Wednesday, September 10, 2014

The limits of expected utility

It is probably not a surprise to most readers of this august periodical to find out that I yield to few people in my appreciation for economic reasoning. Mostly, the alternative to economic reasoning is shonky, shoddy intuitions about the world that make people worse off. Shut up and multiply is nearly always good advice - work out the optimal answer, not what makes you feel good. The alternative is, disturbingly often, more people dying or suffering just so you can feel good about a policy.

But perhaps it may be a surprise to find that I not infrequently end up in arguments with economists about the limits of economic reasoning in personal and ethical situations. There is often a tendency to confuse the 'is' and the 'ought'. We model people as maximising expected utility, usually over simple things like consumption or wealth, because these are powerful tools to help us predict what people will do on a large scale. But for the question of what one ought to do, it is particularly useless to do what some economists do and say, 'Well, I do whatever maximises my utility'. No kidding! So how does that help you decide what's in your utility function? Does it include altruism? If so, to whom and how much? Do you even know? A lot of ethical dilemmas in life come from not knowing how to act, which (if you want to reduce everything to utility terms) you could say is equivalent to not knowing how much utility or disutility something will give you. There's ways to find that out, of course, but those ways mostly aren't economics.

More importantly, this argument tends to sneak in a couple of assumptions that, when brought to the fore, are not nearly as obvious as the economics advice makes them.

Firstly, it's not clear that utility functions are fixed and immutable. This is perhaps less pressing when modeling monopolistic competition among firms, but is probably more first order in one's own life. Could you change your preferences over time so that you eventually got more joy out of helping other people, versus only helping yourself? And if so, should you? It's hard to say. You could think about having a meta-utility function - utility over different forms of utility. For the same amount of pleasure, I'd rather get pleasure from virtue than vice. This isn't in most models, although it probably could be included in some behavioral version of stuff (I suspect it may all just simplify to another utility function in the end). But even to do this requires a set of ethics about what you ought to be doing - you need to specify what behavior is utility-generating behaviour is admirable and what isn't. Philosophers have debated what those ethics should be for a long time, but you'll need to look outside economics to find what they are.

Mostly, people just assume that whatever they like now is good enough. Of course, they're assuming their desires don't raise any particular ethical dilemmas. You can always think about extreme cases, like if someone gains utility over torturing people. Most die-hard economists would probably still not give the torturer the advice to just do what gives them utility. They'd try to find wiggle ways out by saying that they'd get caught, but that just punts the question further down the road - if they won't get caught, does that mean they should do it? You'd probably say either a) try to learn to get a different utility function that gets joy from other things (but what if they can't?), or if they're more honest b) your utility isn't everything - some form of deontology applies, and you just shouldn't torture people for fun simply because you find it enjoyable.

Of course, if you admit that deontology applies, some things are just wrong. It doesn't matter if the total disutility from 3^^^3 dust specks getting in people's eyes is greater, you'd still rather avoid torture. Eliezer Yudkowsky implies that the answer to that question is obvious. How many economists would agree? Fewer than you'd think. I'm probably not among them either, although I don't trust my intuitions here.

But fine, let's leave the hypotheticals to one side, and consider something very simple - should you call your parents more often than you do? For most young people, I'd say the answer is yes, even if you don't enjoy it that much. Partly, it's something you should endeavour to learn to enjoy. Even if this doesn't include enjoying all of the conversation, at least try to enjoy the part of being generous with one's time. Though the bigger argument is ultimately deontological - children have enormous moral obligations to their parents, and the duties of a child in the modern age include continuing to be a support for one's parents, even if you might rather be playing X-Box. If you ask me to reason this from more basic first principles, I will admit there aren't many to offer. Either one accepts the concept of duties or one doesn't.

In the end, one does one's duty not always because one enjoys it, but simply because it is duty. Finding ways to make duty pleasurable for all concerned is enormously important, and will make you more likely to carry it out, but in the end this isn't the only thing at stake. There is more to human life than your own utility, even your utility including preferences for altruism. It would be wonderful if you can do good as a part of maximising your expected utility. Failing that, it would be good to learn to get utility from doing good, perhaps by habit, even if that's not currently in your utility function. Failing that, do good anyway, simply because you ought to.

No comments:

Post a Comment