Review of Living in a Strange World

Caspar Hare’s latest book is a lot of fun

I recently reviewed Caspar Hare’s Living in a Strange World for Ethics. The journal will presumably paywall the official paginated version of my review, but here’s the plain text “pre-print”, in case it’s of broader interest…

Living in a Strange World explores the practical implications of three descriptive hypotheses that we have surprisingly strong reason to believe: (1) Many everyday actions radically change the entire course of the future, by changing who will later exist. (2) In the vicinity of every person there exists a vast number of sentient sub-people, like the being consisting of all of me minus one particular hair. (3) The universe is spatially infinite, and contains infinitely many qualitative duplicates of any state of affairs that had a non-zero chance of being realized in our visible universe. In each case, Hare first motivates the descriptive claim, then explains its potentially radical practical implications, and finally argues that we can block some but not all of those putative implications. Crucially, Hare eschews the “base accommodation” of a mere Moorean shift: “If there are good independent reasons to believe the metaphysics, and there are good, independent reasons to believe that the metaphysics demands of us that we reform, then we must reform” (20). Clearly written, carefully argued, and readable within a couple of afternoons (while rewarding closer study), the book offers a rollercoaster ride of mind-bending analytic philosophy at its finest.

I found the first and third parts especially enjoyable, since the underlying descriptive claims are both credible and definitely concern the real world. The second is more debatable. It hinges on the ‘special composition question’ (SCQ): “Which plural collections… of material things are such that their members make up a material thing?” (109). As Hare charmingly relates, non-philosophers uniformly dismiss this as a bad question. I suspect they’re right to do so, as the competing answers offered by ontologists don’t correspond to any clear difference in reality. If there’s a sense in which we can say there exists a multitude of overlapping individuals sharing my thoughts and experiences, that may say more about our conceptual schemes than about the world itself.

If you don’t share my deflationary metaontology, you may find Part 2 more gripping. And even for deflationists, there is a puzzle about how we can most coherently talk about individual existents without getting ourselves into practical difficulties. (Hare flags that larger people, with more atoms to subtract, will contain vastly more sub-people. If all sentient individuals count equally and independently, then the largest sentient being and its “subbers” collectively form a utility monster.)

Hare’s own solution might be considered a form of stealth deflationism. He argues that what is bad about someone’s suffering is “what makes your fully imagining being [them] in the phenomenal centered way an accurate representation of some aspect of reality” (165). He also argues that to fully imagine being a person just is to thereby imagine being each corresponding sub-person. That’s how he gets the conclusion that what’s bad about any sub-person’s suffering is nothing over and above what’s bad about the person suffering—there’s no further, independent harm or bad to count here. But that seems to be because there’s no independent aspect of reality. When we talk about a sentient sub-person, the sentience in question just is that of the person. But then it’s unclear what the sub-person’s “existence” really amounts to. There’s a person, with some atoms as parts that we can imagine subtracting. But it seems unnecessarily confusing to characterize the result as a sentient individual in their own right.

Granted, the claim that it’s “unnecessary” can’t be fully vindicated without an alternative conceptual framework that’s less apt to confuse. Consider how different the problem looks from the vantage of an explicitly mind-first view like Berkeleyan idealism. There’s no temptation to attribute each being’s mind to any particular collection of atoms. Our bodies are our physical avatars for interfacing with the world, but ontological fussiness about their precise boundaries would seem beside the point. Any suggestion that each sub-body (containing enough of our brains, perhaps) possesses our consciousness gets things backwards. Now relax the assumption of idealism, while holding on to the idea that we (as mental creatures) possess our bodies rather than vice versa. A view along these lines, fully fleshed out, may dissolve the problem.

On Hare’s revisionary view, we literally contain multitudes. The question, “Which individual am I?” falsely presupposes that only one individual is thinking that thought. Once we disambiguate the indexical, we can truly think “I_person am a person,” along with any number of other, equally true, self-attributions of sub-personhood (154). But I think it makes more sense to think of oneself as a basically mental being—a bundle of co-conscious experiences, perhaps—that is merely derivatively associated with a body. Co-conscious experiences don’t seem to be merely optional parts in the way of a stray hair or atom. Our mental boundaries may be more secure than our physical ones.

Let’s turn to the more substantial challenges from the first and third parts of the book. Both may motivate a kind of practical axiological nihilism: a sense that nothing we do will make things “better or worse, on aggregate, for strangers” (67, cf. 189). The fragility of future identities (Part 1) threatens this result if we combine it with the parity view that most lives are “on a par” rather than precisely comparable in welfare value. Since any significant action can be expected to radically change who subsequently comes to exist, widespread parity between possible lives would imply that almost nothing we do could be expected to make the future overall better or worse. (Perhaps existential risk mitigation would be an exception? Having no future lives would surely be worse.) Part 3 suggests that the universe may contain infinitely many lives, in which case it’s obscure how to compare the overall value of different possible outcomes. It seems there will be infinite value (and infinite disvalue) no matter what we do.

Hare’s solution, in both cases, appeals to the surprisingly helpful principle that Reasons are Contrastive: “A consideration is a reason to do X rather than Y, only if it is something that can be said for doing X, but not for doing Y” (69, 191).

We have some reason to help any identifiable individual. The fact that helping would neither increase nor decrease aggregate welfare is not a reason against helping; it is, Hare notes, “not suitably contrastive” (193). Nor does it undermine our reason to help an identified individual, assuming (as seems right) that we ought to care about individuals and not just “how often the qualities associated with [their] thriving are instantiated in the universe” (195). So that’s something.

It’s still a bit unsettling, though. If Hare is right, our reasons of beneficence may be significantly weaker than I’ve always taken them to be. (I care about individuals, but I also care a lot about making things go overall better; I’d feel much less motivated to give to charity if I knew it would fail at the latter goal.)

Hare concludes that “if the world is spatially infinite and suitably homogeneous, then when we are in the business of helping people, we have most reason to help the people we know” (211). Why is this? Curiously, it stems from a kind of impartiality principle that Hare calls Morality Is Not First and Foremost About Relations to Me. Individuals you know can appear in “non-self-involving considerations” that pick them out directly by name. We ordinarily assume that there are non-self-involving considerations about how our actions affect strangers, like “More strangers will live on if I do this, fewer strangers will live on if I do that” (210). But in an infinite and suitably homogeneous world, an infinite number of strangers will live on no matter what we do. The best we can do is appeal to self-involving considerations like “I will save a stranger by doing this,” or “More strangers will live on my planet if I do this.” Such considerations may carry some weight—but less, Hare suggests, than the weight we should give to a non-self-involving consideration like “Jasmine will live on if I do this.”

Why can’t we appeal to the fact that more strangers will live on Earth if I do this? Hare replies that it turns on “whether moral reasoning should be exclusively focused on people” (214). It may seem objectionably arbitrary to be especially concerned about the events on a particular planetary surface as such. (It may be more comprehensible to be especially concerned for those who share a planet with you, but that’s self-involving.)

I wonder if we could instead bracket the unaffected universe, and welcome aggregate improvements among those affected by our choices? The individuals whose well-being hinges on my choices presumably form a finite population, since the infinite spaces lie beyond my causal reach. Suppose that this finite population fares better in aggregate if I act. That’s not about me, it’s about affected people, wherever they may be. A move along these lines might rescue the idea that we have non-self-involving reasons to help strangers, even in an infinite universe. Making things better in aggregate for everyone with a stake in my choice, and no worse for anyone else, would seem to capture what I originally cared about when I hoped to make things better overall. The mere existence of (even infinitely many) other galaxies unaffected by my actions shouldn’t undermine my reasons to help those I can.

Let’s return to Part 1’s challenge from parity. On Hare’s view, we can’t expect to bring about a better (or worse) future, even just considering the finite future population of our planet. The different lives that would exist in either case are too different; the resulting populations are on a par. As a result, Hare thinks, “we know that it is never the case that, in the objective sense of ‘ought’, we ought to give to strangers” (80). But so long as we don’t know too much about the competing candidate lives, we may yet have contrastive subjective reasons to help strangers.

I find it hard to believe that it’s never the case that we objectively ought to help strangers, as I find it hard to believe that this never makes things better in aggregate. For example, I expect that, thanks to the eradication of smallpox, the individuals who exist now are better off than the different individuals who would otherwise have existed in our place. If someone invents a cure for cancer, I would similarly expect the next generation of people to have better lives (on net) than those that would otherwise exist. So Hare’s parity claim seems wrong to me. It may be difficult to make precise welfare comparisons with any confidence, but I don’t see much reason to infer from this epistemic difficulty that there typically is no fact of the matter about which of two lives is in fact better for the person living it.

If I’m right that the barrier to comparing lives is merely epistemic, then subjective betterness is also a (fallible, but reasonable) guide to objective betterness. If all you know about two possible lives is that one would live a year longer than the other, then that is some reason to expect (with probability slightly greater than 0.5) that the longer life is also the objectively preferable life. Similarly, we have some reason to expect (with probability just barely greater than 0.5) that saving someone’s life will result in an objectively preferable outcome for people on the whole.

So I remain optimistic about objective oughts. Massive invisible effects don’t affect the case for beneficence, to my mind, since we’ve no reason to expect the unforeseeable consequences of helping to skew negative. Hare worries about a normative asymmetry, however. He thinks the near-50% chance that we “end up an unwitting villain”—in the sense that our best efforts prove counterproductive in the long run—is “a further strike against making serious efforts to give to strangers” (61).

I think the heightened normative stakes here are symmetrical. It’s certainly unsettling to dwell on the risk that our best efforts prove counterproductive. But as far as others’ interests are concerned, that risk is balanced by the likelihood that our efforts prove even better than expected. From behind a veil of ignorance, but otherwise knowing just what we do about the likely effects of our actions, any reasonable person would want the helpful action to be performed. So it would seem objectionably self-indulgent to refrain merely due to unwillingness to risk oneself proving an “unwitting villain.” That narrative property is not morally significant in comparison to the effects on others, and we’ve seen that the risks to others are decisively outweighed by the expected benefits. So helping others remains worthwhile in expectation.

Consider also that treating inaction as morally “safe” by default makes most sense against a background of general safety and abundance, in which unexpected downsides of high-variance actions can be expected to straightforwardly outweigh unexpected upsides. That is not the situation we are considering here—where the status quo includes far too many global deaths from cheaply preventable illnesses—so we should be wary of misapplying a heuristic that is meant for a different context. However worried you are about being an “unwitting villain,” you should worry even more about being a “passive villain” who easily could have prevented very bad things from happening, but didn’t even try.

Still, if one remains unconvinced by my normative claims, Hare offers another response (besides parity) for mitigating the risk of unwitting villainy. Given counterfactual indeterminacy, there is no unique alternative to pick out as the way things would have gone if you’d done otherwise. Any number of possible worlds (varying radically in long-run value) may count as equally close. So if things turn out unexpectedly badly, there’s no guarantee—only a likelihood—that it would have been better to do nothing. That is, Hare suggests, “not so unwittingly villainous” (63).

Let me conclude with my favorite argument from the book. According to The Great Yes (95), we all ought to be (something like) grateful for every actual event that occurs. Given chaos, every actual individual owes their existence—and all the good it contains—to almost every event that occurred prior to their conception. Assuming that most lives are worth living, the beneficiaries of any event thus vastly outnumber and outweigh the victims. Of course, we should still hope for better events rather than worse ones beforehand, and help victims of injustice when we can. But once the facts are settled, we could not reasonably prefer an alternative that would deprive every actual future person of their entire lives. The Great Yes and Reasons are Contrastive are my two main takeaways from the book.

The ethical implications of revisionary metaphysics form a comparatively neglected topic, but also, as Hare demonstrates, a delightfully interesting and philosophically fruitful one. I rarely come across a book as thought-provoking as Living in a Strange World. I highly recommend it, and wish there were more like it—unless that would rob future people of their existence, of course.