We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
Published online by Cambridge University Press: 26 January 2009
A common objection to consequentialism is that it makes unreasonable demands upon moral agents, by failing to allow agents to give special weight to their own personal projects and interests. A prominent recent response to this objection is that of Samuel Scheffler, who seeks to make room for moral agents by building agent-centred prerogatives into a consequentialist moral theory. In this paper, I present a new objection to Scheffler's account. I then sketch an improved prerogative, which avoids this objection by incorporating a non-proportional account of the relationship between values, costs and reasons.
1 Scheffler, S., The Rejection of Consequentialism, Oxford, 1982Google Scholar.
2 Ibid., p. 5.
3 Ibid., p. 20.
4 Ibid., p. 67.
5 Scheffler, S., ‘Prerogatives without Restrictions’, Philosophical Perspectives, vi (1992), 377–97, at 378CrossRefGoogle Scholar.
6 This weighting may seem implausibly high. However, a much lower weighting would lead to almost all inhabitants of developed countries being overwhelmed by their obligations to distant strangers in the developing world. As one aim of the hybrid view is to provide a less demanding alternative to traditional act consequentialism, this would largely defeat its purpose.
7 See, for instance, Hooker, B., ‘Rule Consequentialism’, Mind, ic (1990), 67–77CrossRefGoogle Scholar; Jackson, F., ‘Decision-theoretic Consequentialism and the Nearest and Dearest Objection’, Ethics, ci (1991), 461–82CrossRefGoogle Scholar, and the broader literaturexs referred to therein.
8 The proof of all this is as follows. If Affluent is not required to donate all her money to Oxfam, then there must be at least one amount of money which meets condition (i). Let x be the smallest such amount of money; that is, the smallest amount of money such that Affluent is permitted to donate x dollars to charity rather than donating any more. It follows that, for any amount of money less than x, Affluent is not allowed to donate that amount of money rather than donating more. If Affluent is not allowed to keep all her money for herself, then x must be greater than zero. So x meets condition (ii) as well as condition (i). Now, if Affluent is permitted to donate x dollars, then she is permitted to donate x dollars rather than donating x + 1 dollars. The marginal utility of that next dollar to Affluent, when multiplied by the extra weight Affluent is permitted to give to her own interests (call the result the weighted cost), must be at least as great as the marginal value of a dollar to Oxfam. Assume for the moment that, at x, the weighted cost is greater than the marginal value. Assume also that both the weighting and the marginal value are constant, whereas the marginal cost is increasing. Therefore, there must be some amount of money (z) less than x, such that the weighted cost to Affluent of donating z dollars is equal to the marginal value of an extra dollar. Affluent would then be permitted to donate z dollars rather than donating z + 1. However, for each dollar we add beyond z + 1, the marginal value remains constant, whereas the weighted cost increases. Therefore, if Affluent is permitted to donate z rather than z + 1, she will be permitted to donate z rather than any amount greater than z. But this would contradict our assumption that x is the least amount of money such that Affluent is permitted to donate no more than that amount. So, by reductio, we can conclude that, at x, the weighted cost is not greater than the marginal value. Therefore, the weighted cost must be equal to the marginal value. So x meets condition (iii). QED. (For this proof, I made the simplifying assumption that money comes in discrete amounts, which I identify with dollars.) These results depend upon two assumptions, which do not seem unreasonable. Given the size of Oxfam's operations as compared to Affluent's resources, we can assume that the marginal good produced by each additional dollar donated is constant. In other words, the difference between the amount of good produced by a donation of x dollars and that produced by a donation of x + 1 dollars will be the same for any value of x between zero and n. By contrast, we can assume that the marginal cost to Affluent of each additional donation of one dollar is increasing. That is, the cost to Affluent of donating an additional dollar once she has already given most of her income will be much greater than the cost to her of donating an additional dollar when she has not given anything as yet. However, if we reject the assumption of increasing marginal cost, then it must be the case either that Affluent is permitted to donate nothing at all or that she is required to give all of her money to Oxfam. The possibility that Affluent is required to make some donation, but not required to bring about the best possible consequences, will be ruled out. (The proof is straightforward. Both marginal value and the weighting the agent is allowed to give to her own interests are constant. If the marginal cost to the agent of donating an extra dollar is also constant, then the relationship between weighted cost and marginal value will be constant. Either weighted cost is always at least as great as marginal value, in which case Affluent is permitted to donate any amount she chooses rather than donating more, or weighted cost is always less than marginal value, in which case Affluent is always required to donate more if she is able to. Either a zero donation is permitted or a total donation is required.) Yet it is precisely this possibility that proponents of the hybrid view seek to defend. Accordingly, the assumption of increasing marginal cost must be retained.
9 I have elsewhere presented a similar objection to rule consequentialism (see Mulgan, T., ‘Rule Consequentialism and Famine’, Analysis, liv (1994) 187–92CrossRefGoogle Scholar; and ‘Two Conceptions of Benevolence’, Philosophy and Public Affairs, xxvi (1997) 63–79Google Scholar).
10 We should note that my claim here is not that Affluent will be required to give ten times as much money in Superefficient Oxfam as in Normal Oxfam, nor that she will be required in Inefficient Oxfam to give only one tenth as much money. Indeed, as a result of diminishing marginal utilities, the differences between the amounts of money she is required to sacrifice are likely to be less than factors of ten. However, the variations in the amounts of money Affluent is required to sacrifice will still be very great. Greater, I would suggest, than we can accept.
11 There are two dimensions to the wrong facts objection. The first is the epistemic claim that Affluent should not need to acquire such detailed empirical knowledge before she decides how much to give to charity. The second claim is that the question of whether or not a given level of donation is morally acceptable ought not to depend upon details regarding the number of starving folk. We might call these two claims the ‘subjective’ and ‘objective’ sides of the wrong facts objection. In the text, I have tended, for simplicity's sake, to run the two claims together. They are obviously mutually supporting. One explanation for the claim that Affluent does not need to plough through the World Bank report before acting would be that the facts in question would not affect the objective rightness or wrongness of her actions. On the other hand, the very fact that the suggestion that Affluent should plough through the World Bank report would strike us as odd may lead us to the view that the facts contained within that report cannot be morally relevant. Despite their intimate connection, the subjective and objective claims are distinct, as someone who distinguished between objective and subjective rightness could make one without making the other. Such a person would accept only one side of the wrong facts objection. (I am grateful to Samuel Scheffler for pointing out to me the importance of this distinction in this context.)
12 See Scheffler, S., ‘Prerogatives without Restrictions’, 378Google Scholar.
13 Furthermore, my argument does not require me to assign some particular value to M. I need only the assumption that M has some particular value.
14 We could, of course, provide a parallel response, to the effect that there is a certain minimum sacrifice which is always required of any agent in any situation. (For instance, perhaps we should always ‘give till it hurts’.) However, this move would threaten to render the agent-centred prerogative effectively redundant, as all the real work would be done by the maximum and minimum levels, not by the prerogative.
15 We should note also that the situations discussed in my Affluent's Ignorance tale are by no means bizarre. They involve none of the alien invasions, miracle technologies or supernatural forces often found in the tales of philosophers. If a theory cannot give the right answers in plausible counterfactual situations, then it must be rejected.
16 For this example, I make the simplifying assumption that each of the first 10 units of cost has a weighting of 2, while each subsequent unit has a weighting of 5. In a more realistic theory, this simple stepwise function would be replaced by a more continuous one, but this does not affect the argument of the text.
17 The proportionality of value thesis is by no means universally held. For rejections of it, see, for instance, Hurka, T., ‘Value and Population Size’, Ethics, xciii (1983), 496–507CrossRefGoogle Scholar; Ng, Y.-K., ‘What Should We Do about Future Generations?’, Economics and Philosophy, v (1989), 235–53CrossRefGoogle Scholar. I should emphasize that I am not suggesting that the arguments of this paper rely upon this thesis, merely that they are not incompatible with it.
18 The rationale for non-proportionality presented in this paragraph has close affinities with Scheffler's own rationale for the agent-centred prerogative itself, as presented in The Rejection of Consequentialism.
19 The hybrid view may, of course, face other obstacles. For instance, Shelly Kagan has raised concerns regarding the relationship between the agent-centred prerogative and the distinction between doing and allowing. (See Kagan, S., ‘Does Consequentialism Demand Too Much?’, Philosophy and Public Affairs, xiii (1984), 239–54Google Scholar. For Scheffler's response, see S. Scheffler, ‘Prerogatives without Restrictions’.) I attempted to combine the view presented in the present paper with a solution to Kagan's dilemma in a paper entitled ‘A Unified Conception of Benevolence’, presented at the conference of the International Society of Utilitarian Studies in New Orleans, March 1997.
20 Previous versions of this paper were read to the Wolfson Philosophy Society and the University of Otago. I am grateful to both audiences for helpful discussion. I am also grateful to the following for comments on earlier drafts: John Broome, Roger Crisp, James Griffin, Brad Hooker, Rahul Kumar, Andrew Moore, Derek Parfit, Samuel Scheffler and an anonymous referee.
