Parfit’s non-realism 🔗

I’ve always thought of Parfit as a moral realist, but volume 3 clarifies that he does not categorize himself that way. He still thinks normative beliefs can be objectively true or false, which in my mind is the essence of realism. But he wants to avoid implying that normative facts have any sort of metaphysically mysterious existence. Rather, he thinks, normative facts—as well as, for example, mathematical facts—are objectively true but not “ontologically weighty”. Unlike physical facts, they “are not made to be true by corresponding to how things are either in the natural world or in some other part of reality.”(OWM3, p. 64).

The whole idea of non-“ontologically weighty” facts itself seems quite mysterious. But it does seem like a fitting way to describe mathematical facts, so I don’t want to dismiss it too quickly.

Here’s one hesitation I have: To me, the clearest normative truths are that happiness is good and suffering is bad. Unlike (perhaps) mathematical truths, the justification for these normative beliefs is not some sort of internal coherence among them. Rather, the justification has to do with what happiness and suffering feel like. Those feelings—those qualia—surely are an “ontologically weighty” part of reality, and until we understand their nature I doubt we can fully understand the nature of the normative truths that are somehow grounded in them. And currently we don’t understand qualia at a metaphysical level at all; we don’t have a good theory of what consciousness is or how it works.

Temkin and internalism 🔗

The first essay in DARM is by Larry Temkin, who seems to agree with Parfit’s basic view but wants to challenge a few peripheral points. In section 3 he argues that externalism (which says you can have a reason to do something even if you lack any ability to be motivated to do that thing) can find more common ground with internalism (which says what you have reason to do is ultimately dependent on what motivates you) than Parfit’s defense of externalism might suggest. Even if we believe that reasons would exist whether anyone was motivated by them or not, we still want to say that…

…if there were decisive reasons to believe, care about, or do certain things, then, in ideal circumstances, a fully and solely rational being … would, in fact, be motivated to believe, care about, or do those things…(DARM, p. 19)

So “externalists should grant, and even insist on, the importance of the internalist insight that there is a link between reasons and motivation.”(DARM, p. 21-22) That seems right. But I think there’s an important difference that could be characterized roughly like this: internalists define reasons in terms of what we’re motivated to do, while externalists think what we’re motivated to do reflects (in part) the degree to which we have (or lack) a capacity to properly grasp and respond to reasons. Parfit says something similar in his response to Temkin, using as an example the decision whether to stop smoking: (certain kinds of) Objectivist might choose to quit smoking because avoiding an early death is a reason to do so, while for (certain kinds of) Subjectivist “the inference runs the other way” (OWM3, p. 53): whether avoiding early death counts as a reason for them depends on whether it would affect their choice.

Temkin and mattering 🔗

In section 4 of his essay Temkin pushes back against Parfit’s worry that “nothing matters” unless normative realism is true. Part of Temkin’s argument is simply that the conclusion is too implausible: it’s clearly true, for example, that it matters whether people have miserable lives or happy and fulfilling lives. I agree, but would draw a slightly different lesson from this. Consider these three propositions:

1. some things matter
2. if normative realism is false, nothing matters
3. normative realism is false

At most two of these can be true. I have the most confidence in A, so if I became convinced of C, I’d reject B. Similarly, though, if I became convinced of B, I’d reject C. And in fact B seems more probable to me than C, so for me this forms an argument that C is probably false (meaning normative realism is probably true). So I agree with Temkin in the sense that I don’t think we need to worry that we might learn that nothing matters. Things obviously matter. But it’s still worth discussing whether it’s logically necessary for normative realism to be true for anything to matter, because if it is, well, that tells us that normative realism is true!

Deontic badness and the Murder-Preventing Argument 🔗

Sadly the “Murder-Preventing Argument” does not refer to an argument that persuades people not to commit murders. It’s Parfit’s term for an argument about whether there’s much “deontic badness” in wrong actions—that is, whether the mere fact that someone acted wrongly makes the world significantly worse, independently of the badness of any of the effects of that action.

Parfit does not necessarily reject the conclusion but he does reject the argument, for an interesting reason. The problem with it he points out is, basically, that even if we prevent someone from committing murder, we haven’t necessarily prevented them from attempting or at least intending to commit murder. And much of the deontic badness (if there is such a thing), Parfit suggests, would reside in the attempt or intent. If in the thought experiment above, someone will try to commit murder whether we stop them or we let them succeed so we can stop the accident instead, then both options involve the same amount of deontic badness (if any). So our choice doesn’t reveal how much deontic badness we think there is (or isn’t) in the murder.

That reasoning does depend on exactly how the thought experiment is fleshed out, but Parfit thinks “there would seldom be an easy way in which we could prevent the moral badness of some murder or attempted murder.” (OWM3, p. 357)

Let me propose another version of the thought experiment:

In this scenario, not only can we prevent the murder; we can prevent anyone from even forming the intention to murder. Whether we feel a strong preference for preventing the murder may tell us something about whether we believe there’s a significant element of deontic badness in it.

(Admittedly, there are still confounding factors. Maybe you think the badness of the murder is drastically reduced because it’s based on convincing false information. Or maybe you think merely having the sort of disposition that would commit murder under certain circumstances is almost as bad as actually trying to murder someone.)

Any-All and False Fame 🔗

Parfit believes in what he calls…

the Any-All Thesis: If anyone has a personal reason to want something to happen, everyone has a corresponding but perhaps much weaker impartial reason to want this thing to happen. (OWM3, p. 321)

This thesis implies:

When anyone has a reason to care about something,
everyone has some reason to care about this thing. (OWM3, p. 316)

Note that “personal reason” here does not include everything the person merely believes to be a reason. In Parfit’s view there’s still an objective fact of the matter about whether someone has a given reason (whether personal or impartial) or not; people can fail to recognize reasons that they actually have, or mistakenly believe they have reasons that they don’t. But not everyone has the same reasons as everyone else. Examples of personal reasons include “reasons to care for our own children”(OWM3, p. 321)—these can conflict with the reasons others have to care about their own children.

The objectivity of personal reasons is important in how Parfit responds to a couple interesting potential counterexamples to the Any-All Thesis:

Do X and/or Y have a personal reason to pursue a goal (posthumous fame) that no strangers would have a reason to want them to achieve? Since Parfit’s view still expects personal reasons to meet some standard of validity, it allows him to either deny the people in these cases even have personal reasons, or argue plausibly that others would indeed have at least some reason to want them to achieve their goal. As he notes, a hedonist would think X and Y are both mistaken to think posthumous fame would benefit them. On the other hand, “desire-fulfillment theories” of well-being could think both X and Y are benefited or that only X is; in any event, that benefit justifies an impartial reason just as it does a personal reason:

If this person’s posthumous fame would be good for him, there would also be agent-neutral value in this contribution to this person’s well-being. We all have impartial reasons to want things to go well for anyone else. (OWM3, p. 326)

Means vs side-effects 🔗

Is using someone as a means to an end morally worse than doing things that just happen to affect that person negatively? Parfit considers, but rejects, the following principle:

the Means and Side Effect Principle: Though we could sometimes justifiably save several people’s lives in some way whose foreseen side effect would be to kill one other person, it would be wrong to kill one person as a means of saving several other people. (OWM3, p. 374)

One of his arguments against it is a clever, if somewhat insensitive, thought experiment inspired by World War 2. I’ll summarize it:

The idea is that the 100,000 people in option 1 are being used as means, while the 200,000 people in option 2 are dying as side-effects, so the “Means and Side Effect Principle” would tell you to go with option 2 even though it kills twice as many people. And that seems wrong.

I think there’s a flaw in the setup. If we grant that the 200,000 people in option 2 aren’t being used as means, there’s a case that the 100,000 in option 1 aren’t either. Since we’re assuming that merely seeing how powerful the bomb is is enough to motivate surrender, the President could argue he’s not “using” the 100,000 since their deaths are incidental to his goal. He’s just trying to detonate the bomb in the place that leads to the fewest overall deaths while still stopping the war, and, in this bizarre scenario, that happens to be in the middle of a city where a bunch of people unfortunately live.

Such a claim would sound like weaselly casuistry, but is it any more arbitrary than claiming that the 200,000 aren’t being used as means in option 2? The whole distinction between ‘means’ and ‘side effects’ feels fuzzy to me, and I’m tempted to dismiss it as entirely subjective. (That would be a reason to reject the Means and Side Effect Principle directly without bothering with these thought experiments at all.) But maybe you can cash it out in terms of whether the decision-maker’s goal would still be achieved in the nearest counterfactual world where the same decision doesn’t cause the effects in question. If so, perhaps we could ‘fix’ this issue with the thought experiment by stipulating that…

Regardless, it just makes no sense to me to assign any moral significance to this distinction. I have the same basic problem with it that I have with deontology in general: it seems to elevate the concerns of the perpetrator above the concerns of the victim. The victim, I would think, cares a lot about the fact that they’re being killed and not much at all about whether their death is a means or a side effect. Perhaps Parfit is hinting at that in his concluding remarks on the thought experiment:

Consider what the US President might later say to some of the grieving relatives of the 200,000 civilians whom the second bomb would kill. He might say, ‘We Americans greatly regret our having killed all these innocent people. But if we had dropped the first bomb, though we would have killed 100,000 fewer people, we would have killed these people as a means of persuading your Government to surrender. It was morally necessary for us to drop this second bomb, though we knew that it would kill more people, in order to ensure that we would be killing people, not as a means of persuading your Government to surrender, but only as a side effect.’ (OWM3, p. 375-376)

Evolutionary debunking arguments 🔗

Sharon Street thinks the fact that our minds were shaped by evolution makes it unreasonable to believe in objective normative truths. Evolution is optimizing for the survival/propagation of genes, not the having of true beliefs, so it would seem like an extraordinary coincidence if the beliefs it shaped us to have just happen to correspond to some objective moral truth. Parfit in volume 2 responded to Street’s argument; she replies briefly in DARM, and he doubles down in reply in OWM3.

A couple other entries in DARM comment on this topic. Richard Yetter Chappell challenges, among other things, the idea “that we are … required to assign equal (or roughly equal) odds to the truth of each possible normative system.”(DARM, p. 156) Rather, “It’s only by employing our normative judgment that we can draw any conclusions about the likelihoods of various normative claims.”(DARM, p. 157)

Katarzyna de Lazari-Radek and Peter Singer also push back on Street’s argument, by appealing to

…the idea that the capacity to grasp moral truths is simply an application of our capacity to reason, which enables us to grasp a priori truths in general, including both the truths of mathematics, and moral truths. For if the ability to grasp moral truths is an aspect of our ability to reason, and to respond to reasons, it is easy to give an account of how it arose. (DARM, p. 289-290)

They argue we should be more suspicious of our moral intuitions when there are clear evolutionary reasons for us to have them (like biases toward our family or tribe), but less suspicious when there aren’t (as in the case of “universal altruism of the sort that is required by the axiom of rational benevolence”(DARM, p. 290)), as the latter intuitions are better explained as a side effect of a general ability to recognize truth. This leads to the following bold conclusion:

Since the claim that egoism is rational clashes with the Golden Rule and with the principle of universal benevolence, and the principle of egoism is subject to a debunking evolutionary explanation, while the impartial principles are not, we have grounds for supporting the impartial principles rather than the egoistic one. (DARM, p. 292)

I’m not sure what I think of all this.

Imperceptibly small harms 🔗

Is it wrong to do things which put really tiny burdens on other people—burdens so minor they can’t perceive them? Parfit thinks it can be, and that this is relevant for e.g. how our actions affect climate change. He talks about this issue in section 185 of OWM3, “Small Effects and Great Harms”.

Surely this convoluted way of doing things doesn’t absolve the torturers of responsibility, which suggests:

it can be wrong to impose pain on people, even if these acts make no one’s pain perceptibly worse. (OWM3, p. 425)

Parfit points out there are two different principles we might use to explain why the torturers’s act—turning a switch that causes an imperceptible increase in pain to 1000 people—is wrong. It could be wrong because of the total amount of harm it causes across those 1000 people, or it could be wrong because it’s “one of a set of acts”(OWM3, p. 425) (a switch-turn from each of the 1000 torturers) which “together impose great pain on one or more people.”(OWM3, p. 425) He calls these the “Single Act View” and “Many Acts View” respectively.

He says “most of us would find the [Many Acts View] more plausible” and that it’s sufficient for addressing “many actual cases” such as air pollution.(OWM3, p. 426) But he thinks the Single Act View is true too. To support this he gives a somewhat lengthy thought experiment (“Drops of Water” and “Case Two”, pp. 426-428); I’ll try to summarize the essential features in abstract form:

The idea is that in both versions, it seems like you should contribute to the pool, but the Many Acts View can only explain why in the first version; for the second version we need the Single Act View.

Convergences? 🔗

In his quest to prove that there is less deep disagreement in ethics/meta-ethics than there might initially appear to be, Parfit in volume 2 argued for some unusual readings of Hume and Nietzsche. Simon Blackburn and Andrew Huddleston, respectively, push back on those interpretations in DARM.

I’m not all that interested in what the historical figures really thought, at least in the context of evaluating Parfit’s convergence claims. Even if it was a misunderstanding to attribute certain views to them that conflict with Parfit’s, those views remain a problem for Parfit’s convergence claims if those views have seemed plausible to lots of informed and thoughtful people.

Trying to show convergence among living philosophers seems more valuable, since they’re still around to clarify their views and confirm whether an attempted reconciliation of their views with those of others represents them fairly or not. Parfit evidently thought that his back-and-forths with two contributors to DARM, Peter Railton and Allan Gibbard, did lead to convergence between himself and each of them. I was worried that Parfit’s intense desire to find such convergence would lead him to interpret their writing tendentiously and exaggerate the extent of agreement. But Railton’s enthusiastic follow-up comments in OWM3 give the impression that he, at least, thought they had indeed come much closer together. Gibbard’s comments in OWM3 leave more room for doubt on that point: Parfit follows them with a further commentary of his own, beginning, “As Gibbard’s remarks above unobviously imply, we have resolved our two deepest meta-ethical disagreements”(OWM3, p. 224, emphasis added).