The research, conducted by University of British Columbia psychologists Will Gervais and Ara Norenzayan, does not take sides in the debate between religion and atheism, but aims instead to illuminate one of the origins of belief and disbelief. "To understand religion in humans," Gervais says, "you need to accommodate for the fact that there are many millions of believers and nonbelievers."

One of their studies correlated measures of religious belief with people's scores on a popular test of analytic thinking. The test poses three deceptively simple math problems. One asks: "If it takes five machines five minutes to make five widgets, how long would it take 100 machines to make 100 widgets?" The first answer that comes to mind—100 minutes—turns out to be wrong. People who take the time to reason out the correct answer (five minutes) are, by definition, more analytical—and these analytical types tend to score lower on the researchers' tests of religious belief.

But the researchers went beyond this interesting link, running four experiments showing that analytic thinking actually causes disbelief. In one experiment, they randomly assigned participants to either the analytic or control condition. They then showed them photos of either Rodin's The Thinker or, in the control condition, of the ancient Greek sculpture Discobolus, which depicts an athlete poised to throw a discus. (The Thinker was used because it is such an iconic image of deep reflection that, in a separate test with different participants, seeing the statue improved how well subjects reasoned through logical syllogisms.) After seeing the images, participants took a test measuring their belief in God on a scale of 0 to 100. Their scores on the test varied widely, with a standard deviation of about 35 in the control group. But it is the difference in the averages that tells the real story: In the control group, the average score for belief in God was 61.55, or somewhat above the scale's midpoint. On the other hand, for the group who had just seen The Thinker, the resulting average was only 41.42. Such a gap is large enough to indicate a mild believer is responding as a mild nonbeliever—all from being visually reminded of the human capacity to think.

Another experiment used a different method to show a similar effect. It exploited the tendency, previously identified by psychologists, of people to override their intuition when faced with the demands of reading a text in a hard-to-read typeface. Gervais and Norenzayan did this by giving two groups a test of participants' belief in supernatural agents like God and angels, varying only the font in which the test was printed. People who took the belief test in the unclear font (a typewriterlike font set in italics) expressed less belief than those who took it in a more common, easy-to-read typeface. "It's such a subtle manipulation," Norenzayan says. "Yet something that seemingly trivial can lead to a change that people consider important in their religious belief system." On a belief scale of 3 to 21, participants in the analytic condition scored an average of almost two points lower than those in the control group.

Analytic thinking undermines belief because, as cognitive psychologists have shown, it can override intuition. And we know from past research that religious beliefs—such as the idea that objects and events don't simply exist but have a purpose—are rooted in intuition. "Analytic processing inhibits these intuitions, which in turn discourages religious belief," Norenzayan explains.

Harvard University psychologist Joshua Greene, who last year published a paper on the same subject with colleagues Amitai Shenhav and David Rand, praises this work for its rigorous methodology. "Any one of their experiments can be reinterpreted, but when you've got [multiple] different kinds of evidence pointing in the same direction, it's very impressive."