In this paper I argue that Aristotle’s discussion of self-love in the Eudemian Ethics (EE) suggests that it was written in response to shortcomings in his treatment of the topic in the Nicomachean Ethics (NE). Specifically, Aristotle modifies his account of self-love between the, criticizing in the EE positions he advocates in the NE.1 This difference is easily explained if the EE came later. But if the NE were later, we would have the mysterious picture of Aristotle advocating a view he had already shown to be flawed, with no explanation for why the change was made and no mention of how to deal with the problems.
In §1 I outline Aristotle’s discussion of self-love NE VIII-IX and its central role in his larger discussion of friendship. In §2 I argue that self-love cannot be reciprocated the way Aristotle requires of the paradigm form of friendship, a problem partitioning the soul will not solve. In §3 I show that Aristotle was aware of this problem with self-love in the EE, and accordingly gave self-love a much smaller role than it received in the NE. Finally, I argue in §4 that the best way to explain the change between EE VII and NE VIII-IX is that the EE is later. While the chronology of Aristotle’s ethical work may be of sufficient interest, my primary goal is to argue, at least on the subject of friendship, that the EE is the more developed, and therefore the more authoritative, work. It is the EE, not the NE, to which we should turn to find Aristotle’s considered judgments.
§1. Self-love in the NE
Aristotle’s discussion of friendship in the NE occupies two books. Books VIII-IX.3 outlines a tripartite classification of friendship, corresponding to three ends of our affections and pursuits: utility, pleasure, and goodness (1155b18, 1156a5). Utility and pleasure friendships are impermanent and directed more toward the object of affection than toward the person providing it (1156a14). Character friendships, on the other hand, are based on a person’s goodness, and are described as perfect or complete (teleia), as the truest or most genuine (malista) form of friendship, and as good and pleasurable absolutely (haplōs agathon hē hēdu).2 As such, character friendship is the primary and proper (prōtōs men kai kuriōs) form, while pleasure and utility friendships only count because of their resemblance to character friendship (1157a31).3 And importantly, character friendship can only exist between two people who similarly virtuous (1156b8).
At IX.4 Aristotle transitions to a conceptual analysis of friendship. He claims that friendship is defined (horizontai) in terms of how a good person feels toward herself. A friend is identified by five characteristics (1166a5-10): (a) friends pursue the real or apparent good for the other’s sake; (b) friends wish each other to live for the other’s sake; (c) Friends live together (sundaigonta); (d) friends make the same choices (t’auta hairoumenon); (e) friends share in their trials and triumphs (ton sunalgounta kai sugchaironta). These characteristics, Aristotle argues, are most clearly found in a good person’s relationship with himself, and therefore friendship is derived (elēluthenai) from self-love. This is not an account that somehow reduces friendships to self-love.4 On the contrary, Aristotle’s aim is to show that self-love is the paradigm case of friendship, which explains and sets the standard for other amicable relationships.5
Immediately after defining friendship in terms of self-love Aristotle states, “Whether or not there really is friendship toward oneself or not, let us dismiss the question for the present” (1166a34). But despite his ambivalence here, Aristotle uses self-love for several key arguments in the NE. Friendship is an extension of self-love, because friends are, in Aristotle’s terminology, “another self” (allos autos). This is not simply a metaphor, but rather a critical, literal piece of Aristotle’s larger project.6 A friend is a friend because one feels the same way about that person as one does about oneself, a feeling made possible because of the similarity between the two objects of one’s affection. Aristotle also cites the extension of self-love to others to explain filial affection: children are part of the selves of parents (1161b21, 28) and brothers are the same self physically separated (1161b33).
Aristotle uses self-love not only for its explanatory power; it also generates normative conclusions. At IX.8 Aristotle questions whether it is right or wrong to love oneself, as the Greek word for self-love, philautia, has pejorative connotations. But in typical fashion, Aristotle distinguishes between two senses of the term. There is a noble way to love oneself: to love one’s nous, the best and dominant part of oneself (1168b29-1169a11). This is something the good person ought to do, since the best part of oneself deserves one’s affections and esteem. This fact also allows Aristotle to argue at IX.9 that the good life will require friends. Aristotle claims that life is intrinsically pleasant and desirable for the good man, and therefore consciousness of and participation in one’s existence is desirable (the full argument occurs at 1170a14-1170b19). But because a friend is another self, a friend’s self will be just as intrinsically desirable as one’s own self. Since a life lacking such intrinsically desirable goods would be incomplete, eudaimonia requires the good man to have friends.7
Without making self-love central to his account of friendship, Aristotle lacks the resources to demonstrate the conclusions to which he argues. Hence Aristotle’s theory of friendship requires self-love. This is unfortunate, because self-love has some serious theoretical problems, to which we now turn.
§2. Problems with self-love in the NE
Fortenbaugh claims that the NE’s treatment of friendship “may be said to show Aristotle at his logical best.”8 This would seem to be a minority view, as NE VIII and IX are frequently derided for their failures to provide cogent arguments that are consistent with the rest of thebook.9 I would like to add to this list. I think there are several problems with self-love in the NE, but in this paper I will limit myself to one: that self-love cannot be reciprocated the way Aristotle claims friendship requires, and therefore cannot count as friendship at all.
Aristotle states several times that friendship is a two-place relationship that does not exist unless it is reciprocated. In other words, it is a necessary condition of X and Y being friends that X reciprocates the good will (eunoia) felt toward him by Y, and likewise for Y’s feelings toward X. This condition in fact serves to distinguish friendship from mere good will (1155b31-36, 1166b30-33. See also 1157b34-58a2). But in self-love, there is only one subject and one set of feelings. It is therefore impossible for self-love to be reciprocated, and therefore impossible for self-love to be a form of friendship.
Aristotle may have suggested a response to this challenge if after dismissing the question of whether friendship with oneself really is possible he wrote, “though it would appear that such friendship is possible, if there are two or more parts” (1166a34).10 We might think that, in a style reminiscent of Plato’s Republic, self-love exists when parts of the soul have the right relation to each other. Aristotle’s liberal demarcation of the soul in NE I and his depiction of noble self-love as love of the best part of oneself (1168b29-1169a11) corroborates this strategy. But it will not solve the problem of reciprocity.
Suppose we divide the soul into nous and a non-rational part and follow Aristotle in NE IX.8 in identifying the self with nous. As Bostock argues, this would obfuscate the difference between how the good person feels about himself and how the bad person feels about himself, since in both cases the person is to be identified with the rational part of the soul.11 Moreover, the reciprocity problem would simply move locations, as a single nous cannot reciprocate its love for itself anymore than a single person can.
So suppose instead that we say that the rational part and the irrational parts of the soul reciprocate love for each other.12 This strategy won’t work either. For one, it is not clear that the non-rational part of the soul is capable of the complex activity that constitutes love, as opposed to mere affection or appetitive desire. Second, the non-rational part of the soul would not seem to be worthy of the nous’s affections. After all, bad self-love is directed towards the wrong parts of the soul (1168a15-25), and we wouldn’t want to make nous guilty of this shortcoming. Third, mutual affection between parts of the soul would not constitute friendship, at least not in its paradigm form. Recall that one way Aristotle distinguishes character friendship is that it is only possible between two people similar in their virtue (1156b8). But nous is superior to the non-rational part of the soul, and therefore could not have an equal friendship with it. Aristotle details several kinds of unequal friendship in NE VIII (child and parent, ruler and ruled, good and regular person, etc), but these fall short of the best kind of friendship, which holds between equals. This is difficult to reconcile with the paradigm status of self-love.
These criticisms are not offered as an attempt to refute Aristotle’s arguments, but rather as worries he would have to address. This is where things get interesting, because Aristotle himself saw the need to address these worries. What may be surprising is that he addresses them in the EE.
§3. Self-love in the EE
I’ve argued that self-love fails to count as a genuine form of friendship because the soul is not structured in such a way to reciprocate it. In the EE Aristotle addresses this challenge head-on. Aristotle admits that the marks of friendship are most clearly found in relation to oneself, just like in the NE.13 But Aristotle’s position on self-love differs dramatically.14
Aristotle writes “Some think each person to be most truly friend to himself, and using this standard they judge his friendship to his other friends” (1240a8). This is exactly the position advocated in the NE. But what follows is quite different. Aristotle continues “According to arguments and the received view (dokounth’ huparchein) of friendship, in some ways they are in opposition, and in other ways they appear the same. For self-love is somehow friendship by analogy, but not without qualification (haplōs)” (1240a12-14). Why does Aristotle think this? He continues, “For being loved and loving is divided into two parts (en dusi gar diēirēmenois)” (1240a15). Such a relationship can only truly apply to the soul insofar as the soul itself has two distinct parts (1240a20), and in comparing self-love with self-restraint or self-justice Aristotle suggests that this is true only in a qualified way. There is a sense in which one can be friend to oneself, in the same way there is a sense in which self-restraint or justice to oneself is possible. But just like in NE V.11/EE IV.11 on justice, this is not a literal sense. Self-love requires two distinct parts, rather than just two senses or two descriptions. This does not lead Aristotle to completely reject the notion of self-love. But he does show much more sensitivity to the problem in the EE than in his NE treatment.15
The demotion of self-love in the EE is not restricted to when it is discussed directly. Recall that in the NE Aristotle uses the notion of friends as other selves to expand self-love to others. In the EE this argument is conspicuously absent. Instead, the notion of other selves is relegated to the last few pages of EE VII.12, where it plays a greatly reduced role. Here Aristotle adapts the proverb “another Hercules” to mean ‘another self’, which friends are supposed to be. But instead of a metaphysical notion of an entity identical to, and therefore as valuable as, one’s own self, Aristotle here uses the much simpler relationship of resemblance. In the same way that ‘another Hercules’ is as strong as Hercules, that is, resembles him in a certain way, a friend resembles oneself in certain ways. But these resemblances are very specific and spread across friends; Aristotle claims that “by nature [friends] are most akin (suggenestaton), with respect to the body similar to one, another with respect to the soul, and with these things [body and soul] according to one part another [similar to] another” (1245a32-34). This passage occurs shortly after Aristotle claims that we should share different activities with different friends according to their differences (1245a19-23). This restricted and far more commonsensical treatment of friends as other selves stands in stark contrast to its robust and paradoxical role in the NE.16
All this shows a set of marked differences between the NE and the EE on friendship. I argued in the last section that the NE discussion had several serious flaws. In the EE, Aristotle not only changes his account to avoid these flaws, but cites these flaws as reasons in support of the EE treatment. This forces us to reevaluate the relationship between the two treatises.
§4. Interpreting Self-love in the Aristotelian Ethics
Aside from Kenny and a small handful of others, the NE is regarded as Aristotle’s last, best ethical treatise. Indeed, there was a period where the EE’s place anywhere in the Aristotelian corpus was doubted. Now, thankfully, this period has past, thanks in large part to Kenny’s work in arguing that the EE is not only genuine Aristotle, but that the common books belong there. Kenny also (tentatively) argued that there was some evidence that the EE was later than the NE. I believe that Aristotle’s treatment of self-love in the two treatises provides more evidence.
Consider first the picture that would result if the NE were in fact the later work: In the EE, Aristotle presents a picture of self-love that is crafted to avoid certain problems with self-love. Sometime later, Aristotle writes NE VIII-IX, where he gives an account that has exactly those problems he criticized earlier. And Aristotle says nothing about these problems. He does not explain why he changed his mind, does not give arguments to refute his earlier position, he does not even acknowledge that these earlier shortcomings exist. If this picture is right, Aristotle knowingly adopted a position he correctly believed to be flawed.
But suppose instead that the EE is Aristotle’s later work. This presents Aristotle in a far more plausible (not to mention charitable) light. In the NE Aristotle gave a theory of friendship which had some problems he did not realize. He later came to change his view to deal with these problems. In doing so, he addresses his original view and explains why it was mistaken. He then presents a new theory which shows he has learned from these mistakes.
We may corroborate the evidence for the EE’s relative lateness with two passages which appear to cite the NE. The first is discussed by Kenny.17 At 1244b29 Aristotle criticizes a passage for being too obscure about knowledge and the desirability of life. Kenny argues that the corresponding passage is located not in some lost work, but at NE 1170a25-b9. The second passage occurs at 1240a, where Aristotle discusses the characteristics of friendship that all apply to the good man’s feelings for himself. Aristotle begins the passage by claiming that these are the characteristics found “in the treatises we usually study” (en tois logois episkopein eiōthamen) (1240a24). Since these marks are laid out in detail in NE VIII-IX, I can think of no better location for the reference.18
Skeptics should note that the idea of NE VIII-IX being written early in Aristotle’s career is nothing new. Annas has argued that the NE treatment of friendship was early, based on the clear influence of Plato’s Lysis and its poor fit with the rest of the book.19 Madigan suggests that NE IX.8 was written earlier than bookX, the latter developing due partly to mistakes in the former.20 Kahn, on the other hand, suggests that NE IX was revised in order to fit with the later book X.21 While Sherman does not directly address the chronology of the Aristotelian ethics, her statement that in the EE “Aristotle adds a new dimension to his discussion of friendship as it appears in the NE and MM” suggests the possibility that the EE is later.22 If it is plausible that NE VIII-IX was written early, and EE VII appears aware of NE VIII-IX (but not vice versa), what could be more reasonable than to conclude that the EE treatment of friendship was a revised draft?
I believe these considerations present a strong case for the NE preceding the EE. But it is important not to overstate the evidence here. If we assume that all the NE books form one treatise and the EE books another, then we could apply chronological conclusions from parts of the texts to the wholes. But it is not clear how justified this assumption is. So instead I can only make the weaker claim that EE VII antedates NE VIII-X. Even if the common books belong with the EE (which I believe Kenny has ably shown to be the case), we are still left with relationship between EE I-III and NE I-IV unexplored. But that is the project for another day.
I want to conclude by stressing my conclusion. Rather than the work of an early, immature Aristotle, parts of the EE seem instead to be the later, and more importantly, more developed work. Aristotle has consciously rejected some of his positions in the NE and given us explanations why. So at least for the books we’ve discussed, the EE gives us Aristotle’s considered views on the matter. I believe that, at least on the topic of self-love, the EE is the better work. More importantly, Aristotle would seem to agree. If this is the case, then the EE should be the work to which we turn for determining what Aristotle’s final word on friendship was.23 The NE has long enjoyed pride of place in Aristotle’s ethical corpus. Kenny has argued that was not true in antiquity, and I join what he called “a small handful of Aristotelian fanatics” in arguing that it should not be true today.24
1 In a longer version of this paper, I argue that this is the case not only for self-love (which has problems with reciprocity and with disinterested affection (i.e., loving another for his sake rather than your own), but also for the role self-love plays in eudaimonia.
2 1156b7and passim, 1157b25, 1157b27
3 Exactly how this relationship is borne out is not obvious. See Fortenbaugh (1975), Jost(1991)
4 The case against a reductive, egoistic notion of friendship in Aristotle is probably best argued in Annas (1977) and (1988).
5 For self-love as the paradigm, see Annas (1977) and (1988). See also Kahn (1981), Kraut (1989), Price (1989), and Rogers (1994). For self-love as the standard, see Pakaluk (1998).
6 Cf. Bostock (2000), esp. p. 176; Hardie (1980), p. 324; Kahn (1981), p. 34; Kenny (1992), pp.46-51; Osburne (2009), p. 1, Stern-Gillet (1995), pp.14-15, and contra Pakaluk (1998), p. 172-73.
7 Cf. Cooper (1977b), p. 293
8 Fortenbaugh (1975), p. 51
9 Notable examples of Aristotle’s logical shortcomings here are (i) the conflict between the psychologically conflicted bad man in NE VIII versus the enkratic, akratic, and vicious agent in NE VII;[See Annas (1977), pp. 553-54; Bostock (2000), pp. 172-74; Pangle (2003), pp. 144-145] (ii) the convoluted argument for the intrinsic desirability of friends with which I ended the last section; [See Cooper (1977b), Kenny (1978), pp. 227-230] (iii) the lack of a viable defense of the proposition that a friend is a second self; [See Annas (1988), pp. 1-2; Hardie (1980), pp. 324-325] (iv) not to mention the now well-worn ambiguity between egoism and altruism See Annas (1988), (1977); Gottlieb (2009), pp. 145-50; Hardie (1980) pp. 326-335; Kahn (1981); Kraut (1989), pp. 78-154, (1988); Madigan (1991); Politis (1993); Price (1989), pp. 103-130; Rogers (1994); Sherman (1987).
10 I say ‘may’ and ‘if’ here because this line is suspected to be an interpolation. See Rackham (2003:1926).
11 Bostock (2000), p. 174
12 Another option would be to divide the soul into two rational parts, but that would introduce more problems than it solves.
13 Though, interestingly, the claim that friends make the same choices is omitted in the EE. I should also like to note that in his discussion of friends feeling the same feelings, Aristotle introduces the qualification “if this were possible, and if not, as nearly the same (eggutata) [as possible ]” (1240a39). This should address what Annas refers to as “the derision from the commentators, who tirelessly point out that one cannot, logically, regard another person’s thoughts, pains, etc, exactly as one does one’s own.” Annas (1977), p. 542. Is this perhaps another place where the EE is more developed, possibly in response to shortcomings in the NE?
14pace Rowe (1971), who wrote that “If there is little difference between the two accounts of the individual virtues, there is even less between those of friendship. There may be slight differences of emphasis; but there is not a single point at which the doctrines of the EE and EN differ to any significant extent” (p. 52).
15 There is one line where the text claims that the absolutely good man seeks to be friend to himself, because “as has been said, in him there exist two things which by nature desire to be friends and which it is impossible to separate” (1240b29-30). I am not convinced this line was written by Aristotle rather than a later commentator, because Aristotle in fact did not just say that there were two factors. Rather, he said that self-love was only analogously a kind of friendship specifically because friendship requires two parts, which Aristotle strongly implies self-love does not capture. But even if this is genuine Aristotle, it does not refute the present thesis. The phrase “du echei en autōi” is in the very least ambiguous between the claim that there are two distinct parts of the soul and the claim that the soul has distinct descriptions. Moreover, Aristotle claims at NE 1102a34 that when parts are distinct they can be separated in reality, while two factors or descriptions cannot be so separated. So if it in fact impossible to separate the two factors of the soul in the good man, this suggests that these factors are not distinct parts, and therefore that the affection a good man feels toward himself is not literally friendship.
16 I am forced to disagree with Stern-Gillet (1995) on this point, where she claims that Aristotle’s use of other selves in the EE is paradoxical. She interprets the argument as follows: “In the same way as some other person can be as Herculean as Hercules, my friend can be as much myself as I am” (p. 12). I believe she made the mistake of reading this passage of the EE so as to correspond to the treatment of other selves in the NE. My reading is not only more plausible, but also makes Aristotle less guilty of a non-sequitur in his choice of illustrative proverbs.
17 Kenny (1978), pp. 226-229. Kenny also discusses Aristotle’s EE rejection of an identification being a human being and her nous in the NE, as discussed above.
18 A third possible reference is EE 1236b17-21, where Aristotle seems to be responding to the claim at NE 1157125-32 that utility- and pleasure- friendships are derivative forms of friendship in virtue of their resemblance with the claim that all friendships count in virtue of pros hen homonymy (Jost (1991), p. 36). See also Fortenbaugh (1975).