Summary

I illustrate and defend actualist object views as my conception of radical empathy, of being concerned exactly with what we would actually care about. As a kind of asymmetric person-affecting view, the most important implication for cause prioritization is probably lower priority for extinction risk reduction relative to total utilitarianism.

I illustrate with an example where I find actualist object views handle changing preferences better than other views (more).
An important implication for cause prioritization is the (Procreation) Asymmetry: we have reasons to prevent miserable lives, but not to create happy or fulfilling lives, for the sake of what those lives would actually care about (more).
1. This is not (necessarily) antinatalist or pro-extinction, but it would probably lead to less priority for extinction risk reduction, compared to total utilitarianism (more).
I highlight some problems for actualism, addressing one dilemma here, and leaving issues with the Repugnant Conclusion and replacement for another piece (more).

Acknowledgements

Thanks to Ariel Simnegar, Lukas Gloor, Justis Mills, Tori, JackM and Vasco Grilo for helpful feedback. All errors are my own.

An example for changing preferences

There’s an adult named Alice and a child named Bobby. In the outcome Adopt, Bobby’s parents are killed and Alice meets and adopts Bobby, coming to care deeply about his welfare — her concern for his welfare is the preference P —, but she only met and adopted Bobby because his parents were killed in Adopt. In Adopt, their deaths have caused him some trauma, and he’d have been better off if they hadn’t been killed, in the outcome Survive. Still, Alice would be an excellent adoptive mother, and Bobby’s life would still seem very good overall to her in the world in which his parents are killed and she adopted him (Adopt), so she’d have a favourable or positive attitude towards his life and welfare (in Adopt). However, Alice, in Adopt, would be genuinely concerned for Bobby’s welfare and what’s best for him, and would know he’d have been better off had his parents not been killed and prefers that they wouldn’t have been killed. In Adopt, she’d therefore prefer Survive. Her attitude matches her impression of Bobby’s overall welfare.

Survive: Bobby’s parents survive, Alice never meets Bobby

Adopt: Bobby’s parents die, Alice adopts Bobby

P never exists

P exists

P is a favourable attitude (e.g. pleasure, approval, appreciation) towards Adopt

According to P, Survive > Adopt

P: Alice’s preference for Bobby’s welfare.

If Survive didn’t seem better to Alice, I’d think it wasn’t just Bobby’s welfare that P was really about, or that P was the only difference, but what else she gained by being with him. Her love for Bobby could be more selfish than selfless. But I will assume here her love for him is in fact much more selfless than selfish, so it’s regrettable to her, for his sake, that his parents were killed. And she could tell you this herself in Adopt.

I will focus on what seems better on behalf of Alice’s preference P, and not how others are made better or worse off except insofar as this is reflected in Alice’s preference P. So, I’m setting aside what’s best for Bobby and his parents from any perspective besides P, and ways Alice might care about them or anything else other than through P. Other reasons would typically be important and worth weighing in practice.

On behalf of Alice’s preference P, it seems to me that it would be better had Bobby’s parents survived. This is a direct intuition I have about the case. But there are two similar actualist object view arguments I would give in defense of this intuition.

Deliberation path argument

If we imagine choosing Adopt ahead of time, anticipating Alice’s preferences (and ignoring others), we take them on and find that Survive would have been better for P, because that’s what P would have actually preferred. But since we haven’t actually chosen Adopt yet, we can just switch to Survive, and then imagine choosing it instead.
If we imagine choosing Survive, we wouldn’t prefer Adopt on behalf of P because P doesn’t actually exist in Survive, so it gives no actual reason to choose otherwise, and we’d stick with our choice of Survive.

All paths end in Survive. Wherever you end up is what’s best or what you should do, so we should choose Survive on behalf of P, if we ignore other reasons.^[1]

Best in the outcome argument

In Adopt, P would prefer Survive. Adopt is a worse outcome to P if it happens.
In Survive, P doesn't actually care either way, because it doesn’t actually exist. Survive is no worse than Adopt, nor Adopt worse than Survive.

There is therefore only one outcome that’s best according to P, according to what P would actually care about: Survive. This is what matters, so Survive is better on behalf of P.^[2]

On the other hand, as far as I can tell, most consequentialist views and axiologies in the literature won’t find Survive better on behalf of P.

Survive: Bobby’s parents survive, Alice never meets Bobby

Adopt: Bobby’s parents die, Alice adopts Bobby

P never exists

P exists

P is a favourable attitude (e.g. pleasure, approval, appreciation) towards Adopt

According to P, Survive > Adopt

P: Alice’s preference for Bobby’s welfare.

Some assign P or her relationship with Bobby positive value on her behalf, and so find Adopt better on her behalf. Her attitude towards Bobby and his welfare is positive, after all, and it is an important relationship. But again, she’d prefer Survive in Adopt according to P, so this would be objectionable to her based on P.

Some ignore P or their relationship, because it’s not the right kind of thing to count morally, or under a presentist or necessitarian preference-affecting view, because P doesn’t yet or necessarily exist, respectively. Or, they might find the two options incomparable on her behalf or on behalf of P. But indifference isn’t sensitive to the fact that she'd prefer Survive in Adopt all else equal, and she would have reason to object in Adopt. And incomparability just seems too strong: even if her preferences don’t strictly agree between the two outcomes, she would find Adopt objectionable and wouldn’t mind Survive.

On some views, like antifrustrationism (Fehige, 1998) and negative preference utilitarianism, on Alice’s behalf, Survive beats Adopt because P or Alice and Bobby’s relationship has negative value in Adopt.^[3] Then, this negative value assignment doesn’t match her own (attitude) towards Bobby and his welfare in Adopt, which is positive. Furthermore, on behalf of P alone, Bobby also dying with his parents would be just as good as Survive and also beat Adopt. However, on behalf of P, she’d disagree with this in each of Survive, Adopt and if Bobby dies, being indifferent in Survive or if Bobby dies, and strictly preferring both Survive and Adopt to Bobby dying in Survive.^[4]

If there were other differences to account for, including the direct interests of Bobby and his biological parents, but also possibly other differences between Survive and Adopt, P could tip the balance in one direction or the other. And P should favour Survive.

So, please don’t let Bobby’s parents die on behalf of Alice’s preference P. That’s exactly what she would disprefer. And please don’t also let Bobby die on behalf of P, either. There’s no world where that would seem better to her due to P.

The Asymmetry

The Asymmetry (or the Procreation Asymmetry) is an intuition that has been characterized as follows (McMahan, 1981):

[W]hile the fact that a person's life would be worse than no life at all (or 'worth not living') constitutes a strong moral reason for not bringing him into existence, the fact that a person's life would be worth living provides no (or only a relatively weak) moral reason for bringing him into existence.

There are different versions of the Asymmetry, some of which are not antinatalist.^[5]

One objection to views which satisfy any version of the Asymmetry is that any argument for the worseness of the presence of some features like suffering, complaints or victims can be reversed into an argument for the betterness of the presence of some symmetric features like flourishing, anti-complaints/gratitude or beneficiaries (MacAskill, 2022, p.172, Beckstead, 2013, pp.80–81).^[6] But this doesn’t apply to actualist arguments, and actualist arguments can imply the Asymmetry (Parsons, 2002, St. Jules, 2019, Cohen, 2020, Spencer, 2021). I will illustrate here.

Suppose P is an individual’s aversive, unpleasant, disapproving or other disfavourable or negative attitude towards an experience, other parts of their life or their life as a whole. For example, P is the unpleasantness of a painful sensation (or a disposition to find that sensation painful), so is a preference against the presence of that sensation, the object of the preference. Consider:

an outcome Nonexistence without the individual, and therefore without the pain and its unpleasantness, i.e. without the object of P and P itself, and
an outcome Pain with the pain and its unpleasantness, i.e. with the object of P and P itself,

and no other affected preferences between the two:

Nonexistence

Pain

The individual doesn’t exist

P never exists and its object is absent

P exists, its object is present, and it counts against the object

P: Nonexistence > Pain

Deliberation path argument

We can imagine starting in Nonexistence or Pain, but all roads lead to Nonexistence. That’s pretty metal.

Best in the outcome argument

The preference P says nothing in Nonexistence, because it doesn’t exist there. Nonexistence does best according to all of the preferences in Nonexistence.
If Pain were to occur, it would exist and favour Nonexistence. Pain does worse according to a preference in Pain, and no better according to any other in Pain.

There is only one outcome that’s best according to all the preferences in it, according to what everyone would actually care about: Nonexistence. So, on an actualist object view, Nonexistence is better.

On the other hand, suppose P is an individual’s attracted, pleasant, approving or other favourable or positive attitude towards an experience, other parts of their life or their life as a whole. For example, P is the disposition to find a particular sensation pleasurable, so is a preference for the presence of that sensation, the object of the preference. Consider:

an outcome Nonexistence without the individual, and therefore without the sensation and the disposition to find it pleasant, i.e. without the object of P and P itself, and
an outcome Pleasure with the sensation and its pleasantness, i.e. with the object of Pleasure and P itself,

and no other affected preferences between the two:

Nonexistence

Pleasure

The individual doesn’t exist

P never exists and its object is absent

P exists, its object is present, and it counts in favour of the object

P: Pleasure > Nonexistence

Deliberation path argument

We have two separate paths, one starting from Nonexistence, and another starting from Pleasure, and we can either end in Nonexistence or in Pleasure. So each is best, or permissible.

Best in the outcome argument

Again, the preference P says nothing in Nonexistence, because it doesn’t exist there. Nonexistence does best according to all of the preferences in Nonexistence.
If Pleasure were to occur, P would exist and favour Pleasure. We’d have reason in Pleasure to choose Pleasure or to have chosen Pleasure. But this wouldn’t matter if Nonexistence were to occur, because P wouldn’t exist.

Each outcome is best according to every preference in it, according to what everyone would actually care about. So, on an actualist object view, each is best, or permissible.

Asymmetry and extinction

An important implication of the Asymmetry for cause prioritization is that our reasons to reduce the risk of extinction will be weaker than on total views. If we allow extinction, all of the potential “good” lives that never come to exist won’t actually care to have existed instead, because they won’t care about anything at all.

On the other hand, it doesn’t necessarily follow that, even ignoring indirect reasons, it would be better to cause or allow extinction for the sake of future people, in case at least one will have a bad life, or in case there’s a risk that at least one will have a bad life. That would plausibly follow on much stricter antinatalist Procreation Asymmetries, but my arguments here are compatible with non-antinatalist Procreation Asymmetries. There are multiple different ways to generalize my actualist arguments here to more complex cases, and I will discuss some in another piece.

Some problems for actualism

There are cases where we can’t do what everyone would prefer, but this is often a matter of conflicting preferences and making tradeoffs. We can’t always make everyone happy. And even one person can have their own conflicting preferences. We can’t always do best for each of someone’s preferences.

However, Bykvist (2007, 2010) proposed an unusual dilemma for actualist views:

Suppose that you know that if you get married, then you will prefer being unmarried to being married. You will adopt certain perfectionist ideas about marriage and think that your marriage does not live up to the standards. However, if you stay unmarried, you will accept less exacting requirements and prefer being married to being unmarried. Now, if you get married, then you ought not to get married, since you will then prefer not being married, whereas if you do not get married, you ought to get married, since this is what you will then prefer.

If how informed or rational a preference is matters, let’s assume that neither preference is more informed or rational than the other; they are adopted for reasons that don’t come from your (in)experience with marriage.

No matter which outcome you choose or consider to be better, it will be worse to you in that outcome. This looks like a dilemma for actualism.

Still, it seems resolvable in a way that’s intuitively satisfying. We’re concerned with following your preferences, being guided by what you would actually care about. Your preferences in each outcome point to the other outcome. In this case, we could just ask which you would prefer more: to be unmarried if you marry, or to marry if you’re unmarried? And then just pick whichever you would prefer more.^[7] Equivalently, we can just ask whichever you would disprefer least having done it.^[8]

Spencer (2021) proposed an interpersonal dilemma with the same structure:

Misery or Moremisery. We are deciding whether to create Misery or another person, Moremisery. We know that Misery, if created, would lead a miserable life; that Moremisery, if created, would lead an even more miserable life; and that nobody other than Misery or Moremisery will be affected by our choice.

And we can resolve it the same way: Moremisery prefers Misery more than Misery prefers Moremisery, so we should choose Misery.

For different versions of actualism and related discussion, see Hare, 2007 (especially weak actualism), St. Jules, 2019, Cohen, 2020 and Spencer, 2021. Also similar are the narrow views described by Thomas (2019, section 5.2) and Pummer (2024). Object versions of these views would usually agree with the actualist object view arguments made in this piece. However, I don’t find these specific views — or any other view I’ve read about online — entirely satisfying. Most have some specific problems. With more than two options, most of these views can lead to the Repugnant Conclusion, the Very Repugnant Conclusion and replacement (St. Jules, 2024).

In another piece, I will motivate and outline a new approach to population ethics and the ethics of changing preferences that can overcome these problems and better match person-affecting intuitions. The approach generalizes Dasgupta’s method (Dasgupta, 1994, Broome, 1996, St. Jules, 2024) to one that can generate choice rules over more than 2 options from any binary choice rule or set of pairwise comparisons. The solution again seems to lie in just being more radically empathetic. Radical empathy is the gift that keeps on giving.

^{^}
I will generalize this reasoning in another piece.
^{^}
The argument is consistent with weak actualism: an option X is permissible if and only if it’s best according to the preferences in X, after aggregating them (Hare, 2007, Spencer, 2021). I will discuss this view, problems with it and potential solutions in another piece.
^{^}
For example, because the preference P in Adopt is partly frustrated, we could assign it negative value in Adopt, but no or 0 value in Survive, because it doesn’t exist in Survive.
^{^}
However, she wouldn’t disprefer Bobby dying if Bobby dies so wouldn’t have reason to object to it.
^{^}
Antinatalist versions of the Asymmetry may not allow lives of positive well-being to offset lives of negative well-being, or the possibility of positive well-being to offset the risk of negative well-being. This can be avoided with other versions. Thomas (2022) defines:
The Group-Level Asymmetry. Between creating no one, and creating some additional people, with no effect on those who independently exist…
1. If the average wellbeing of the additional people would certainly be negative, we ought not to create them.
2. If the average wellbeing of the additional people would certainly be positive, it is permissible but not required to create them.
and
The Expectational Asymmetry. Between creating no one and creating some additional people, with no effect on those who independently exist…
1. If the additional people would all have negative expected wellbeing, we ought not to create them.
2. If the additional people would all have positive expected wellbeing, it is permissible but not required to create them.
We can combine the two:
The Expectational Group-Level Asymmetry. Between creating no one and creating some additional people, with no effect on those who independently exist…
1. If the expected total wellbeing of the additional people is negative, we ought not to create them.
2. If the expected total wellbeing of the additional people is positive, it is permissible but not required to create them.
We could also consider similar asymmetries applied to preferences instead of directly to whole-individual well-being.
^{^}
MacAskill (2022, p.172) writes:
If we think it’s bad to bring into existence a life of suffering, why should we not think that it’s good to bring into existence a flourishing life? I think any argument for the first claim would also be a good argument for the second.
Beckstead (2013, pp.80–81) responds to “the victim requirement” for victims of existence with the possibility of beneficiaries of existence, e.g. those glad to have been born, and poses anti-complaints as the symmetric counterparts of complaints.
^{^}
Bykvist (2007) also defends the dependence of an action’s moral status on whether or not it is performed, as long as the theory can still guide action:
It seems paradoxical to say that an action's normative status - whether it is right, wrong, or obligatory - depends on whether or not it is performed. In this paper, I shall argue that in itself this dependency is not paradoxical. I shall argue that we should not reject a normative theory just because it implies this kind of dependency. Not all dependencies of this kind are bad, or at least not bad enough to warrant wholesale rejection. Instead, we should reject a theory when this dependency makes it a poor guide to action, in particular, when the dependency makes it impossible for agents to comply with the theory.
^{^}
On their equivalence:
1. In outcome Marry, you have a preference P for Unmarried over Marry. You prefer Unmarried and disprefer Marry by the strength of P.
2. In outcome Unmarried, you have a preference Q for Marry over Unmarried. You prefer Marry and disprefer Unmarried by the strength of Q.
Then,
You prefer Unmarried in Marry (P) more than you prefer Marry in Unmarried (Q)
iff
You disprefer Marry in Marry (P) more than you disprefer Unmarried in Unmarried (Q).
iff
P in Marry is stronger than Q in Unmarried