Cicero on Epicurean Friendship: A Reappraisal 109

POLITEIA VOL. 1, No 2, SPRING 2019

APPENDIX

Cicero on Epicurean Friendship:

A Reappraisal
Phillip Mitsis
New York University

Given the occasion, I thought I might take the opportunity to revisit my first pub-
lished paper on Epicureanism1 originally begun in a Lucretius seminar at Cornell in
1978 taught by Elizabeth Asmis. The paper's argument now seems to me in retrospect
to exhibit a certain naive confidence in Cicero's reliability as a source, and though I still
stand by some of its overall conclusions about Epicurean friendship, my views about
the nature of the path that one needs to take to get to those conclusions has changed
considerably.2 Readers will best judge whether a forty-year gap in looking at these
arguments, apart from merely complicating matters, has led to any progress.
At De Finibus Bonorum et Malorum 1.65-70, Cicero, through the character of Tor-
quatus, offers our most detailed and seemingly systematic surviving account of Epi-
curean views of friendship. Brad Inwood has forcefully reminded us, however, that
this work has a particular focus and that its overall structure and arguments clearly
reflect it.3 Cicero is examining what the fines (limits, ends, criteria, etc.) of goods and
of evils are in various rival philosophical theories and in the first two books he sets
himself the project of presenting and criticizing the Epicurean claim that pleasure
serves as a final end or goal of our actions. One recurring problem in his account is
that he glosses the notion of finis with terms having different valences and, thus, those
wishing to extract from Cicero's discussion Epicurus' original view of friendship are
faced with the problem that he often describes the relation of friendship to pleasure
in terminology imprecise enough to encompass different theoretical outlooks. By the
same token, his discussion presents several further formidable obstacles and again no
consensus has emerged about how to overcome them.4 One problem is that from the
outset we are confronted with descriptions of friendship in a Latin vocabulary that are
often emotionally charged5 in a way reminiscent of, say, Montaigne speaking about
his friendship with La Boétie. This is aided, no doubt, by the perceived etymological
connections between amicitia and amor that Cicero plays on (cf. De Amicitia 8). We
cannot assume, of course, that this directly captures the particular emotional nuances
of Epicurus' original Greek.
Perhaps the most difficult and problematic feature of Cicero's presentation, how-
ever, is that it is structured around the claim that true friendship requires us to love
Phillip Mitsis 110

friends as much as ourselves (e.g. . . . aeque amicos et nosmet ipsos diligamus . . .

Fin.1.67). Some would argue that this shows that Cicero is viewing questions of Epicu-
rean friendship through the lens of solidifying professional disputes over generations
both within and without Epicurean circles. On the other hand, Alain Gigandet,6 in
a brilliant intervention, has made a strong case for these philosophical controversies
going back to Epicurus himself, though not necessarily in this exact form. It is this
slippage, however, between Cicero's preoccupations and the original wider context of
these philosophical questions, along with the inexactness of Cicero's vocabulary and
potential distortions of his rhetorical methods that continues to bedevil contemporary
scholarship.
What is clear, however, is that Cicero relentlessly uses the requirement of friends
loving one another as much as themselves throughout his criticism of Epicurean views
(2.82-86) and in the service of his overall argument that hedonism is incompatible
with the normative fabric of Roman ethical and political life.7 Even if some Epicu-
reans think they can meet this requirement for friendship, he claims, they are being
inconsistent with their overall commitment to hedonism, while others come to real-
ize this and give up hedonism in order to save their commitments to true friendship.
Again, we have no direct evidence that such a requirement ever served as this kind of
lynchpin in Epicurus' original account. Moreover, it is hard not to be suspicious of
the notion of any Epicureans giving up hedonism. Thus, one might certainly wonder
whether Cicero is constructing and reifying Epicurean "positions" in order to be able
to go after them later in his own response with a very blunt cleaver.
To be sure, Torquatus' presentation of these positions sometimes seems to present
actual points of contact with surviving texts of Epicurus. Unfortunately, attempts
to show precise correspondences are mostly frustrating. For example, he claims that
Epicurean friends will rejoice in their friends' joy as much as their own and be equally
pained by their anguish (1.67-68). At SV 56 we find the claim that the wise man will
suffer (Ἀλγεῖ μὲν ὁ σοφὸς) no more than a friend who is being tortured. However
literally we are meant to take Epicurus' claim here, it is perhaps plausible to infer sym-
metrical attitudes on the part of a sage to his friend's pleasures as well. We might
therefore conclude that Torquatus is correctly reflecting a general position of Epicurus
himself in saying that a wise man will feel in the same way towards his friend as he
does towards himself (Quocirca eodem modo sapiens erit affectus erga amicum quo
in se ipsum . . . Fin. 68 ff). Not all scholars would be willing to make such an infer-
ence, however, without a more secure and directly corresponding text. Moreover, it is
an even greater jump from the affective claim that one will feel a friend's pain no more
than one's own to the general evaluative claim that Epicureans will value or, in Cicero's
terminology, "love" friends as much as themselves. This is fairly typical of the elusive
nature of Torquatus' account. He seems to sometimes draw near to known texts of
Epicurus, but none of the major positions he outlines maps directly on to them. Nor,
unfortunately, does his exposition straightforwardly fill in gaps that might help us put
those texts coherently together with any confidence.
Cicero on Epicurean Friendship: A Reappraisal 111

Torquatus begins his exposition by saying that friendship has been debated by
Epicureans in three ways (1.66). Many have taken this to mean that there was a dis-
pute among Epicureans themselves and that the three positions Torquatus presents
are meant to represent reified factions or historical layers within the Epicurean school.
This again seems open to question, I think, since the three positions do not necessarily
conflict and to a certain extent seem to address different aspects of friendship and dif-
ferent kinds of friends. It might just as well be the case that these three "positions" have
been extracted from a variety of argumentative contexts and from arguments that Epi-
curus and later Epicureans used in the course of their debates with rival schools. Cice-
ro seems to take particular relish, for instance, in the claim that the second Epicurean
position on friendship appears to have conceded a crucial criticism of the Academics
(1.69). Yet, this supposed position overlaps with the two other accounts in key ways
and, unlike them, it offers a causal empirical explanation of the origins of friendship of
the sort one could imagine in an Epicurean text dealing with the psychological, social,
and historical origins of friendship. It is not certain, then, that these three "positions"
are anything more than constructions by Cicero based on a selective presentation of
Epicurean arguments taken out of their immediate context.
We can make a parallel with Epicurus' account of justice. In trying to show how
the virtues are compatible with hedonism, Torquatus emphasizes that, like the rest
of the virtues, justice is connected to a life of pleasure and tranquility. This aretaic
conception of virtue as a virtuous trait of character has correspondences in Epicurus'
theory (Ad Men. 132), but in other sources, Epicurus develops a contractual theory
of justice.8 The relation of these two strands of Epicurus' thinking about justice is left
unexplored in Cicero's account, but it would be easy to construct a kind of Epicurean
disputatio of the following form: "some say justice is a contract", while "others, giving
in to Stoic criticism, say it is an inner virtue exemplified by sages," and "a third group
says that justice has its empirical/historical origins in rival groups beginning to feel
pity for each others' children" (cf. DRN 5. 1119-1123). These three attested positions
about justice show something important about Epicurus' overall theory, yet however
ultimately consistent, they hardly represent distinct positions of three competing his-
torical Epicurean groups.
Of the three positions on friendship described by Torquatus, the first is one that
many scholars have taken to best capture at least some elements of Epicurus' original
position, though it is itself problematic as set out, and Cicero only says (2.82) that he
seemed (mihi videbar) to recognize a dictum of Epicurus himself in it. This is further
complicated by the fact that the initial part of the dictum he recognizes--that friend-
ship cannot be sundered from pleasure--is said by Torquatus to hold for all three po-
sitions at a general level both as he opens and closes his exposition (cf. 1.70, without
pleasure "no principle (institutionem) for friendship can be found."). Cicero, however,
continues with what seems to be an allusion to KD 289 and a polemical one at that, and
further links this general principle to the claim that without friends one cannot live in
safety and without fear--which parallels the criticism one sees figuring prominently,
Phillip Mitsis 112

say, in later Stoic sources such as Seneca.10 In any case, Cicero goes beyond what KD
28 explicitly says to the extent that we may have reason to suspect that he is only mak-
ing use of the pretense of an allusion for his own purposes and to anchor a common
charge made by Epicurus' opponents.
It is also difficult to assess the various ways in which Cicero is distancing his own
views from those of his character, Torquatus, and the effect that literary tropes of pre-
sentation have on the accuracy of his account. Arguably at other key places in Tor-
quatus' exposition of Epicurus' doctrines, Cicero seems to be undercutting the argu-
ments of his literary creation in a manner similar to, say, the way that Plato often
handles Socrates' interlocutors.11 Thus, it is difficult to assess how much of the confu-
sion that scholars have seen in some of the positions outlined by Torquatus is being
foisted on Epicurus by Cicero and is merely part of a general strategy of showing how
Epicurean accounts of the virtues and friendship are inconsistent with hedonism.
With these preliminary features of Cicero's presentation in mind, it might be help-
ful to being by first setting out in more detail the two accounts of Epicurean friend-
ship marked out by Cicero as being later developments. There is not really enough
independent evidence to corroborate his claim, but both offer inklings of views that
are sufficiently different from any of the surviving glimpses we have in Epicurus' texts
that it has led some scholars to find Cicero's assertion plausible. However, as I have
suggested, these positions might just as well represent extracts of arguments culled
out of context from works by Epicurus or his followers that we no longer have. So, at
least in my view, caution is in order about them being actual distinct later Epicurean
"positions," however useful they may ultimately prove in shedding light on Epicurus'
original views.
I begin with the last and briefest of the three accounts (1.70). Torquatus here de-
scribes friendship as a certain compact (foedus quoddam) among the wise to love their
friends no less (nec minus) than themselves. By itself, of course, this admits of a range
of interpretations. Such a pact, for instance, might be innocent of any ulterior motives
for furthering some other self-interested goal. So, for instance, we might imagine two
sages becoming something like 'blood brothers' on recognizing their mutual virtue.
Like the vow of an ideal marriage, such a pact merely puts a seal on a particular re-
lationship as intrinsically valuable, without making it the first step in a wider egoistic
strategy. Of course, such a pact, just as in marriage, might also be used as a means to
some further goal, and this is certainly one construal that Cicero offers in his criticism
at 2.83. There is some reason for thinking that such a construal might not be a fair
account of the Epicurean position, however, since Epicurus in his account of compacts
of justice gives no independent binding force to contracts themselves and argues that
they are only valid so long as they reflect an individual's good. Thus, it does not seem
likely that an Epicurean sage would need to try to bind another sage in a strategic con-
tract of friendship in the hopes of furthering some long-term interest, especially since
the contract in itself, if the parallel with contractual justice holds, might have no inde-
pendent force and becomes void if it harms his interests. While it might initially seem
Cicero on Epicurean Friendship: A Reappraisal 113

plausible to see these contracts as part of an indirect strategy for maximizing pleasure,
it is not at all clear that Epicurean sages would either need or abide by them as part of
an indirect selfish strategy. This is not to say, of course, that contracting sages could
not be genuine friends and would not be committed to each other in sub-optimal cir-
cumstances. It is just that the Epicurean notion of a contract does not on its own serve
to underwrite or justify such commitments.
At the same time, Cicero also seems to leave open, however ironically, the possibil-
ity that such Epicurean compacts might not be purely strategic, and he suggests that
the kinds of attitudes that they underwrite should be extended to loving all the virtues
intrinsically as well (2.83). Whether this is ironic or not on his part, Cicero thinks that
he has shown that either such compacts are made strategically for further self-interest-
ed goals, in which case the intrinsic value of the compact and, hence, true friendship is
undermined, or, if the compact is indeed innocent of further self-centered goals, then
like an intrinsic love of virtue, it is incompatible with Epicurus' hedonism. In trying to
peer through Cicero's account to an original Epicurean argument, it is difficult to see
why sages would need to make contracts with one another for either strategic or be-
nign reasons. If contracts are strategic, they may not be binding on their own. Nor is
it clear how one can enter a contract to love another as oneself, unless one already does
so. If the love fades, moreover, that would seem to outweigh any contractual force.
Conversely, if the contracts are merely benign symbolic tokens of mutual love, one
wonders why Epicurean sages would go in for such outer trappings in the first place.
We have a certain amount of information about Epicurus and his relations with his
most important followers. We also hear from Philodemus about relations of friend-
ship among the gods.12 Nowhere do we hear about Epicurean sages making explicit
contracts of friendship with their fellows, nor gods with gods, for either symbolic or
strategic reasons. One reason for this, perhaps, is that neither sort of contract seems
to make much sense in an Epicurean context. Arguments from silence can cut both
ways, of course, and it might be that Cicero is indeed reporting a genuine later Epicu-
rean position. Nonetheless, as we have seen, there are too many obstacles to make any
secure inferences of any particular substance from this account to Epicurus' original
position.
In turning to Torquatus's second account, which is not explicitly limited to sages,
we see the same tensions presented between hedonism and the intrinsic valuation
of friends, but--at least in Cicero's reckoning--more clearly resolved against Epicu-
rus' hedonism. Torquatus relates that this group, though sharp enough (satis acuti­-
--apparently unlike Epicurus himself, cf. 2.81), are more timid (timidiores) than the
first group because they fear that friendship will be crippled by the demands of a self-
regarding hedonism. In fact, they seem to agree with Cicero in this and are described
as coming to this "more humane" (2.82) conclusion because of their fear of Academic
(vestra convicia), censure. Torquatus explains that, for these Epicureans, friendship
finds its beginnings in our initially seeking pleasure from one another, but by growing
accustomed (usus) to each other, familiarity (familiaritas) results, from which blos-
Phillip Mitsis 114

soms (efflorescere) such mutual love that we come to love friends for their own sakes,
even if there is no advantage (utilitas) from their friendship (I.69).
This account is striking for several reasons. On the one hand it seems a remarkable
anticipation of the long empiricist tradition of associationist theories beginning with
Locke and Hume, then on to Bentham, Mill, and Bain, and finally to contemporary
connectionists. Associationists tie the cognitive armature of individuals to their ex-
periential and causal history and treat psychological development as a mechanical em-
pirical process. In general, Epicurus was keen on such mechano-empirical explana-
tions of origins and we find glimpses of associationism throughout his philosophy, for
instance in his account of the development of language, society, concepts, etc. On this
telling, Epicurean friends come to value each other in their own right, like Epicurean
sages, but these attitudes of intrinsic valuation arise, not through conscious rational
choice, but through a causal and experiential process involving habit, familiarity, and
growing intimacy.
What is more striking, however, and almost unheard of in the wider context of an-
cient ethical theorizing, is Cicero's suggestion that the criteria grounding these friend-
ships trump Epicurean friends' conception and pursuit of their own telos, pleasure.
While Peripatetics, Stoics, and even some Cyrenaics might have had to do some fancy
footwork to justify valuing friends for themselves in the context of their overall theo-
ries of self-regarding good, such an explicit rejection, if that is what it is, of their own
favored view of the telos by these Epicureans is nothing short of astonishing. We might
have expected, for instance, that on coming to the realization that hedonism cannot
explain disinterested friendship, they might have either given up such friendship or
shopped around for a rival theory to explain how it could remain part of their concep-
tion of their own good. These Epicureans, however, although described as satis acuti,
persist in consciously maintaining an inconsistent set of doctrines because of their
supposed fear of Academic reproach and, thus, it seems, are satisfied with remaining
Epicureans only in name.
All of this is decidedly odd, and there are further difficulties in Cicero's picture.
He concludes at 2.82 that this group of Epicureans abandons hedonism, though not
Epicureanism, but confusingly changes terminology in criticizing their views. Where-
as he has Torquatus stating that such friendships blossom from association even if
there is no utility (nulla utilitas) in them, he describes such friendships in his critique
(2.82) as occurring, even with all expectation of pleasure disregarded (etiam omissa
spe voluptatis). While we might suppose that utilitas and voluptas are being used in-
terchangeably in the two accounts, two potential problems present themselves. First,
some scholars13 have seen a connection between this second account and the claim in
SV 23, “Every friendship is choiceworthy for itself, though it has taken its beginning
from benefit.” (Πᾶσα φιλία δι ἑαυτὴν αἱρετὴ [Usener: ἀρετή MSS] ἀρχὴν δὲ εἴληφεν
ἀπὸ τῆς ὠφελείας). One thing that is not clear in its compressed Greek, however, is
whether friendship continues to remain beneficial while being choiceworthy (and
presumably pleasurable) despite taking its origin from benefit; or whether it remains
Cicero on Epicurean Friendship: A Reappraisal 115

choiceworthy (and pleasureable?) even apart from benefit. Here ὠφελεία has seemed
to some to correspond to utilitas. In Torquatus's account, though, we have an explicit
wedge driven between utilitas and friendship that is not found in SV 23; and it is un-
clear whether this might be to serve Cicero's purposes in showing that the requirement
of loving a friend intrinsically is incompatible with hedonism. One could imagine,
however, given the flexibility of Cicero's vocabulary here, some friendships not being
advantageous or producing utility in some sense, but still being pleasurable and thus
choiceworthy.
More problematic, however, the notion of Epicureans giving up hedonism in the
face of Academic opprobrium certainly raises suspicions and it is hard to imagine a
concrete group of them consciously continuing to do so as ongoing adherents of the
school. Other explanations of this "position" beckon, and it might well be, for instance,
that Epicurus in a particular argumentative context gave an associationist account of
the origins and growth of friendships through familiarity, some of which might and
others which might not foster utility or an individual's pleasure. This does not mean, of
course, that Epicurus then endorsed those that do not, rather than, say, advise against
them and insist that they be eliminated like other sorts of troubling personal relations.
This is surely what we would expect and in many ways it strains credulity to believe
that any Epicureans would hold that troubling friendships should be maintained, re-
gardless of their genesis. Indeed, Epicurus is clear about the costs of such troubling
friendships (SV 56‑57; cf. DL 10.121b). Nor is the psychological explanation of the
genesis of a particular behavior sufficient to recommend it for an Epicurean; it must be
held up for rational evaluation and it is this that ultimately must justify it.
In any case, as presented, this second account is easy pickings for Cicero, since, on
the one hand, these Epicureans violate a cardinal rule of ancient theorizing about the
telos, while they concede exactly what Cicero is keen to show, i.e. that hedonism can-
not support the requirement that friends treat each other non-instrumentally. To the
extent that they do so, however, they are more acute than their master in Cicero's eyes
since they accept a dictum that he never did (2.82), i.e. that we can love our friends for
their own sakes without the expectation of pleasure.
We can now turn to Torquatus's first and most extensive account. For many schol-
ars his exposition has seemed incoherent and they have concluded that Cicero either is
unfairly foisting an inconsistent argument on Torquatus or that his account merely re-
flects the inconsistency of Epicurus' original position. Torquatus begins with the claim
that some Epicureans deny that the pleasures of our friends are to be sought after as
much as our own (1.66). Moreover, they further deny what seems to be the case to
some of their nameless critics, that such a view undermines the stabilitas of friendship.
He then concludes his argument by claiming that the sapiens will feel (erit affectus)
exactly towards his friends in the same way as he feels towards himself and will un-
dertake the same efforts for his friend's pleasure as for his own (1.68). At first glance,
there seems to be a fairly straightforward contradiction between his opening claim and
his conclusion, and we might wonder about the intervening steps. These would seem
Phillip Mitsis 116

to be initially purely self-interested and strategic. Reason itself (ratio ipsa), he claims,
advises that one acquire friends because a solitary life is filled with metus (fear) and
insidiae (snares), and friendships offer reassurance and the promise of pleasure. Thus,
since (1) a life of secure and continuing pleasure is not possible without friends; and
(2) friendship itself requires loving friends as much as we love ourselves; as a conse-
quence, (3) intrinsic mutual love between friends is brought about (efficitur) and it is
connected with pleasure (connectitur). This central argument is followed by the claim
above (1.68) about the sapiens that seems to serve as a purely parenthetical amplifica-
tion. Then Torquatus goes on to reiterate his rather imprecise claim about the relation
between friendship and pleasure in (3). He does so by drawing a parallel to his earlier
conclusion about the virtues and says that both virtue and friendship always inhaer-
erent (adhere, cling to, stick fast, etc.) to pleasures (1.68). Given that Torquatus had
earlier argued that the virtues are purely instrumental to pleasure, this would seem to
suggest that friendship is as well. But there is also a significant disanalogy. Virtues are
valued only instrumentally, whereas friends value each other as much as themselves.
At first blush, Torquatus account seems to be a jumble of claims that perhaps
reflect arguments from different contexts. The concluding assertion about the sage
seems to have been tacked on to the argument in a way that is both unexplained--why
this sudden claim about the sapiens?-- and also pleonastic. By the same token, Tor-
quatus interjects a strong non-instrumental requirement about friends' mutual regard,
but leaves its exact relation to hedonism unclear. Finally, the account begins and ends
with seemingly opposing views about how friends are to view and act towards one
another's pleasures. For many, all this is sufficient to suggest the inability of these later
Epicureans to coherently explain their justification of friendship, and it also has been
tempting to read this confusion back onto Epicurus himself.
It might be helpful to begin, however, by distinguishing two competing views con-
cerning the opening move of Torquatus' argument. He says that some Epicureans
deny that the pleasures relating to our friends are to be striven for, desired, etc. (expe-
tendas) as much as our own. Nonetheless he believes that they are still able to easily
defend a conception of friendship that does not totter. One problem is the scope of
this initial claim. Is Torquatus summarizing an overall position from the outset or is
he merely beginning to build a larger argument? Dorothea Frede, for instance, takes
the latter view. She argues that the claim about the lesser value of friends' pleasures
is just the first step in a developmental argument in which friends later come to rec-
ognize that in order to maintain friendships they need to come to value each other's
pleasures as much as their own. What distinguishes this argument from association-
ism, it seems, is that it appears to be a more rational and conscious process in which
individuals begin from and do not lose sight of their own self-interest, even if they
eventually realize that they must give up a narrow pre-occupation with it. Frede also
suggests that such an account is not necessarily contradictory.14 The pleasures of our
friends are not initially as desirable as our own, but they take on equal value in the
course of our recognition that a warrant of real friendship comes to be provided by
Cicero on Epicurean Friendship: A Reappraisal 117

our shared emotions. Our friends' pleasures, that is, gain an equivalent value to our
own in the context of new bonds of emotional mutuality. To Frede's picture one might
also add the observation that in this account it is only within the parameters of such
mutual relations that friends come to "love each other as much as themselves . . . aeque
amicos et nosmet ipsos diligamus . . ."), whereas in associationism, friends love others
propter se ipsos (1.69) and mutuality is not explicitly raised as a necessary criterion for
attitudes that are not purely self-interested. It may not be prudent, given the potential
imprecision in Cicero's renditions, to lean too heavenly on these possible differences in
the attitudes of friends toward one another in these two accounts, but neither should
they be discounted.
In many ways, Frede's is initially an attractive suggestion, but it still leaves unex-
plained exactly how this sea change in our attitudes towards friends' pleasures occurs
and why it is justified. Torquatus merely says "idcirco et hoc ipsum efficitur in amicitia
. . ." (1.67), without giving an account of its mechanisms or justification. More worry-
ing is that Frede maintains that "[t]he case of friendship is, indeed, exactly analogous
to that of the virtues."15 This is in line with Torquatus' assertion, but the virtues clearly
are only a means to pleasure--a necessary means, perhaps--but still only a means. But
surely if the virtues are exactly analogous to friendship, it makes friends purely instru-
mental to one's own pleasure as well. This is perhaps what we might have initially ex-
pected if we take Torquatus' opening sally to summarize this position, but such a view
seems to directly undercut the kinds of disinterested attitudes apparently underwritten
by the mutuality of loving a friend as much as oneself.
One notable line of argument has attempted to address this difficulty straight on
and to bring coherence to Torquatus' account in a different way.16 Eschewing a develop-
mental view of the passage, Timothy O'Keefe offers an analysis that endorses an initial
expectation that the argument will show us how friendship is compatible with friends
ultimately not valuing the pleasures of their friends as much as their own. Whereas
Frede must hold that the opening denial of the equal desirability of the pleasures of
our friends (quae ad amicos pertinerent) reflects loose writing on Cicero's part--since,
strictly speaking, those whose pleasures we view as being less valuable are not yet
"real friends"--, O'Keefe takes this claim at face value and thinks that it underlies the
whole argument in a crucial way. He does so by relying on an argument that, at least
in embryonic form, has become prominent in contemporary defenses of indirect con-
sequentialism.17 Of course, the coherence of contemporary indirect consequentialism
is hotly debated18, but O'Keefe attempts to find in Torquatus' exposition something
other than a mere farrago of Epicurean confusions or Ciceronian misunderstandings.
O'Keefe makes a distinction between our first-order and second order strategies
as friends.19 My overall strategy in friendship as an Epicurean is still to maximize my
pleasure, but I see that my best first-order strategy for doing so is to treat my friends'
pleasures as being equal to my own. On occasion, perhaps, this first order strategy
may not directly pay off, but by following it as a general rule, I will tend in this way to
best maximize my pleasures overall. On this account, only my own pleasures remain
Phillip Mitsis 118

intrinsically valuable, but in making my decisions, I recognize that the best way of
achieving my second-order or higher-level goal is to behave in a way that treats my
friends' pleasures equally with my own.
For O'Keefe, Epicurean friends are not, however, engaging in what he calls "double-
think", but instead recognizing that in one sense only their own pleasure has intrinsic
value, while in another sense they do indeed love their friends as much as themselves.
For the purposes of this discussion, I will eschew this language of "doublethink", since
I prefer the more neutral phrase used in contemporary discussions, "one thought too
many." Unlike "doublethink", it does not raise the wider specter of societal or ideo-
logical manipulation. Unlike some contemporary political or economic theorists who
see in two-level theories a potential for manipulating individuals for a larger good,
Epicurus never suggests that it would be good for a society of Epicurean friends as a
whole, in order to unwittingly maximize both their own and the larger social good,
to be manipulated into first order other-regard and to give up any further complicat-
ing second-order thoughts about the priority of their own individual interests. In an
Epicurean context, each individual must consciously adopt the kind of double strategy
proposed by O’Keefe.
So the question arises, why does O'Keefe think that this double strategy does not
lead to the problem of holding one thought too many? At first glance, it certainly
might appear that when I am acting, I must be able to hold in my thoughts both the
priority of my own pleasure and the equivalence of my own pleasure to that of my
friends. O'Keefe tries to meet this challenge by appealing to a distinction between
beliefs and behavior. He claims that Epicureans behave towards friends and treat them
in ways that display an equal concern for their interests, but they do so maintaining
the second-order goal of maximizing their own pleasure. This is meant to be different
from the claim that the Epicurean friend often will need to act with a kind of duplicity
by behaving outwardly in one way, while believing something else.20
In support of his claim, O'Keefe argues that Torquatus' terminology of "diligo" and
"affectus erga" can be read as behavioral or dispositional terms. Thus, when Torquatus
says in 1.68, "Quocirca eodem modo sapiens erit affectus erga amicum quo in se ipsum
. . ." it can have the more generic meaning that the sage 'will be disposed' in the same
way toward his friend as toward himself, not necessarily that he "will feel in the same
way toward his friend as toward himself." I find O'Keefe's reinterpretation of the Latin
unlikely, but more telling in this particular instance is its overall context. This claim
about the sage follows directly from (quocirca) the preceding claim that "we rejoice in
our friends joy as much as in our own, and are equally pained by their sorrows (1.67-8;
Rackham). These joys and sorrows hardly seem to be merely a description of ways of
treating or being disposed towards others (cf. SV 57, Ἀλγεῖ μὲν ὁ σοφὸς.)
More important, it is hard to see how O'Keefe's argument that these are best under-
stood as behavioral terms can ultimately evade the problem of these wise Epicurean
friends having one thought too many. O'Keefe sometimes speaks as if when Epicu-
reans are treating their friends' pleasures in manner equivalent to their own, they are
Cicero on Epicurean Friendship: A Reappraisal 119

merely behaving in a particular way. But, of course, Epicureans think that all of our
actions and behaviors are the result of particular cognitive states, so there can be no
clean division between second order thoughts and first order behaviors.21 When we are
acting on behalf of our friends' pleasures, we clearly must have corresponding beliefs
about their value. But it is unclear how those beliefs would not be in tension with our
second-order beliefs about the primacy of our own pleasures. By the same token, even
if we were to follow O'Keefe in cashing out our first-order relations to friends in terms
of dispositions, when we act we would still have one disposition too many--a disposi-
tion to treat our friends' pleasures as equivalent to our own, but still another general
disposition to treat our own as being more important. That is, I doubt that an Epicu-
rean can claim that our diathesis to pursue our own pleasure as primary is somehow
silenced when we are disposed to treat friends' pleasures equivalently to our own. The
same holds for the suggestion that Epicureans in one sense take their own pleasures
to be primary and in another view their friends' pleasures as having equal value.22 In
whatever sense they are taking these twin commitments, they are saddled with one
commitment, one sense, one disposition, or one belief too many.
O'Keefe further argues that some surviving bits of evidence (SV 34, SV 39) are
compatible with his account of first- and second-order strategies. As far as I can tell,
however, the thrust of these passages has other aims. At SV 34 Epicurus says that it is
not so much the help of friends that we need, but the confidence of that help. But I can
hold both of these thoughts at the same time since they are perfectly compatible and,
in any case, what would the first order strategy be in order to insure such confidence?
Should I make it a rule not to help friends in order to bolster their confidence in me or
should I aim to help them? SV 39 describes reciprocity and says that both those who
are always asking for favors and those who never do get it wrong. The passage is not
explicitly limited to relations of friendship, pace O'Keefe, but again, the worry is one
of finding a proper balance. There is no suggestion that one needs to aim at one thing
in order to gain another.
Annas raises the more general objection that Epicureans do not engage in multi-
level strategies in their deliberations, because they always are supposed to do just one
thing: to always check each and every prospective choice against one's final goal, plea-
sure.23

Εἰ μὴ παρὰ πάντα καιρὸν ἐπανοίσεις ἕκαστον τῶν πραττομένων ἐπὶ τὸ τέλος τῆς
φύσεως, ἀλλὰ προκαταστρέψεις εἴτε φυγὴν εἴτε δίωξιν ποιούμενος εἰς ἄλλο τι, οὐκ
ἔσονταί σοι τοῖς λόγοις αἱ πράξεις ἀκόλουθοι. KD 25 (cf. Men. 130, ταῦτα πάντα
κρίνειν καθήκει; Men. 132 τὰς αἰτίας ἐξερευνῶν πάσης αἱρέσεως καὶ φυγῆς) 

Unless at every appropriate moment you refer each one of your actions to the telos
of nature, but first divert your avoidance or pursuit to some other thing while acting,
your words will not correspond to your actions. (KD 25).
Phillip Mitsis 120

O'Keefe argues that the conception of deliberation we find in KD 25 does not pre-
clude a two-level strategy. First of all, Epicurus, he claims, is not suggesting that we
evaluate each and every one of our choices for its hedonic payoff. He is only recom-
mending that we evaluate our choices, desires, etc. at a more general level, a level that
approaches second-order reasoning about our goals. Whether the Greek allows this
is certainly arguable, since Epicurus' almost formulaic use of πᾶς in these contexts
certainly seems to rhetorically suggest "each and every" choice, action, etc. Even if we
were to grant O'Keefe the generality of these claims, however, it would still leave Epicu-
reans with the problem of balancing two more general and conflicting demands when
faced with individual decisions. So, for instance, I might generally come to believe that
I should treat my friends' pleasures on a par with my own and, of course, I might also
believe that my friends' pleasures are less valuable than mine in general. What do I do
now? In some sense, the whole point of indirect consequentialism was to distinguish
motives of individual actions from their overall consequences. O'Keefe's version of
a two-level theory, however, ascribes the same overall motivation for adopting both
of these general desires, i.e. to maximize my individual pleasure. In distinguishing
levels, he depends on a distinction not so much between motives and consequences,
but in valuing a friend's pleasure and acting on behalf of it. As we have seen, however,
Epicurus does hold such a view, given his account of the relation of beliefs and action.
For the Epicurean, our actions are grounded in and flow from our beliefs in the kind
of straightforward manner familiar from a long line of cognitivist/materialist theories.
As we are acting on behalf of our friends' pleasures, we are not somehow merely act-
ing or play-acting in a cognitive vacuum. In O'Keefe's account, Epicureans are un-
necessarily saddled with contradictory thoughts, dispositions, desires, etc. Therefore,
whatever difficulties we may face in putting together the various pieces of Torquatus's
account, O'Keefe's sophisticated attempt at finding coherence in it, I think, ultimately
falters.
One seemingly unavoidable conclusion in the face of these unsuccessful attempts at
finding coherence in Cicero's account is that he either is wittingly or unwittingly foist-
ing confusions on Epicureans or he is reporting, however tendentiously, the tensions
inherent in Epicurus' and his followers thinking about friendship. For the moment, it
seems that we do not have an entirely convincing way of choosing between these not
particularly appealing alternatives and in some sense, it may be that any account of
Epicurean friendship, if it tries to take Ciceronian evidence seriously, will inevitably
be limited by some combination of these shortcomings. Of course, Epicurus would
certainly not be alone among ancient philosophers in offering an account of friend-
ship whose various elements seem to involve us in a maze of arguments that can seem
to be merely a case of ignotum per ignotius. At a general level, however, I would still
maintain that Cicero's evidence should incline us to look in Epicurus' texts for signs
of disinterested friendship and to not dismiss the possibility merely on the grounds of
his hedonism. I have no confidence that this will lead to a smoothly coherent theory;
but it would be a mistake to try to bracket these problems in attempting to understand
Cicero on Epicurean Friendship: A Reappraisal 121

any ancient ethical theory,24 since they take us to the heart of what it means to pursue
eudaimonia in a world inhabited by others whom we value.

Endnotes
1
"Epicurus on Friendship and Altruism" Oxford Studies in Ancient Philosophy vol. 5 (1987) pp.
127-153.
2
I try to give a full discussion of the evidence in "Friendship", in The Oxford Handbook of Epi-
curus and Epicureanism, ed. Phillip Mitsis (Oxford, 2020), a section of which forms the basis for
this paper. I thank OUP for permission use portions of that discussion here.
3
For many acute observations about Cicero's rhetorical methods and the nature of his argu-
ments against Epicurean hedonism see Brad Inwood, "Rhetorica Disputatio: The Strategy of
"de Finibus" II." Apeiron: A Journal for Ancient Philosophy and Science 23, no. 4 (1990): 143-64.
4
Recent comprehensive discussions are that of Dorothea Frede, " Epicurus on the importance of
friendship in the good life (De Finibus 1.65-70; 2.78-85)," in Cicero's De Finibus: Philosophical
Approaches, eds. Julia Annas and Gábor Betegh (Cambridge, 2015) pp. 96-117 and Holger Es-
sler, Die Lust der Freundschaft und die Lust des Freundes von Epikur bis Cicero in Michael Erler
and Wolfgang Rother, Philosophie die Lust. Studien zum Hedonismus (Basel, 2012) pp. 139-60.
5
Fin. 1.65 "amoris conspiratione consentientes"; Fin. 1.67 "amicos et nosmet ipsos diligamus";
Fin. 1.69 "tamen ipsi amici propter se ipsos amentur"; etc.
6
Alain Gigandet. “Épicure, la philia et les philoi: un réexamen.” in Verite Et Apparence: Me-
langes En L'Honneur de Carlos Lévy, Offert Par Ses Amis Et Ses Disciples (Latinitates 8) Turnhout:
Brepols, 2016, pp. 89-106. David Konstan claims that there are different registers between φίλος
and φιλία in Greek conceptions of friendship. The latter, in Konstan's view, encompasses wider
social and familiar relations that include concerns about ἀσφάλεια, justice, and so on, while
φίλος corresponds to “friend” in the sense of an intimate, personal relation. Konstan argues that,
for Epicurus too, φιλία is an affective, stable relation of other-regard, love, and altruism associ-
ated with happiness that arises across a wide range of social relations. It is often best studied in
its origins and early development. The relations of φίλοι, in contrast, are based for Epicurus on
relations of personal loyalty and mutual utility. It was only in the centuries following Epicurus'
death that the relations of φίλοι began to be discussed in terms not only of mutual utility, but
also of philia as well. Such things as emotional attachment or, for instance, sincerity, as evi-
denced in Philodemus’s Peri Parresias, increasingly began to become topics for reflection among
later Epicurean φίλοι as they were faced with living with one another in a close community.
David Konstan, Friendship in the Classical World (Cambridge, 1997): 53 ff. Gigandet argues that
φίλος and φιλία are intertwined in crucial ways in Epicurus' thinking about friendship from the
very beginning.
7
Carlos Lévy assesses the role that Cicero's moral and political views play in his presentations of
Epicurean doctrines in an important discussion, "Cicero" in The Oxford Handbook of Epicurus
and Epicureanism, ed. Phillip Mitsis, (Oxford University Press, 2020).
8
For the general problem of the relation of aretaic and contractual justice see Phillip Mitsis , Epi-
curus' Ethical Theory. The Pleasures of Invulnerability, (Cornell University Press, 1988) chapter 3.
9
KD 28 "The same thinking that equips us to be confident that nothing terrible is either eter-
nal or even for very long, gives us within these same limits to understand friendship’s secu-
rity (safety, reliability, secureness) as being most complete (ἀσφάλειαν φιλίας μάλιστα κατεῖδε
συντελουμένην)." Many scholars interpret ἀσφάλειαν φιλίας as “the security provided by
Phillip Mitsis 122

friendship,” no doubt because of the prominence of ἀσφάλεια in connection with justice and
the security it provides from others. However, it is not clear whether φιλίας is functioning as
an objective or subjective genitive and Epicurus might just as well be pointing to the sureness
or reliability of friendship itself. Torquatus’s paraphrase at Fin. 1. 68-69 (amicitiae praesidium
esse firmissimum.) with its hints at rather unepicurean Roman military imagery emphasizes the
security that friendship provides from the hostility of others.
10
In the scholarship, friendship and the desire for security are often strongly linked. Some (e.g.
Matthew, Evans, “Can Epicureans Be Friends?” Ancient Philosophy 24 (2004): 407-24) have used
Seneca to claim that security from others is, in fact, the primary goal of Epicurean friendship.
But this merely takes onboard uncritically later Stoic polemics and mixes together Epicurus'
arguments about justice with those of friendship. Seneca’s use of Epicurean friendship as a foil
for Stoic views has been much studied mostly with negative conclusions about his reliability in
reporting Epicurus’ views. (e.g. Wolfgang Brinckman, Der Begriff der Freundschaft in Senecas
Briefen, .”Diss.: Cologne, 1963; Paola Gagliardi, Un legame per vivere: (sul concetto di amicitia
nelle lettere di Seneca), Atti e memorie - Università degli studi della Basilicata. Introduzione di
Giovanni Polara. Congedo: Lecce, 1991. The question of whether friendship has its origins in
inopia, indigentia, imbecillitas, (cf. Laelius, 26) and ultimately in the desire for help and protec-
tion is a fairly standard topos. But questions about origins need to be distinguished from claims
about the point or goal of friendship. Seneca nowhere in Ep. Mor. 9.8, for instance, mentions
pleasure as the goal of Epicurean friendships and we have no evidence in Epicurus' own texts
for the claim that rather than being a consequence of friendship, security is its point or aim. Cf.
G. Roskam, Λάϑε βιώσας: On the Vicissitudes of an Epicurean Doctrine. (Leiden, 2007) for an
important and wide-ranging discussion of the evidence.
11
Cf. "Epicurei e Cirenaici a confronto" with Enrico Piergiacomi, in Phillip Mitsis, Libertà, pia-
cere, morte. Studi sull'Epicureismo e sulla sua influenza (Carocci, 2019) 109-113.
12
I take up the theological dimensions of Epicurus' ethics generally in P. Mitsis, La teoria etica di
Epicuro. I piaceri dell'invulnerabilità (L'Erma di Bretschneider, 2019), chapter 6. I am much in-
debted to Enrico Piergiacomi's Storia delle teologie atomiste: da Democrito a Diogeniano, (Roma,
Sapienza Università Editrice, 2017) and to his
"Παρ' ἡμᾶς . Note di confronto su due studi recenti sull’etica di Epicuro" in«Syzetesis», 2.2 (2016),
pp. 131-154 
13
Eric Brown offers an extremely careful examination of the evidence in "Epicurus on Friend-
ship (Sententia Vaticana 23)," Classical Philology 97 
(2002), pp. 68-80. Also Essler, H. (2012)
‘Die Lust der Freundschaft und die Lust des Freundes von Epikur bis Cicero’. In M. Erler and W.
Rother (eds.). 2012. Philosophie der Lust: Studien zum Hedonismus. Basel. 139-160.
14
Frede p. 106. See as well Stern-Gillet, Suzanne "Epicurus and Friendship" Dialogue 28.2 275-
288 (1989) for a defense of the claim that a desire for ataraxia necessarily leads to mutual bonds
of trust and self-sacrifice.
15
Frede, p.106
16
O'Keefe, Tim. “Is Epicurean Friendship Altruistic?” Apeiron: A Journal for Ancient Philosophy
and Science 34, no. 4 (2001): 269-305.
17
In a stimulating and far ranging discussion Julia Annas explores the possibility of such a two-
level theory to explain the evidence. She argues that Epicurus needs such a theory, but does not
produce it. O'Keefe argues that Epicurus does in fact produce it. Cf. Annas, Julia. The Morality
of Happiness. (Oxford, 1993) pp. 240.
18
See, for instance, Mason, E. (1998). Can an Indirect Consequentialist Be a Real Friend? Eth-
ics, 108(2): 386-393.
Cicero on Epicurean Friendship: A Reappraisal 123

19
O'Keefe p. 294.
20
Such an account is developed by Matthew Evans (“Can Epicureans Be Friends?” Ancient Phi-
losophy 24 (2004) pp. 407-24, but his claim that non-instrumental friends would be too much of
an hedonic liability is rather peremptory. William Baird’s thought experiment and conclusion
about how Epicureans in Valuationville will achieve ataraxia more readily than those in Self-
opolis offers a more plausible alternative. William Baird, “Friends with Benefits: Other Regard
in Epicurean Ethics,” (ScholarWorks @ Georgia State University, 2011) p. 30 ff.
21
Cf. the discussion of Attila Nemeth, Epicurus on the Self (Routledge, 2017).
22
O'Keefe tries to bolster his account by making an analogy to the Stoics' distinction between
the goal and the target of their moral craft. The Stoics, however, are appealing to a craft with an
internal goal, i.e. I fulfill the moral craft when I properly aim at preferred indifferents, indepen-
dently of whether I successfully hit the target or not. The Stoics are keen to show how external
results are not part of the goal of their moral craft. In contrast, consequences are the goal for
Epicureans. By merely aiming at their friends' pleasure correctly, Epicureans are not thereby
fulfilling their goal, which remains separate and external from the process itself of aiming. Cf.
Phillip Mitsis, "Moral Rules and the Aims of Stoic Ethics" The Journal of Philosophy vol. 83, no.
10 (1986) pp. 556-57.
23
Julia Annas, The Morality of Happiness (New York, 1993) p. 241 ff.
24
I am indebted to Alain Gigandet and David Konstan for many conversations about
Epicurean friendship over the years. I am also pleased that this paper has served as the
occasion for a new friendship with Professor Dimitrios Andriopoulos and I wish to thank
him for all of his efforts in overseeing this volume and for arranging such an unexpected
honor, for which I am deeply grateful.