Hostname: page-component-7c8c6479df-xxrs7 Total loading time: 0 Render date: 2024-03-28T06:10:53.168Z Has data issue: false hasContentIssue false

A Defence of Average Utilitarianism

Published online by Cambridge University Press:  20 March 2015

MICHAEL PRESSMAN*
Affiliation:
University of Southern Californiampressman@gmail.com

Abstract

Seemingly every theory of population ethics is confronted with unpalatable implications. While various approaches to the subject have been taken, including non-consequentialist approaches, this area has been dominated by utilitarian thought. The two main approaches to population ethics have been total utilitarianism (‘TU’) and average utilitarianism (‘AU’). According to TU, we should seek to bring about the state of affairs that maximizes the total amount of happiness. According to AU, we should seek to bring about the state of affairs that maximizes average per capita happiness. Both theories have been afflicted by seemingly strong objections, and as a result, numerous variations and hybrids have been introduced. Despite the widespread disagreement in the field, though, a near consensus has developed in rejecting AU as an absurd view. In this article, however, I will go against the grain and argue that AU is the theory of population ethics that we should endorse.

Type
Research Article
Copyright
Copyright © Cambridge University Press 2015 

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 Parfit, Derek, Reasons and Persons (Oxford, 1984)Google Scholar.

2 See Huemer, Michael, ‘In Defence of Repugnance’, Mind 117 (2008), pp. 899933CrossRefGoogle Scholar; Ryberg, Jesper, ‘Is the Repugnant Conclusion Repugnant?’, Philosophical Articles 25 (1996), pp. 161177Google Scholar; Ryberg, Jesper, ‘The Repugnant Conclusion and Worthwhile Living’, The Repugnant Conclusion: Essays on Population Ethics, ed. Ryberg, J. and Tännsjö, T. (Dordrecht, 2004), pp. 239–56CrossRefGoogle Scholar; Tännsjö, Torbjorn, ‘Who are the Beneficiaries?’, Bioethics 6 (1992), pp. 288–96Google Scholar; Tännsjö, Torbjorn, ‘Why We Ought to Accept the Repugnant Conclusion’, Utilitas 14 (2002), pp. 339–59CrossRefGoogle Scholar, reprinted in The Repugnant Conclusion, ed. Ryberg and Tännsjö, pp. 219–38.

3 See Hurka, Thomas, ‘Value and Population Size’, Ethics 93 (1983), pp. 496507CrossRefGoogle Scholar.

4 See Temkin, Larry S., ‘Intransitivity and the Mere Addition Paradox’, Philosophy and Public Affairs 16 (1987), pp. 138–87Google Scholar.

5 There are different views about what precisely counts as ‘a utilitarian approach to population ethics’ or ‘a utilitarian account’, and there is not an obvious fact of the matter. Here and in what follows, I leave this definitional question largely untouched, and, when I refer to accounts as ‘utilitarian’ or ‘non-utilitarian’’, I mean to refer to paradigmatic cases of each – rather than cases at the margins where an account's utilitarian or non-utilitarian status is controversial. For one interesting discussion on this topic see Parfit, Reasons and Persons, app. I.

6 Mackie, John, Ethics: Inventing Right and Wrong (New York, 1977)Google Scholar.

7 Parfit, Reasons and Persons, p. 394.

8 Parfit, Reasons and Persons, p. 359.

9 Some would deny that this case gives reason to reject person-affecting moralities, arguing that the second child would be worse off in the state of affairs where the girl does not wait to have a child. However, it seems odd to say that this child is harmed by not being brought into existence, and maintaining this position would require one to accept additional implausible positions as well. For further discussion, see Broome, John, Weighing Lives (Oxford, 2004)CrossRefGoogle Scholar.

10 See Melinda A. Roberts, ‘Person-Based Consequentialism and the Procreation Obligation’, The Repugnant Conclusion, ed. Ryberg and Tännsjö, pp. 99–128.

11 See Nils Holtug, ‘Person-Affecting Moralities’, The Repugnant Conclusion ed. Ryberg and Tannsjo, pp. 129–62, for a discussion of the difficulty – in fact, the impossibility – of constructing a plausible person-affecting morality.

12 Narveson, Jan, ‘Utilitarianism and New Generations’, Mind 76 (1967), pp. 6272, at p. 68CrossRefGoogle Scholar.

13 Parfit, Reasons and Persons, p. 394.

14 Boonin-Vail, David, ‘Don't Stop Thinking About Tomorrow: Two Paradoxes about Duties to Future Generations’, Philosophy and Public Affairs 25 (1996), pp. 267307, at p. 268CrossRefGoogle Scholar.

15 Broome, Weighing Lives.

16 Narveson, Jan, ‘Moral Problems of Population’, The Monist 57 (1973), pp. 6286, at p. 73CrossRefGoogle Scholar.

17 Throughout the article, reference will be made to ‘happiness levels’ and ‘quantities of value’. While there are many interesting and important issues surrounding how to determine what happiness level an individual has and what these numbers actually represent, this article will attempt to bracket these issues. I will simply make the seemingly plausible assumption that sense can be made of numerical values representing people's happiness.

18 This is a very slight variation of an example offered by Broome, Weighing Lives, p. 146.

19 It does not seem as though the arbitrary choice of how to order individuals should have an effect on the betterness ranking of two worlds, and in cases such as the comparison between E and F, it seems that it would. If the conditional good account allows for the comparison between E and F, it seemingly would say that (4, 5) is better than (4, 1, 6), but worse than (4, 6, 1). One might think that if, among the arbitrary orderings of individuals within each world, there are orderings that provide different prescriptions, then perhaps this indicates that the two worlds are equally good. However, not only does this inference seem not to cohere with the intuitions that result in the seeming contradiction, but there would be extreme cases that could be constructed that would make this rule seem implausible. For example, consider a comparison between A (2) and B (9, 9, 9, 9, 9, and 1.99). One of the arbitrary orderings of B would put 1.99 in the first slot, and since this means that there would be at least one ordering for a prescription in each direction, the result, on this rule, would then be that A is equally as good as B.

20 Among the approaches that have been offered are views that (1) deny the transitivity of ‘equally as good as’’, (2) maintain that goodness is relative and not absolute, (3) deny that betterness is fully determinate, and (4) no longer try to find an axiological account of the neutrality intuition. See Broome, Weighing Lives, p. 149, for further discussion of the notion of conditional good, and of the failure of these other approaches.

21 In fact, the rejection of the neutrality intuition is what enables AU to be coherent in the ways discussed in the previous paragraph – ways in which a position embracing the neutrality intuition would not be coherent.

22 One might think that the ‘anti-more-is-better intuition’ – the foundation of my argument for AU – begs the question in a debate with a proponent of TU. In a sense this is true, but in a harmless sense. As I say in the text, I think that the majority of proponents of TU are prima facie sympathetic to the anti-more-is-better intuition, but that they then abandon AU because they are persuaded by one or more of the main objections to AU that I will discuss. Thus, in this sense, I am not begging the question, because the target of my arguments is someone who would buy my premise – and not a committed TU proponent. As I mention in this paragraph, though, my argument does not rest on the anti-more-is-better intuition, and it could entice a reader to believe AU even if they do not share this intuition prima facie. This is because there are prima facie difficulties with other views – such as the Repugnant Conclusion.

23 Each of these could, in turn, be further precisified by stipulating which temporal population is relevant.

24 Interestingly, it is worth flagging that some object to AU1, arguing that it is susceptible to an intrapersonal analogue of the Repugnant Conclusion. This point could suggest one of two things. First, it could suggest that we should espouse a different version of AU. Second, it could cast doubt on AU1's rejection of TU. This inquiry is beyond the scope of this article, but I do believe that there are plausible replies to both of these concerns.

25 Parfit, Reasons and Persons, p. 420.

26 Parfit, Reasons and Persons, p. 420, citing McMahan, Jeff, ‘Problems of Population Theory’, Ethics 92 (1981), pp. 96127CrossRefGoogle Scholar.

27 Parfit, Reasons and Persons, p. 420.

28 This example does not depict the individuals ‘who will live in the distant future’ as being very far into the future – it does not depict a string of values coming after the fourth value and before the fifth. This, however, is just for the sake of simplicity of presentation, and the point is the same. Also, note that the third individual is slightly better off in the cases where the child is had. This is supposed to represent the happiness a parent gains by having the child. A similar point, however, would apply even if the parent was not depicted as being made better off by having the child.

29 Parfit, Reasons and Persons, p. 419.

30 Huemer, ‘In Defence of Repugnance’.

31 Huemer, ‘In Defence of Repugnance’.

32 Huemer, ‘In Defence of Repugnance’, p. 920.

33 Huemer, ‘In Defence of Repugnance’, p. 921.

34 ‘V’ means value, the second letter denotes the world, and the numeral denotes the temporal part.

35 Huemer, ‘In Defence of Repugnance’, p. 921. In addition to arguing that duration of welfare is equivalent to widespreadness of welfare, Huemer also argues that intensity of welfare is equivalent to duration of welfare (and thus to widespreadness of welfare). For the purposes of our discussion here, however, we need not discuss this other argument.

36 Huemer, ‘In Defence of Repugnance’, p. 921.

37 Mill, John Stuart, Utilitarianism (Indianapolis, 1979/1861)Google Scholar.

38 Parfit, Reasons and Persons, p. 361.

39 There certainly has been discussion about the relevance of future persons, merely possible persons, and persons that are contingent relative to the set of options, but the fact that TU must answer the question of which population is relevant does not seem to get mentioned in critiques of AU. Discussions of AU's shortcomings do not make clear that the choice of the relevant population is a question that should be answered prior to the debate between AU and TU, and thus should be answered without being influenced by a thinker's commitment to AU or TU.

40 In other words, while the following is not the case for AU, TU satisfies what one might call a monotonicity condition, according to which population X is better than population Y if and only if the combined population X & Z is better than Y & Z. However, this property of TU notwithstanding, as just argued, even unaffected values (e.g. population Z) are still part of the population relevant to ethical inquiry.

41 I’m open to the idea of including other conscious creatures as well, but this would not change the structure of the theory.

42 I owe thanks to an anonymous reviewer for making the following point and for stressing its importance. Although I refer throughout this article to ‘person-affecting intuitions’, it is important to recognize the differences between various types of person-affecting intuitions. Among the different possible formulations are the following two. First, x is better than y, if (x is better than y for some actual person and worse for no one). Second, x is better than y, only if x is better than y for some actual person. Unless stated otherwise in what follows, I will intend a third formulation – the conjunction of the first and second formulations. It is helpful to notice when making comparisons, though, which conjunct of the conjunction is operative in making a particular claim. In most cases, however, it is the second formulation that is doing the work.

43 It is important to note here (and throughout section V.D when I make claims analogous to this) that I am not saying that a committed proponent of TU would find A preferable to C. This is certainly not the case since TU holds that C is preferable to A. Those about whom I’m speaking are those who find B preferable to A and who then conclude that what must be influencing them and what must be doing the work in an explanation of their intuitions is a sympathy for TU.

44 In the versions of Reverse Egyptology and Egyptology discussed here, a parent experiences a gain in happiness as a result of having the child in question. However, even if the version analysed were one in which a parent did not experience a gain in happiness, the analysis would be similar, and the conclusion the same: our intuitions in these cases are not dictated by AU or TU, but by person-affecting intuitions.

45 Parfit, Reasons and Persons, p. 165.

46 These points should be sufficient, but there is also an explanation of how these points regarding Egyptology can be understood not only as being axiological points, but also as points that apply to the context of decision-making.

Since it still might be difficult for some to accept that past values are relevant for ethical inquiry, I will provide an example to give further intuitive support for this claim – an example which will also show that sense can be made of Cases 3 and 4, even as they are written. This example will illustrate two points:

First, a reader who is sympathetic to evidential decision theory (EDT) will take the example to show that choices between 3A and 3B and between 4A and 4B are choices that can confront us. According to EDT, even TU, in practice, might at times provide different prescriptions based on whether it takes a temporally neutral or current/forward-looking form, and the examples suggests that a proponent of EDT should espouse a temporally neutral morality. This illustrates the important point that any persuasive power that the Egyptology objection provides over and above the persuasive power of Reverse Egyptology is, at best, limited to impressing causal decision theorists.

The example, however, is not introduced only for the eyes of proponents of EDT. Its implications are broader: Even if temporally neutral and current/forward-looking versions of TU never actually have diverging prescriptions, if they did, a temporally neutral version would be more plausible. As such, even in cases where the values of individuals in the past are ‘unaffected’, they are as relevant and important as the unaffected values in the future. Although one might try to draw a distinction between the relevance and importance of happiness that will be experienced and happiness that was experienced, no such distinction can be made.

Genetic Twin Example. You live in a small city, and there's a possibility that you have a gene that causes two behaviours – neither of which would be expressed by many individuals without the gene: (1) in one's thirties, one will work extraordinarily hard to bring about a 20 per cent gain of happiness for each of one's fellow citizens, and (2) in one's fifties one will be destructive and bring about a 80 per cent loss of happiness for each of one's fellow citizens. You know that, a century ago, your genetic twin lived in a city with five times the population of yours. However, you do not know whether he had the particular bad gene, and thus whether he manifested these two behaviours. You do, however, know that he lived to age ninety. You, on the other hand, have been infected with a virus that will kill you in your forties, before you could become destructive to your fellow citizens. What should you do? Should you bring about the 20 per cent gain to your population before your death?

Since bringing about the 20 per cent gain to your population is strong evidence that your brother brought about a net 60 per cent loss to his population, it seems as though your choice is between two states of affairs:

Presented with this situation (and assuming EDT), a current/forward-looking version of TU would provide a different prescription than would a temporally neutral version of TU. The current/forward-looking version would prescribe helping one's citizens, but the temporally neutral version would prescribe not helping one's citizens: According to EDT, what's relevant is the probability of what occurred in the past conditional on your doing an act. Conditional on your not helping your citizens it's probable that your brother did not have the gene and that his citizens had values of ten. Conditional on your helping your citizens it's probable that your brother did have the gene and that his citizens had values of six. So according to EDT and a temporally neutral morality, one should not help one's citizens. It seems plausible to me that it would be better not to help one's citizens, and thus to bring about state of affairs B. (Of course, while some find this plausible, others find this claim implausible, and thus view this as a reductio against EDT. People's intuitions on this are far from unanimous.)

Wherever one's intuitions lie on this question, this example should provide further intuitive support for the claim that past individuals should be included in the population relevant for ethical inquiry – even to a causal decision theorist who denies that a temporally neutral and current/forward-looking TU would, in practice, ever diverge. This is for the following reason: just as we can hope for future individuals to be happy, so too can we hope that past individuals were happy. Thus, even in cases unlike the above example, cases where current assessments place identical strings of values for past people in each of the states of affairs being compared, these values should still be taken into consideration by an impersonal morality – be it AU or TU. Once the scope is widened beyond not-unaffected individuals, there do not seem to be grounds to exclude from the relevant population any individuals, past, present or future. As argued earlier in section V, the fact that unaffected individuals in the past cannot be affected does not distinguish them from individuals in the present or future who, it has been stipulated, are unaffected by an act.

47 An additional concern that one might raise (and that an anonymous referee did raise) for a temporally neutral AU view is that it implies variants of the Repugnant Conclusion. The referee's words are worth providing in full: ‘Suppose that, in the past, there existed 10 billion people with lives barely worth living. Each had a lifetime well-being of 1. We have two choices regarding who will exist in the future. In outcome A another 10 billion people will exist, with only marginally better lives than the people in the past. Their lifetime well-being will be 2. In outcome B, 50 million people will exist, with an extremely high lifetime well-being of 100. This means that the average well-being is 1.5 in A, and 1.49 in B. AU therefore favours A. If a main intuition behind AU is “the idea that while we want people to be as happy as possible, we do not necessarily have reason to bring into existence additional happy people” and that “when it comes to happy people, more is not better”, the implication that A is better than B seems troubling. If a smaller and very happy total population is preferable to a larger and less happy one, why shouldn't we have this preference also regarding future populations?’ This is a reasonable concern that is important to raise. The temporally neutral AU theorist does, however, have the following response: the average happiness must be considered for the whole state of affairs, and choosing to focus on some subset of the entire state of affairs will always enable one to describe cases such as the above. Consider a variation of the case offered, but where the world is just beginning, and both options A and B also include the creation of the original 10 billion people at well-being level 1. In this case, it seems that the AU intuition is that A is better than B. However, as I have argued in this article, the only difference between choice one and choice two is whether the unaffected (and unaffectable!) people are in the past or in the future. It seems that this difference is immaterial to which state of affairs is better, and thus that the two cases should be treated alike. It has been argued that the temporally neutral version is (perhaps counter to initial intuitions) the more plausible version. Thus, as attractive as the intuition in the reviewer's objection may be, it seems that revising it is the least difficult of various routes to reflective equilibrium.

48 Parfit, Reasons and Persons, p. 393.

49 Narveson, ‘Utilitarianism and New Generations’, p. 68).

50 Temkin, Larry S., Rethinking the Good (Oxford, 2012), pp. 319–23Google Scholar.

51 As Temkin says, though, the original source of the sort of example that he discusses is Stuart Rachels. Rachels's original example was described in his unpublished Philosophy, Politics, and Economics thesis, ‘A Theory of Beneficence’ (Oxford University, 1993), and Rachels, further discusses the topic in his ‘Counterexamples to the Transitivity of Better Than, Australian Journal of Philosophy 76 (1998), pp. 7183CrossRefGoogle Scholar. In my discussion, I will continue to refer to Temkin's example and discussion, but it should be remembered throughout that Rachels is the original source.

52 Not only has the foregoing discussion attempted to provide an explanation of why one might have asymmetrical intuitions in the positive and negative utility contexts, but it has aimed to show that – once made aware of this explanation – one should revise one's view and maintain a position that has a symmetrical treatment of the two contexts. It is important to note, however, that even this – if accepted – is not (by itself) enough to show that we should reject our prima facie TU-friendly intuitions in the Two Hells context to make them consistent with AU-friendly intuitions in the positive utility context. One could consistently make the opposite adjustment instead. While I do not have enough space here to articulate my explanation fully, the following is my view in a nutshell. First, I believe I have provided strong reasons (throughout this article) to accept average utilitarianism in the positive utility context. If plausible, these provide strong reasons to maintain AU-friendly intuitions in the positive utility context and to revise one's TU-friendly intuitions in the Two Hells context – as opposed to making the opposite adjustment. Second, my explanation for the different intuitions in the positive utility context and the Two Hells context operates by showing how person-affecting intuitions are seeping in and how our person-affecting intuitions provide different responses in the context of benefits and in the context of harms. The difference, however, in brief, is that we care more about the effect for a person in the harm context than in the benefit context. Thus, since I have argued that person-affecting intuitions should be rejected, it is our intuitions in the context of benefits – and thus, in this case, the positive utility context – that more closely approximate an impersonal morality, and, thus, it is these that we should maintain. For this reason, our intuitions in the Two Hells context are more suspect, and it is the intuitions in this context that should be rejected. I owe great thanks to an anonymous reviewer for pointing out the need to provide an explanation for why one should revise one's conflicting intuitions in the manner I suggest we should, as opposed to doing so in the opposite manner.

53 See section IV.

54 I am deeply indebted to Ralph Wedgwood for his comments – comments that have been both numerous and extraordinarily helpful. I am also grateful to two anonymous reviewers for their valuable input.