Monday, February 9, 2015

An altruistic prisoner's dilemma

Jeff Kaufman talks about an ethics trade that he sometimes does with a friend.
I have a friend who is vegan for animal welfare reasons: they don't think animals should be raised for food or otherwise suffer for our benefit. On the other hand, they used to really enjoy eating cheese and miss it a lot now that they're vegan. So we've started trading: sometimes I pass up meat I otherwise would have eaten, and in exchange they can have some cheese.
This is a win-win for the vegan, since they get to have some cheese, and there is no net harm to animal welfare.  It is not clear what's in it for Jeff though, except for his idiosyncratic preference to have such trades.  I am not sure this is an interesting scenario by itself, since, in general, any trade is possible with sufficiently idiosyncratic preferences.

Therefore, I propose a similar scenario, which I'll call the vegan/omnivore dilemma.

Suppose we have Vivian, who is cares about animal welfare, but likes eating meat, and Oscar, who doesn't care about animal welfare, but is so-so on meat.  Both people are altruistic, in the sense that they care about the total utility, not just their personal utility.  However, Vivian thinks the total utility includes the welfare of animals, while Oscar does not.

Let's say that if Vivian eats meat, that results in 2 utils compared to eating vegetables.  If Oscar eats meat, that results in 1 util.  Since both people are altruistic, these utils are seen from both of their perspectives.  For each person that eats meat, that results in a net loss of 3 utils, but only from Vivian's perspective.  Here's the outcome matrix:

Structurally, this game is identical to the prisoner's dilemma.  Vivian prefers to eat veggies, because eating meat is a net harm.  Oscar prefers to eat meat because he likes meat.  And yet, if each person acted upon their preferences, we'd get the outcomes in the upper right corner (1 util from Oscar's perspective, -2 utils from Vivian's).  From both people's perspectives, this is worse than the outcome in the lower left corner (2 utils from Oscar's perspective, -1 util from Vivian's).

This is interesting, since the typical statement of the prisoner's dilemma assumes selfish preferences.  Usually we have two prisoners, each of which wants to reduce their own jail time, and does not care one iota about the other prisoner.  But in the vegan/omnivore dilemma, both people are acting upon a measure of total good.

One possible resolution to the prisoner's dilemma is for both prisoners to start to think about the other prisoner, and exercise a bit of altruism.  However, in the vegan/omnivore dilemma, each player believes they are already being altruistic, and that the other person is simply incorrect about their assessment of what is good.  Must each person exercise "meta-altruism", and help the other person fulfill their misguided ideals?

I'm inclined to think this might be a case where "defecting" is the morally correct choice.

Although the typical prisoner's dilemma can be resolved by introducing altruistic preferences, the vegan/omnivore dilemma can be resolved by introducing selfish preferences.  Suppose each person selfishly prefers having money, and doesn't care how much money the other person has.  Then each player could "buy" the other player's cooperation.  If the prices are right, there doesn't need to be any actual exchange of money.  This is basically the situation described by Jeff Kaufman.  However, for a number of practical reasons this resolution isn't very satisfactory.

1 comment:

miller said...

In this dilemma, "defecting" means acting according to your own preferences. For Vivian, to defect is to eat veggies. For Oscar, to defect is to eat meat.