As seen on some really big websites I am scared and confused to have been featured on!

Virtue Ethics: Not Practically Useful, Either

I’ve been trying to understand some of the responses to my review of After Virtue. Tell me whether this is about right:

The problem of “doing the right thing” consists of two subproblems.

First, knowing what the right thing is. Do we legalize or ban abortion? Do we press the switch in the trolley problem or not?

Second, behaving correctly in situations where we do know what the right thing is. For example, going to visit a friend in the hospital even though the hospital is far away. Not cheating on your taxes even though you could use the money. Working hard even though no one is checking up on you.

The proposal was that virtue ethics doesn’t claim to be a solution to the first problem, but is a uniquely excellent solution to the second, and in fact the solution people actually use. Am I understanding this correctly?

Because if that’s true, I still disagree.

Virtue Ethics Is What People Do

I am not really good at thinking in terms of good or bad people. I can do so in very edge cases, like Kim Jong-il or St. Holden. But I tend to process all the people I know and have remotely okay interactions with as “good people”, leading to conversations like:

Me: Oh, cool, Bob is coming over soon! Bob is great!
Friend: But didn’t Bob [do X, Y, and Z]?
Me: Well, yes, but other than that he’s great.
Friend: And didn’t he [do A, B, and C]?
Me: You can’t just keep taking all these things Bob does out of context!
Friend: Okay, what has Bob done that was good?
Me: He…well…he…you know! Bob! He’s great!

When I deviate from this, it almost always tends to be in terms of actions, not qualities. Like “Bob is great, except that he posts really annoying things on Facebook all the time”. Or “Bob is great, except that he has some really horrible politics.”

When I think about my own morality, it’s almost never in terms of whether I am or am not a virtuous person (I have at least one subagent that’s always convinced I’m a virtuous person, and at least one subagent that’s always convinced I’m terrible and deserve to die, and doing good things paradoxically strengthens the latter subagent for some reason).

It’s usually in terms of – I guess it feels like a missile must feel locking on to a target. If my mind is sufficiently calm and predisposed to goodness, I think “Wait a second, that’s the morally correct thing to do”, lock on to one option, and then feel really good about it – a serene feeling.

I can’t always do this. If I’m sufficiently angry, part of me thinks “I bet I could lock on to what’s good and do it…but that would probably involve turning the other cheek, or compromising, AND THEN THESE JERKS WOULD GET AWAY WITH IT.” And then my mind makes up all sorts of game theoretic justifications for why it’s more important to punish defectors than to do what feels like the right thing at this precise moment. Or if I’m sufficiently exhausted, I think “If I started worrying about what’s good now, then I’d probably have to do it, and that would be really arduous.”

On the one hand, this is no doubt a very idiosyncratic report; I don’t expect my experience of morality to be similar to anyone else’s. On the other hand, this is an idiosyncratic report and I don’t expect my experience of morality to be similar to anyone else’s. So if you say “Virtue ethics is the way people naturally think about morality!”, that’s either a typical mind fallacy or you’re going to have to do a much better job explaining virtue ethics.

My experience of morality is contray to traditional virtue ethics in almost every way. It doesn’t feel like it depends on my social roles. It doesn’t strike me as divisible – that is, it feels like solid goodness and words like “continence” or “prudence” don’t do anything to me. It doesn’t strike me as the same feeling that occurs when I consider important but non-ethical questions like procrastination. It doesn’t strike me as performed in a community or according to a narrative. It’s just not virtue ethics.

The Practice of Making People More Moral

The other claim is that virtue ethics is the science of making people better – a process for helping people refine their existing moral intuitions and overcome temptation more effectively.

If that’s true, it’s a science much like medieval medicine was a science – totally untested and not especially likely to bear any resemblance to reality. If people are actually looking for ways to become more moral, I bet an hour’s search would find about thirty of them that are more likely to work than adopting virtue ethics.

These could be broadly divided into beliefs and practices. In terms of beliefs, I think the most useful would be a belief in the Devil, moral realism, humility, and various kinds of magical thinking.

In terms of practices, there would be willpower training (pretty much anything that requires willpower counts as willpower training, but let’s say exercise as the stereotypical example), rationality training, keeping a journal, talking about morality, making friends, joining a group with some sort of interest in morality, cutting yourself off from bad influences, making yourself happier (happy people are more moral), learning relaxation/stress-busting techniques, and reading fiction.

All of these things would make people more moral in different directions and in different ways. For example, I bet reading works of fiction about poor people in the Third World would make you more likely to donate to charity, and contemplating virtue ethics and the just polis would make you more likely to make you get involved in local politics. Which of these you recommend is very closely linked to whether you think giving to charity to the Third World is more or less important than getting involved in local politics (hint: there is only one answer to this question which is not really stupid).

In terms of the best all-around practice for increasing morality I would nominate meditation, especially lovingkindness meditation. David Chapman, who knows ten zillion times more about Buddhism and meditation than I do, suggests metta bhavana, tonglen, and chöd. Even very generic meditation ticks several of the boxes above – relaxation, willpower training, and happiness – but these are said to (and have some evidence of) specifically increasing your ability to love and care about other people.

This seems like probably the best thing you can do for morality short of ground it objectively. If people love and care about others, they end up automatically ticking the two boxes I claim are required to end up more-or-less utilitarian – grounding morality in the world and caring about other people. If you’ve got that, it kind of lowers the degree to which morality even needs to be grounded objectively; it would still be nice, but we can trust people to do the right thing even if it isn’t.

Virtue ethics doesn’t satisfy either of these criteria, and in fact, we find that throughout history a lot of really terrible people have been very good virtue ethicists (the Spartans come to mind). So although many of the commenters here want to virtue ethics from its failure to ground morality by saying it removes the need to ground morality, I think virtue ethics can’t even do that right and there are a lot of things that are much better.

Edit: Something I said in the comments that might clarify my position. I think even this is giving virtue ethics too much credit, since it’s not just “use our inborn moral sense” but a host of claims about making lists of virtues and studying teleology – but on the principle of steelmanning an opponent’s argument:

Imagine that instead of virtue ethics we’re talking about grammar. In most cases, we have a natural grammar sense – that is, the real reason I don’t say “Me is Scott” is because it just sounds wrong.

In most cases this is good enough. In some cases it isn’t – for example, sometimes we have to teach grammar to foreigners who lack this intuitive grammar sense. Or sometimes there are edge cases where we’re really not sure what word to use. Or we want to program a computer to write with proper grammar. Or we want to set editorial policy for a newspaper. Or sometimes we’re just genuinely curious how grammar works.

There is a point to having a science of grammar where smart people say “Oh, it looks like the predicate nominative form is used in this way.”

And inventing a “virtue grammar”, where people say “But you’re ignoring all normal grammar usage in favor of a few silly edge cases! Real people don’t talk about predicate nominatives! Just use your natural grammar sense!” is a total waste of everyone’s time. Yes, natural grammar sense usually works well, but shouting “Hey, natural grammar sense often works well!” contributes nothing to the field and is just distracting people from actually figuring out how grammar works.

In pretty much all fields except ethics, everyone has agreed that the proper thing to do is to be happy when our natural senses are good enough, but also create a formal study of the field in order to go beyond what our natural senses can tell us. I don’t understand why we can’t also do this for ethics.

This entry was posted in Uncategorized. Bookmark the permalink.

33 Responses to Virtue Ethics: Not Practically Useful, Either

  1. Verchoo says:

    Looking for methodologies, rather than goals, might be on the wrong side of neurodiversity. There might be justice training that makes most people act more justly, or it might be that everyone should guide themselves by the same star but find maps that are locally useful.

    Do you think virtue ethics could be a superset of utilitarianism? What would that relationship look like?

    Report comment

  2. im says:

    -I add one more boolean to utilitarianism: prioritization.

    Report comment

  3. Joe says:

    It sounds like you’re just renaming “virtue ethics” to “basically being a good guy ethics”. It’s like you don’t like the philosophical jargon involved in “virtue ethics” ala MacIntyre so you rephrase it in more superficial language with a little eastern meditation to maintain geek chic status. I guess I’m just not sure how your morality is much different then virtue ethics. You still use language like right, wrong and goodness.

    Report comment

    • Berry says:

      Virtue Ethics does not have a monopoly on Moral language.

      Report comment

    • Scott Alexander says:

      ….what?

      I’m saying I *don’t* judge people’s morality in a way that relates to “whether they’re a good guy” or not and that my definition of “goodness” is very different from that of virtue ethicists’.

      Report comment

      • Joe says:

        Well I would consider your reluctence to think ill of other people as a virtue, namely charity. Also I’m not sure why you would not think of your morality as being performed in a community, especially if you’re polygamous. If you don’t believe you’re trying to be the best person you can be I think you’re selling yourself short. I must have missed your definition of “goodness” in the OP. I struggle with the two subagents you discribed, so I sympathize. I can’t imagine my life as not part of a narrative so I’ll just trust that you’re being honest about that. My beef with “Virtue Ethics” is that it is too generic. Everyone feel obligated to be a better person so why not spend time debating more specific moral philosophies?

        Report comment

  4. jimrandomh says:

    I feel like, in saying “virtue ethics is bad”, based on the specific virtue ethics you’ve encountered, you’re doing something akin to saying “consequentialism is bad” after reading consequentialist writing by Clippy. Consequentialism, deontology, and virtue ethics are not moral frameworks; they are *categories* of moral frameworks, or rather, categories that individual fragments of moral knowledge fall into, according to whether they do their good/bad classification on events, on actions, or on people.

    In order to use virtue-ethical moral knowledge to pick actions, you have to translate and relate it into deontological form, with a possible detour through consequentialism. If you try to be a Pure Virtue Ethicist, discarding all your consequentialist and deontological knowledge and building it up from a virtue ethical foundation, you’re going to have a bad time.

    But as I see it, that’s really not what virtue ethics is for. Virtue ethics is almost entirely a third-person morality: it is for evaluating the goodness of others, to decide whether they should be empowered and allied with. It is the native form for evaluating people, in the same way that consequentialism is the native form for evaluating events and deontology is the native form for evaluating actions. And if you want to decide who to empower and ally with as a consequentialist or a deontologist, you are going to have to translate your moral knowledge into virtue-ethical form first.

    My position is that consequentialism, deontology, and virtue ethics are *all* valid, and any complete moral framework must gracefully handle fragments of moral knowledge in all three forms. I think a surprisingly large amount of our moral knowledge comes from hero/villain classifications given to us in fiction, spread across weak-evidence relations like “the goals of good people are good”.

    (Mainstream status: Probably not, but I only study philosophy, not history of philosophy as published by dead fools, so I really have no idea.)

    Report comment

    • anodognosic says:

      I’ve had a similar thought for a while: this kind of syncretism is the only way to really make sense of metamorality. Sometimes it feels like a copout, because it’s bound to create conflicts, if not outright contradictions, that I don’t have the first clue how to solve. And still, in its imperfection, it still strikes me as superior than any single class of moral framework.

      Report comment

    • Paul Torek says:

      Absolutely, on the three forms. However, in fairness to Scott, “virtue ethics” is more often used in philosophy as the name of a school of thought rather than a set of ethical concerns. Although, sometimes the latter. Ah, the clarity of jargon!

      Report comment

    • John Bray says:

      I’ve always seen virtue ethics as more of a first person guidance on a fulfilling life, leaving questions of goodness as something to be gauged by the individual as they utilise their capacities. For me virtue has always been about individual human fulfilment and the broader notion of how good or bad a person is will be decided by whether their actions adhere to a cultural norm. I’m a relativist though so I’m uncomfortable using good/bad to describe people and I think some actions are seemingly universally reprehensible due to the fact that all human beings share physiological and possibly some basic social traits (though I’m really not sure about this one to be honest). What we share, in being human beings provides an ethical backdrop for most cultures to condone or condemn certain behaviours (e.g. child abuse, murder etc.). Virtue ethics isn’t connected to this level of ethics, for me it can only be used as a personal guide to realising capacities and finding fulfilment through that.

      Report comment

  5. Benquo says:

    Classical virtue ethics is not trying to do the same thing as modern moral philosophy. Aristotelian virtue ethics is not primarily about how to give others exactly their due, it’s about how to live an AWESOME life.

    Report comment

  6. Vladimir says:

    I am not really good at thinking in terms of good or bad people.

    Come on. If you’re walking in a dark alley in the middle of the night, I’m sure there are people whom you’d be glad to see walking around because it would make you feel safer, and also people whom you’d be terrified to see walking towards you. Similarly, there are people whom you’d readily trust with your life, and people whom you wouldn’t trust with a ten dollar bill. There are also people who you know are your loyal friends and allies, and also people who would stab you in the back the moment it seemed like a profitable option. And so on.

    Every social interaction you ever have is based on an implicit judgement of virtue and character — and if most of them don’t involve recurring re-evaluations of whether someone is in the “good” or “bad” people category, this is only because these judgements can usually be made reliably and early, and people’s character tends to remain constant. (And also because you probably do a good enough job of insulating yourself from really bad people, aided of course by the institutions within which you live and work.)

    You seem to consider the entire normal system of human morality as trivial and uninteresting, and focus on extreme and controversial cases as the proper subject matter of moral philosophy. It’s true that natural virtue ethics can’t provide a conclusive answer to many of these problems, whereas utilitarian and other abstract theories purport to be able to. But if this motivates you to turn to the later, consider the following points:

    1. If this abstract theory provides answers for the extreme and controversial cases, then it should provide answers for everyday common cases as well. But here we see that these abstract theories are of little use, often providing no useful answer or plainly absurd answers, and requiring tortured rationalizations and special pleading just to get them to clear the bar of ordinary common sense. (Not to mention what happens, as I pointed out in my previous comments here, when the personal interest of an abstract moralist hinges on a conflict between his abstract moral theory and common-sense virtue ethics.)

    2. In many of these problems, we simply must admit that absent appeals to some religious or other supernatural authority, there simply is no “right” answer, and the problem really is about a clash of fundamental values and preferences. This can end up only in a peaceful and thorough separation where people admit their differences and let good fences make good neighbors, or in a conflict — whether political or violent — where one side eventually wins and gets to coerce the other. Abstract moral theories are nothing but ideological weapons in conflicts of this sort, and they — especially utilitarianism — can always be made to support any ideological position whatever, assuming only that its advocates are skilled enough in lawyering. In the light of this, it’s absurd to take these abstract theories and proclaim them as authoritative over the entire human life, including that major part of it where common-sense virtue ethics works smoothly and by consensus.

    Report comment

    • Oligopsony says:

      2. In many of these problems, we simply must admit that absent appeals to some religious or other supernatural authority, there simply is no “right” answer, and the problem really is about a clash of fundamental values and preferences. This can end up only in a peaceful and thorough separation where people admit their differences and let good fences make good neighbors, or in a conflict — whether political or violent — where one side eventually wins and gets to coerce the other. Abstract moral theories are nothing but ideological weapons in conflicts of this sort, and they — especially utilitarianism — can always be made to support any ideological position whatever, assuming only that its advocates are skilled enough in lawyering. In the light of this, it’s absurd to take these abstract theories and proclaim them as authoritative over the entire human life, including that major part of it where common-sense virtue ethics works smoothly and by consensus.

      Surely you can’t be claiming that “common-sense virtue ethics” isn’t itself an example of this.

      Report comment

    • Scott Alexander says:

      “Come on. If you’re walking in a dark alley in the middle of the night, I’m sure there are people whom you’d be glad to see walking around because it would make you feel safer, and also people whom you’d be terrified to see walking towards you. Similarly, there are people whom you’d readily trust with your life, and people whom you wouldn’t trust with a ten dollar bill. There are also people who you know are your loyal friends and allies, and also people who would stab you in the back the moment it seemed like a profitable option. And so on.”

      Yes. And I am just saying I think mostly in terms like “Bob would probably steal my ten dollar bill” and not as “Bob is a bad person” let alone “Bob lacks the virtue of sincerity”.

      “Every social interaction you ever have is based on an implicit judgement of virtue and character — and if most of them don’t involve recurring re-evaluations of whether someone is in the “good” or “bad” people category, this is only because these judgements can usually be made reliably and early, and people’s character tends to remain constant. (And also because you probably do a good enough job of insulating yourself from really bad people, aided of course by the institutions within which you live and work.) “

      Every social interaction *could* be framed in terms of deontology – “is this interaction going to violate the rules of responsible friendly behavior” (Alicorn definitely thinks this way) – or in terms of consequentialism – “is this interaction going to leave me better or worse than before?” – or in terms of virtue ethics – “is this a good person”. I’m saying that I don’t naturally think in the virtue ethics terms any more than I naturally think in the deontological terms. The fact that in theory someone could associate these interactions with virtue ethics means nothing.

      “You seem to consider the entire normal system of human morality as trivial and uninteresting, and focus on extreme and controversial cases as the proper subject matter of moral philosophy.”

      Sort of. Imagine that instead of virtue ethics we’re talking about grammar. In most cases, we have a natural grammar sense – that is, the real reason I don’t say “Me are Scott” is because it just sounds wrong.

      In most cases this is good enough. In some cases it isn’t – for example, sometimes we have to teach grammar to foreigners who lack this intuitive grammar sense. In other cases our grammar sense misfires, or is uncertain. Or sometimes we’re just genuinely curious how grammar works.

      There is a point to having a science of grammar where smart people say “Oh, it looks like the predicate nominative form is used in this way.”

      And inventing a “virtue grammar”, where people say “But you’re ignoring all normal grammar usage in favor of a few silly edge cases! Real people don’t talk about predicate nominatives! Just use your natural grammar sense!” is a total waste of everyone’s time. Yes, natural grammar sense usually works well, but shouting “Hey, natural grammar sense often works well!” contributes nothing to the field and is just distracting people from actually figuring out how grammar works.

      In pretty much all field except ethics, everyone has agreed that the proper thing to do is to be happy when our natural senses are good enough, but also create a formal study of the field in order to go beyond what our natural senses can tell us. I don’t understand why we can’t also do this for ethics.

      Report comment

      • David Gerard says:

        To continue that analogy, total utilitarianism looks very like adopting a theory because it has numbers in, or could have numbers in if you eventually develop it enough, because numbers are automatically cooler – and treating all the ridiculous results and singularities that result as fantastically important results rather than evidence that the theory is maybe not so good.

        Report comment

      • Vladimir says:

        In pretty much all field except ethics, everyone has agreed that the proper thing to do is to be happy when our natural senses are good enough, but also create a formal study of the field in order to go beyond what our natural senses can tell us. I don’t understand why we can’t also do this for ethics.

        I’m not saying we can’t do it in principle, but that:

        (1) Existing attempts to do so have produced horribly bad theories, which contradict our natural senses and produce absurd conclusions. (And not in the same way as counterintuitive scientific findings, where our natural senses can be shown to be misleading. Rather, their implications are correctly perceived by our natural senses as absurd — and further scrutiny only makes them look worse.) The rational course of action in this situation is to discard these bad theories and start from scratch.

        (2) If we actually do start from scratch in our study of ethics and make sure to stick to the reality of what human beings are, not to metaphysical pies in the sky and sophistries useful only for signaling and lawyering, we will end up — or at least have to start with — something resembling virtue ethics.

        To use your language analogy, utilitarian and deontological theories are as if some linguists were to devise a formal grammar based on their favored abstract philosophical principles instead of observing how people actually speak — and ended up with a grammar that produces meaningless random noises. Imagine them then insisting that their formal grammar is in fact the only correct way to speak English, and somehow winning high intellectual prestige and influence with this nonsense. They would probably often show off their sophistication by producing these noises instead of normal speech — but on those occasions when it would be in their interest to be understood, they would still speak normal English.

        Report comment

        • Scott Alexander says:

          Aside from the things I addressed in my new post, we seem to disagree a lot on how bad utilitarian and deontological ethics work. As far as I can tell they work about as well as Newtonian physics – they get the right result in the overwhelming majority of cases, but break down in certain weird edge cases where in fact they might still be salvageable.

          The libertarian version of deontology – don’t initiate force or fraud, otherwise do what you want – is enough to be a relatively decent person. I happen to like government and think some force is useful, but it doesn’t seem like a stunning obvious failure.

          Likewise, rule utilitarianism correctly generates ideas like don’t steal, don’t kill, form just governments, give to charity, et cetera. You can argue that this might be biased and people might be sneaking their own intuitions in, but the argument you’re actually making – that no one can get it to work – seems empirically wrong.

          Report comment

  7. Oligopsony says:

    I don’t know very well what it is to be a good person, but I think I do have something of a better idea of what it is to be a good boyfriend, student, communist, son, friend. It seems to me that Satre was right about “personhood,” but these other, more relational, identities also seem richer in content. I’m not all that great at regulating my behavior, but inasmuch as I do it is through such means.

    Of course this doesn’t mean you work this way! But we’re all about charitable reading here: “everybody practically thinks with virtue ethics” can be steelmanned into “for at least some of us, virtue ethics is how we actually think.” Of course, if this is hardwired (or might as well be), then that’s not an argument for others adopting virtue ethics but us orthopraxic virtue ethicists to come up with better virtue ethics. In this technologies like meditation seem like they are indeed useful!

    Report comment

  8. ozymandias42 says:

    I feel like some of this might be a Brains Are Different From Each Other issue. Because like… I’m a consequentialist, but my morality-as-implemented is basically deontologist, because I like rules and I feel satisfied when they are followed properly. So to me “give ten percent of your income to charity” feels like it comes from the same place and creates the same satisfaction as “turn in your homework on time” and “the word ‘earth’ is generally lowercase but capitalized when used to refer to the planet.” The major difference is that I feel like a bad person when I break moral rules but not when I miscapitalize ‘earth.’

    But otoh I presume someone else obtains a similar satisfaction from virtue ethics and I don’t really care why you do good things as long as you do them. So. (I also wonder if other ethical systems are psychologically optimal for other people. Ethic of care?)

    Report comment

  9. Salem says:

    There is a point to having a science of grammar where smart people say “Oh, it looks like the predicate nominative form is used in this way.”

    And inventing a “virtue grammar”, where people say “But you’re ignoring all normal grammar usage in favor of a few silly edge cases! Real people don’t talk about predicate nominatives! Just use your natural grammar sense!” is a total waste of everyone’s time.

    Actually, I think your grammar example shows precisely the strength of virtue ethics. Real people don’t talk about predicate nominatives, and much of “grammar” is artifical and contradictory rules retro-fitted onto human communication in a confused muddle of descriptivism and prescriptivism – sound familiar?

    The problem is, you lack charity towards the “virtue grammarians.” They don’t just say “use your natural grammar sense.” They add “And if you don’t have a natural sense, then try and imitate someone who you think speaks well, until you get a natural sense of your own.” And, if you have ever learned a foreign language, that’s exactly what works in practice, whereas you will never learn how to speak properly just by studying semi-arbitrary rules in a classroom. And forget about learning when it’s acceptable to break the rules of grammar! That’s why modern grammarians don’t try and lay down rules for language like in days of old, but are themselves “virtue grammarians.”

    The correspondance to ethics should be obvious.

    Report comment

    • Alex says:

      The same thought occurred to me, but yet we have no problem with linguists who study syntax right?

      Report comment

    • Scott Alexander says:

      Real people don’t usually talk about predicate nominatives, but there are certainly times when people need to crack open a grammar book, and as soon as you have reached that time, saying “Use your natural grammar sense!” no longer helps. So writing a grammar book that just says “Use your natural grammar sense” is a waste of time.

      …especially if you Japanese English of student are and you of grammar sense natural this like works.

      Report comment

  10. sixes_and_sevens says:

    I seem to have a problem with virtue ethics. I broadly subscribe to a utilitarian framework. For the overwhelming majority of moral problems I critically assess, this is what I use to do the assessing. Many of these problems are contrived hypotheticals, some are around politically-divisive issues or esoteric real-world scenarios. A few are around the actions of the people I interact with socially.

    A very, very small number (maybe one a month) are based around my own personal moral dilemmas, as they pertain to my actions and how that will impact other people. I have a moderate amount of training and education in related fields and am generally a pretty instrumentally effective person, but as soon as I have a personally-motivated moral dilemma, all of that flies out the window and I start worrying about what kind of person my decisions make me.

    It’s almost like my mind has some sort of factory for manufacturing possible ignoble motives I might have for any given course of action. “What if I’m doing this because of [x]? What kind of person does that make me?”

    I would love to not do this, because it’s patently ridiculous, but I have no idea how to go about abandoning it.

    Report comment

  11. Gilbert says:

    Disclaimer: Everything I know about Virtue Ethics I read at Unequally Yoked. Your last post was the straw that finally made me order After Virtue, but it’s not here yet and I’m talking through my incompetent backside.

    Anyway, here are two of Leah’s posts on Virtue Ethics, both of them pre-conversion-but-already-on-the-trajectory.

    My understanding of her main point is that a person’s inner motivational structure is itself a moral issue even if it doesn’t make much difference in externally observable actions. As the joke goes, a Kantian should hate the moral law so that there is more virtue in following. And the reason it’s so funny is that following the moral rules over one’s inner grumbling is good in a fake-it-till-you-make-it way, but not the end result we want to arrive at. Of course this pitch is best for deontolgists and like me and an older version of Leah, but optimizing some consequences is also just a rule, so it isn’t totally irrelevant there either.

    Report comment

  12. MugaSofer says:

    You know, I could swear I remember reading some of your old blog posts that mentioned (in passing) instinctively making decisions based on whether they were what a virtuous person would do, and having to force yourself to actually evaluate the expected utility. Maybe I’m misremebering, but perhaps you’re focusing too much on the jargon of virtue ethics? Or the Fundamental Attribution Error?

    Report comment

    • Scott Alexander says:

      Maybe? I don’t remember this, but I frequently forget things I’ve said.

      It wouldn’t surprise me if I occasionally thought virtueethically – or sometimes deontologically, or sometimes consequentially.

      Report comment

  13. Federico says:

    Traditional ethics & meta-ethics is like mainstream music theory: more hinderance than help. Virtue ethicists argue along such lines as, “Please don’t learn music theory from books, because Rameau’s harmonic approach is incorrect. Why not instead transcribe a Bach chorale or join an orchestra?” I see the good sense herein.

    Civilisation depends on a system of social norms, in addition to and in mutual causation with laws. Attempts to transform popular ethics, radically and discontinuously, are therefore dangerous and often malicious. Should the elite subscribe to false, highfalutin ethical or meta-ethical theories—and indeed they are mostly false—rather than common sense, they might sanction such normative disruption.

    However, Westergaard’s Introduction to Tonal Theory seems a legitimate form of musical pedagogy. Music must employ certain rules and constructs—the Universe is lawful. Might ethics & meta-ethics be a similar case? In the other thread I proposed a spine for a rebooted ethics & meta-ethics (“fundaments”). It raises difficult questions, but I think these are the right questions and distinctions.

    Now statistical thermodynamics was originally derived in reverse: Gibbs assumed a property of thermodynamic microstates that happened to produce empirically “correct” answers. Only decades later did E.T. Jaynes lend weight to this assumption; Gibbs had implicitly used the Maximum Entropy principle. Li & Vitanyi in turn justified MaxEnt, by proving that it approximates Solomonoff Induction.

    I thus disagree with (what appears to be) Vladimir’s characterisation of theories that reconcile abstract theory with real human behaviour as “tortured rationalization”. To work backwards is not always to rationalise, especially when one employs the best available evidence.

    My own ethical behaviour is indeed far from utilitarian. Like us all I am influenced by emotion, imitation, prudence, natural proclivities and enculturation. Utilitarianism is the typical solution of the most analytic and executive part of human brains, when these influences converge upon their belief that the brain obeys definite physical laws.

    Virtue ethicists correctly point out the obvious: no gestalt human has ever behaved like a utilitarian. Secondly, naïve utilitarianism fails to consider TDT. Thirdly, utilitarians rarely shut up and multiply (quite as much as they should) when they treat theoretical trolley problems. The hedonic calculus is heavily future-loaded, due to a possible singularity or other massive hedonic event, so the welfare of fat men, other trolley-victims and their families is equally negligible, i.e. almost entirely of instrumental and not terminal value.

    Utilitarianism is not hereby refuted. The solution, working backwards, is that no human brain is adequately treated as a single, stable decision-agent. This insight, though simple, is not part of mainstream ethics & meta-ethics; arguably it is found in Buddhism. This may be because reductionism-of-humans is a recent trend, because the idea is discomfiting, or because it is difficult to generate alternative criteria for selfhood. I proffer that the agent explicitly or implicitly determines its identity circularly, according to its own utility; Yudkowsky has taught us that instrumental and epistemic rationality are inextricable.

    This theory of fundaments, in addition to mass-consumption ethics, is useful for several reasons.

    1. Many of us assign non-negligible probability to a singularity or similar event in the near future. These extreme scenarios are outside the operating window of virtue ethics.

    2. Sophists sometimes attempt to manipulate one’s notions of virtue. An explicit theory of ethics & meta-ethics upon which to fall back, even if rarely used, forms an intellectual defense.

    3. Moral realism is deeply intuitive to humans. It probably evolved to solve Parfit’s Hitchhiker problems; TDT motivates quasi-deontological behaviour in such cases. We no longer live in tribes, so now moral realism encourages people to hold highly specific beliefs about how massive and arbitrary groups of humans ought to behave—we all should morally, if not prudentially, do X, Y and Z. This is incompatible with necessary decentralised decision-making. Richard Joyce’s moral error theory decreased my susceptibility to such ideas; but moral anti-realism is too abstruse to be part of virtue ethics.

    4. Emotion is the mind-killer, even for moral anti-realists. Emotion lends credible commitment to moral claims—deontology is rationalised emotion—and without the rationalisation it still promotes rival sub-agents. Those cognisant of selfhood bias might therefore find methods to dampen upstart sub-agents, such as meditation and lifestyle choices. Only the Buddhist tradition, to which few Westerners are naturally inclined or exposed, stresses such virtues.

    The virtue ethicist’s critique of fundaments blends into a critique of utilitarianism.

    I am skeptical of preference utilitarianism—maximisation of “collective utility”. Such a goal is not strictly impossible. One might simulate an arena, into which is thrust a statistically representative subset of Platonian decision-agents. One runs the simulation for a certain time, and measures the net change in the arena environment by various metrics. Thereafter, one optimises a widening volume of space pari passu. Such a goal has extremely high Kolmogorov complexity, and is therefore improbable. The soi-disant preference utilitarian must have an inaccurate map of his fundaments.

    Hedonic utilitarianism is far simpler: create more happy and fewer sad qualia. I perceive three claims that might be questioned.

    A. Qualia exist, and are separable from humans-as-decision-agents.

    B. One can increase the “measure” of certain qualia, i.e. this hypothesis is false.

    C. Rational decision-agents assign probabilities to qualia concerning whether they are felicitous.

    Virtue ethics, almost no less than hedonic utilitarianism, assumes A, B and C. The premise of such virtues as mercy and kindness is that torture chambers tend to bring about less felicitous qualia than Disneyland. I’ve never known anyone to excuse real, tangible cruelty on the basis that interpersonal hedonic comparison is impossible.

    I cannot conceive of my reaction to persuasive evidence that A, B or C were untrue. I would no longer be able to criticise Auschwitz, torture or rape. I might be left with e.g. a preference for symmetry in mathematical objects (although since 0 is not a probability, this might yet be outweighed by a utilitarian shot in the dark). So I should invoke the virtue-ethicist’s own argument from common sense: almost no humans act as though they doubt A, B or C, and actions speak louder than words.

    If qualia are irreducible, our confidence in felicific calipers might never equal our confidence in voltmeters and fuel gauges. However, we have at minimum such likely proxies as activity of mu opioid receptors or their correlate in an artificial brain.

    Vladimir also suggests that once TDT and the sub-agent perspective are brought to bear, hedonic utilitarianism is so remote from gestalt human behaviour that it cannot be distinguished from other forms of consequentialism in the same framework.

    Were this true, benefits #3 and #4 would continue to render fundaments worthwhile. In cases #1 and #2, I submit that a lot of concrete decisions do hinge on the terminal goal of the executive sub-agent—all sub-agents not being created equal. One with a complex terminal goal, who understands TDT and has no selfhood bias, might endorse utopian political ideas since events in his lifetime are inherently important; a simple hedonic utilitarian should be more conservative. In some cases, such beliefs really matter. I expect that Eliezer Yudkowsky’s executive sub-agent would have chosen a different life were its terminal goal less altruistic, and I don’t doubt that he dwells upon fundaments. Due to the social and scientific novelty of our decision-environment, he and other contrarians and futurists lack obvious and available models and cultures from which to draw appropriate virtues.

    Report comment

    • Vladimir says:

      Hedonic utilitarianism is far simpler: create more happy and fewer sad qualia.

      The problem is that hedonic utilitarianism also immediately diverges from natural human morality and values.

      To put it as briefly as possible, the main reason for this is that in an intelligent and social species, behaviors and characteristics that originally evolved to address straightforward utilitarian game-theoretical and coordination problems eventually become terminal values in their own right. These behaviors and characteristics are, roughly speaking, the same things that have been traditionally called virtue and character.

      This is why the optimum point of hedonic utilitarianism — a universe tiled with wireheads — seems so repulsive. It’s a vision of a universe filled with beings completely empty of virtue and character, which makes it little better, if not worse, than if it were just filled with inanimate matter.

      Report comment

      • Federico says:

        Thank you for the interesting discussion.

        Eliezer Yudkowsky has taught that the repugnant conclusion seems palatable, should one replace “lives barely worth living” by the more accurate “lives barely worth celebrating“. One’s perception of orgasmium might be equally fragile, biased by tiny conceptual inaccuracies.

        Several heuristics suggest to me that hedonic utilitarianism is a credible terminal goal.

        Firstly, many intelligent and social humans scorn virtues and character traits that others cherish. Machiavelli disclaimed friendship, sincerity and faithfulness; men of quotidian politics enact his ideas. Such diverse authors as H.L. Mencken and Edward de Vere have scorned women and romance. Ugliness and obscurity have supplanted truth and beauty in art.

        Outside ascetic religious traditions, few are keen to abandon every virtue and trait, but jointly they put a ceiling on the proposition’s implausibility. Each is cynical about what he knows best. The only universally cherished trait is compassion; I know of no celebrated psychopath.

        Secondly, it is difficult to imagine how we might view our world were we more intelligent. I turn to my perception of sandbox video games like Grand Theft Auto. A 5-year-old is immersed in such a game; adult humans recognise that it is populated by simulacra with a limited repertoire of mechanical responses, whose interactions generate somewhat complex phenomena. Intelligence-enhancement, without transformation of our goal-ness substrate, might cause us to view such virtues as love, friendship, faith, music, art and accomplishment as another such unimpressive suite of rote behaviours. We should marvel at how our lesser selves were too enthralled to notice the aliasing, glitches and stultifying repetitiveness.

        Thirdly, religious concepts of heaven do not stress a plethora of virtues and behaviour. They are memetically evolved, and therefore more trustworthy than rationally constructed utopias. Consider Thomas Morell’s lyric from Jephtha:

        Farewell, ye limpid springs and floods,
        Ye flow’ry meads and leafy woods;
        Farewell, thou busy world where reign
        Short hours of joy and years of pain.
        Brighter scenes I seek above
        In the realms of peace and love.

        Complex or local terminal goals venerate the “busy world”, which Iphis is keen to escape. A Universe tiled with orgasmium more closely resembles a realm of peace and love—not peace, love and each other evolutionary trait with which our spirits are encumbered.

        Buddhism is yet less sympathetic to complex futures. Here are verses from the Hsin hsin ming:

        The Great Way is not difficult;
        just avoid picking and choosing!

        Only when you neither love nor hate
        does it clearly reveal itself.

        To see its truth
        be neither for, nor against.

        Conflicts between longing and loathing
        are a disease of the mind.

        Gain and loss, right and wrong
        away with them once and for all!

        The Perfect Way, like vast space
        lacks nothing, has nothing in excess.

        When the mind does not discriminate
        all things are as they really are.

        Entering the deep mystery of suchness
        releases us from all attachments.
        Viewing all things in their oneness
        we return to our original nature.

        This state wherein all relations have ceased
        is indescribable by analogy.

        The mind in full accord with the Way
        drops off its selfish preoccupations.

        This better invokes the optimum point of hedonic utilitarianism than the LessWrong wiki. Orgasmium, or Nirvana is a highly conscious and blissful sentience. These are qualia without attachment—virtues or character traits.

        Eliezer Yudkowsky, whose executive sub-agent also claims to have a complex terminal goal and even flirts with a local terminal goal, trivialises Zen Buddhism. He has co-opted its language for his Way of rationality, whose “koans” are logical puzzles and not deliberately wrong questions designed to aid meditation. This is unwise, since Buddhism would improve his community’s fundaments and even his approach to Friendly AI.

        Report comment

  14. Federico says:

    It is interesting, albeit political, to consider what virtues ought to circulate at different levels of society. In Wikipedia’s list of political slogans, I found the following which enjoin broad virtues rather than narrow policy aims:

    Deus, Patria, e Familia — Salazar

    Ein Volk, ein Reich, ein Führer — Nazi Germany.

    Liberté, Égalité, Fraternité — French Revolution.

    Ma Mati Manush (“Mother, Motherland, and People”) — All India Trinamool Congress

    Patria o Muerte — Cuban Revolution.

    “Think global, act local” — Various

    ¡Una, Grande y Libre! — Franco

    Two or three virtues must be optimal, in this particular crude context. To contain urges by mutual opposition, a more desirable slogan would juxtapose contrasting virtues. “Liberté, Égalité, Familia”, or “Ma Mati Libre!” To design one from scratch, what about “Liberty, Security and Grace”?

    Some more virtues:

    Humility — One shouldn’t rock the boat.

    Charity — One should give back more than one takes.

    Care — Liberty is the luxury of self-discipline.

    Grace — Anyone can be coarse.

    Enterprise — America’s business is business

    Enlightenment — Greatness is between man and himself.

    One serious, and I daresay memetically evolved benefit of certain religions—and the virtue of grace—is that they direct status-competition and other striving inwards. Nothing is more harmless than a Buddhist or a hipster—except for a Buddhist hipster.

    Report comment