People are extraordinarily skilled at spotting cheats—much better than they are at detecting rule-breaking that does not involve cheating. A study showing that just how good we are at this adds weight to the theory that our exceptional brainpower arose through evolutionary pressures to acquire specific cognitive skills.

The still-controversial idea that humans have specialized decision systems in addition to generalized reasoning ability has been around for decades. Its advocates point out that the ability to identify untrustworthy people should be favored evolutionally since cheats risk undermining the social interactions in which people trade goods or services for mutual benefit.

To test whether we have a special ability to reason about cheating, Leda Cosmides, an evolutionary psychological test called the Wason selection test, which tests volunteers' ability to reason about "if/then" statements.

The researchers set up scenarios in which they asked undergraduate volunteers to imagine they were supervising workers sorting applications for admission to two schools; a good one in a district where school taxes are high, and a poor one on an equally wealthy, but lightly taxed district. The hypothetical workers were supposed to follow a rule that specified "if a student is admitted to the good school, they must live in the highly taxed district".

Half the time, the test subjects are told that the workers had children of their own applying to the schools, thus having a motive to cheat; the rest of the time they were told the workers were merely absent-minded and sometimes made innocent errors. Then the test subjects were asked how they would verify that the workers were not breaking the rule.

Cosmides found that when the "supervisors" thought they were checking for innocent errors, just 9 of 33, or 27 percent, got the right answer—looking for a student admitted to the good school who did not live in the highly-taxed district. In contrast, when the supervisors thought they were watching for cheats, they did much better with 23 of 34, or 68 percent getting the right answer.

This suggests that people are, indeed more adept at spotting cheat than at detecting mere rule-breaking. Cosmides says, "any cues that it's just an innocent mistake actually inactivate the detection mechanism."

The result is what you would expect if natural selection had favored this specific ability in early, pro-social humans—and is not at all what would happen under selection for generalized intelligence, Cosmides says. "My claim is that there is nothing domain-general in the mind, just that that can't be the only thing going on in the mind."

Other psychologists remain skeptical of this conclusion. "If you want to conclude that therefore there's a module in the mind for detecting cheater, I see zero evidence for that," says Steven Sloman, a cognitive scientist at Brown University in Providence, Rhode Island. "It's certainly possible that it's something we learned through experience. There is no evidence that it's anything innate."