Hostname: page-component-77c89778f8-swr86 Total loading time: 0 Render date: 2024-07-17T19:17:13.207Z Has data issue: false hasContentIssue false

An Actualist Explanation of the Procreation Asymmetry

Published online by Cambridge University Press:  18 September 2019

Daniel Cohen*
Charles Sturt University
*Corresponding author. Email:


While morality prohibits us from creating miserable children, it does not require us to create happy children. I offer an actualist explanation of this apparent asymmetry. Assume that for every possible world W, there is a distinct set of permissibility facts determined by the welfare of those who exist in W. Moral actualism says that actual-world permissibility facts should determine one's choice between worlds. But if one doesn't know which world is actual, one must aim for subjective rightness and maximize expected actual-world permissibility. So, because one should expect actual people to be worse off than they could have been if one creates a miserable child, creating a miserable child is subjectively impermissible. And because one should expect actual people to be at least as well off as they could have been if one fails to create a happy child, failing to create a happy child is subjectively permissible.

Copyright © Cambridge University Press 2019

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.)


1 Peter Singer introduces these two variants of utilitarianism, which he calls the ‘total’ and ‘prior existence’ views (Singer, Peter, Practical Ethics, 1st edn. (Cambridge, 1980))Google Scholar. For a survey of the difficulties faced by various theories in justifying the asymmetry, see Roberts, Melinda A., ‘An Asymmetry in the Ethics of Procreation’, Philosophy Compass 6 (2011), pp. 765–76CrossRefGoogle Scholar.

2 McMahan, Jeff, ‘Problems of Population Theory’, Ethics 92 (1981), pp. 96127CrossRefGoogle Scholar.

3 Broome, John, Ethics Out of Economics (Cambridge, 1999), p. 230CrossRefGoogle Scholar.

4 Broome, Ethics, p. 168.

5 This view is defended in Williamson, Timothy, Modal Logic as Metaphysics (Oxford, 2013)CrossRefGoogle Scholar.

6 Benatar, David, Better Never to Have Been (Oxford, 2006)CrossRefGoogle Scholar.

7 Harman, Elizabeth, ‘Can we Harm and Benefit in Creating?’, Philosophical Perspectives 18 (2004), pp. 89113CrossRefGoogle Scholar, at 98.

8 See McMahan, Jeff, ‘Asymmetries in the Morality of Causing People to Exist’, Harming Future Persons: Ethics, Genetics and the Nonidentity Problem, ed. Roberts, Melinda and Wasserman, David (New York, 2009), pp. 4968CrossRefGoogle Scholar, at 57; Parfit, Derek, Reasons and Persons (Oxford, 1986), pp. 525–6CrossRefGoogle Scholar.

9 But everything I say is compatible with non-utilitarian views. One might take moral actualism, as I have formulated it, to be merely a ‘theory of beneficence’; that is, a theory describing the moral reasons generated by welfare considerations alone. So, when I say that some action is ‘permissible’, according to actualism, one might interpret this as meaning ‘permissible, in so far as permissibility is determined only by welfare considerations’.

10 For earlier discussions of moral actualism, see Bigelow, John and Pargetter, Robert, ‘Morality, Potential Persons and Abortion’, American Philosophical Quarterly 25 (1988), pp. 173–81Google ScholarPubMed; Harman, Elizabeth, ‘Creation Ethics: The Moral Status of Early Fetuses and the Ethics of Abortion’, Philosophy and Public Affairs 28 (2000), pp. 310–24CrossRefGoogle Scholar; Narveson, Jan, ‘Utilitarianism and New Generations’, Mind 76 (1967), pp. 6272CrossRefGoogle Scholar; Parsons, Josh, ‘Axiological Actualism’, Australasian Journal of Philosophy 80 (2002), pp. 137–47Google Scholar; and Warren, Mary Anne, ‘Do Potential People Have Moral Rights?’, Canadian Journal of Philosophy 7 (1977), pp. 275–89CrossRefGoogle ScholarPubMed. I am especially indebted to Parsons’ conceptualization of these issues.

11 Carlson, Erik, Consequentialism Reconsidered (Dordrecht, 1995), p. 100CrossRefGoogle Scholar. Carlson attributes the principle to Wlodek Rabinowicz.

12 But suppose that I am choosing between Φ(Wn) and Φ(Ws) and that I have a 0.5 credence that I will take each option. It follows (given moral actualism) that I should have a 0.5 credence that both Φ(Ws) and Φ(Wn) are permissible, and a 0.5 credence that only Φ(Wn) is permissible. But if I have a 0.5 credence that Φ(Ws) is impermissible, surely I shouldn't do it! This line of thought is tempting, but note that it tacitly employs a second kind of permissibility: an account of what to do when one is uncertain about which of one's options are ‘objectively’ permissible. So, such a view requires a full characterization of how ‘subjective permissibility’ works. (I will describe such a theory later.) One concern with this approach to subjective permissibility is that it's doubtful whether one can form stable unconditional credences about what one will do under the circumstances. If one decides that Φ(Ws) is indeed subjectively impermissible, one's credence that one will do it should drop. But then, as one's credence that Wn is actual rises, so too should one's credence that Φ(Ws) is objectively permissible (assuming moral actualism). But as one's credence that Φ(Ws) is objectively permissible rises, the case for its subjective permissibility improves, and so one should increase one's credence that one will do it. But as one's credence that Ws is actual rises, one's credence that Φ(Ws) is objectively permissible should drop (assuming moral actualism). And so on…

13 This formulation is ambiguous, as ‘the actual world’ can be understood either as an indexical, picking out the world in which the agent is located, or as rigidly designating some particular privileged world. However, we should prefer the former interpretation, as moral actualism would otherwise have odd implications for the moral obligations of merely possible people. Specifically, it would imply that a merely possible person is obligated to maximize our welfare (assuming ours is the unique actual world), and to ignore the welfare of her world-mates. For instance, on this interpretation, actualism would entail that Sherlock Holmes is required to ignore completely Watson's interests, in so far as Watson isn't actual, and to do what he can to maximize the interests of Tom Cruise, even though Tom Cruise doesn't exist in his world.

14 There are other, less plausible, alternatives to moral actualism which I won't consider. For instance, one might argue (a) that there is one particular world such that I should do what is permissible relative to this world; (b) that I may perform an action as long as it is permissible relative to some world; or (c) that I may do only what is permissible relative to all the worlds I am choosing between.

15 Hare, Caspar, ‘Voices from Another World: Must We Respect the Interests of People Who Do Not, and Will Never, Exist?’, Ethics 117 (2007), pp. 498523CrossRefGoogle Scholar. Hare uses the label ‘strong actualism’ for the view I have been calling ‘moral actualism’, i.e. the view that an action is permissible iff it is at least as good as any alternative for actual people. (Hare himself uses the term ‘moral actualism’ for the weaker claim that an action if actually performed is permissible iff it is at least as good as any alternative for actual people.)

16 Why are these two claims equivalent? According to moral actualism, Φ(Wx) is permissible iff it is permissibleW@. Now, if I were to perform Φ(Wx), then W@ would be Wx. So, if I were to perform Φ(Wx), then Φ(Wx) would (according to moral actualism) be permissible iff it were permissibleWx.

17 See Hare, ‘Voices’, pp. 503–5, for a detailed development of this second objection. See also Bykvist, Krister, ‘The Benefits of Coming into Existence’, Philosophical Studies 135 (2007), pp. 335–62CrossRefGoogle Scholar, at 351–2.

18 For a useful overview of arguments for and against the possibility of moral dilemmas, see Terrance McConnell, ‘Moral Dilemmas’, Stanford Encyclopedia of Philosophy, <> (2018).

19 See Scanlon, Thomas, What We Owe to Each Other (Cambridge, MA, 1998), p. 390Google Scholar.

20 One might wonder why I don't employ a more standard account of subjective permissibility, according to which an action is subjectively permissible iff it maximizes expected value. Suppose I am choosing between Φ(Wh) and Φ(Wn), and that I know that the expected value of Φ(Wh) (i.e. the valueWh of Wh) = x, and that the expected value of Φ(Wn) (i.e. the valueWn of Wn) = y. A practical problem is that I don't know how to compare x and y, as each value is relative to a different world; a deeper, theoretical, problem is that there doesn't appear to be any fact of the matter about how to make such comparisons.

21 Of course, one doesn't always know exactly which world will obtain as a result of a given action. But we can easily extend subjective actualism to deal with this more familiar kind of uncertainty in the following way:

Φ is subjectively permissible iff its expected permissibility (the sum, for every Wx, of: the degree of permissibilityWx of Φ(Wx) * the probability that Wx is the case, conditional on Φ) is at least as great as that of every alternative.

In the procreation cases with which we are concerned, the probabilistic expectations involved are trivial, because it is stipulated exactly which world will obtain, given each option. But even given these stipulations, it is still appropriate to appeal to one's expectations in cases where one is ignorant about which of one's options is objectively permissible (because one is ignorant about which action one will, in fact, perform).

22 Subjective actualism can deal with these problem cases because it determines the subjective permissibility of an option by comparing its degree of objective permissibility, if performed, with that of each of one's alternatives. In contrast, because weak actualism takes the permissibility of an option to be entailed by its outright normative status, if performed, weak actualism is unable to deal with the problem cases. Weak actualists might want to borrow this comparative approach in order to deal with the problem cases, but their view will then simply collapse into subjective actualism.

23 Thus formulated, subjective actualism is extensionally equivalent to the Harm Minimization principle defended in McDermott, Michael, ‘Utility and Population’, Philosophical Studies 42 (1982), pp. 163–77CrossRefGoogle Scholar. However, as Harm Minimization is a theory of objective permissibility, this theory is unable to explain how objective and subjective permissibility can intuitively diverge in procreation cases, even when the effects of one's actions are known. My view can do this. For instance, given a choice between Φ(Wh) and Φ(Wn), my view says that Φ(Wn) is subjectively permissible, but also that Φ(Wn) would be objectively impermissible if Φ(Wh) were chosen. This resonates with the commonly expressed sentiment that, for every actual happy child, it would have been impermissible, in a sense, not to create that child.

24 Parfit, Reasons, ch. 16.

25 Jackson, Frank, ‘Decision-Theoretic Consequentialism and the Nearest and Dearest Objection’, Ethics 101 (1993), pp. 461–82CrossRefGoogle Scholar, at 462–3.

26 Parfit, Reasons, p. 367.

27 Michael McDermott describes a variation of Harm Minimization theory which is designed to generate plausible implications in this case (McDermott, Michael, ‘Harms and Objections’, Analysis, 79 (2019), pp. 436–48CrossRefGoogle Scholar). A similar variation could possibly be adopted by proponents of the individualist view.

28 Otsuka develops a theory which similarly grounds permissibility in the ‘complaints’ a person has in a world, relative to other available worlds. His view also has the implication that permissibility is choice-set dependent (Otsuka, Michael, ‘How it Makes a Moral Difference that One is Worse Off than One Could Have Been’, Politics, Philosophy & Economics 17 (2018), pp. 192215CrossRefGoogle Scholar).

29 Of course, there are usually many more than two possible children I might create, given a particular procreative option. At the very least, there are as many possible children as there are possible combinations of available gametes, given the circumstances of a procreative option. However, to keep things manageable, let's maintain the fiction that there are just two possible particular children I could create on this occasion.

30 Nevertheless, in those unusual cases where the assumption does hold true, subjective actualism has the counterintuitive implication that both options are subjectively permissible. Perhaps Φ(Wh+) is intuitively obligatory only if Happy+, if she existed, would be much happier than Happy would have been, if she existed. I set this complication aside for present purposes.

31 The solution to the non-identity problem sketched here is similar to that defended in Hare, Caspar, The Limits of Kindness (Oxford, 2013)CrossRefGoogle Scholar.

32 I am grateful to audiences at Charles Sturt University, The Australian National University, the 2014 Society for Applied Philosophy Conference, and the 2014 Rocky Mountain Ethics Conference, for their useful feedback. I would especially like to thank the following people, who each played a significant role in helping me develop the ideas presented in this article: Ralf Bader, Adam Bales, Wylie Breckenridge, Campbell Brown, John Cusbert, Hilary Greaves, Toby Handfield, Patrick Kaczmarek, Robyn Kath, Morgan Luck, Josh Parsons, Teruji Thomas, Anabella Zagura, and two anonymous referees for Utilitas. This work was partly funded by the Leverhulme trust (RPG-2014–064) and the Australian Research Council (DP110101810).