In his book Social Evolution, Robert Trivers enlarged on his 1976 evolutionary theory of self-deception. Self-deception is

The anthropologist Lionel Tiger says something similar in Optimism: The Biology of Hope. The connection to the sort of constructive irrationality we have just been discussing is seen in Trivers's paragraph about 'perceptual defense':

The relevance of this to the wishful thinking of religion should need no spelling out.

The general theory of religion as an accidental by-product — a misfiring of something useful — is the one I wish to advocate. The details are various, complicated and disputable. For the sake of illustration, I shall continue to use my 'gullible child' theory as representative of 'by-product' theories in general. This theory — that the child brain is, for good reasons, vulnerable to infection by mental 'viruses' — will strike some readers as incomplete. Vulnerable the mind may be, but why should it be infected by this virus rather than that? Are some viruses especially proficient at infecting vulnerable minds? Why does 'infection' manifest itself as religion rather than as… well, what? Part of what I want to say is that it doesn't matter what particular style of nonsense infects the child brain. Once infected, the child will grow up and infect the next generation with the same nonsense, whatever it happens to be.

An anthropological survey such as Frazer's Golden Bough impresses us with the diversity of irrational human beliefs. Once entrenched in a culture they persist, evolve and diverge, in a manner reminiscent of biological evolution. Yet Frazer discerns certain general principles, for example 'homoeopathic magic', whereby spells and incantations borrow some symbolic aspect of the real-world object they are intended to influence. An instance with tragic consequences is the belief that powdered rhinoceros horn has aphrodisiac properties. Fatuous as it is, the legend stems from the horn's supposed resemblance to a virile penis. The fact that 'homoeopathic magic' is so widespread suggests that the nonsense that infects vulnerable brains is not entirely random, arbitrary nonsense.

It is tempting to pursue the biological analogy to the point of wondering whether something corresponding to natural selection is at work. Are some ideas more spreadable than others, because of intrinsic appeal or merit, or compatibility with existing psychological dispositions, and could this account for the nature and properties of actual religions as we see them, in something like the way we use natural selection to account for living organisms? It is important to understand that 'merit' here means only ability to survive and spread. It doesn't mean deserving of a positive value judgement — something of which we might be humanly proud.

Even on an evolutionary model, there doesn't have to be any natural selection. Biologists acknowledge that a gene may spread through a population not because it is a good gene but simply because it is a lucky one. We call this genetic drift. How important it is vis-a-vis natural selection has been controversial. But it is now widely accepted in the form of the so-called neutral theory of molecular genetics. If a gene mutates to a different version of itself which has an identical effect, the difference is neutral, and selection cannot favour one or the other. Nevertheless, by what statisticians call sampling error over generations, the new mutant form can eventually replace the original form in the gene pool. This is a true evolutionary change at the molecular level (even if no change is observed in the world of whole organisms). It is a neutral evolutionary change that owes nothing to selective advantage.

The cultural equivalent of genetic drift is a persuasive option, one that we cannot neglect when thinking about the evolution of religion. Language evolves in a quasi-biological way and the direction its evolution takes looks undirected, pretty much like random drift. It is handed down by a cultural analogue of genetics, changing slowly over the centuries, until eventually various strands have diverged to the point of mutual unintelligibility. It is possible that some of the evolution of language is guided by a kind of natural selection, but that argument doesn't seem very persuasive. I'll explain below that some such idea has been proposed for major trends in language, such as the Great Vowel Shift which took place in English from the fifteenth to the eighteenth century. But such a functional hypothesis is not necessary to explain most of what we observe. It seems probable that language normally evolves by the cultural equivalent of random genetic drift. In different parts of Europe, Latin drifted to become Spanish, Portuguese, Italian, French, Romansche and the various dialects of these languages. It is, to say the least, not obvious that these evolutionary shifts reflect local advantages or 'selection pressures'.

I surmise that religions, like languages, evolve with sufficient randomness, from beginnings that are sufficiently arbitrary, to generate the bewildering — and sometimes dangerous — richness of diversity that we observe. At the same time, it is possible that a form of natural selection, coupled with the fundamental uniformity of human psychology, sees to it that the diverse religions share significant features in common. Many religions, for example, teach the objectively implausible but subjectively appealing doctrine that our personalities survive our bodily death. The idea of immortality itself survives and spreads because it caters to wishful thinking. And wishful thinking counts, because human psychology has a near-universal tendency to let belief be coloured by desire ('Thy wish was father, Harry, to that thought', as Henry IV Part II said to his son*)).

There seems to be no doubt that many of the attributes of religion are well fitted to helping the religion's own survival, and the survival of the attributes concerned, in the stew of human culture. The question now arises of whether the good fit is achieved by 'intelligent design' or by natural selection. The answer is probably both. On the side of design, religious leaders are fully capable of verbalizing the tricks that aid the survival of religion. Martin Luther was well aware that reason was religion's arch-enemy, and he frequently warned of its dangers: 'Reason is the greatest enemy that faith has; it never comes to the aid of spiritual things, but more frequently than not struggles against the divine Word, treating with contempt all that emanates from God.'⁸⁵ Again: 'Whoever wants to be a Christian should tear the eyes out of his reason.' And again: 'Reason should be destroyed in all Christians.' Luther would have had no difficulty in intelligently designing unintelligent aspects of a religion to help it survive. But that doesn't necessarily mean that he, or anyone else, did design it. It could also have evolved by a (non-genetic) form of natural selection, with Luther not its designer but a shrewd observer of its efficacy.