Acting on Behalf of Another

Abstract This paper provides an analysis of the phrase ‘acting on behalf of another.’ To do this, acting on behalf is first distinguished from ‘acting for the sake of another,’ the latter being a matter of other-directed motivation, the former of what we call ‘normative other-directedness’—i.e., acting on the claims and duties of the other. Second, we provide a distinction between two kinds of acting on behalf of another: representation as other-directedness plus normative replacement, and normative support as other-directedness without normative replacement. Third, the paper offers conditions of appropriateness for both types of acting on behalf.


Introduction
We encounter claims that something is done "on behalf" of another person, several other persons, or a group of persons, on an everyday basis. "When you see granny, greet her on my behalf" is as natural a request as speaking up on behalf of victims of oppressive structures and demanding compensation on their behalf. Philosophers, too, use this phrase a lot in a variety of contexts. They discuss political representation and legal trusteeship as paradigmatic instances of speaking and acting on behalf of those represented or entrusted by (Pitkin 1967;Miller 2014;Markovits 2014;Salkin 2020). Other philosophers discuss reactive attitudes that are felt on behalf of a victim or on behalf of the moral community (e.g., Strawson 1974, Darwall 2006, Wallace 2019 as well as, more specifically, the possibility of forgiveness on behalf of a victim of wrongdoing (e.g., Radzik 2010). In yet another context, theorists stress that empathy can be described as an emotion felt on behalf of another (e.g., Maibom 2007Maibom , 2009Prinz 2011).
Despite its ubiquity, the meaning of the phrase 'acting on behalf of another ' has not yet been thoroughly analysed in the philosophical literature and, indeed, what is meant by it is not intuitively clear. An exception is the important work done on political representation and fiduciary relationships that we are also going to consider in more detail in the paper. But, as we shall try to illustrate in what follows, acting on behalf does not simply come down to representation. Agents also act on another's behalf when they speak up for her or defend her rights-when they, as we shall say, offer normative support to another person. Both normative support and a specific form of representation (that we will call normative representation) are captured by 'acting on behalf of another,' and, as we will show, they share the feature of normative other-directedness while also differing in relevant respects.
The ubiquity of the phrase and its unacknowledged ambiguity warrant a philosophical analysis of what acting on behalf of another can mean, thus providing a preliminary step toward an ethics of so acting. In the following we will offer such an analysis and contrast normative representation and normative support as two ways of acting on behalf of another. The aim of the paper, then, is to disambiguate two notions of acting on behalf of another. We will use paradigmatic cases, such as fiduciary or proxy agency as well as speaking up on behalf of a victim, to delineate the relevant territory and draw the relevant distinctions. As these examples will illustrate, the two ways of acting on behalf of another are unified by characteristically involving what we call 'normative otherdirectedness'; an action on behalf of another requires a form of standing that is derivative from the claims, duties, and/or normative powers of another, and the person who acts on another's behalf has claims, duties, and/or normative powers that are derivative from those of the party on whose behalf she acts.
We proceed as follows. After distinguishing acting on behalf of another from the related cases of 'acting for the sake of another' or 'acting in her interest' (section 1), we argue that there are, centrally, 1 two different ways an agent can act on behalf of another (section 2): she can act as a normative representative (section 2.1) or she can offer a specific kind of normative support by responding to the other's claims on her behalf (section 2.2). Both are described by saying that an agent acts on behalf of another, and, as we will show, they possess a common core in the guise of normative other-directedness. What distinguishes them is that normative representation, unlike normative support, involves what we call normative replacement: while in both cases someone responds to the claims of another person, only in normative representation does she take up the other's position in the relevant normative framework and hence stands for the other; in cases of normative support she supplements the other, she stands by the side of or with the other in the relevant normative framework. We thence analyse normative representation as normative otherdirectedness plus normative replacement and normative support as normative other-directedness without normative replacement. To finalize our analysis and bring the distinction between both kinds of acting on behalf of another into sharper relief, we offer a first discussion of preconditions and conditions of appropriateness of acting on behalf of another (section 3).

Acting for the sake of another and acting in the interest of another
The Oxford English Dictionary's first entry under 'behalf' explains "on behalf of someone" as "in the interests of a person, group, or principle," and Merriam-Webster also has an entry explaining 'on behalf of someone' as 'for the benefit of someone' or 'in support of someone.' 2 In other words, 'on behalf of' is often used synonymously with 'for the sake of' of 'in the interest of.' Collecting money on behalf of the poor or campaigning on behalf of a friend who is running for office both are cases where one acts for the sake or in the interest of another. However, these cases do not involve acting on behalf of another in the same way in which political representation or speaking up on behalf of a crime victim does. As should become clear below, what the political representative does when 1 In this paper, we focus on what we take the be the two important paradigmatic ways of acting on behalf of another. There might well be others, and we're, in fact, hoping that the distinctions we draw here will help to further fine-grain our understanding of the various ways in which we can act on behalf of others.
An anonymous reviewer has pointed out to us that sometimes a distinction is made between acting "on behalf" of someone and acting "in her behalf." Indeed, sometimes acting in someone's behalf is understood to mean what we describe as acting for her sake or to her benefit whereas acting on her behalf is, by contrast, understood rather along the lines of representing her. However, writers and dictionaries differ as to whether there really is an accepted distinction between the two phrases and the formulation "in behalf of" seems not to occur anymore in British usage (see, e.g., Soanes and Stevenson 2005, 148, Merriam-Webster online https://www.merriam-webster.com/dictionary/behalf#usage-1, Cambridge Dictionary online https://dictio nary.cambridge.org/dictionary/english/behalf, Wiktionary https://en.wiktionary.org/wiki/on_behalf_of#English, all accessed June 13th 2022, 6:24 CET. Given the unclear linguistics, we will not use 'in behalf.' Nonetheless we are very grateful for the reviewer's pointer which, among other things, drew our attention to the difference between 'acting for the sake of' and 'acting in the interest of.' speaking on behalf of her representees or what the person does who speaks up on behalf of a crime victim are not adequately described as acting for the sake of those others or in their interest (although it may also be this).
In this section, we want to distinguish the three cases of acting for the sake of someone, acting in her interest, and acting on her behalf. All three can be described as actions done for another, though in different senses.
What, then, is acting for the sake of someone? Consider the following two examples: (A) John and Sally sit at a table working. Sally drops her pencil, possibly not even noticing it. John, being a nice and helpful person, picks it up and hands it to her.
We may say that John picks up the pencil for Sally's sake. What John is doing here is responding to a kind of interest or need of, and doing something to the benefit of, Sally, and we can take him to be doing so because he is a nice person and cares about Sally. Contrast (B): (B) Chris and Sally are working, and Sally drops her pencil. Chris then picks it up because, as it happens, he will have a favour to ask of Sally and wants to put her in a good mood. He acts out of egoistic motives.
We would not say that Chris acted for Sally's sake. Rather, Chris is acting for his own sake and the situation happens to be such that meeting Sally's interests is what Chris's egoistic motives require.
What these examples teach us is that acting for the sake of another is a matter not of performing a certain type of action or of meeting the other's interests but a matter of the right kind of motivation. You act for the other's sake if your action is motivated by concern for the other's interests.
If acting for the sake of another is, primarily, a matter of being motivated by concern for another, and not simply of a kind of action, it will be compatible with but independent of acting on another's behalf. We will argue below that actions on behalf of another are normatively derivative from the other's claims, duties, or normative powers, and that there are two fundamentally different kinds of such actions. However, whether an action is on behalf in either of the two ways is independent of whether the action is done out of altruistic motives, i.e., for the sake of another. That is, one can act on behalf of another without being concerned with benefitting the other. Acting on behalf of another, as we understand it, describes the normative underpinnings of one's so acting understood in terms of where one's standing to do so derives from, independent of any motivational setup.
With this understanding in place, another possibility comes to mind that also falls under the dictionary understanding of 'on behalf' but is different from the phenomenon we are interested in. This is acting in someone's interest or to her benefit whether or not one does so for her sake. It is plainly possible to collect money on behalf of refugees but not so much for the sake of refugees as for the sake of your partner whose political career will benefit from your charity work, and in (B), Chris acts to Sally's benefit but not for her sake. You can adopt someone's benefit as your aim without being motivated by an altruistic concern for her. This is also distinct from acting on the other's behalf in the sense we are interested in. In fact, it is a conceptual possibility of our analysis that one can, in acting on behalf of another, act against the interests of another.
What matters for our purposes is that whether an action is done out of altruistic motives and whether it is done to another's benefit are questions orthogonal to whether the action is done on the other's behalf. Actions for the sake of another are motivationally, but not per se normatively, otherdirected; the agent's standing to perform them is not usually derivative from the other's claims. The same goes for acting in another's interest or to her benefit which also does not imply derivative standing. We will therefore restrict our understanding of 'on behalf of' to cases of normative otherdirectedness as defined in the following.

Acting on behalf of another
What is it to act on behalf of another person, then? We argue that there are two different ways A can act on behalf of B: A can act (1) as a normative representative, or A can (2) offer normative support by responding to the claims of B on B's behalf. What unifies these two ways of acting on behalf of another is that they involve what we call 'normative other-directedness': 3 Normative Other-Directedness: A's action X is normatively other-directed if and only if the claims, duties, and/or normative powers A acts on in X-ing are derivative from B's claims, duties, and/or normative powers, and acting on such derivative claims, duties, and/or powers is constitutive of X-ing.
With this in mind, let us begin by taking a closer look at (1): ubiquitous instances of acting on behalf as a normative representative.

Normative representation
Normative representatives can be said to act on behalf of their representees because they, at least to some extent and regarding at least some domains, take up the representees' normative position and act from it. Representatives act on behalf of their representees by 'normatively replacing' them: Normative Replacement: A normatively replaces B within a domain D iff (some of) B's claims, duties, and normative powers within D are transferred to A such that they can be acted upon by A but not B (for the time of the replacement), while B remains the primary bearer of the claims, duties, and normative powers in question such that A's standing to act on them remains derivative.
Instead of B voicing her claims, enacting her powers, or performing her duties, A takes up this role and does it in B's stead (within a certain scope and with a certain degree of discretion). That is, A moves from being a third-party to being a relevant second-party to a given normative situation-from bystander to representative. Depending on context, this has different consequences for the validity and adequacy of A's representation of B, B's standing to recall or instruct her representative, and the norms subsequently governing A's actions. In all cases, however, B will be partly replaced by A within the normative landscape. It follows that some actions B could casually have performed in the absence of a representer will now be impossible for B because it has become A's rather than B's part to perform actions of this type. At the same time, the representative's actions are always directed at and dependent on the representee because they involve the exercise of powers or the possession of duties or claims that are the powers, claims, and duties of the representee, transferred within the scope of representation to the representative. Normative replacement is therefore always normatively other-directed.
Let's consider a set of examples to get a better grip on the phenomenon of normative representation and its key characteristics, including legitimisation, partial and full normative replacement, as well as normative disagreement: Fiduciaries: Fiduciary relationships are paradigm cases of partial normative replacement. Take, for example, the lawyer-client relationship, where the latter in effect acts as the legal fiduciary of the former (within a given domain). 4 The lawyer normatively replaces the client in 3 We derive this term from the notion of other-directed emotions or motivations which is often found in the psychological and philosophical literature on empathy (e.g., Darwall 1998;Aaltola 2014;Steinberg 2014;de Vignemont and Jacob 2016). It is used to pick out emotions that are intentionally directed at another person or her mental states and motivations that aim at satisfying another person's desires or interests. As we have seen, 'for the sake of' is a way of picking out motivationally otherdirected actions. 4 For a detailed discussion of fiduciaries as representatives, see the insightful discussion of Miller (2018).
the relevant area and acts in the client's stead: for example, when you entrust your lawyer with managing your property or conducting negotiations in your stead. 5 While the lawyer is certainly accountable to her client and can lose her standing to represent her, all else being equal, the lawyer replaces her client in the relevant area (e.g., a certain legal domain). 6 To establish such a relationship, usually explicit authorization in the form of mutual consent (usually by signing a contract) is necessary. This authorisation of a lawyer as one's fiduciary involves the transfer of certain powers, claims, and duties to her, instating her as one's partial normative representative. This is a case of normative replacement insofar as it is now the lawyer and not the client any longer who is positioned to act on these powers, claims, and duties. For as long as my lawyer is my authorised trustee in charge of my property, she can make transfers and sales, but I cannot; and for as long as she is authorised to negotiate in my stead, I will not be able to negotiate directly for myself. The lawyer is acting in my stead, and I can return to my normative position only by withdrawing her authorisation. The lawyer's actions, thus, are still normatively dependent on the client, but the lawyer's powers are no longer at the client's discretion; the client cannot, without revoking her lawyer's powers, enact them herself. Quite to the contrary, and in contrast to the below cases of normative support, as the result of the relevant mode of authorisation, a lawyer can have the relevant power to act on behalf of her client irrespective of whether the client approves or disapproves of this power at a given time. Moreover, and implicit in these considerations is the further idea that, while having to normatively orient themselves at claims and interests of their clients, 7 lawyers themselves possess a certain level of discretion as to how best represent and hence act on behalf of their client's claims. The lawyer can now speak on her client's behalf, do their business, enact her powers, consent, or demand what is her client's due, on her behalf. More so, the lawyer might now also carry out relevant duties on behalf of the client, e.g., to be present at certain meetings, or fill in reports. She can even use her representative powers to place the representee under obligations, e.g., when signing a contract on her behalf. All of this suggests that there can be normative disagreement between the representative and the represented. What is said here carries over to other fiduciary relationships, including prominently parents acting as the fiduciaries of their minor children.
Proxies: Similar to cases of fiduciary agency are proxy agents. Typical examples of proxy agents are spokespersons: 8 they speak in their representee's stead and questions are directed to the spokesperson rather than their natural addressee. Another case familiar from the literature is proxy voting (Alger 2006, Green-Armytage 2015, where someone's voting power may be transferred to a proxy who is then authorised to cast a vote in her stead. In both cases, the proxy acts in the representee's stead and thus partly normatively replaces her-the duty to answer questions is transferred to the spokesperson, as is the power to reject certain questions, and the power to cast a vote is transferred from the representee to the proxy; the former is not any longer in a position to cast a vote. Proxy voting is a particularly neat case of what we have in mind with normative replacement: the power to cast a vote is quite literally transferred from one agent to another, but the proxy's power remains derivative from the primary standing of her representee to do so on her own behalf. The proxy, when acting on the representee's behalf, thus acts in her stead, and normatively replaces her with regard to certain powers, duties, or claims. Political Representation: The case of political representation is partly similar, though often more complex than proxy agency. 9 Political representatives, like proxies and fiduciaries, act on their representee's behalves by enacting powers, claims, and duties that are derivative from their representees'. Thus, political representatives can cast votes in parliament, speak at political congresses, or sign international treaties on behalf of their constituents, and they can only felicitously do so because they are representing their electorate in these domains. Political representation thus also involves an authorisation of some to act on behalf of others in some area, and the represented others consequently have no direct say in these areas (e.g., no one is going to negotiate a treaty with a people, one is going to negotiate with the people's representatives). What makes this case more complex than proxy agency is that the powers, claims, and duties of political representatives do not directly mirror those originally held by the representees. No individual citizen ever had the power to sign an international treaty or cast a vote in parliament. Nonetheless, at least on traditional views of political legitimacy, the representatives' standing is derivative standing, even though the (e.g., contractualist) transfer of powers from citizens to representatives involves the creation of new or changed powers, claims, or duties. Without going into this complexity-a task we leave to substantive political theorising-we can therefore say that political representation as commonly conceived also involves a partial normative replacement of primary claim holders by representatives. This is the sense in which political representatives can be said to be acting on their representee's behalf (and not [only] for their sake or in their interest). 10 These, and more, examples involve acting on behalf of another in the sense of normative representation. What unites them is that the representative, to some extent and regarding some subjects, replaces the representee in the normative space: the representative becomes the one to voice and execute the representee's claims, to respond to her duties, and to exercise her normative powers. The representative is acting on behalf of the representee because her actions are normatively other-directed-the standing to engage in them is derived from the representees' claims, powers, and duties.
What we have seen from the examples, however, is that representation can take very different forms. Often, but not always (think of parents representing their minor children) must representation be directly authorised by the representee, and the representee will have discretion over whether and whom to authorise to represent her and within which scope. The representee may or may not have the standing to call back her representer. Likewise, the representee's wishes may or 9 Thanks to an anonymous reviewer who has drawn our attention to these complexities. 10 We are of course aware that political representation has received extensive treatment in the literature. See, e.g., Pitkin (1967) who also devotes an entire chapter on the notion of 'acting for.' As such, we are not in this section assuming to uncover new conceptual territory, but to build on the literature on representation and, more importantly, place it in relation to other forms of acting on behalf that have not yet been analysed. For other important work on (political) representation, see e.g., Mansbridge (2003), Saward (2006Saward ( , 2008Saward ( , 2010, Urbinati (2000). may not be binding for the representer or she may have more or less discretion. The scope of representation can be global, as in the case of parents and their very young children, or more local, extending to merely a limited subject matter, as is typically the case for lawyer-client relations or proxies. These important differences do not prevent a common analysis of these cases: the representative's actions are always normatively other-directed and the representative (partially) normatively replaces the representee. With this in mind, we now turn to a different way of acting on behalf of another.

Normative support
While representation means appropriating another's normative position by normatively replacing the other, there is another kind of normatively other-directed agency that also goes by the label 'acting on behalf of another.' We call it 'normative support.' An agent normatively supports another by normatively backing up in actions (often including speech) another's claims. That is, the claims of another are explicitly or performatively supported, affirmed, or enforced by the actions of someone who is not herself the claim bearer but is rather asserting and enacting the claims of another. She does so by affirming or enforcing these claims, especially where they are contested. Such actions are normatively other-directed because the normative power to press the claim and demand compliance derives from the person's claim that one normatively supports. Only in virtue of this claim can a supporter have the standing to, e.g., demand compliance with it. It significantly differs from representation, however, in that the claims bearer is not normatively replaced. Whereas a representative is acting on another's behalf by acting in her stead, normative supporters remain third parties. They support the other's claims, but do not replace or represent her. That is to say, normative supporters stand with but not for the person on whose behalf they act. Again, we will approach this category by offering a number of examples.
Neighbourly Disagreement: Anna, Bert, and Caterine are neighbours, each living in an apartment on the same floor of the same building. Anna habitually leaves her dirty shoes at her door outside her apartment in the common hallway. Bert is annoyed at the view and smell and confronts Anna. Anna and Bert end up arguing heatedly about whether it is legitimate to leave one's shoes in the hallway. Caterine steps in. She shares Anna's position and demands Bert to accept Anna's right to leave her shoes outside, thus speaking up for Anna's claim.
In doing so, Caterine is not merely stating a fact. Rather, she is taking sides and playing a part in enforcing Anna's rights and demanding respect for them. Inasmuch as the conversation Anna, Bert, and Caterine are having is not purely academic (about the "juridical" fact of the matter) but a practical disagreement and a negotiation of claims, Caterine's actions are normatively guided by Anna's claims. 11 Only on the basis of those claims can she demand acceptance from Bert. Therefore, Caterine's actions are normatively other-directed. They serve the affirmation and enforcement of another's claims. However, Anna is not normatively replaced. It is not the case that when Caterine enters the conversation speaking up on Anna's behalf that Anna has no role to play anymore. If Caterine were representing Anna, only one of them would be in a position to confront Bert. What happens in the example, however, is that Caterine comes to Anna's support. Caterine acts as a third party. She is not representing Anna and is not bound by Anna's instructions or interests. In addition, the fact that she finds herself on Anna's side rather than Bert's is not a matter of a representation relation but of whose (assumed) claims she supports. Thus, Bert now faces two opponents, rather than Caterine as a representative of his actual opponent, Anna. The latter would be the case if a lawyer were speaking up on behalf of Anna as her client, thus representing her. Lawyer and client do not act as distinct parties whereas Anna and Caterine, in the example, are two 11 More precisely, Caterine's actions will be guided by what she believes Anna's claims to be. parties who are united by their cause. Anna is defending her own claims, and Caterine comes to defend and back up Anna's claims.
Stealing: Dominique leaves her laptop in the seminar room while going to the bathroom. When Peter attempts to steal it, Dominique's fellow students step in. They call Peter out, demand him to leave the laptop where it is, and claim that "This is Dominique's laptop. No one may take it." They explicitly blame Peter for trying to steal it. When Dominique returns and notices what happened, a fellow student demands that Peter apologise to her. Dominique, too, demands an apology from Peter.
Again, bystanders express and enforce someone else's, namely Dominique's, claims, and again they do not do so by representing Dominique but by standing by her side. What this example shows is that normative support is possible both in the presence and the absence of the claims bearer. It also shows another feature of vicarious agency that is not normatively replacing: the students can demand that Peter apologise to Dominique-that is, they not only enforce Dominique's claim to her property, they also direct the normative consequences of the event to Dominique, not to themselves. That is to say, they demand of Peter what is owed to Dominique. Nonetheless, they do not speak for Dominique like a representative would. Dominique's return to the room does not make their demand infelicitous. Since they act as third parties, Dominique and her fellow students can independently of each other make demands of Peter based on Dominique's claims. 12 Racism in the metro: Kyle overhears how Sarah is racistly insulted by a man in the metro. He also notices that Sarah pretends not to have noticed, keeping her cool. However, Kyle considers the man's remark inacceptable and a severe wrong done to Sarah and does not want to let the man get away with it. From her behaviour he assumes that Sarah would rather let it pass. Nonetheless he loudly confronts the man about his remark.
Again, we can plausibly say that Kyle spoke up on Sarah's behalf. He confronts the man not only for being racist but specifically for the personal disrespect he expressed towards Sarah. The moral anger and indignation he experienced at the violation of Sarah's claims was what authors have described a vicarious resentment or resentment on Sarah's behalf (Strawson 1974;Wallace 2019). In doing so, Kyle did not demand on his own behalf that the man not insult Sarah, but did so on Sarah's behalf. 13 Notably, despite the fact that Sarah and Kyle disagree about the appropriate course of action and Kyle must even assume to be acting against Sarah's will, his action is nonetheless done on Sarah's behalf. This goes to show that normative other-directedness does not depend on agreement of both parties or approval by the claims bearer. Whether Kyle's action was morally appropriate given Sarah's (presumed) disapproval is a substantive question, and we can imagine this to be sensitive to the details of the case; there might not be a clear rule about the moral import of such disagreement. We have seen above that representatives, too, might sometimes be allowed to disagree with the represented without losing their status as representatives and sometimes also without acting inappropriately. This case is different, however. Where a representative normatively 12 A reviewer has pointed out to us that this example could also be read as a case of normative representation-the bystanders stepping into Dominique's position and speaking in her stead. We resist this interpretation and understand such an intervention as a case of normative support without normative replacement. To see why, note that, unlike for representatives, bystanders could not receive apologies on Dominique's behalf and their blame and demand will not turn infelicitous the moment Dominique returns and speaks up for herself. 13 Things can get more complex, and it is of course possible for Kyle, in addition to making a demand on behalf of Sarah, to demand that the man refrain from insulting others on his own behalf. As such, Kyle might demand on his own behalf that fellow members of the moral community ought not to be insulted by others. This is entirely compatible with the fact that he also demands this of others on behalf of his fellow members, i.e., Sarah. See below for a discussion of the metaethical conditions of normative support, including note 20.
replaces her representee, they do not act independently as two distinct parties-the only one bearing the relevant standing with regard to the action in question is the representative. In contrast, Sarah and Kyle remain distinct parties even throughout the acting-on-behalf. As such, the scene in the metro may plausibly evolve into an open argument between all three persons about how to appropriately treat each other. What matters here is not merely the standing of the person speaking on Sarah's behalf-Kyle-but in fact both the standing of Sarah to make normative demands for herself and Kyle's standing to do so on her behalf. For Sarah never ceases to possess her standing to make normative demands in the relevant situation by having transferred it to Kyle. Quite to the contrary, Kyle has standing to make these demands only because of, and in support of, Sarah's standing to do the same. A representative, by contrast, gains her standing precisely because the representee gives up hers.
Moral Anger and Third-Party Forgiveness: When discussing third-party forgiveness philosophers often do not clearly distinguish between bystander forgiveness and forgiveness that consists in the forswearing of anger felt on the victim's behalf as normative support. Bystander forgiveness occurs when bystanders observe wrongdoing and experience moral outrage but come to foreswear it and thereby forgive the perpetrator. In forgiving, they restore their own relationship to the perpetrator irrespective of the victim (this is what Pettigrove 2009 and Walker 2013 discuss as "third-party forgiveness"). These bystanders do not forgive on behalf of the victim because the moral anger they felt was not on the victim's behalf. This contrasts with cases where third parties have acted as normative supporters and have thus felt and expressed anger on behalf of the victim. If you experience moral anger on the victim's behalf, possibly confronting the perpetrator on her behalf, your aggression is based on the victim's claims-you are angry not so much about the transgression as for the victim. Subsequently, you can forswear this kind of anger if you judge that the victim's claims have been satisfied. This can happen when the perpetrator adequately apologises, makes amend, etc. Importantly, while your anger was on the victim's behalf, your forgiveness in such a case will not be on behalf of the victim. While your moral anger was guided by the victim's claims, thus felt and expressed on her behalf, foreswearing it, thereby forgiving the perpetrator, does not respond to the victim's claims. Rather, insofar as it is appropriate to forgive those who have made up for their wrongdoing, it is a response to the perpetrator and her claim to being forgiven. This will be possible even if the victim herself does not forgive, although presumably the victim's view will have a special import on the adequacy of third-party forgiveness (cf. Radzik 2010, 81). 14 The above cases exemplify a second meaning of acting on behalf of another, different from normative representation. Normatively supporting another and representing her have in common that they are instances of normatively other-directed agency: the agent who supports or represents another derives her standing from the other's claims and gains normative powers derived from the other person's. The structural difference, as we have seen, is that in representation, the represented is normatively replaced by the representative. Her position in normative space-the position from which 14 Third-party forgiveness as discussed here is significantly different from the controversial case of forgiving on behalf of a dead person. Philosophers sometimes discuss such cases of forgiveness on behalf of a victim who cannot forgive herself anymore because she is dead or unable to communicate. If someone were killed in a traffic accident caused by another's inattentive driving, the culpable driver might ask the victim's children for forgiveness. Theorists disagree whether the victim's children can plausibly offer forgiveness on behalf of the deceased person, but some consider it a plausible possibility (Griswold 2007;Hughes and Warmke 2017; for a differing view, see Walker 2013). We do not want to take a stand on this matter but merely point out that if one considers forgiveness on behalf of the dead a possibility, it involves a kind of representation, rather than normative support. If the culpable driver asks the victim's children for forgiveness on the victim's behalf, and if they (attempt to) offer such forgiveness, they are addressed as and (attempt to) act as the victim's representative: they "step into" her normative situation and act in the position of the person to whom the wrong was done. The controversial nature of these cases arises precisely because it is unclear whether such representation can be legitimate.
she acts on her claims, duties, and powers-is taken up by her representer. Representer and representee do not act as independent parties. Normative supporters, in contrast, enter the scene as and remain third parties taking sides based on the other party's claims. 15 They speak up on behalf of an affected party, but they aim not at representing her but at supporting her claims. Often, though not necessarily (remember Kyle), they will thus be speaking in addition to and alongside the affected parties. 16

Preconditions and conditions of appropriateness
In this final section, we attempt to bring out the difference between representation and normative support more by attending to their normative preconditions as well as to considerations that bear on their respective moral appropriateness. The questions thus are: What makes something a case of representation/support rather than attempted or pretended representation/support? And, independently, what makes a case of representation/support morally (in)appropriate? Let us note, however, that the following considerations are not meant to be exhaustive. We only take ourselves to begin to address the relevant normative and appropriateness conditions, and the main purpose of this section is to bring the difference between representation and support into sharper relief. To do so, consider one last example: Guilty Partner: Laura returns from work in a very bad mood. When her neighbour Alfred accidentally bumps into her in the staircase, she insults and even pushes him. Laura's partner Trish overhears the scene and feels ashamed, even guilty, for Laura's behaviour. She steps in, addresses Alfred, and says, "I am so sorry. I apologise." She also turns to Laura and says, "This wasn't right," and "You should apologise." In this scene, we can see Trish oscillating between two different roles. In parts of what she says to Alfred, she casts herself in the role of Laura's representative. When she apologises, she apologises on Laura's behalf, in just the way we would expect Laura to do. Being Laura's partner, Trish, at least to some extent, considers herself justified in representing her, offering an apology in her stead. However, when she condemns Laura's behaviour and tells her to apologise, she acts not as Laura's representative but as Alfred's normative supporter. She re-affirms Alfred's claims in condemning Laura's violations of them, and she speaks up on Alfred's behalf towards Laura, demanding an apology to Alfred.
With regard to each of Trish's utterances, we can ask two distinct questions: (1) Does it constitute a case of acting on Laura's behalf, and in virtue of what, or is Trish merely trying or pretending to do so, and (2) is her doing so (morally) appropriate? These questions bring out, respectively, the normative preconditions and the conditions of appropriateness of speaking on behalf of another and can be asked independently for cases of normative representation and normative support. We will pay most attention to (1) and only remark on (2) in passing.
Given what we have argued so far, question (1) can be split into two consecutive questions: What is required for an act to count as done on behalf of another, and, if it does, what is required for it to be an act of representation or support, respectively? As we have argued, what constitutes an act as performed on another's behalf is normative other-directedness: the act must constitutively involve acting on claims, duties, and/or normative powers, and these must be derivative from the claims, duties, and/or normative powers of another party. A precondition on acting on someone's behalf, therefore, is that the normative entities the act presupposes are actually in place, and that they belong to the other party. Consider Trish's apology to Alfred. It is a condition on successful apologies that the apologising party has standing to apologise (Smith 2008, 52-55). It is a first condition for Trish to apologise on Laura's behalf, therefore, that Laura has standing to apologise. Trish acts on Laura's behalf in virtue of performing an act that constitutively requires such reference to Laura's normative position. Two things here matter for Trish's utterance to be (a) a felicitous apology and (b) on Laura's behalf: that there is standing to apologise, and that it is, primarily, Laura's, not Trish's. Similarly, her demand that Laura apologise to Alfred is a demand on Alfred's behalf insofar as it is based on Alfred's claim. If Alfred had no claim to an apology, Trish's demand would be unfounded, and if it were not Alfred who had such a claim but Trish herself, her demand would not be one on Alfred's behalf.
Obviously, this does not suffice for the validity of either utterance. An additional precondition must be met, and it can be brought out by the following question: In virtue of what is it, or can it be, Trish's business to respond to Laura's duty to apologise or Alfred's claim to an apology? Here, normative representation and normative support come apart. Start with representation, for which the crucial issue seems to be whether Trish steps into Laura's normative position, i.e., normatively replaces her, legitimately. If in their relationship it is (explicitly or implicitly) agreed upon that the partners may, under certain circumstances, speak for each other, such representation seems legitimate. In contrast, Trish's apology to Alfred would not only immediately be voided if Laura demanded Trish to not speak in her stead. Even more clearly, if Trish had asked Laura whether she may apologise to Alfred on her behalf and Laura had denied this, any such apology could merely pretend to be on Laura's behalf. The necessary legitimation would be missing and by apologising nonetheless Trish would merely pretend it to be in place.
This precondition of representation is particularly obvious in institutionalised settings. Lawyers can represent their clients, but only if authorised to do so. If one were to act as another's lawyer without such authorisation, one would merely be pretending to be representing her. Consequently, anything one did as the other's representative would be infelicitous and have no normative import. Note, however, that direct authorisation by the representee is merely one way in which representation can be legitimised. Parents, for example, can speak on behalf of their children, but need no direct authorisation to do so; rather, a (partly institutionalised) practice is in place that determines who can legitimately speak on behalf of minor children. Whether authorisation is required, which form it must take, and which other ways of legitimising a representation relation there are will depend on the specific context and the norms governing it. Often, as in relationships, these norms will be implicit and vague, and subject to constant renegotiation (cf. Scheffler 2015). Nonetheless, if someone takes up another's normative position, speaking or acting as her representative, this always requires that one legitimately adopts this role where legitimacy is determined by the background norms governing the specific context. 17 17 We can leave it open here whether legitimacy in this sense must always be morally justified. There have been (and indeed are) societies where husbands could without authorisation or consent act as representatives of their wives, and certainly we now judge this to be a very unjust practice. However, given that the practice was in place, it gave rise to norms of representation that did, in fact, enable husbands to represent their wives in many situations (they could sign contracts on behalf of their wives, etc.). One way of looking at this situation is that therefore the husbands did really act as representatives although neither their particular representative agency nor the practice behind it were justified. A different way of analysing this might hold that because the practice was unjustified, husbands did not actually take over their wives claims and normative powers but were merely universally taken to be so. They were thus not really representing, although everyone assumed them to be. Whether legitimate representation is in this way generally disabled by immoral norms of representation is a question for further discussion.
As we have seen, representation often only covers a given domain: a lawyer acts as the client's representative only within a specified domain/range of actions, e.g., when enforcing certain contractually specified demands, etc. Normative representation is usually local and what we might call rangelimited, not global (with the important exceptions of parents acting on behalf of their children or caretakers representing severely impaired individuals). The typical range-limit of representation derives from the fact that representatives are usually only legitimated to represent their representees in certain domains. It is the degree of the legitimisation that determines the scope of representation.
What about normative support, then? What is it that makes "This wasn't right" and "You should apologise" a case of normative support-of speaking up on Alfred's behalf? What is not presupposed for normative support-we have already seen this in Racism in the Metro-is any kind of authorisation by Alfred, nor, indeed, that Alfred does not object to Trish speaking up for him. Guilty Partner does not tell us about Alfred's thoughts, but this does not seem to affect whether we can say that Trish is speaking on Alfred's behalf.
Rather, Alfred's claims are Trish's moral business in the sense that she has standing to make demands on his behalf in virtue of a more general standing members of the moral community have to make demands and enforce claims on each other's behalf. While the grounds of this standing may form an independent matter of investigation, it seems clear that we have it from the special salience of those cases where, due to special circumstances, we lack it. The standing to blame (and with it, the standing to demand) is undermined by hypocrisy or complicity and seems not to penetrate close personal relationships of others. These cases have received extended philosophical discussion. 18 The fact of their salience, however, points out how common the assumption is that third parties, merely by virtue of being fellow members of the moral community, normally do have standing to concern themselves with the claims of others. Normative support requires the truth of this assumption: for Trish to make a valid demand on Alfred's behalf, not only must Alfred have a valid claim, we must also grant Trish standing to concern herself with it.
To sum up: acting on behalf of another requires that the claim, duty, or normative power acted on indeed is in place and that it belongs to the person on whose behalf one acts. In addition, the other's normative position must be relevantly one's business. Normative representation requires a form of authorisation for a transfer of the representee's claims, duties, and powers to the representer. Representation must be legitimate. Normative support requires a form of standing for third parties to concern themselves with another's claim-a form of standing we commonly grant each other as members of the moral community.
Importantly, neither variant of acting on behalf of another requires, as a precondition, moral appropriateness. This is so even for representation, although it requires legitimacy. It is quite possible that someone can legitimately speak for another while at the same time she should not do so. Even if Trish can legitimately speak on Laura's behalf, whether she should do so depends on a number of additional considerations. Will she thereby really solve or mitigate the conflict? Will it be good for Laura? Might the better way to go be to talk Laura into apologising, rather than doing it for her? There may well be different views how to respond to the situation though, and even Laura and Trish might be disagreeing. While the representee's view has some bearing on what it would be appropriate for the representer to do, the strength of this consideration will differ from case to case, and often not be quite determinate. We have seen that some representatives, e.g., parents of young children, have substantial leeway and discretion in how to interpret the representee's position whereas others have significantly less leeway and limited discretion, e.g., clearly instructed lawyers. Between partners like Laura and Trish, where there are no explicit or even institutionalised rules for such cases, this will be sensitive to many details of the situation and their relationship and will at least to some degree also be subject of negotiation.
What is important to see, however, is that at least many cases of inappropriate representation are not thereby infelicitous. Parents or lawyers or even partners may well be blamed for how they represent their respective representees, but this alone does not change the fact of the matter that they are representing them and that when they act on their representees' behalf, the normative consequences of these acts will stand. Lawyers, e.g., can at least under some conditions sign contracts on their clients' behalf, and even if all agreed in retrospect that they should not have done so, the contract is not thereby voided. Their representative act was inappropriate, and they are blameworthy for committing it, but the act was not infelicitous. Inappropriateness renders representation infelicitous only if appropriateness and legitimacy depend (partly) on the same considerations. Thus, if a lawyer is authorised only to sign one specific agreement on her client's behalf but she exceeds this, this transgression of her responsibilities is both blameworthy and infelicitous.
We have already seen in Racism in the Metro that this is similar for normative support. Here, too, different considerations bear on the significance of acting on behalf of another and the wishes of the person on whose behalf one is acting do not by themselves always settle this matter. Especially in cases where the victim of a wrong is not fully aware of its nature or gravity, it may be justified to speak on her behalf even if she does not want one to do so.
What makes such disagreement possible and maybe also justified there in representation cases is that the representer is legitimately representing the representee, and that this legitimacy and the subsequent normative replacement need not require the representer to follow the representee's wishes or preferences. In normative support cases, it is not a legitimate representation relation that allows for justified acting in disagreement with the claim holder. Rather such disagreement is possible and often justified because the supporter acts as a third party. While she expresses the other's claims, she acts independently, from her own position and point of view.
Because she remains a third party, one might worry whether a normative supporter's actions are in the end really other-directed as defined above and can hence be genuinely understood as actions on behalf of another person? That is, one might now think that someone who normatively supports someone else is better understood as in fact responding to a general moral principle or rule determining what we owe to each other, and only in a second step, if at all, to the claims of a person that can be derived from it. Thus, when Kyle is seemingly speaking up on Sarah's behalf, Kyle is, in fact, not responding to any of Sarah's particular claims, but rather pointing to the violation of, and in turn demanding compliance with, a general moral principle that simply affects Sarah and in virtue of which Sarah can be said to have certain claims. After all, or so the worry might go, this is precisely what it means to remain a third party-to remain, so to speak, on the sidelines. As a result, normative support might best be understood not as other-directed but rather as 'principle'or 'ruledirected' action.
This apparent worry allows us to make an important clarification here concerning otherdirectedness: When someone offers normative support to another person, he or she is not merely responding to a general moral principle and its demands. Quite to the contrary, as pointed out above, a normative supporter like Kyle is responding to the specific claims of a particular person, and her standing to enforce this claim is derived from the claim holders primary standing. While this kind of support can be based only on the valid claims of another, it is genuinely other-directed insofar as it is derivative of the claim-bearers standing to press these claims on their own behalf. Thus, in supporting Sarah, Kyle is advancing Sarah's standing as a source of valid claimssomething that a third party who only points to a general moral principle cannot do. To make it more explicit, consider the following comparison between a normative supporter and what we might call an observing third party: whereas Kyle-or anyone else offering normative support-will point to the fact that it was Sarah in particular who has been wronged by the other person's racist insult and that the other should at least offer an apology to her, thereby responding to the disregard of Sarah's valid claim not to be insulted, an observing third party who is merely pointing to the violation of a general moral principle will not respond to the disregard of Sarah's valid claim and may simply come to the judgment that it was wrong to insult Sarah, period. In doing so, an observing third party whose actions are principle-directed might be doing something importantpointing out the violation of a general moral principle-and in doing so give credence to Sarah's claims, but he or she won't yet be offering normative support by, e.g., demanding that the perpetrator apologize to Sarah and hence be acting on behalf of Sarah and in support of her claims. 19 This, only a normative supporter can do.
Of course, this is not to deny that the observing third party can have concern for the interests of the affected party. In pointing out the violation of a general moral principle, the observing third party might well point out that the interests of the affected party that stand to be protected under the general moral principle have been violated. And while this concern for the interest of the affected party is very much toward the other, and maybe even for the sake of the other in that it expresses concern for the benefit and loss thereof of the other, it is not normatively other-directed, because the standing it presupposes does not derive from the standing of the affected party to press claims and exercise normative powers on her own behalf and can thus not be understood as being on behalf of the other person. Normative support is neither aimed at advocating general moral principles nor furthering the interests of another, but rather at enforcing the claims of another and demanding compliance with them. 20 Normative other-directedness and the very idea of normatively supporting another thus presupposes a relational understanding of the moral domain. According to relational understanding of the moral domain, our moral obligations are fixed by the valid claims of those with whom we share membership in the moral community and are embedded in what R. Jay Wallace calls, a "moral nexus" (Wallace 2019). Moral obligations, according to a relational understanding of morality, are thus best captured as directed obligations-that is, obligations which correspond with the claims of those individuals to whom they are owed. 21 Thus, to be morally obligated on the relational understanding of morality just is to be obligated to some particular individual. It is within the relational framework that we can locate and explain normative other-directedness and normative support, something that a nonrelational view, according to which moral obligations are fixed by general moral principles irrespective of the claims of particular individuals, simply can't do, thereby leaving unexplained what we take to be an important phenomenon.

Conclusion
We have argued that acting on behalf of another must be distinguished from acting for another's sake-the former being a matter of normative other-directedness, the latter of altruistic motivations -and that normative representation and normative support constitute two different kinds of acting on behalf of another. In representation, the represented is normatively replaced by the representative who acts for the represented whereas normative support does not involve replacement; rather, the supporter acts alongside the individual who is being supported and thus as a third party. Both instances of acting on behalf of another are unified by their normative other-directedness. We have identified legitimacy in the sense of being authorized as a representative according to the relevant set of norms as the central normative precondition of normative representation, whereas normative support rests on a form of standing we (defeasibly) have as members of the moral community. We hope to here have provided a clarification of what acting on behalf of another can mean and pointed out an important distinction to be kept in mind when philosophers talk about acting on behalf of another.