The non-identity problem is an ethical problem that springs from the fact that our choices can affect not only the wellbeing of future people, but also which specific people will be around to enjoy it.
For a canonical case, consider:
WILMA: Wilma has a condition which will cause any child she conceives to have incurable blindness. However, taking an inexpensive medicine for two months will cure this condition, so that any child she conceives thereafter will have typical eyesight. Disregarding her doctor’s advice, Wilma forgoes the medication and soon conceives a daughter named Pebbles. Pebbles has incurable blindness, but her life is well worth living. Wilma bears the costs of all necessary accommodations for Pebbles’s blindness.1
Most people, when presented with Wilma’s case, judge that she did something wrong by conceiving a blind child instead of a sighted one. But when pressed to say more, it’s hard to pin down what exactly Wilma did wrong. On a widely accepted definition of ‘harm’, you harm someone if and only if you make them worse off than they otherwise would’ve been. But if that’s how we’re defining ‘harm’, Wilma didn’t harm Pebbles, since Pebbles wouldn’t even have existed had Wilma conceived another child.
And if Wilma didn’t harm Pebbles, it’s hard to see how she could have wronged her. But if Wilma didn’t harm Pebbles—and we stipulate that she didn’t harm herself or society or whatever—it’s hard to see what she could have done anything wrong at all. Thus, the non-identity problem.
There are roughly a gazillion attempts to solve the non-identity problem. One of the main ones—the one I’ll dunk on here—says the problem lies in the original definition of ‘harm’ as: ‘making someone worse off than they otherwise would’ve been’.
In place of that definition, philosophers have defended a slew of alternatives:
Someone suffers a harm if and only if they’re badly off.
Someone suffers harm if and only if either they suffer ill-being or their well-being is lower than it was before2.
Someone suffers harm if they’re subjected to something that holds negative prudential value for them3.
An action harms a person if it causes pain, early death, bodily damage, makes them deformed, etc., even if they wouldn’t have existed had the action not been performed.4
Etc, etc, etc.5
Why accept these alternative definitions of ‘harm’? The arguments tend to be that they dodge counter-examples that plague rival accounts, that they’re relatively simple and unified, and that—supposedly—they solve the non-identity problem.
After all, you might think, the non-identity problem is only a problem so long as we think Wilma doesn’t harm Pebbles. But on these alternative definitions of ‘harm’, Wilma does harm Pebbles. The aforementioned accounts don’t say Pebbles is harmed if she was made worse off than she otherwise would’ve been, or made worse of than she was before, or made worse off compared to this or that baseline. Rather, they say Pebbles suffers a harm merely if she exists in some bad state, or something like that. And since blindness is plausibly bad (intrinsically, socially, whatever), conceiving Pebbles in a state of future blindness is a harm to her.
Call this strategy: the harm-based solution.
Typically, opponents of the harm-based solution respond by giving tailored counter-examples to the aforementioned accounts of harm. But this strategy is tedious, and possibly fruitless: a fairly mainstream takeaway from the Gettier literature is that there might no be a Gettier-free definition of any philosophically interesting, natural language concept, HARM included.
The strategy I suggest goes deeper: if it works, we can do away with this entire genre of solution to the non-identity problem in one fell swoop, without getting into the nitty gritty of finding a counter-example-free definition of ‘harm’.
This strategy is suggested at points by David Boonin in The Non-Identity Problem and the Ethics of Future People, and showcased in his reply paper to Molly Gardner6. Here’s how it works.
Suppose we accept the following constraint on solutions to the non-identity problem: a solution to the non-identity problem should be rejected if it implies anti-natalism. This constraint shouldn’t be respected dogmatically, or anything, but it should factor heavily in our deliberations. After all: anti-natalism strikes most as even more implausible that the implausible conclusion that Wilma did nothing wrong. We don’t want to chow down on the larger bullet to avoid biting the smaller one.
Given that constraint, it seems to me that all harm-based solutions to the non-identity problem should be rejected because they must either imply anti-natalism or be saddled with ad-hoc restrictions to avoid that implication.
Why think all harm-based solutions imply anti-natalism? Consider the following case:
CAREER: Wilma wants one child at most, but knows she’d get just as much life satisfaction from pursuing several fulfilling careers—and she can’t do both. If she has a child, the child will be blind.
I’m this case, it doesn’t seem like Wilma would wrong Pebbles should she choose to have her. Even if a non-standard account of HARM turned out to satisfy the concept HARM best, and thereby entailed that Wilma had harmed her child by conceiving him, that wouldn’t be a *wrongful* harm.
But the only difference between CAREER and the standard non-identity case is that in the standard non-identity case, Wilma’s second option is conceiving a sighted child, as opposed to pursuing a fulfilling career. But whether or not some sighted child—who is not the blind child—was Wilma’s second option doesn’t seem, in principle, like the kind of fact that could transform Wilma’s blameless harming of Pebbles into a wrongful harming.
One reason for thinking this is that the formal intuition just seems right. But for those who prefer case-specific intuitions to more abstract, formal ones, I think it can also be supported by cases. Consider, e.g.,
FERTILITY COACH: The case is the same as before, with one specification. One of the (short-term) careers Wilma would go into if she didn’t have a blind child is fertility coaching. She knows that if she goes into this profession, she’ll make possible the conception of a sighted child by a different couple of a different race six years later. (I add the “of a different race six years later” part to ward off the intuitional confusion Boonin warns about in his book, where—he alleges—we have trouble holding the non-identity facts clear in our mind’s eye as we reflect on our intuition about the case.)
In FERTILITY COACH, the counterfactual is (for all intents and purposes) the same as the in the standard non-identity case: *if* Wilma doesn’t conceive Pebbles, *then* she will bring about the conception of a sighted child. But the truth of this counterfactual doesn’t seem like it could (or does) make it the case that Wilma wrongfully harms Pebbles, rather than merely harming her in a non-comparative sense.
One might object that the counterfactual is different in the two cases—“will conceive” vs “will bring about the conception of (by another couple)”. But while that is a difference, it seems like an irrelevant difference; and there are compelling arguments against so-called solutions to the non-identity problem that try to make parental duties the issue.
Most of the energy directed at refuting harm based solutions to the non-identity problem comes in the form of giving counter-examples to this or that non-standard analysis of HARM. But if this kind of blanket strategy works, as I think it does, then we can spend more time on better solutions, and less time worrying about how ‘harm’ should be defined.
The Wilma/Pebbles example comes from David Boonin, who adapts it from a case given by Derek Parfit. This condensed version is quoted from: Noggle, R. (2019). Impossible obligations and the non-identity problem. Philosophical Studies, 176(9): p. 2372.
Cf. Unruh, C. F. (2023). A hybrid account of harm. Australasian Journal of Philosophy, 101(4), 890-903.
Cf. de Villiers-Botha, T. (2020). Harm as Negative Prudential Value: A Non-Comparative Account of Harm. SATS, 21(1), 21-38.
Cf. Harman, E. (2004). Can we harm and benefit in creating? Philosophical Perspectives, 18, 89-113.
For an account that’s too long to summarize in a bullet point, see: Gardner, M. (2015). A Harm-Based Solution to the Non-Identity Problem. Ergo 2(17), 427-444.
Boonin, D. (2019). Solving the Non-Identity Problem: A Reply to Gardner, Kumar, Malek, Mulgan, Roberts and Wasserman. Law, Ethics and Philosophy, 7, 127-156.
Hey Amos, I am on substack again. Do you think non-identity problem refutes or creates major problems for Walter Block and Hans Hermann Hoppe type deontological libertarianism?
I am reading this paper by Lukasz Dominiak - https://repozytorium.umk.pl/bitstream/handle/item/5537/Libertarianism%20and%20Obligatory%20Child%20Support.pdf?sequence=1 - and it looks like deontological libertarianism of Hoppe and Block is extremely implausible.
Thanks for the quick review of the issue with this article! Re the medication example, it strikes me that in one set of cases, it just affects of the development of a foetus _from_ the conception, in which case it can be - given a certain view of personhood - be a matter of giving medication to a person without changing the identity of that person. Obviously this has its own implications for personal identity since the moment of conception preceeds psychological activity (and therefore any psychological continuity), but perhaps the fact that it _does_ fit our intuition that we would just be medicating an unborn person is revealing itself for our notions of personhood and personal identity...
The alternative seems to be that you alter _who_ gets conceived... in which case there are huge questions regarding embryo screening and whether embryo selection alienates parent from child, objectifies child/refuses to treat children/potential children as respected people, etc etc. In terms of embryo screening, I do think there is an answer that could allow for some embryo screening even with strong Christian ideals, at least in cases where IVF is already being done. In the process of IVF, multiple embryos are created. Only one or two can be implanted without posing serious risks to the implanted others' prospects of survival. This situation is analogous to having one or two doses of a life saving medication and five or so potential recipients. It makes intuitive sense that we may give the medication on the basis of the likely contribution of different candidates to wider society (e.g., embryo selection on basis of predicted IQ (by the by, polygenic scoring on IQ is pretty poor at the moment, only explaining ~4% of phenotypical trait variation)). Some would debate whether selecting on potential contribution to society fails to treat each individual as an end in themselves, or fails to respect each individual, etc. But what would be more widely intuitively accepted is giving the life saving doses on the basis of likelihood of the medication working - that seems to follow a basic intuition of treating each as equally worthy of treatment, and giving it in a manner to maximise the amount of life. Under such a situation, embryo screening to pick the one with the best predicted chance of making it to full term seems to fit even with pretty strong Christian (and similar) principles. Seems a bit of a digression, but I think the same could apply in the case of medication that affects the nature of what possible person gets conceived. If you're happy with discriminating medication on the basis of predicted impact on the wider world, then the blindness medication makes sense. If not, then it figures to be reluctant to do so - and wouldn't be incurring harm on Pebbles in the process.
PS I appreciate I may have misunderstood the thought experiment as to whether Pebbles is a different person or not, depending on whether blindness is averted... or if the problem _is_ just because Pebbles would be the same person, but is not acknowledged as a person whilst in the womb. (In the latter case, I would suggest that the fact we think of it as harming Pebbles suggests that our intuitions on personhood do accept that a person's meaningful life does begin before birth, and that we had caveats etc to get around this when it conflicts with other desires.)