(A repost from my visit to Crooked Timber).

In a couple of recent posts, Matt Yglesias has raised the question of how consequentialists should handle “other-regarding” preferences. He gives two examples. The first is about the possible execution of Saddam Hussein

My own take on the punishment issue leads to a somewhat paradoxical result. … If Iraqis would feel better with him executed, then go for it…
I like to think of this as a wise and sophisticated point of view, but the trouble is that my preferences depend on other people’s preferences. As long as not very many people agree with me, that’s fine, but if some huge portion of the world were to decide I was right, then you’d wind up with an unfortunate self-reference paradox. Sadly, consequentialist attitudes tend to have these kind of results and I think that if I were smarter I would dedicate my life to resolving the problems.

The second is about the preferences of people who are repulsed by overtly gay behavior. Matt concludes that their preferencesmust be counted, although they should be argued against.

This is an issue of considerable practical interest to resource and environmental economists, because of the popularity of stated preference methods for evaluating public goods such as environmental preservation. I find these methods problematic and one big problem is the treatment of other-regarding preferences.

This is why I have an article on the topic in the American Journal of Agricultural Economics, (PDF and algebra alert). Not, I imagine the kind of journal that philosophers like Matt read with any regularity

In this paper, I show that the kind of disinterested (the jargon term is ‘nonpaternalistic’) altruism considered in the Iraq example, (I want whatever the Iraqis want) should not be counted in a consequentialist evalation. The argument proceeds by comparing what you’d get by looking at individual preference statements with those from M members of a mutually altruistic household or community. You’d hope that, in any operational procedure these would be the same, especially when each member of any given household wants the same thing.

This is true of a voting procedure, for example. We get the same outcome if we allow each member of each household to vote individually, or if we let them collectively cast M votes (remember that in this simple example they all vote the same way)(.

By contrast, if you try to implement a consequentialist assessment taking account of the mutual altruism of the household, you end up in a complete mess. In the case of perfect altruism, each individual gets counted M times over, once for themselves and once for every other family member. Things get even worse when some groups have negative altruism towards others (I want them to get whatever they don’t want).

So Matt’s Iraqi example can be resolved reasonably easily by saying that we should evaluate the consequences for Iraqis, on the standard assumption that they know their own preferences, and that Matt’s altruistic preferences should be disregarded. Some difficulties arise when we ask whether Kuwaitis, Iranians etc should have a say also, but I don’t hink they are insuperable.

The second problem is much trickier, and can’t be resolved at the level of abstraction we are currently using, simply because we don’t have a self-evident criterion for classing things as self-regarding or other-regarding. Consider, in addition to Matt’s example of overtly gay behavior and compare the cases of public nudity and smoking in enclosed public places. I guess that most people reading this would want to permit the first, limit the second and prohibit the third but that obviously would not have been the case fifty years ago.

As Matt mentions, Mill tried to bluster his way past this one at the beginning of On Liberty, but he also showed the correct approach with the rest of his discussion . Rather than seeking a first-principles argument that other-regarding preferences should not count, Mill gives consequentialist arguments to suggest that we are all better off if society defines a sphere of self-regarding actions and lets individuals choose for themselves within this sphere. This is true even if, in some short term sense, aggregate utility would be increased by imposing conformity with social norms. Hence, once we have decided that marriage is (largely) within the private sphere* and that homosexuals and heterosexuals should have equal rights, we should disregard, for policy purposes, the preferences of people who are uncomfortable about this. [You can strengthen this case with rule-utilitarianism if you want to, but I don’t think it’s necessary]

*Of course, there’s a huge feminist debate about this, but I don’t think it’s crucially relevant to the point I’m making.

Reference: , ‘Individual, household and community willingness to pay for public goods’, [AJAE 1998, 80(1), 58-63.