Testing the Late Pleistocene Arctic Origins of East Asian Psychology using Ancient and Modern DNA

doi:10.21203/rs.3.rs-7696395/v1

Download PDF

Research Article

Testing the Late Pleistocene Arctic Origins of East Asian Psychology using Ancient and Modern DNA

https://doi.org/10.21203/rs.3.rs-7696395/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 04 Nov, 2025

Read the published version in Adaptive Human Behavior and Physiology →

You are reading this latest preprint version

This study tests climatically driven selection on behavioral polygenic scores (PGS) using global modern genomes together with ancient datasets from both Western and Eastern Eurasia. PGS were computed for Extraversion, Agreeableness, and Neuroticism, and temporal trends were modeled with hinge regressions at 12,000 and 19,000 years BP while adjusting for latitude and coverage. In modern worldwide populations, all three traits showed latitude-associated differentiation, consistent with cold-climate adaptation. In Western Eurasian ancient genomes, Extraversion and Neuroticism both declined with sample age before the Last Glacial Maximum (pre-knot slopes ≈ − 0.018 and − 0.045 to − 0.051 SD/kyr), then reversed direction after 12k–19k BP (slope shifts Δβ ≈ +0.036 to + 0.104, all p < 0.001), indicating post-LGM relaxation of selection. Agreeableness showed the complementary pattern: increasing before the knots and attenuating thereafter. A latent factor extracted from the three traits, oriented to reflect higher Agreeableness and lower Extraversion/Neuroticism, captured these coordinated trends at the population level (Δβ ≈ −0.17, p < 0.001), supporting the single-trait findings. In the Eastern Eurasian sample, limited pre-Holocene coverage precluded hinge tests, but within the Holocene, ancestry components (e.g., Jōmon, Tibetan, Northeast Chinese) predicted trait scores in directions broadly consistent with Arctic adaptation, whereas latitude effects were weaker once admixture was accounted for. Collectively, both trait-specific and factor-level results support the hypothesis that late Pleistocene cold intensified selection against Extraversion and Neuroticism while favoring Agreeableness, with Holocene climates reversing or relaxing these pressures.

Ancient DNA

Polygenic Scores

Factor Analysis

Hinge Regression

Climatic Adaptation

East Asia

Behavioral Evolution

Current evolutionary psychology frameworks focus primarily on universal savannah adaptations (Tooby & Cosmides, 1990) or recent Holocene cultural developments like rice agriculture and Confucianism (Nisbett, 2003; Talhelm et al., 2014). This leaves a critical gap: 40,000 years of post-Out-of-Africa adaptation to diverse Pleistocene environments, particularly the extreme conditions of the Arctic. Despite mounting evidence from genomics (Mao et al., 2021), paleoclimatology (Jin et al., 2020), and evolutionary medicine (Price & Lannon, 2023) documenting intense selection pressures during this period, psychological models have largely overlooked these formative millennia.

Studies examining polygenic adaptation in cognitive, personality, and physiological traits using ancient DNA have been limited to the past 12,000 years due to the scarcity of older samples available at the time of publication (Piffer & Kirkegaard, 2024a; Piffer, 2025).

David Sun (2025) addresses this gap with a compelling theory centered on Late Pleistocene Arctic environments. During the Last Glacial Maximum (∼26–19 kya), ancestral Northeast Asian populations inhabited the harsh Arctic and subarctic regions of Siberia. Sun argues that these extreme conditions—characterized by bitter cold, resource scarcity, and survival dependence on small-group cooperation—created powerful selective pressures that shaped core behavioral phenotypes still evident today. Similarly, Qi et al. (2023) demonstrate that cold climates promoted social bonding and complex group behaviors in Asian colobine primates, suggesting parallel evolutionary mechanisms in humans.

The theory proposes four key psychological adaptations that emerged from Arctic survival demands: reduced Extraversion for energy conservation in confined winter shelters, suppressed Neuroticism for emotional control under life-threatening conditions, heightened Agreeableness to maintain group harmony and prevent fatal conflicts, and enhanced visuospatial abilities for navigation and tool use in sparse environments. Most significantly, Sun contends these Arctic-forged traits spread throughout East Asia as these populations migrated southward during the Holocene, establishing psychological foundations that preceded and potentially enabled later cultural innovations like Confucianism and intensive agriculture.

Sun's framework draws support from multiple disciplines. Genetic studies reveal Arctic-adapted alleles (EDAR V370A, LEPR) and elevated population divergence in East Asians before 15,000 years ago (Bryk et al., 2008; Mao et al., 2021; Coop et al., 2009). Ethnographic parallels between modern Inuit and East Asian populations show striking similarities in emotional suppression, conflict avoidance, and risk-averse decision-making (Briggs, 1970; Draguns et al., 2000; R. Nelson, 1969). Modern polar research stations provide a natural experiment, with personnel developing remarkably similar psychological profiles—emotional stability, reduced assertiveness, and cooperative behavior—under comparable environmental pressures (Gunderson, 1974).

However, Sun's theory has not been subjected to rigorous population-genetic testing or examined across historical timescales, leaving predictions untested.

To move beyond coarse time–latitude models, we summarize ancestry structure - especially in Eastern Eurasia - using model-based ADMIXTURE and then relate those ancestry components to personality PGS. This approach lets us ask which lineages actually carry the predicted “Arctic” psychological profile, instead of attributing differences to geography alone. In practice, we estimate individual ancestry fractions and merge them with PGS at the individual level, enabling regressions that include date and coverage controls. This design aligns directly with our questions about lineage-specific selection versus ecological proxies and tackles the research questions from three different angles:

First, an admixture lens tests Sun’s claim that adaptations concentrated in interior, highly seasonal Northeast Asian lineages rather than “cold per se.” By decomposing genomes into components (e.g., Jōmon, Northeast Chinese, Tibetan, Afanasievo–Yamnaya) and entering those components as predictors of the agreeableness–emotional-stability composite, we can identify which ancestries track the predicted profile and which do not. In our results, several Northeast Asian–derived components show positive associations (e.g., Jōmon, Tibetan, Northeast Chinese), whereas Afanasievo/Yamnaya is not reliably positive—pinpointing lineages consistent with the hypothesis.

Second, adding admixture clarifies whether apparent spatial trends are confounded by ancestry. Indeed, once ancestry components are included, latitude effects in the Eastern dataset weaken or vanish, indicating that between-population psychological differences are better explained by ancestry histories than by location alone. This directly adjudicates “ecology vs. ancestry” alternatives that time-only or latitude-only models cannot resolve

Finally, because our Eastern dataset is mostly Holocene, admixture provides power precisely where pre-LGM time contrasts are sparse: it lets us test the hypothesis via who (ancestry lineages) rather than when (pre- vs post-LGM) in that region, while remaining comparable to our Western hinge-regression analyses. We therefore foreground admixture in the Introduction, with full implementation details in Methods (software, merging procedure) and the Eastern Eurasia section (K-level and component labeling).

Related regression designs—linking polygenic scores in ancient individuals to sample age (years BP) and ancestry fractions—have been applied previously in Eastern Eurasia (e.g., Piffer, 2025). Our analyses focus on personality-relevant PGS, integrate ADMIXTURE components explicitly, add Western time-series hinge tests, and include additional controls (coverage, latitude) and reliability checks.

Testing Arctic Psychology at Scale

We present the first comprehensive evaluation of Sun's Arctic adaptation theory through four primary predictions. If Arctic environments truly shaped core East Asian psychological traits, then reduced Extraversion and Neuroticism combined with elevated Agreeableness should characterize not only modern East Asian populations, but also Arctic groups worldwide, Native Americans - which descended from Siberian migrants (∼15 kya; Sikora et al., 2019) - and ancient Northeast Asian genomes spanning the transition from Pleistocene to Holocene cultures. Under extreme, chronic threat in small interdependent groups, high emotional reactivity would impair coordination and risk management, whereas lower Neuroticism - i.e., stronger emotional control - would stabilize decision-making and social harmony. Comparative work on cold-climate primates and ethnography of polar teams converges on this prediction (Qi et al., 2023; Gunderson, 1974). Alternative hypotheses are possible, but the theoretical expectation is explicit: cold, resource-scarce winters favored emotional restraint.

We test this prediction by comparing the polygenic scores for Extraversion, Neuroticism and Agreeableness between these ancestral groups and others such as Europeans, Africans and South Asians.

A corollary of this prediction is that the polygenic signatures of reduced Extraversion, reduced Neuroticism, and elevated Agreeableness constitute an integrated adaptive complex driven by shared Arctic selection pressures. Consequently, these traits should covary systematically across populations, loading onto a common latent factor reflecting their coordinated evolution. This predicted factor structure would provide empirical evidence that Arctic adaptation shaped not just individual traits, but a syndrome of co-adapted psychological characteristics.

Additionally, this theory predicts that post-Last Glacial Maximum genomes (~ 19 kya) should exhibit lower Extraversion polygenic scores compared to pre-LGM samples.

Our third prediction is that Southeast Asian populations, carrying higher proportions of Austronesian-related ancestry (Yang et al., 2020; He et al., 2022), will show diluted expression of these Arctic-adapted psychological traits compared to their Northeast Asian and circumpolar counterparts, with corresponding signatures detectable in ancient DNA records.

Our fourth prediction is that the effect of cold adaptation on personality is gradual, not dichotomous (i.e., exclusive to Arctic populations). Consequently, we predict a gradient of personality traits across human populations corresponding to latitude (distance from the equator). This prediction is based on the polygenic architecture of psychological traits and the action of selection on standing genetic variation.

To examine temporal patterns in polygenic scores, we modeled sample age as a continuous variable (Years BP) rather than grouping observations into discrete periods. Specifically, we employed hinge (piecewise) regression models with knots placed at 12,000 BP and 19,000 BP. These thresholds were chosen because they correspond to major paleoclimatic transitions: 12,000 BP marks the onset of the Holocene, when global temperatures rose sharply following the Younger Dryas, while 19,000 BP marks the end of the Last Glacial Maximum (LGM), the coldest phase of the last glacial cycle. By testing slope changes at these points, the models evaluate whether genetic trends align with known environmental shifts.

In the hinge regression framework, the trajectory of standardized PGS values is estimated continuously over time, with the possibility of a slope change at the knot but no forced discontinuity in level. Scatterplots of Years BP against PGS (z-scores) were created for visualization, with fitted hinge regression lines overlaid to illustrate estimated pre- and post-knot trends.

For each personality trait PGS, we regressed standardized scores on (a) sample age relative to the knot (in kyr), (b) the hinge term capturing slope change beyond the knot, and (c) covariates for latitude and sequence coverage to adjust for geographic and data quality effects. The pre-knot slope, the post-knot slope (derived from the pre-knot slope plus the hinge coefficient), and the significance of the slope change together indicate whether temporal trajectories in PGS shifted direction at 12k or 19k BP.

Through integrated analysis of ancient and modern genomic data, we evaluate Sun's Total Evolutionary Ecologies model, which emphasizes pre-Holocene environmental pressures in psychological evolution. Our approach addresses fundamental questions that current frameworks cannot answer: Do Arctic-adapted psychological traits persist independently of later cultural developments like Confucianism and agricultural intensification? Can we detect genomic gradients that mirror the psychological trait clines observed from Northeast to Southeast Asia? Does Sun's theory account for psychological patterns in Native American populations separated from Asian populations for over 15,000 years?

By testing these mechanisms empirically, this study moves beyond Holocene-focused models toward a more complete evolutionary framework that incorporates the formative influence of Pleistocene environments on human psychology.

Population genetics rationale for using split-half reliability of polygenic scores

Population differences in the mean of a quantitative trait due to positive covariances – that is, (cross-population) linkage disequilibrium (LD) – between distant variants can arise for polygenic traits under divergent selection. In practice, this happens when alleles with similar effects are driven to similar frequencies within populations across multiple loci (Latta, 1998; Le Corre and Kremer, 2003; Ma et al., 2010). The other component of genetic differentiation in quantitative traits, Fst, does not take into account this covariance of allelic effects and it was shown to be usually very small for highly polygenic traits subject to divergent selection pressures (Berg and Coop, 2014).

Fst quantifies genetic differentiation through the average variance of individual allele frequencies across populations. Formally, for locus i with allele frequency p_i in a given population, Fst depends on Var(p_i), which is non-negative and discards sign information. Consequently, when estimating population mean differences in a polygenic trait – defined as Δµ = Σ_i β_i Δp_i where _βi is the effect size and Δp_i is the allele frequency difference – the summation exhibits sign dependence that Fst ignores.

Crucially, two distinct scenarios demonstrate Fst's limitations:

When β_i and Δp_i are consistently aligned (e.g., β_i > 0 and Δp_i >0 for trait-increasing alleles), Δµ is large.

When β_i Δp_i terms have opposing signs (e.g., β_i > 0 with Δp_i < 0 at some loci, while β_j < 0 with Δp_j >0 at others), summation cancels out (Δµ ≈ 0).

Fst is identical in both scenarios if Σ(Δp_i)² is equal, yet only scenario (1) shows trait differentiation. This sign-agnostic property causes systematic underestimation of trait differences when alignment exists. Conversely, high Fst (e.g., > 0.15) can coexist with Δµ ≈ 0 when Cov(β, Δp) ≈ 0, as demonstrated by Le Corre & Kremer, 2012. The covariance structure driving alignment can be assessed via split-half reliability of polygenic scores (G = Σ_i β_i p_i), measured as Corr(G_A, G_B) for disjoint locus partitions A and B.

To avoid dependence on any single partition, we computed split-half reliability with 1,000 randomized SNP partitions using the splitHalf routine in R and report the mean correlation across splits. This assesses the internal consistency of PGS across disjoint SNP sets; it is not taken as exclusive evidence for selection, as such covariances can also reflect assortative mating or stratification

Modern genome samples

Samples were collected from a variety of public databases and were classified by the technology employed to genotype the DNA (micro-array chips vs whole-genome sequencing (WGS).

The WGS dataset was comprised the following: 1000 Genomes (The 1000 Genomes Project Consortium, 2015), gnomAD (Karczewski et al., 2020), SweGen (Rentoft et al., 2019), Genome of the Netherlands (GoNL, nlgenome.nl), 1000 Polish genomes (Kaja et al., 2022), NARD (Yoo et al., 2019), Turkish Genome Project (Alkan et al., 2014), Taiwan Genomes (Hsu et al., 2021), Mexico City Prospective Study (Ziyatdinov et al., 2023), Korean Variant Archive (Lee et al., 2022), ABraOM (Naslavsky et al., 2022), DanMAC5 (https://danmac5.cpr.ku.dk/), and a Saudi Arabian WGS study of 302 individuals (Malomane et al., 2025) covering 51 ethnic groups in total.

The array dataset was compiled from a variety of open access online resources (see Piffer & Kirkegaard, 2024b). We imputed the datasets after QC using TOPMed imputation server (Das et al., 2016) with Minimac imputation (Fuchsberger et al., 2014), covering 102 ethnic groups in total.

Ancient genome samples

The imputed samples were derived from two prior studies (Piffer & Kirkegaard, 2024a; Piffer, 2025). The western Eurasian ancient DNA dataset was augmented with additional genomes. Imputation procedures followed the methodology outlined in Piffer (2025) and Piffer & Kirkegaard (2024a). Raw sequencing data (.bam and .fastq files) were obtained from the European Nucleotide Archive (ENA) and NCBI using the following accession numbers and corresponding publications: PRJEB30874 (Olalde et al., 2019); PRJEB30985 (Villalba-Mouco et al., 2019); PRJEB36208 (Rivollat et al., 2020); PRJEB37057 (Agranat-Tamir et al., 2020); PRJEB37660 (Saupe et al., 2021); PRJEB39040 (Prüfer et al., 2021); PRJEB39134 (Hajdinjak et al., 2021); PRJEB53564 (Antonio et al., 2024); PRJEB58642 (Villalba-Mouco et al., 2019); PRJEB59976 (Ruoyun et al., 2024); PRJEB6272 (Lazaridis et al., 2014); PRJEB64496 (Bennett et al., 2023); PRJEB77116 (Ravasini et al., 2024); PRJNA1023512 (Tian et al., 2024); SRP132561 (Amorim et al., 2018); SRP423598 (Vyas et al., 2023); PRJEB81467 (Lazaridis et al., 2025); PRJEB81468 (Nikitin et al., 2025).

Admixture analysis

Admixture components for the samples were computed using ADMIXTURE software (Alexander et al., 2009), a powerful tool for estimating individual ancestries from multilocus SNP genotype datasets. The ADMIXTURE analysis works by decomposing genotype data of each individual into fractions representing potential ancestral populations. Admixture data was merged with the PGS dataset in R by organizing individuals based on their FAM

identifiers.

GWASs used for polygenic scores

We utilized the largest and most recent GWAS on the Big Five personality traits, encompassing 46 cohorts with 611,000 to 1.14 million participants of European and African ancestry (Schwaba et al., 2025). Polygenic scores accounted for approximately 9% of the variance in each trait, increasing to 10%–16% after correcting for measurement unreliability. Genetic influences on personality were highly consistent across geographic regions, raters (self vs. close others), age groups, and assessment methods—though Agreeableness showed lower consistency across geography and rater instruments. A key strength of this GWAS was that polygenic scores maintained predictive validity in within-family analyses, suggesting minimal confounding by shared environmental factors.

Polygenic scores (PGS) were calculated using the most recent and extensive GWAS for each trait. To filter variants, we applied clumping and thresholding (C + T) using PLINK 1.9 (Chang et al., 2015), setting a standard GWAS p-value threshold of p < 5 × 10 − 8 with a linkage disequilibrium (LD) of r2 < 0.1. Allele frequencies were computed using PLINK 2.0 (Chang et al., 2020). The frequency files were loaded into R (version 4.4.1) (R Core Team, 2023) and merged with the GWAS summary files to compute the polygenic

scores.

Summary of GWAS data

Tables 1 and 2 present the following data for each GWAS:

The number of independent SNPs after linkage disequilibrium (LD) clumping and thresholding (r² < 0.1 and p < 5 × 10⁻⁸).

The number of matching SNP IDs found in the whole-genome sequencing (WGS) and WGS + Array datasets.

The split-half r of the resulting polygenic scores.

Table 1
WGS dataset
GWAS	N independent SNPs	N matching IDs	% matching	Split-half r
Agreeableness	39	31	79.49	0.262
Extraversion	258	219	84.88	0.786
Neuroticism	703	563	80.08	0.716

Table 2
WGS + Array dataset
GWAS	N independent SNPs	N matching IDs	% matching	Split-half r
Agreeableness	39	15	38.46	0.039
Extraversion	258	114	44.19	0.241
Neuroticism	703	286	40.68	0.266

Modern WGS sample

Using whole-genome sequencing data, we derived polygenic scores for traits A, E, and N across a globally diverse sample of 51 modern populations.

East Asians had the lowest Extraversion and Neuroticism PGS and the second highest Agreeableness PGS, following Indigenous Mexicans (“Mexico IMX”) and other admixed Amerindian groups (Suppl. Figures 1,2,3).

The factor analysis of three polygenic scores (Agreeableness [AGR], Extraversion [EXT], and Neuroticism [NEU]) revealed strong evidence for a single underlying genetic factor. The correlation matrix showed expected patterns, with AGR negatively correlated with both EXT (r = -0.76) and NEU (r = -0.75), while EXT and NEU were positively correlated (r = 0.93).

All variables demonstrated excellent factor loadings (|0.78–0.97|), with EXT showing the strongest loading (0.97) and AGR the weakest (-0.78). The model explained 82.1% of the total variance, indicating excellent explanatory power. Residual diagnostics confirmed perfect model fit, with no absolute residuals exceeding 0.00 (all < 0.05 threshold).

Reliability analysis using Omega Total (ω = 0.93) indicated exceptional internal consistency for the single-factor solution.

The populations with the highest factor scores were Papuans and East Asians, followed by Amerindians (Fig. 1).

A modest but statistically significant correlation emerged between latitude and the personality factor score (PFS) (r = 0.38, p = 0.027). However, this linear association masked a more complex curvilinear pattern (Fig. 2), wherein European populations—despite residing at higher latitudes—exhibited lower PFS values than East Asians. This nonlinearity suggests that latitude alone cannot fully explain global PFS variation and implies the involvement of additional biogeographic or cultural factors.

Since continental ancestry was a key driver of the relationship, we ran a regression using superpopulation as a categorical predictor (with East Asians [EAS] as the reference group) and Personality Factor Score (PFS) as the dependent variable. The model explained 97.5% of variance in PFS (F(5, 26) = 199.97, p < 0.001), with all superpopulation coefficients showing statistically significant differences from the reference group (East Asians) (Table 3). African ancestry showed the strongest negative association (β = -2.80, p < 0.001), followed by Middle Eastern/North African (β = -1.35, p < 0.001), while European (β = -0.72, p < 0.001) and South Asian (β = -0.67, p < 0.001) ancestries showed more moderate effects.

Table 3
Regression Results Table. D.V.= PFS
Superpop	Estimate (β)	Std. Error	t-value	p-value
(Intercept)	0.95	0.058	16.386	< 0.001
European	-0.717	0.084	-8.493	< 0.001
African	-2.8	0.092	-30.563	< 0.001
American	-0.546	0.136	-4.019	< 0.001
Middle East/N.Af	-1.353	0.116	-11.673	< 0.001
South Asian	-0.673	0.104	-6.445	< 0.001
Adj. R²	0.97

WGS + Array sample

Using whole-genome sequencing and imputed array data, we derived polygenic scores for traits A, E, and N across a globally diverse sample of 102 modern populations.

Among the groups analyzed, Arctic people (Athabaskans) and Native Americans had the highest polygenic scores (PGS) for Agreeableness, while Africans had the lowest. For Extraversion, Southeastern Asian populations (e.g., Miao, Gelao-Li-Zhuang) showed the lowest PGS, whereas Africans had the highest. Similarly, Mongolic and Turkic groups (e.g., Salar, Baoan) exhibited the lowest Neuroticism PGS, while Africans again scored the highest (Suppl. Figures 4–6).

The factor analysis of three polygenic scores revealed strong evidence for a single underlying genetic factor. The correlation matrix showed expected patterns, with AGR negatively correlated with both EXT (r = -0.51) and NEU (r = -0.67), while EXT and NEU were positively correlated (r = 0.85).

All variables demonstrated excellent factor loadings (0.647-1), with EXT showing the strongest loading (1) and AGR the weakest (-0.65). The model explained 70.5% of the total variance, indicating excellent explanatory power. Residual diagnostics confirmed perfect model fit, with no absolute residuals exceeding 0.00 (all < 0.05 threshold).

Reliability analysis using Omega Total (ω = 0.87) indicated exceptional internal consistency for the single-factor solution.

The populations with the highest factor scores were Central Asians (Sherpa, Salar, Tibetan, Baoan), followed by South Asians and East Asians (Fig. 3).

A modest but statistically significant correlation emerged between latitude and the personality factor score (PFS) (r = 0.47, p < 0.001).

Figure 4.

Since continental ancestry was a key driver of the relationship, we ran a regression using superpopulation as a categorical predictor (with East Asians [EAS] as the reference group) and Personality Factor Score (PFS) as the dependent variable. The model explained 84.3% of variance in PFS (F(9,85) = 57.11, p < 0.001)) (Table 4). African ancestry showed the strongest negative association (β = -2.80, p < 0.001), followed by Middle Eastern/North African (β = -1.35, p < 0.001), while Central Asian (β = -0.023, p = 0.92) and Arctic (β = -0.34, p = 0.175) ancestries were not significantly different.

Table 4
Regression Results Table. D.V.= PFS
Variable	Estimate	Standard Error	t value	p value
(Intercept)	0.846	0.096	8.81	0
AFR	-2.798	0.14	-19.939	0
Admix. American	-0.93	0.22	-4.223	0
Arctic	-0.339	0.248	-1.367	0.175
Central Asian	-0.023	0.22	-0.104	0.917
EUR	-0.605	0.132	-4.578	0
MENA	-1.348	0.163	-8.254	0
Nat. Am.	-0.478	0.248	-1.925	0.058
SAS	-0.41	0.14	-2.921	0.004
SE Asian	-0.399	0.188	-2.119	0.037
Adj. R²	0.843

Ancient DNA

Western Eurasian ancient DNA

The summary statistics regarding Years BP, Coverage, Latitude and Longitude are presented in Table 5.

Table 5
Summary statistics
Variable	n	mean	min	max
Years BP	3950	3417.420127	181	44255
Coverage	4116	1.323574953	0	69.4
Latitude	4130	47.10338112	30.649857	69.65
Longitude	4130	15.71160062	-50.1	91.02565

Agreeableness

We ran a regression with Agreeableness PGS as the dependent variable and Years B.P., Latitude, Longitude, and Coverage as independent variables. The model was statistically significant (F(4, 3932) = 4.901, p < .002). Years BP showed the strongest association (β = 0.051, p < .001), indicating that Agreeableness PGS decreased in more recent epochs. Latitude and Coverage were non-significant (Table 6).

Table 6
Regression results table. D.V.= Agreeableness PGS
Variable	Estimate (β)	Std. Error	t-value	p-value
(Intercept)	0	0.016	0	1
Years BP	0.051	0.016	3.184	0.001
Latitude	-0.027		-1.660	0.097
Coverage	0.002		0.119	0.905
Adj. R²	0.003

Extraversion

We ran a regression with Extraversion PGS as the dependent variable and Years B.P., Latitude, and Coverage as independent variables. The model was statistically significant (F(4, 3932) = 22.960, p < .001). Latitude showed the strongest association (β = −0.13, p < .001), indicating that Extraversion PGS decreased farther from the equator. Coverage (β = −0.035, p = .027) had smaller but significant effects, while Years B.P. was non-significant (β = −0.02, p = .272) (Table 7).

Table 7
Regression results table. D.V.= Extraversion PGS
Variable	Estimate (β)	Std. Error	t-value	p-value
(Intercept)	0	0.016	0	1
Years BP	-0.018	0.016	-1.091	0.272
Latitude	-0.129	0.016	-8.064	< 0.001
Coverage	-0.035	0.016	-2.216	0.027
Adj. R²	0.019

Neuroticism

We ran a regression with Neuroticism PGS as the dependent variable and Years B.P. (Table 8). Latitude and Coverage as independent variables. The model was statistically significant (F(4, 3932) = 27.888, p < .001). Latitude showed the strongest association (β = −0.120, p < .001), indicating that Neuroticism PGS decreased farther from the equator. Coverage (β = −0.068, p < .001) had smaller but significant effects.

Table 8
Regression results table. D.V.= Neuroticism PGS
Variable	Estimate (β)	Std. Error	t-value	p-value
(Intercept)	0	0.016
Years BP	-0.065	0.016	-4.103	< 0.001
Latitude	-0.120	0.016	-7.499	< 0.001
Coverage	-0.068	0.016	-4.285	< 0.001
Adj. R²	0.020

Assessing the effects of the post-Last Glacial Maximum on AGR, EXT, NEU PGS and the PFS.

To examine temporal patterns in polygenic scores, we modeled sample age as a continuous variable (Years BP) rather than grouping observations into discrete periods. Specifically, we employed hinge (piecewise) regression models with knots placed at 12,000 BP (Holocene onset) and 19,000 BP (end of the Last Glacial Maximum) to test for slope changes in the relationship between age and standardized PGS values. This approach allows for a continuous trajectory with a potential change in slope at key paleoclimatic thresholds, without imposing an artificial discontinuity in the fitted line. Scatterplots of Years BP against PGS (z-scores) were created for visualization, with fitted hinge regression lines overlaid to illustrate the estimated temporal trends.

In these models, the standardized PGS was regressed on (a) sample age relative to the knot (in kyr), (b) the hinge term capturing slope change beyond the knot, and (c) covariates for latitude and sequence coverage to adjust for geographic and data quality effects. The pre-knot slope, the post-knot slope (derived from the pre-knot slope plus the hinge coefficient), and the significance of the slope change at the knot together indicate whether temporal trajectories in PGS reversed direction at 12k or 19k BP.

Agreeableness

Inspection of the plot revealed that temporal trends differed among the two groups (Fig. 4).

Using hinge regressions with knots at 12,000 and 19,000 years BP, we observed consistent patterns across specifications. Prior to the knot, the Agreeableness PGS increases significantly with age (12k model: β = +0.0300 SD per kyr, p = 3.41×10⁻⁵; 19k model: β = +0.0267 SD per kyr, p = 7.99×10⁻⁵). At the knot, the slope shifts downward (12k model: Δβ = −0.0383 SD per kyr, p = 0.0068; 19k model: Δβ = −0.0466 SD per kyr, p = 0.0172), yielding post-knot slopes of − 0.0083 SD/kyr (12k) and − 0.0199 SD/kyr (19k). Latitude and coverage were not significant in either specification. Overall explanatory power is small (R² ≈ 0.5% for both thresholds), but the slope reversal is statistically reliable: Agreeableness PGS trends upward with age within the younger interval and trends downward beyond the knot.

Table 9
Regression results table. D.V.= Extraversion PGS
Term	12k				19k
	Estimate	SE	t	p	Sig	Estimate	SE	t	p	Sig
(Intercept)	0.4094425	0.118576	3.453	5.60E-04	***	0.5734016	0.1408466	4.071	4.77E-05	***
Date_kyr	0.0300297	0.0072375	4.149	3.41E-05	***	0.0267066	0.0067633	3.949	7.99E-05	***
hinge_knot	-0.0383457	0.0141525	-2.709	0.00677	**	-0.0466476	0.0195716	-2.383	0.0172	*
Latitude	-0.0030607	0.0023353	-1.311	0.19006		-0.0032266	0.0023323	-1.383	0.1666
Coverage	0.0002267	0.0064623	0.035	0.97201		0.0005875	0.006461	0.091	0.9276

Extraversion

Inspection of the plot revealed that temporal trends differed among the two groups (Fig. 5).

Using standardized polygenic scores as the dependent variable, hinge regression models with knots at 12,000 and 19,000 years BP produced highly similar qualitative patterns (Table 10). In both models, the slope of the PGS trajectory was significantly negative prior to the knot (12k model: β = −0.0188 SD per kyr, p = 0.0087; 19k model: β = −0.0182 SD per kyr, p = 0.0067). After the knot, the slope changed direction, with a significant positive increment (12k model: Δβ = +0.0358 SD per kyr, p = 0.0108; 19k model: Δβ = +0.0537 SD per kyr, p = 0.0056). This yielded post-knot slopes of + 0.0169 SD/kyr (12k) and + 0.0356 SD/kyr (19k). Latitude and coverage were consistently negative and significant predictors across both models ( ≈ − 0.019 SD per degree latitude; ≈ −0.014 SD per unit coverage). Model fit was modest (R² ≈ 0.019 for both thresholds), with the 19k knot offering a slightly better fit and a steeper positive post-knot slope. Taken together, these results indicate a reversal in the temporal trend of standardized PGS values, with a decline toward the knot followed by a subsequent increase, whether the boundary is placed at the onset of the Holocene (12k BP) or near the end of the Last Glacial Maximum (19k BP).

Table 10
Regression results table. D.V.= Extraversion PGS
Term	12k					19k
	Estimate	SE	t	p	Sig	Estimate	SE	t	p	Sig
(Intercept)	0.749489	0.117506	6.378	2.00E-10	***	0.62819	0.13953	4.502	6.92E-06	***
Date_kyr	-0.018816	0.007172	-2.624	0.00874	**	-0.01817	0.0067	-2.711	0.00673	**
hinge_knot	0.035754	0.014025	2.549	0.01083	*	0.05373	0.01939	2.771	0.00561	**
Latitude	-0.019265	0.002314	-8.325	< 2e-16	***	-0.01925	0.00231	-8.331	< 2e-16	***
Coverage	-0.013684	0.006404	-2.137	0.03267	*	-0.01398	0.0064	-2.184	0.029	*

Neuroticism

Inspection of the plot revealed that temporal trends differed among the two groups (Fig. 6).

Using standardized Neuroticism PGS as the dependent variable, hinge regressions with knots at 12,000 and 19,000 years BP show highly similar qualitative patterns (Table 11). In both models, the pre-knot time slope is significantly negative (12k model: β = −0.0509 SD per kyr, p = 1.4×10⁻¹²; 19k model: β = −0.0448 SD per kyr, p = 2.38×10⁻¹¹). At the knot, there is a significant positive change in slope (12k: Δβ = +0.0820 SD per kyr, p = 4.96×10⁻⁹; 19k: Δβ = +0.1043 SD per kyr, p = 7.64×10⁻⁸), yielding post-knot slopes of + 0.0312 SD/kyr (12k) and + 0.0594 SD/kyr (19k). Latitude and coverage are consistently negative and significant covariates in both specifications (latitude ≈ − 0.0189 SD per degree; coverage ≈ − 0.026–0.027 SD per unit). Intercepts differ across models (significant at 12k; n.s. at 19k), but are not substantively meaningful given centering/scaling. Overall, the results indicate a reversal in the temporal trend of Neuroticism PGS—declining up to the knot, then increasing thereafter—with the 19k knot implying a somewhat steeper post-knot rise.

Table 11
Regression results table. D.V.= Neuroticism PGS
Term	12k				19k
Term	Estimate	SE	t	p	Sig	Estimate	SE	t	p	Sig
(Intercept)	0.482895	0.117219	4.12	3.87E-05	***	0.211527	0.1393	1.519	1.29E-01
Date_kyr	-0.050849	0.007155	-7.107	1.40E-12	***	-0.044815	0.006689	-6.7	2.38E-11	***
hinge_knot	0.082007	0.013991	5.862	4.96E-09	***	0.104248	0.019357	5.386	7.64E-08	***
Latitude	-0.018938	0.002309	-8.203	3.14E-16	***	-0.018643	0.002307	-8.082	8.39E-16	***
Coverage	-0.026316	0.006388	-4.119	3.88E-05	***	-0.027069	0.00639	-4.236	2.33E-05	***

Factor analysis at the group level

At the population level, inter-trait correlations among mean PGS were modest (E–N: r = 0.27; A–N: r = − 0.19; E–A: r = − 0.03). Factorability was borderline by KMO (0.503) but supported by Bartlett’s test (χ²(3) = 10.50, p = 0.0148). Parallel analysis indicated a single common factor. A one-factor solution (scores obtained via principal-axis factoring) captured a weak but coherent dimension dominated by Neuroticism with smaller, opposing contributions of Extraversion and Agreeableness; for interpretability, scores were oriented so that higher values indicate greater agreeableness and emotional stability (A↑, N↓, E↓).

Hinge regressions of the group-level factor (z-scored) on time (Years BP), with knots at 12,000 and 19,000 BP, controlled for latitude and coverage and used precision-motivated weights (√N, softly scaled by coverage) (Table 12).

Knot at 12,000 BP. Model fit: R² = 0.418 (adjusted R² = 0.393), F(4, 92) = 16.52, p = 3.10×10⁻¹⁰. The pre-knot slope was + 0.109 SD per kyr (SE = 0.027, t = 4.07, p = 9.97×10⁻⁵). The hinge term was − 0.168 SD per kyr (SE = 0.042, t = − 4.03, p = 1.17×10⁻⁴), implying a post-knot slope of − 0.058 SD per kyr. Latitude and coverage were positive covariates (Latitude: β = +0.060 SD/°, SE = 0.010, t = 5.86, p = 7.28×10⁻⁸; Coverage: β = +0.216 SD/unit, SE = 0.064, t = 3.36, p = 0.0011).
Knot at 19,000 BP. Model fit: R² = 0.386 (adjusted R² = 0.360), F(4, 92) = 14.48, p = 3.32×10⁻⁹. The pre-knot slope was + 0.083 SD per kyr (SE = 0.0248, t = 3.33, p = 0.00123). The hinge term was − 0.175 SD per kyr (SE = 0.0536, t = − 3.26, p = 0.00159), yielding a post-knot slope of − 0.092 SD per kyr. Latitude and coverage again entered positively (Latitude: β = +0.060, p = 1.32×10⁻⁷; Coverage: β = +0.227, p = 8.78×10⁻⁴).

Table 12
Regression results table. D.V.= “Arctic” Personality Factor
Term	12k					19k
	Estimate	SE	t	p	Sig	Estimate	SE	t	p	Sig
(Intercept)	-2.095	0.543	-3.86	0.000211	***	-1.767	0.638	-2.769	0.0068	**
Date_kyr	0.11	0.027	4.069	0.0000997	***	0.083	0.025	3.334	0.00123	**
hinge_knot	-0.168	0.042	-4.025	0.000117	***	-0.175	0.054	-3.255	0.00159	**
Latitude	0.06	0.01	5.856	0.0000000728	***	0.06	0.011	5.718	0.000000132	***
Coverage	0.216	0.064	3.364	0.00112	**	0.227	0.066	3.44	0.000878	***
Model R²	0.418					0.386
Adj. R²	3.93E-01					0.36
F-statistic	16.52					14.48
Model p	3.10E-10					0.00000000332

Overall, the factor (higher = more agreeable and emotionally stable) increases prior to each knot and declines thereafter, with covariates behaving consistently across specifications.

Eastern Eurasian ancient DNA

The summary statistics for Years BP, Coverage, and Latitude are reported in Table 12. Because only a handful of genomes predate 12,000 BP, we did not attempt to model LGM-specific effects in this dataset. For comparability with earlier work (Piffer, 2025), we applied the same K = 8 admixture model. The components were labeled according to their distinctive frequency patterns across populations. For example, one component dominated the ancestry of Afanasievo and Finnish individuals (> 90%), was common in Xinjiang (~ 50%) and Turkic/Upper Paleolithic Siberians (> 25%), but rare in East Asians; this was therefore labeled “Afanasievo-Yamnaya”. Another component was nearly fixed in the Jomon (~ 90%) yet occurred at lower frequencies in Koreans (~ 10%) and minimally in other East Asians, leading us to label it “Jomon”. The overall structure of these ancestry components across cultural groups is illustrated in Fig. 8.

Table 12
Summary statistics
Variable	n	mean	min	max
Date	937	3089.009104	141.5	40000
Coverage	1105	1.75371659	0.001	32.291
Latitude	1104	40.21091058	3.533	70.7
Longitude	1104	104.2600836	22.4	204.41*

*Because a negative value would not make sense in this context, we compute longitude east on a 0-360 scale.

Agreeableness

We ran a regression with Agreeableness PGS as the dependent variable and Years B.P., Latitude, Longitude, and Coverage as independent variables. The model was statistically significant (F(3, 933) = 2.713, p = 0.044). Only Coverage had a significant effect on Agreeableness PGS (Table 13).

Table 13
Regression results table. D.V.= Agreeableness PGS
Variable	Estimate (β)	Std. Error	t-value	p-value
(Intercept)	0	0.033
Years BP	-0.045	0.033	-1.353	0.177
Latitude	-0.007	0.035	-0.212	0.832
Coverage	0.088	0.033	2.668	0.008
Adj. R²	0.005

Jomon and Tibetan ancestries had a significant positive effect on Agreeableness PGS (Fig. 9).

Extraversion

We ran a regression with Extraversion PGS as the dependent variable and Years B.P., Latitude, Longitude, and Coverage as independent variables. The model was statistically significant (F(4, 932) = 5.458, p < .001). Only longitude was a significant predictor of Extraversion PGS(β = -0.158, p < .001), indicating that more easterly populations had lower Extraversion PGS (Table 14).

Table 14
Regression results table. D.V.= Extraversion PGS
Variable	Estimate (β)	Std. Error	t-value	p-value
(Intercept)		0.033
Years BP	0.020	0.033	0.603	0.546
Latitude	0.010	0.033	0.295	0.768
Coverage	0.017	0.033	0.501	0.617
Adj. R²	-0.002

All the ancestries except Tibetan tended to have lower Extraversion PGS than the reference group (Southern Chinese) (Fig. 10).

Neuroticism

We ran a regression with Neuroticism PGS as the dependent variable and Years B.P., Latitude, Longitude, and Coverage as independent variables. The model was statistically significant (F(4, 932) = 3.592, p = .006). Only longitude was a significant predictor of Neuroticism PGS(β = -0.117, p < .001), indicating that more easterly populations had lower Neuroticism PGS (Table 15).

Table 15
Regression results table. D.V.= Neuroticism PGS
Variable	Estimate (β)	Std. Error	t-value	p-value
(Intercept)	0	0.032
Years BP	0.044	0.033	1.326	0.185
Latitude	0.009	0.033	0.287	0.774
Coverage	-0.039	0.033	-1.171	0.242
Adj. R²	0

All the ancestries except Afanasievo/Yamnnaya tended to have lower Extraversion PGS than the reference group (Southern Chinese) (Fig. 11).

Factor analysis at the individual level

We analyzed standardized individual PGS for Agreeableness (AGR), Extraversion (EXT), and Neuroticism (NEU). We inspected the trait correlation matrix and evaluated sampling adequacy (KMO) and sphericity (Bartlett’s test). We then ran a maximum-likelihood, unrotated one-factor model with regression-based scores. Because Extraversion showed a near-zero loading and parallel analysis indicated no reliable common factor, we operationalized a two-indicator latent dimension using AGR and the sign-reversed NEU (− NEU) to summarize “agreeableness–emotional stability.” Factor scores were computed via regression and z-standardized for analysis. As a robustness check, we also computed the first principal component (PC1).

Correlations among traits were small (AGR–EXT r = − 0.04; AGR–NEU r = − 0.17; EXT–NEU r = 0.02). KMO was borderline (overall = 0.50; MSAs: AGR = 0.50, EXT = 0.55, NEU = 0.50), while Bartlett’s test rejected the identity matrix (χ²(3) = 34.79, p = 1.35×10⁻⁷). Parallel analysis suggested zero common factors. A one-factor ML model nonetheless yielded weak structure (loadings: AGR = 0.631, EXT ≈ 0, NEU = − 0.274; SS loadings = 0.477, 15.9% variance). Consistent with these diagnostics, we summarized individual differences with a two-indicator factor based on AGR and − NEU, oriented so that higher scores indicate greater agreeableness and emotional stability; PC1 was retained for robustness checks.

We regressed the individual latent score (higher = more agreeable & emotionally stable) on standardized covariates and standardized ancestry components (with Southern_Chinese omitted as the baseline) (Fig. 12). Temporal and data-quality controls were null (Date: β = −0.04, p = .21; Latitude: β = 0.07, p = .33; Coverage: β = −0.01, p = .74). In contrast, several ancestry components showed positive associations (standardized betas, 95% CI): Tibetan β = 0.23 [0.12, 0.33], Northeast Chinese β = 0.20 [0.12, 0.27], Jōmon β = 0.18 [0.12, 0.24], SE Asian β = 0.14 [0.05, 0.23], Arctic β = 0.13 [0.03, 0.22], and Siberian β = 0.11 [0.01, 0.21]; Afanasievo/Yamnaya was not significant (β = 0.10, p = .057).

Factor analysis at the population level

At the culture level, we mapped populations to predefined culture labels, removed empty/NA labels, and then averaged standardized PGS for Agreeableness (AGR), Extraversion (EXT), and Neuroticism (NEU) within each culture. We conducted a maximum-likelihood, unrotated one-factor analysis on the culture-level means and extracted regression-based factor scores. Sampling adequacy (KMO) and Bartlett’s test indicated the data were suitable for a common factor. Factor scores were z-standardized across cultures (reported as factor_z); the sign of the factor is arbitrary.

Culture-level correlations (traits).

AGR–EXT: r = − 0.345; AGR–NEU: r = − 0.452; EXT–NEU: r = 0.527. These are appreciably stronger than at the individual level, consistent with noise reduction after aggregation.

Sampling adequacy at the culture level was acceptable (KMO = 0.64; item MSAs: AGR = 0.71, EXT = 0.65, NEU = 0.61), and Bartlett’s test rejected the identity matrix (χ²(3) = 9.86, p = 0.0198). A one-factor maximum-likelihood solution (unrotated) on culture means yielded loadings of − 0.544 (AGR), 0.634 (EXT), and 0.831 (NEU), explaining 46.3% of the variance (SS loadings = 1.389). As scaled, higher scores correspond to higher Neuroticism and Extraversion and lower Agreeableness, noting the factor sign is arbitrary.

At the culture level, a one-factor ML solution explained 46.3% of the variance (SS loadings = 1.389). We oriented the factor so that higher values indicate greater agreeableness and emotional stability; under this orientation the factor loads positively on AGR and negatively on NEU and EXT.

Across cultures, the z-standardized factor (higher = more agreeable & emotionally stable) shows a broad spread with one extreme outlier (Fig. 13). The highest values are observed for Arctic (1.05), Tianyuan (1.02), NE China (0.84), Jomon (0.75), Hexi_corridor (0.74), and Korea (0.64), followed by Afanasievo (0.59), SE Asia (0.49), Mongol (0.49), and Siberian (0.23). Near the grand mean (≈ 0) are Tibet (− 0.01), S China (− 0.07), and Turk (− 0.12). Negative scores cluster for Island SEA (− 0.25), Kofun (− 0.38), Amur (− 0.52), Boysman Russia (− 0.60), and Xinjiang (− 0.63), with UP Siberia lower still (− 0.83). Finnish is a pronounced low outlier (− 3.44).

This study presents the first genomic examination of David Sun's (2025) Arctic adaptation hypothesis, which proposes that Late Pleistocene environmental pressures in Northeast Asia selected for reduced Extraversion, suppressed Neuroticism, and elevated Agreeableness - traits subsequently dispersed throughout East Asia. Our analyses reveal coordinated evolution of these traits and biogeographic patterning consistent with cold-climate selection. However, key predictions of Sun’s model received mixed support, and several contradictions necessitate significant theoretical refinement.

Evidence for an integrated trait complex

The strongest support for an integrated evolutionary response emerges from our factor analysis of modern genomes. We identified a robust latent factor characterized by high loadings on low Extraversion (0.97), low Neuroticism (0.89), and high Agreeableness (0.78). This personality factor score (PFS) suggests coordinated evolution of these traits under shared selective pressures, aligning with Sun’s proposal of a co-adapted psychological syndrome. Crucially, this complex shows peak expression in populations with deep Northeast Asian ancestry (East Asians, Native Americans; Suppl. Figures 1–6), consistent with descent from populations exposed to intense Pleistocene cold stress.

Climatic and geographic correlates

Latitude effects support cold adaptation as a key selective driver, though not limited to Arctic populations. Globally, we found significant positive correlations between PFS and latitude (r = 0.38–0.47, p < 0.05), with Central Asian and Siberian groups (e.g., Sherpa, Salar) showing particularly high scores (Fig. 3). The elevated PFS in Sherpa populations may also reflect adaptation to high-altitude cold climates.

In ancient Western Eurasia, hinge regressions revealed sharp climatic contingencies. At the population level, a single latent factor extracted from Extraversion, Agreeableness, and Neuroticism polygenic scores showed a clear biphasic trajectory. With knots placed at 12,000 and 19,000 years BP, both models yielded significant slope breaks (Δβ ≈ −0.168 to − 0.175, both p < 0.002), indicating that the factor increased with sample age prior to the knots and declined thereafter. Specifically, pre-knot slopes were positive (β ≈ +0.083 to + 0.110 SD per kyr, both p < 0.001), while post-knot slopes turned negative (β ≈ −0.058 to − 0.092 SD per kyr). These reversals are consistent with the hypothesized late Pleistocene–early Holocene transition in selection pressures (Fig. 7).

The models also revealed robust spatial effects. Latitude exerted a strong positive influence on the factor (β ≈ +0.060 SD/°; p < 1e − 7), in line with predictions that Arctic-adapted personality profiles would be more pronounced in higher-latitude populations. This finding indicates that the temporal breakpoints were embedded within a persistent spatial cline: colder and more northerly regions consistently favored higher scores on the composite of Agreeableness and Emotional Stability. Coverage contributed a smaller but reliable positive effect (β ≈ +0.216 to + 0.227 SD/unit; p < 0.001), which is best interpreted as a data-quality correlate rather than a substantive driver.

Overall model fits were moderate (R² = 0.39–0.42), underscoring that the break in temporal slope is both statistically robust and theoretically consistent. With the factor oriented to load positively on Agreeableness and negatively on Extraversion and Neuroticism, the trajectory matches the a priori predictions: late Pleistocene cold intensified selection against Extraversion and Neuroticism while favoring Agreeableness, whereas Holocene conditions reversed or attenuated these trends. The larger post-knot decline in the 19,000-year model is compatible with versions of the hypothesis that locate the inflection nearer the Last Glacial Maximum. These conclusions remain subject to standard cautions about ecological inference and residual structure; nevertheless, the convergence of results across knot placements, weighting schemes, and spatial covariates strengthens the link between the observed breakpoints and the theoretically specified transitions.

Patterns in the Eastern dataset cannot currently be linked to pre-/post-LGM dynamics because we had only a handful of pre-12 kya genomes and therefore did not test hinge (12k/19k) effects in this region. Within the available (mostly Holocene) time range, regression models showed weak or absent main effects of latitude and time. Agreeableness was largely explained by coverage; apparent longitudinal gradients in Extraversion and Neuroticism disappeared after controlling for ancestry components; and several Southeast Asian ancestries were associated with lower Extraversion and Neuroticism than Southern Chinese (Figs. 8–9), contradicting the dilution hypothesis. The persistence of low Extraversion and Neuroticism in Eastern populations, even without post-LGM reversal, suggests that these traits were stabilized through Holocene cultural processes (e.g., rice agriculture, Confucian institutions) rather than counterselected, unlike in Western Eurasia.

In the ancient East Asian dataset, cultures with clear ties to Late Pleistocene Northeast Asian lineages (Arctic, Tianyuan, NE China, Jomon, Korea) sit high on the agreeableness–emotional-stability factor, but Amur, Boysman Russia, and Upper Paleolithic Siberia fall lower despite cold climates. This discrepancy suggests that temperature alone is not the operative driver; what matters is the specific social-ecological package associated with interior, highly seasonal, resource-scarce winter ecologies that Sun’s model emphasizes. Amur and Boysman are coastal/maritime contexts with different subsistence profiles, social networks, and seasonal risk structures; UP Siberia is older and plausibly pre-selection-peak relative to the LGM window.

Second, our ancestry signals point to which cold-adapted lineages matter. The individual-level regressions show positive associations for Northeast Chinese, Tibetan, Jōmon, Siberian, Arctic components, but Afanasievo/Yamnaya is not reliably positive. Cultures with greater Western Eurasian/Steppe contribution may therefore trend lower on this factor even in cold settings, whereas cultures anchored in NE Asian ancestry trend higher. That is consistent with Sun’s claim that the relevant pressures were borne by Northeast Asian populations overwintering in interior subarctic zones, not by all high-latitude populations uniformly.

Third, factor emergence at the culture level (≈ 46% variance) versus weakness within individuals fits the idea that we are detecting a population-structured syndrome. Averaging reduces per-individual aDNA noise and reveals between-culture differences that track ancestry histories and ecologies. The low score for Finnish—outside the core East Asian phylogeny—also cautions that our mixed panel contains lineages shaped by distinct selection histories, further suggesting that “cold ≠ same selection.”

These results do not falsify the Arctic-adaptation hypothesis; rather, they clarify its scope conditions. In the ancient East Asian dataset, high scores on the agreeableness–emotional-stability factor are most consistent with Northeast Asian–derived, interior winter ecologies, whereas some cold-but-maritime (e.g., Amur, Boysman Russia) or distinct cold lineages (e.g., UP Siberia) do not show the same profile. Thus, cold exposure per se is insufficient; the relevant selective pressures appear lineage- and ecology-contingent—particularly those tied to interior, highly seasonal, resource-scarce overwintering—consistent with Sun’s framework. The stronger factor structure at the culture level further suggests a population-structured syndrome that becomes visible when individual-level noise is averaged. A fuller assessment of ecological covariates and alternative ancestry parameterizations would be valuable in follow-up work, but is beyond the scope of the present study.

Critical limitations and contradictions

Despite broad consistencies with Sun’s framework, our results challenge his Temporal Exposure Emphasis (TEE) model, which posits that Arctic/sub-Arctic populations should display stronger trait profiles due to prolonged cold exposure. Athabaskans do not differ significantly from East Asians in PFS (β = −0.34, p = 0.175; Table 4), and Native Americans show only marginal differences (p = 0.058). This implies that the trait complex characterizes Northeast Asian-descended populations broadly, without amplification by extended time in Arctic ecologies. Moreover, the absence of a latitudinal gradient in ancient East Asia stands in contrast to the clear latitude effects in ancient Western Eurasia.

Methodological limitations further constrain interpretation. Reliability of Agreeableness PGS remains low (split-half r = 0.039), reducing confidence in its population-level differences. Our reliance on latitude as a proxy for cold stress oversimplifies climatic complexity; direct measures of winter severity and humidity are needed for stronger inference. Finally, ancient East Asian coverage remains shallow before 12k BP, precluding robust tests of pre-LGM dynamics outside Western Eurasia.

Toward a refined model

The geographic distribution patterns defy expectations of Southeast Asian dilution and instead point to regionally specific adaptations. Western Eurasian genomes show clear biphasic selection, with LGM-driven suppression of Extraversion and Neuroticism followed by Holocene reversals. Eastern Eurasian genomes, in contrast, show persistence without reversal, implying that cultural co-option maintained the trait complex after the LGM. This resolves the “European Enigma”: low PFS in high-latitude Europeans reflects Holocene counterselection rather than lack of Pleistocene adaptation.

Our findings therefore support a layered evolutionary model:

Biogeographic foundations: broad reduction in Extraversion and Neuroticism across Northeast Asian-descended populations.

Pleistocene amplification: intensified selection against these traits during the LGM in Western Eurasia.

Holocene divergence: reversal of selection in Western populations, but cultural reinforcement of the complex in East Asia.

Following Sun (2025), we interpret Confucianism not as the origin but as the cultural codification of Pleistocene-derived values—cooperation, emotional restraint, conflict avoidance—that persisted because they remained adaptive in dense agricultural societies.

Future directions

Future work should prioritize sequencing high-coverage Siberian genomes older than 19k BP, which would allow direct testing of in-situ selection during the LGM. The apparent Southeast Asian anomaly demands clarification through larger ancient DNA datasets that can separate polygenic score limitations from genuine regional adaptations. Finally, modeling how Holocene cultural transitions reinforced, rather than replaced, cold-adapted psychological traits could illuminate why these characteristics persist in East Asia today. Moving beyond latitude proxies to incorporate direct climatic variables will be essential for disentangling environmental from cultural selection forces.

We confirm the existence of a genetically coordinated personality profile—low Extraversion, low Neuroticism, and high Agreeableness—in Northeast Asian-descended populations, likely shaped by ancestral cold stress. Yet its persistence in East Asia, reversal in Europe, and unexpected expression in Southeast Asia challenge a strictly Arctic-focused model. Future frameworks must integrate deep-time climatic selection, Holocene contingencies, and regional cultural adaptations to explain the mosaic of human personality variation. Sun’s hypothesis provides a crucial starting point, but our results call for its expansion into a more nuanced model of psychological evolution.

Author Contribution

I am the sole author of this work and was responsible for all aspects of the study, including conceptualization, methodology, formal analysis, data curation, software, validation, visualization, writing—original draft preparation, and writing—review and editing.

Funding. The authors did not receive support from any organization for the submitted work.

Genomes Project: The 1000 Genomes Project Consortium. (2015). A global reference for human genetic variation. Nature, 526(7571), 68-74. https://doi.org/10.1038/nature15393
Agranat-Tamir, L., et al. (2020). The genomic history of the Bronze Age Southern Levant. Cell, 181(5), 1146–1157.e11. https://doi.org/10.1016/j.cell.2020.04.024
Alexander, D.H., Novembre, J., & Lange, K. (2009). Fast model-based estimation of ancestry in unrelated individuals. Genome Research, 19, 1655–1664. https://doi.org/10.1101/gr.094052.109
Alkan, C., Kavak, P., Somel, M. et al. Whole genome sequencing of Turkish genomes reveals functional private alleles and impact of genetic interactions with Europe, Asia and Africa. BMC Genomics 15, 963 (2014). https://doi.org/10.1186/1471-2164-15-963
Amorim, C. E. G., Vai, S., Posth, C., et al. (2018). Understanding 6th-century barbarian social organization and migration through paleogenomics. Nature Communications, 9(1), 3547. https://doi.org/10.1038/s41467-018-06024-4
Antonio, M. L., Weiß, C. L., Gao, Z., Sawyer, S., Oberreiter, V., Moots, H. M., Spence, J. P., Cheronet, O., Zagorc, B., Praxmarer, E., Özdoğan, K. T., Demetz, L., Gelabert, P., Fernandes, D., Lucci, M., … Pritchard, J. K. (2024). Stable population structure in Europe since the Iron Age, despite high mobility. eLife, 13, Article e79714. https://doi.org/10.7554/eLife.79714
Bennett, E. A., Parasayan, O., Prat, S., et al. (2023). Genome sequences of 36,000- to 37,000-year-old modern humans at Buran-Kaya III in Crimea. *Nature Ecology & Evolution*, 7(12), 2160–2172. [https://doi.org/10.1038/s41559-023-02211-9](https://doi.org/10.1038/s41559-023-02211-9)
Berg, J. J., & Coop, G. (2014). A population genetic signal of polygenic adaptation. PLoS genetics, 10(8), e1004412. https://doi.org/10.1371/journal.pgen.1004412
Briggs, J. L. (1970). Never in anger: Portrait of an Eskimo family. Harvard University Press.
Bryk, J., Hardouin, E., Pugach, I., Hughes, D., Strotmann, R., Stoneking, M., & Myles, S. (2008). Positive selection in East Asians for an EDAR allele that enhances NF-κB activation. PLoS ONE, 3(5), Article Article e2209. https://doi.org/10.1371/journal.pone.0002209
Chang, C. C., Chow, C. C., Tellier, L. C., Purcell, S. M., & Lee, J. J. (2020). PLINK 2.0: Expanding the scope of genomic analyses. Preprint available at: https://www.cog-genomics.org/plink/2.0/.
Chang, C. C., Chow, C. C., Tellier, L. C., Vattikuti, S., Purcell, S. M., & Lee, J. J. (2015). Second-generation PLINK: Rising to the challenge of larger and richer datasets. GigaScience, 4(1), 7. doi:10.1186/s13742-015-0047-8
Coop, G., et al. (2009). The role of geography in human adaptation. PLOS Genetics, 5(6), e1000500.
Das, S., Forer, L., Schönherr, S., Sidore, C., Locke, A. E., Kwong, A., Vrieze, S. I., Chew, E. Y., Levy, S., McGue, M., Schlessinger, D., Stambolian, D., Loh, P.-R., Iacono, W. G., Swaroop, A., Scott, L. J., Cucca, F., Kronenberg, F., Boehnke, M., … Fuchsberger, C. (2016). Next-generation genotype imputation service and methods. Nature Genetics, 48(10), 1284–1287.
Fuchsberger, C., Abecasis, G. R., & Hinds, D. A. (2014). minimac2: faster genotype imputation. Bioinformatics, 31(5), 782–784.
Gunderson, E. K. E. (1974). Psychological studies in Antarctica. In E. K. E. Gunderson (Ed.), Human Adaptability to Antarctic Conditions (Antarctic Research Series, Vol. 22, pp. 115–131). Washington, DC: American Geophysical Union. https://doi.org/10.1029/AR022p0115
Hajdinjak, M., Mafessoni, F., Skov, L., et al. (2021). Initial Upper Palaeolithic humans in Europe had recent Neanderthal ancestry. Nature, 592, 253–257. https://doi.org/10.1038/s41586-021-03335-3
He, G.-L., Wang, M.-G., Li, Y.-X., Zou, X., Yeh, H.-Y., Tang, R.-K., Yang, X.-M., Wang, Z., Guo, J.-X., Luo, T., Zhao, J., Sun, J., Hu, R., Wei, L.-H., Chen, G., Hou, Y.-P., & Wang, C.-C. (2022). Fine-scale north-to-south genetic admixture profile in Shaanxi Han Chinese revealed by genome-wide demographic history reconstruction. Journal of Systematics and Evolution, 60(4), 955–972. https://doi.org/10.1111/jse.12715
Hsu, J. S., Wu, D. C., Shih, S. H., Liu, J. F., Tsai, Y. C., Lee, T. L., Chen, W. A., Tseng, Y. H., Lo, Y. C., Lin, H. Y., Chen, Y. C., Chen, J. Y., Chou, T. H., Chang, D. T., Su, M. W., Guo, W. H., Mao, H. H., Chen, C. Y., & Chen, P. L. (2023). Complete genomic profiles of 1496 Taiwanese reveal curated medical insights. Journal of advanced research, S2090-1232(23)00405-8. Advance online publication. https://doi.org/10.1016/j.jare.2023.12.018
Hui, R., et al. (2024). Genetic history of Cambridgeshire before and after the Black Death. Science Advances, 10, Article eadi5903. https://doi.org/10.1126/sciadv.adi5903
Kaja, E., Lejman, A., Sielski, D., Sypniewski, M., Gambin, T., Dawidziuk, M., Suchocki, T., Golik, P., Wojtaszewska, M., Mroczek, M., Stępień, M., Szyda, J., Lisiak-Teodorczyk, K., Wolbach, F., Kołodziejska, D., Ferdyn, K., Dąbrowski, M., Woźna, A., Żytkiewicz, M., Bodora-Troińska, A., … Sztromwasser, P. (2022). The Thousand Polish Genomes-A Database of Polish Variant Allele Frequencies. International journal of molecular sciences, 23(9), 4532. https://doi.org/10.3390/ijms23094532
Karczewski, K. J., Francioli, L. C., Tiao, G., Cummings, B. B., Alfoldi, J., Wang, Q., ... & MacArthur, D. G. (2020). The mutational constraint spectrum quantified from variation in 141,456 humans. Nature, 581(7809), 434-443. https://doi.org/10.1038/s41586-020-2308-7
Latta, R. G. (1998). Differentiation of allelic frequencies at quantitative trait loci affecting locally adaptive traits. The American Naturalist, 151(2), 283–292
Le Corre, V., & Kremer, A. (2003). Genetic variability at neutral markers, quantitative trait loci and trait in a subdivided population under selection. Genetics, 164(3), 1205–1219.
Le Corre, V., & Kremer, A. (2012). The genetic differentiation at quantitative trait loci under local adaptation. Molecular Ecology, 21, 1548–1566. https://doi.org/10.1111/j.1365-294X.2012.05479.x
Lazaridis, I., Patterson, N., Mittnik, A., et al. (2014). Ancient human genomes suggest three ancestral populations for present-day Europeans. Nature, 513, 409–413. https://doi.org/10.1038/nature13673
Lazaridis, I., Patterson, N., Anthony, D., et al. (2025). The genetic origin of the Indo-Europeans. Nature, 639(8044), 132–142. https://doi.org/10.1038/s41586-024-08531-5
Lee, J., Lee, J., Jeon, S., Lee, J., Jang, I., Yang, J. O., Park, S., Lee, B., Choi, J., Choi, B. O., Gee, H. Y., Oh, J., Jang, I. J., Lee, S., Baek, D., Koh, Y., Yoon, S. S., Kim, Y. J., Chae, J. H., Park, W. Y., … Choi, M. (2022). A database of 5305 healthy Korean individuals reveals genetic and clinical implications for an East Asian population. Experimental & molecular medicine, 54(11), 1862–1871. https://doi.org/10.1038/s12276-022-00871-4
Ma, X. F., Hall, D., Onge, K. R., Jansson, S., & Ingvarsson, P. K. (2010). Genetic differentiation, clinal variation and phenotypic associations with growth cessation across the Populus tremula photoperiodic pathway. Genetics, 186(3), 1033–1044. https://doi.org/10.1534/genetics.110.1208734
Malomane, D. K., Williams, M. P., Huber, C. D., Mangul, S., Abedalthagafi, M., & Chiang, C. W. K. (2025). Patterns of population structure and genetic variation within the Saudi Arabian population. bioRxiv. https://doi.org/10.1101/2025.01.10.632500
Mao, X. et al. (2021). The deep population history of northern East Asia from the Late Pleistocene to the Holocene. Cell, 184(12), 3256–3266.
Matsumura, H., et al. (2021). Craniometrics reveal "two layers" of prehistoric human dispersal in Eastern Eurasia. Scientific Reports, 11, 20312.
Naslavsky, M.S., Scliar, M.O., Yamamoto, G.L. et al. Whole-genome sequencing of 1,171 elderly admixed individuals from Brazil. Nat Commun 13, 1004 (2022). https://doi.org/10.1038/s41467-022-28648-3
Nikitin, A. G., Lazaridis, I., Patterson, N., et al. (2025). A genomic history of the North Pontic Region from the Neolithic to the Bronze Age. Nature, 639(8044), 124–131. https://doi.org/10.1038/s41586-024-08372-2
Nisbett, R. E. (2003). The geography of thought. Free Press.
Olalde, I., Mallick, S., Patterson, N., Rohland, N., Villalba-Mouco, V., Silva, M., Dulias, K., Edwards, C. J., Gandini, F., Pala, M., Soares, P., Ferrando-Bernal, M., Adamski, N., Broomandkhoshbacht, N., Cheronet, O., Culleton, B. J., Fernandes, D., Lawson, A. M., Mah, M., Oppenheimer, J., … Reich, D. (2019). The genomic history of the Iberian Peninsula over the past 8000 years. Science (New York, N.Y.), 363(6432), 1230–1234. https://doi.org/10.1126/science.aav4040
Piffer, D. (2025). Directional selection and evolution of polygenic traits in Eastern Eurasia: Insights from ancient DNA. Twin Research and Human Genetics, 28(1), 1–20. doi: 10.1017/thg.2024.49
Piffer, D., & Kirkegaard, E. O. W. (2024a). Evolutionary trends of polygenic scores in European populations from the Paleolithic to modern times. Twin Research and Human Genetics, 27, 30–49. doi: 10.1017/thg.2024.8
Piffer, D., & Kirkegaard, E. O. W. (2024b). Polygenic Selection and Environmental Influence on Adult Body Height: Genetic and Living Standard Contributions Across Diverse Populations. Twin research and human genetics, 27, 265-282. https://doi.org/10.1017/thg.2024.43
Price, P. A., & Lannon, M. (2023). The evolution of chronic primary angle closure glaucoma in the Inuit and oriental eye. Journal of Community Medicine & Public Health, 7(1), Article Article 290. https://doi.org/10.29011/2577-2228.100290
Prüfer, K., Posth, C., Yu, H., et al. (2021). A genome sequence from a modern human skull over 45,000 years old from Zlatý kůň in Czechia. Nature Ecology & Evolution, 5, 820–825. https://doi.org/10.1038/s41559-021-01443-x
Qi, X.-G., Wu, J., Zhao, L., Wang, L., Guang, X., Garber, P. A., Opie, C., Yuan, Y., Diao, R., Li, G., Wang, K., Pan, R., Ji, W., Sun, H., Huang, Z.-P., Xu, C., Witarto, A. B., Jia, R., Zhang, C., … Li, B. (2023). Adaptations to a cold climate promoted social evolution in Asian colobine primates. Science, 380(6648), eabl8621. https://doi.org/10.1126/science.abl8621
R Core Team. (2023). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.R-project.org/
Ravasini, F., Kabral, H., Solnik, A., et al. (2024). The genomic portrait of the Picene culture provides new insights into the Italic Iron Age and the legacy of the Roman Empire in Central Italy. Genome Biology, 25(1), 292. https://doi.org/10.1186/s13059-024-03430-4
Rentoft M, Svensson D, Sjödin A, Olason PI, Sjöström O, Nylander C, et al. (2019) A geographically matched control population efficiently limits the number of candidate disease-causing variants in an unbiased whole-genome analysis. PLoS ONE 14(3): e0213350. https://doi.org/10.1371/journal.pone.0213350
Rivollat, M., et al. (2020). Ancient genome-wide DNA from France highlights the complexity of interactions between Mesolithic hunter-gatherers and Neolithic farmers. Science Advances, 6(22), Article eaz5344. https://doi.org/10.1126/sciadv.aaz5344
Saupe, T., Montinaro, F., Scaggion, C., Carrara, N., Kivisild, T., D’Atanasio, E., Hui, R., Solnik, A., Lebrasseur, O., Larson, G., Alessandri, L., Arienzo, I., De Angelis, F., Rolfo, M. F., Skeates, R., Silvestri, L., Beckett, J., Talamo, S., Dolfini, A., . . . Scheib, C. L. (2021). Ancient genomes reveal structural shifts after the arrival of Steppe-related ancestry in the Italian Peninsula. Current Biology, 31(12), 2576–2591.e12. https://doi.org/10.1016/j.cub.2021.04.022
Sikora, M., et al. (2019). The population history of northeastern Siberia since the Pleistocene. Nature, 570(7760), 182–188.
Schwaba, T., Sullivan, M. L. C., Akingbuwa, W. A., Ilves, K., Tanksley, P. T., Williams, C. M., Dragostinov, Y., Liao, W., Ackerman, L. S., Fealy, J. C. M., Hemani, G., de la Fuente, J., Gupta, P., Stein, M. B., Gelernter, J., Levey, D. F., Võsa, U., Ausmees, L., Realo, A., Estonian Biobank Research Team, Vaht, M., Allik, J., Esko, T., Mõttus, R., Vainik, U., Jonsdottir, G. A., Thorleifsson, G., Gunnarsson, Á. F., Bjornsdottir, G., Thorgeirsson, T. E., Stefansson, H., Stefansson, K., Cheesman, R., Qin, Q., Corfield, E. C., Ask, H., Torvik, F. A., Ystrom, E., Tesli, M., Boomsma, D. I., de Geus, E. J. C., Hottenga, J.-J., Melo, D. C., Snieder, H., Hartman, C. A., Xia, C., Campbell, A., Luciano, M., Deary, I. J., Hill, W. D., Jang, S.-K., Vrieze, S. I., Abecasis, G., Lupton, M. K., Mitchell, B. L., Viher, P. V., Colodro-Conde, L., Martin, N. G., Medland, S. E., Derks, E. M., Wormington, B., Kaprio, J., Silventoinen, K., Palviainen, T., Gidziela, A., Rimfeld, K., Plomin, R., Malanchini, M., Dick, D. M., Aliev, F., COGA Collaborators, The Spit for Science Working Group, Wesseldijk, L. W., Ullén, F., Mosing, M. A., Kranzler, H. R., Nunez, Y., Beck, S., Polimanti, R., Edwards, T., Giannelis, A., Willoughby, E. A., Lee, J. J., McGue, M., Terracciano, A., Marongiu, M., Fiorillo, E., Cucca, F., Sutin, A. R., van der Most, P. J., Oldehinkel, A. J., Kretschmer, T., Shabalin, A. A., Docherty, A. R., Krueger, R. F., Freilich, C. D., Mishra, B. H., Lehtimäki, T., Raitakari, O. T., Kähönen, M., Saarinen, A., Dobewall, H., Keltikangas-Järvinen, L., Berger, K., Herrera-Rivero, M., Streit, F., Awasthi, S., Witt, S. H., Tuhkanen, J., Räikkönen, K., Eriksson, J. G., Lahti, J., Davies, G., Redmond, P., Taylor, A., Corley, J., Russ, T. C., Ciullo, M., Nutile, T., Ding, J., Qian, Y., Tanaka, T., Ferrucci, L., Zillich, L., Sirignano, L., Harden, K. P., Genç, E., Gajewski, P. D., Getzmann, S., Fraenz, C., Peñate, J. E. S., Lis, S., Hall, A. S. M., Schmahl, C., Herpertz, S. C., Abdellaoui, A., Nivard, M. G., & Tucker-Drob, E. M. (2025). Robust inference and widespread genetic correlates from a large-scale genetic association study of human personality. bioRxiv. https://doi.org/10.1101/2025.05.16.648988
Sun, D. (2025). Arctic instincts? The Late Pleistocene Arctic origins of East Asian psychology.Evolutionary Behavioral Sciences, 19(3), 227–265. https://doi.org/10.1037/ebs0000373
Talhelm, T. et al. (2014). Large-scale psychological differences within China explained by rice versus wheat agriculture. Science, 344(6184), 603–608.
Tian, Y., Koncz, I., Defant, S., Giostra, C., Vyas, D. N., Sołtysiak, A., Pejrani Baricco, L., Fetner, R., Posth, C., Brandt, G., Bedini, E., Modi, A., Lari, M., Vai, S., Francalacci, P., Fernandes, R., Steinhof, A., Pohl, W., Caramelli, D., Krause, J., Izdebski, A., Geary, P. J., & Veeramah, K. R. (2024). The role of emerging elites in the formation and development of communities after the fall of the Roman Empire. Proceedings of the National Academy of Sciences of the United States of America, 121(36), e2317868121. https://doi.org/10.1073/pnas.2317868121
Tooby, J., & Cosmides, L. (1990). The past explains the present: Emotional adaptations and the structure of ancestral environments. Ethology and Sociobiology, 11(4–5), 375–424.
Villalba-Mouco, V., van de Loosdrecht, M. S., Posth, C., Mora, R., Martínez-Moreno, J., Rojo-Guerra, M., Salazar-García, D. C., Royo-Guillén, J. I., Kunst, M., Rougier, H., Crevecoeur, I., Arcusa-Magallón, H., Tejedor-Rodríguez, C., García-Martínez de Lagrán, I., Garrido-Pena, R., Gilbert, E., Jones, E. R., Lalueza-Fox, C., & Haak, W. (2019). Survival of Late Pleistocene hunter-gatherer ancestry in the Iberian Peninsula. Current Biology, 29(7), 1169–1177. https://doi.org/10.1016/j.cub.2019.02.006
Villalba-Mouco, V., van de Loosdrecht, M.S., Rohrlach, A.B. et al. A 23,000-year-old southern Iberian individual links human groups that lived in Western Europe before and after the Last Glacial Maximum. Nat Ecol Evol 7, 597–609 (2023). https://doi.org/10.1038/s41559-023-01987-0
Vyas, D. N., Koncz, I., Modi, A., Mende, B. G., Tian, Y., Francalacci, P., Lari, M., Vai, S., Straub, P., Gallina, Z., Szeniczey, T., Hajdu, T., Pejrani Baricco, L., Giostra, C., Radzevičiūtė, R., Hofmanová, Z., Évinger, S., Bernert, Z., Pohl, W., Caramelli, D., … Veeramah, K. R. (2023). Fine-scale sampling uncovers the complexity of migrations in 5th-6th century Pannonia. Current biology : CB, 33(18), 3951–3961.e11. https://doi.org/10.1016/j.cub.2023.07.063
Yang, M. A., Fan, X., Sun, B., Chen, C., Lang, J., Ko, Y.-C., Tsang, C., Chiu, H., Wang, T., Bao, Q., Wu, X., Hajdinjak, M., Ko, A. M.-S., Ding, M., Cao, P., Yang, R., Liu, F., Nickel, B., Dai, Q., … Fu, Q. (2020). Ancient DNA indicates human population shifts and admixture in northern and southern China. Science, 369(6501), 282–288. https://doi.org/10.1126/science.aba0909
Yoo, S. K., Kim, C. U., Kim, H. L., Kim, S., Shin, J. Y., Kim, N., Yang, J. S. W., Lo, K. W., Cho, B., Matsuda, F., Schuster, S. C., Kim, C., Kim, J. I., & Seo, J. S. (2019). NARD: whole-genome reference panel of 1779 Northeast Asians improves imputation accuracy of rare and low-frequency variants. Genome medicine, 11(1), 64. https://doi.org/10.1186/s13073-019-0677-z
Ziyatdinov, A., Torres, J., Alegre-Díaz, J. et al. Genotyping, sequencing and analysis of 140,000 adults from Mexico City. Nature 622, 784–793 (2023). https://doi.org/10.1038/s41586-023-06595-3

No competing interests reported.

Download PDF

Journal Publication

published 04 Nov, 2025

Read the published version in Adaptive Human Behavior and Physiology →

Editorial decision: Accepted
30 Sep, 2025
Editor assigned by journal
30 Sep, 2025
Submission checks completed at journal
30 Sep, 2025
First submitted to journal
23 Sep, 2025

You are reading this latest preprint version

Testing the Late Pleistocene Arctic Origins of East Asian Psychology using Ancient and Modern DNA

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Testing Arctic Psychology at Scale

Methods

Population genetics rationale for using split-half reliability of polygenic scores

Modern genome samples

Ancient genome samples

Admixture analysis

GWASs used for polygenic scores

Results

Summary of GWAS data

Modern WGS sample

WGS + Array sample

Ancient DNA

Western Eurasian ancient DNA

Agreeableness

Extraversion

Neuroticism

Agreeableness

Extraversion

Neuroticism

Factor analysis at the group level

Eastern Eurasian ancient DNA

Agreeableness

Extraversion

Neuroticism

Factor analysis at the individual level

Factor analysis at the population level

Discussion

Evidence for an integrated trait complex

Climatic and geographic correlates

Critical limitations and contradictions

Toward a refined model

Future directions

Conclusion

Statements and Declarations

Author Contribution

References

Additional Declarations

Status:

Journal Publication

Version 1