1 References, unless otherwise indicated, are to the Nicomachean Ethics (NE); the translations are generally those of Terence Irwin (Indianapolis, IN: Hackett 1985). Throughout this paper, ‘friendship,’ ‘love’ and their cognates will be used to translate ‘philia’ and its relatives. A review of the difficulties in translating the term can be found in John Cooper, ‘Aristotle on Friendship,’ in A. Rorty, ed., Essays on Aristotle's Ethics (Berkeley, CA: University of California Press 1980) 301-40.

2 1155b31, 1156b7-10, 1159a10, 1164a34, 1166a3, 1168a34, 1168b3; cf. also 1156a13.

3 Cooper, ‘Aristotle on Friendship,’ argues that pleasure- and utility-friendships have a substantial non-instrumental component. This is the point of distinguishing between essentially instrumental friendships (that have non-instrumental features, as in Cooper's example of the businessman who becomes willing to do small favors for a regular customer) and friendships that are essentially non-instrumental. Although Cooper's treatment tends to assimilate the different kinds of friendships to one another (on this point, see also note 17), it is clear that utility- and pleasure-friendships fall on the essentially instrumental side of the divide. In section V we will see a more principled way of spelling out the distinction between these and friendship for the sake of the friend.

So this is not to say that (true) friends don't have their uses. Aristotle is aware that friends have instrumental value (most notably at 1099a31-b2). The friend is a benefactor in the hour of need, and an opportunity for the exercise of virtue when fortune is generous (1155a5ff, 1169b11-16). At one point, Aristotle describes having friends as ‘the greatest external good’ (1169b10), a characterization that may suggest to some instrumental rather than intrinsic value. It is thus worth recalling that Aristotle thinks that external goods (such as health) can have both intrinsic and instrumental value, and that the latter need not preclude the former. See note 6.

4 Now it might be objected that while this is a natural way for us to understand friendship, it is not obviously so for Aristotle. Aristotle takes friends to love one another for virtue, and since one virtuous person is, qua virtuous person, like another, friends must be, as it were, a fungible commodity. Such a position conflicts so strongly with our pretheoretical understanding of friendship and with the general tone of Aristotle's discussion of it that its only support can be our inability to find a preferred construal of the claim that one loves a friend for his virtue. As we shall see, an alternative is available. Moreover, this reading will be extremely hard put to solve the puzzle I am about to pose, i.e., to explain why one loves just those virtuous people that one does love out of the many that one might.

5 For the former, see note 2. For the latter, 1166a30-33, 1166bl, 1169b6, 1170b6f, 1171b33; cf. also 1171a20.

6 To be sure, there do appear to be instrumental uses of one's friends: Aristotle states (to translate as literally as possible) that ‘many things are done just as through instruments also as through friends and political power’ (1099blf), and that ‘your friend, since he is another yourself, supplies what your own efforts cannot supply’ (1169b6). But in the same, somewhat stilted way, one can make instrumental use of oneself (as in ‘the movement of limbs that are the instruments’ - 1110a16). It is probably more appropriate to regard this kind of instrumentality as a sort of extended agency: ‘what our friends achieve is, in a way, achieved through our agency, since the origin is in us’ (1112b28). See note 3.

7 Jennifer Whiting presents such a view - although not as an interpretation of Aristotle - in ‘Friends and Future Selves,’ Philosophical Review 45 (1986), 547-80.

8 As does Whiting, ibid.

9 Cf. Hartman, Edwin Substance, Body and Soul (Princeton, NJ: Princeton University Press 1977) 10-56Google Scholar; Whiting, Jennifer ‘Form and Individuation in Aristotle,’ History of Philosophy Quarterly 3(1986), 359-77.Google Scholar

10 I'm not claiming that one desires the good for one's friend for his own sake only because he is another self. One might, for example, feel good will towards Masons, and so also to one's Masonic friends.

11 Before mentioning the tricky question, a caveat: Actually, we desire apparent goods for our apparent selves, and the base, Aristotle recognizes, may, in addition to mistaking apparent goods for goods, mistake their apparent selves for their real selves, as when they try to ‘gratify their appetites and in general their feelings and the non-rational parts of the soul’ rather than reason (1168b20; cf. 1168b30-1169a3). Moreover, the very base may not even desire goods for themselves at all - partly because they find themselves unbearable, and partly because they may lack coherently organized personalities, i.e., full-fledged selves (1166b2-27). So, more precisely, when we desire the good for our own sakes, we do so because we are ourselves.

Now while Aristotle often enough seems to commit himself to the view that we desire the good for our own sakes because we are ourselves, other remarks indicate that we desire the good for ourselves for our own sakes on account of our own virtue. These views need not be taken as contradicting one another if the latter can be taken as sketching the internal structure of the former. Such a construal would also be convenient for the line of exegesis I am advocating in another way, since I am about to argue that it is causing the friend's virtues that makes him my friend. My attempt to be literal about the ‘another self’ locution should then make me want to say that it is causing my own virtues that makes me myself; and the kind of account of the relation between one's virtues and one's selfhood I have in mind would allow me to say something sufficiently like this. Providing a full-fledged account of this kind would take me too far afield, so I will not do it here. However, I’ll defend a more modest version of the claim in section V.

12 Aristotle seems to be contrasting companionly friendship with that of relatives and possibly with the political friendships of the immediately preceding passages, so I take the scope of ‘companionly friendship’ to be what we ordinarily take as friendship, with the proviso that Aristotle would have in mind genuine, i.e., virtuefriendship.

13 After comparing the relation between parents and children to that between gods and men, Aristotle brings to our attention the crucial fact that gods and parents are respectively ‘the causes of [the] being’ of men and children (1162a4-7). And at 116lal7 Aristotle describes the father as ‘the cause of his children's being.’

14 Cooper, 330f.

15 The phrase ‘human mirror’ is Cooper's, adapted from the Magna Moralia (MM). The ‘human mirror’ argument seems to me the more dubious of the two. First, in making it out Cooper draws heavily on the MM (1213a10-26), which should be used with caution when trying to establish the views of the NE. For even if, as some deny, the MM is authentic, it may represent a stage in the development of Aristotle's thought distinct from that of the NE. Cooper appears to regard the MM passage as a more straightforward form of the argument of NE 1169b27-1170a4 and of 1170a15-b7. But it is not clear to me 1) that the two just-mentioned passages both contain the same argument or 2) that either NE passage contains the argument of the MM passage. Indeed, the contrast between the generally admitted obscurity of the NE passages and the relative clarity of the MM passage casts doubt on the claim.

Second, Cooper prefers his view to Ross's alternative construal of 1169b27-1170a4, which he regards as circular. But it is not clear that the argument, properly understood, is circular. Cooper points out that, on Ross's construal, ‘Aristotle simply assumes, altogether without explicit warrant, that the good man will have friends’ (318) - and this is an inappropriate assumption in an argument for having friends. But Cooper conflates the project of demonstrating ‘the value to a person of his having friends’ (318), with the distinct project of providing someone who has no friends with reasons to acquire them. Were Aristotle's aim the latter, rather than that of explaining to the virtuous person why his friends are valuable, the argument would indeed be circular. But notice that the argument, construed in this way, renders friendship essentially instrumental: a friendship formed in order to attain the pleasure promised by the argument would be one of the friendship-analogs, rather than true friendship.

If, instead, part of being virtuous is contributing to the virtue of others, and if - as we shall see - this has as a causal consequence that the virtuous man acquires friends, then the alleged circularity disappears. For then it is simply a fact that the virtuous man does have friends. Raising the question of the value of these friends is in such a case not at all the same as asking for reasons to make friends. So Ross-like interpretations, which do not require excursions to the MM and are thus to be preferred, remain open.

Given the foregoing, one might wonder why I adopt the ‘human mirror’ story at all. The reason is that even if these views were not in the foreground when the NE was composed, Aristotle may still have had ‘human mirrors’ in mind when thinking of the ways in which friends affect each other's virtue. There are undoubtedly many ways in which this happens, and Aristotle need not have enumerated in the NE all those he had discovered. For our purposes it suffices to supply plausibly Aristotelian examples of such causal relations.

16 Recall Physics 202b5, and see note 18 below.

17 When Cooper takes the function of these arguments to be that of ‘defend[ing] the value of friendship only by showing that for human beings it is a necessary means to attaining certain broadly valuable psychological benefits’ (Cooper, 332, italics mine) he is providing an unsatisfactory, because (solely) instrumental, account of friendship. And, not incidentally, an account that cannot explain the non-interchangeability of friends. Cooper's position in ‘Aristotle on Friendship’ seems to me schizophrenic: he denies that he is producing an essentially instrumental account of friendship, and in the first part of the paper seems more or less successful; but the latter half produces what is, denials notwithstanding, just such an account. Since the two parts of the paper were originally separately published papers (see Cooper's note 1), there may be an historical explanation for the final version's split personality. I will confine myself here to explaining why the story that Cooper gives is in fact essentially instrumental.

Mirrors - even human mirrors - are tools used for a specified purpose; this is true despite Cooper's flat denial on p. 333. Those who cooperate with one in a shared activity are, on the account presented, instruments towards one's performance of that activity. Cooper distinguishes between shared activities like chess, in which co-participants need have solely instrumental value, and shared virtuous activities, which require intimate acquaintance in order to ascertain and apprehend the co-participant's virtue; such intimate acquaintance is, Cooper thinks, sufficient for friendship. A similar line is taken regarding ‘human mirrors.’

This is an interesting argument, but its links will not bear the strain that Cooper puts on them. For one thing, intimate acquaintance is by no means sufficient for friendship (as anyone who has ever shared his living quarters will testify). For another, ascertaining the virtue of another need not require intimate acquaintance. Virtues are something like dispositional psychological states. We can imagine that psychologists discover ways of reading these states off of a person's nervous system. They then build virtuometers, which quickly and reliably evaluate a person's virtue index. (If we want to make the story less anachronistic, we can imagine that the gods, who see into the souls of men, sponsor Virtue Evaluation Oracles in downtown Athens.) Under such circumstances it is very easy for virtuous persons to find guaranteed-to-be-virtuous co-participants in virtuous activities. But it would be unreasonable to describe such people - people like the stranger with whom I have agreed, on the basis of his 98.7 Composite Virtue Index, to cooperate in building a temple - as friends. Such devices could also, with a little more story-telling, be made to take the place of ‘human mirrors’; instead of using one's friend to study one's own virtue, one could take virtuometer readings. Stories like this one show that ‘human mirrors’ and co-participants in virtuous activities are, qua occupiers of these roles, simply instrumental means to an end; they can be replaced by gadgets like virtuometers or by virtuometer-certified strangers because they are already no more than tools, and any tool, qua tool, is replaceable by anything else that serves the same purpose. Cooper's account, then, construes friends as essentially instruments, and as a fungible commodity.

18 A question that might be raised is whether my account can be extended to ‘civic friendship’ - the somewhat weaker form of philia that is found between fellowcitizens. (Even if it cannot, this is not an objection to the account; for Aristotle thinks that ‘we should set apart the friendship of families and that of companions’ [1161bllf].) Now, the citizens of the Greek polis took one of its primary functions to be that of educating its citizens. This education involved not merely the transmission of skills and information but a broader mission, involving the inculcation of virtues, that we can call the shaping of character. This widely shared opinion was also held by Aristotle. (Cf., e.g., 1099b29-32, 1102a7-10, 1103b2-7, 1155a23, 1179b32-1180a15, 1180a35-b7.) If, as I have suggested, one's character, constituted at least partly by one's virtues, is at least part of what makes one who one is, then the polis is (partly) responsible for its citizens being who they are.

Of course the polis is not a person. To say that the polis is responsible for a person's virtues and character is to say that the responsibility is shared by his fellow citizens who, with him, constitute the polis. Each of this person's fellow citizens is (in a small degree) responsible for his virtues, and thus (to a similar degree) for his being who he is. If so, then each citizen is a procreator of his fellow citizens, and this will lead him to exhibit (a suitably weak form of) love towards them. Civic friendship, it turns out, can also be understood as derivative from the relation between a procreator and his creature.

19 And they do not - barring the possibility that the corrupted friend has become another person (see note 22)- cease to have been one another's procreators. This may explain one's residual goodwill towards an (uncorrupted) ex-friend (1165b31-35).

20 It may be possible to substantiate the claim on its second, stronger reading. While I won't attempt to do so here, it seems possible to defend the view that virtues are not merely characterizing but also individuating features, perhaps in the context of a reading that commits Aristotle to individual forms.

21 A similar implication is possibly carried by 1157a15, which describes friends for utility as persons who ‘were never friends of each other, but of what was expedient for them’; one who befriends another for his virtue is, we may suppose, a friend of the other. Cf. also 1164a10.

22 If one were to argue for the stronger version of the claim (see note 20, above), one might want to treat this case as one in which the corrupted friend has become a different person, reading the passage against Aristotle's discussions of still-more-radical transformations that fail to preserve identity at 1159a9ff and 1166a20-24.

23 This may not always be true if these descriptions are given a ‘rigid reading’: what makes someone useful to me may be his virtues. A non-rigid reading is intended.

24 1102a3; 1140a27 is seen as supporting the claim that the eudaemonia in question is one's own.

25 This is, if I correctly understand him, Vlastos's view. (Personal communication and unpublished notes.)

26 This reconciliation of Aristotle's theory of friendship with his eudaemonism may help to resolve an ambiguity I have been leaving open. We said that the procreator has the same being as his creature; and this is likely to prompt the reader familiar with Aristotle to ask whether the being they share is species being or individual bdng, i.e., whether the procreator and his creature are the same in kind or the same in number. Our discussion suggests that the more strongly one reads eudaemonism as a form of psychological egoism, the more one will be motivated to adopt the latter construal. Diogenes Laertius ascribes to Aristotle the saying that friends are ‘one soul inhabiting two bodies’ (V.20); on such a reading, this would literally be Aristotle's view.

27 It has occurred to me that Frost may have had Aristotle's views of friendship in mind when he wrote this poem. Frost would have shared the exegetical orthodoxy of his time in taking Aristotle to hold that things of a kind have their form in common but are individuated by their matter. Virtues are specified by the form: if brooks had forms, a brook's virtues might include being full of clear, flowing water. Frost seems to be saying that he loves Hyla Brook not for its brookish virtues (which it possesses scantily, if at all) but for the matter that makes it differ from other brooks: its dried mud and dead leaves. What we love is not the form, and the virtues specified by it, but the individuating matter (hule). (The pun on hyla and hule has been thought of by others - e.g., Rorty, Richard Philosophy and the Mirror of Nature [Princeton, NJ: Princeton University Press 1979], 40, 45, 64fGoogle Scholar - and it is not extravagant to suppose that it may have occurred to Frost as well.)