Deontology and Preferability
Neglecting an important concept
Having previously lamented the (unnecessary) conceptual limitations of the utilitarian tradition, it’s time to turn to the rival deontological tradition. Whereas many utilitarians focus on value to the point of neglecting other significant normative concepts, my sense is that deontologists’ hyper-focus on permissibility can lead to similar problems. I wrote ‘Mark Significance with Attitudes’ because critics of utilitarianism so commonly assume that “a morally significant difference must make a difference to how we should act.” So I think almost everyone in ethics could benefit from paying more attention to fitting attitudes in general.1 But at least deontologists tend to be receptive to this point, once it’s brought to their attention. What’s often more difficult is to get them on board with the more specific concept of preferability (or fitting preference). Many seem to view the concept with suspicion.
This creates some difficulties for me, because my most innovative works-in-progress involve making the case for—and exploring the implications of—placing preferability at the very center of our moral theorizing. The most common negative feedback I’ve received, both in conversation (e.g., talk Q&As) and from anonymous journal referees, is that the respondent either doesn’t fully understand or else refuses to engage with my central concept.2 I hope this post will help to make clear both (i) what the concept is, and (ii) why it’s philosophically fruitful to engage with.
What is Preferability?
Irreducible Preferability
One complaint I’ve received is that the concept of preferability is “ambiguous” between two possible reductive understandings: ranking in terms of impartial value and ranking in terms of deontic status. I now take care to explicitly clarify that the concept is not analytically reducible to either—it is, rather, a substantive normative question what overall preferences are truly warranted:
I use ‘preferability’ in the fitting attitudes sense: W2 is preferable to W1 if and only if it is uniquely fitting to prefer W2 over W1, all things considered. This claim is not conceptually reducible either to claims about value or to claims about permissibility and impermissibility. It’s a conceptually wide open question what it is fitting to prefer, and how these fittingness facts relate to questions of permissibility and value.
Preferability ≠ Value
This is already entailed by the above, but is worth hammering home just because of how commonly people get confused about this: preferability orderings may diverge from impartial value orderings (at least for non-consequentialists).
This shouldn’t be surprising: If you think that other things besides impartial value (e.g. deontic constraints) truly matter, then you presumably think that moral agents ought to care about more than just impartial value, and thus sometimes should prefer a less-valuable outcome over a more-valuable one, on the basis of these further considerations. Deontologists are free to have, and to recommend, deontologically-flavored preferences. The basic concept of preferability is theory-neutral on its face, begging no questions.
Outcomes include actions
Another common mistake is expressed in the following sort of objection:
“Of course I prefer that five survive rather than that only one does. But as a deontologist, I think it matters how this outcome is brought about. It would be wrong to violate a deontic constraint, even to bring about an otherwise preferable outcome. So, for deontologists, there’s just no connection between preferability and obligation.”
Again, this misunderstands preferability. An “otherwise preferable” outcome may be rendered outright dispreferable by the very feature that one brackets in order to make the “otherwise” evaluation. Stop bracketing morally relevant considerations, and a connection between preferability and obligation may be restored.
My concept of preferability applies over act-inclusive possible worlds, not general outcomes (understood in abstraction from how they are brought about). Sure, everyone can generally prefer that more lives be saved rather than fewer. That’s not in dispute. The question is whether deontologists should specifically prefer that one person be killed in order to prevent five other killings. This is an interesting and conceptually open question!
Why Preferability Matters
It’s unavoidable
It would be absurd to deny that to there can be truths about preferability. We clearly should prefer that innocent people not suffer, all else equal. Compare a possible future in which a child gets struck by lightning with an alternative in which they don’t. Nobody has acted wrongly in either case. But we clearly ought to hope and prefer that the child not be struck by lightning (all else equal). This is a datum of common sense, and any theorist who denies it is in the grip of a theory.3
There’s undeniably more to morality than just the deontic assessment of actions. As moral agents, we should care about more than just right- and wrong-doing. A complete moral theory must have something to say about what broader preferences are fitting, virtuous, or morally called-for, and some of these preferences will concern events or states of affairs, not just actions, that affect people’s well-being. No sane and decent view can deny this.
Now, just because there are preferability facts, it doesn’t follow that any particular theorist has to talk about them. You could develop a theory that focuses exclusively on other (e.g. deontic) normative properties. But any such incomplete theory had better at least be coherently completable. If (as I argue in some of my work) certain verdicts about the deontic statuses of actions cannot be coherently combined with any plausible further claims about the relative preferability of various possible worlds, then those initial verdicts cannot all be true. So I don’t think that the in-principle relevance of preferability facts to ethical theory can be denied.
It’s helpful
This section is inevitably going to be more contestable, since it involves evaluating philosophy—a notoriously disputable undertaking. My current book project, Beyond Right and Wrong, will hopefully demonstrate by example the philosophical value of ethical theorizing in terms of preferability. For now, let me just highlight a few key themes:
(1) The question, how should bystanders feel about optimific rights violations? is an extremely deep and puzzling challenge for deontologists to grapple with. I think it may even be the single deepest challenge to deontology as such. But conceptual blinkers prevent many from even recognizing the challenge.4
(2) Foregrounding preferability over deontic status is especially important for charitably understanding consequentialist theories, as I explain in Bleeding-Heart Consequentialism. Many popular “objections” fall apart once translated into these terms.
(3) It’s a very helpful frame for fully grasping the importance of beneficence—or, more generally, prioritizing within the permissible. Many intuitively supererogatory actions are nonetheless extremely worth doing!
(4) It can also help bring to light why we shouldn’t always be overly concerned to avoid wrongdoing per se. For example, in deontic leveling-down cases, we should prefer acting (merely comparatively) wrongly over not having any good options at all. And even when an act is outright harmful (e.g. purchasing factory-farmed meat), it could be worth prioritizing more effective ways of helping over preventing oneself from repeating that harmful wrongdoing in future.
Conclusion
Many of these claims are reasonably contestable, and it’s fine if you disagree with some or all of them. My point is not (just) that thinking about preferability will help you to grasp important truths (though I do think that it helps), but more broadly, that it suggests an interesting way of thinking about ethics—one that can be recognized as philosophically fruitful and worth engaging with, even if you disagree on various matters of detail.
Perhaps virtue ethicists do better here?
I’ve found that 10 minutes of back-and-forth conversation can often win over a skeptic, but one doesn’t have that opportunity when dealing with journal referees!
I have actually had a referee openly object that my paper “presents it as obvious that we should “prefer” that children not get struck by lightning, even if we have no power to prevent it.” Yes, this is obvious! (My wife laughed out loud when I read her that one.)
My view is that principled deontologists should prefer a permissible alternative over an optimific wronging, even when this results in more subsequent moral violations. But it’s an interesting, contestable question, which warrants further attention and discussion. In particular, it’s important that further research explores the costs and benefits of each answer here. That’s difficult when so many referees refuse to tolerate such discussion, due to misunderstandings of the sort addressed above.
Dear Richard,
First of all: hi! I've been reading your (excellent) blog for many years. I also think we overlapped at Princeton way back in the day, and may even have met, but there is no reason why you should remember me.
Regarding this post:
1. I agree strongly with the central thesis, that preferability is an important and somewhat neglected concept in ethics -- neglected especially by 'deontologists,' particularly those of Kantian inspiration. As you say, much of the suspicion derives from the fear that talk of preferability is smuggling in a tendentious form of consequentialism. Maybe it's worth noting, though, that many deontologists are also suspicious of idea that the maximally thin notion of a preference used by many philosophers, decision theorists, etc., has any central role to play in our philosophical psychology. Not sure which suspicion is more fundamental. I have two thoughts here about ways of disarming some of the knee-jerk skepticism (some) deontologists feel towards preferability-talk. First, it might be better to talk about desirability, and, especially, about what it is fitting to *want.* It's less stilted and jargony (to my ear), and so more clearly has a basis in our commonsense ethical evaluations. Second, in my experience some deontologists can be warmed up to the relevant notion of preferability (or desirability) by being reminded that it is an open theoretical possibility that facts about which preferences over states of affairs would be fitting are at least sometimes explained by facts about other fitting attitudes. So, the fact that I should (/it would be fitting to) prefer to not cut up the one to save the five could be explained by the fact that it is fitting to respect the one as a person with a sacred-magical-dignity-aura (or some such).
2. Relatedly, I think it might strengthen your case here if you were a bit more ecumenical about the relation between preferability and preference (or desirability and desire). It's just overwhelmingly plausible that there is some true and interesting biconditional of the form: A is preferable to B iff it is G to prefer A to B, where G is some good normative or evaluative status. Maybe G is *fitting*-- I know that's your view (or your preferred bit of ideology). But it could be *virtuous* or *rational* or any number of other things (or, indeed, several at once), and the general point you are making will still stand. Or am I missing something? This isn't really meant as a criticism -- perfectly fine to use your preferred ideology on your blog! Just a thought about how to pitch the general point to the widest possible audience.
3. I also agree (I think) with the spirit behind the claim that "'how should bystanders feel about optimific rights violations?' is an extremely deep and puzzling challenge for deontologists to grapple with." But I'm sympathetic to some kind of agent-neutral deontology, so I think the question, as posed here, carries a false presupposition. That is, I think the bystander, with no special connections to any of the victims/potential victims, shouldn't prefer that we cut up the one to save the five (that it wouldn't be virtuous or fitting for him to so prefer, etc.). And I think something is impartially better iff it is impartially preferable. So I'm inclined to think that there are no optimific rights violations, at least on the reading of "rights violations" where such violations are necessarily subjectively impermissible. But the general point -- that thinking through how they should feel about, say, someone else cutting up the one to save the five is a productive and challenging line of inquiry for deontologists -- seems clearly right to me. I'm not trying to convince you of the truth of agent-neutral deontology, or the associated verdicts about preferability and permissibility. I realize that you have sophisticated views about these matters which I haven't engaged with at all.
Anyway, thanks for your interesting post.
Hi Richard! Mark here. I always enjoy your work on Ethical Theory and EA/Longtermism, so thanks for the thought-provoking essay! I'm mostly in agreement, and I admire your intricate, multi-category, non-simplistic approach to Ethical Theory as compared to what is perhaps the vast majority of folks in Ethics who flail in oversimplification due to their limited categories or notions or yardsticks. But some responses, some to play devil's advocate.
(A side comment & question first: I love the fact that you always link to your other Essays. It makes things nice and tidy and makes me fantasise of a hierarchy or structured Web of "Chappell's Ethical System". So I'm curious: do you think there is ONE paper (or one small set of papers) which form the "foundations" of your Ethical thought, from which all your other papers fall out more or less as logical corollaries of them? And—if this isn't too bold!—could you create a mindmap of how all your papers fit together, logically speaking? I think fellow admirers of your work would greatly appreciate seeing how this whole labyrinth fits together! I sincerely think you have an admirable intricate take of Ethics and "how it hangs together" (to borrow that Sellarsian phrase) and I think it would be a shame if other philosophers weren't given the opportunity of being able to take it all in (to grasp the forest over the trees) on 1 page or 1 jpg. I think more philosophers would be more inclined to follow your intricate approach if they could appreciate it easily/quickly first before delving into the details.)
1) What is the significance of talking/thinking in terms of preferability ("X is preferable", "X is preferable over Y") vs talking/thinking in terms of normative reasons? ("There is normative reason to prefer X" or "There is normative reason for Bob to prefer X" or "There is more normative reason for X to exist")
2) Given that, at the end of the day, what we want our Ethical Theory to do, at a minimum, is to tell us how we ought *to live* (i.e. to answer the Socratic question) or what we have most reason *to do*, couldn't we actually dispense (in our foundations, not in practice) with preferability and value and treat them as mere heuristics or guides to talking about reasons for action? After all, it seems that people's preferences or emotions or desires are simply not within their direct, voluntary control. Only their actions. And given the ought implies can principle (or a suitable generalisation thereof), how could normativity/reasons attach to something outside the sphere of choice? (to borrow a phrase from Epictetus)
And personally, I would like to minimise the number of posits or ontological commitments I make, simplicity/parsimony or Occam's Razor, I think, being the only principled way of not having to consider a whooole bunch of (perhaps infinite) crazy, wild philosophical theses. So it is with metaethics. I want to make the fewest number (or smallest variety) of posits about what fundamental normative facts/properties/truths there are, and I think normative reasons for action might be enough as far as Ethical Theory goes. Why countenance *irreducible* normative facts about preferability or value or permissibility when I can get all the plausible verdicts from just reasons for action? Sure, perhaps I can't describe, simpliciter, the normative difference between a possible world in which a child is lightning-struck and an otherwise identical world in which he isn't. But (1) This isn't even action-guiding and so not really normative or not helping to answer the Socratic question (2) I can still approximate towards that desirable/plausible verdict: I can still say "IF it were within my power to cause a child to be lightning-struck, I have normative reason to not do so (or extremely less reason to do so than to not do so)."
3) Suppose it were given that it is impossible for Bob to prefer X. (E.g. perhaps because there simply does not exist any chain of cause-and-effect (or continuous fourdimensional spacetime worm) leading from the Big Bang to the [nonexistent] event (the mental state) of Bob's preferring of X—plus the Kripkean fact that Bob has his origins essentially/necessarily: from the Big Bang).
Would Bob still have reason to prefer X? (Seems not, given the Ought Implies Can principle, or a suitable generalisation thereof) But then again, our default/commonsense intuitions beckon us to answer: Yes (e.g. if we take X="The child not to be lightning-struck")
It seems to me, as someone with sympathies for Necessitarianism, that I would answer Yes only when thinking of an utterly abstract (and so, nonexistent) agent. But when thinking about that very person, Bob himself, I must consider what is possible and what is impossible for his mental states to do. But when we do Ross-style comparisons (or thought experiments) of pairs of possible worlds with only one independent variable is modified, one is tantalised to think in the hypothetical abstract rather than messy, real, concrete Reality where sometimes it's just the case that a person couldn't have preferred X no matter what. His mental/neural states were necessitated to be thus-and-so with ironclad necessity by cause-and-effect or the laws of Nature or a continuous fourdimensional spacetime worm that had always already been in such-and-such condition.
Plus, it seems to me that the rational (or ideally rational) person would just accept Reality as It [eternally or fourdimensionally] is (*), rather than prefer a nonexistent, hypothetical world over It. A "Reality" which is just like Reality but with 1 variable magically changed is just as fictitious as Heaven/Nirvana/Paradise. This (*) statement is of course apt for misunderstanding! It is by no means equivalent to saying "the rational person would just do nothing, or, think the present time-slice is superior in value to all other time-slices".