Our ancestors did not always have religion, yet many religious practices seem to have very ancient origins. So when did religion first evolve? Archaeologists have long been fascinated by this question. One indication is burial. Some experts believe this began as far back as 200,000 years ago with the Neanderthals, but the motivation for such cacheing of bodies is ambiguous. So most archaeologists more cautiously define the appearance of religion by looking for evidence of grave goods in burials, since these at least unequivocally imply belief in an afterlife. Deliberate burials of this kind do not occur much before 25,000 years ago. Such burials imply a sophisticated theology, so we can safely assume that these were preceded by a long phase of less sophisticated religious belief. But without evidence on the ground, can we see any further back than this?
I have suggested that there is another way to get an unexpected insight into the question. It comes from asking what kind of mind is required to hold religious beliefs. Take the statement: "I believe that god wants..." To grasp this an individual needs theory of mind - the capacity to understand that another individual (in this case, god) has a mind of his own. Philosophers call this "second-order intentionality" because such statements contain two notions of intent: I believe and god wants. But we need more than this to build a religion.
Third-order intentionality allows me to state: I believe that god wants us to act with righteous intent. At this level, I have personal religion. But if I am to persuade you to join me in this view, I have to add your mind state: I want you to believe that god wants us to act righteously. That's fourth-order intentionality, and it gives us social religion. Even now, you can accept the truth of my statement and still it commits you to nothing. But add a fifth level (I want you to know that we both believe that god wants us to act righteously) and now, if you accept the validity of my claim, you also implicitly accept that you believe it too. Now we have what I call communal religion: together, we can invoke a spiritual force that obliges, perhaps even forces, us to behave in a certain way.
So, communal religion requires fifth-order intentionality, and this also happens to be the limit of most people's capacity as indicated by research done by myself and my colleagues. I think this is no coincidence. The majority of human activities can probably be dealt with using second or third-order intentionality. The two extra layers beyond this undoubtedly come at some considerable neural expense. Since evolution is frugal, there must be some good reason why we have them. The only plausible answer, so far as I can see, is religion. And that's where this line of reasoning can throw light on the origins of religious belief.
As far as we know, all other animals are locked into first-order intentionality, with the exception of great apes who are just about able to cope with second order. If you look at the brains of humans and other animals you find that the level of intentionality they can achieve scales linearly with the volume of grey matter in their frontal lobes (a particularly important part of the brain's processing units). This can be used to work out the level of intentionality our extinct ancestors were capable of - provided you have a fossil skull from which you can measure the overall volume of the brain.
Plotting these values onto a graph, the evidence suggests that as early as 2 million years ago, Homo erectus would have aspired to third-order intentionality, perhaps allowing them to have personal beliefs about the world. Fourth-order intentionality - equating to social religion - appeared with archaic humans around 500,000 years ago. And fifth order didn't appear much before the evolution of anatomically modern humans around 200,000 years ago - early enough to ensure that all living humans share this trait, but late enough to suggest that it was probably a unique adaptation.
In a separate strand of research, my colleagues and I have also found a relationship between the size of the brain's neocortex and social group size in primates. Interestingly, this "social brain hypothesis" predicts that around the time our ancestors evolved the capacity for fifth-order intentionality their community sizes would have exceeded about 120 individuals. Religion may have evolved to provide the mechanism for bonding them into a coherent social unit.