« IQ or Financial Incentives? | Main | Gerd Gigerenzer »
August 16, 2004
The Social Dilemma of Exaggerating the Positive
THE SOCIAL DILEMMA OF EXAGERATING THE POSITIVE-OF PEOPLE AND OF POLICIES
Imagine you are invited to a party with ten strangers and asked to play an unusual game. Upon arrival you are presented with a choice. You can either 1) choose to receive $10 now, which you may keep, or 2) choose to receive $20, which you must entirely give away by evenly dispersing $2 dollars to each of the other 10 strangers.
If everyone votes for (2), everyone leaves with $20. However, if only you vote for (2) and everybody else votes (1), they leave with $12 each, and you leave with nothing. Many other combinations of votes are possible.
Carnegie Mellon professor Robyn Dawes speculates on this problem ...
Now suppose that this game is repeated over a number of trials. What happens? Many colleagues around the world and I have conducted similar games, and the results are quite predictable. Especially if people are able to communicate with each other and make commitments prior to the first trial, almost all give the money away. But what happens on later trials where people know what they did on the previous trial and can infer from their payoff what other people did collectively? The rate of cooperation (giving) over trials is highly predictable. If this rate does not reach 100% on the first trial, some people note that they are making less money than those who keep their $10 stake and they then switch from being cooperators to being defectors. Subsequently, more people do so, often at an accelerating rate. By the end of ten trials, virtually no one is giving away the money. What has happened is that behavior has stabilized on the sink of universal defection. The one exception is that if people do not find out what happened after the first trial, where generally a majority honor the commitments to cooperate, subjects avoid the sink. When a few people know, however, that they are being "suckered" by others when they give away their $10 they stop doing it. (It is even possible to "reset" the situation by having an additional discussion prior to some trial, but even then over trials without discussion, the group ends up in the sink.)
An observer looking at behavior at the beginning of this degenerating process and ascribing the results to personality characteristics would conclude that people are altruistic and cooperative, while someone observing the end of the process and making similar personality attributions would conclude that people are generally selfish and uncooperative. But it is the situation itself that yields the behavior, at least on the part of anyone attempting to behave "rationally".
Posted by DSN at August 16, 2004 03:02 PM