Love and the Value of a Life

Kieran Setiya
University of Pittsburgh

In The Sovereignty of Good, Iris Murdoch placed among the facts “which
seem to have been forgotten or ‘theorized away’” in contemporary phi-
losophy “the fact that love is a central concept in morals”: “We need a
moral philosophy in which the concept of love, so rarely mentioned now
by philosophers, can once again be made central” (Murdoch 1970, 2, 45).
Murdoch’s complaint could be made today. Although philoso-
phers sometimes write about love and ethics, they do so mainly in con-
nection with partiality or close relationships. They are interested in the
obligations and prerogatives of love. Love is a fact with which morality
must engage and by which it is potentially challenged. But its domain is
local. Outside the context of close relationships, love can be ignored. It is
not at the root of obligation, as such, but an aspect of our personal lives
that calls for moral reflection. On this approach, one could have an ade-
quate view of the nature and grounds of morality without having much to
say about love.
It may be a mistake to put love at the center of ethics or to give it
an essential role in the explanation of moral duty. But there is a tradition

For discussion of these topics, I am grateful to Samuel Asarnow, Chris Bobonich, Michael
Bratman, John Broome, Jorah Dannenberg, Cian Dorr, Peter Graham, Marah Gubar,
Caspar Hare, Liz Harman, Pamela Hieronymi, David Hills, Zena Hitz, Kristen Inglis,
Troy Jollimore, Simon Keller, Niko Kolodny, R. J. Leland, Anna-Sara Malmgren, Sarah
McGrath, Dan Morgan, Sara Mrsny, Mike Otsuka, Japa Pallikkathayil, Govind Persad,
Philip Pettit, Jacob Ross, Grant Rozeboom, Tamar Schapiro, Mark Schroeder, James
Shaw, Sam Shpall, Sharon Street, Ken Taylor, Michael Thompson, and Valerie Tiberius;
to participants in my spring 2012 seminar on love and ethics at the University of Pitts-
burgh; to audiences at Stanford University and the University of California at Davis; and
to the editors of the Philosophical Review.

Philosophical Review, Vol. 123, No. 3, 2014

DOI 10.1215/00318108-2683522
q 2014 by Cornell University

251
 KIERAN SETIYA

of thinking otherwise, on which Murdoch goes on to draw: the tradition

of morality as love for one another. The commandment to love one’s
neighbor as oneself belongs to the statement of the laws in Leviticus,
and its influence can be traced, through Christianity, to St. Augustine,
Kierkegaard, and the ethics of agape.1 But it is even less discussed in
secular moral philosophy than the golden rule.
This essay will not defend the ethics of love. But it prepares the way.
Its method is exploratory: I argue from attractive premises to puzzling
conclusions, conclusions that force us to confront the place of love in
moral philosophy. Although my target is love’s moral import, much of the
essay is independent of that. In section 1, I discuss the rationality of love:
whether there are reasons for love and what they could be. Rejecting
meritocratic conceptions, I argue that another’s humanity is enough to
justify loving them, though there may be other reasons, too. In the fol-
lowing section, I argue that it is rational to act with partiality to those you
love, moved by features of the circumstance apart from their relationship
with you. The love that is justified by another’s humanity involves the
disposition to act in this way. Nor is the rationality of being so disposed
conditional on love. In section 3, I draw out a consequence of these claims
for a contested issue in moral philosophy: should the numbers count?2
If I could either save one stranger from drowning or a group of three, am
I required to save the greater number? My conception of love conflicts
with the claim, which many find obvious, that I should save three instead
of one.

1. Love Is Not Irrational

Though we are said to love things that are not human beings, to love
justice, or philosophy, or baseball, we will focus on love for one another.
Even here, we should be struck by the variety of love. As well as romantic
love, there is the love involved in friendship, parental and filial love, love
of one’s brother or sister, of a distant relative, of one’s colleagues or the
members of one’s religious community, and love of one’s neighbor, if
there is any such thing. Philosophers who try to state the essence of love
typically do so in ways that are highly abstract, or that apply, in full, only to
some varieties of love. Thus, Robert Nozick wrote, in a well-known essay,

1. On love of neighbor in Judaism, see Goodman 2008; and for Christian con-
ceptions, Outka 1972 and Wolterstorff 2011.
2. This issue was made prominent by Taurek 1977; see also Anscombe 1967.

252
 Love and the Value of a Life

“What is common to all love is this: Your own well-being is tied up with that
of someone (or something) you love,” which covers many cases, if not all,
but he went on to emphasize the desire for union with the beloved, which
applies at most to some (Nozick 1989, 68, 70). If I love my children self-
lessly, I may hope for union with them only if it is in their interests. My
desire to be with them is conditional on whether they benefit from their
relationship with me.
Perhaps inspired by the model of parental love, a recurring theme
is that to love someone is to care about them, so that their interests count
as reasons for you to act: “Loving someone or something essentially means
or consists in, among other things, taking its interests as reasons to serve
those interests” (Frankfurt 2004, 37).3 But this is not sufficient for love
since I may care about the interests of strangers without loving them. And
it also seems unnecessary. David Velleman (1999, 353) calls the picture of
love as the desire to benefit, or the desire for union, “a sentimental fan-
tasy—an idealized vision of living happily ever after.” Not only is it “easy
enough to love someone whom one cannot stand to be with, . . . [in] most
contexts, a love that is inseparable from the urge to benefit is an
unhealthy love, bristling with uncalled-for impingements” (Velleman
1999, 353). This is overstated and a bit unfair. Concern for the good of
another need not involve the desire to benefit her in ways that infringe on
her autonomy, or when the benefit is not a matter of urgent need. But
Velleman is right that there can be love without the desire for union—he
gives the example of a difficult relative—and that there can be love with-
out the desire for another’s good. Romantic love can be both selfish and
possessive, and while it may then be flawed, it does not cease to be love.
No adequate conception of love can ignore or deny its diversity.
What then can we say, in general terms, about the nature of love? It will be
enough for us that a common element of love, in many of its forms, is
vulnerability to the needs and interests of another human being: the
special way in which most of us care about our children, parents, friends,
and romantic partners.4 In its degree and its directedness, this goes
beyond moral concern. We would do more for our loved ones than for

3. See also Kolodny 2003, 136 –37; Jollimore 2011, 112– 13; and for a long list of
related views, Velleman 1999, 352.
4. Velleman (1999, 361) suggests that love “makes us vulnerable to the other” but
also that it “arrests our tendencies toward emotional self-protection [and] disarms our
emotional defenses.” The first point seems right; the second is as narrow and sentimental
as the claims that Velleman rejects. It may be despite myself that I am emotionally vulner-
able to someone I love: my emotional defenses are down but not “arrested” or “disarmed.”

253
 KIERAN SETIYA

just anyone, and their suffering affects us more. This is true even of dis-
tant friends, or ones we rarely see. At the same time, love is shaped by our
relationships to those we love. To echo Velleman (1999, 353), while I may
“think of myself as an agent of [my children’s] interests,” I do not relate to
my friends, or to my wife, in that paternalistic way.
Our question is whether there are or can be reasons for love, in this
paradigmatic form. This is often denied, as by Harry Frankfurt in The
Reasons of Love : far from being a response to reasons, or to “the inherent
value of its object,” love is “the originating source of terminal value”
(Frankfurt 2004, 38, 55). There are several issues to distinguish here.
Frankfurt makes two claims: that love is the ultimate source of practical
reasons and that love is not a response to objective value. The first claim
may entail the second, but the converse does not hold. One could deny
that there are reasons for love without finding in love “the origins of
normativity” (Frankfurt 2004, 48). In any case, I will set aside the argu-
ment that there are no reasons for love because there are no reasons that
do not depend on love; Frankfurt’s critique of objective value is not our
present concern. We should also distinguish the claim that there are
reasons for love from the more specific contention that they consist in
valuable qualities of the beloved, qualities she has apart from her relation
to you, such as beauty, vivacity, or intelligence. Some of Frankfurt’s argu-
ments are directed specifically at the latter: “I can declare with unequiv-
ocal confidence that I do not love my children because I am aware of some
value that inheres in them independent of my love for them” (Frankfurt
2004, 39). This is consistent with there being reasons for love that turn on
Frankfurt’s relationship with his children, not their intrinsic character.
Having marked this possibility, we will consider, first, the view that
Frankfurt most explicitly rejects: that there are reasons for love and that
they are qualities of the person one loves that distinguish her from other
people. In Love’s Vision, Troy Jollimore (2011, 25 – 26) defends a qualified
version of this claim: “Loving someone is, in large part, a kind of positive,
appreciative response to her in virtue of her attractive, desirable, or oth-
erwise valuable properties. . . . Love is thus a matter of reason, insofar as it
is a response to something external that attempts to be adequate to the
nature of its object.” But love is “something between an appraisal and a
bestowal”: “It is an appraisal insofar as it takes itself to be responding to
independent, preexisting values”; what is bestowed is “the sort of close,
generous, and imaginative attention that allows valuable features of this
sort fully to reveal themselves” (71 –72). On this view, the valuable qual-
ities of another constitute reasons for love, to which one responds in

254
 Love and the Value of a Life

loving her appropriately. Love then involves appreciative vision, attention

to the beloved that brings out qualities otherwise “muted, subtle, or dif-
ficult to discern” (71).5
In light of the diversity of love, we should be wary of equating it
with appreciative vision. That is often involved in love, but it is not essen-
tial. Parents can be unsparingly critical, and while that may not be for the
best, it need not conflict with their claim to love. What is more, their
children may be happy to reciprocate. But we can set this aside. More
troubling is the claim that distinctive qualities of the person you love are
what justify you in loving her. The idea that there are reasons for love and
that such reasons consist in the merits of the beloved is notoriously prob-
lematic. Jollimore (2011, 15 – 18) gives a lucid account of these problems
in the opening chapter of his book.6 If you have qualities that are reasons
for me to love you, and that justify my love, does it follow that anyone
who is aware of these qualities should love you, too? (The problem of
universality.) That I should love anyone who has these qualities? (The
problem of promiscuity.) That if someone else has more of these qual-
ities, I should love her instead, or more? (The problem of trading up.)
That if you lose these qualities, I should stop loving you? (The problem of
inconstancy.) None of these implications seems right. It is therefore
tempting to reject the picture of love as a rational response to reasons,
which are distinctive virtues of the person you love.
Jollimore is not convinced. He responds to the difficulties by insist-
ing that reasons for love need not require it: reasons can make “objects or
options eligible for choice without making it mandatory that agents select
those particular options” ( Jollimore 2011, 93 – 94).7 In a different idiom,
we can distinguish “insistent” from “noninsistent” reasons.8 The latter
make it appropriate or rational to respond in a certain way, without re-
quiring that response. By contrast, insistent reasons to f are either deci-
sive, implying that one ought to f, or contribute to such requirements, as
pro tanto reasons. Insistent reasons combine to yield stronger reasons and
fix what one should do, all things considered. If we think of people’s
merits as noninsistent reasons for love, we solve most of the standard
problems. It does not follow from the fact that your qualities make you

5. For love as a response to reasons, see also Jollimore 2011, 95, 118.
6. See also Kolodny 2003, 139 –41.
7. See also Jollimore 2011, 118, 121; his terminology is adapted from Raz 1999
[1997].
8. See Kolodny 2003, 163, following Kagan 1989, 378– 81.

255
 KIERAN SETIYA

worthy of love, and in that sense justify me in loving you, that others
should love you, too, that I should love anyone who shares those qualities,
or that I should love those who have more of them instead of, or more
than, you ( Jollimore 2011, 123 – 34).
The problem of inconstancy, however, remains. If reasons for love
are noninsistent but not redundant, it must be possible to lack such
reasons, and so for love to be unwarranted. On the most natural view, it
is rational to love someone only if one’s beliefs about them would, if true,
provide a reason.9 Suppose I love you because you are funny and con-
siderate, but as time goes by, you change. You no longer make me laugh,
and you think only of yourself. Does it follow that I should not love you
anymore? Not immediately. For love may inhibit the cold reappraisal of
your qualities, and it may involve a disposition to find new virtues as you
change.10 Consider, then, the more extreme case in which I see that my
reasons are gone, and I cannot find replacements. This may be rare, but
as Jollimore (2011, 139) concedes, it is not impossible. Here the quality
theorist must deny that it is rational to love. And this is a mistake. It is not
that love must continue in the face of radical change—that love is not love
which alters when it alteration finds—but that unconditional love is not
irrational. (Here and throughout, I use “irrational” in the broad sense
that connotes a failure of practical reason, not the narrow sense that
suggests internal conflict or culpability.)11 It is not irrational to love my
wife with a constancy that would survive the loss of her admirable qual-
ities, the things that drew me to her in the first place or that attract me
now. Nor is it irrational to love one’s children, come what may.
In his most direct response to this concern, Jollimore (2011, 139 –
41) argues that we can love people for qualities they used to have. We
need not focus on present virtues: that you were once funny, or kind, can
be a reason for me to love you now. In effect, this transforms the tempo-
rary into the permanent. The property of having once been kind is one
you cannot lose. If this is my reason for loving you, the problem of incons-
tancy cannot be raised. But this only takes us so far. Even if it works, the

9. In principle, it might be rational to love someone in the absence of such beliefs,

even though I should not love them, all things considered, if there is no reason for love.
In effect, the absence of distinctive merits would be a defeating condition. Although
I don’t discuss it explicitly, the arguments below refute this view. It, too, conflicts with
the rationality of unconditional love.
10. These points are stressed by Jollimore in responding to the problem of incon-
stancy; see Jollimore 2011, 138– 39.
11. I discuss this contrast in Setiya 2012, 8.

256
 Love and the Value of a Life

turn to history cannot solve a related problem: that of error. It is not

irrational to love my wife with a constancy that would survive not only
the loss of her admirable qualities but the discovery that they were never
real. If someone can lose these qualities, there is a more peculiar case in
which they were absent all along. You never acted from true kindness,
always with an ulterior motive; your “jokes” were unintentional; my mem-
ories are false. The point is not that my love must survive in this scenario: it
is rational for it to end. But it is not a mistake to go on loving my wife when
I learn that I was wrong about her from the start. This kind of commit-
ment conflicts with the need for valuable qualities, even past ones, as
reasons for love.
There is a deeper complaint here, against love as a response to
positive qualities. To repeat what I said before: if reasons for love are
noninsistent but not redundant, it must be possible to lack such reasons,
and so for love to be unwarranted. There must be people who are not
worthy of love, or circumstances in which, not being aware of their virtues,
it is irrational to love them. This implication is profoundly troubling. It
can seem less so if we concentrate on friendship and romantic love, or
conflate reasons for love with reasons for maintaining a relationship. That
there are people I shouldn’t marry or date or be friends with seems not so
strange. But are there people whose parents shouldn’t love them? What
prevents that, if only merit justifies love?12
It can also seem less strange that there are people we should not
love if we think of love as concern for another’s interests and conceive
those interests in morally neutral terms. Suppose it would make my child
happy to succeed in some awful project. It may be true that I should not
want to help him or promote the satisfaction of his desires. But the
expression of parental love is not compelled to take that form. When
I love my child, my concern for his well-being may be morally
conditioned.13 It may be out of love that I try to prevent him from
doing something wrong, to undermine his projects, even at the cost of
utter misery. If he succeeds, it may be out of love that I try to change him,
to make him repent or turn himself in. What I want for my child is not
pleasure but the goods of life, and so the good of moral virtue. According
to Jollimore (2011, 142), if “one’s beloved were to become evil . . . love for

12. The response I am inviting is the opposite of Freud’s, in Civilization and Its Dis-
contents : “A love that does not discriminate seems to me to forfeit a part of its own
value . . . not all men are worthy of love” (Freud 2010 [1930], 82).
13. On this point, see Wolterstorff 2011, 76– 78.

257
 KIERAN SETIYA

him might require its own extinction.” Of course, love might end, and this
might be rational. But it is not required. It is not a mistake to love one’s
child in the face of terrible wrongdoing, though one’s love for him may
be transformed.14
What I have argued so far is that if there are reasons for love, they
do not consist in a person’s particular merits. The standard problems
show that such reasons are not insistent. And unless they are redundant,
noninsistent reasons struggle with constancy and error. It is not irrational
to persist in loving someone, even as you learn that their positive qualities
have faded or were never there. Love is not in that way meritocratic.
Nor are the arguments for the quality theory especially strong. It is
true that if you ask me why I love my wife or child, I may have something to
say, and what I cite in answer to your question may look like reasons:
“because she is funny”; “because he is full of joy.”15 But there is a certain
awkwardness in putting things this way. It is more natural to say that this is
“what I love” about him or her than why I love them. My remarks can be
interpreted not as attempts to justify love but as expressions of it. Jolli-
more is right to connect love with appreciation, at least in a familiar range
of cases. It is often part of romantic or parental love to be sensitive to, and
moved by, the positive qualities of one’s partner or child. It is then a
consequence of love that you can point to distinctive merits in the person
you love. In doing so, you show the quality of your attention to that
person. That is why it is especially apt, and especially touching, to respond
by citing subtle virtues, not manifest strengths, and to focus on idiosyn-
crasies, not properties that are widely shared. Your awareness of these
qualities shows the depth of your appreciation.
It does not follow from any of this that there are no reasons for
love, just that they do not consist in our distinctive merits. To anticipate,
my view is that another’s humanity is sufficient reason to love them: no
one is unworthy of love. But there are complications here, best explored
through the idea of relationships as reasons for love. In “Love as Valuing a
Relationship,” Niko Kolodny (2003, 146) rejects the quality theory but
defends the existence of such reasons: “Love is . . . rendered normatively

14. Not surprisingly, reflection on love in the face of evil has been part of the agapic
tradition. What does it mean to love one’s enemy? “However difficult to apply in practice,
for the sake of the neighbour one may have to resist his exploitation as well as attend to his
needs” (Outka 1972, 21).
15. For this argument, see Jollimore 2011, 18– 20; and for a reply to which I am
sympathetic, Kolodny 2003, 138– 39.

258
 Love and the Value of a Life

appropriate by the presence of a relationship. Love, moreover, partly

consists in the belief that some relationship renders it appropriate, and
the emotions and motivations of love are causally sustained by this belief
(except in pathological cases).” It is crucial to separate two parts of Kolod-
ny’s view. The first is a theory of relationships as insistent reasons for love.
The fact that you are my child, or that we have a certain history together,
can be a reason to love you that makes it unreasonable not to.16 This claim
seems plausible to me, though I will not defend it at length. If someone
fails to love their own child, or their love for their partner or friends is
utterly capricious, this shows a defect of character. Nor is there a substan-
tive distinction between defects of character in one’s response to the facts
and defects of practical reason.17 There is no room to say that the indif-
ferent parent is ethically flawed but not irrational, at least not in the broad
sense of “irrational” that connotes a failure to respond to reasons. As
Kolodny insists, it does not matter to this point that love is not a voluntary
act. We can have reasons to respond in ways that are not voluntary, or up
to us, as in reasons for emotion and belief (Kolodny 2003, 138).
Kolodny’s second claim is much more dubious. He denies that it is
possible to love someone without believing that one’s relationship with
them is a source of practical reasons.18 The idea of relationships as insis-
tent reasons for love does not entail this claim about the nature and
conditions of love itself. And the claim is open to objection. The obvious
problem is that relationships of friendship and romantic love partly con-
sist in, and so cannot precede, the emotional vulnerabilities of love itself.
How can the love involved in friendship rationally begin? It cannot be
justified by friendship or involve the belief that we are already friends.
Friendship cannot be a reason for the emotional vulnerabilities through
which it comes to be.
Kolodny (2003, 169 – 70) replies to this objection with an incre-
mental view. We first interact as acquaintances in ways that justify, and are
seen by us to justify, ongoing interaction and the beginnings of concern.
As we build our relationship, the reasons intensify and the responses
deepen. Eventually, we reach the point that his theory of love describes.
No doubt some relationships are like this. They grow steadily and recip-
rocally. But does it have to be that way? What about momentous, instan-

16. On relationships as insistent reasons, see Kolodny 2003, 167– 68.

17. This is the argument of Setiya 2007.
18. He goes on to specify, in some detail, the content of this belief; see Kolodny 2003,
150 –53.

259
 KIERAN SETIYA

taneous love? Or love that is unrequited and unjustified by a history of

interaction? What about Dante and Beatrice?19 The Sorrows of Young
Werther (Goethe 2012 [1774])? Stendhal’s “passionate love” for Mathilde
Dembowski (Stendhal 1975 [1822])? Rather than treat such love as
pathological, since the relationship is not plausibly thought to justify its
force, we should deny Kolodny’s second claim, which does not follow
from his first. We should disentangle the idea of relationships as insistent
reasons for love from the alleged condition of love, that one take one’s
relationship to be a source of reasons. In doing so, we make room for love
at first sight and from a distance.
The simplest view to take here, and the one that I accept, is that it is
sufficient to justify love that its object is another human being. It would be
a mistake to love a baseball card more than I love my child, but there is no
one it is irrational to love.20 Any one of us would be justified in loving any
other. It is consistent with this claim that relationships are insistent
reasons for love, but not that they make love “normatively appropriate”
if that means it would be inappropriate otherwise. Kolodny’s argument
for the latter claim turns on the fact that love can seem out of place. What
should we make of the wife who loves her abusive and uncaring husband
(Kolodny 2003, 137)? Suppose I wake up with feelings of intense concern
for one of my son’s classmates, a child I hardly know?21 What these
examples show, however, is not that love can be irrational, but that it
can take misguided forms. What is irrational is to love someone in the
wrong way: possessively, perhaps, or self-destructively. It is not a mistake to
love them at all. This point reflects, again, the diversity of love, which
need not aim at benefit unconditioned by morality. The abused wife
should not want to advance her husband’s interests, as he conceives
them; most likely, she should end the marriage. It is easy to imagine

19. This case is raised against Kolodny in Stump 2006, 26– 27.
20. For the baseball card example, see Jollimore 2011, 22 –23. A more difficult case
is love for nonhuman animals. Nothing I say in the rest of the essay depends on this—my
claim is that humanity is sufficient, not necessary, to justify love—but I doubt that it is
rational to love one’s dog with the degree of partiality one has for one’s wife or child. To
anticipate section 2: in ordinary circumstances, you should not save your dog at the cost of
a stranger’s life.
21. This is a variation on Kolodny 2003, 144 –45. Kolodny’s treatment differs in two
ways. First, the child he considers is a complete stranger, known only by his name on the
class roster. I think it matters to love whether its object is available for singular thought.
Since I am not acquainted with him, this may not be true of the child in Kolodny’s case.
Second, his topic is whether my feelings for the child could even count as love. Our
question is whether, if they did, such love would be irrational.

260
 Love and the Value of a Life

oneself saying to her, “You need to move on; forget about him; let go.”
And that might be right. Her husband should not occupy her thoughts.
But she need not cease to love him, and so to wish for him, sincerely and
profoundly, a better life. The case in which I love a stranger’s child may
seem disturbing. But again, it is easy to misconstrue. It would be wrong to
love the child as I love my own, to regard myself as the agent of her
interests, or to invade her privacy. But love does not require these things.
It need not compel attention or affect what one does from day to day.
Compare my attitude to friends I have not seen or thought about for
years. Still, I care about them more than I care about just anyone; in
fact, I continue to love them. Is it wrong to feel that way about the strang-
er’s child? My obligations to lost friends may differ from my obligations to
her, but it is not irrational to be moved by the child’s well-being as I am
moved by theirs. I do not need a relationship to justify this response.
That we do not need such reasons is confirmed by cases of amne-
sia. Kolodny (2003, 141 – 43) suggests that forgetting a relationship would
tend to extinguish love. He may be right about this, but the effect is not
inevitable: as I have argued, love is possible in the absence of such beliefs.
Imagine, then, a case in which I forget how I relate to my wife and child.
Awakening from a coma, having lost all memory of our past relationship,
I still know them by name and find myself deeply stirred. For reasons I
do not understand, I love this woman and this boy. Is my love for them
irrational? Assuming that it is not, relationships are not required as
reasons for love.
It might be thought that when it persists in this way, love is justified
by its origins: I used to know of a relationship that is a reason for love. But
again, we can imagine otherwise. To take a more fantastic case, suppose
that, instead of suffering amnesia, I discover that we came into existence
an hour ago and that our “memories” are false: we have no past rela-
tionship. Is love then irrational? I think the answer is still no. If you
were to learn that your shared past is an illusion, would you stop loving
the people you took to be your friends and family? Would you be making a
mistake? (If you struggle to imagine how you could discover this, suppose
instead that you remain deceived. Do you lack sufficient reason to love
these people? Should you stop?)
We know from the rationality of love at first sight, and from a
distance, that love can be justified without an existing relationship. The
same point holds in the more peculiar cases I have described. It is rational
to love someone without believing that you have a relationship with them,

261
 KIERAN SETIYA

and the absence of a relationship is not a defeating condition, in light of

which love must end.
No doubt some will be unmoved. Taking a more cynical view of
love at first sight, or in the face of a bad relationship, amnesia, or skeptical
delusion, they will insist that we need relationships as reasons for love.
I argue against this further in section 2. For now, suppose that we do
not. The cases considered so far support a simple view. That you are
another human being is enough to justify loving you, if not to require
such love. This reason does not fade in the absence of a relationship or
of any distinctive merit. The upshot is a conception of love that has affin-
ities with Velleman’s theory in “Love as a Moral Emotion.” For Velleman
(1999, 366), “respect and love [are] the required minimum and optional
maximum responses to one and the same value,” that of personhood or
rational will. The object of respect and love is not a property, as this for-
mulation might suggest, but its instance in a particular human being. In
effect, personhood is seen as an insistent, decisive reason for respect, and
a sufficient, noninsistent reason for love. One way to illuminate the pic-
ture of love I am proposing is to compare it briefly with this account.
First, Velleman says that love is “optional,” suggesting not only that
it is permitted but that it is not required. When he asks why we love some
people and not others, he does not appeal to reasons but to contingent
facts of salience and the limits of emotional energy (Velleman 1999,
370 – 72). In contrast, I am open to the view that relationships can be
insistent reasons for love.
Second, where Velleman cites personhood as the ground of love,
my view cites humanity. The difference matters to love for human beings
who are not persons in the Kantian sense, such as infants and the severely
impaired. But the contrast may be overdrawn. I hold that humanity is
sufficient reason for love, not that it is necessary; I do not object to love for
rational beings that are not human. And Velleman (1999, 365) allows that
“love is felt for many things other than possessors of rational nature,”
claiming only that love for persons is distinctive in being a response to the
value someone has “by virtue of being a person or, as Kant would say, an
instance of rational nature.” Although I am skeptical of this, I need not
press the point. The rest of my argument goes through if we replace
“humanity” with “personhood”: it does not turn on cases in which they
come apart. More generally, my view is that whatever property gives us
moral status of the sort that commands respect, it is this property that
justifies love. Any theory of moral status that attributes it to rational
human beings will do the work my argument requires.

262
 Love and the Value of a Life

As this remark suggests, I think Velleman is right to connect love

and its reasons with the morality of respect. It is not clear to me exactly
what he means when he calls love a “moral emotion” and describes it as
“an arresting awareness of [the] value” of personhood or rational will
(Velleman 1999, 360). On my account, love need not involve a belief in
the value of humanity, a belief that it justifies love or demands respect. But
a weaker thought is tempting. If humanity is a reason for love, just being
human is enough to warrant the intense concern that we devote to our
close friends and family. Can it then be rational to hold other human
beings in absolute disregard? Can we take something to be a sufficient
reason for love but not a conclusive reason for respect? Although it is not a
contradiction, this claim may be normatively incoherent. At least, there is
room to make the case. The idea that every human being is worthy of
love, that it would be rational for anyone to love them, is a picture of the
value in human life that may have moral implications.
We will come back to the morality of love. I end by noting a final
contrast with Velleman, about the reasons for which we act when we act
from love. Considering the case in which a man could save a stranger or
his drowning wife, but cannot save both, Velleman writes that of course he
should save his wife but “the reasons why he should save her have nothing
essentially to do with love” (Velleman 1999, 373).22 These reasons turn
instead on the “mutual commitments and dependencies of a loving rela-
tionship” (373). For Velleman, there is thus a great divergence between
what justifies love, which is rational personhood, and what justifies par-
tiality in action. In the following section, I defend an opposing view, on
which the reasons for which we act when we act from love need not appeal
to relationships, but can draw simply, and directly, on the needs of those
we love.

2. Acting from Love

In his well-known discussion of the drowning wife, Bernard Williams was
concerned to deny that morality is involved in the husband’s decision,
that it must give permission for him to save her. As he put the point, in a
memorable passage: “[This] construction provides the agent with one
thought too many: it might have been hoped by some (for instance, by his
wife) that his motivating thought, fully spelled out, would be the thought
that it was his wife, not that it was his wife and that in situations of this kind

22. The example comes from Williams 1981 [1976], 17–18.

263
 KIERAN SETIYA

it is permissible to save one’s wife” (Williams 1981 [1976], 18). As ensuing

treatments have shown, Williams’s claim about this case is puzzling. His
phenomenology seems right: the husband should not dwell on or pause
to debate whether it is permissible to save his wife. But it does not follow
that he will save her whether or not it is permissible to do so, or that love
requires him to act that way.23 He can save her because she is his wife while
being disposed to refrain if the only way to save her, in the circumstance, is
by doing something morally wrong.
Our topic is not Williams on moral permissibility but a more rad-
ical claim that others have made. In a letter quoted by Liam Murphy,
Derek Parfit wrote: “It’s odd that Williams gives, as the thought that the
person’s wife might hope he was having, that he is saving her because she
is his wife. She might have hoped that he [would save] her because she was
Mary, or Jane, or whatever. That she is his wife seems one thought too
many” (Murphy 2000, 140n36). This complaint is echoed by Harry Frank-
furt (2004, 36n2): “I cannot help wondering why the man should have
even the one thought that it’s his wife. Are we supposed to imagine that at
first he didn’t recognize her? Or [that] he didn’t remember that they
were married, and had to remind himself of that? It seems to me that the
strictly correct number of thoughts for this man is zero.” Frankfurt’s
description is perhaps unfair. It is absurd to imagine these thoughts
going through the man’s head, but as we saw with Williams, that does
not mean he is not aware that the woman is his wife or that this is not
among the reasons for which he acts. Still, that is close to the position I
accept. In acting from love—as, for instance, in saving one’s wife—one
need not act on the basis of one’s relationship to the object of love. That
she is your wife may not be “one thought too many” but it is one thought
more than you need.
There are two arguments for this. The first builds on the results of
section 1. There I argued that the fact of another’s humanity is enough to
justify love. When you love someone for this and no other reason, and you
act from love, you do not act on a relationship to the one you love. Per-
haps you cross the room to speak to someone you love at first sight. The
thought of a relationship is, in this case, one thought too many, simply
because it is false. It cannot be essential to acting from love that one act for
that reason. What about the cases of amnesia and the illusory past? Is it
irrational to favor the interests of your wife and child, or the people you

23. For compelling arguments to this effect, see Herman 1993, chaps. 2 and 9.

264
 Love and the Value of a Life

took them to be, in situations of this kind? That is hard to believe, on the
assumption that you still love them, as you did before, and that you are
justified in doing so. If partiality in emotion is warranted without the need
for a past relationship, so is partiality in action.24 Awakening from a coma
with your memories destroyed, or discovering that they were fake, you
love the person in front of you, and you are justified in doing so. Would it
then be wrong to give more weight to their interests than to those of other
people to whom you have no emotional response? Are you justified in
feeling love only so far as you do not act on it? Surely not. If love is
warranted, what is warranted is a response that affects how you behave,
and the reasons for the response transmit to that behavior. Velleman is
wrong to draw so sharp a line between the reasons for love and the reasons
for acting from love. We do not need relationships to justify either one.25
The second argument is independent of these concerns. It does
not rest on, and in fact supports, the view that we do not need relation-
ships as reasons for love. In exploring this argument, it will be useful to
have before us a schematic example of partiality in action. Three strang-
ers are stranded at sea, about to drown. Your wife, M, is drowning, too.
But the strangers are too far away for you to save all four. If you save your
wife, the strangers will drown. If you let her drown, you can save the
three.26 The circumstance is otherwise unexceptional. You have no
special obligation to the strangers; nor do they have a right to aid beyond
the rights of any human being. Their life or death would have no unusual
consequences, good or ill. They are not about to cure a terrible disease or
start an unjust war. Nor is your wife. Finally, neither the strangers nor your

24. See also Kolodny (2003, 176), though he draws the opposite conclusion, that
neither response is justified except by the relationship.
25. Note that the reasoning here is specific to the case of love. I do not claim that
reasons for emotion always count as reasons for the behavior that emotion would naturally
prompt: one might have reason to be angry but no reason to express one’s anger in action.
Velleman (1999, 353– 54) motivates the divergence in the present case by contrasting the
attitude of love, which is directed at an individual, with the desire to benefit her, which
aims at a specific outcome. As we saw in section 1, however, this contrast is overdrawn,
since the desire to benefit someone need not have the unhealthy, invasive character to
which Velleman objects. A disposition to be moved by the needs of the beloved—to be
moved to act, that is, not just to feel—is paradigmatic of love. It is in fact unclear how
sharply we differ on this score. Velleman (1999, 373) suggests that reasons for love may
come apart from reasons for acting from love, but what he actually says is that the reason
why you should save your wife turns on your relationship with her. As will become clear,
I agree with the second claim, but not the first.
26. This case is adapted from Anscombe 1967, 17.

265
 KIERAN SETIYA

wife are moral saints or ethically depraved. Like many, I believe that, in
this circumstance, it is permissible to save your wife: you would be justified
in doing so. Something stronger may be true, that you should save your
wife, all things considered. It would be wrong to let her drown, even to
save three lives. But I will focus on the weaker claim. The grounds that I
invoke do not explain why you should save your wife instead of the stran-
gers: they are sufficient, not decisive, reasons to act.
According to the weaker claim, you would be justified in saving
your wife, in the circumstance described. If you know that you are in this
circumstance, it would be rational to save her. There is a further question.
When you dive into the water, by what facts about the circumstance is it
rational to be moved? Should you save her because she is your wife or on
the ground of her relationship with you? I do not think you have to be
moved in this way in order to be rational. When you dive into the water
and save your wife from drowning, it is rational to be moved simply by the
fact that she is in need and by other facts about the circumstance—the
lack of exceptional conditions—not by the fact that she is your wife.
Some take a hard line here, holding that it is not merely inessential
to acting from love that one act on a relationship, but that it is antithetical.
As Kolodny (2003, 157) notes, the idea of saving your wife because of your
relationship can seem to “[reflect] a kind of alienation from one’s be-
loved,” a failure to act out of genuine concern for her. There is something
oddly and perhaps disturbingly self-regarding in deciding to save your
wife because of how she relates to you. According to Philip Pettit (1997,
155): “The characteristic explanation of a lover’s behaviour towards a
beloved is not the recognition of the fact of loving her, nor the recog-
nition of the fact of any related features, but the fact of loving itself.”
Unlike Pettit, I do not say that being moved by the belief that you love
someone, or that she bears some relationship to you, is incompatible with
acting from love. Love, and the commitment to act on it, may be genuine
but conditional. What I say is that it is rational to act from love without
being moved by such beliefs: to act for the sake of someone you love,
moved by a concern for her that is not contingent on how she relates to
you. That is how I would want my wife to act if she were in your place: to
save me not because I am her husband or because we have the relation-
ship we do, but spontaneously, just because it is me. As Pettit (1997, 158 –
59) argues, the reason for which she acts, in that case, is a singular

266
 Love and the Value of a Life

thought—“Kieran is drowning”—that need not mention my relation-

ship to her.27
The conclusion I am urging is that when you save your wife from
drowning in the case described above, it is not irrational to be moved by
features of the circumstance—that she is in danger, that you have no
special obligation to the others, and so on—that do not include the
fact that she is your wife. When you are moved in this way, your love for
M makes you responsive to her needs, regardless of her relationship with
you. It is in that way unconditional.
In light of this, we can give the further argument promised in
section 1, against the need for relationships as reasons for love. Suppose
that we need such reasons. Another’s humanity is not enough to justify
love. Instead, what makes it rational to love someone is a belief about their
relationship with you. In general, however, when an attitude is rational
only because you believe that p, it is a condition of acting rationally on that
attitude that you be moved, in part, by that belief. (Such motivation need
not be self-conscious: you need not think of the belief as a ground on
which you are acting, but your action depends on it.) It follows from these
assumptions that it is irrational to save your wife, acting from love, unless
you are moved by a belief about your relationship. But, as I have argued,
that is not the case. Since it is rational to act from love, when you save your
wife, without being moved by the belief that you love her or that she is
your wife, we do not need relationships as reasons for love.28
The emerging picture is this. As I argued in section 1, and as I have
argued again, while relationships may be insistent reasons for love, such
reasons are not required. Nor is love justified by a person’s distinctive
merits, past or present. The fact of their humanity is sufficient. I need no
more reason to love my wife than that she is another human being. What
is more, it is not irrational to act with partiality to those you love, moved
by features of the circumstance apart from their relationship to you.

27. See also Keller 2013, 92– 96. The point here is not just that acting from love is
distinct from acting for the sake of love, where that means acting in order to develop,
sustain, or strengthen love. (Compare Stocker 1981, 754 –55, on acting for the sake of
friendship.) That fact could be absorbed by denying that the reasons for which you act
when you act from love are teleological, while insisting that they cite your relationship
(Stocker 1981, 763). Nor is it that your relationship serves as an “enabling condition” for
reasons, not a reason to act (Keller 2013, 134– 36). The point is that your relationship
need not be part of what moves you at all.
28. A similar argument works against the quality theory, assuming that it is rational to
save your wife without regard for her distinctive merits.

267
 KIERAN SETIYA

The love that is justified by your wife’s humanity involves the disposition
to act in this way.
My argument rests on a final claim, which is harder to articulate
and defend. Perhaps the best way to approach it is by comparison. The
position I accept is in one way less radical than Pettit’s. He insists that, in
acting from love, one is not moved by a belief about love or one’s relation
to its object. I have argued only that one need not be. But Pettit (1997,
156) does not deny that when you act from love, love plays a role in
justifying what you do: “A lover in the proper sense will have no need
for . . . reflective thoughts [about love or relationships] in order to be
motivated to pursue the beloved’s good. And this, despite the fact that
the lover’s behaviour may be justified by the fact that they love the person
favoured.”29
The role of love here may be compared with the role of desire in a
common conception of instrumental reasoning. When you act on the
desire for E, you do not act on the belief that you desire E. Instead you are
moved directly by the belief that M is a means to E and by the desire for E
itself. Still, the desire plays a justifying role. It would not be rational to act
on your belief about means if you lacked the relevant desire. Nor is it
rational to act on the mere belief that you desire E, and the belief that M is
a means to E, since the first belief could be false. On this conception,
desire plays a “background” role in practical thought. The rationality of
being moved by means-end beliefs is conditional on the desire for E, but
that desire does not appear in the content of a belief by which you
are moved.30 For Pettit, love plays a similar role. Its presence makes it
rational to favor the interests of those you love, as when you save your wife
from drowning, but it need not figure in the content of a belief on which
you act.
Although I agree that beliefs about love need not be involved in
acting from love, I do not see how love could do the work that Pettit
describes. How could love make it rational to act on another’s interests
in the way you do when you save your wife if it would otherwise be
irrational to do so? Take the disposition you manifest when you save
your wife from drowning at the cost of three lives, a disposition to give
more weight to her needs than to those of other people, limited by special
obligations, unusual consequences, and extremes of virtue and vice. This
disposition leads you to save her in light of beliefs about the circumstance

29. See also Pettit 1997, 159.

30. This account draws on Pettit and Smith 1990.

268
 Love and the Value of a Life

that do not include the belief that she is your wife. Is the rationality
of being so disposed conditional on love? Unlike Pettit, I believe that
it is not.
We can defend this claim, and see more clearly what it means, if
we ask how love relates to the disposition in question. Does love involve
this disposition, or is it a state of mind—a sentiment or attitude—distinct
from it? On the most plausible conception, being disposed to give priority
to her interests partly constitutes your love for your wife. Unlike desire,
which along with belief is an “input” to means-end reasoning (on the
conception sketched above), love is not an input to a pattern of thought
in which you are also moved by beliefs about the circumstance. Rather, it
consists, in part, in your propensity to be so moved. If this propensity is
irrational, it is irrational to love your wife. One way to put this point is that
the structure of love and its relation to practical thought are more like
that of a character trait than of an attitude like desire.31 Virtues and vices
are defined by the grounds to which they make one sensitive, the facts one
is likely to detect or comprehend about one’s circumstance, how one
responds to them in agency and affect. Traits of character are not material
for practical reasoning, premises for practical inference of other kinds,
but reflect or control the shape of one’s practical thought.32 The same is
true of love.
No doubt, in ordinary cases, there is more to love than the dispo-
sition to be moved by another’s needs in the way you are when you save
your wife. Love has affective dimensions and egoistic ones: it influences
how we feel and what we want for ourselves. These aspects vary widely.
As we saw in section 1, love comes in many shapes and forms. It does not
follow from this that the rationality of the dispositions involved in love
turns on the presence of the others, that these dispositions are rational
only when they come together. Suppose it is true that, in loving your wife,
you are disposed to feel about her in certain ways, to desire her company,
and so on. The question is whether it is rational to give priority to her life
over the lives of three, when the circumstance is otherwise normal—no
special obligations, unusual consequences, extremes of virtue or vice—
only because these affective dimensions are in place. This is what I deny.
It would not destroy the rationality of your action, when you dive into
the water and save your wife, if your love for her had lost its affective force,
if it had become increasingly dispassionate and narrow in scope, no more

31. Compare Pettit 1997, 154 –55.

32. I defend this conception of character at greater length in Setiya 2007, 70 –75.

269
 KIERAN SETIYA

than a deep sensitivity to her needs, constrained by duty. Nor do I see how
being disposed to feel a certain way about your wife could make this
sensitivity rational—could make it rational to save her life at the cost of
three others—if it would otherwise be wrong. What justifies you in saving
your wife is not your emotional state or your desire to be with her. I
conclude that the disposition to give priority to her needs is categorically
rational, in this sense: its rationality does not depend on any further
attitude or any aspect of love beside itself.
In this section, I have argued for two principal claims. First, that
what is justified by another’s humanity is not just love, but love of a kind
that involves partiality in action. When you save your wife from drowning
in the case we have discussed, it is rational to act from love on the basis of
beliefs about the circumstance that fail to register the fact that she is your
wife. Second, that the disposition you manifest in doing so would be
rational even in the absence of other dispositions that constitute love.
Its rationality does not turn on its affective environment, or on emotional
responses apart from the response involved in the disposition itself. This
is what I mean when I say, in contrast with Pettit, that the disposition to
save your wife is not justified, or made rational, by love.

3. On Whether Love Counts

It is time to acknowledge a surprising consequence of this account. Go
back to our example. Three strangers are about to drown; your wife, M, is
drowning, too. But you cannot save all four. If you rescue your wife, the
strangers will drown. The circumstance is otherwise unexceptional. You
have no special obligation to the strangers; nor do they have a right to aid
beyond the rights of any human being. The life or death of those involved
would have no unusual consequences, good or ill. Nor are there pressing
questions of virtue, vice, and moral desert. In this circumstance, it is
permissible to save your wife: you would be justified in doing so, even at
the cost of three lives. So far, many would agree. But now ask why
you would be justified. It is striking that, according to the argument of
section 2, the rationality of your action does not depend on your relation-
ship to M. You are rationally disposed to save her when you recognize
features of the circumstance—that she is in need, that there are three
strangers, no special obligations, and so on—apart from the fact that she
is your wife. This kind of partiality is justified by the fact that she is another
human being. What is more, the rationality of your action does not
depend on aspects of love that go beyond the disposition you manifest

270
 Love and the Value of a Life

in saving M, a disposition to give priority to her needs. It is rational to be so

disposed even if you lack the affective or other dimensions of love. It
follows that it would be rational for a stranger, S, confronted with the
same situation—able to save M or three others, but not all four—to share
your disposition: to give more weight to her interests than to those of
other people, and thus to save her on the basis of beliefs about the cir-
cumstance that ignore M’s relationship to him. That M is another
human being is enough to justify this disposition. In general, though,
when the disposition to f in light of certain beliefs is practically rational,
and those beliefs are true, it is not the case that one should not f. This is
not to say that one should do so, since it might also be rational to refrain.
But it does mean that, in saving M instead of the others, S does not fail to
act as he should, despite the fact that all four of them are strangers to him.
Like you, S has sufficient reason to save M from drowning, at the cost of
three lives.
For the most part, this argument consists in the direct application
of claims from section 2. There is one step that may give pause. Is it really
inconsistent with the rationality of being disposed to save M, and the
absence of false beliefs, that S should refrain from doing so? Might the
fact that M is a stranger be a decisive reason to let her drown, in order to
save the three? The answer is that if this were a decisive reason not to save
M, it would not be rational for you to be disposed to save your wife on the
basis of beliefs that ignore her relationship to you. Where the fact that p
is a decisive reason not to f, it is irrational to be disposed to f on the basis
of beliefs that are consistent with believing that p. True, there are cases
in which it is rational to f, given what you know, although the fact that p is
a decisive reason not to f. But these are cases in which the rationality of
your action turns on ignorance of whether p or on the belief that not-p.
Neither condition is met by the disposition you manifest in saving your
wife. You are not ignorant of your relationship nor are you moved by the
belief that she is your wife. It follows that being a stranger to M is not a
decisive reason for S to let her drown, even though the cost of saving her is
three lives. It is inconsistent with the rationality of being disposed to save
M, and the absence of false beliefs, that S should refrain from doing so.
We can state the reasoning here informally. You are justified in
saving M by facts about the circumstance apart from your relationship
with her: that she is in need, that you have no special obligation to the
others, and so on. S shares these reasons with you. Their force is not
contingent on the relationship; nor does the rationality of acting on
them depend on aspects of love beside the disposition to do so. Thus,

271
 KIERAN SETIYA

they are sufficient to justify S in saving M. If he were to save her for these
reasons, he would be acting as he should.
We have derived from the account of love in section 2 a form of
skepticism about saving the greater number. In the case of S, who can save
one or three, most believe that he should save three. Like Elizabeth
Anscombe (1967) and John Taurek (1977), we have been led to deny
this.33 There may be circumstances in which S should save the others, as
when they have a special claim or right to aid and he would wrong them by
saving M.34 Perhaps he should save them when the consequences of not
doing so would be awful, or where M is morally depraved. But in the
situation we are considering, none of this obtains. In terms of our discus-
sion: where you would be justified in saving your wife by facts about the
circumstance apart from her relationship with you, S would be justified
in saving her, too.
The denial that S should save the greater number in cases of this
kind is one element of what has come to be known as “numbers skepti-
cism.” But that term suggests more radical claims, which I do not accept.
To begin with, it is consistent with my view that S would act both rationally
and permissibly if he were to save the group of three, simply because there
are more of them. The conclusion of my argument is that he has sufficient
reason to save M, not that his reason is decisive. While it is rational to love
M, as another human being, it is equally rational to love the others. And
for all I have said, it is rational to love none, and to give equal weight to the
interests of all. A stranger whose practical thought is in this way rational
would act so as to save the greater number, on the ground that more
needs will then be met. The sense in which the numbers do not count
is given by denying that they must, not by the claim that is irrational to
care how many lives you save. Here I disagree with Taurek (1977, 310),
who makes the stronger claim: “The numbers in themselves simply do not
count for me. I think they should not count for any of us.”
Taurek (1977, 303, 306) goes on to say that, faced with a choice
between saving one and saving three, he would elect to toss a coin. If it
comes up heads, he would save one; if it comes up tails, he would save
three. Some find in these remarks a doctrine of fairness that requires us

33. Taurek’s own examples involve the distribution of medicine that one owns, and so
introduce complexities about the strength of property rights; see Kavka 1979, 288 –91.
Our case avoids this.
34. See Taurek 1977, 298, 300– 301, 310– 13; the absence of such constraints is the
principal theme of Anscombe 1967.

272
 Love and the Value of a Life

to give equal chances to the four who could be saved. That is not how I
read Taurek. He is describing what he would do, not prescribing a prin-
ciple for others. But whatever Taurek’s view, the principle of equal
chances does not follow from the argument above. The point of this
argument is that just as you could simply save your wife, no coin toss
required, so S could simply save M.
More generally, the idea of numbers skepticism suggests complete
antipathy to aggregation and to arguments that appeal to the “imperson-
al good.” This is certainly part of Taurek’s stance.35 And it is emphasized
in recent work, which draws on Scanlon’s (1998, 229 – 41) contractualism.
Like Taurek, the contractualist is skeptical of aggregation: the principles
describing what we owe to each other must be justified to each individual,
a justification that cannot aggregate claims. It is a puzzle for contractual-
ism how to avoid Taurek’s verdict about saving the greater number.36
Whatever the outcome of that debate, the argument above does not
depend on it. This argument does not appeal to the danger of illicit
aggregation, to the demands of justification to all, or to doubts about
the intelligibility of the impersonal good. It is perfectly consistent with my
view that there are conditions in which the numbers should count, in
which it would be wrong for S to save your wife because of facts about the
aggregate or impersonal good. He may have sufficient reason to save her
at the cost of three lives, but not at the cost of a hundred, or a thousand, or
a million. What follows is only that, where S would have decisive reason to
let M drown, in order to save the others, you would not be justified in
saving her by facts about the circumstance apart from her relationship to
you. Where you would be justified in saving M, but S would not, the fact
that she is your wife must play a justifying role. (Likewise, it is the fact that
M is your wife that explains why you are not merely justified in saving her,
but ought to do so; here you differ from S.)
Finally, it does not follow from my argument that S acts rationally
in saving M, no matter what his reasons. What would be rational is for him
to save her because he shares your disposition, the one that was specified
in section 2. It would not be rational for S to save M because she is to the
left of him, while the others are to the right, or to save her just because he
feels like it. If the disposition he manifests in saving M is the disposition to
satisfy his whims—if his reasons are irrelevant or worse—it is irrational

35. See Taurek 1977, 303– 10.

36. For a critical appraisal, see Otsuka 2006.

273
 KIERAN SETIYA

for S to save your wife, leaving three to drown.37 Nor does it follow that the
disposition to respond to her needs could rationally vanish as quickly as it
came. There may be insistent reasons for love to persist, at least to some
degree. If he saves your wife from drowning, S cannot regard her as a
stranger, though the contours of his commitment are unclear.
The fact that S could save your wife irrationally prompts a further
question, about what it is permissible to do. Presumably, if he acts on
grounds that justify his action, S acts permissibly, as you act permissibly in
saving your wife. But suppose his motivation is irrational, since he acts on
a whim or for some illicit reason? Some deny that the grounds on which
we act are relevant to the permissibility of our actions, except through
their predictive or expressive significance; what matters is the existence of
sufficient reason.38 Others suggest that while sufficient reason may ren-
der a type of action permissible in a circumstance, whether one acts
permissibly turns on one’s reasons for acting.39 I won’t attempt to settle
this dispute. It is enough to deny that S should save the greater number
and to hold that he acts permissibly if he saves M on suitable grounds.
Although my conclusion differs from Taurek’s, my argument fol-
lows a thread in his discussion. Considering “pairwise comparison” —the
idea that when aggregation is not at stake, we should compare the size of
harms—Taurek (1977, 301) writes as follows:
If it would be morally permissible for B to choose to spare himself a certain
loss, H, instead of sparing another person, C, a loss, H 0 , in a situation
where he cannot spare C and himself as well, then it must be permissible
for someone else, not under any relevant special obligations to the con-
trary, to take B’s perspective, that is, to choose to secure the outcome most
favorable to B instead of the outcome most favorable to C, if he cannot
secure what would be best for each.40

The same point could made with numbers, as when M could save her own
life or the lives of three. To paraphrase Taurek: if M is not required to

37. Does it matter to the rationality of his action why S is disposed to give priority to
M? Only if his reason affects the content of his disposition, and so the reason for which he
acts when he saves her from drowning. There are delicate questions here about when and
why a reason is ineligible that I won’t try to resolve.
38. This is the view proposed in Scanlon 2008, chaps. 1 –2, following Thomson 1991,
292 –96; Thomson 1999, 514 –18. These discussions are bound up with more specific
issues, about the doctrine of double effect, that we can set aside.
39. For a defense of this idea, see Hanser 2005.
40. Because he rejects pairwise comparison, Taurek is immune to the intransitivity
argument of Otsuka 2004, 421 –26.

274
 Love and the Value of a Life

sacrifice herself to save the others, it must be permissible for someone

else, not under any special obligations to the contrary, to take M’s per-
spective, that is, to save her instead of three if he cannot save them all.
Taurek’s principle can seem obtuse. Why can’t the fact that M would be
saving herself be morally relevant? What about the idea of an “agent-
relative permission” that allows M to refuse this sacrifice without permit-
ting a stranger to refuse it on her, and others’, behalf ?41 But our reflec-
tions suggest that Taurek is right. Shifting from what is permissible to
what is rational, the point is that it is rational for anyone to love M, and so
to “take her perspective,” simply because she is another human being. If
this is rational, it is rational to take M’s needs as reasons of special weight,
and therefore rational to save her in the circumstance described.
At the end of section 1, I said that my conception of love involves a
picture of the value of a life. The picture is one in which every human
being is worthy of love, so that it would be rational for anyone to love
them. The subsequent arguments spell out part of what this picture might
involve. We can come to grips with it by asking: From the point of view of
a stranger, do the needs of every human being count the same? Are the
reasons that derive from them equally strong? If the answer were yes, it
would be irrational for a stranger to give more weight to the needs of M
than to those of anyone else.42 According to my argument, however, this is
not the case: it would be rational for S to give more weight to M. Does it
follow that M matters more? S may be moved more strongly by her needs,
but do they provide him with stronger reasons in the normative sense? If
the answer were yes, it would be irrational for S to give as much weight to
the interests of others. And again, that is not the case: it would be rational
for S to respond in the same way to the interests of all concerned. It
follows that, from the point of view of a stranger, the reasons that derive
from the needs of different people are not equal in strength, nor are
some stronger than others. Instead, such reasons are incomparable.43
This is consistent with the claim that everyone matters equally, in a sense:
no one matters more than anyone else, and each is equally deserving of
concern. Still, there is a radical consequence for morality. If every human
being is worthy of love, and it is rational to treat them with partiality, then

41. This objection is forcefully pressed in Parfit 1978 and Kavka 1979, 286– 88.
42. One should be moved to the same degree by reasons of equal weight, and more
decisively by stronger reasons. I defend and elaborate these claims in Setiya 2007, 12 –13,
though the terminology differs.
43. I explore the prospect of incomparable reasons in Setiya 2007, 77– 79.

275
 KIERAN SETIYA

Taurek is right, at least to this extent: we are justified in saving one when
we could save three. This fact does not reflect the value of relationships,
or an agent-relative reason, but the irreplaceable worth of human life.

4. Conclusion
At the beginning of this essay, I said that its method was to argue from
attractive premises to puzzling conclusions. The premises are claims
about the nature of love, and of its reasons. The conclusion is close to
Taurek’s. Faced with the choice between saving one and saving three, all
of whom are strangers to you, it is not the case that you should save three.
If that conclusion is absurd, as many believe, one of the premises of the
argument must be false.
We can set out the steps, schematically, as follows. First, another’s
humanity is sufficient reason for love; you need not bear a special
relationship to them. Second, what is justified by another’s humanity is
a disposition to give priority to their needs in the absence of special
obligations, unusual consequences, or extremes of virtue or vice. The
rationality of being so disposed does not depend on aspects of love beside
the disposition itself. Finally, when the disposition to f in light of certain
beliefs is practically rational, and those beliefs are true, it is not the case
that one should not f. It follows that a stranger has sufficient reason to
save your wife, even at the cost of three lives.
Although I have argued for the first two steps, in sections 1 and 2,
I would not claim that my case for them is irresistible. Some philosophical
arguments are designed to compel assent, others to explain why someone
might believe that p without aiming to convert the skeptic. My argument
falls toward the latter end of the spectrum. I hope that it will move those
who feel the pull of Taurek’s view but are uncertain of his reasons. It will
not persuade those who think he is just confused, or that it is evident that
the stranger must save three. Now that they know where the argument is
going, skeptics will deny that it is rational to love just anyone; that it is
rational to act with partiality to those you love, except on the basis of their
relationship with you; or that it is rational to be so disposed if one lacks the
affective and other dispositions characteristic of love.
I don’t have more to say about this now. Even though it can be
resisted, I think there is value in setting out the argument above. What I
want to do in closing is to reconsider the absurdity of Taurek’s view. How
great an objection is it to the picture of love that emerged from sections 1
and 2 if it denies that one should save the greater number? Is there any

276
 Love and the Value of a Life

way to make this plausible, to weaken our attachment to numbers in cases

of the relevant kind? I think we can learn here from an aspect of Taurek’s
essay that is insufficiently discussed.
Although his argument has prompted much reflection on our
duties to friends and loved ones, Taurek does not in fact appeal, as I
have done, to the case in which you save your wife. The example he
gives involves you and David; and while he refers to David as a friend,
he does not have in mind a deep commitment. David is a mere acquaint-
ance, someone you know and like, not someone with whom you spend
much time or with whom you share your life.44 Your relationship with
him is superficial: it is not a source of obligation (Taurek 1977, 296 – 97).
Think of David as the person who makes your coffee at Starbucks, or
who drives the bus you take to work, someone you like but with whom
you exchange no more than friendly words and a casual smile. Still, it
seems to me, as it does to Taurek, that you have sufficient reason to save
him, even at the cost of several lives. Try to picture it. On one side, three
anonymous faces; on the other, David, whose face you know. You are not
obliged to save him. Should you let him drown in order to save the others?
If the circumstance is otherwise normal—no special obligations, unusual
consequences, or extremes of virtue or vice—I don’t see that you
should.45 As Taurek argues, however, if you can save David, whom you
know and like, against the claims of three, “the moral force of such
[claims] must be feeble indeed” (Taurek 1977, 298). As I would put it,
such claims are incomparable. It is not that there is a reason to favor those
you know and like over those you have never met. It is rather that, in the
absence of special conditions, there is sufficient reason to give priority to
the needs of any human being. This is what you do for David, not irra-
tionally, and what you might do for anyone else.
None of this is meant as a further argument. The point of
approaching the case in this way is to shift our perception of it, not to
prove that my conclusion holds. There is a perspective from which the
denial that one should save the greater number seems not callous or
indifferent to the value of human life but to express an intelligible con-
ception of that value. This conception is one in which love is, as Murdoch
(1970, 2) claimed, “a central concept in morals.” It is part of the value of a
life that every human being is worthy of love: it would be rational for

44. David is called a “friend” in Taurek 1977, 296– 97, 299; and “someone [you] know
and like” in Taurek 1977, 295– 98.
45. For the opposing claim, see Parfit 1978, 289 –90.

277
 KIERAN SETIYA

anyone to treat them with the partiality we devote to our friends and
partners, parents and children. This fact has practical implications. We
often have the opportunity to help others at little cost to ourselves, and
the obligation to do so. My conclusion does not conflict with this. But it
does suggest that, in deciding who to help, we need not give weight to
numbers and relationships alone. Even if your money would save more
lives in distant countries, it does not follow that you should give to Oxfam
rather than to local charities or individuals in need, not because there is
insistent reason to favor those who are close to home, but because it is
rational to give priority to anyone.46
Finally, we can ask how this conception of human value bears on
the commandment to love one’s neighbor as oneself. Is there any point
to this formula once we reject the severe interpretation on which it
demands complete impartiality?47 Why not talk instead about reverence
or respect? In my view, these questions are premature. To answer them,
we would need to say more about the nature of self-love as the model
for love of one’s neighbor. And we would need to explore the argument
sketched toward the end of section 1. According to that argument, what
is required of us in relation to every human being is not love of the kind
you have for your wife or child, but a species of respect. It deserves to be
called “love,” first, because the demand for it is explained by the fact
that we are worthy of love; second, because, like love, it cares about the
other for her own sake, not in relation to anything else; and third,

46. There are complications here. First, in order to be rational when you save the few,
you must act on the disposition to give priority to their needs. Second, if the wealth that
enables you to help other people depends on social injustice, they may have claims on you
that distinguish this case from the one we have discussed. Like Anscombe (1967), I have
assumed that there are no special obligations to consider. For much the same reason,
Taurek’s conclusion does not apply to questions of social policy, where claims of justice in
distribution are impossible to avoid. See the discussion of the coast guard in Taurek 1977,
310 –13. This is one reason why it matters whether our obligations to those in need are
obligations of justice or not.
47. Although it has been put forward—as in Cottingham 1983; Reeder 1998—this
interpretation is not orthodox. In the Jewish tradition, “love your neighbor” is often
equated with the golden rule: “as yourself ” means “as you wish to be loved by them”
(see Goodman 2008, 12– 15; and for doubts about this equation, Outka 1972, 292 –94;
Goodman 2008, 152; Wolterstorff 2011, 98– 99). For Aquinas, “as” denotes the manner,
not the degree of love: to love your neighbor as yourself is not to give your interests equal
weight (see Oderberg 2008, sec. 3). And in his classic study of agape, Gene Outka never
imagines that love of neighbor might prohibit partiality. It restricts the scope of our
special relationships—you cannot harm an innocent stranger in order to help your
friends—but it neither explains nor precludes them (Outka 1972, 268– 74).

278
 Love and the Value of a Life

because the reason for it is the same as the reason for love. You should
regard your neighbor as worthy of love, and so as claiming respect, on
the same ground that justifies your self-love: not the fact that you are
you, but the fact of your common humanity. I do not know whether
anything like this could be true, or what it would say about the character
of respect. Nor is it essential to the argument of this essay. But the
prospect of such insights, however dim, is part of why I am willing to
take seriously a picture of love that leads me to deny that numbers
count, at least in cases of a certain kind, a position I hope seems less
absurd than it did before.

References
Anscombe, G. E. M. 1967. “Who Is Wronged?” Oxford Review 5: 16–17.
Cottingham, J. 1983. “Ethics and Impartiality.” Philosophical Studies 43: 83–99.
Frankfurt, H. 2004. The Reasons of Love. Princeton: Princeton University Press.
Freud, S. 2010 [1930]. Civilization and Its Discontents, trans. J. Strachey. New York:
Norton.
Goethe, J. W. 2012 [1774]. The Sorrows of Young Werther, trans. D. Constantine.
Oxford: Oxford University Press.
Goodman, L. E. 2008. Love Thy Neighbor as Thyself. Oxford: Oxford University
Press.
Hanser, M. 2005. “Permissibility and Practical Inference.” Ethics 115: 443–70.
Herman, B. 1993. The Practice of Moral Judgment. Cambridge, MA: Harvard Uni-
versity Press.
Jollimore, T. 2011. Love’s Vision. Princeton: Princeton University Press.
Kagan, S. 1989. The Limits of Morality. Oxford: Oxford University Press.
Kavka, G. 1979. “The Numbers Should Count.” Philosophical Studies 36: 285 –94.
Keller, S. 2013. Partiality. Princeton: Princeton University Press.
Kolodny, N. 2003. “Love as Valuing a Relationship.” Philosophical Review 112:
135– 89.
Murdoch, I. 1970. The Sovereignty of Good. London: Routledge.
Murphy, L. 2000. Moral Demands in Nonideal Theory. Oxford: Oxford University
Press.
Nozick, R. 1989. The Examined Life. New York: Simon and Schuster.
Oderberg, D. S. 2008. “Self-Love, Love of Neighbour, and Impartiality.” In The
Moral Life: Essays in Honour of John Cottingham, ed. N. Athanassoulis and
S. Vice, 58 –84. Basingstoke: Palgrave Macmillan.
Otsuka, M. 2004. “Skepticism about Saving the Greater Number.” Philosophy and
Public Affairs 32: 413 – 26.
———. 2006. “Saving Lives. Moral Theory, and the Claims of Individuals.” Phi-
losophy and Public Affairs 34: 109– 35.

279
 KIERAN SETIYA

Outka, G. 1972. Agape: An Ethical Analysis. New Haven: Yale University Press.
Parfit, D. 1978. “Innumerate Ethics.” Philosophy and Public Affairs 7: 285–301.
Pettit, P. 1997. “Love and Its Place in Moral Discourse.” In Love Analyzed,
ed. R. Lamb, 153– 63. Boulder, CO: Westview Press.
Pettit, P., and M. Smith. 1990. “Backgrounding Desire.” Philosophical Review 99:
565– 92.
Raz, J. 1999 [1997]. “Incommensurability and Agency.” In Engaging Reason,
46– 66. Oxford: Oxford University Press.
Reeder, J. P. 1998. “Extensive Benevolence.” Journal of Religious Ethics 26: 47 –70.
Scanlon, T. M. 1998. What We Owe to Each Other. Cambridge, MA: Harvard Uni-
versity Press.
———. 2008. Moral Dimensions: Permissibility, Meaning, Blame. Cambridge, MA:
Harvard University Press.
Setiya, K. 2007. Reasons without Rationalism. Princeton: Princeton University
Press.
———. 2012. “Internal Reasons.” In Internal Reasons: Contemporary Readings,
ed. K. Setiya and H. Paakkunainen, 1– 34. Cambridge, MA: MIT Press.
Stendhal. 1975 [1822]. Love, trans. G. Sale and S. Sale. London: Penguin.
Stocker, M. 1981. “Values and Purposes: The Limits of Teleology and the Ends of
Friendship.” Journal of Philosophy 78: 747– 65.
Stump, E. 2006. “Love, by All Accounts.” Proceedings and Addresses of the American
Philosophical Association 80: 25 – 43.
Taurek, J. 1977. “Should the Numbers Count?” Philosophy and Public Affairs 6:
293– 316.
Thomson, J. J. 1991. “Self-Defense.” Philosophy and Public Affairs 20: 283 –310.
———. 1999. “Physician-Assisted Suicide: Two Moral Arguments.” Ethics 109:
497– 518.
Velleman, J. D. 1999. “Love as a Moral Emotion.” Ethics 109: 338–74.
Williams, B. 1981 [1976]. “Persons, Character, and Morality.” In Moral Luck,
1– 19. Cambridge: Cambridge University Press.
Wolterstorff, N. 2011. Justice in Love. Grand Rapids, MI: Eerdmans Publishing.

280