About 12,000 years ago human societies went big; tribes and villages grew into vast cities, kingdoms and empires within just a few millennia. For such large and complex societies to take root, people needed to maintain social cohesion and cooperation, even among complete strangers. What enabled this, many researchers have argued, was religion.

Such a religion, the idea goes, would work particularly well if it established standards of morality and behavior—and enforced them with the threat of supernatural punishment. This may involve so-called big gods who care about who is naughty or nice, like in the Abrahamic religions. Or, as in the Buddhist concept of karma, religions can enforce morality through what is dubbed “broad supernatural punishment”—spontaneous consequences that occur without the intercession of conventional big gods.

But a new study, published Wednesday in Nature, casts doubt on the role these kinds of “pro-social” religion play in enabling large-scale societies. “It’s not the main driver of social complexity as some theories had predicted,” says Harvey Whitehouse, an anthropologist at the University of Oxford and one of the lead authors of the study.

Instead the study suggests pro-social religions appeared after complex societies had already emerged. Although these religions may have helped sustain and grow large societies, the analysis makes the case that they were not necessary for societies to expand in the first place.

Such pro-social religions may dominate today's world—but they did not do so throughout much of humanity’s past, when people typically tried to appease temperamental gods via prayer and sacrifice rather than trying to be good. Pro-social religions were “really not common," says Edward Slingerland, a historian and religious scholar at the University of British Columbia who was not involved in the new study. “Yet when they do arise, they spread.”

The pro-social religions hypothesis—which the new study’s authors also call the “moralizing gods” hypothesis—holds that these religions spread because they kick-started the big societies that eventually came to dominate large parts of the world. People in small tribes and villages were often biologically related or otherwise well acquainted, so it was relatively easy for residents to hold one another accountable when it came to cooperation and doing reciprocal favors. But big societies require cooperation between strangers who might interact only once, Slingerland says. So a team of researchers including Slingerland and Ara Norenzayan, a psychologist at U.B.C., had previously proposed that pro-social religions (pdf)—ones with big gods, in particular—might have been needed to foster enough cooperation to get complex societies off the ground. Experiments have found people psychologically primed to think about pro-social religions are more cooperative. Historical studies have also suggested the rise of such religions coincided with increasing social complexity.

But other evidence has been mixed, finding differences between the two types of pro-social religions. For example, a 2015 study of Austronesian societies found big gods emerged after the rise of complex societies—although the concept of broad supernatural punishment had appeared before. A 2017 study of Viking-era Scandinavia found similar results.

These previous studies focused on specific geographic regions—but the new one covers the globe, relying on a database called Seshat and spanning 414 societies from 30 geographical regions across 10,000 years of history. To build this database, researchers went through published academic literature covering various societies. They tabulated the extent to which a society believed in pro-social religions—along with various indicators of social complexity including population, territorial size and the presence of legal codes and courts.

In the new study the authors coded this information as numerical ratings, and used statistical techniques to calculate a number reflecting each society's complexity level. When the researchers compared the growth of social complexity with the times that pro-social religions appeared, they found that in the vast majority of societies such religions emerged later—after societies swelled to a population of about a million. “This paper has done substantial damage to the big gods hypothesis, which fits in with what we had previously found,” says Russell Gray, an evolutionist at the Max Planck Institute for the Science of Human History, who was part of the Austronesian study but not the new one. The new results also weaken the idea that broad supernatural punishment is needed for complex societies, he says. “This is by far the best thing I’ve seen out of the much-hyped Seshat project so far.”

But Slingerland is skeptical. He points out that many of the entries in the Seshat database do not list any expert consultation. “That just worries me,” he says. Many elements in the database—such as belief in pro-social religions—are subject to interpretation and debate, even among experts who have spent careers studying a particular society and time period, he notes. Without experts to check every data entry, he adds, the database could be underestimating uncertainty or mischaracterizing a society. “I’m not saying the data is all wrong,” he says. “It’s just that we don’t know—and that, in a way, is just as bad because not knowing means you can’t take seriously the analysis.”

The authors of the Nature paper did consult with dozens of experts—but it would have been impossible to recruit scholars to vet all of the 47,613 records they analyzed, says Patrick Savage, an anthropologist and statistician at Keio University in Japan, who led the new study along with Whitehouse and Pieter Francois, an anthropologist at Oxford. Even without expert verification of every entry, Savage says the team remains confident in the quality of its data and analysis.

Where there were uncertainties or disagreements, Savage adds, the researchers re-ran the statistical analysis several times with multiple possible values of the data. Each time they still found pro-social religions emerged after complex societies. The conclusions were so strong and consistent that any discrepancies or errors in the database would not have been significant enough to make much of a difference, he says.

If the study stands up to scrutiny, it does not mean moralizing gods had no bearing on complex societies; they could have helped enforce cohesion and cooperation as these societies conquered others and encompassed a more diverse populace. “It’s not all bad news for the moralizing gods hypothesis,” Whitehouse says.

Other aspects of religion could also have helped enable the rise of social complexity. For example, Whitehouse speculates the standardization of religious rituals could have been crucial. According to his team's analysis, archaeological evidence for rituals often corresponds to the times societies got big, before pro-social religions appeared. “When it comes to the initial rise of social complexity,” the study’s authors wrote, “how you worship may ultimately have been more important than who you worship.”