War is an interesting scenario they mention. Even when faith-inspired overconfidence of survival or even divine reward for martyrdom backfires on an individual level and increases likelihood of death and lack of direct descendants, it could be good for the propogation of genes that contribute to that overconfidence if this act is advantageous to your kin.
Even if that's not the case and such genes will become less frequent, delusion in war (and in peacetime) certainly seems to have been good for religious growth in a more direct way, the propogation of another, non-dna confined kind of replicator, namely the religious meme. Perhaps that's because unlike dna, there's no cost to the frequency of the replicator for being wrong (so long as its unfalsifiable). Also, this replicator doesn't depend for its frequency so directly on the survival and reproduction of the vehicles, human bodies (to use Dawkins' terminology of replicators and vehicles), so it's interests aren't so dependant on how good or bad it is for them.
I also think overconfidence of survival in war would have to have a different role in these guys' mathematical model to overconfidence in reward for martyrdom (since being wrong that you will be rewarded after death has no effect on your survival, it is merely having the belief, overconfident or not, that matters).
A read something else on a delusion related note in the Selfish Gene by Richard Dawkins recently. He suggests somewhere near the beginning that since being able to decieve others will lead to the evolution of the ability to spot signs that someone is lying, we might also expect to have evolved the ability to hide from our own concious mind that we are lying, so as not to give away such clues.
So maybe we can blame evolution for religious founders and apologists believing their own lies
![Tongue](https://www.councilofexmuslims.com/Smileys/custom/tongue.gif)