Ancient DNA Shows Schizophrenia Risk Purged Over 10,000 Years

TL;DR

• A new ancient-DNA time series from Akbari, Reich, and colleagues shows strong directional selection in West Eurasians for lower schizophrenia polygenic scores (PGS) over the last 10,000 years. ¹
• Interpreting the effect sizes in ordinary epidemiologic units, the best-fit trend implies schizophrenia may have been 5–10× more common in some Holocene populations than today—roughly “every twentieth adult,” not “one in a few hundred.” ²
• Schizophrenia risk variants are disproportionately enriched in regions of the genome that have experienced human-specific positive selection, implying the trait is a failure mode of something adaptive rather than pure genetic garbage. ³
• These dynamics map almost one-to-one onto the Eve Theory of Consciousness (EToC): recursion and selfhood as recent, fitness-enhancing innovations that opened a new valley of insanity, with the Holocene as a global clean-up operation. ⁴
• The Neolithic Y-chromosome bottleneck—where ≈95% of male lineages vanish between ~7000–5000 BC—slots neatly into the same story as a sex-skewed selection wave on male failure modes of selfhood. ⁵

Companion article: For a broader analysis of how the Akbari & Reich findings impact intelligence, bipolar disorder, education, and income alongside schizophrenia, see “Holocene Minds on Hard Mode”.

“For many generations, there must have been a tug-o-war between Adam, his daemons, and Eve; god would not go down without a fight.” ⁴
— Eve Theory of Consciousness

Schizophrenia as the Shadow of Eve#

If you buy the Eve Theory of Consciousness, you already accept three controversial premises:

1. Recursion is the core computational trick behind self-awareness, inner speech, deliberation about the future—the human condition. ⁴
2. Women crossed the recursion threshold first, with men following later, mostly in the Holocene. ⁴
3. Consciousness was not a gentle UX update; it was a phase transition in how brains are wired to model themselves, and therefore a generator of new, specifically human failure modes.

Schizophrenia, in this frame, is not random madness. It is what happens when recursion misfires—when the self-model fractures and the old bicameral gods refuse to leave quietly.

That’s the mythic claim. What the new Akbari–Reich paper adds is a time series: a literal graph of how hard evolution has been pushing back against that failure mode in the last 10,000 years. ¹

The figure we started from (their panel for schizophrenia PGS) shows three ancestral West Eurasian strands:

• Western hunter-gatherers (orange) with high schizophrenia PGS around 8–9k years before present.
• Early Anatolian farmers (green) with notably lower scores.
• Steppe pastoralists / Indo-Europeans (blue) arriving mid-Holocene with elevated scores again. ⁶

Overlaid on top is a black smoothed curve through all ancient genomes and a red regression line: both trend steeply down over time. The statistics (γ ≈ −0.84, highly significant) say the obvious out loud: schizophrenia risk alleles have been systematically drained from the West Eurasian gene pool.

EToC has been arguing for years that schizophrenia only becomes possible once there is a self to break—a Holocene disease of failed recursion. ⁴ Akbari and Reich just handed us a movie of evolution trying to turn the volume down.

How Much More Common Was Schizophrenia?#

Before we plug this into Genesis, we need to translate the population-genetic Esperanto of “polygenic scores” into something like incidence rates.

Modern baseline#

Globally, schizophrenia affects on the order of 0.3–0.7% of people at some point in life, depending on methodology. ²
That’s roughly 1 in 150–300 adults.

Incidence—the number of new cases per year—clusters around 15 per 100,000 person-years, with wide variation between sites. ⁷

So “normal modern world” is: in a town of 10,000 people, maybe 30–60 adults alive right now with schizophrenia, and 1–2 new cases per year.

What the PGS slope implies#

Polygenic scores are standardized; a shift of +1 SD in liability, in case–control GWAS, often corresponds to about a doubling of odds for the trait. Exact numbers vary, but odds ratios of ~2 per SD are routine. ⁸

Reading off Akbari–Reich’s schizophrenia panel, the difference between early Holocene high-risk groups and present-day West Eurasians is on the order of 1.5–2 SD of PGS. The regression line drops from ~0.8 at 9k BP toward ~0 by the present; the smoothed curve kisses slightly above and below that. ¹

Brutal back-of-the-envelope:

• Suppose each +1 SD in schizophrenia PGS multiplies the risk by ≈2×.
• A +1.5–2 SD gap therefore corresponds to a 3–4× to 4–6× increase in incidence.
• If we take the modern baseline as ~0.5% lifetime risk, 4–6× gives 2–3%, and the upper tail of plausible mappings hits ~5%.

So a central mental image that is consistent with the genetics is:

In some Holocene populations, something like 1 in 20 adults might at some point have tipped into a schizophrenia-like state, rather than 1 in a few hundred.

Two important caveats:

1. The mapping from PGS to incidence is noisy and context-dependent. These scores are trained on modern societies with antipsychotics, obstetrics, and urban stressors that did not exist in 6000 BC. ⁸
2. Selection likely acts on liability more broadly—milder psychotic traits, cognitive side-effects, social dysfunction—not just diagnosed cases.

But the direction is clear: Holocene humans were measurably more psychosis-prone, and evolution has been grinding away at that liability ever since.

From an EToC perspective, that’s exactly what you would expect if early recursion was a dangerous technology that had not yet been debugged.

Schizophrenia as a Failure Mode of Recursion#

Standard psychiatric genetics was already converging on a paradox: schizophrenia is highly heritable, reduces reproductive fitness, and yet is not ultra-rare. ⁹

• Lifetime risk ~0.7%, chronic course, and large functional impairment. ¹⁰
• A huge Swedish population study found patients with schizophrenia have fertility ratios around 0.25–0.5—far fewer children than their unaffected siblings. ¹¹

Under equilibrium, that’s a recipe for the trait to dwindle into the noise. Instead, GWAS finds hundreds of common risk alleles of tiny effect, many in regions that show signatures of positive selection in humans. ³

Liu et al. (2019) explicitly compared risk alleles in modern humans to archaic genomes and found evidence that schizophrenia risk variants have been negatively selected in recent humans relative to Neanderthals and Denisovans. ¹²

That’s already suspicious. Something about Homo sapiens’ specific cognitive upgrade both:

• Introduced new architectural vulnerabilities (so Neanderthals weren’t doing this), and
• Conferred enough upside that evolution kept pushing into the risky regime and then trimming back the worst failures.

EToC translates that into plain language:

• The package of recursive abilities—self-reflection, narrative, time travel, counterfactual planning—came online relatively late, probably in stages, with women ahead of men. ⁴
• Every new way of modeling yourself is also a new way of mis-modeling yourself. If your self-model fractures or assigns the wrong source tags to internal voices, you get hallucinations, persecutory delusions, “voices of gods.”
• There is no schizophrenia in a world without selves; there is only possession. (Jaynes comes close here, but EToC insists recursion and “I-ness” are of a piece. ⁴)

The genomic enrichment of schizophrenia loci in human-specific selected regions is exactly what you’d expect if the trait is a shadow cast by a beneficial re-wiring of cortex rather than an ancient pathogen lingering for no reason. ³

Akbari & Reich’s result just time-stamps the clean-up: the strong downward trend in PGS over the last 10k years says that once recursion stabilized, selection went to work pruning its worst edge cases.

Timeline: Eve Theory Meets Holocene Genetics#

To make the correspondence clearer, here’s a deliberately reductive timeline.

This is not a proof; it’s a pattern match. But it’s a disturbingly good one.

The Snake Cult as a Selection Engine#

In EToC, the Snake Cult of Consciousness is not a single archaeological site; it’s a game-theoretic attractor: any ritual complex that discovers how to push young adults over the introspective threshold with maximum drama—song, fasting, sex, torment, and yes, snake venom. ⁴

Two key claims matter for our purposes:

1. Self-awareness can be induced. Psychedelics plus deliberate symbolic framing can make the self “flip on” in those whose brains are on the threshold.
2. The intervention is risky. Only a fraction of initiates land in a stable attractor; others bounce back to pre-self states or fall into chronic fragmentation.

If you run that ritual for centuries in small, competitive groups, you get a massive selection gradient on stable recursion:

• Men whose brains can support a coherent self after initiation become exceptionally capable, and likely more attractive mates and leaders.
• Men who tip into psychosis either die, fail to reproduce, or are marginalized.

Power et al.’s fertility data show how brutal the gradient can be even in modern Sweden: people with schizophrenia have between one quarter and one half as many children as their siblings. ¹¹

Now add the Neolithic Y-chromosome bottleneck, where effective male population size drops to the equivalent of roughly one reproducing man for every 15–20 women, while female lineages stay diverse. ⁵

Mainstream work explains this as:

• Patrilineal, patrilocal kinship systems.
• Inequality and clan warfare, concentrating reproductive success in a minority of males. ¹³

EToC adds one more layer: what traits are being selected within those patrilineal clans?

If the competitive advantage of a lineage depends partly on how well its leaders can coordinate, plan, and manipulate symbols—i.e., on stable recursion—then clan war and patrilineality become the ecological stage on which the Snake Cult runs its experiment at planetary scale.

Akbari & Reich show that schizophrenia-related alleles are among those being steadily culled during this period. ¹
Karmin and later work show that male lineages undergo a synchronized culling worldwide. ⁵

It’s not crazy to see these as different cuts through the same historical process.

Why the Akbari–Reich Curve Is EToC-Shaped#

Zoom in on three features of the schizophrenia PGS curve that are especially suggestive from an EToC lens.

1. High initial risk in hunter-gatherers#

The Western hunter-gatherer group sits at the high end of schizophrenia PGS in the Akbari–Reich plot. ¹

In EToC, these are precisely the people living through the transition: small bands, intense ritual culture, pockets of recursive self-awareness, and what Jaynes would call gods shouting in people’s heads.

The model predicts lots of unstable partial selves—people who can do some recursion but not enough to integrate it, a prime recipe for psychotic experiences. The high PGS looks like a genetic fossil of that.

2. Anatolian farmers as local minima#

Early farmers from Anatolia show distinctly lower schizophrenia PGS than either the hunter-gatherers or the later Indo-European pastoralists. ⁶

Farming imposes a different selection regime:

• More stable, kin-dense communities.
• High premium on long-term planning and delayed gratification.
• Possibly less need for extreme initiation rituals once selfhood becomes more common in childhood.

In other words, a partial escape from the valley of insanity: recursion is starting to stabilize, and the worst failure modes are being weeded out.

EToC would read this as an early pocket where the “self meme” has largely fixed, and the main selective work is now on fine-tuning rather than brute-force induction via venom and terror.

3. Indo-Europeans re-introduce risk—and it gets culled#

Then the Indo-European steppe populations show up with elevated schizophrenia PGS again, and yet over the combined West Eurasian time series, the overall trend continues downward. ¹

So we get the following dynamic:

1. A new group arrives with a higher load of “bad recursion” alleles.
2. Admixture temporarily bumps the regional liability up.
3. The same selection gradient that’s been operating for millennia grinds it back down.

This is exactly what you’d expect if the underlying cognitive architecture is now locked in—everyone is playing the recursion game—but negative selection keeps shaving off lineages whose implementation is too fragile.

How Strong Was the Selection?#

Akbari & Reich quantify selection strength in terms of γ, their trend parameter; values around −0.8 with tiny P-values indicate a strong, consistent push across millennia. ¹⁴

We can translate this loosely into a selection coefficient on schizophrenia liability:

• Power et al.’s fertility ratios suggest that diagnosed schizophrenia today has something like a 50–75% fitness cost, depending on sex. ¹¹
• If early Holocene populations had more people in the high-liability tail—say, 5% instead of 0.7%—the aggregate fitness drag could easily reach several percent of total reproductive output.

A few percent may not sound like much, but in population genetics that’s enormous. A selection coefficient of just 1–2% acting consistently over hundreds of generations is more than enough to drive large allele-frequency shifts, especially when the trait is highly polygenic.

EToC’s bet was always that the ascent of self-awareness was recent enough and strong enough to leave detectable selection scars. ⁴
Akbari & Reich are showing those scars in real time.

What This Does Not Prove (Yet)#

This is where we have to resist theologizing our own pet theory.

The data do not prove:

• That snake-venom initiation rituals existed in exactly the way EToC sketches them. That’s a specific conjecture hanging on suggestive ethnography and the sheer ubiquity of snake deities. ⁴
• That women’s earlier adoption of recursion is the proximal cause of the patterns in schizophrenia PGS; the ancient DNA doesn’t stratify by sex and cognitive phenotype (yet).
• That the Y-chromosome bottleneck is driven primarily by recursion stability rather than by more prosaic patrilineal politics.

What the data do show, independent of any mythic gloss:

1. Schizophrenia liability has been under sustained negative selection in humans. ¹²
2. The last 10,000 years in West Eurasia saw a large, monotonic drop in schizophrenia PGS. ¹
3. Male lineages experienced an extreme Holocene bottleneck, consistent with a world where very few men did much of the reproducing. ⁵

If EToC is even roughly right, those facts were predicted in spirit: a late, dangerous cognitive upgrade, followed by a long epoch of selection for stable implementations.

So the epistemic status here is not “settled” but “the parameter space just shrank a lot.”

Where the Theory Could Fail#

A theory is only interesting to the extent it risks being wrong.

Here are a few ways EToC’s reading of schizophrenia could be falsified or at least boxed in:

1. Ancient schizophrenia incidence stays flat. If future work manages to infer psychosis liability more directly—for example via endophenotypes or brain-relevant polygenic scores—and finds no Holocene spike, then the EToC “valley of insanity” starts to look more metaphor than model.
2. Risk alleles pre-date recursion. If high schizophrenia liability turns out to be equally present in archaic humans clearly lacking recursive culture, that weakens the link between the trait and selfhood.
3. Sex-specific data disagree. EToC predicts sex-skewed dynamics: earlier, stronger selection on stable recursion in women, then a catch-up wave hitting men in the Holocene. If ancient DNA with sex-stratified PGS trends goes the other way, the story needs revision.
4. Alternative traits explain the selection signal better. Many schizophrenia loci are pleiotropic; if it turns out the main driver of selection was something orthogonal (say, infection resistance), then the neat fit with recursion becomes more tenuous. ¹⁵

Note that all of these are, in principle, empirically addressable. One nice thing about anchoring consciousness in genetics and archaeology is that Nature has opinions.

FAQ#

Q 1. Does this mean people in 6000 BC were “crazier” than us? A. Not in the Hollywood sense; rather, more people’s brains probably sat closer to the psychosis threshold, with a higher rate of frank schizophrenic breakdowns and milder subclinical symptoms woven into religious and shamanic life.

Q 2. Could culture alone explain the decline in schizophrenia risk? A. Culture surely matters, but the Akbari–Reich result is explicitly about allele frequencies shifting over millennia, which requires differential reproductive success—not just different upbringing or stress exposure. ¹⁴

Q 3. How does the Y-chromosome bottleneck tie into this? A. Around 7000–5000 BC, male lineages collapsed to what looks like one successful patriline for every ~15–20 women, implying intense selection on male traits; EToC suggests one such trait was the ability to survive selfhood without catastrophic psychosis. ⁵

Q 4. Isn’t schizophrenia too heterogeneous to be a single “failure mode of recursion”? A. Clinically, yes; genomically, risk alleles converge on brain development and synaptic function in association cortex and DMN-adjacent regions—exactly where you’d expect recursion-related circuitry to live, making “fragile self-model” a natural organizing metaphor. ⁹

Q 5. What new data would most help test the Eve Theory angle? A. Sex-stratified ancient PGS for schizophrenia and related traits, better dating of recursive language emergence, and cross-regional comparisons (e.g., Eastern Eurasia) to see whether similar selection patterns track independent “awakening” trajectories. ¹⁶

Footnotes#

Sources#

1. Akbari, A., et al. “Pervasive findings of directional selection realize the promise of ancient DNA to elucidate human adaptation.” Genome Research (2024). ¹⁴
2. Saha, S., et al. “A Systematic Review of the Prevalence of Schizophrenia.” PLoS Medicine 2(5) (2005): e141. ⁷
3. World Health Organization. “Schizophrenia.” Fact sheet, 2025. ²
4. Power, R. A., et al. “Fecundity of Patients With Schizophrenia, Autism, Bipolar Disorder, Depression, Anorexia Nervosa, or Substance Abuse vs Their Unaffected Siblings.” JAMA Psychiatry 70(1) (2013): 22–30. ¹⁷
5. Owen, M. J., et al. “Genomic findings in schizophrenia and their implications.” Molecular Psychiatry 28 (2023): 1–17. ¹⁸
6. Srinivasan, S., et al. “Genetic markers of human evolution are enriched in schizophrenia.” Biological Psychiatry 80(4) (2016): 284–292. ³
7. Liu, C., et al. “Interrogating the Evolutionary Paradox of Schizophrenia.” Frontiers in Genetics 10 (2019): 389. ¹²
8. Karmin, M., et al. “A recent bottleneck of Y chromosome diversity coincides with a global change in culture.” Genome Research 25(4) (2015): 459–466. ⁵
9. Guyon, L., et al. “Patrilineal segmentary systems provide a peaceful explanation for the post-Neolithic Y-chromosome bottleneck.” Nature Communications 15 (2024). ¹³
10. Cutler, A. “Eve Theory of Consciousness (v2).” Vectors of Mind (2023). ⁴
11. Cutler, A. “The Snake Cult of Consciousness.” Vectors of Mind (2023). ¹⁹
12. Hudon, A., et al. “Exploring the intersection of schizophrenia, machine learning, and genomic data: a scoping review.” JMIR Bioinformatics and Biotechnology (2024). ²⁰
13. Bhugra, D. “The global prevalence of schizophrenia.” PLoS Medicine (2005). ²¹
14. CNRS. “Social change may explain decline in genetic diversity of the Y chromosome at the end of the Neolithic period.” Press release, 2024. ²²
15. Aporia Magazine. “Overwhelming evidence of recent evolution in West Eurasians.” 2024. ⁶