Moral Partiality and Duties of Love

: In this paper, I make a case for the view that we have special relationship duties (also known as “associative duties”) that are not identical to or derived from our non-associative impartial moral obligations. I call this view “moral partialism”. On the version of moral partialism I defend, only loving relationships can normatively ground special relationship duties. I propose that for two capable adults to have a loving relationship, they must have mutual non-trivial desires to promote each other’s interests or ﬂourishing and to respect each other’s core values. Along the way, I critically ascertain three alternative accounts of what normatively grounds special relationship duties and argue that my proposed view avoids the problems plaguing the alternatives.


Introduction
It is a fundamental and widely acknowledged truth that we can have duties (or obligations or responsibilities) to people with whom we share a special relationship, duties we do not have to those with whom we do not share a special relationship, e.g., .1,2  For example, Felix's familial relationship with his mother may be the basis of his associative responsibility to visit her in the hospital, Edna's close friendship with Kaleb may be the source of her associative duty to co-sign his mortgage loan, Lilo's sibling relationship with her brother may be the basis of her associative obligation to bail him out of jail, and Homer's parental relationship with his child Eden may be the source of his associative duty to rescue her from a fire before he saves other people's children.
I take "special relationship" to cover any personal association between two people that is not purely transactional (unlike an interaction with a cashier at a supermarket) and is based on a shared history (unlike interactions with strangers).Intimate partner relationships, friendships, familial relationships, and collegial relationships are prime examples of special relationships.I take duties to be pro tanto moral reasons, which we may or may not accept as such.As duties are pro tanto, they can override less weighty duties in particular situations.The weightiest duties are all-things-considered duties.If someone acts on a duty that outweighs a second duty in a particular situation, the less weighty duty retains its normative force.Thus, although violating the less weighty duty is justified and excuses the actor from blame, violating that duty is still wrong.
The view that sharing a special relationship with another person can give rise to duties we would not otherwise have is not a matter of controversy.Rather, the disagreements driving the debate regarding associative duties concern the normative status and ground of these duties. 3Call the view that all associative duties are identical to or derived from non-associative moral duties "moral impartialism", 4 and call the view that not all associative moral duties are identical to or derived from non-associative moral duties "moral partialism". 5To avoid confusion, I shall sometimes refer to associative duties that are not identical to or derived from non-associative duties as "associative duties per se".As moral impartialism holds that all associative duties are identical to or derived from non-associative duties, this view entails that if one duty is weightier than another, this cannot merely be because of some fact about a special relationship we share with another person.Moral partialism denies this. 6,7 It is important to distinguish at the outset two different varieties of moral partialism: minimal and substantive partialism. 8In its minimal form, moral partialism holds that the very existence of a special relationship, such as an intimate partner relationship, a friendship, a familial relationship, or a collegial relationship, engenders associative duties per se. 9 John Cottingham [36] glosses this view as follows: Those picked out for special treatment are specified not in terms of some descriptive (and therefore universalizable) quality or feature that they possess, but in terms of some particular relationship which they have to the agent.Thus, in the fire case, my decision to favour my child is based simply on the fact that she is my daughter: there is a non-eliminably particular, self-referential element in my rationale for selecting this child rather than some other.[36] (pp.[358][359] In its substantive form, moral partialism holds that for us to have associative duties to people with whom we share a special relationship, the relationship must possess intrinsic or relational features that not all special relationships instantiate, and that can justify the associative duties, see, e.g., [5,14,19].10 Thus, sharing a special relationship with another person does not in and of itself engender associative duties.It merits emphasis that substantive moral partialism does not take associative duties to always carry greater weight than non-associative impartial moral duties.Rather, the view merely holds that associative duties can sometimes carry greater weight than non-associative duties.In this paper, I argue that moral impartialism and minimal moral partialism are untenable and develop a variant of substantive moral partialism.

Clarifying Moral Impartialism
Moral impartialists hold that associative duties and rights are identical to or derived from impartial non-associative moral obligations and rights, e.g., [1,12,16,[27][28][29].Importantly, impartialists agree with partialists that special relationships can give rise to duties and rights we would not otherwise have.For example, intimate partners commonly share experiences and entrust each other with sensitive information about themselves (e.g., their successes, failures, ambitions, fears, flaws, and fetishes), which makes them particularly vulnerable to betrayal, denigration, exploitation, or other kinds of emotional and physical harm by their partner [12,[37][38][39]45].Partialists and impartialists agree that because of our special vulnerability to intimate partner harm, our intimate-partner relationships often give rise to associative duties, such as duties of loyalty, trust, and forgiveness, which we do not have to strangers.However, unlike partialists, impartialists hold that all special relationship duties derive from impartial non-associative moral obligations, along with empirical facts about the relevant type of relationship.For example, impartialists might argue that the loyalty and trust duties we owe to our intimate partners derive from the general moral principle that we ought to take special care not to harm particularly vulnerable individuals, along with empirical facts about intimate partner relationships.
The two main classes of requirements of impartial modality that ground special relationship duties, according to moral impartialists, are general and transactional moral obligations. 11General moral obligations are the general requirements of impartial morality owed to the moral community or the state, such as moral obligations not to commit battery, threaten others with a gun, or invade their privacy [46,49,50].The rational default expectation is that as moral agents, we will fulfill our general moral obligations because we stand in reciprocal relations of mutual accountability to each other [49,50].When a wrongdoer thwarts this expectation without a valid excuse, we normally hold the wrongdoer accountable for the transgression, for example, by subjecting them to reactive attitudes, like resentment or indignation [50].
The second main class of requirements of impartial modality that moral impartialists argue can serve as a source of special relationship duties are the transactional moral obligations we acquire during our transactions with others (e.g., when exchanging goods and services) [46,47,51].During our transactions with others, we exchange direct or indirect (implied) speech acts that serve to bind us to each other in transactional obligations as obligees and obligors.For example, when a vassal took an oath of fealty in a feudal society in Medieval Europe, he committed himself to surrender and provide services to the lord, thus binding himself in a transactional obligation to the lord.The vassal simultaneously acquired a claim right to be invested with a fief or tenure at their lord's court in exchange for his servitude.Likewise, if Olga hires Konrad as a general contractor to build her home, she binds herself in a transactional duty to Konrad to pay him, and Konrad acquires a claim right to receive a payment for the work Olga hired him to complete.The main types of speech acts that serve as sources of transactional duties and rights are those John Searle [52] dubbed "commissives" and "declaratives".Commissives are speech acts through which a speaker conditionally or unconditionally commits themselves to a course of action or the cessation of a course of action.By committing themselves in this way, they bind themselves as an obligor to an obligee in a transactional duty, and the obligee acquires a claim right to the obligor's fulfilling their commitment.Speech acts, such as signing a contract, negotiating a higher pay, promising to quit smoking, pledging one's help, betting on a sports team, volunteering for a charity, guaranteeing the quality of a product, proposing marriage, exchanging vows, offering to pay for a meal, accepting an invitation, pleading guilty, refusing to comment, or leaving a will, are prime examples of commissive speech acts, through which an obligor acquires a duty and an obligee acquires a claim right.
A declarative speech act brings a new state-of-affairs into existence, provided the speaker is duly authorized and recognized as authorized by the audience.The declarative speech acts of interest to us here introduce, eliminate, or alter moral (or legal) rights or duties, provided that the speaker has the relevant authority or power.Speech acts, such as baptizing a baby, setting a curfew, sentencing a criminal, absolving a debt, annulling a marriage, promulgating a decree, vetoing a decision, grounding a child, subpoenaing a witness, canceling a meeting, abrogating a treaty, permitting one's neighbors to use one's pool, and pronouncing two individuals wife and husband, are prime examples of declarative speech acts, through which an authority introduces, eliminates, or alters someone's moral (or legal) rights or duties.As some speech acts belong to both categories, the commissive and declarative speech act categories are not mutually exclusive.Representative examples of speech acts that belong to both categories include consenting to sex, refusing medical treatment, prohibiting the wearing of shoes inside, accepting a job, resigning from a job, and forgiving someone.
All sides can agree that some of our special relationship duties are identical to or derived from our general and transactional moral duties.For example, because everyone has a right not to be assaulted, we all have an impartial moral duty not to assault those with whom we share a special relationship.Because we are morally obligated to keep our promises to our promisees, a fortiori we are morally obligated to keep our promises to those with whom we share a special relationship.However, moral impartialists make a much stronger claim, viz., that non-associative moral duties are the sole source of associative, or special relationship, duties.For example, impartialists might construe special relationship duties owed to intimate partners, family members, friends, and colleagues as deriving from general moral obligations, such as a husband's duty not to rape his spouse, and transactional moral duties, such as your sister's duty to keep her promise to quit smoking.Entering into an agreement or contract is a quintessential example of a commissive speech act.For two parties to enter into an agreement or contract, one party must, at a minimum, make an offer, and the second party must accept that offer [53].By making an offer, the first party makes a conditional commitment to a course of action, and by accepting the offer, the other party finalizes the agreement or contract, which introduces mutual and interdependent duties and rights.Special relationship duties owed to intimate partners, friends, and colleagues, an impartialist might argue, derive from general moral obligations, such as obligations to be honest about important issues, take special care not to harm particularly vulnerable individuals, and refrain from knowingly spreading false information about others, whereas transactional duties derive from commissive and declarative speech acts, such as the speech acts of agreeing to be exclusive, promising to attend a friend's birthday party, and volunteering to serve on a departmental committee.
Special relationship duties owed to dependent children and other family members, impartialists might argue, likewise derive from general moral obligations, such as obligations to care for one's dependent children and to respect other people's rights to autonomy and self-expression, and transactional moral duties derived from commissive and declarative speech acts, such as the speech acts of setting curfews, calling time-outs, announcing house rules, inviting family members to visit, and accepting or declining invitations.
In short, moral impartialists make the controversial claim that because our nonassociative moral duties are the sole source of our associative duties, there are no associative duties per se.In the next section, I will provide three arguments against moral impartialism.

Against Moral Impartialism
My first argument against moral impartialism is based on the fact that articulating that people with whom we share a special relationship need assistance or are in a bad way often can excuse us from fulfilling our existing transactional duties [17,26].To borrow an example from R. Jay Wallace [26] suppose Felix agreed to give a keynote address at an upcoming conference and therefore has a pre-existing non-associative duty to give the keynote at the conference.Telling the conference organizers that his mother is in the hospital would normally excuse him from complying with his pre-existing duty.The fact that articulating that his mother is in the hospital can excuse him from satisfying his pre-existing duty to the conference organizers shows that his responsibility to his mother is weightier than his pre-existing duty.
In the above case, the dissimilarity of Felix's conflicting duties makes it hard to eliminate the possibility that a moral impartialist could somehow find a way to derive Felix's duty to his mother from his general and transactional moral obligations.However, a different case can achieve the desired result.Suppose that Xander, who has two indistinguishable cars, has promised both his wife Cordelia and the general contractor currently overseeing some home renovations that they can borrow one of his cars the following day.Cordelia and the contractor have an essential business meeting some 50 miles away the next day but in opposite directions.That evening, Xander realizes that someone just stole one of his cars.In the envisaged case, the fact that someone just stole one of Xander's two cars and that he has promised his wife Cordelia that she could use one of the two cars the following day can excuse him from fulfilling his duty to his contractor.Conversely, the fact that someone just stole one of Xander's two cars and that he has promised the contractor that he could use one of the two cars the next day cannot excuse Xander from fulfilling his duty to his wife.Since the only relevant difference between Xander's duties to Cordelia and the contractor is that Xander shares a special relationship with Cordelia but only a transactional relationship with the contractor, it must be because of the special relationship he shares with Cordelia that his duty to her is weightier than his duty to the contractor.As we saw in Section 1, moral impartialism is committed to the view that a duty we owe to one person cannot override a duty we owe to a second person if the only relevant difference between the duties is that we share a special relationship with the first person but not with the second person.Since the only relevant difference between the duties Xander has to Cordelia and the contractor is that Xander shares a special relationship with Cordilia but not with the contractor, this example casts doubt on moral impartialism. 12 My second argument against moral impartialism is a variant of Bernard Williams's [25] famous "one thought too many" objection to Kant's requirement that we act only on maxims that can be regarded as valid for anyone similarly situated [54,[56][57][58].The Kantian requirement, Williams argues, conflicts with the idea of acting on the basis of a special relationship one shares with another person.Consider the following lifeboat case loosely based on Williams's original [5].
(The Kantian Husband) A Kantian can save only one of several drowning people-all strangers except his wife.He judges that the maxim "I will prioritize saving my drowning wife over saving a drowning stranger" is universalizable and saves his wife.However, when he tells her why he rescued her rather than one of the strangers, she is anything but happy.She insists that given their special relationship and the particular circumstances, he had an exceptionless duty, and not merely a warrant, to save her.
On a standard reading of Kant, Kant holds that for us to act rationally, we must deliberate and act following the Categorical Imperative.On this reading of Kant, acting for the reason that one has a special relationship with another person without checking one's action for universalizability is practically irrational, even if doing so would make no difference to one's choice.In (The Kantian Husband) example above, the husband would have performed the same action (viz., saved his wife) if he had acted solely for the reason that he shares a special relationship with her and does not share a special relationship with any of the strangers.However, if he acted solely for that reason, he would have acted irrationally on this interpretation of Kant, as it would have been a mere incident that his action conformed to the Categorical Imperative.
Williams's [25] "one thought too many" concern about Kantian ethics is that it would be absurd to insist that we must always first consider whether an action is permissible by the Categorical Imperative, and if it is permitted, only then may we choose to perform that action, for instance, the action of saving our drowning spouse rather than a stranger.I agree with Williams that this outcome would be undesirable.Morality cannot demand we pause to do ethics while our loved ones drown.A promising reply to Williams is to deny that the Categorical Imperative is a decision-making procedure that should be used in any situation to determine whether an action is generally morally permissible in that situation rather than a metaethical method for identifying one's general moral duties [59,60].
While this suggestion circumvents Williams's "one thought too many" concern, it still does not ease the wife's objection in envisaged example [20].The wife's objection partly stems from the fact that her husband did not value their special relationship enough to consider her worthy of special treatment.Instead, he prioritized his general (non-associative) moral duties and only decided to save his wife after practical deliberations confirmed that doing so was morally permissible.Most of us think that the wife has a valid point when she insists that her husband had a special relationship duty to save her, see e.g., [58]. 13Yet, if the wife is correct that the husband has a special relationship duty to save her, there can be circumstances in which associative duties give us reasons to act in a particular way, even if general and transactional moral (non-associative) duties do not demand that we act in that way.But moral impartialism holds that associative duties are identical to or derived from non-associative moral duties.Thus, the idea that the husband had a duty to save his wife, despite not having a non-associative duty to save her, is at odds with moral impartialism.
My third argument against moral impartialism is based on the intuition that, sometimes, we have duties to help the people with whom we share a special relationship, even if doing so infringes on our non-associative moral requirements [61].In defense of this view, Dean Cocking and Jeanette Kennett [61] invite us to consider the narrative presented in the movie Death in Brunswick (1990).In this movie, Carl accidentally stabs Mustapha, and in a panic, he reaches out to his best friend, Dave, who initially suggests that they call the police and explain what happened.However, Carl-who has been in trouble with the law in the past-implores Dave not to call the police, fearing that he will be arrested again.Dave understands Carl's point and helps him move the body to the cemetery where Dave works.Dave then finds a coffin in a newly dug grave and squeezes Mustapha's corpse into it next to the other body.When Mustapha's distressed wife and son later confront Dave and Carl, they insist that they have no idea what happened to Mustapha.Thus, while helping Carl avoid a prison sentence, Dave commits some serious general moral (and legal) offences: he secretly disposes of a dead body, interferes with a grave, desecrates a corpse, and lies about Mustapha's death to his family.Despite these general moral wrongs, Dave has not failed as a friend.Even if Dave's impartial (non-associative) moral duties outweigh his friendship duties in the envisaged case, there are other cases in which a person's special relationship duties clearly override the general requirements of moral impartiality.I will look closer at an example of this kind in the next section.Cases in which a person's associative duties override their non-associative duties are, as we have seen, problematic for moral impartialism.In light of these considerations, we have reason to reject moral impartialism.This leaves us with minimal and substantive moral partialism.In the next section, I offer two reasons to reject minimal moral partialism.

Against Minimal Moral Partialism
Against moral impartialism, I have argued that not all our special relationship duties are identical to or derived from our non-associative general and transactional moral obligations.The only viable views of associative (or special relationship) duties are thus minimal and substantive moral partialism.Minimal moral partialism, recall, holds that the mere existence of special relationships, such as intimate partner relationships, familial relationships, friendships, or collegial relationships, entails that the parties to the special relationship have mutual associative duties and rights [26].Substantive moral partialism, by contrast, holds that for us to have associative duties to a person with whom we share a special relationship, the relationship must possess features that not all special relationships instantiate and that can justify the associative duties.
In this section, I will provide two considerations against minimal moral partialism.As we have already seen, articulating our duties to the people with whom we share a special relationship can often excuse us from fulfilling our pre-existing transactional obligations.As this suggests that our associative duties can carry greater weight than our non-associative moral duties, this speaks against moral impartialism [17,26].However, we can repurpose a version of this argument as an argument against minimal moral partialism.Suppose, once again, that Felix has agreed to give a keynote address at an upcoming conference and that he, therefore, has a moral responsibility to give the keynote.Telling the conference organizers, "My mother is in the hospital," would ordinarily excuse him from satisfying his pre-existing duty to give the keynote.But the only reason that uttering the sentence "My mother is in the hospital" can excuse Felix from his pre-existing duty to give the keynote address is that an utterance of this sentence conveys what Kent Bach [62] calls an "impliciture." 14After all, it is not Felix's mother's presence in a hospital that excuses him from complying with his moral obligation, as witnessed by the fact that uttering the sentences in ( 2)-( 4) below would fail to excuse Felix from fulfilling his pre-existing obligation, even though ( 2)-( 4) entail (1).Nor is it his mother being sick in the hospital that excuses him from satisfying his pre-existing obligation, as witnessed by the fact that uttering the sentences in ( 5)-( 6) below would fail to excuse him from complying with his pre-existing obligation, even though ( 5)-( 6) entail (1).Rather, (1) can only excuse him from fulfilling his pre-existing obligation if it conveys something like the content of the sentence in (7) below.

1.
My mother is in the hospital.

2.
My mother is visiting a friend in the hospital.

3.
My mother is in the hospital until she is off duty at 7 p.m.

4.
My mother is in the hospital for her quarterly Botox injections.

5.
My mother is sick in the hospital.I hope the poison I poured into her coffee is strong enough to kill her. 6.
My mother is sick in the hospital.But I won't visit her, as she has a restraining order against me coming within 100 feet of her.

7.
My mother is sick in the hospital, and my visiting her benefits her, which really matters to me.
Our practices of referring to our duties to those with whom we share special relationships in order to be excused from our existing non-associative moral obligations thus undermine minimal moral partialism, as they suggest that it is not merely the sharing of a special relationship with another person that is the source of associative duties.Instead, to ground associative duties, the special relationship must possess certain features not instantiated by all special relationships and that can justify the associative duties.
The second argument against minimal moral partialism is that few of us would be willing to violate our impartial moral obligations to help just any family member, friend, or coworker.Rather, when we are willing to violate our non-associative moral obligations to help another person, it is usually because they are particularly important to us.For example, in Cocking and Kennett's [61] example derived from Death in Brunswick, where Dave commits some serious general moral (and legal) wrongs to save his close friend Carl from returning to prison, it seems pretty clear that Dave would not have violated his impartial (non-associative) moral obligations for just any person with whom he happened to share a special relationship, say, one of his coworkers at the graveyard.The philosophical literature is rife with thought experiments making the same point.Here is one such thought experiment: Henrietta cannot afford the expensive drugs that could save her sister-in-law Sally's life.After several unsuccessful attempts to raise money, Henrietta decides to steal the life-saving drugs from CVS, thereby committing a severe non-associative moral (and legal) wrong to save her sister-in-law's life.
What motivates Dave and Henrietta to violate their impartial (non-associative) moral obligations to help Carl and Sally, respectively, is not that they share special relationships with them.Indeed, we can easily imagine friends and in-laws who would never violate their impartial (non-associative) moral obligations to help each other.Instead, what motivates Dave and Henrietta to help Carl and Sally, respectively, is that the special relationship they share with them possesses special properties that not all special relationships possess and that can justify the associative duties.Since minimal moral partialism holds that the mere sharing of a special relationship with another person suffices to generate special relationship duties and rights between the involved parties, the preceding considerations speak against minimal moral partialism.That raises the following question: what grounds special relationship duties if we cannot derive them from impartial moral obligations?I consider this question in the next section.

Korsgaard on Special Relationship Duties
Christine Korsgaard [55] has proposed that our associative and non-associative duties have similar sources.Both types of normativity, she argues, are grounded in our selfconception.I find Korsgaard's proposal to be very promising but reject it as it stands.In the spirit of Kant, Korsgaard argues that because we humans are self-conscious and, thus, have a conception of ourselves, we alone are in a position to take a step back from our most intense desire to reflect on whether we endorse that desire. 15If we do not endorse it, we are in a position to (freely) decide whether or not to act on the desire and hence whether or not our desire is a reason for us to act.A desire is a reason for us to act only as long as it withstands reflective scrutiny, or to put it differently, a desire is a reason for us to act only as long as we decide that, on reflection, we endorse that desire.
For Kant, deciding whether a desire is a reason to act requires determining whether acting on the desire is something we will be law [54,56,57].The latter is Kant's Categorical Imperative. 16Unlike Kant, Korsgaard [55] distinguishes Kant's Categorical Imperative from Kant's Moral Law, also known as the Law of the Kingdom of Ends or Republic of All Rational Beings.Kant's Moral Law demands that we act only on maxims (i.e., desires) that "all rational beings could agree to act on together in a workable cooperative system" [55].The Categorical Imperative, by contrast, tells us to act on a maxim only if we will that that type of act become law.Here, to "will" that something becomes law means to reflectively endorse that it becomes law.Reflective endorsement, Korsgaard [55] argues, requires being true to how we conceive of ourselves.However, as she notes, there are a plethora of ways in which we can conceive of ourselves; for instance, we may imagine ourselves as a citizen of the kingdom of ends, as someone's friend, lover, or child, as someone's psychiatrist, teacher, or babysitter, or as a member of a social identity group, religion, profession, or nation, or as a workplace bully or a citizen of a kingdom where only our interests matter [55].The different ways of thinking of ourselves, Korsgaard suggests, can be understood as descriptions under which we value ourselves, find our lives to be worth living, and our actions worth undertaking.We can think of these descriptions as comprising our practical identities.According to Korsgaard, each of our practical identities is a source of reasons and obligations, where the obligations spring from what we must do (or not do) to be true to our practical identities.For example, being true to one's practical identity as a son may require one to visit one's mother when she is sick, and being true to one's practical identity as a sister may require one to bail one's brother out of prison.
According to Korsgaard [55], only those obligations that originate in our conception of ourselves as citizens of the Kingdom of Ends are general (non-associative) moral requirements.Duties arising from our portrayals of ourselves as parents, friends, spouses, or brothers-in-arms are not identical to, or derivative from, our general (non-associative) moral requirements but are associative duties per se.Korsgaard thus rejects moral impartialism, the view that all our special relationship duties are identical to, or derivative from, our non-associative moral responsibilities.Moreover, as sharing a special relationship with another person cannot, on its own, ground our special relationship duties, Korsgaard's view entails a commitment to substantive moral partialism.
Although Korsgaard [55] holds that associative duties are not moral, she leaves open the possibility that associative duties may sometimes carry greater weight than non-associative impartial moral requirements.When obligations that spring from different practical identities conflict with one another, Korsgaard argues, the practical identities that are most important to us take priority.
While I strongly sympathize with Korsgaard's [55] suggestion that practical identities can matter to normativity, I do not think her proposal succeeds.Korsgaard emphasizes that only some of the various ways we might think of ourselves are our practical identities, namely the descriptions under which we value ourselves, find our lives worth living, and find our actions worth undertaking.But this qualification does not go far enough.
One problem with Korsgaard's [55] proposal is that it fails to rule out the idea that self-destructive practical identities can ground associative duties.Suppose that Lucy is in an intimate partner relationship with Rick, who is verbally abusive to her and plainly does not care about her.Although Lucy cares about Rick, the lack of reciprocity precludes their intimate partner relationship being a loving or caring one.Despite the verbal abuse inflicted by her partner, Lucy values herself, finds her life worth living, and finds her actions worth undertaking under the description "Rick's intimate partner".Lucy's associative role as Rick's intimate partner, thus, constitutes one of her practical identities. 17Things are quite different for Rick.It is not the case that Rick values himself, finds his life worth living, and finds his actions worth undertaking under the description "Lucy's intimate partner".Thus, Rick does not have a practical identity as Lucy's intimate partner.Korsgaard's view, thus, has the unwelcome consequence that Lucy's practical identity as Rick's intimate partner vests special relationship duties in her to Rick, such as duties of loyalty, trust, and forgiveness, but because being Lucy's intimate partner is not among Rick's practical identities, Rick has no special relationship duties of loyalty, trust, and forgiveness to Lucy.What's more, Korsgaard's [55] view entails that Rick has claim rights to Lucy's loyalty, trust, and forgiveness, but that Lucy has no right to lay claim to Rick's loyalty, trust, or forgiveness.Not a satisfactory result.
Wallace [26] marshals a different argument against self-conception views like the view advocated by Korsgaard [55].His concern about self-conception views is that attempts to ground our duties in our self-conceptions make it impossible to violate the duties we recognize we have.If we recognize we have a duty to another person, yet we decide not to fulfill it, then, he argues, that shows that complying with this obligation is not among those things that are most important to who we are.However, if we cannot violate an alleged duty we think we have, then that is just another way of saying that we do not have the duty in the first place.Wallace [26] acknowledges that self-conception views can accommodate cases in which we vitiate an obligation because it is incompatible with another duty that has greater significance to us than the first.However, he argues, champions of self-conception views still face the intractable problem of explaining how we can violate the weightiest, or all-things-considered, duties we recognize we have.
As Korsgaard's [55] proposal lacks the resources to address these objections adequately, we should look for a different account of what grounds associative duties.In the next section, I propose an alternative to Korsgaard's variant of substantive moral partialism.I suggest that caring/loving relationships normatively ground our associative duties.For a special relationship between capable individuals to be loving, I argue, each party to the relationship must have a mutual core-value desire to promote the other's flourishing (or well-being) and to respect the other's core values.In the subsequent section, I show that my variant of substantive moral partialism does not fall prey to the above objections.

Moral Partialism Redux
Care and love play central roles in my account of what grounds of our special relationship duties.Before I provide the details of my proposal, a word on the meanings of "care" is in order.In one sense, "care" refers to a warm or loving attitude.In a different sense, "care" refers to something one must provide for another person.Suppose Homer hires a babysitter, Heidi, to care for his young daughter Eden so that he can attend a colloquium dinner.As a babysitter for hire, Heidi is morally obligated to provide proper care for Eden.However, doing so is compatible with her not having a warm or loving emotional attitude toward Eden.In this sense, "care" is semantically akin to "care" as a legal duty.In the legal sense, a duty of care is a fiduciary responsibility an individual or organization has to people who depend on them in a certain capacity.For example, a doctor has the duty of care to give patients proper medical attention; a babysitter has the duty of care to attend to the child's needs; and a gym owner has the duty of care to maintain a safe environment for employees and customers.Those in charge of an organization also have the duty of care to make prudent decisions that promote the organization's interests.A breach of duty of care happens when a person or organization's failure to meet the relevant standard of care causes harm to someone who depends on them in the relevant capacity.As we will see, there are affinities between care as an emotional attitude and care as something people and organizations have a legal duty to provide.The key difference between care as an attitude and care as a legal duty is that care as a legal duty is not an emotional attitude but conduct and omissions that meet a relevant care standard.The conduct and omissions that meet the relevant care standard do not usually express an emotional attitude of care, although they could do so.
Here, I use "care" to refer to a distinctive loving attitude.In ordinary parlance, the expressions "care about" and "care for" can both be used to denote the loving attitude in question.The question here arises: When we care about someone, is our distinctive loving attitude an attitude of love?Or, to put it differently: When we care about someone, does that entail that we love them?The answer to this question depends on how we disambiguate "love."For a disambiguation of "love," there is hardly any better place to look than Aristotle.Indeed, if "love" refers to the kind of love Aristotle dubs "philía" (ϕιλία), the answer to our question is in the affirmative: to truly care about another human being is to love them. 18Aristotle uses the noun philía to refer to love in a generic sense that comprises spousal, parental, familial, and friendship love [65] (NE VIII.12 1162a16-27; EE VII.3 1238b23-25, VII.10 1242a31-32; NE VIII.7 1158b11-12, VIII.10 1161a4-5; NE IX.7 1168a19).The denotation of "philía" is a loving, or caring, relationship, which commits Aristotle to what we might call the "reciprocity view," the view that love depends for its existence on a reciprocal loving relationship.However, Aristotle does not conceive of philía as a sui generis (or irreducible) loving relation that obtains between two people.Rather, he takes philía to be composed of affective attitudes and desires.The verb form "philein" (τὸ φιλεῖν) refers to a directed affective attitude of caring, such as a mother loving her child [65] (NE VIII.8 1159a26-35), whereas the verbal nominalization "philêsis" (ϕίλησις) refers to a reciprocated affective attitude of caring, such as mutual affection and care in an intimate partner relationship [65] (NE IX.7 1168a19). 19Aristotle also uses philêsis to ascribe affective features, like warmth, kindliness, or lovingness, to personal and transactional relationships, such as those between a blacksmith and a loyal customer or a mother and her child [65] (NE VIII.2 1155b28, VIII.3 1156a6, IX.1 1163b29-1164a21, IX.7 1168a19).
Aristotle sometimes speaks of the act of loving (philein) someone or something as requiring wanting what we regard as good for its sake and not for our own sake [65] (NE II.4 1155b31, II.4 1156b9-10, VIII.7 1157b28-33, VIII.8 1159a26-35; DA 403a16-18; Rh II.4 1380b34-1381a3).As an illustration of a sacrifice of self-interests, consider Aristotle's example of a mother who gives away her child because doing so is in the child's best interest [65] (NE VIII.8 1159a27-35; cf., Rh II.4 1381a1-2).The mother sacrifices her deep interest in being part of the child's life because her love for her child moves her to act in her child's best interest rather than in her own interest.It may, thus, seem that Aristotle's philía is a selfless and self-sacrificing form of caring [65] (NE IX.4 1166a5-6; MM II 1211b19-21).Philía would, then, by its very nature be the polar opposite of sexual love (eros), a passionate, greedy, and self-indulgent form of affection [67].However, this is not how Aristotle's understands philía [68].Loving someone may require sacrificing one's interests and values for the sake of the beloved's interests or values [65] (NE VIII.2 1155b17-1156a5, VIII.3 1156b8-24, VIII.13 1163a10-1163a23), but it does not typically require pure selflessness and self-sacrifice [65] (NE VIII.2.1155b27-1156a3, 1155b34-1156a5, IX.8).In the context of friendship, for example, philía tends to elicit desires for reciprocity and emotional intimacy.One reason we strive for the kinds of friendships that nurture love is at least in part that they help advance our own interests as an explicit end [65] (NE IX.8).Thus, while philía, in Aristotle's sense, may require selflessness and self-sacrifice, it more commonly requires a balance of giving and receiving.Philía, in Aristotle's sense, is, thus, a good candidate to be the distinctive loving attitude I will argue grounds special relationship duties.In what follows, I shall use the terms "loving relationship" and "caring relationship" synonymously, while keeping in mind that for a relationship to be loving or caring (and, thus, for it to manifest philía), it must manifest either a unidirectional attitude of caring or loving (philein) or a bidirectional attitude of caring or loving (philêsis).
Against this backdrop, I am now in a position to lay out my view of what normatively grounds associative (or special relationship) duties.I propose that only caring/loving relationships can ground associative duties.My initial gloss on caring/loving relationships runs as follows: two people can only share a caring/loving relationship if they have a mutual desire to promote each other's interests and respect each other's values."Respect", here, is to be understood in Stephen Darwall's [69] recognition sense.To respect the other's values is to give serious consideration to those values when making decisions that affect the other person.This initial gloss of caring/loving relationships is still deficient.Indeed, desires can be short-lived, faint, pointless, random, and of little importance to their bearer.But, surely, such desires are not weighty enough to ground associative duties.I propose that the desires that ground associative duties are distinctive kinds of values (or valuational states).Following John Doris [70], I shall take values (or valuational states) to be desires that "exhibit some degree of strength, duration, ultimacy, and non-fungibility, while playing a determinative-justificatory role in planning" [70].Your desire exhibits ultimacy if you desire the object of the desire at least partly for its own sake.For example, if you desire good health at least partly for its own sake, your desire for good health is ultimate.Your desire plays a determinative-justificatory role in planning if it partly determines and partly justifies what you plan to do.For example, if you plan to visit the gym regularly to remain in good health, your desire for good health partly determines and justifies what you plan to do.Finally, your desire exhibits non-fungibility if you consider the object of your desire to be irreplaceable.For example, if you want your marriage to last, and you regard your spouse as irreplaceable, then desire exhibits non-fungibility.For Doris [70], our values are closely bound to a valuational account of self-direction or agency. 20Doris's valuational theory locates the exercise of agency in the expression of the actor's values, where behaviors express values when they are guided by a value-relevant goal.As we can exercise our agency on the basis of fungible desires, e.g., a desire for chili fries rather than garlic fries, Doris does not require that desires exhibit non-fungibility to count as values.Since the objects of values (or valuational states) that exhibit non-fungibility are the values that matter most to us, I will call such values "core values" (or core valuational states).Doris [70] suggests non-fungibility as a criterion to determine whether we truly value something or whether it is merely hard to resist it or a convenient way of satisfying some general need. 21He writes: If the object of desiring can be replaced without loss-if life can go on pretty much as it did-then that object is not an object of value.[. ..]Significant human relationships [. ..] aren't supposed to be so easily interchangeable: the jilted and bereaved may learn to love again, but some of what they lose may seem quite irreplaceable.The non-fungibility requirement could also be erected as a bulwark against the somewhat unattractive thought that "mere needs" are values.If alternate modes of nutrition were readily available to humans, it might be that people could forgo eating with no loss; eating is the object of desire, yet doesn't have the "specialness" associated with value.(That gourmets, gourmands, and other foodies could not switch without loss makes the point; they do value eating.)In the end, a mere needs exclusion might not be mandatory; in conditions of scarcity, the objects associated with mere needs can start to seem pretty special.[70] Although Doris [70] does not employ the notion of a core value, what he says here seems more pertinent to core values than "trivial" values.When a desire satisfies the four constraints needed for it to be a core value, I will say that the desire is a "core-value desire".Given the notion of a core-value desire, we can say that if a special relationship is loving/caring, the involved parties must have mutual core-value desires to promote the other person's interests (or flourishing) and to respect the other's core values.
As it stands, my account fails to address the scope properties of the desires that define caring/loving relationships.One reason this is problematic is that on one scope reading of these desires, being in a caring/loving relationship requires knowing all of the interests and core values of the other person, which would be an unnecessarily strict requirement with which we probably can never comply.To illustrate this point, suppose that Lilo asks her best friend Stitch to help her with a homework assignment.As happens to be the case, it is in Lilo's best interest to complete the assignment, but Stitch does not know this fact.So, he helps her out, and by doing so, he thwarts her interests.Stitch might repeatedly, unknowingly thwart her interests, thus severely setting back her interests in the long run.In the envisaged scenario, Lilo and Stitch's claim that they care deeply about each other is false.Not a satisfactory result.Intuitively, failing to promote the other person's interests because of ignorance of their interests should not entail that the relationship is not caring or loving.Likewise, failing to respect the other person's values due to ignorance should not imply that the relationship is not caring or loving.
By shifting our attention to the scope properties of desires, we can circumvent the undesirable implications that would otherwise follow in cases in which we do not know all the interests or core values of the other person."Stitch has a core-value desire to promote Lilo's interests" can be given a wide-scope (de re) or a narrow-scope (de dicto) reading. 22On a wide-scope (de re) reading, "Stitch has a desire to promote Lilo's interests" can be read as follows: there are some interests, X, that are Lilo, and Stitch wants to promote X.On a narrow-scope (de dicto) reading, "Stitch has a core-value desire to promote Lilo's interests" can be read as follows: Stitch wants to promote Lilo's interests X, whatever they may be.On the de dicto reading, failing to promote a person's interest based on ignorance does not preclude that the special relationship can ground associative duties.Thus, the mutual core-value desires to promote the other's interests that mark loving relationships must be de dicto desires.
"Stitch has a core-value desire to respect Lilo's core values" is subject to the same scope ambiguity as "Stitch has a core-value desire to promote Lilo's interests".On a wide-scope (de re) reading, "Stitch's core-value desire to respect Lilo's core values" can be read as follows: there are some values X that are Lilo's, and Stitch wants to respect X.On a narrowscope (de dicto) reading, "Stitch's desire to respect Lilo's core values" can be read as follows: Stitch wants to respect Lilo's values X, whatever they may be.On the de dicto reading, failing to respect the other person's values based on ignorance does not preclude that the special relationship can ground associative duties.So, the mutual core-value desires to respect each other's core values that mark loving relationships must be de dicto desires.Thus, for a special relationship to be caring/loving does not require knowing all of the other person's interests and values.However, not knowing the other person's interests and values may preclude a relationship from grounding associate duties when this creates a reasonably foreseeable risk of the other person suffering a severe moral wrong.
Mutually wanting to promote each other's interests and to respect each other's core values is sufficient but not necessary for a relationship to be caring/loving, as we can reasonably assume that we can have caring/loving relationships with individuals who lack the capabilities required to respect other people's core values or to promote their interests, e.g., young children [65] (NE VIII.7 1158b12-1158b28).For simplicity's sake, let us call individuals with the required combined capabilities "capable individuals" and those who do not have them "incapable individuals".The requirements for relationships between capable and incapable individuals are less strict than those for relationships between capable individuals.A special relationship between a capable and an incapable individual is only caring/loving if the capable person has a core-value desire to promote the other's interests and respect their core values (if any).My primary focus in what follows is on special relationships between capable individuals.Thus, unless otherwise noted, I shall assume that special relationships involve capable individuals.In light of the above considerations, we can articulate the notion of a loving/caring relationship as follows: (Caring/Loving Relationship) A special relationship between two capable individuals is caring/loving, just when: they have mutual (de dicto) core-value desires to promote each other's interests and to respect each other's core values.
It merits emphasis that a relationship can be loving/caring even if one person attaches greater importance to their core-value desires to promote the other person's interests and to respect their core values than the other person does to theirs.For a special relationship between two capable individuals to be loving/caring, each person's core-value desire must surpass a specific threshold for loving/caring simpliciter.I elaborate on the gradability of love and care below.
At first glance, it may seem that my notion of a caring relationship is rationally incoherent.After all, promoting a person's well-being and respecting their values can come into rational conflict.For example, a father may know that becoming a lawyer like himself is in his son's best interest (all-things-considered). Suppose, however, that his son's core-value preference is to become a chicken farmer.If the father respects his son's core-value preference to become a chicken farmer, he cannot rationally have a core-value desire to promote what he knows to be in the son's best interest, viz., becoming a lawyer.
Although there is indeed an appearance of rational conflict here, this appearance is-at least for the most part-illusory.Respecting what we know to be the other person's core values is generally the right thing to do and generally takes priority over promoting what we know to be in their best interest.Promoting what we know to be in the other person's best interest while disrespecting their values is often unduly paternalistic.Strong paternalism may be tolerable in exceptional circumstances, for instance, if a person is self-destructive, perhaps because they value taking their own life more than anything else, but not in cases in which a person has a core-value desire to do something that is not self-destructive yet does not maximize their flourishing to the greatest possible extent either.

Destructive Identities and Non-Violateable Associative Duties
One advantage of my proposed version of moral partialism is that it avoids the objections to views that attempt to ground associative duties in how we conceive of ourselves, such as Korsgaard [55].In this section, I begin by showing that my account avoids the problem of destructive identities that threatens Korsgaard's proposal.I then show that my account does not fall prey to Wallace's [26] objection to self-conception views.
Recall that on Korsgaard's [55] view, there are many ways we can think of ourselves; for instance, we may think of ourselves as someone's friend, lover, or child.Several of these ways of conceiving of ourselves, she suggests, can be understood as descriptions under which we value ourselves, find our lives to be worth living, and find our actions to be worth undertaking. 23The latter descriptions, she argues, comprise our practical identities, which, in turn, ground our special relationship duties.
While I am sympathetic to Korsgaard's [55] self-conception view, I disagree with her claim that practical identities normatively ground special relationship duties.Suppose, once again, that Lucy and Rick are in an abusive intimate partner relationship, as Rick is emotionally abusive to Lucy.Despite (or perhaps because of) the emotional abuse, Lucy has fallen victim to self-deception, which has resulted in her valuing herself, finding her life worth living, and finding her actions worth undertaking under the description "Rick's intimate partner".Lucy's associative role as Rick's intimate partner, thus, constitutes one of her practical identities.Thus, on Korsgaard's proposal, Lucy's practical identity as Rick's intimate partner normatively grounds associative duties she owes to him.But it is not the case that Rick values himself, finds his life worth living, or finds his actions worth undertaking under the description "Lucy's intimate partner".His associative role as Lucy's intimate partner is, therefore, not among his practical identities.Thus, Korsgaard's [55] proposal will deliver the verdict that Lucy has associative duties to Rick, such as duties of loyalty, trust, and forgiveness, but that Rick has no such obligations to Lucy.What's more, Korsgaard's view entails that Rick has claim rights to Lucy's loyalty, trust, and forgiveness, whereas she has no rights whatsoever to claim the same loyalty, trust, and forgiveness in return.This is an unpalatable consequence.
My variant of moral partialism avoids this counterintuitive result.In my view, special relationships must be caring/loving to give rise to associative duties and rights.As Rick is emotionally abusive to Lucy, their intimate partner relationship is not caring/loving and, therefore, cannot normatively ground any associative duties or rights.Thus, in my view, Lucy has no duties to be loyal to, trusting of, or forgiving of Rick, and Rick has no claim rights to Lucy's loyalty, trust, or forgiveness.This implication is clearly less unacceptable than one that binds victims of intimate partner abuse to their abusers in special relationship duties of loyalty, trust and forgiveness, while letting the abusers off the hook.
Let us now turn to Wallace [26] objection to self-conception views like the version defended by Korsgaard [55].According to Wallace, attempts to ground our associative duties in our self-conceptions make it impossible to vitiate the all-things-considered associative duties we recognize we have.If we have conflicting associative duties to different people, we must violate one of them.However, knowingly failing to comply with an all-things-considered associative duty we recognize we owe to another person shows that our association with that person is not of the utmost importance to us.Thus, our associative role with respect to that person is not among the self-conceptions under which we value ourselves, find our lives worth living, or find our actions worth undertaking, which means that our associative role with respect to that person cannot normatively ground any associative duties.Thus, what we thought we owed to the other person was never something we owed to them in the first place.Or so the objection goes.
At first glance, Wallace's [26] objection to self-conception views may seem to carry over to my version of moral partialism.Recall that associative duties arise from special relationships in which both parties have a (de dicto) core-value desire to promote the other's interests (or flourishing) and to respect the other's core values, where having a (de dicto) core-value desire to promote the other's interests requires deeply valuing doing so.To see why Wallace's objection may at first seem to carry over to my view, suppose that Rosie recognizes that she has an associative duty to attend her close friend's book signing.However, overcome by her akrasia, Rosie vitiates this duty and stays at home and watches television instead.It may seem that Rosie's breach of duty shows that promoting her friend's interests and respecting his values are not among her core values, which would seem to mean that she never owed him any special relationship duties in the first place.
My proposed account has a natural reply to this objection.In the envisaged case, Rosie recognizes that she has an all-things-considered associative duty to attend her friend's book signing but breaches the duty due to akrasia.When Rosie succumbs to her intense desire for self-gratification, against her better judgment, this outcome subtracts from how much she cares about her friend.However, it does not fully eradicate Rosie's affective caring attitude, provided she does not make a habit of it.
The reason that care can survive the occasional weakness of will turns on its gradable nature.Unlike pregnancies, college degrees, and electoral wins, caring about another person is not an on-off matter but occurs in degrees.Statements, like those in ( 8)-( 14), testify to the gradability of "care" and "love", given that "love" can replace "care" without infelicity.

8.
The person Lucy cares most about is Rick.9.
Stitch cares equally about all his sisters.10.Rosie cares more about Ziggy than she does about Coco.11.Coco cares more about her wife now than ever before.12. Rosie would have cared much more about Ziggy if she had not cheated on her. 13.Lily cares deeply about her best friend.14.Hank and Viggo care a lot about each other, considering that they are ex-lovers.
Familiar examples of gradable verbs and adjectives are "trust", "honest", "large", "tall", and "extravagant".One mark of gradables is that they occur with degree morphology, such as "too much", "more than", and "a little".As the sentences in ( 8)-( 14) demonstrate, "care" occurs with degree morphology.A second mark of gradables is that their meaning tends to depend on a comparison class.For example, a home that would be extravagant in Camden, New Jersey, would probably not be extravagant if located on Fisher Island in Miami, Florida.The statement in ( 14) above shows that "care" can change in meaning as its comparison class changes.
The upshot is that one person can care about a second person, even if they do not perfectly fulfill their special relationship duties.Caring about another person merely requires that one's (de dicto) core-value desire to promote the other's interests and respect their core values surpasses a certain threshold.For example, two individuals who differ in height can both be tall simpliciter.Similarly, if Ziggy cares more than Rosie about their friend, Ziggy and Rosie can still accurately be said to care about their friend simpliciter.As noted, the meaning of gradable verbs and adjectives often depends on a comparison class.For example, how tall a person needs to be to count as tall simpliciter depends on a comparison class for tallness.At six feet six inches, Michael Jordan is tall for an American man but not for an NBA basketball player.Likewise, how much we need to care about another person to care about them simpliciter depends on the comparison class used.Often, different types of relationships come with a comparison class of their own.Caring a lot about one's child may require exceeding a higher threshold for care than caring a lot about one's neighbor due to the different comparison classes for neighborly associations and parent-child relationships.Given these considerations, I am now in a position to respond to Wallace's [26] objection.Knowingly violating an all-things-considered associative duty to a close friend, owing to akrasia, does not mean we do not care simpliciter about our close friend unless we keep violating our all-things-considered associative duties to them.It merely means we are not perfect.As my variant of moral partialism can make sense of what it means to violate an all-things-considered associative duty, it is not vulnerable to Wallace's [26] objection.

Conclusions
In this paper, I have made a case for a version of moral partialism, the view that not all associative (or special relationship) duties are identical to or derived from non-associative moral duties.Associative duties, I have argued, are grounded in mutually caring or loving relationships.For two capable individuals to share a caring or loving relationship, I argued, they must mutually and non-trivially want (de dicto) to promote the other's interests (or flourishing) and respect the other's core values, as non-trivially wanting (de dicto) to promote the other's interests requires deeply valuing doing so.Moreover, I have shown that my proposed account of special relationship duties avoids the pitfalls of three alternative views, i.e., the view that associative duties are identical to or derived from non-associative impartial moral duties (i.e., moral impartialism), the view that special relationships by their very existence generate special relationship duties (i.e., minimal moral partialism), and an alternative self-conception view defended by Christine Korsgaard. 24 I speak of non-associative moral duties, which include general and transactional duties, because I hold that special relationship, or associative, duties are a special kind of moral duties.However, my arguments are consistent with the view that associative duties are associative duties per se, but not moral duties, see, e.g., [26].
5 Scheffler [18] calls the two views "cosmopolitan egalitarianism" and "cosmopolitan non-egalitarianism," whereas Wallace [26] calls the two views "reductionism" and "non-reductionism."6 [1,12,16,[27][28][29] (among others) express sympathies toward some version of moral impartialism, as the view is defined in this paper.At least some of associative duties seem to derive from second-person epistemic duties.Thus, if we owe it to a stranger not to violate her right to epistemic justice, then we also owe it to our good friend Iris not to violate her right to epistemic injustice, see [30][31][32][33][34][35].Thus, it is a reasonable guess that moral impartialists do not ultimately want to maintain that all associative duties are identical to or derived from non-associative moral duties.For simplicity's sake, however, I will set aside second-person epistemic duties.8 [3][4][5]14,[17][18][19][20][21][22][23][24][25][26][36][37][38][39][40][41][42][43][44] (among others) express sympathies toward some version of partialism, as the view is defined in this paper.Disagreement remains among partialists as to whether associative duties per se are moral duties or sui generis non-moral duties.9 A common (teleological) argument for minimal partialism is that acknowledging special relationship duties is instrumental to or necessary for having the relevant special relationships [36].10 Examples of intrinsic features of special relationships that not all special relationships possess include being (partly) constituted by a relationship history, mutual love, or mutual trust, whereas examples of relational features of special relationships that not all special relationships possess include being each other's favorite skiing companions, being the special relationship of your highest priority, or being out in the open (as opposed to secret).For example, a secret love affair is a special relationship that does not possess the relational feature of being openly acknowledged.

11
Transactional moral obligations are commonly referred to as bipolar obligations [46][47][48].Bipolar obligations have two poles: they are directed and relational obligations that makes an obligor indebted to an obligee.Bipolar obligations provide the obligee with legitimate claims and demands of the obligor.Thus, as an obligee bound in bipolar obligation to an obligor, a person has individual second-person authority to demand on their own behalf that the obligor fulfill their binary obligation to them.Thus, transactional and characteristic associative duties are examples of bipolar obligations.As I want to divorce associative duties from transactional duties, I will not use the term "bipolar obligation" here.Stephen Darwall [46] contrasts bipolar obligations with non-directed unipolar moral obligations, which he refers to as "moral obligations per se (or period)."The obligations that I call "general moral obligations" are what Darwall [46] calls "moral obligations per se" Moral obligations per se are never owed to anyone in particular.The normative status of a moral obligation per se derives from the representative second-person authority of the moral community.Any moral agent has representative authority to demand on behalf of the moral community that any moral agent complies with moral obligations per se.Thus, if you violate a moral obligation per se, you do not wrong anyone in particular, nor are you responsible to anyone in particular for your wrongdoing.Rather, you have done something wrong, and you are accountable for your wrongdoing from the point of view of the moral community.Darwall [46] assimilates bipolar obligations to civil laws and moral obligations per se to criminal statutes.12 It might be argued that a Kantian can overcome this issue and subsequent difficulties by invoking Kant's distinction between perfect and imperfect duties.Perfect duties are duties admitting "of no exception in favor of inclination" [54], whereas imperfect duties admit of exceptions.For example, we have a perfect duty not to torture children for fun and an imperfect duty to sometimes and to some extent assist or help others.It is doubtful, however, that appealing to Kant's distinction between perfect and imperfect duties can help a Kantian discriminate between a duty to one's contractor (or to a stranger) and a similar duty to one's wife, since Kant would deny that one has any imperfect duties to one's loved ones, which one does not have to strangers or those with whom one shares a purely transactional relationship.However, Korsgaard [55] proposes a way for a Kantian to embrace moral partialism.I address her proposal below.Thanks to an anonymous reviewer for pressing this point.13 Velleman [58] argues that the husband is morally obligated to save his wife for reasons of their shared history.14 Impliciture, in Bach's [62] sense, should not be kept apart from Paul Grice's [63] conversational implicature.In the case of conversational implicature, the speaker says one thing but pragmatically conveys something different.For example, if you sarcastically say to your coworker who is perpetually late "You are always very punctual," you say one thing but convey the opposite.In the case of impliciture, a speaker says one thing, but conveys something more specific.For example, if you say "I have not had breakfast," you convey something more specific than what you say (i.e., that you have not had breakfast today).
15 Thus, for Kant, if we act on a desire, we do so freely [54,55,64].16 A critic might argue that other Kantian principles, such as the Formula of Humanity [54], which requires treating others as ends in themselves and not as mere means, are especially apt to special relationships.However, this is not obviously true.On Kant's interpretation, the Categorical Imperative and the Formula of Humanity are both formulas of universalizability and treated as mutually derivative [54,[56][57][58].Thus, the Formula of Humanity cannot be used to justify giving priority to saving your own child from the fire before you save other people's children.17 Associative roles differ from institutional roles.But many associative roles have institutional role counterparts.For example, being a parent is an associative role one acquire if one has children.Being a parent is also an institutional role defined by institutional parental duties and rights.Being a friend, by contrast, is an associative role with no institutional role counterpart.However, which institutional role we have is a contingent matter.For example, being a friend could have been an institutional role defined by duties and rights, such as the right of friends to file taxes together or file for a friendship divorce.According to Aristotle, intimate partner relationships and close friendship often manifest several of the four types of love: philía (a relation of caring), eros (sexual love), storge (empathy), and agape (Godly love) [5,40].This was especially true in Ancient Greece, where sexual relationships (pederasty) between an adult male and a younger boy often continued as character friendships with a sexual component long after the younger boy had reached adulthood, as in the case of the friendship between the Iliad's epic heroes Achilles and Patroklos [66].As Aristotle puts it in the Nicomachean Ethics, "many [pederastic] couples continue the relationship, if, as a result of spending time together, they come to love each other's character, because they are of similar character" [65] (NE VIII.4.1157a 9-11).20 Doris [70] is mostly concerned with a valuational account of self-directed (or autonomous) agency, but develops a valuational account of the self that depends on values manifesting non-fungibility in Chapter 8.

21
For an argument for the view that the irreplaceability test can be used to determine whether something is an important value to us, see also [71].22 Interestingly, the distinction between the de re and the de dicto readings of sentences containing attitudinal and other modal operators dates back to the medieval French scholastic philosopher Peter Abelard's Dialectica and Logica "Ingredientibus."[72].23 As Korsgaard [55] understands "practical identity," a conception of ourselves under which we value ourselves, find our lives worth living, and find our actions worth undertaking need not be fully consciously available to us and hence need not involve self-consciousness.

24
I am grateful to Raja Halwani and three anonymous reviewers for this journal for invaluable comments on an earlier version of this paper, and to Tim Bayne, Ronald de Sousa, John Doris, Brian Leiter, Arina Pismenny, and Michael Slote for insightful discussion of these and related issues.I am also grateful to audiences at Humboldt University, the University of Miami, the University of Oslo, and a symposium at the 2016 Eastern division meeting of the APA, for the opportunity to present some of these ideas and for excellent discussion and feedback.