Memes vs genes

Comments on Substack.

Few popular science books launch fields of research. Richard Dawkins’ The Selfish Gene is one of them. The field it launched was memetics,¹ and it is named after a word coined in the book:

We need a name for … a unit of cultural transmission, or a unit of imitation. “Mimeme” comes from a suitable Greek root, but I want a monosyllable that sounds a bit like “gene”. I hope my classicist friends will forgive me if I abbreviate mimeme to meme.

Here, Dawkins uses “meme” to describe any idea that can spread—nationalistic pride, the drowning child argument, Despacito.

Before reading the book, I had not known that 1) he coined the word “meme” (of the Internet meme), and 2) the analogy between memes and genes runs any deeper than “memes compete for attention, like genes do for resources”. In fact, the analogy offers at least two surprising parallels between genes and memes: memes don’t need to arise from intelligence design, and they can be co-adapted.

The first parallel: successful memes can be created unintentionally.

One of the key ideas in The Selfish Gene is that no organism plans to have a feature suited for its environment. Rather, genes — the recipe/source code for the animal’s features — accumulate errors while copying DNA from parent to child. These errors sometimes cause a phenotypic change — a change in the quality of the dish/the behavior of the compiled program. When the phenotypic change improves prospects for reproduction, the gene spreads. No intelligent design — no conscious process picking good genes — necessary. Just the genes likely to get passed on, getting passed on. A bit like rational behavior getting selected for in markets.

Similarly, the most successful memes may not have been designed to spread. Earworms are a great example: no author of one, I assume, wants their song to become an earworm. Most great scientific works are unintended memes. Newton’s goal with inventing calculus was not to create widely-adopted mathematical machinery, but to model the orbits of planets. In general, scientists aim for correctness, not popularity. That the memetic success of great science is unintended is best shown by the fact that much of it has to be protected by patents.

“Attention Is All you Need” and “Goto Statement Considered Harmful” have birthed catchphrases in the CS community, but their authors probably didn’t intend this: they just found the titles neat. The word “meme” is an excellent meme, too. But Dawkins just wanted to show the versatility of the replicator idea by coining it.

Dawkins argues that arguably the most successful memes ever — religions — may have been unexpected hits as well. The idea of hell fire was probably not “planned deliberately by a Machiavellian priesthood trained in deep psychological indoctrination techniques”. Rather, it was an “unconscious meme” that ensured its own survival with its resilience, which the Durants colorfully describe in The Lessons of History:

One lesson of history is that religion has many lives, and a habit of resurrection. How often in the past have God and religion died and been reborn! Ikhnaton used all the powers of a pharaoh to destroy the religion of Amon; within a year of Ikhnaton’s death the religion of Amon was restored. Atheism ran wild in the India of Buddha’s youth, and Buddha himself founded a religion without a god; after his death Buddhism developed a complex theology including gods, saints, and hell. Philosophy, science, and education depopulated the Hellenic pantheon, but the vacuum attracted a dozen Oriental faiths rich in resurrection myths. In 1793 Hébert and Chaumette, wrongly interpreting Voltaire, established in Paris the atheistic worship of the Goddess of Reason … In America the rationalism of the Founding Fathers gave place to a religious revival in the nineteenth century.

The second parallel: memes can be co-adapted.

In carnivores, the genes for sharp teeth, front-facing eyes, and short intestines are “co-adapted” to form a gene complex. None of these genes is individually great. In a herbivore, they would be a resource drag at best, and actively harmful at worst. Rather, the gene for sharp teeth works for its genetic climate — that is, with the other genes in the carnivore’s body.

Similarly, memes may get co-adapted into big meme complexes. An example is an organized church — its architecture, rituals, music, and written traditions work well together, but individually any of them might not spread very far. Similarly, in science, a slow dribble of anomalies — none of which individually gets many citations — causes a Kuhnian paradigm shift, which packages the papers reporting the anomalies into a highly cited meme complex. Technologies which complement each other are co-adapted in this sense, too: a breakthrough in neural networks architecture is resurrected by an improvement in chip design, and vice versa.

Memetic engineering

The meme-gene analogy ends here in The Selfish Gene. But you can extend it to think about “memetic engineering”. That genes are not consciously designed doesn’t mean they cannot be consciously designed. Humans have edited the cotton genome to make it resistant to pests, and the rice genome to fortify it with beta-carotene. By studying what distinguishes the best-spreading genes from the bad, they have designed their own killer genes. For example, gain-of-function researchers study viruses to increase their transmissivity.

Similarly, you could study the best-spreading memes to design your own. Economists, sociologists and practitioners have indicated the marketing lessons from religions. The EA Forum has case studies of effective (and ineffective) advocacy. And, of course, Twitter is the memetic engineer’s model organism: like an E. coli culture, it cuts feedback loops dramatically shorter.