Duty of Assitance and Models of Responsibility

Our aim in this paper is to discuss the problem of overdemandingness of assistance-based models of responsibility in the discussion of world poverty. We shall take Singer’s approach as paradigmatic and argue that the degree of its overdemandingness creates an impasse in the theoretical discussion. This, we hold, can only be solved by combining more models of responsibility, e.g., the causal model. This means denying one of Singer’s main methodological presuppositions: the separation between factual and normative aspects in his argument. However, our strategy still presupposes a background of assistance-based models, which is only to be specified by other models. With this we will show that the accusation of implausibility raised against Singer’s approach is not a matter of the content of the duty it imposes. Rather it is a matter of the moral force of the rationale provided by it. Key-words: Duty of Assistance; Models of Responsibility, Peter Singer, Poverty Resumo: Nosso objetivo neste artigo é discutir o problema da demanda excessiva colocada pelos modelos de responsabilidade de assistência na discussão sobre pobreza mundial. Tomaremos a abordagem de Singer como paradigmática e argumentaremos que seu grau de exigência gera um impasse teórico na discussão. Este impasse, acreditamos, só pode ser resolvido recorrendo a outros modelos de responsabilidade, e.g., o modelo causal. Isso implica rejeitar uma das principais pressuposições metodológicas de Singer, a saber, a separação entre aspectos factuais e normativos da discussão. Nossa estratégia, porém, ainda pressupõe o pano de fundo do modelo de assistência, que deve apenas ser especificado por outros modelos. Com isso pretendemos mostrar que a acusação frequente de implausibilidade levantada contra a abordagem de Singer não é principalmente uma questão do conteúdo do dever imposto. Mas sim da força moral das razões oferecidas pelos modelos de assistência para fundamentar deveres. Palavras-chave: Dever de Assistência; Modelos de Responsabilidade, Peter Singer, Pobreza Resumen: Nuestro objetivo en este artículo es discutir el problema de la demanda excesiva de los modelos de responsabilidad de asistencia en la discusión de la pobreza mundial. Tomaremos el abordaje de Singer como paradigmática y argumentaremos que su grado de exigencia crea un impasse teórico en la discusión. Este impasse, creemos, solo se puede resolver combinando más modelos de responsabilidad, por ejemplo, el modelo causal. Eso implica negar uno de los principales presupuestos metodológicos de Singer: la separación entre los aspectos factuales y normativos de la discusión. Sin embargo, nuestra estrategia aún presupone un trasfondo de lo modelo de asistencia, que solo deben ser especificado por otros modelos. Con eso mostraremos que la acusación frecuente de implausibilidad que se levanta contra el abordaje de Singer no se trata principalmente del contenido del deber impuesto. En verdad se trata de la fuerza moral de las raciones ofrecidas por los modelos de asistencia para fundamentar deberes. Palabras-claves: Dever de asistencia; Modelos de Responsabilidad; Peter Singer; Pobreza 1 Graduated in Philosophy from the Federal University of Pará (UFPA). Master's student at PPGFil, Federal University of Santa Catarina. Diversitates International Journal (ISSN: 1984-5073) Vol. 13, N.1, Janeiro/Junho (2021), p. C01-C16


Introduction
In discussing global poverty, we have no doubt that the affluent countries in the world possess the necessary resources to alleviate it considerably or even to put it to an end. The relevant question, however, is deciding which agents should be chosen to shoulder its costs. The responsibility approach provides useful tools for solving this problem. According to Miller, the question of responsibility is to find binding principles between two parties (A & P) that have enough moral weight to ground a special duty of A in alleviating a particularly bad situation in which P finds itself 2 .
Here we shall focus on the model of assistance-based responsibility. Assistance-based models 3 rely on the idea of capacity to shoulder some or all costs to alleviate a certain bad situation 4 . That is, if one can, then one has a moral obligation to do it. A paradigm approach of this model of responsibility is Singer's Principle of Sacrifice, which states that: " […]if it is in our power to prevent something very bad from happening, without thereby sacrificing anything of comparable moral importance, we ought, morally, to do it" 5 .
For most of his critics, Singer's principle is implausibly overdemanding. Some argue that, e.g., by demanding from agents much more than they are actually willing to give, Singer's Principle is too counterintuitive to generate a binding sense of remedial responsibility towards the poor worldwide. This is because it is argued that we only fell morally moved to discharge a duty to people who we consider close to us by some special relation. Singer has often shielded himself from these critics with plausible arguments. Since the first publication of Famine, Affluence and Morality, he has repeatedly stated that it is not his goal to do good to our present moral intuitions -which rely on relations of the kind. In fact, Singer thinks that precisely because of these intuitions our criteria of distribute justice are far too minimalist for the dimension of world poverty and our capacity to act accordingly.
Still, we cannot deny that his critics raise valid points; hence the impasse persists. As Lichtenberg argues 6 , this may be a sign that the debate is theoretically depleted and, therefore, is no longer fruitful. We agree with this comment, therefore, in this paper we shall take a different approach. We think that assistance-based models are overdemanging. The reason for this, however, is a moral one: the notion of capacity, on which assistance-based principles rely cannot offer morally strong reasons to match the size of the duty they usually demand. However, from this, it does not follow that affluent 2 Miller,"Distributing Responsibilities",454. 3 Kreide also refers to it as "Expediency" (Kreide 2007). Hereafter I shall adopt Barry and Øverland's terminology and refer to this model exclusively as "assistance-based". 4 Barry and Øverland,Responding to Global Poverty: Harm,Responsibility,and Agency,[1][2]Miller,"Distributing Responsibilities", 460-1. 5 Singer, "Famine, Affluence Morality," 5-6. 6 Lichtenberg, "Famine, Affluence, and Psychology." C3 nations do not have a duty to the poor worldwide that exceeds what we presently think is required, as Singer's critics seem to suggest. This is because the lack of moral force of this model can be complemented by combining more models of responsibility, i.e., adducing to other morally relevant factors that ground responsibilities of A towards P.
From this we will conclude that a duty imposed over A is not by itself overdemanding, but only in relation to its moral rationale. And this can be lessened or aggravated assembling a number of other morally relevant reasons, such as A having contributed to P's situation to some extent (causal principle). With this we will conclude that the moral demands posed by assistance-based principles are not implausible, rather the idea of capacity on which it sustains its conclusions is too weak to match their requiring moral weight. It requires the normative force of other models of responsibility, e.g., Pogge's causal model or Miller's special relations model.
We think this strategy complements assistance-based models of responsibility also because it stays on the background of assistance-based models. By choosing to do this, it does not succumb to Singer's greatest insecurity regarding factual aspects, that is, that these might be used to limit our duties of assistance towards the world poor. This strategy involves, however, denying one of Singer's main presuppositions, namely, the drastic separation between moral and factual aspects of this discussion. I.e., we think it is possible to draw moral imperatives from factual aspects, because most of the times the latter are imbibed with normative assumptions in such a way that they cannot be so easily kept apart 7 . Therefore, enhancing the moral strength of assistance-based principles is a fruitful approach to this problem.
These introductory remarks being made, let us advance to meeting our goals. To do that, first we need to sketch Singer's assistance-based model of responsibility.

Assistance-based models and Singer's Principle
According to Barry and Øverland 8 , this type of responsibility is undisputedly important when we think of global poverty. Even if A has not in any way contributed to P's situation, or if P himself has brought about the situation upon him, P still has a rightful claim of assistance based on its needs to lead a decent life or to provide a decent life to its citizens 9 . Therefore, it is undeniable that the rich nations have a moral duty to help alleviate the conditions of the poor nations, even if they hadn't contributed to it in any extent.

C4
On moral grounds, assistance-based models assume that solving global problems, such as poverty, is a matter of general responsibilities that hold for every individual or country worldwide.
However, since those problems present urgent situations, these general responsibilities must be allocated in a more qualified manner, in order to account for a division of labor that can best account for its degree of urgency 10 . Therefore, assistance-based models rely on two main ideas as the basis for assigning duties: effectiveness and costs or capacity. Only the more resourceful parties are singled out as responsible for delivering aid to a particular situation. This is because they are best positioned to most effectively shoulder some or all costs of solving it, since the solution could be delivered at a relatively lower costs for them 11 .
Since assistance-based models offer what we can call a more forward-looking perspective, we can say that it presents a relative advantage to others. Contribution-based or causal models, e.g., look too much to past interferences for assigning duties. Causal responsibility may sometimes be problematic, because it may fail to properly equate responsibility for bringing a particular situation about and capacity to solve it 12 . That is, it fails to resolve problems when bearers of causal responsibility are currently incapable of discharging their duties 13 . Furthermore, it can assign no obligation whatsoever for agents that did not participate in bringing unjust situations about, even though they may come out as possible candidates for delivering aid 14 . This is the case of natural disasters, such as tsunamis, earthquakes, etc., which no agent is responsible for. When Singer's proposal first appeared in his famous article "Famine, Affluence and Morality" (1979) it was precisely his goal to demonstrate why affluent nations and its citizens are singled out as having an assistance-based responsibility to deliver aid to the problem of world poverty. And, therefore, to demonstrate why it is inacceptable that affluent agents do only so little for it. This is the role of the Principle of Sacrifice in Singer's formulation of the duty of assistance.
Singer's argument has a very straightforward form. He argues that (1) suffering from harms due to extreme/absolute poverty is something bad; and (2) If it is in our power to prevent this kind of suffering, without putting in risk anything of equal/similar 15 moral value; (3) not helping the poor cannot be a justified conduct 16 . 10 Barry, "Global Justice: Aims, Arrangements, and Responsibilities", 230 11 Kreide, "Neglected Injustice: Poverty as a Violation of Social Autonomy", 171; Miller,"Distributing Responsibilities",460. 12 Other subtler problem of the causal approach is that of specifying how exactly one can be held causally responsible for a situation, i.e., how one defines "contribution" to it. This may raise lengthy conceptual discussions. Cf., e.g., Øverland 2016, Ch. 4 & Pogge 2007. 13 Miller, "Distributing Responsibilities", 460. 14 Kreide, "Neglected Injustice: Poverty as a Violation of Social Autonomy", 174. 15 This implies two variations of the principle a strong and a moderate one (Singer 2016, p.27-8). 16 Singer, "Famine, Affluence Moral.," 5-6.

C5
Singer's argumentation is oriented to a radical change in our current conceptual scheme. That means, since our standards of distributive justice are low, we should not try to meet them in our theoretical discussions about the matter. Bearing this project in mind, it is important to grasp the negative form of the (3), for it is meant to reject the supererogatory idea that not aiding the poor is not blameworthy, since we do not have a duty entailing it.
From this argument he extracts the following principle: "[…]if it is in our power to prevent something very bad from happening, without thereby sacrificing anything of comparable moral importance, we ought, morally, to do it" 17 .
Singer's argument has what has been called a non-derivative form. According to Cullity 18 , to argue that non-contribution to solve global problems is wrong we need to identify a morally relevant relation between affluent agents and the world poor. To meet this goal, two strategies emerge: there are what he calls the derivative arguments and the non-derivative arguments. The former are the arguments that try to stablish such a morally significant relation based on reasons that involve affluent people collectively, such as belonging to a community or having cultural relations to other communities. By contrast, non-derivative arguments try to stablish a relation of the kind more directly, i.e., between the individuals or other agents themselves without recurring to other specific moral reasons, such as special relations 19 . A common strategy employed by those who take this path is to argument by analogy, such as Singer's famous pond analogy: "[…] if I am walking past a shallow pond and see a child drowning in it, I ought to wade in and pull the child out. This will mean getting my clothes muddy, but this is insignificant, while the death of the child would presumably be a very bad thing" 20 .
Examples of the kind are usually called life-saving analogies 21 . Their goal is precisely to identify the relevant moral relation mentioned above, namely, A having the power to act immediately to help P. Next it tries to argue why in other similar but more remote cases we would have the same or similar moral duty to act upon it. According to Cullity, this argument, in Singer's case takes a subsumptive form. Arguments of such form "[…] treat the task of justifying moral judgements about particular actions as the task of identifying general moral principles under which those judgements can be subsumed as instances" 22 .
C6 judgement regarding our duty of assistance to the poor and our duty to assist the drowning child are of a same class. Therefore, the same conclusions about our duties of assistance should be drawn in both situations. This being so, the moral duties yielded in the example of the pound are meant to be assumed as duties affluent agents have to aid the poor 24 . For the affluent agents are hypothetically in the same moral relation to the poor as a person able to help and a drowning child.
Other reasons that appear to differentiate both scenarios are not taken as morally relevant.
Note that in Singer's analogy very few assumptions are made. It does not, as it normally occurs in assistance-based principles, consider any mitigating factors: we do not have any special relations to the child. The analogy also does not consider anything about the origins of the situation, i.e., how this state of affairs has come to existence in the first place or if it is the case that our presence contributes or has contributed in any way for the existence of this state of affairs 25 . It only assumes that we are in a special position to put the situation right, because we plainly have the capacity to wade in and save the child from drowning.
Singer's argument, in particular, is a very demanding one, precisely because it has the subsumptive form 26 . If we accept that making lesser sacrifices to prevent a child from drowning is our moral duty in that case, we should also consider our moral duty to abdicate considerable amount of our resources to save the lives of poor people abroad. In Singer's initial formulation, this implies that we ought to donate until marginal utility. That is, until the point that by donating more, we'll cause an amount of harm to ourselves that is equal (or similar) to the harm we are trying to prevent. More specifically, it means that we have to donate until the point that by donating more we'll be at the same poverty level as the worst-off people in the world 27 .
This of course would imply making great sacrifices regarding our current way of living, but surely these would be morally irrelevant regarding the possibility of putting a very bad situation right.
This, of course, implies a radical reformulation in our idea of duty to the poor. Nevertheless, for Singer, if we accept the premises of his argument, but not this conclusion, there's an incoherence in our system of moral intuitions, which we ought to eradicate.
As we can see, Singer's conclusion is not very easy to digest and his critics try to show why we cannot draw it from its premises. This consists in a number of strategies. We shall concentrate in three of them, the first is a methodological one. It mounts (1) an attack against the possibility itself of reaching a convincible conclusion with an argument by analogy. The other two criticize (2) the effectiveness of assistance models and (3) its vagueness. 24 Lichtenberg, "Famine, Affluence, and Psychology.", Section II 25 Barry and Øverland,Responding to Global Poverty: Harm,Responsibility,and Agency, Ibid., 15 27 Arneson, "What Do We Owe to Distant Needy Strangers?"

C7
Problems with assistance-based models

Problems in arguing by analogy
As we have seen assistance-based models usually rely on capacity for assigning duties to the agents they think are best positioned to alleviate the situation. The duties imposed to A by such principles have an utilitarian bias, since the costs A has to assume will presumably be less relevant relatively to A's capacity. Therefore, by picking out the best suitable agents, we maximize the good done, without making relatively great sacrifices 28 . This being so Singer's approach is a classical consequentialist one, which usually arrive at moral conclusions based on calculations. For that, from a methodological point of view, he must assume that the same moral variables come into play in every analogous situation 29 . That is, it must assume some kind of universal principle, or value neutral measurement rod, on which critical ethical reasoning finds its basis 30 . This assumption can be tackled by objecting against its underlying conception of definition from a Wittgensteinian point of view. It can be said that the argument employed by Singer's life-saving analogy is problematic, for it presupposes an essentialist conception of definition, which can be regarded as problematic for ethical reasoning.
An important argument in Wittgenstein's Philosophical Investigations (PI) 31 is his critique of the conception of real or essentialist definition. According to this conception, for a definition to be adequate, it must precisely circumscribe the property F, by virtue of which every Fx is Fx. This must work in such a way to make the definition applicable without exceptions to all relevant cases 32 . The argument Wittgenstein draws against it is that, rather than being a necessary methodological assumption in philosophy, the conception of real definition turns out to be a dogmatic requirement 33 . That is, it is an unnecessary assumption that we take every manifestations of phenomena falling under a concept to have a common and unifying property. Rather, if we redirect the orientation of our thinking 34 in a more contextually sensible direction, we will find that most cases bear only familiar resemblances to each other. We could still identify centers of variation, however, for a relevant variety of concepts these are not fix and rigid centers, as it is demanded in real definitions. This is the case for concepts such as "game", "language", but also for ethical and aesthetical concepts 35 . 28 Cullity, "THE LIFE-SAVING ANALOGY", 12 29 This is explained above by noting that his argument has a subsumptive form. 30

C8
For Wittgenstein, such concepts have an open-texture: words such as "beautiful" when applied to landscapes have different meanings than when applied to sculptures and other objects of art. By the same token, for him, concepts such as "good" are also bound to the act they qualify 36 . Therefore, their meaning can only be completely grasped with the word referring to this act, which is contextually given.
As we have seen, the method of arguing in question draws its conclusions assembling two moral judgements under the same moral concept. This being so, the conclusion to be drawn from the conception of family resemblance applied to ethical concepts is that such comparisons are problematic 37 . For it can be plausibly assumed that the relevant ethical concepts in analogy have contextually different meanings and, therefore, cannot be compared in the intended way.
This objection can, however, be overturned. For this impossibility can only be said to hinder the analogy if the differences between the ethical conceptions in both cases are indeed to be considered morally relevant. And even if this is the case, it does not follow that we are impeded to stablish a moral duty this way, since both cases under the analogy can still be said to be similar -although not equal -in a moral relevant sense. That is, even if we could not say that we have the exact same duties, we can still argue that we have similar duties 38 . Hence, it is still open to debate whether these differences are morally relevant or whether they have a relevant difference in degrees of similarity.
As we shall see this is precisely what the next objections aim at. They try to identify a disanalogy between the both cases that ultimately cancels the possibility of arguing by analogy.

The imperceptibility objection
This objection tries to show that there is a serious disanalogy in non-derivative arguments, such as Singer's. It also criticizes the efficiency of discharging duties of assistance by donating to aid agencies. As noted above, assistance-based models rely on capacity for selecting the most suited agents to put a certain bad situation right, provided that shouldering its costs will amount to relatively low sacrifices for them. The advantage of such model is that relying on capacity might be a strategy for solving problems in the most effective way, for there may be situations where more specific duties cannot be so easily assigned. Since the imperceptibility objection launches an attack at its greatest advantage it might represent a serious case for assistance-based models.
According to Cullity 39 , the imperceptibility objection criticizes the moral basis of assistancebased models for donating to agencies that focus on promoting emergencies relief programs. The 36 (AWL 35) 37 Kuusela, "Wittgenstein, Ethics and Philosophical Clarification.", 54 38 Cullity, "THE LIFE-SAVING ANALOGY.", 12 39 Cullity, "THE LIFE-SAVING ANALOGY", 4-5 C9 author asks us to imagine a case in which I give my money to aid-agencies that supply food for camps.
Hopefully, the money I donated will be used to increase the amount of food, which in turn will be distributed to everyone in the camp. But the extra food my donation will allow the agency to buy will not exactly be used to feed one extra person. Rather, it will be diluted in the total amount of food available in camp, possibly representing only a tight increase in it. If this is the case the effect of my contribution will be in fact imperceptible as an increase in the good done to the beneficiaries of my donation. Therefore, had I refrained from donating food to this particular agency it would not follow that these people would starve. This represents a serious disanalogy in Singer's example. For the subsumptive form of his argument wants to lead us to the conclusion that not donating is similar to letting the child drown in the pond, when, as the situation above tries to show, it is not the case 40 .
Moreover, if the increase in the amount of food my donation enables to buy is to be as thin as this objection suggests, it would be imperceptible to such an extent that its effect could be considered irrelevant. Since anyone 41 that addresses the problem individually is minuscule in comparison to its extent, it follows that the money I donate individually will probably represent a greater loss for me than it could be a gain for the beneficiated poor 42 . Moreover, if my contributions to aid agencies will not have a more concrete effect how can I say that by not donating to them I fail to act for the world poor (e.g., by failing to fulfill my duty of assistance or violating their right not to be hungry) 43 ? Therefore, assistance-based models recurring to life-saving analogies fail to stablish a plausible analogy to ground duties. This objection, however, has only limited scope, since, if it is correct, it only demonstrates that, first, individual and uncoordinated actions towards the problem are wasteful, but that does not apply for collective and coordinated actions 44 . If non-derivative arguments try to stablish moral relations individually, it does not follow that we have to discharge the correspondent duties also individually. Second, that donating through aid-agencies that promote emergencies relief programs is inefficient leaves untouched other forms of donating, e.g., through political action or even through aidagencies that aim at preventing future harm to the poor by providing structural solutions. 45 Hence, unless the critique shows every single way of donating to be ineffective, it does not represent a real problem to assistance-based models 46 . 40 Ibid., 5 41 As pointed out, arguments like these, at least in this context, are non-derivative, i.e., the relation is stablished individually according to my economical capacity of donating. 42 Cullity, "THE LIFE-SAVING ANALOGY", 5; Shue, "Mediating Duties.", 695 43 Cullity, "THE LIFE-SAVING ANALOGY.", 6 44 Shue, "Mediating Duties.", 697 45 Which is in fact what most of them are concerned with (Cullity 2002, 4). 46 Lichtenberg, "Famine, Affluence, and Psychology.", Section II.

C10
Even acting in community, however, it is most likely that our efforts to deal with the problem of poverty are still quite inferior than what is required. What this shows is that we need some kind of selectivity in approaching it 47 . This may be provided by deriving duties from special relations.

Special relations
The special relations approach, other than providing a criterion of selectivity, is also a common strategy for criticizing Singer's conclusion. It consists in pointing out morally relevant factors that differentiate our moral relation towards P in each case and arguing that these factors amount to a disanalogy between the two cases. By doing this, the critique states that trying to derive equal or similar duties from comparing both cases is unwarranted.
For Miller, arguments like Singer's amount to a kind of ethical universalism that rests on an implausible assumption. That is, that when we act morally we should consider only rational ethical principles and do not let our sentiments and common moral intuitions influence our judgement. However, for him, this is an idealistic argument, since for the majority of human beings only the former factors have the moral weight to motivate us to act ethically 48 .
One important strategy of specifying responsibilities is to find special relations between agents. An important approach 49 for this is the cause or the contribution-based model. The causal model focuses on whether an agent can be held as responsible for bringing about a certain state of affairs to assign duties of remedial responsibility 50 . It tells us to look for the causes of poverty, e.g., conditions of production, government structures and the international rule system 51 .
According to Pogge 52 -its main proponent -the attribution of causal responsibility can be done in many ways that do not necessarily involve a direct interference, which would indeed greatly limit the scope of its applicability. This is because causally responsible agents can always adduce to other elements that also causally contributed to bringing a certain situation about, thereby, alleviating the cost of moral responsibility they have to weight. The causal approach can be detailed in a number of ways that include factors that might appear as morally relevant. For instance, acts of omission and the maintenance of existing of social institutions can also be considered as bearing causal responsibility for creating or maintaining certain states of affairs.
If a successful causal relation can be stablished between A and P, the causal approach offers a very strong rationale for assigning duties. For no one would object that having contributed to some 47 O'Neill, Faces of Hunger: An Essay on Poverty, Justice, and Development., 160O'Neill. 48 Miller, On Nationality, 57-8 49 Another relevant approach would be to employ the community principle. According to Miller (2001, p.462), the community principle states that when people are linked by special ties of any type a community might have (e.g., shared activities and commitments, common identities, common histories) they have a special duty to one another. 50 Miller, "Distributing Responsibilities", 455 51 Kreide, "Neglected Injustice: Poverty as a Violation of Social Autonomy.", 172 52 Pogge, "Severe Poverty as a Human Rights Violation.", 15-7 C11 situation is a weak reason to assume a great deal of responsibilities for remedying it 53 . In this way, it also helps with the vagueness of some discourses for assigning duties that fail to bind agents to do or not to do something 54 . This, as we have seen, was a significant problem of assistance-based approaches.
The special relations approach appears at first sight to be fruitful, since they appeal more directly to our common moral intuitions in ascribing responsibility, instead of appealing to abstract and artificial devices 55 . Indeed, according to Shue 56 , approaches of the kind are the only way of assuring that affluent agents have a positive and perfect duty to the poor. That is, that affluent agents have a morally justified duty to assist the poor and that the poor have a rightful claim of assistance.
Nevertheless, Singer rejects considerations of the kind for these comments usually rely on arguments regarding factual aspects to ground duties 57 . He argues that these reasons should not be incorporated in our moral reasoning, because they have no bearing in our moral obligations and only contribute to cloud our judgement about them. Singer fears that this kind of strategy can serve as an excuse for affluent agents assuming their due responsibility as having great capacity to alleviate the problem of world poverty 58 . Which is definitely a reasonable worry, since the definition of "cause", e.g., can -even though not always -be limited to such an extent that actually makes more difficult to assign duties. In this sense, the causal approach greatly depends on how precisely the notion of "cause" is analytically specified 59 .
However, this is not only supposedly a bad strategy for convincing others to assume more responsibilities, it can be disputed from a philosophical point of view as well. Surely it is important to draw a distinction between the moral and factual or psychological aspects of the discussion. For it would be a clear case of naturalistic fallacy to draw moral imperatives directly from factual premises 60 . If this is so, we can rightfully say that these critics barely touched Singer's normative conclusion that, even if you do not have any relevant factors towards P, assistance-based responsibility imposes a duty of making great sacrifices 61 to save P 62 . 53 Kreide, "Neglected Injustice: Poverty as a Violation of Social Autonomy.", 174. 54 Pogge, "Severe Poverty as a Human Rights Violation.", 14. 55 Miller,"Distributing Responsibilities",462. 56 Shue,"Mediating Duties.",703. 57 Singer, "Famine, Affluence Moral.", 7 58 Singer, Peter. "Poverty, Facts, and Political Philosophies: Response to 'More Than Charity. '" Ethics & International Affairs 16, no. 1 (2002): 121-24. doi:10.1111/j.1747-7093.2002 This also implies that it might be problematic to apply this model to individual agents (Pogge 2007, 17). But it must be applied to greater collective agents. Therefore, the causal approach also shifts the problem of poverty from an individual perspective to a collective perspective, and this brings the question of the applications of the collective models of responsibility. 60 Singer, "Famine, Affluence Moral.", 14-5 61 It is, of course, disputable whether we would actually need to make such large sacrifices even if we consider capacity alone as a principle for assigning duties Barry and Øverland, Responding to Global Poverty: Harm, Responsibility, and Agency, Ch.2 & CULLITY, The Moral Demands of Affluence. 62 Barry and Øverland,Responding to Global Poverty: Harm,Responsibility,and Agency,14;Lichtenberg,"Famine, Affluence, and Psychology.", Section II C12 Nevertheless, sometimes we cannot sharply distinguish moral from factual or psychological aspects in ethical discussion. For instance, we could also reject drawing the conclusion that from A's capacity or resourcefulness it follows necessarily that it has a duty to sacrifice its assets to solve problems it did not caused 63 . This shows rather that most of the factual reasons we give for assigning possible duties are imbibed with very strong normative assumptions 64 . Because of this, we think, the critics raise valid points. Indeed, the special relations approach was an intuitively good option for dealing with the problem of selectivity introduced by the imperceptibility objection. By rejecting it, it seems that the discussion reached an impasse.

Combining More Models of Responsibility
Surely there may be good reasons for separating the moral task of grounding duties from the question whether we will actually discharge them. But the overdemandingness 65 of Singer's initial formulation leaves the latter question completely open. As Lichtenberg argues, this impasse may be a sign that this debate is no longer fruitful. In any case, if the objections from his critics do not attack directly Singer's conclusion, they at least illuminate what bothers people in it.
By adducing to morally relevant reasons that distinguish the two cases, they seem to object not to the overdemandingness of the sacrifices themselves to be made regarding the problem of world poverty. Rather, what makes assistance-based models burdensome is the fact that a single notion of responsibility based on capacity is an insufficient reason for demanding such sacrifices 66 . Surely, if we had other kinds of special relations -such as having caused P's situation -to the party requiring assistance, it would be undisputable that the costs of putting the situation right would not be unfairly high 67 . We can specify this argument following the distinction drawn by Barry and Øverland 68 between required costs (R) and necessary costs (N) 69 . R is the cost due to which A has to be held morally responsible to alleviate P's condition. N is the cost A actually has to shoulder to alleviate or solve P 's situation. Only when N is bigger than R the costs of acting are overdemanding. In these situations,

C13
we can rightfully argue that A has no responsibility of shouldering the entire cost for alleviating P's condition. Hence, we can only morally require P's assistance when R equals or is bigger than N.
A possible way of demonstrating this is to take, for example, contribution-based models. This is because it provides an important shift in our perspective regarding the fair distribution of resources.
This proposal does not focus exclusively on a fair distribution of resources, since this aspect presupposes a rightful claim over these 70 . If P's resources are not fully own, the idea of fair distribution must also accommodate the duty of compensation.
We can show how this works in Singer's own example. As we know, as a result of the outbreak of the Bangladesh Liberation War, almost nine million of Bengali refugees suffered and died from lack of food, shelter and medical care. At the time -as it is now for any situation -it was plainly in the power of most affluent nations to deliver assistance in order to drastically reduce this amount of suffering. According to Singer's data, Britain, for example, gave the highest amount of monetary aid to try to alleviate the problem -ca. £14,750,000. On the other hand, it also directed ca. £275,000,000 to the development of a supersonic transport of the project of the Anglo-French Concorde 71 .
For Singer, since our present standards of distributive justice are low, we do not regard England's monetary decisions as wrong in any way. Rather we regard it as charitable or supererogatory, we think that the country has only a weak distributive duty of justice in this case. Therefore, we would be demanding too much of it if we said that it has to make large sacrifices to help alleviate the situation.
However, as Barry and Øverland 72 point out, even if we accept that, it does not follow from it that England has no duty whatsoever of assuming overdemanding costs. This is because assistancebased models assume that A is rightfully entitled to all of its resources and assets. If it be proved that England had to some extent contributed for bringing that situation about 73 , in addition to its initial assistance-based duty, it would also have a contribution-based duty to shoulder all costs in dispute.
From this point of view, as we mentioned, we obtain a shift in the conception of justice that we regard as underlying the case. That is, we went from a distributive conception of justice to a rectificatory one 74 . If A's resources are not fairly own, the idea of fair distribution must also accommodate the duty of compensation.
C14 offer for relating a particular agent to it. Only R changed, due to other morally relevant considerations that were brought to play 75 . This means that the overdemandingness of N is a function of R. That is, a duty imposed over A is not by itself overdemanding, but only in relation to the moral weight represented by R. And R can be lessened or aggravated due to a number of morally relevant reasons assembled under it. This means that the duties posed by assistance-based principles are not implausible, but the idea of capacity on which it sustains its conclusions is too weak to match R to N. This moral force in turn does not depend on the contents of the special duties we identify in a given situation 76 . In this context, the rationale of the causal approach provides a strong change of perspective in this aspect, which enhances the normative force of assistance-based models.

Concluding Remarks
Indeed, there are some problems in assistance-based models to ground a duty of assistance to the poor worldwide. However, to recur to the special relations approach alone is even more problematic because, as noted, it does not correlate necessarily to other factors useful for that grounding obligations. This can limit its scope of applicability. On the other hand, if we focus solely on the general responsibilities ascribed by assistance-based models, it may seem that we are placing too demanding moral obligations on agents in complex cases where in fact this is not so. This may present a problem in formulations of duty since the overdemandingness of his formulation can also be drawn by those who want to reject the duty of making large sacrifices to help the poor abroad in the first place 77 . The alternative sketched here, in turn, could revert Singer's strategy of avoiding excuses the other way around, since it stills works on the background of the general notion of responsibility of assistancebased duties. Therefore, as Kreide 78 suggest, assistance-based models can and should be complemented by other models of responsibility.