Skip to main content.
October 23rd, 2006

Running and Metaethics

One of the nice features of being in Canberra is that I get to go for runs around Lake Burley Griffin with Nic Southwood and talk (or wheeze) about philosophy. A nice feature of talking about philosophy while running is that, when I’m actually just out of breath and can’t talk, I can pretend that the reason why I’m not talking is because I’m being terribly deep, and having a good hard think about what the best thing to say next is. The last time we went for a run, we got to talking (wheezing) about The Moral Problem. All of the confused parts of what follows are due to me. All of the lucid parts are due to Nic. (Well, except for the ones that are due to Michael Smith or to Brian.)

Here’s a way of stating Michael Smith’s view in TMP that people often get away with:

Aing in C is (morally) right iff our ideally rational selves would advise us to A in C.

(I’ve said this in philosophical company and not had anyone complain about it, and I’ve been in conversations where somebody else said it and all of the rest of us let it pass without complaint.)

Here are two concerns about that view:

1) It doesn’t distinguish the advice that our idealized selves would give on moral grounds from the advice that our idealized selves would give on any other kind of grounds – comic, aesthetic, prudential, or whatever. And so it’s not going to succeed in picking out the morally right. Instead, it’ll pick out something like the advisable all-things-considered.

2) Suppose you think that moral reasons don’t always trump other sorts of reasons. Then you’ll think that there are cases where the (morally) right thing to do is to B, but your ideally rational self would, on account of the stronger countervailing nonmoral reasons, advise you to A. In cases like this (if there are any), the Smithian view above will misclassify Aing as morally right, since it’s the action that our ideally rational selves would, all things considered, advise us to perform.

(There’s a lot of room for filling in cases, and, obviously, a lot of room for disputing about whether particular cases really are examples of the relevant phenomenon. A contentious, but not obviously crazy, example that Brian and I use elsewhere, for other purposes, is a situation where it’d be a little bit morally bad, but really funny, to throw a pie in the face of some undeserving victim.)

Now, the view that people (including me) often get away with attributing to MS isn’t, as far as I can tell from a quick scan, actually endorsed by him anywhere in The Moral Problem. What he actually says is, “our A-ing in circumstances C is right if and only if we would desire that we A in C, if we were fully rational, where A-ing in C is an act of the appropriate substantive kind: that is, it is an act of the kind picked out in the platitudes about substance” (p184, his italics). (I’ve replaced phis with ‘A’s, since I’m a blogging amateur and don’t know how to get the greek letters from my word file into the post.)

It’s not clear, though, how much help this is in handling the two concerns above.
The first objection is clearly what the italicized bit of Smith’s official view is designed to avoid. The move is to distinguish the moral oughts from the nonmoral ones by carving off a domain of actions, such that our idealized selves’ advice about which of those actions to perform is moral advice. (Presumably the same will happen for other sorts of oughts – other domains of behavior will be carved off as the domains of prudential, comic, aesthetic, etc. advice.) (I’m going to be a little bit sloppy about the distinction between what our idealized selves would desire and what they would advise in what follows. I don’t think anything bad will come of it, and I’m too much in the habit of thinking about MS’s view in terms of advice to be able to self-edit reliably…)

The problem is that distinguishing moral and non-moral oughts by appeal to the kinds of actions to which they apply just doesn’t seem like the right way to go. For (pretty much) any kind of action, there can be both moral and non-moral (and various different kinds of non-moral) reasons for performing that kind of action. Sometimes the (predominant) reasons why we ought to cut our hair, sell our shares in Exxon, throw a pie at Brian, eat our vegetables, etc. (or to refrain from doing these things) are prudential. Sometimes the (predominant) reasons why we ought to do (or refrain from doing) these things are moral. If my idealized self would advise me to eat my vegetables for exclusively prudential reasons, then I ought, prudentially, to eat my vegetables, but it’s not the case that I ought, morally, to eat my vegetables. If my idealized self would advise me to eat my vegetables for exclusively moral reasons, then I ought, morally, to eat my vegetables, but it’s not the case that I ought, prudentially, to eat my vegetables. At least, that seems like the natural thing to say. But we can’t say it on Smith’s account. Whether my idealized self’s advising me to A means that I morally ought to A or not depends, on Smith’s account, only on what kind of action Aing is, and not at all on the considerations on the basis of which my idealized self would advise me to perform that sort of action.

It’s also pretty clear that the official view isn’t going to help with (2) – what we need there is, again, a way of identifying a distinctively moral class of reasons for advising one action over another, rather than a way of identifying a distinctively moral realm of behavior, about which all advice is moral advice. Even on Smith’s official view, so long as Aing is an action of the right substantive type, our ideally rational selves advising it – for whatever reason – will be enough to guarantee its rightness. That’ll be enough to render impossible the sort of situation described in (2), where the moral reasons just barely favor Bing, but since they’re outweighed by stronger nonmoral reasons to A, our ideally rational selves would, all things considered, prefer that we A. To the extent that we think this sort of situation is possible, we should be as suspicious of the official view as we were of the not-quite-official one.

So here’s the summary of what Nic and I are worried about, I think (at least, here’s what I’m worried about as a result of talking (wheezing) to Nic about this stuff): Smith’s view looks like it’s trying to draw the distinction between the moral and the prudential, aesthetic, etc. in the wrong place: between sorts of actions, rather than between sorts of reasons for action. And that looks like it’s going to deliver some bad results. If we think that there can be both moral and nonmoral reasons for or against doing more or less anything, and that moral reasons don’t always win when the two conflict, we should expect a lot of misclassification. In cases where the action’s of the relevant substantive type, but the reasons for or against performing it, are, in this particular case, exclusively prudential, it’s going to misclassify actions as right (or wrong) when in fact they’re just prudent or imprudent. When the action’s of the right substantive type, and weak moral reasons against are outweighed by stronger nonmoral reasons in favor, the action will be misclassified as morally right. And when the action’s of the wrong substantive type, but there are strong reasons for doing it that are exclusively or primarily moral, the action will fail to be classified as morally right, even though it seems like it ought to be.

(One way to resist this is to say something fancy about what the substantive types are, such that there can’t be both moral and nonmoral reasons for and against performing actions of the relevant types. Maybe that’ll work. I can’t see real clearly how it would go, and I’m concerned that there’ll always be counterexamples, but I don’t have a good argument in hand that it can’t be done. But my guess is that, even if it works, the fancy things one will have to say will include building stuff about reasons into the action types. And if you’re going to do that, why not just start by talking about reasons?)

We’ve knocked around a couple of ideas about how to futz with MS’s view in response to this, but maybe I’ll leave that until after people have had a chance to say why we’ve got it all wrong about the trouble for the official view and there’s no need for any futzing…

Posted by Andy Egan in Uncategorized


This entry was posted on Monday, October 23rd, 2006 at 5:27 am and is filed under Uncategorized. You can follow any responses to this entry through the comments RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

3 Responses to “Running and Metaethics”

  1. Kenny Easwaran says:

    Good to see that you’ve joined in!

    I don’t have much of anything to say philosophically about this, but if you type the ampersand sign, followed by “phi”, followed by a semicolon, the result looks like this: φ

  2. Michael Smith says:

    I’m almost certainly going to regret this, as in the past I’ve resolved NEVER to submit to a blog. But here goes.

    Let’s start by adding some more structure.

    The basic unit of analysis is the claim ‘It is desirable that p.’ More precisely, the basic unit of analysis is the claim that it is intrinsically pro tanto desirable that p, and we analyze this in the way you say when you give the official analysis, with the extra assumptions that the desires in question are intrinsic (because it is the concept of intrinsic desirabilty that’s up for analysis) and that they are the sorts of desires that can be overridden by other desires (because it is the concept of pro tanto desirability that’s up for analysis). The upshot is that the intrinsic desires in question will be desires which have a certain strength as compared with other intrinsic desires that you have when you’re fully rational, and the strengths of these intrinsic desires vis a vis each other will fix the relative degrees of intrinsic pro tanto desirability in question. All things considered desirabilty will be a function from all of the different things that are pro tanto desirable.

    With this basic unit in place, we analyze the concept of a normative reason for action in something like the following way: ‘An agent has a normative reason to A in C iff his A-ing in C is an action that he has the option of performing that would bring about a state of the world that is desirable.’ I think of this analysis as Moorean in spirit. This is because, in true consequentialist-in-the-Moorean-sense spirit, the concept of a reason for action (ie the concept of what we ought to do) gets reduced to the concepts of an option and desirability (ie the concept of ability and the concept of value). In order to type reasons for action, we would therefore need to type the substantive kinds in the desirable outcomes, and here, I take it, much for the reasons you suggest, it will be important that we focus on substantive kinds among the intrinsically desirable outcomes.

    Won’t this take care of the examples you consider? Let me run through a single example just to illustrate. For me to have a prudential reason to eat my vegetables is for me to have the option of eating my vegetables where the outcome of my doing so is something that my fully rational self intrinsically desires to obtain and where that outcome is of the appropriate substantive kind to count as prudential. I take it that this means that the outcome is something like, that I am healthy, maybe. For me to have a moral reason to eat my vegetables is for me to have the option of eating my vegetables where the outcome of my doing so is something that my fully rational self intrinsically desires to obtain and where that outcome is of the appropriate substantive kind to count as moral: that people in general be happy, perhaps.

    What do you think?

  3. Andy Egan says:

    Thanks for the reply, Michael! Long delay in reply on account of being out of town giving talks and stuff…

    That sounds to me like it ought to work. Let me make sure I understand the story, though: It looks like your response has got two moving parts: talking about pro tanto desires rather than all things considered desires, and distinguishing substantive kinds of outcomes rather than substantive kinds of actions.

    The pro tanto desire move was one that Nic and I talked about, but we were still worried because it\‘s not much help if your substantive kinds that are distinguishing between the moral and the non-moral are kinds of actions – if the action\‘s of the right substantive kind, every reason for or against performing it will wind up getting counted as a moral reason. The second part of the response – having the substantive kinds be kinds of outcomes rather than kinds of actions – looks like it handles that worry.

    (It\‘s interesting that the first move points to a spot where my sliding between the desires and the advice of our ideally rational selves really is troublemaking. I\‘d been reading the desire-talk in the quote I took from TMP as being about all things considered desire, rather than pro tanto desire, which is real natural if you\‘re really thinking, in the background, about advice rather than desire all the time (as I was).)

    Is the story about rightness in terms of having a moral reason, then, that it\‘s right for S to A in C iff S has a decisive moral reason (or a bunch of collectively decisive moral reasons) to A in C?

Leave a Reply

You must be logged in to post a comment.