The Ethics of Partiality

: Partiality is the special concern that we display for ourselves and other people with whom we stand in some special personal relationship. It is a central theme in moral philosophy, both ancient and modern. Questions about the justification of partiality arise in the context of enquiry into several moral topics, including the good life and the role in it of our personal commitments; the demands of impartial morality, equality, and other moral ideals; and commonsense ideas about supererogation. This paper provides an overview of the debate on the ethics of partiality through the lens of the domains of permissible and required partiality. After outlining the conceptual space, I first discuss agent-centred moral options that concern permissions not to do what would be impartially optimal. I then focus on required partiality, which concerns associative duties that go beyond our general duties to others and require us to give special priority to people who are close to us. I discuss some notable features of associative duties and the two main objections that have been raised against them: the Voluntarist and the Distributive objections. I then turn to the justification of partiality, focusing on underivative approaches and reasons-based frameworks. I discuss the reductionism and non-reduc-tionism debate: the question whether partiality is derivative or fundamental. I survey arguments for ‘the big three’, according to which partiality is justified by appeal to the special value of either projects, personal relationships, or individuals. I conclude by discussing four newly emerging areas in the debate: normative transitions of various personal relationships, relationships with AI, epistemic partiality, and negative partiality, which concerns the negative analogue of our positive personal relationships.


Introduction
The focus in this paper is partiality: the special concern that we display for ourselves and other people with whom we stand in some special personal relationship, in terms of both our actions and our attitudes.The over-arching question for us will be: 'What, if anything, may justify partiality?' Take a moment to reflect on those endeavours that are centrally important in your life.Like most people, you probably have personal commitments, goals, and hobbies as well as personal relationships you value with your family, friends, and colleagues.You see these endeavours as giving you good reasons for actions and attitudes that other people lack-you are partial to your hobby of playing the drums even though you could spend your time doing something better from an impartial point of view.And when it comes to your relationships, you see yourself as owing more to your children, spouse, and friends than to people who are completely unrelated to you.You feel joy and excitement when your pursuits and intimates fare well and sadness and grief when they fare poorly.Intuitively, partiality is an obvious and justified fact of life.
2 A good starting point is Aristotle's NE, Bks VIII and IX.See also Annas (1993), p. 27, and  Cottingham (1998).Scheffler's (1982) The Rejection of Consequentialism is a good example of this approach (the clue is in the book title).See also Williams (1981), Stocker (1976), Wolf (1992), and Jeske  (1997).favour partiality.Impartial theories such as consequentialism have famously been charged with being too demanding by turning us into nodes in a maximizing calculus of value or alienating us from ourselves. 4But Kantian or other deontological theories may also seem objectionably impartial, abstracting from the identity of persons (or basic commitments) and thereby failing to make room for a particular person's agency. 5fending partiality can give us breathing space from the stringent demands of morality and make it a more human phenomenon as opposed to a distant and austere one. 6The stakes in this debate are high: insofar as some consequentialist or deontological impartialist theories are unable to justify partiality, we may face a seismic shift at the level of normative theorizing, being forced to move away from these theories in search of more plausible alternatives. 7In other contexts, partiality may put pressure on other important moral ideals, such as those of moral equality and freedom. 8The ethics of partiality therefore interacts with some issues at the very core of ethics.
In light of the heterogeneity of the partiality literature, I find it helpful to divide the topic into permissible partiality on the one hand and required partiality on the other.Permissible partiality concerns your permissions not to do what would be impartially optimal.Required partiality concerns associative duties that go beyond your general duties to others and require you to give special moral priority to people who are close to you, such as your family, friends, or other intimates.
Below, I first describe the conceptual space (Section 2) and then give an overview of the domains of permissible and required partiality (Section 3 and 4).I then review some of the most influential lines of defence for permissions and requirements and disentangle the different levels at which these can occur (Section 5).I conclude with some recent trends in the literature on the ethics of partiality (Section 6).

Big Picture
There are four different 'big picture' positions: No Partiality: Partiality cannot plausibly be defended.
On this view, there is no breathing space or escape from morality's stringency. 9The most forceful defence of this can be attributed to Kagan (1989), though he argues for the stronger claim that no moderate morality can be justified, where moderate morality is understood to include permissions to deviate from doing what would be impartially optimal as well as any constraints on doing what is impartially optimal.10Permissible Partiality Only: There exist only permissions (not requirements) to deviate from doing what would be impartially optimal.
On this view, you are sometimes permitted to forgo doing what is impartially optimal, but you are never required to do so.11For example, you might sometimes be permitted to focus on your interests or spending time with your good friend instead of doing what impartial morality would demand.However, on this view, you would never do wrong by nonetheless choosing to do what would be impartially optimal.
Required Partiality Only: There exist only associative duties (not permissions) to give special priority to your intimates (plausibly, in addition to other general duties).
Conversely, this view says that you can only be required, and never merely permitted, to act with partiality. 12On the one hand, this view does seem intuitively plausible when considering some special personal relationships.For example, we might think that the parents are mostly under special requirements to care for their children.However, on the other hand, this view seems especially controversial insofar as it includes partiality to oneself, since most think that we only have permissions as opposed to requirements of partiality to ourselves.13Full Partiality: There exist permissions to deviate from doing what would be impartially optimal and associative duties to give special moral priority to your intimates.
This picture is what most people intuitively endorse and what is probably most closely aligned with commonsense morality.On this view, you have permissions to deviate from doing the best, but in some cases, you are also required to act in certain ways toward your intimates.

Permissible Partiality
A helpful way to think about the domain of permissible partiality is in the context of the previously mentioned fight for breathing space from impartiality's stringent demands.Much of the literature here reacts to impartial theories such as maximizing consequentialism or Kantian theories, trying to make room for the optionality of acting on our personal point of view.

Options
More formally, permissible partiality concerns

Agent-Centred Moral Options (henceforth just 'options'):
Permissions to bring about outcomes that are impartially suboptimal because they promote your own or your intimates' interests.
Options are also commonly referred to as 'prerogatives' or simply 'permissions'.They permit you to act non-optimally if you choose to do so, but they do not require that you act in a non-optimal way.'Interests' here can be understood narrowly in terms of our self-concern, desires, or preferences, or more broadly in terms of objective goods such as valuable personal relationships.14A central feature of options is that they are agent-relative; they are indexed to you and consequently give you permission to help your friend rather than a stranger, but they do not necessarily give someone else permission to do the same.15

Options to do what?
Hardly anyone who thinks that options exist denies the existence of Self-Favouring Options: Permissions to bring about an impartially suboptimal outcome because it favours your own interests; 16,17 and Other-Favouring Options: Permissions to bring about an impartially suboptimal outcome because it favours your intimates' interests.
For example, you might be permitted to give one unit of happiness to yourself instead of giving five units to a stranger, or you might permissibly give two units of happiness to your long-term friend instead of giving five to a stranger.
More controversial is the existence of so-called

Self-Sacrificing Options:
Permissions to bring about an impartially suboptimal outcome because it discounts your own interests. 18he idea here is that you can sometimes permissibly forgo a greater benefit to yourself for the sake of providing a lesser benefit to someone else.For example, you might forgo five units of happiness for yourself to give your friend one unit of happiness.
The reason that the existence of these self-sacrificing options is more controversial is that they appear antithetical to the idea that moral closeness increases our options to favour. 19It is because your intimates are closer to you that it seems plausible that you can favour intimates more than strangers.Likewise, commonsense morality also suggests that you can significantly favour yourself over intimates and strangers because you are 'closest to yourself'-an ultimate intimate.But if this picture is true, then why should you ever be permitted to sacrifice your own interests for others?
More recently, some have additionally defended the existence of Other-Sacrificing Options: A permission to bring about an impartially suboptimal outcome because it discounts your adversaries' interests. 20ese options concern permissions to act or not act in certain ways toward the moral opposite of your intimates: your moral adversaries.In dealings with your adversaries, you might sometimes be permitted to give lesser weight to their interests than to the interests of strangers.For example, you may permissibly give a stranger one unit of happiness instead of giving three units to your adversary.You might, for example, think that someone who is rude to you for no reason may be treated by you in this way.
One of the challenges for defenders of other-sacrificing options is to explain plausibly why the response of discounting our adversaries is not an impartial phenomenon.We might think that the agent-neutral notion of desert can explain why your bully's interests should be discounted.

Required Partiality
Debate about required partiality can arise in the context of impartiality, but various other entry points exist as well, ranging from the commonsense idea that 'we owe more' to people who are, in some sense, closer to us to the justification of political obligation.21

Required partiality concerns
Associative Duties: Requirements to act in certain ways in virtue of the special relationships in which you stand to your intimates. 22sociative duties are also sometimes referred to as 'special obligations' or 'duties of partiality'.In contrast to options, they are 'presumptively decisive' agent-relative considerations and require you to be partial to your intimates.On most conceptions, to fail to discharge your duties is to wrong your intimate. 23sociative duties are duties you owe to people who are close in virtue of some special relationship (be it a personal history, affection, love, or group membership), such as your children, parents, siblings, spouses, friends, colleagues, peers, compatriots, or even members of the same species as you.Going beyond general or natural duties to others, these duties may require that you do distinctive things for your intimates, give extra weight to them, or do more for them of what you would have to do for strangers.

The Strength of Associative Duties
Associative duties not only are more extensive than general duties to other people but also have much greater strength.The strength of associative duties can be understood in terms of three further features.24 (1) Stringency First, associative duties are not so easily overridden by considerations of cost to oneself as positive duties to others.For example, although you may be expected to bear some cost to provide assistance to a stranger, you may be expected to bear greater costs to provide comparable assistance to your brother or child.So, you might have to risk your life to save your child from drowning in a river, but you might not be required to incur the same risk to your life if the person at risk is a stranger.
(2) Gravity A second feature of associative duties is that they often take precedence over general duties in cases where the two conflict.If both a sister and a stranger need assistance, and you can help only one of them, you might be required to help the sister, even if you would have been required to help a stranger who was the only person in need of assistance.This holds even if the sister's need is less urgent than that of the stranger.
(3) Overridingness Third, associative duties are sometimes considered to have a different threshold at which they can override other duties.For example, the threshold at which a positive duty can override a negative duty is sometimes lower if the positive duty is to an intimate than it would be if the positive duty were to a stranger.For example, you might sometimes be required or permitted to harm someone in order to provide a badly needed benefit for your sister or child, though this might be wrong for you to do in order to provide a comparable benefit for a stranger.Conversely, the threshold at which a positive duty can override a negative duty is sometimes higher if the negative duty is to a family member than if it is to a stranger.It may sometimes be permissible to inflict a lesser harm on one stranger to prevent a much greater harm to another stranger when it would be wrong to inflict the same lesser harm for the same reason on your sister or child.

Two Objections to Associative Duties
There are two important objections to associative duties.Both highlight how a commitment to partiality can come into conflict with other ideals, such as freedom or equality.

The Voluntarist Objection
The voluntarist objection is directed at the special duties that are supposed to arise from non-consensual actions. 25It is, in effect, an objection on behalf of the individual who is supposed to be bound by associative duties.As the grown child of my parents, I might be required to impose much greater cost on myself to help them than the cost I would have to impose on myself to help a stranger, yet I did not choose to become their child. 26Associative duties, if they can apply to agents absent any relevant consensual act, would, according to the voluntarist, therefore constitute unreasonable constraints on some of the individuals they bind.
If we are not prepared to deny that children have associative duties towards their parents, there are three main ways to reply: (i) attempt to show that all the associative duties in question really are contractual or promissory in nature; (ii) show that, while not all associative duties are contractual, they nevertheless resemble contractual duties in a relevant respect; or (iii) bite the bullet and maintain that, in some cases, associative duties can arise absent an agent's consent.27

The Distributive Objection
The distributive objection focuses not on the unreasonable burdens that being bound by associative duties seem to impose on an individual but instead on the objectionable feature of supplying participants in special relationships with benefits that appear to be unreasonable. 28This is, in effect, an objection on behalf of the individuals who neither are bound by nor benefit from associative duties.
By way of illustrating the distributive objection, imagine a three-person society that consists of A, B, and C. Imagine that all three owe general duties toward each other.Suppose now that A and B start to form a special personal relationship.This relationship brings them a number of benefits.For one thing, participation in the relationship may itself be inherently rewarding and contribute to the participants' flourishing.The relationship also changes the structure of their duties, providing A and B with associative duties to one another.Once A and B enter into a special relationship, C is disadvantaged in at least in three ways: (1) where A and B might have benefited C supererogatorily, their focus will now be on each other; (2) when it is impossible for them to perform their general duties both to each other and to C, it is C who will always lose out; and (3) sometimes the associative duties that A and B owe to one another will override their general duties to C.
The distributive objection thus sees associative duties as providing additional advantages to people who have already benefited from participation in rewarding groups and relationships, and it views this as unjustifiable when the provision of these additional advantages is to the disadvantage of those who are needier either because they are not themselves participants in rewarding groups or relationships or because they have significantly fewer resources of other kinds to begin with.
One response to the distributive objection is to insist that upon acquiring associative duties, you do not owe less to others but simply more to your intimates.So, for example, in addition to your duty of benevolence, you might have additional duties to your intimates.But discharging these duties need not mean that you are required to do less for strangers. 29

Justifying Partiality
The literature on justifying partiality can be confusing.One reason is that it is not always clear whether the aim is to justify permissions, requirements, or both.A second reason is that the debate about justification takes place at different levels.Before we discuss the respective grounds of associative duties and options, let's consider two 'high-level' frameworks within which partiality can be justified.

Underivative and Reasons-based Approaches According to
Underivative Approaches, agent-centred options and associative duties are grounded in independent and underivative permissions and duties (as opposed to normative reasons). 30ese approaches justify partiality by appeal, not to more basic factors such as normative reasons or 'ought other things equal', but to fundamental notions that are themselves permissive or deontic.For example, a version of this approach might say that you have a prima facie duty of beneficence to promote the interests of everyone impartially, but in addition to this duty, you have underivative permissions to either pursue or not pursue your own interest.In some cases, your permissions outweigh your duty of beneficence, granting you options not to bring about the impartially optimal outcome.As for required partiality, underivative approaches might hold that your associative duties are strengthened general duties of beneficence. 31

According to
Reasons-based Approaches, options and associative duties are grounded in normative reasons.
The starting point of this approach is typically to say that there are two fundamentally different kinds of reasons: agent-neutral reasons and agent-relative reasons. 32In the case of permissible partiality, options are then grounded in the conflict between agentrelative and agent-neutral reasons and their incommensurability.For example, according to one view, you have agent-neutral reason to do what is best from the impartial perspective and agent-relative reason to do what is best from the agent-relative perspective.You have a permission to act impartially suboptimally in some situation just in case your agent-neutral and agent-relative reasons favour conflicting courses of action.When this is the case, you are permitted to take on either perspective and do either what is agent-neutrally optimal or what is agent-relatively optimal. 33 the case of required partiality, reasons-based approaches ground associative duties in 'presumptively decisive' agent-relative normative reasons to act in special ways with respect to our intimates.These approaches must then offer a story about why certain grounds give rise to such distinctively strong reasons.For example, as we shall see below, we might think that certain valuable personal relationships based on mutual intimacy and love can give rise to reasons of this sort.

The Reductionism-Non-reductionism Debate
A second foundational issue that is orthogonal to the above discussion is whether partiality can be justified reductively by appeal to more basic moral principles or must be justified non-reductively by appeal to sui generis options or duties.The importance of this issue goes back to the stakes mentioned in the introduction, above, about the correct moral theory.If partiality can be accounted for derivatively, then we might be 31 Hurka (2016) defends this view in Section 5. 32 Why are there two distinct kinds of reasons?According to one view, agent-relative reasons are grounded in the intrinsic value of permitted projects (Nagel 1989) or in the more general value of having 'space' in which to pursue our personal lives (Scheffler 1982). 33See Bader (2020) for a recent defence of a decision-theoretic form of dual-ranking actconsequentialism based on this framework of normative reasons.Other versions of dual-ranking act-consequentialism are developed by Sen (1982) and Portmore (2003).
able to keep our commitment to impartial theories such as consequentialism; however, if partiality must be accounted for non-derivatively, we might face a seismic shift at the level of normative theorizing.
The reductionist-non-reductionist debate is more prevalent in the arena of required than of permissible partiality.The reason for this is that derivative approaches to the existence of options do not really account for options but rather debunk them.For example, according to Railton's (1994) self-effacing view, there exist agent-neutral reasons to promote the good impartially, but it turns out that, as a matter of fact, being partial to our intimates will accomplish what is impartially best. 34This, however, does not yield options-just a requirement to be partial when that is what impartial morality demands.
For required partiality, there are two camps available.According to Reductionism, associative duties are reducible to more basic special duties.This camp says that associative duties are reducible to more basic duties that morality typically countenances and that are themselves grounded in more basic moral principles.These may have to do with the moral significance of i) promises and contracts, ii) compensating benefactors by expressing gratitude, iii) providing aid to those we have made particularly vulnerable to our actions, or iv) a general duty of beneficence. 35

According to
Non-reductionism, there exist sui generis associative duties.This camp says that some of our associative duties are irreducible to other, more basic moral duties in the way that the reductionist suggests.For example, some non-reductionist views maintain that intimate personal relationships between two people may ground sui generis duties of love. 36prominent starting point for non-reductionism proceeds by appeal to the phenomenology of being partial.Versions of non-reductionism often say that, in paradigmatic instances of legitimate partiality when we are discharging our associative duties, our motives are, at the very least, an indication of the ground of associative duties. 37Others emphasise the intrinsic value of harmony between our motives and our reasons and argue that self-effacing theories that maintain a discrepancy between the two cannot be appropriately action-guiding. 38e debate between reductionists and non-reductionists is at an impasse. 39Reductionists typically flat-out deny that what motivates our adherence to our associative duties must also ground its justification.Others may suggest that our motives indicate that our existing basic moral duties ought to be discharged to a greater degree. 40Progress could be made by better distinguishing the content of our associative duties from the content of general duties to determine to what extent what we owe our intimates cannot be captured reductively.

The Big Three
The previous sub-section considered partiality at the more general level of inputs and outputs, but there is also the further substantive question: what are the things that give us reasons or underivative permissions/duties?What are the grounds of partiality?Much of the literature addresses this question.
Three main contenders are available: the Project View, the Relationship View, and the Individuals View.

According to the
Project View, partiality is justified by appeal to the special value of ground projects.
Bernard Williams (1981) has famously suggested that at least some of an agent's ground projects are fundamental components of the agent's identity and give meaning 37 See, for example, Keller (2013), p. 27: 'It would be depressing if in paradigmatic cases of partiality, we are systematically misperceiving our reasons.[…] It is difficult to see what sort of philosophical argument could convince you that when you act out of your special concern for your intimates, the things that really provide you with reasons are different from the things that seem to provide you with reasons.Stocker (1976), pp. 453-455, and(1998), p. 173. 39 See Wallace (2012), p. 183, who notes that '[...] there is, in my view, no knock-down argument in favour of the non-reductionist position.'For a response, see Owens (2012). 40Bazargan-Forward (2018) develops an account along these lines.Hurka (2016) argues that love can be a sui generis strengthener of basic moral duties.
to the agent's life. 41Projects must involve a certain amount and continuity of goaloriented agency, since they would not otherwise be distinguishable from mere preferences, desires, or wants.Thus, for you to have a project with central importance for your life, it must necessarily be the case that certain of its constitutive elements are sufficiently pervasive within your life.
One challenge that the Project View faces is that it cannot account for associative duties. 42If your project is to play the drums, then you may permissibly do so instead of doing other impartially better things, but it does not seem that you would be doing something wrong if you decided to give up the project.In response, the Project View can say that some joint projects may generate duties and that many of our personal relationships can be construed as such joint endeavours. 43Alternatively, we might go fully disjunctive and restrict the Project View to permissible partiality alone.On this view, associative duties are justified in reductionist fashion by appeal to other more basic duties (see section 5.2); for example, having possibly put someone in a vulnerable position with regard to you may create certain duties on your part toward them. 44cording to the Relationship View, partiality is justified by the appeal to the special value of personal relationships.
There are different versions of the Relationship View.Some say that personal relationships are valuable (or valued) for their own sakes and that this fact explains why they can justify partiality (permissions, duties, or both); 45 others appeal not to the special value of relationships but maintain that they have inherent and irreducible reasongiving force. 4641 Williams (1981) pioneered the Project View.Other writers who have defended it include Wolf (1982, 1992), MacIntyre (1984), Sandel (1982), Stroud (2010), and Betzler (2013).Sandel and MacIntyre focus particularly on commitments to the political community that has formed one's identity and to which one owes one's allegiance. 42See Scheffler (2004), pp. 258-260. 43 Stroud (2010), pp. 143-149. 44 Wolf (1992), p. 247, gives a defence along these lines.For critique, see Keller (2013), Ch. 2.
The Relationship View has strong commonsensical support, but it also faces challenges.One worry is that in paradigmatic cases where we show special concern for our intimates, it does not seem that we are thinking about the relationship but rather about the individual to whom we are partial.This might suggest that the Relationship View does not fully accommodate the phenomenology of partiality. 47One way to defuse this worry is to draw attention to the fact that even the Relationship View can still make intimates-the persons to whom we are partial-the focus of valuing. 48cording to the Individuals View, partiality is justified by appeal to the special value of individuals.
The Individuals View says that partiality is justified by the individuals with whom our special relationships are shared. 49Rather than acting as reasons for treating those persons better than others, our relationships act as enablers-that is, as background conditions that explain why some facts count as reasons for a given agent. 50e main challenge for the Individuals view is that it seems to accomplish too much by explaining partiality away.It does not seem able to explain why we should give moral priority to our intimates over strangers but instead denies that strangers have claims on us in the first place.This might seem to 'throw the baby out with the bathwater'. 51 response to some of the worries mentioned above, we might also adopt hybrid or intensifier views.These views justify partiality by appeal to, for example, a combination of the reason-modifying role that our projects or personal relationships can play. 52

Conclusion
Let me conclude with some a few remarks about the future of the ethics-of-partiality debate.
One growing area of research concerns the normative transitions of personal relationships.Most work on personal relationships to date has focused on their justification and distinguishing those relationships that can give rise to partiality versus those that cannot.However, often personal relationships are not static but dynamic and evolve over time: they can grow and flourish, stagnate, deteriorate, come to an end or, as we see below, change in their valence.Relevant questions in this debate are: What are different types of transitional processes?What is the normative significance or breakups, divorces, or other friendships 'turned sour'?
Another area concerns relationships in the digital age and with artificial intelligence.Are relationships with robots on social media relationships of the sort that could justify partiality? 53How do they relate to other paradigmatic relationships?And how can they be justified?
Beyond the issue of whether and how we should show display partiality in terms of our actions, there are also underexplored questions concerned with epistemic partiality: are there forms of epistemic bias that are required by certain personal relationships? 54f so, which types of relationships require epistemic partiality and why?Lastly, as I mentioned previously, recent scholarship has turned to the negative mirror image of partiality to our intimates: the realm of negative partiality to our moral adversaries. 55Negative partiality raises various new issues in the ethics-of-partiality debate: What forms of negative partiality exist?Are there permissions to discount our adversaries, and could there also be duties to do so?Can 'the big three' account for negative partiality?What is the relationship between positive and negative partiality?Can you favour your friends as much as you can dis-favour your enemies, both normatively and epistemically?I hope to have left you with an appetite to dive in. 53See Jeske (2018) for a recent discussion of friendship and social media. 54See Keller (2004) and Stroud (2006 on epistemic partiality. 55 See Brandt (2020), Lange (2020), and Brandt and Lange (manuscript) on negative partiality.See also Owens (2012), pp.111-13 on rivalry and Blundell (1989) and Kolodny  (2010), section 4, on harming enemies.