And yet, I don't think the selfish gene hypothesis captures the entirety of the idea of the unit of selection. I subscribe to multi-level selection theory, which has been spearheaded over the last few decades almost single-handedly by David Sloan Wilson (from whom I actually learned the stuff). As its name suggests, this broader theory argues that natural selection can take place at the level of genes, individual organisms or even groups, provided certain conditions are met.
Dennett seems to favor a meme-based approach; Dawkins likes to think that it's either a programming bug or simply a byproduct of some other adaptation; and others think that religion is a legitimate adaptation on its own right. One of the most interesting (and possibly correct) hypotheses about the evolution of religion, you will not be surprised to hear, is Wilson's group selection theory: while competition is not always best within groups (because free-riders will exploit the nice altruistic suckers and drive them to extinction), it is extremely efficient when it comes to competition between groups. So, while religion may not always be all that good for the individual members of a religion, it definitely gives strength, resources, cohesion and power to groups. This might also be why religious people can't shut up about their beliefs :)
Anyway, in this somewhat bizarre TEDTalk presentation, psychologist Jonathan Haidt, while trying to say something profound about religion and self-transcendence, actually has more interesting stuff to say about group selection.
Try explaining massive-scale war on the selfish gene hypothesis... possible, but not entirely convincing.
If you're curious about David Sloan Wilson's approach to the scientific study of religion, you might want to check out this fascinating lecture.