FOXP2

In one line: FOXP2 was crowned "the language gene" on the strength of two amino acids — and then the evidence quietly took the crown back.

---

The gene

FOXP2 ("forkhead box P2") sits on the long arm of chromosome 7 (7q31). It doesn't build the machinery of speech directly; it's a transcription factor — a regulatory switch that turns hundreds of other genes on and off. It came to science through the KE family, a three-generation London pedigree in which about half the members inherit a severe speech-and-language disorder (developmental verbal dyspraxia / childhood apraxia of speech: orofacial-motor deficits plus expressive and receptive language problems). In 2001 the cause was pinned to a single change in FOXP2 — R553H, arginine to histidine, in the protein's DNA-binding (forkhead) domain.

The crowning

In 2002 came the headline finding: human FOXP2 differs from the chimpanzee's by just two amino acids (T303N and N325S — both distinct from the KE family's R553H). The same paper argued the gene had been swept through our species by recent positive selection, roughly coincident with the dawn of language. The story was irresistible: two changes, and humanity could speak. The press dubbed it the language gene.

The mechanism

What FOXP2 actually does is subtler — and not about grammar. It is expressed in the basal ganglia (the striatum/caudate) and shapes corticostriatal circuits. KE-family members show structural and functional differences in the caudate. When researchers engineered mice to carry the two human amino acids, the mice didn't talk: their striatal neurons grew differently, dopamine shifted, and they moved faster from declarative to procedural learning — i.e., they automated motor routines sooner. FOXP2's domain is learned, sequenced movement, not syntax.

The fall — and what's really true

Then the grand narrative came apart. In 2007, ancient DNA showed Neanderthals carried the same two changes — so they can't be what made modern humans uniquely able to speak. In 2018, a large, globally diverse genetic survey found no evidence of a recent selective sweep at FOXP2; the original signal was an artifact of sample composition. And FOXP2 turns out to be a deeply conserved, pleiotropic gene — expressed in lung, gut and heart as well as brain, and required by songbirds to copy their songs. It isn't "the gene for language." It's one node in a vast network — the first thread we pulled.

Sources

Full claim-by-claim evidence is in references.md. Primary anchors:

• Lai et al. 2001, Nature — FOXP2 identified; R553H in the KE family.
• Enard et al. 2002, Nature — two human-vs-chimp substitutions; the (later-refuted) selection claim.
• Watkins et al. 2002, Brain; Vargha-Khadem et al. 2005, Nat Rev Neurosci — basal-ganglia phenotype.
• Krause et al. 2007, Current Biology — Neanderthals share the two derived changes.
• Haesler et al. 2007, PLoS Biology — FoxP2 knockdown impairs songbird vocal imitation.
• Enard et al. 2009, Cell; Schreiweis et al. 2014, PNAS — humanized-mouse corticostriatal/procedural effects.
• Atkinson et al. 2018, Cell — no evidence of recent selection at FOXP2.

Accuracy note: This episode is deliberately built against three traps in the popular FOXP2 story. (1) The KE mutation R553H is not one of the two human-vs-chimp changes (T303N/N325S) — different positions, different meaning (disease vs. evolution). (2) There was no recent selective sweep, and the two changes are shared with Neanderthals — so "two amino acids made modern humans speak" is unsupported. (3) FOXP2 is not "the language gene" and there is no evidence it confers syntax or recursion; it is a pleiotropic regulator of vocal-motor / procedural-sequence learning.