Before I start, it is helpful to briefly note what love cannot be if we take the phenomenon of the beloved’s irreplaceability seriously: love cannot be an attitude that is fully explained and justified by value judgments about the beloved’s beauty, wit, or other such valuable qualities. For if to love another person means no more than judging them to be beautiful or funny, say, and desiring and caring for them as a result, then why shouldn’t lovers be willing, in principle, to replace their beloveds for persons with relevantly similar or better qualities, and desire and care for them instead? In fact, according to the quality theory, it seems that lovers should be thus willing, in both an expectant and a rational sense of ‘should’. But (analogously to the compensation-scenario above) we neither expect lovers to be willing to replace their beloveds, nor judge them to be irrational if they are not. In fact, as we saw in the previous section, we expect lovers not to be thus willing and if they are, we either doubt that they are actual lovers or we doubt whether they are in their right minds. The quality theory not only contradicts but implies the polar opposite of what we have said so far. So we should rule it out.Footnote 8
One response might be to distinguish between deontic and non-deontic reasons in this context.Footnote 9 Deontic reasons for some attitude x both justify and rationally require x, whereas non-deontic reasons for x only justify but do not rationally require x. If the beloved’s valuable qualities only provide the lover with non-deontic reasons for love, they justify the lover’s love without rendering the absence of love irrational. So if, in a relevantly similar situation, the lover did not respond with love, the lover would not thereby be irrational. This way, we can explain why we should not be surprised and why it would not be irrational, if lovers refuse to replace their beloveds on the basis of comparative value judgments (despite their love being based on and justified by value judgments that are in principle open to comparison).
The distinction between deontic and non-deontic reasons does not solve the irreplaceability problem for the quality theory, however. On this picture, while there would not be anything wrong with a lover refusing to replace their beloved in the above fashion, there would also not be anything wrong with a lover who did not refuse. Both options are equally rational and compatible with love here. But the second option is not in fact compatible with love, so the irreplaceability problem remains. The three accounts of love I discuss in the following are alternatives to the quality theory that try to address this problem.
I turn to volitional accounts first, according to which love does not imply any value judgments but is a purely volitional phenomenon. Harry Frankfurt (2004, 2006) offers the most developed of these accounts, so I focus on his. But my main point applies to volitional theories in general.Footnote 10
According to Frankfurt, love is unconditional care for the beloved, and value judgments about the beloved are neither necessary nor sufficient for this kind of care. Love has causes, but no reasons, in the sense that from the lover’s point of view, there does not have to be anything about the beloved that speaks in favour of and justifies her love. Instead of having reasons, love generates practical reasons, insofar as it determines the lover’s final ends: to care for something unconditionally means to value its continued existence and wellbeing as a final end, rather than instrumentally or, more generally, non-finally (where ‘valuing something’ just means being disposed to act on its behalf or to promote it in certain ways, rather than judging it to be valuable in some respect).Footnote 11
This also provides an explanation for the beloved’s irreplaceability. For whatever we value as a final end is irreplaceable to us – in fact, this is partly what it means to be a final end. Consider non-final ends first. Among the non-final ends are means to other ends, and something we value as a mere means is valuable to us only insofar as it produces the end.Footnote 12 It does not matter to us as the particular thing it is. The same is true of other non-final ends: we do not value them as the particular things they are, but only relative to other ends. Hence, it makes sense for us to be willing to replace them for anything that stands in the same relation to these other ends (for example something that produces them just as or more effectively). But something we value as a final end, we do value as the particular thing it is and not just as a placeholder for something else. And this entails that we take it to be irreplaceable: nothing but this particular thing will do for us.
So Frankfurt’s account meets condition one: it can explain the subjective aspect of the beloved’s irreplaceability.Footnote 13 However, it does not have the resources to explain the moral aspect of the phenomenon - it does not meet condition two. Think about the parents in the above scenario again who, in a moment of confusion, agree to replace their child even though they love their child. According to Frankfurt’s account, by accepting the replacement deal, the parents would be disregarding their own final ends. To that extent, they would be practically irrational. But at most this amounts to some kind of ‘wronging’ of themselves. Why should it make them feel guilty towards their child? Why would they feel as if they had done something wrong to their child by relating to her as if she were replaceable?
Since their love is not a reflection on how valuable they believe their child to be - since it is in fact compatible with their love that they think of their child as objectively value-less - it seems that at least qua lovers, the only reason they have for relating to their child as irreplaceable in thought and action is the fact that they happen to find themselves in this particular psychological state, a state which makes them want to do so. In other words, there is no further normative standard that comes along with loving their child that could explain their guilt towards her.Footnote 14
I now turn to accounts that keep some key ideas of the quality theory. According to relational accounts, love does imply value judgments, but the judgments in question are about the beloved’s relationship with the lover and their particular ways of interacting. Niko Kolodny (2003) argues that to love a person is to believe that one has a valuable type of relationship with them, for instance. More recent examples are Katy Abramson’s and Adam Leite’s (2011) view that love is a reactive emotion, or Hitchem Naar’s (2017) view of love as based on subject-relative reasons.
In all of these approaches, lovers would not be willing to replace their beloveds for persons with similar or even better ‘intrinsic’ qualities, because their love is a response to what has happened between them and their beloved in particular. The properties relevant for explaining and potentially justifying love are these relational ones. In the following, I argue that such accounts have trouble meeting both condition one and two. With respect to condition two - the moral aspect - the problem is not that they do not introduce any normative standards at all, but that they do not introduce the right kind of standard.
I start with Kolodny’s relationship theory. Insofar as relationships are constituted by historical events connecting the relata in question, they are unique to the relata. But we have some kind of unique relationship with many (if not with all) persons and, clearly, not all of these unique relationships explain and / or justify love. Hence, according to Kolodny (2003: 150-1), only particular types of valuable relationships are relevant - the ones that count as romantic relationships, friendships, familial relationships, etc. But if these types of values end up doing the explanatory work, why should lovers not be willing to replace their beloveds for persons with whom they share equally or more valuable relationships?Footnote 15
Once more, imagine parents who lose a child. If they have another child at some later point, their relationship to this new child will be equally valuable to the relationship they had with the child they lost. Still, as mentioned in Sect. 1, the second child will not be a compensation for the first one: it is not as if the second child cancels out the loss of the first, in the way a new copy of a book can (sometimes) cancel out the loss of the first. How could Kolodny’s theory explain this? The parents are equally well off with regard to their participation in valuable relationships after they have a second child. So the new relationship should compensate for the loss of the first - and it might in fact do that, but that is not the same as compensating for the loss of their child. This remains unexplained, so condition one is not met.Footnote 16
It is also unclear whether Kolodny’s account meets condition two. Take the parents in the previous example again, who love their child but agree to the replacement deal in a moment of confusion. According to Kolodny, they would deprive themselves and the child of the valuable relationship they share with their child, so they would harm both themselves and the child. However, not every harm constitutes a wrong, so not every harm explains and justifies guilt.
Even if the harm did constitute a wrong, however, for instance because the reasons for inflicting it were insufficient, it is not clear that Kolodny’s theory can give the right answer for why. What distinguishes this case from other cases in which people end valuable relationships for insufficient reasons? It seems that his theory cannot explain what is specifically wrong with treating one’s beloved as replaceable and why lovers qua lovers would feel guilty about this in particular (rather than generally about having ended the relationship with their beloved for insufficient reasons).
Abramson & Leite (2011) and Naar (2017) offer a different approach. According to them, love is (or is at least in some ways analogous to) a ‘reactive attitude’ in Peter Strawson’s (1962) sense of the term.Footnote 17 A paradigmatic reactive attitude is gratitude: we feel and have reason to feel gratitude towards someone if they have done something nice to us, but not if they are just a nice person in general. Similarly, the thought goes, we love and have reason to love another person only if they have acted in certain valuable ways towards us - simply being a person with valuable qualities in general is not enough. So just like gratitude, love is not just a form of valuing of the beloved, but something we in some sense also owe to someone under certain circumstances, or that we should give to them.Footnote 18
It is this second aspect that explains the beloved’s irreplaceability: in this picture, the beloved is irreplaceable to the lover in the same sense the person to whom we owe gratitude is irreplaceable to us. Just as we owe gratitude to particular persons who have treated us well and cannot pay this debt by giving gratitude to someone else, no matter how qualitatively similar this other person is or how well they have treated us, we owe love to particular persons and cannot pay this debt by giving love to someone else instead - again, no matter how qualitatively similar they otherwise are or in what lovable ways they have treated us. (If this other person has treated us in lovable ways, too, then we owe them love, too. But that just means that they are irreplaceable for us, too.) Since what matters in this picture is what actually happened between two persons, the earlier replacement scenarios are ruled out.
Still, I think this does not capture the right sense in which lovers relate to their beloveds as irreplaceable. In Sect. 1, I argued that lovers value their beloveds as irreplaceable: when they lose their beloved, they grieve (in part) because they have lost something that is of irreplaceable value to them. But if the beloved is irreplaceable to the lover in the same sense as someone we owe gratitude to is irreplaceable to us, we cannot explain why losing the beloved injures the lover in this way. Valuing and owing are different relations, and in this account, the latter carries the explanatory weight. The fact that the lover also values the beloved for having treated the lover in lovable ways is neither necessary nor sufficient for explaining the beloved’s irreplaceability here: the owing relation is enough to explain it, and without reference to owing, the account would not fare any better than the initial quality account with respect to explaining irreplaceability. So condition one is not met.
What about condition two? On this picture, if the parents in the earlier example replace Ada for Eva in a moment of confusion, they would fail to give Ada what they owe her – if not love itself (since they still love her), then loving care and attention. This is a reason to feel guilty. But it does not sufficiently explain the parents’ guilt. We can imagine that before the replacement takes place, Ada’s parents come to their senses and reverse their decision. In that case, they would not have stopped giving Ada what they owe her, and yet, they would still feel guilty simply for having related to her as replaceable for a moment.Footnote 19
An objectivist account such as David Velleman’s (1999) explains the beloved’s irreplaceability by reference to the beloved’s objective end-hood. Velleman agrees with Frankfurt that lovers take their beloveds to be ends in themselves. However, unlike Frankfurt, he thinks that they take them to be objective ends in themselves. On his picture, love is grounded in the same value judgment as respect (in the Kantian sense of Achtung) - namely in the judgment that the beloved is a person, in the sense of a bearer of the value of personhood. Persons qua persons are objective ends in themselves and as such, are objectively irreplaceable. Insofar as love is a relation to the beloved as a person, lovers relate to their beloveds as objectively irreplaceable.
This is what distinguishes love and respect from other attitudes: when we relate to others in other ways – as objects of emotions such as joy or fear, say, or as our colleagues, bus drivers, doctors, etc. – we might also relate to them as persons but this is not constitutive of these kinds of relations. By contrast, on Velleman’s account, it is partly constitutive of both love and respect that they relate to their objects as persons.
According to Velleman, the only difference between love and respect is that whereas love is a seeing of the other as a person, respect is just the theoretical acknowledgement of their personhood. ‘Seeing’ is not necessarily meant metaphorically here: even though personhood is not something that could be directly visible, Velleman (1999: 371) takes it to be indirectly visible in the way someone walks and talks, the way they hold themselves, and so on. It also ‘disarms our emotional defences’ (1999: 361) towards the person seen, and so, fittingly, whether or not we actually come to see someone else’s personhood is not under our control (‘fittingly’, as it makes sense that we cannot control when our emotional defences are disarmed). This explains why we do not always love everyone we respect (even though on his picture, it seems that we would have reason to love everyone we respectFootnote 20) and it also explains why, unlike respect, love cannot be morally required.
Velleman’s account captures something important which the previous two did not: it explains why the beloved is irreplaceable not just for the lovers, but also for non-lovers, as outlined in Sect. 1. The beloved’s irreplaceability and the irreplaceability any person has for us qua a person are intimately conceptually connected. It might be somewhat surprising that what explains the beloved’s irreplaceability is a property that the beloved shares not just with some but with all persons. But since this property is responsible for every person being irreplaceable, the earlier replacement scenarios are immediately blocked. For a person to be irreplaceable means that they cannot be replaced by anything or anyone, not even other, equally irreplaceable persons. Respecters relate to others insofar as they irreplaceable, and so do lovers.
The account also meets condition two – the moral aspect –, because on this picture, what makes the beloved irreplaceable is itself also of moral value. In Velleman’s Kantian framework, we ought to relate to objective ends in themselves as irreplaceable, in a moral sense of ‘ought’. If being a lover means actually seeing that the beloved is an objective end in itself and thus objectively irreplaceable, it also implies believing that one ought not to relate to the beloved as replaceable.
The objectivist account depends on a broadly Kantian conception of persons as objective ends in themselves. It is beyond the scope of this paper to develop a proper defence of this conception, and it is not strictly needed for my current purposes.Footnote 21 The fact that a lot of us intuitively agree with this conception, as outlined in the passages in Sect. 1 on other people’s deaths, and as many of our every day practices suggest, provides a sufficient basis for taking the approach seriously.
The problem with Velleman’s account is that it does not meet condition one, i.e. it does not explain the subjective aspect of the beloved’s irreplaceability. If love is a seeing of the beloved as a person, as compared to a mere theoretical acknowledgement of their personhood, we cannot explain why lovers grieve over the loss of their beloved in a way non-lovers who nevertheless respect the beloved do not: the distinction between seeing and theoretically acknowledging someone else’s personhood does not capture this difference. The difference it captures is analogous to the difference between witnessing a car crash and reading about it on the news. The former is a more intense and shocking experience, an experience that might well disarm our emotional defences, to use Velleman’s term. But the grief lovers feel when they lose their beloveds is not just a more intense and shocking form of the experience that the non-lover has. As outlined in the previous section, it is substantively different because it involves a personal injury. It is more as if one would be involved in the crash oneself.
Neither volitional, nor relational, nor objectivist accounts are able to explain both the subjective as well as the moral aspect of the beloved’s irreplaceability. Frankfurt’s account meets condition one, but fails to meet condition two; the relational accounts do not meet either condition satisfactorily; and Velleman’s theory meets condition two, but fails to meet condition one.