Nothing is ever lost: belief systems are built on previous ones, but the earliest were not religions in our sense of the term
One of Robert Bellah's central ideas is that "nothing is ever lost". We are built like the cities of Troy on our previous selves. Every night we sleep uneasily on their rubble. Religious and ethical systems are also built on their predecessors, and this is true both socially and psychologically. Even the most abstract theologian will refresh her mind in ritual and, if she doesn't, will soon cease to believe.
This isn't because ritual dulls the mind and kill doubts, but because belief is an emotional fact, quite as much as disbelief can be. To hold that something is true and not to care in the least that it is so is not to believe it in any religious or important sense.
With this in mind, it's clear that, to look for the roots of religion, we have to look at the roots of language. This is another link to ritual, for Bellah believes that language and music come from the same roots. He talks about "musilanguage" as an early form of communication, in which the tones, the tunes and the rhythms convey a great part of the meaning. Musilanguage is still present in our lives today, of course. We can see it in everything that is lost when conversation is reduced to an exchange of plain text.
Both, he believes, arose first in play. Mammalian play is very important to him, for it is the first place in evolution where animals are free from deficiency consciousness. It is gratuitous behaviour. Play is undertaken for its own sake. It can be extremely serious: the opposite of "play" is not "serious", or even "real", but "work", which is something undertaken for ends outside itself. But play operates in its own time, and within its own rules, neither of which need have anything to do with the world outside.
It's worth noting in passing here that, although this is an "evolutionary" account of psychology, it is nothing like what's known as "evolutionary psychology". Bellah does not postulate a "play module", or explain that children who are good at games grow up to have more descendants than the nerds.
One other thing about play, which makes it important in the earliest forms of religion, and of language, is that it is essentially embodied. It starts with actions, and only later, if at all, moves on to ideas. It never entirely loses the connection to action. This is one way into the importance of ritual. There are certain things we know only because our bodies have done them, and some kinds of knowledge are inexplicable in words. Bellah is too high-minded to mention sex in this context, but anyone who was an unhappy teenage virgin can remember what such an absence of knowledge felt like. He talks instead about bicycle-riding, a less exciting example of a skill we can learn but never express in words.
Religious practice is full of gestures, of course: hands are clasped in prayer, or the whole body kneels, or lies flat in token of submission. These things are also performed for an audience. Solitary prayer is a very late and highly developed form of ritual. It starts as something communal. Music, dancing and storytelling are all embodied reactions to the world and, when you join in, they become a kind of explanation, too.
Nothing is ever lost. A theory is not a better version of a story, nor even a story cleansed of worthless decoration. I know there are people who believe it is. The progress from myth to theory is one of the central rationalist stories. But this is itself a myth more than a factual, historical account. It gains its power and coherence from a belief in progress. If that fails, there is nothing but confusion left. So what are the earliest religions we can find? The first thing they were not is "religions". They didn't have theologies. They didn't have priesthoods or scriptures. They didn't have explicit theories of how the world is. They may not even have had creation myths, in the sense that, without an order of time, creation might have happened yesterday. Living after Christianity as we do, we assume that creation myths have to explain why the world is not perfect. This may not have seemed to our ancestors a question worth asking.
What they did have was fierce egalitarianism. The first questions that ritual answered were not concerned with where we come from, but how to live as a community, which survives by sharing and co-operation. Literate religions have often been used to justify inequality. But the earliest religions of which we know, or which we can reconstruct, made all their participants equal in rituals. There was no one in them like a bishop, nor anyone like a king. The long and twisting road from there to the first dawn of universalist ethics is one that Bellah is trying to map. It passes through kings, and priest-kings who were even worse than bishops.
Brown, Andrew. 2012. "Religion in Human Evolution, part 2: faith, language, music and play". The Guardian. Posted: July 23, 2012. Available online: http://www.guardian.co.uk/commentisfree/2012/jul/23/robert-bellah-religion-human-evolution