To an evolutionary biologist, such pan-specificity combined with continuous variation strongly suggests that one is dealing with an evolutionary adaptation.
For another example, people in the US largely speak English, with a subset that speak Spanish, and a few other languages represented in scattered groups. That does not mean we should talk about English as an adaptive product of evolution. Language, definitely - there's clearly a heritable biological element to that ability. Similarly, religion may easily be a consequence of a universal trait like curiosity (we want answers to questions, religion provides them, so it spreads - even if the answers are all wrong) or empathy (we are social animals, we like community activities, religion hijacks that communal urge), but religion itself is but one replaceable instance, an epiphenomenon that too many people mistake for the actual substrate of the behavior.
Alan has commented on this blog before, and I gather that he tends to accept the validity of group level evolutionary biological processes in humans to a greater extent than I, e.g.; W.D. Hamilton's notorious use of kin selection in the 1970s to model human cultural procceses. That being said, I would add one point that Myers did not:
continuous traits are invariably those whose selective implications are weak
. That is, if a continuous trait is under directional selection the variation will quickly be exhausted as the loci which control heritability will "fix" to one allelic variant. This is why I am skeptical of direct selection for religiosity qua religiosity, the fact that religious sentiment seems to exhibit a range, and that environmental modulation can quickly generate a non-trivial number of atheists, implies that this just isn't a trait which has been strongly selected for (contrast this with heterosexuality in human males, removing the social ban against homosexuality will not, I suspect, result in a dramatic increase in the percentage of homosexuals because the bio-behavorial constraint against male-male sexual relations is rather powerful). But, I have issues with Myers' response as well. He seems to be making an analogy between religions and languages, suggesting that just as we abandon English for French given appropriate social contexts, so we can abandon religion for no religion, given the appropriate social contexts. Though this has some validity, as a normally distributed trait conventionally has a large component of environmental variance (as I believe religion does), another analogy between language and religion is that specific languages and religions are culturally inherited or inculcated, but the propensity for language and religion is genetically rooted. There is a weakness with the analogy insofar as language competency is one of the strongest arguments that one can make for a tightly integrated mental module with a deep biological root. In other words, language is a genetic trait, "wild type" humans all have an equal facility with it at its basic level.^1 In contrast, religiosity seems to be a heritable trait, there is variation within the population, some of it (perhaps as much as half) controlled by genetic variation. How can religion be genetically coded, in part, if there was no directional selection? Myers offers the answer when he suggests it is a byproduct of other traits, such as empathy. It seems likely that as highly social animals we have been strongly selected for empathy, though still to various degrees (and there may be mixed strategies within the population, some biologically rooted). Traits which exhibit a normal distribution are generated by a process where innumerable independent genetic loci act as "random variables" which contribute some of the effect to the trait. If religion emerges as a byproduct out of the milieu of human traits such as empathy, agency detection, curiosity, etc., then the variation within these traits naturally will lead to a variation in the realized range of any byproduct traits. Finally, there is also the implication that religious belief may be a natural emergent property of the human mind so long as these various characteristics remain part of the human cognitive toolkit. This is why I am skeptical that religious belief can be eliminated through resocialization, humans vary in their propensity toward religious belief and at some point along the range of propensities within the population the social-environmental pressures need to be very powerful to tilt the field against avowed religious belief. The totalitarian Communist states come to mind. But, in places like the former Soviet Union religion made an immediate come back after the removal of positive inducements toward atheism. Large minorities, such as in parts of Europe, can emerge who are areligious or atheist, but in free societies it seems that the majority of humans will default toward a form of supernaturalism. Addendum: Questions about the 'adaptive' value of religion also depend on the level of analysis. If supernaturalism is a human modal state it seems likely culture will reshape it in the aggregate into organized religions of some sort, and between these religions selective processes may work. But, that does not imply that on the psychological level religion was initiated by selection for belief in God, as such. In other words, religion is not an adaptation, but Christianity, Buddhism, etc., may be exaptations. 1 - Of course, though basal linguistic capacity is universal, there remains a heritable range on top of this in regards to the ability to use words with in a clever manner.