educative rationale of the friendship would be lost -and I do not mean educative as in "aesthetic education," but rather in the sense of "moral education." While I say this as a neo-Aristotelian, I repeat my discontent that Aristotle himself did not consider nonextreme character shifts of this kind as offering valid reasons for the discontinuation of deep friendships, and that he instead fastened only on more radical and rare examples. That said, Marc may have been too quick in seeing the change in Serge as a reason for terminating their friendship. When friends change their values, we have the option of also changing our values accordingly rather than rigidly sticking to our guns (as Amélie Rorty explores 34 ). Still, even in such cases, the decision is educational as much as psychological. We need to ask ourselves about the additional educational resources such a change would create in our trajectory toward self-cultivation, and also weigh the educational barriers that it might potentially erect.
Because dissolutions of friendships on grounds of subtly changing developmental paths are probably much more common than Aristotle envisaged, and such dissolutions are often profoundly painful, one may have some sympathy with the UK head teacher who discouraged pupils from forming "best friendships" so as to avoid the pain and upset caused by fallings out. 35 While, from an Aristotelian perspective, this would count as an overreaction (throwing the baby out with the bathwater), extensive research findings that highlight the long-term negative consequences of (deep) friendship-endings for pupils' sociomoral development and school success provide grist for the head teacher's mill. 36 Jessica Isserow is quite right that on any plausible account of the grounding of deep friendships, enhancing their credentials "is not something to be swiftly achieved by cutting ties with those friends who are deemed morally subpar, and replacing them with the good Samaritans and humanitarians of the world." 37 In that sense, Aristotle was on target about the unique endurance of character friendships. However, once the friendship has become perverted from its educational purpose, this may legitimately lubricate the slide toward the friendship's dissolution (or at least its downgrading to a mere pleasure or utility friendship) long before the friend has turned incurably bad.

How Socially Unequal Friendships May Lose Their Educational Value and Become Dissolved
The thesis about education as the raison d'être of deep friendships has, I submit, even more initial plausibility in the case of unequal friendships, such as between a teacher or professor and a student. World literature teems with examples of young learners soaking up wisdom and reflective virtues from mentors while the mentors rejuvenate their own character with inputs from the youthful exuberance and open-mindedness of their mentees. However, Aristotle makes things slightly difficult for himself with three limiting conditions he places upon unequal character friendships. Two of those I have briefly touched upon before: first, that such friendships are, by necessity, doubly unequal, with inequality in social standing implying inequality in virtue; second, that the mentee is not allowed to initiate the friendship with the explicit aim of learning from the mentor, to avoid the risk of instrumentalization. Instead, the mentee has to be motivated to love the mentor for the mentor's own sake (and vice versa). The third condition is slightly odd. It takes the form of an equalizing principle, according to which unequal friendships only work if the stronger party is "loved more than he loves; for when the loving reflects the comparative worth of friends, equality is achieved in a way" (NE, 221 [1158b26-29]). So while the weaker party gets "more profit" (in terms of character growth), the stronger party gets more honor and devotion (NE, 236-237 [1163b1-15]). I call this principle "odd" because Aristotle later makes the empirical claim that "benefactors seem to love their beneficiaries more than the beneficiaries love them [in return]," just like the craftsman "likes his own product more than it would like him if it acquired a soul" (NE, 250-251 [1167b16-35]). Thus, the equalizing principle itself seems to fall afoul of minimal psychological realism, which is not good news for the strict ethical naturalist, Aristotle.
These three limiting conditions notwithstanding, Aristotle seems to be surprisingly oblivious to the various obstacles that may stunt and decimate unequal friendships as "schools of virtue." To avoid unnecessary controversy, let me focus 36. Ibid.,187. 37. Isserow,"On Having Bad Persons as Friends," 3107. in the remainder of this section on unequal friendships of the kind that are most explicitly educational, in order not to beg the question whether all are. I am thinking here of friendships between teachers/professors qua mentors and their mentees. Two penetrating studies provide examples of those. Amy Shuffelton recounts the friendship that she, as a teacher, developed with an artistically inclined, nerdy, and lonely elementary school pupil in Krakow some time agoa friendship that has persisted over many years, with the former pupil being a young man in his late twenties when the article was written. 38 Paul Weithman elaborates on "academic friendships" between professors and the undergraduate students in their classes. 39 Weithman refrains from defining those as full-blown character friendships in the Aristotelian sense, as they normally disband at the end of the academic year; instead, he relies rather obliquely on references in Aristotle to a higher form of utility friendships that "would seem to depend on character" more than one sees in ordinary "mercenary-type" utility friendships (NE, 233-234 [1162b23-27]). However, Weithman fails to account for cases where the student continues to rely on the mentorship of the professor long after graduation and where their friendship even lasts as long as both live, as is not infrequently the case in my own experience. It would be churlish to deny such associations the status of unequal but complete character friendships in the Aristotelian sense. Another way to put this point is to say that the "terminating conditions" of friendships between professors and (graduated) students are not as clear-cut as those, say, between a doctor and (cured) patients. 40 Shuffelton spends considerable time discussing the objection that friendships of this kind breach the teacher's duty of impartiality toward all her students. She correctly argues that a teacher can give all her students excellent service while giving some of them (who, for some reason, need personal friendship rather than just instruction) more of it. In some cases of particularly vulnerable students, the truly caring teacher will rush in unhesitatingly where others are unwilling to tread. In my previous article on character-friendship problems mostly overlooked by Aristotle, I identified specific barriers that unequal mentor-student friendships may encounter. 41 For one thing, because such friendships tend to be motivated by concerns that involve enjoyment by the mentor in spending time with a young person, as well as considerable academic benefits accruing to the mentee from the association with the more learned person, there is serious danger of the friendship never reaching the higher level of character friendship or, worse still, of degenerating gradually into mere pleasure or utility friendship even if it does actualize the higher ideal at the outset. For another thing, because of the well-known phenomenon of role inertia -a phenomenon that is routinely scrutinized in social psychology -it is highly likely that even after the mentee has caught up with the mentor in terms of character development, their association continues to be stuck in the terms in which it was originally forged. In such cases, a potentially equal character friendship becomes misconceived by one or both parties as unequal, thus undercutting its capacity to (continue to) be educationally productive. As Aristotle himself acknowledges, one of the biggest dangers to a friendship is if the two friends do not conceive of the friendship as belonging to the same type: of complete versus incomplete (NE, 243 [1165b5-11]).
As already noted, Aristotle is reticent about possible eros-philia conflicts that often lurk around the corner in these cases, as in other kinds of character friendships. 42 It is not so much that he excludes cross-gender friendships (for example, he explicitly states that a husband and his wife can enjoy "friendship for virtue, if they are decent. For each has a proper virtue, and this will be a source of enjoyment for them" [NE, 232 (1162a25-27)]). It is rather that he does not share modernity's concerns with the faint undercurrent of excitement often floating close to the surface in friendships between the genders and how this undercurrent may gradually plunge any budding character friendship into a maelstrom of misunderstandings, jealousies, and conflicts. While modern scholars have studied this problem in detail, 43 none of them has focused specifically on the detrimental effects of eros-philia conflicts on the educational "regulative ideal" of character friendships. I will end this section, therefore, with a brief discussion of a problem that is more explicitly educational and to which, I submit, unequal character friendships of the kind I have been considering are particularly susceptible. Let us call it the problem of paternalism. Since the Enlightenment, the ideal of education as the cultivation of the critically reflective, autonomous individual has permeated almost all respectable educational theories in the Western world, in one guise or another. Any educational interventions smacking of paternalism are anathema to the ideal of autonomy. Obviously, the ideal of autonomy was not familiar to Aristotle in its modern form. However, he famously demanded that virtuous persons not only perform the right actions, but also perform them for the right reasons and from the right motives: knowing them, taking intrinsic pleasure in them, and deciding on them for themselves (NE, 40 (1105a30-34)]. The underlying assumption is that ethical acts do not have moral value unless they are guided by the agent's own phronesis; and although phronesis-guidance cannot be equated 42. Shuffelton, "Philia and Pedagogy 'Side by Side'," 220. with autonomy in the modern sense, it is incompatible with the idea of a moral agent being paternalistically controlled by an external puppet master.
Despite the high-brow ideals to which both Aristotelian educators andmore radically still -their modern counterparts are meant to aspire according to the phronesis and autonomy mantras, every seasoned educator knows that no educational activities can get off the ground without some measure of paternalism involved. In a sense, all students beginning the study of a new topic subject themselves to initiation into certain qualities of mind, hitherto unknown to them, by putting themselves under the teacher's authority and guidance. As Weithman unapologetically puts it, the professor should try to make her partnership with students formative; it is her job to form their intellectual tastes. 44 However strongly the ideal of autonomy figures as the eventual goal, and however much the learning process is conceptualized as subjection to normativity itself rather than subjection to the teacher's authority as a guide to that normativity, there is no escaping the fact that the pedagogical relationship involves, by its very nature, elements of paternalism -even if that paternalism is self-chosen and fully complied with by the student. This fact forces upon the teacher, whether she wants it or not, certain motivational and dispositional structures 45 that seem to be quite alien to a relationship between character friends, even of the unequal kind. For example, it is incumbent on the teacher to impart certain knowledge, a particular mindset, and specific skills, and the measure of whether or not that has been successful seems to have very little to do with the mutual collaborative self-cultivation of character that is meant to be the raison d'être of Aristotelian character friendships.
What I have been saying here is not that the role of a teacher/professor is essentially incompatible with the role of an unequal character friend as Aristotle understands that role. There are, I believe, countless examples of mentor-mentee friendships where those two roles have been brought into alignment. I am simply calling attention to a warning signal that failed to register on Aristotle's radar. Although character friendships with a socially superior party are often entered into precisely to avoid the paternalism that typically characterizes unequal associations, there is great danger that friendship with a teacher or professor is liable to become, to be perverted into, the very vice that it was created to resist. It is not so much that friendship with a student undermines the teacher's paternalistic authority, 46 but the other way round: that the teacher's authority can undermine the friendship.

Concluding Remarks
The aim of this article has been to retrieve (deep) friendship as an educational concept. While the most obvious place to pan for gold is Aristotle's account of character friendship, whose value is essentially moral-educational, I have focused here more on the negative side of the story: how viewing friendship as an educational concept helps make sense of why many deep friendships are so vulnerable to dissipation and rupture. In a nutshell, putting an educational construction on friendship comes with as many liabilities as assets, and those may in some cases not only enfeeble the specific friendship in question but fully cancel its advantage. To be sure, one of the glories of Aristotle's moral and educational theory is its focus on the value of friendship; at the same time, however, one cannot but wish that his assessment of its pros and cons had been more measured and evenhanded.
I will end this article with two observations. The first is that the somewhat rose-tinted view of character friendships as one-sidedly rewarding "schools of virtue" that one finds in Aristotle may be but one more instantiation of his general overestimation of the characterological stability of people brought up in good habits. The idea of the (increasingly) unified moral agent on an (unproblematically) upwardly mobile trajectory toward full phronesis, with the help of trusted friends on the same journey, belies the extent to which human life is rarely a quiescent stretch, characterized by psycho-moral integration, but rather is one that bends with the winds of time and fortune -and where dilemmas lurk around every corner, including dilemmas about how to best interact with one's friends. Because we are constantly changing as persons, it is no wonder that, as we enter new unforeseen situations, the educational benefits that we derive from associations with different friends fluctuate and may, in some cases, make us drift apart from people who, at previous junctures, were our "other selves." The very point of education is to enact change, and it would be remarkable if the change that my friend enacts in me, or vice versa, always kept us close together rather than prying us apart. In many cases, drifting away from one another will be a slow process, just as the effects of education rarely hit us in an epiphanic instant. In the words of Samuel Johnson, the "most fatal disease of friendship is gradual decay, our dislike hourly increased by causes too slender for complaint, and too numerous for removal." 47 The second observation is that education is not, by its nature, an essentially comfortable process. Socrates did not only liken himself to a midwife but also to a gadfly and a torpedo fish. Our desire for self-verification is often stronger than our desire for self-cultivation, and even when moral educational guidance is provided by a character friend nonpaternalistically, it may touch a raw nerve. In the words of novelist Patrick White, friendship is "two knives." "They will sharpen each other when rubbed together, but often one of them will slip and slice off a thumb." 48 There are times when we are simply not ready to have our thumb sliced off and where guidance from our friends is seen -rightly or wrongly -as being surplus to requirements. "Friendship is something that won't be boxed. It wriggles." 49 Aristotle did his best to "box" friendship conceptually into different types and he advanced the discourse considerably by doing so. However, when attempting to box character friendship as the complete type and to put an educational spin on it, he tried to control and contain a phenomenon that, like all education, has an inherent tendency to wriggle and get out of hand. It is not so much that Aristotle was wrong about the educational value of deep friendships; indeed, I think his insights were fundamentally right. It is rather that he underestimated the extent to which even potentially educationally beneficial friendships are fraught with risk and remain, for various reasons, liable to conflicts and disruptions.