Residents of Second Life—an online computer game in which players can do almost everything they can do in real life, such as buy and sell property, take classes and date—tout their world’s realistic settings and social opportunities. Now a growing number of scientists are beginning to take notice and are bringing their human behavior research into the virtual world.

Second Life allows researchers to study scenarios that they cannot in real life, such as placing a person in someone else’s body, changing the laws of physics or even performing experiments that are otherwise ethically taboo. Communications scientist Nick Yee of the Palo Alto Research Center, who uses Second Life as his primary laboratory, says that the setting could provide new ways to explore people’s feelings about age, sex or race. Another group of researchers at University College London recently repeated Stanley Milgram’s notorious 1963 experiment—in which participants were asked to administer apparently lethal electric shocks to another volunteer—in a virtual-reality setting. The results were similar to those of the original experiment; although the participants became uncomfortable, many continued administering shocks at the request of the researchers. Computer scientist Mel Slater, who led the experiment, says that virtual reality is more realistic than Second Life but agrees that, like virtual reality, the game has the potential to be a powerful research tool.

Dmitri Williams, a communications professor at the University of Southern California, says that online games such as Second Life also offer an unprecedented chance to gather large amounts of accurate behavioral data. “In these worlds,” Williams explains, “you have the equivalent of cameras recording people’s every move.”

Some experts, however, caution that it is too early to say for sure whether experiments done in virtual worlds can be applied to real behavior. A recent study from Yee’s group demonstrated that many people respond to social cues such as personal space and eye contact much as they would in real life. But in other cases, such as risk-taking behavior, people behave very differently in games, because the cost of death is relatively insignificant. “We need to find out which situations do match up [with reality] and which don’t,” Williams says. “We’re not even close to that yet.”