March 9, 2010

This Post is Old!

The post you are reading is years old and may not represent my current views. I started blogging around the time I first began to study philosophy, age 17. In my view, the point of philosophy is to expose our beliefs to rational scrutiny so we can revise them and get better beliefs that are more likely to be true. That's what I've been up to all these years, and this blog has been part of that process. For my latest thoughts, please see the front page.

Deontic Utilitarianism, Liberty Utilitarianism, and Deontologism

(Cross-posted to Southern California Philosophy.)

I just came across the following passage by J.J.C. Smart in Smart and Williams' Utilitarianism: For and Against:

What Bentham, Mill and Moore are all agreed on is that the rightness of an action is to be judged solely by consequences, states of affairs brought about by the action. Of course we shall have to be careful here not to construe 'state of affairs' so widely that any ethical doctrine becomes utilitarian. For if we did so we would not be saying anything at all in advocating utilitarianism. If, for example, we allowed 'the state of having kept a promise', then a deontologist who said we should keep promises simply because they are promises would be a utilitarian. And we do not wish to allow this (p. 13).

This line of thought is horribly confused. A similar confusion is manifested in Amartya Sen's Inequality Reexamined (incidentally also p. 13), where he says that libertarians such as Robert Nozick are centrally concerned with equality of liberty or of rights. Let me try to clarify this muddle. Deontologism is the view that some actions are right or wrong independent of their consequences. An essential feature of libertarianism (especially of the Nozickian variety) is the view that there is a deontological prohibition on the violation of rights - i.e., the violation of rights is always wrong irrespective of the consequences. Now consider the following views:
  • Deontic Utilitarianism: Agents ought to act in such a way as to maximize the total number of morally good actions, and minimize the total number of morally bad actions.

  • Liberty Utilitarianism: Agents ought to act in such a way as to minimize the number and severity of violations of individual rights/freedoms.

Deontic utilitarianism is not a form of deontologism, and liberty utilitarianism is not a form of libertarianism. (If, as Sen does, we switch to speaking of equalizing rather than maximizing, the situation will not change.) In order to deal with both at once, assume that the deontic utilitarian believes that violations of the right to free speech are morally bad. Now consider the following scenario:
Joe intends to give a speech (on his own property and at his own expense) to which the general public will be invited. In his speech, Joe will rail against Islam, using emotional appeals to persuade his audience that wherever Islam is publicly promoted terror and violence follow. Joe will not promote violence against Muslims in his speech. Sarah knows of Joe's intentions. She further knows that if Joe gives his speech, it will (very likely) start a political movement which will (very likely) result in a ban on the publication and distribution of (peaceful) Muslim literature. However, she also knows that Joe's speech is very unlikely to lead to violence in any form. Sarah is able, by coercive force, to prevent Joe from giving his speech. By violating Joe's free speech rights just once, she can protect the free speech rights of all the Muslims in her country (there are a lot of them) for the foreseeable future. She knows of no other way she can accomplish this. What should she do?

A genuine deontologist (such as myself) would counsel Sarah to refrain from acting: the ends do not justify the means. But a deontic utilitarian or a liberty utilitarian would counsel Sarah to prevent Joe's speech. It follows that deontic utilitarianism is not a form of deontologism, and liberty utilitarianism is not a form of libertarianism.

Posted by Kenny at March 9, 2010 12:28 PM
TrackBack URL for this entry:


This isn't quite right--considerations like the ones you raise have been used to show that deontological theories can't be represented as recommending value-maximization for some agent-neutral, time-neutral value. But if we allow the relevant value to be agent and time relative (e.g., it assigns special disvalue to MY violating rights NOW, versus my violating rights later, or someone else's violating rights at now or some other time), then we can accommodate cases like the ones you mention; deontological theories can still be represented as recommending that agents maximize value (albeit for an unusual looking sort of value).

Michael Smith's written a lot on this, see the following, for example:

Posted by: Daniel at March 17, 2010 9:55 PM

Daniel - That's clearly right. But allowing 'the state of having kept a promise' to enter the equation isn't going to collapse the distinction, as Smart suggests. Smart's suggestion would amount to trying to maximize the number of instances of promise-keeping, which does not correctly capture the deontological picture.

Posted by: Kenny at March 17, 2010 10:38 PM

I haven't read the Smart, but the proposal I had in mind (the sort discussed by Smith) doesn't just amount to assigning value to the "the state of having kept a promise" and the like.

On the Smith sort of proposal, promise-breakings by me now would be weighted much more heavily than promise breakings by me later, or other people at other times. So I wouldn't break a promise now to stop myself (or somebody else) from breaking promises in the future, since the promise-breakings by me now would count for more disvalue than those other promise-breakings.

If you want to say that no matter how many future broken promises (or murders) I could prevent by promise-breaking (or murdering) now, then you'll need to make the relevant value being maximized pretty funky (it will have to either involve some infinities, or something else that serves the same role), but you can still put it in terms of value maximization.

Other people who rely on the possibility of "consequentializing" all moral theories (often by using bizarre value functions) include Andrew Sepielli (see here: and Jacob Ross (see here:

Posted by: Danie at March 18, 2010 7:24 AM

I agree with much of what Daniel says about consequentializing, but I don't think that what Jake, Ted Lockhart, me, and other "normative uncertainty" people are doing relies on it. We don't need to characterize some deontological theory as saying that a) one ought to act so as to produce the best consequences, and that b) my killing 1 now is a worse consequence (relative to me) than is your killing 20 in the future. We can just talk in terms of reason strength: there are stronger reasons for me to avoid killing 1 now to prevent you from killing 20 in the future than there are to do the complementary action. There is no reason to think that reason strength is less amenable to cardinalization than consequence-value is; and it makes no less sense to speak of maximizing expected reason strength than it does to speak of maximizing expected consequence-value.

Posted by: Andrew Sepielli at April 11, 2010 3:53 PM

Post a comment

Return to