The debate over why belief evolved is between byproduct theorists and adaptationists. You might think that the byproduct theorists would tend to be nonbelievers, looking for a way to explain religion as a fluke, while the adaptationists would be more likely to be believers who can intuit the emotional, spiritual and community advantages that accompany faith. Or you might think they would all be atheists, because what believer would want to subject his own devotion to rationalism's cold, hard scrutiny? But a scientist's personal religious view does not always predict which side he will take. And this is just one sign of how complex and surprising this debate has become.

Angels, demons, spirits, wizards, gods and witches have peppered folk religions since mankind first started telling stories. Charles Darwin noted this in "The Descent of Man." "A belief in all-pervading spiritual agencies," he wrote, "seems to be universal." According to anthropologists, religions that share certain supernatural features -- belief in a noncorporeal God or gods, belief in the afterlife, belief in the ability of prayer or ritual to change the course of human events -- are found in virtually every culture on earth.

This is certainly true in the United States. About 6 in 10 Americans, according to a 2005 Harris Poll, believe in the devil and hell, and about 7 in 10 believe in angels, heaven and the existence of miracles and of life after death. A 2006 survey at Baylor University found that 92 percent of respondents believe in a personal God -- that is, a God with a distinct set of character traits ranging from "distant" to "benevolent."

When a trait is universal, evolutionary biologists look for a genetic explanation and wonder how that gene or genes might enhance survival or reproductive success. In many ways, it's an exercise in post-hoc hypothesizing: what would have been the advantage, when the human species first evolved, for an individual who happened to have a mutation that led to, say, a smaller jaw, a bigger forehead, a better thumb? How about certain behavioral traits, like a tendency for risk-taking or for kindness?

Atran saw such questions as a puzzle when applied to religion. So many aspects of religious belief involve misattribution and misunderstanding of the real world. Wouldn't this be a liability in the survival-of-the-fittest competition? To Atran, religious belief requires taking "what is materially false to be true" and "what is materially true to be false." One example of this is the belief that even after someone dies and the body demonstrably disintegrates, that person will still exist, will still be able to laugh and cry, to feel pain and joy. This confusion "does not appear to be a reasonable evolutionary strategy," Atran wrote in "In Gods We Trust: The Evolutionary Landscape of Religion" in 2002. "Imagine another animal that took injury for health or big for small or fast for slow or dead for alive. It's unlikely that such a species could survive." He began to look for a sideways explanation: if religious belief was not adaptive, perhaps it was associated with something else that was.

"Fear of death is an undercurrent of belief. The spirits of dead ancestors, ghosts, immortal deities, heaven and hell, the everlasting soul: the notion of spiritual existence after death is at the heart of almost every religion. According to some adaptationists, this is part of religion's role, to help humans deal with the grim certainty of death. Believing in God and the afterlife, they say, is how we make sense of the brevity of our time on earth, how we give meaning to this brutish and short existence. Religion can offer solace to the bereaved and comfort to the frightened."

ROBIN MARANTZ HENIG, The New York Times Magazine