Why would (otherwise intelligent) scholars believe in
I do not know if many scholars of religion still believe in gods or spirits, but I know that a great many of them believe in the existence of religion itself – that is, believe that the term "religion" is a useful category, that there is such a thing as religion out there in the world, that the project of "explaining religion" is a valid scientific project. Naturally, many of the scholars in question will also say that religion is a many splendored thing, that there are vast differences among the varieties of religious belief and behavior. Yet they assume that, underlying the diversity, there is enough of a common set of phenomena that a "theory of religion" is needed if not already available.
One might think this unfortunate and obdurate tendency to believe in the scholarly equivalent of unicorns is chiefly confined to theologians or other marginal scholars. That is not the case. Indeed, quite a lot of people these days argue for a "scientific explanation of religion". In preparation for this they gather the best and most up-to-date scientific gear, from genetics and evolutionary biology to, inevitably, neuro-imaging.
I applaud the use of such tools in general and deplore it all the more in this particularly futile pursuit.
Fang epic recitation – a matter of "religion"?
There really is no such thing as "religion". Most people who live in modern societies think that there is such a thing out there as "religion", meaning a kind of social and cognitive package that includes views about supernatural agency (gods and suchlike), notions of morality, particular rituals and sometimes particular experiences, as well as membership in a particular community of believers and the constitution of specific organizations (castes of prests, churches, etc.). All this, as I said, is thought to be a "package", where each element makes sense in relation to the others, given a coherent and explicit doctrine. Indeed, this is the way most major "religions" – Islam, Hinduism for instance – are presented to us, the way their institutional personnel, many scholars and most believers think about them.
But all this is a recent invention. Most of human evolution took place in small-scale communities that did not have any religious institutions. This was also the case of most human groups outside modern economic development until recently, and it is still the case in remote places outside the direct influence of modern states. In all these places, there is no unified domain of "religion". True, there may be various ideas about superhuman agents, there may be ideas about morality (often not connected with those agents), there may be notions about ritualized sequences that must be performed (some with and many without a connnection to spirits etc.), there may be community affiliation (generally unrelated to morality or superhuman agency), but there is nothing that would justify putting all these things together.
"Religion" is the recent invention of special organizations that flourished in early states, typically in literate societies. These institutions grouped ritual specialists who collectively tried to set up a corporate monopoly on the provision of particular services – and gradually associated stable doctrine, ritual standardization, excusivity of services and other aspects of corporate branding.
That is why scientific explanations of what happens in religions and what is usually described as "religious" end up deflating the term, and showing that there is nothing sui generis to these kinds of thoughts, norms of practices. For instance, many well-intentioned people have told us that "religion" creates costly signaling, but that is of course found in many other contexts of human communication, and is often not found in "religion". The same goes for other arguments, e.g. that "religion" requires suspension of disbelief, that it creates social cohesion (or social fission), that it creates a unique kin of experience, etc. All these features are sometimes found in association with superhuman agent beliefs, and often without such beliefs, such that the category "religious" explains little if anything.
Note that this is emphatically not a matter of "definition". Many people in the business of explaining religion would say that we need a better, broader, more empirical, whatever, definition of religion. But this seems largely misguided. The problem is one of ontology, not terminology. Square triangles and unicorns may be very clearly defined – they simply do not exist.
In many human languages and cultures there is no local term for "religion", and where there is one it is invariably a term created or hijacked by specific literate organizations, as the label for their specialty. What does it mean, when a particular language has no equivalent for something other people find obvious?
It may point to either one of two diametrical situations. Consider the contrast between "syntax" or "economics" on the one hand, and "sport" or "divorce" on the other. Even though most human languages had no term for "syntax" until recently, they all had a syntax. Even though the notion of economics is a scientific category, all human societies have economic processes. But in a place where there is no term that could be translated as "soccer" or "divorce", it is quite certain that no-one plays sports or gets divorced.
Studying "religion" in the spirit- or god- or ancestor-related stuff that is found before and outside religious institutions is like studying "sport" in a place where there is no such concept. One can certainly find that the people in that place sometimes do strenuous physical activities, that at other times they compete in achieving difficult tasks, that they laughingly throw objects at each other, that they often support their lineage against others… But there is no single time and place where people compete in playful strenuous and difficult physical tasks and others watch and support one of the competing sides. "Sport" is one of these institutions that some human groups have and others do without.
Obviously, the same goes for "religion". But if there is no such thing, why go on about it? Why do (otherwise intelligent) scholars want to use that term?
I see only three possible reasons. A rather sinister one, from a scientific viewpoint, is ideological motivation. Claiming that "religion" is not just a specific institution of particular times and places, but a stable feature of humankind, may be a normative statement, implying that all (or all fully developed) human polities should have religious institutions, that the state of most tribal societies in that regard is somehow abnormal or primitive. This is not just a fantasy of mine. For religious institutions, it is obviously important to claim that there is such a thing as "religion" ( which these institutions are in the business of explaining, managing, etc.). If you don't accept that, there is no reason for those institutions' social role or indeed existence. This may explain why the discipline of "religious studies" (in places where it exists) is invariably penetrated by religious apology, no matter how heroic the efforts of serious academics to defend genuine scholarship against that insidious invasion (see books by Don Wiebe).  A more depressing explanation is that people who talk about "religion" simply have not done their homework and studied their anthropology. This would apply to most journalists' use of the term. I have many a painful recollection of trying to explain to journalists, e.g. that it makes no sense to ask whether Neandertals had "religion", that "animists" as counted in world surveys are bnot members of a "religion" – all to no avail.  Or it could be a convenient way of pointing to what you are studying, without really being committed to the existence of that particular unicorn. Which leads me to a recent exchange on the ICCI website. A while ago, Harvey Whitehouse worte a post on the many varieties of stuff commonly found under the label "religion". Maurice Bloch posted a spirited reply, along lines rather similar to what I am arguing here, although with more asperity. As Dan Sperber and Emma Cohen commented, Whitehouse and Bloch were talking at cross-purposes.
Well, then, shto delat' ? What is to be done?
It makes some pragmatic sense, and it is almost forgiveable, if you are in the business of attracting grants or selling books, to talk about "the brain and religion", "the evolution of religion", "how religion works", "explaining religion", or even, if you are really desperate, of "religion explained".
All this is harmless if your scholarship then proceeds to deflate the notion and explain why your empirical studies have to focus on genuine natural kinds, like costly signaling, counter-intuitive concepts, monopolistic specialists guilds, coalitional psychology, imagined agents, etc.
Our situation is difficult in that there is a great amount of social demand for naturalistic explanations of "religion", all the more so in a world made more dangerous by religious fanatics. Obviously, meeting that demand does not imply that we believe in "religion". But simply deflating the misleading concept seems dangerously close to "having nothing to say about religion". People who are worried about the dangers of modern zealotry may tend to find the statement that "there is no such thing as religion" rather academic. So we have to engage in a particularly delicate rhetorical exercise, showing that cognitive science and evolution have a lot to say about what people usually call "religion", and gently leading people to the realization that "religion", like aether and phlogiston, belongs in the ash-heap of scientific history.