Anthropologist Pascal Boyer has an essay in the 23 October 2008 issue of Nature (Being human: Religion: Bound to believe?) in which he explores the cognitive traits that sustain religious identification and belief.
We should not try to pinpoint the unique origin of religious belief, because there is no unique domain for religion in human minds. Different cognitive systems handle representations of supernatural agents, of ritualized behaviours, of group commitment and so on, just as colour and shape are handled by different parts of the visual system. In other words, what makes a god-concept convincing is not what makes a ritual intuitively compelling or what makes a moral norm self-evident. Most modern, organized religions present themselves as a package that integrates all these disparate elements (ritual, morality, metaphysics, social identity) into one consistent doctrine and practice. But this is pure advertising. These domains remain separated in human cognition. The evidence shows that the mind has no single belief network, but myriad distinct networks that contribute to making religious claims quite natural to many people.
The findings emerging from this cognitive-evolutionary approach challenge two central tenets of most established religions. First, the notion that their particular creed differs from all other (supposedly misguided) faiths; second, that it is only because of extraordinary events or the actual presence of supernatural agents that religious ideas have taken shape. On the contrary, we now know that all versions of religion are based on very similar tacit assumptions, and that all it takes to imagine supernatural agents are normal human minds processing information in the most natural way.
Improved understanding of the neuropsychology underlying religious belief will be helpful if rational, secular alternatives are ever to supplant religion as the predominant theme around which communities are organized.
A tip of the hat to Mind Hacks.