You Can Make a Utility Function for Literally Anything

The idea behind the "rational choice" theory of decision-making, and its applications in numerous fields including both economics and international relations (the latter of which I'm currently taking a course on), is that individual people are rational actors who behave so as to maximize the attainment of their goals, to maximize their "utility." Now, since economists came up with this theory (and have been using it for centuries!), there's also a tendency within the field to assume that individual preferences are extremely standard. People want food, shelter, to continue living. They want money. They might want a few other things, too, like a decent romantic relationship (or, as my professor cynically boiled that notion down, they want sex), but for examining economics or international relations you don't need to worry about that stuff. And of course, people are all quite similar in their wanting of these basic things, so we can treat people as more-or-less interchangeable.

Now, of course, this leads to a picture of the world that's radically inaccurate. Human beings are wildly complex. Their desires are myriad and mysterious, often even to themselves. They behave in weird and unpredictable (or at least, damn hard to predict) ways. And they're all spectacularly different from each other, even when they're quite similar. The psychology of the individual, or in the words of Bertie Wooster, "what they're like," matters quite a lot. So of course the people who believe, rightly, in the psychology of the individual as an important sort of thing form various sorts of movements, like behavioral economics and political psychology, that rebel against the extreme modelling of rational-choice. And they're certainly right to do so, given how the phrase "rational-choice" has played out historically.

But consider the following passage from the article on political psychology I'm currently reading:

"Most rational choice approaches argue that people will seek to maximize their utility; they may not get the outcome they prefer for a whole host of external reasons, but they do not neglect actions designed to maximize their self-interest. From this perspective, it is difficult to explain why rational actors would engage in self-destructive behavior, such as drug addiction, without perverting the notion of self-interest."

That last sentence is only true if we assume that our rational choice theorist is also a paternalist, i.e. someone who assumes he knows what another human being's utility function is. In this particular case, our paternalist is insisting that the drug addict cannot possibly have a utility function which prefers addiction to sobriety. There's no particular reason, from a theoretical standpoint, to assume this. (When making public policy we might decide to be paternalistic in a case such as this, and decide that this person's utility function should prefer that they get clean, but that's a different story.) You can make a utility function for literally any set of preferences. Suicide can easily be explained within a rational-choice approach by simply asserting/assuming that, at the time of suicide, the person in question preferred to die. In my Law & Society class we're about to contrast rational-choice with "normative" theories of decision-making, in which people are motivated by an internal sense of morality. There's no conflict here! Just say that people, or some people anyway, derive utility from behaving consistently with their internal sense of morality. Now a rational-choice approach will show you a world in which people are motivated by their internal sense of morality. However in the entire world you want to explain people's decision-making processes, whatever preferences or desires or animal instincts you ascribe to people, you can just make a utility function incorporating those features. It's really that easy.*



*On a theoretical level, anyway. If you then actually wanted to make a mathematical model of such a utility function, you might find it impossibly complex. In fact you probably would; like I said, we humans are unbelievable complicated people.


Also, as an aside, this is part of what I dislike about the word "rational." Rationality is a characteristic of reasoning, of getting from premises to conclusions (or in this case, getting from preferences to outcomes). You can't evaluate premises, or preferences, on the grounds of rationality, and if you can, that only means you haven't followed the chain of derivation all the way back to its start, to the true premise. So I can't say that the drug addict here is being irrational, because he or she might be rationally following a utility function of which I have a very low opinion. But I could say that they're being unreasonable, the word John Rawls uses to judge the merits of premises. The preference for drug addiction, or the preference for death, can easily be described as unreasonable. I'm not sure exactly what kind of objective criteria we can introduce here, unlike in the case of rationality, but ultimately the world's a subjective place, and we're subjective beings.

UPDATE: Like this, see, a later paragraph from that same article:

...many seemingly chaotic, bizarre, or odd occurrences arise from the rational calculations of different people espousing different preferences. For example, Westerners may have a hard time understanding why someone would commit suicide for a political goal, but from another perspective where life is bleak and the prospect for improvement slim, suicide may seem like a rational act in order to make life better for the family members who remain.

If "rationality" is all we care about, then either we must commit the error of paternalism as to preferences, claiming that everyone must have the same standard preferences and therefore suicide is irrational, or admit that the suicider in that example is in fact rational. But that sounds like an approval of their actions, an admission that they've behaved in a perfectly proper fashion, because after all, it's rational! I might say, on the other hand, that the preferences that lead to suicide are unreasonable. Or they might be; some might want to argue with me on that point, and that's their right. But the point is, you can critique preferences, and premises, even though you can't call them irrational.