John Wilkins points me to a piece by Pascal Boyer,* Being human: Religion: Bound to believe?:
So is religion an adaptation or a by-product of our evolution? Perhaps one day we will find compelling evidence that a capacity for religious thoughts, rather than 'religion' in the modern form of socio-political institutions, contributed to fitness in ancestral times. For the time being, the data support a more modest conclusion: religious thoughts seem to be an emergent property of our standard cognitive capacities.
If there is one thing you can say for philosophy, the discipline imposes a clarity and precision in terminology. The philosophical fixation on method might mean that progress, however you measure it, is slow. But, it will also minimize confusion to the greatest extent possible (though I would offer that the material which modern philosophy often focuses on is well nigh intractable without presuppositions which are not universal). Other fields do not focus so much on methodology, and this causes problems. Boyer, as a cognitive anthropologist, has a "mind-centric" perspective. His discussion of religion necessarily begins at the first building block, the atomic unit if you will, of religious phenomena. His work regularly acknowledges the ubiquity of institutional religion, but its focus is on the cognitive foundations of religious cognition. I believe this causes a problem for the modern reader, especially intellectual types who might read works which attempt to scientifically analyze human phenomena which a "normal" person would never consider examining in such a manner. Many who are interested in this topic may very well be the types for whom religion as a mental experience is most alien. For them, religion is first and foremost an institutional phenomenon girded by a set of belief axioms. Cognitive anthropologists make a point emphasizing the implicit subconscious element of cognition (arguably, most cognition is subconscious), but after a lifetime of conceiving of religion as a formal institutional structure which promulgates a set of beliefs, a minimalist and thin psychological model is I think relatively difficult to internalize for many. This is why I think assertions such as "religion will always be with us" should be taken with a grain of salt. Religion understood as a set of intuitions which predisposes humans to naturally accept the existence of supernatural entities seems to be an emergent feature of human cognition. It sits at the intersection of normal human mental skills such as Theory of Mind and Social Intelligence. Religion understood as the institutional and belief-oriented framework which takes supernatural phenomena as a presupposition is a different creature altogether, and may not be so inevitable. This is where a little cross-cultural history is worthwhile. In the West there is a rough model with proceeds from diffuse paganism, to the more crystallized forms of Christianity, to a world which has experienced the Enlightenment where the monopoly of the One True Religion has been broken. But there are other paths not taken. Consider East Asia, where institutional religion has traditionally been a much weaker force. In China, South Korea and Japan the commanding heights of political power had only a weak association with any religious tradition, and so rather than monopoly there was a default pluralism. The exception to this when religions are perceived to suborn public order or began to rival the political center in power (in other words, religious persecutions in East Asia have had explicitly political prior conditions, not theological ones). With the rise of Christianity in places like Korea a more explicit confessionalization has now become normative, but even to this day the power of institutional religion remains weak. All this is not to say that East Asia is an abode of materialist rationalism. On the one hand, its elite culture has often leaned in this direction, but astrology, geomancy and other occult practices are widespread. In other words, Boyers' model holds insofar as despite the lack of close affiliation to institutional religions a generalized supernaturalism remains common. The key is that supernaturalism is not posterior to religion, it is prior. Recurrent aspects of religion such as ritual, community-building, doctrine and millenarianism are often found in political movements such as Marxism. But after all this is stripped away the core supernatural bias remains. Pascal Boyer to a great extent is actually talking about the inevitable persistence of supernaturalism, not religion. * Boyers' Religion Explained is highly readable.
