The search for altruism

The games economists play

If you’re a college grad (or even in college now), you’ve probably seen tons of ads calling for volunteers for various experiments. When I was in school, I never took my professors up on it. Twenty bucks didn’t seem worth the two-mile trek to where an experiment was conducted, and if they offered more than $100, I figured they must be doing something so weird to test subjects that I didn’t want to show up. (I later found out that an economist might call that line of thinking “adverse selection”)

If you did go, and an economist or sociologist was running the show, you might have been asked to answer a few survey questions or play a few games. The testers were trying to understand how you made decisions in the face of problems that closely mirrored real world situations. Collecting sample data from the real world is always better than trying to duplicate the world in a lab, but unfortunately, it’s much harder for scientists to control variables that might taint the experiment.

In 1982, a few economists running tests in a lab forgot to account for one major variable. And that tainted our understanding of “altruism” for decades.

Do we have no compassion at all?

During most political science and economics curriculums, professors take some time to teach game theory. Even if you didn’t study those courses, you might have a basic understanding of it. The prisoner’s dilemma is probably the most classic example.

Games like the Prisoner’s Dilemma work under the assumption that humans act in their own best interests. Even when they seem to perform an action that helps another, they’re really hoping for something in return. True “altriusm”, in other words, doesn’t exist.

In the early 1980s, economists decided to test if that assumption was true. To do that, they invented a new game, dubbed “Ultimatum”. The concept was simple. Two test subjects—we’ll name them Ben and Tim—have to decide how to split a $20 pot of money. Ben never sees or speaks to Tim other than to transmit his offer.

Ben begins by proposing a split. Tim can choose to accept or reject the offer. If he accepts, the money gets split in the way Ben proposed. If he rejects it, neither of them get anything.

It’s pretty easy to see how traditional game theory would have Ultimatum play out. Since something is better than nothing, Tim is supposed to accept any offer above $0. Since Ben knows the best interests of Tim, he is predicted to offer only $0.01. He gets to keep the majority, and Tim, while unhappy, should accept rather than walk away with $0.

You can probably see where this is going. The game never worked out that way. It turned out that Tim, far from acting rational, rejected most offers of less than a few dollars. He’d rather stick it to Ben for low-balling than walk away with a piddling amount.

Ben, on the other hand, rarely made an offer of just $0.01. He’d offer several dollars—sometimes even splitting the pot 50/50.

Was this evidence of altruism? Maybe. But you could argue that Ben wasn’t emotionally deaf and knew Tim might reject an offer perceived as insulting. A new experiment had to be devised.

From Ultimatum to Dictator

What economists came up with next was simple. A new game, dubbed “Dictator”, would have the same two test subjects, a similar pot of money, but a one-sided transaction. Ben would simply decide how to split the pot. He could keep $19 for himself and give $1 to Tim. He could keep $10 for himself and give $10 to Tim. It didn’t matter. Tim couldn’t accept or reject the offer. His take was left completely in the hands of an altruistic or greedy Ben.

Again, the economists were surprised by the results. A completely rational and self-interested Ben would keep all or nearly all the money for himself. But these Bens tended to give at least a few dollars to Tim. Sometimes they’d split it 50/50. In a variation of the game, where both Ben and Tim worked for the money, Ben was even more likely to give Tim half.

“A ha!” said economists. “Surely this proves that pure altruism exists.”

But…

Alas, the economists—working in a lab, having designed a game that neutralized every unwanted variable they could have possibly imagined—forgot to address one, crucial variable: themselves. Ben didn’t want to look like a pig. Even though Tim would never know who he was, the guy sitting in the room with a pencil and clipboard did.

Since the original Ultimatum and Dictator experiments, many economists have tried to duplicate the results outside of lab settings and failed. Outside of a lab, where the test subjects don’t necessarily know they’re being tested, people are much less apt to be generous.

One experiment, cited in the book SuperFreakonomics, sought to see if baseball card dealers at a convention ripped off customers. When the dealers knew they were being tested, they gave the customers a decent deal. When they didn’t know it was a test, they started taking advantage of them.

That’s not ironclad evidence that altruism is dead. But it does show that the search goes on. And that a lot of lab-based economics experiments are a load of crap.

Further reading:

SuperFreakonomics by Levitt and Dubner

Dictator game giving: altruism or artefact? by Nicholas Bardsley

Tagged as: altruism, charity, game theory