George Williams, a founder of the gene-centric school, claimed “Adaptation is always asymmetrical; organisms adapt to their environment, never vice versa.” He was wrong.

Richard Dawkins proposed the idea of “extended phenotypes”—a phenotype is the subset of an organism’s genetic traits that develop— to describe environmental changes caused by organisms. Derek Bickerton says, for Dawkins a “beaver’s dam [is] just as much an expression of beaver genes” as its tail. But that’s far from the whole tale of gene-environment interactions.

Organisms, and their genes, can face selective pressures from created elements in their environments. Intelligently created nonrandom factors have substantially altered human genes; for example, after only a few thousand years of dairying, adult lactose tolerance has spread to 98% of Swedes but just 7% of Chinese. Protocultural tool use in Galapagos finches has led to their beaks being adapted for using cactus spikes to extract grubs. Unlike Dawkins’s extended phenotypes, this “gene-culture coevolution,” as E. O. Wilson calls it, incorporates more than is in an organism’s genes, accounting for nongenetically transmitted factors like toolsrules, and socially acquired second nature skills.