TL;DR
- Breeder’s Equation math shows that even a tiny selection differential (-or + 0.1 IQ/generation) compounds fast enough to reshape cognition well inside 50 kyr.
- Empirical data say selection on IQ is ~ ± 0.6–0.8 points / generation today, falsifying the “S ≈ 0” premise directly.
- Ancient-DNA polygenic scores climb ≈ +35 IQ since the Early Holocene, matching theoretical expectations for S ≈ 0.2.
- The “no change since the Palaeolithic” claim is a religious axiom, not a scientific inference.
1 Quotes & the Blank-Slate Catechism#
“There’s been no biological change in humans in 40 000 or 50 000 years. Everything we call culture and civilization we’ve built with the same body and brain.” — Stephen Jay Gould oai_citation_attribution:0‡PMC
“Pre-historic people were our equals in mental capacity; the difference is purely cultural.” — David Deutsch, Within Reason podcast, 2025 oai_citation_attribution:1‡YouTube
Such pronouncements are treated as self-evident truths in many classrooms. Below I treat them as falsifiable hypotheses and run the numbers.
2 The Breeder’s Equation on a Beer-Coaster#
[ \Delta Z ;=; h^{2},S ]
- ΔZ – change per generation in the trait mean.
- h² – heritability (≈ 0.5 for IQ1).
- S – selection differential (mean of breeders – population).
2.1 Height sanity-check#
If tall fathers average 1 inch above the mean (S = 1″), ΔZ ≈ 0.8″/generation. Ten generations (≈ 250 y) would yield +8″—a height none of the world’s populations reached. So for height, Σ S over 50 kyr must hover near zero.
2.2 IQ thought-experiment#
Set S = +1 IQ (below test-retest error). With h² = 0.5, ΔZ = 0.5. After 10 generations: +5 IQ. After 400 generations (~10 ky): +200 IQ—clearly absurd. Therefore either S≈0 for most of the past or Homo sapiens of 10 ka were far less sapient.
3 Selection Now: Two Independent Lines#
| Data stream | Estimated S (IQ pts/gen) | Note |
|---|---|---|
| Fertility × test scores, UK/US twin data | -0.8 | Meta-analysis of sibship size vs IQ2 |
| Polygenic-score change across U.S. birth-years 1931-1990 | -0.6 (-0.055 SD) | Genomic selection scan3 |
Magnitude ≈ 1 in the relaxed selection regime of the late 20th C. Claiming S < 0.1 for every prehistoric generation is special pleading.
4 Selection Then: Ancient-DNA Signals#
Large aDNA datasets (~7 000 genomes) let us track polygenic scores (PGS) for hundreds of traits. Findings most relevant here:
- Cognitive PGS rose ~0.5 SD since 10 ka—≈ +35 IQ if linear scaling holds. oai_citation_attribution:2‡Cambridge University Press & Assessment
- Half of the twelve strongest Holocene sweeps involve cognition-linked traits (educational attainment, intracranial volume, etc.).
- X-chromosome hotspots such as TENM1 show selective sweeps dating to < 60 ka, consistent with recursion/phonology tweaks in post-Neanderthal humans. oai_citation_attribution:3‡PubMed
Plugging S ≈ 0.2 into the Breeder’s Equation reproduces the +35 IQ trajectory neatly.
5 Objections—& Why They Fail#
| Blank-Slate Objection | Rebuttal |
|---|---|
| “S = 1 IQ is too high.” | The empirical magnitude of S today is ~0.8 |
| “h² is low.” | Twin/adoption work → h²(IQ) ≥ 0.5; Turkheimer’s First Law4. |
| “IQ ≠ intelligence.” | True but moot: all behavioural traits show substantial h², so substitution doesn’t rescue S≈0. |
| "+5 SD is window-dressing." | Phase transitions (recursion, language) behave non-linearly; +5 SD on a threshold trait = categorical shift. |
| “Archaeology already shows symbolism @ 100 ka.” | Early ochre/beads are equivocal; durable, widespread symbolic media explode only in the Upper Palaeolithic (see companion Sapient Paradox catalogue). |
6 Implications#
- Sapient Paradox resolved. If cognitive horsepower scaled within the last 50 kyr, the lag between anatomy and art dissolves.
- Self-domestication timeframe. Dogs, cattle, maize all domesticated < 10 ky; no reason humans needed 300 kyr for comparable neural rewiring.
- Ideological fallout. Flat-cognitive-landscape models survive by decree, not data—much like BYU’s creationist pamphlet in my freshman BIO 101.
7 Conclusion#
Population genetics, modern fertility data, and ancient genomes converge on one boring conclusion: intelligence has been under directional selection throughout the Late Pleistocene and Holocene. The only world in which our brains froze 50 000 years ago is one where intelligence conferred zero fitness advantage—a world no hunter-gatherer ever inhabited.
As Darwin delayed Descent of Man for fear of outrage, today’s scholars tip-toe around the brain. But the genome doesn’t care. The numbers are in; the blank slate is out.
Footnotes#
Sources#
- S.J. Gould quote in Has Human Evolution Stopped? oai_citation_attribution:6‡PMC
- D. Deutsch & A. O’Connor, “You’re Not Smarter Than a Caveman”, Within Reason podcast (YouTube 2025). oai_citation_attribution:7‡YouTube
- Hugh-Jones D. et al., 2024 selection scan – see footnote 3. oai_citation_attribution:8‡Gwern
- Skov L. et al., “Extraordinary selection on the human X chromosome…”, Cell 2023. oai_citation_attribution:9‡PubMed
- Piffer D. “Evolutionary Trends of Polygenic Scores in European Populations…”, Twin Res. 2023. oai_citation_attribution:10‡Cambridge University Press & Assessment
Estimates of h² for psychometric g range 0.5–0.8 in adulthood (see Polderman et al., Nat. Genet. 2015). ↩︎
Lynn R. Dysgenics (1996) synthesised >25 fertility-IQ datasets; mean r ≈ –0.2 ⇒ S ≈ -0.8 IQ. ↩︎
Hugh-Jones D. & Kohler H-P. “Natural Selection across Three Generations of Americans”. 2024 pre-print. oai_citation_attribution:4‡Gwern ↩︎
Turkheimer E. “Three Laws of Behaviour Genetics”. Curr. Dir. Psychol. Sci. 9 (2000) 160-164. oai_citation_attribution:5‡SAGE Journals ↩︎