Tuesday, March 31, 2009

The science of karma

A fascinating new study from Germany finds that revenge doesn't pay.

In a sane world, this would be no surprise. Energy spent on making someone else's life crap, is energy that's not working for oneself, or one's loved ones. In economic terms, it's simply a waste of resources. From Othello to Captain Ahab to the Count of Monte Cristo, our culture is full of illustrations of how revenge is a dish best not eaten at all.

The fact that this isn't considered blindingly obvious is probably because there are so many batshit-insane people around nowadays. In particular, believers in that beguiling branch of maths known as "game theory" - a category that, nowadays, includes almost everyone concerned with social sciences. Game theory, as popularly interpreted, implies that if someone does something bad to you, you should retaliate proportionately against them.

As far as it goes, the theory is fair enough. But it doesn't go nearly as far as many people think it does. Game theory, like every mathematical theory I've ever heard of, is based on a vast pile of mostly-unspoken assumptions that simply aren't true in the real world. For instance, there's the assumption that all players are following the same rules, which are not liable to change. Then there's the assumption that the number of players is constant - nobody will enter or leave the game mid-way. Or that all players have the same motivation. In most economic and social contexts, these assumptions are patently absurd. But they're so seldom stated that it's easy to forget they're there.

Now, thanks to Prof Dr Armin Falk of Bonn University and his co-authors, we have concrete evidence that "negative reciprocity" is truly counter-productive. In particular, the study finds, vengeful people are much more likely to find themselves unemployed.

What does work, on the other hand, is "positive reciprocity". Turns out that, regardless of the level of motivation, returning good behaviour leads to a much more profitable relationship than punishing bad behaviour.

What I like about this study is that it doesn't care about means or mechanisms. It doesn't try to prove, or even imply, that reward is a better motivator than punishment, or that relationships based on threat of punishment are unhealthy. It doesn't even try to argue with game theory. Maybe these things are true, maybe they aren't; that's not the point here. The point is that there is a measurable difference in outcomes.


castlerook said...

Actually, I've seen discussions of Game Theory that reach a similar conclusion.

Here's a good example:

Anonymous said...

I misremember that some of the earlier Game Theory experiments didn't pan out when women were the subjects, so the experimenters put it down to women being unreliable.

I can't easily find proof on Google though.


vet said...

castlerook, I've seen that story before - it's the mainstream interpretation I was talking about, which identifies "measured revenge" as the optimal strategy. And I'm not denying it works in that specific game setup. But does it work in real life? Or in non-iterated games?

I'm a fan of Hofstadter's analysis of the Prisoner's Dilemma - I love the idea of "superrationality". In practice, I think it's much more common than game theorists think it is. Every time a stranger walks into a shop, or every time you buy something online, both sides have the opportunity to betray one another - using a fake credit card, not shipping the goods, whatever. The Game Theory analysis says that, in those cases, people should pretty much always betray one another. But in real life, it's quite rare for that to happen. Obviously the laws and police have something to do with that, but they can't account for all of it.

S: when the RAND Corporation played these games with their staff in the 1950s, they found that the highly paid economists and mathematicians consistently ratted each other out, but the secretaries and mailroom types and other lowly workers didn't. At the time they put this down to their being uneducated, and therefore of no account.

My idea is that the basic game theory analysis doesn't take account of all the payoffs. For some people, there is self-evidently an added payoff of some kind in considering yourself a nice person. I think that's true for nearly everyone - everyone, basically, who's not a psychopath.

Anonymous said...

Thanks for the reply. Good as always.

Ruby Apolline said...

Really interesting.

While I am by no means a sociobiologist (evolutionary psychologist in today's parlance), I do think studies exploring "genetic" or group genetic behaviors are of interest in beginning to determine physiological structures that may guide decision making and that must be taken into account for any accurate predictions of human behavior.

All this is to say, Hofstadter's "superrationality" sounds to me like Robert Wright's (author of a couple e psych books I've read)truth-telling as evolutionarily adaptive behavior in social animals. As I recall, he describes many non-human animal societies with elaborate mechanisms for determining cheaters and liars and expelling them from the group, which is bad. We may already have these scripts running somewhere and, as such, do not behave as the blank slates game theory might have us be.

vet said...

Ruby, I'm not familiar with Robert Wright, but last week we attended a couple of lectures by Dr de Waal, billed as one of the world's leading primatologists, who promised to talk about "morality" in primate societies. In fact he didn't, really. He did mention reciprocity, but it seemed to be quite firmly in the realms of "you share your food, I'll pick your lice" - there was no mention of "community values", ostracism or other punishments for transgressors. I found that quite disappointing, at the time. But I think it was chiefly because "morality" is mostly a philosophical question, not a psychological one, and he has no taste for that.

Anonymous said...

re. "positive reciprocity" - You may have already seen the article on BBC news 'Chimpanzees exchange meat for sex'. Lest you jump to the concept of Chimp Ho's, This is a long-term exchange, so males continue to share their catch with females when they are not fertile, copulating with them when they are.