Brain-Imaging Study Solves the Auction ‘Overbidding’ Mystery


by Jim Schnabel

November 19, 2008

Economists have observed for decades that participants in auctions tend to bid more than is predicted by traditional theories of rational economic behavior. A recent  "neuroeconomics" study in Science, combining modern brain-imaging technology with behavioral experiments, blames this seemingly irrational behavior on an auction’s social context.

“The conventional wisdom is that there is some pleasure in winning so that people are both bidding based on how much they value what they might get, and just to win,” explains Colin Camerer, a neuroeconomist at the California Institute of Technology who did not participate in the research. “This study indicates the opposite: People get caught up in ‘auction fever,’ and if it seems likely they will win, they bid more to be sure they don’t lose. So the overbidding is like a kind of insurance payment.”

In the study, Rutgers University psychologist Mauricio Delgado and colleagues used functional magnetic resonance imaging (fMRI) to look at changes in activity in the brains of 17 volunteers, mostly students, as they participated in two kinds of economic “game.”

The first was an auction game with a partner, the second a lottery game with no partner. In both cases the choices and total winnings available to a participant were the same; the only difference was that the auction had a competitive social context, in which competitors met each other before playing the game.

As expected, Delgado and his colleagues found that the students bid significantly higher in the auction setting than in the lottery setting. To try to understand why, they looked at how the fMRI responses varied with wins and losses in each setting, and found a clue in a region of the brain known as the striatum.

Previous experiments had shown that the fMRI response in a key portion of the striatum increases or decreases with monetary gains or losses. In this case, the gap in response between winning and losing was much larger in the auction setting, but apparently not because the auction involved a greater “joy of winning.” The positive responses to wins were virtually the same in both settings. But the negative responses to losses were pronounced in the auction setting and almost non-existent in the lottery setting.

“We believe that the contemplation of losses in this dynamic [auction] game where social competition exists is being processed in the striatum and leading to increased overbidding in future trials,” says Delgado.

To bolster this theory, Delgado and his colleagues decided to test it with a more traditional behavioral experiment (not reliant on brain imaging). In this new experiment, volunteers competed in two auction settings. For the first, they were told that the winner—the one who accumulated the most money in a series of bids—would receive a bonus of $15. In the second, all players would be given an extra $15 at the outset, but only the auction winner would keep that money. In monetary terms, of course, both settings were identical; the only difference was in the psychological “framing” designed to evoke the hope of winning the bonus versus the fear of losing it.

Consistent with their hypothesis, Delgado and his colleagues found that the “lose-the-bonus” group bid significantly higher than the “win-the-bonus” group or the control group (which had no bonus).

“There is a large literature on theories of what sorts of rules changes should or should not matter,” says Camerer. “None of these theories would predict that adding an upfront payment they can lose would make any difference, but it does.”

“This is the first type of experiment to use the fMRI evidence to then change the rules of an economic game and see if the rule change matters,” he adds.

The results add to a fast-growing body of evidence that suggests that human economic behavior is not simply determined by monetary goals and calculations but can be influenced, for example, by social goals or by emotions triggered in a social context.

“In general, the potential influence of social factors is discounted in traditional economic models,” says Delgado.

“Neuroeconomics is trying to improve the model using evidence of how the brain works,” says Camerer. “It is still a tiny trend in economics; only about 10 economists are actively involved in producing these data. But it has a lot of potential.”