In most religions—and arguably anything worth being called a religion—God is not just an impersonal force or creator. He has a mind that humans can relate to. Maybe you’re not gossiping on the phone with him late at night, but he has personality traits, thoughts, moods, and ways of communicating with you. If you didn’t know what a mind was or how it worked, not only would you not understand people, you would not understand God, and you would not be religious.

That’s the theory, anyway. Scientists who study religion have come to agree that belief in God (or gods) relies on everyday social cognition: our ability—and propensity—to think about minds. (See chapters 6 and 7 of my book.) Which means if you are autistic, and unable to “mentalize,” you would be an atheist. New research published today in PLoS ONE provides fresh evidence for this claim.

But first, the existing evidence.

Jesse Bering, in a 2002 paper, noted that in autobiographical accounts written by people with high-functioning autism, God is more a principle than a person. He/it provides order but isn’t much concerned with human affairs—the idea of him satisfies the intellect rather than the emotions. Temple Grandin, for example, described God as the entanglement of millions of interacting particles.

In line with such a conception of the divine, Simon Baron-Cohen, who proposed the mindblindness theory of autism, told me that “sometimes I meet people with autism who are religious, but their motivation is driven more by the rules (the system) in theology rather than the anthropomorphizing.”

One outcome of the ability to mentalize is the ability to think teleologically—to see the purpose of objects or events. (Rocks and rainstorms have no purpose, but shovels and showering do.) I found one blog post by a woman with Asperger’s syndrome who wrote that as a child, “The world I perceived was a random, self-sufficient system. It wasn’t built; it grew. (When I was little, I thought houses and roads were some kind of large plant that grew out of the ground; if you had told me people made them I would’ve been thunderstruck).” She didn’t get that some things were created for a reason.

When people see an event as divine intervention, or a result of intelligent design, they’re just letting their teleological bias run amok. They’re attributing purpose where there is none. Bethany Heywood, in collaboration with Jesse Bering, found in her Ph.D. research that even atheists tend to say that certain things happened to them “for a reason,” e.g., to teach them a lesson. But subjects with Asperger’s gave significantly fewer teleological responses than a control group did, and several even expressed confusion regarding the questions about purpose. One, misinterpreting a prompt for “a coincidence you saw meaning in,” wrote, “in practical application, I wear nice clothes and make my hair presentable. Coincidentally people are more friendly towards me.”

The strongest connection between atheism and autism before now was a paper presented at a conference last year by Catherine Caldwell-Harris and collaborators at Boston University. Survey respondents with high-functioning autism were more likely than control subjects to be atheists and less likely to belong to an organized religion. (They were also more likely to have religious ideas of their own construction, perhaps something similar to Temple Grandin’s.) And atheists were higher on the autistic spectrum than Christians and Jews. But the researchers were not able to demonstrate that mentalizing deficits were responsible for the connection.

That’s where the new paper comes in. Ara Norenzayan and Will Gervais of the University of British Columbia and Kali Trzesniewski of UC Davis report on four studies. The first study replicates the finding of the BU research: 12 autistic and 13 neurotypical adolescents took part, and the neurotypical subjects were ten times as likely to strongly endorse God.

The other three studies went further. They included hundreds of participants from a variety of demographics in the US and Canada and used various measures of belief in God and of mentalizing abilities. The results of all three followed the same pattern.

First, people with higher scores on the Autism Spectrum Quotient (items included “I am fascinated by numbers,” and “I find social situations [difficult]”) had weaker belief in a personal God. Second, reduced ability to mentalize mediated this correlation. (Mentalizing was measured with the Empathy Quotient, which assesses self-reported ability to recognize and react to others’ emotions, and with a task that requires identifying what’s being expressed in pictures of eyes. Systematizing—interest in and aptitude for mechanical and abstract systems—was correlated with autism but was not a mediator.) Third, men were much less likely than women to say they strongly believed in a personal God (even controlling for autism), and this correlation was also mediated by reduced mentalizing.

“It’s hard to have an experience of God in your life unless you think of him as a person, with mental states, who you can pray to, who will answer your prayers, who cares about you,” Norenzayan told me a couple years ago, when he was conducting this research.

He and his collaborators point out that mentalizing deficits are of course not the only path to atheism. There are also cultural and educational influences—exposure to other skeptics, say—and cognitive style—some people are more likely to use rationality to second-guess superstition.

But the finding that religion and mentalizing are so tightly bound emphasizes an argument I make in my book: magical thinking is just a natural extension of the everyday thinking that makes us (neurotypically) human. As far as the brain is concerned, God is one of us.