A while ago I put out a post outlining my disagreements with Bronski’s thesis that the rise of Liberalism can be attributed to a build up in mutational load due to relaxed purifying selection following the industrial revolution [1]. I’ve come across some new evidence since then that strengthens the case against it, and I no longer really have any doubts about how strong of a contributing factor mutational load represents.
Before going over the new evidence I’ve come across, I’d like to address two shortcomings of my original article:
“Even should one of them manage to win the baby race, genetic drift alone should stop many of the mutants from passing along the mutated portion of their genomes” [1, ‘Genotypic Liberalism VS Mutation’].
This sentence betrays a basic ignorance of how genetic drift should behave. Short of purifying selection, mutational load should be expected to accumulate more or less linearly with every consecutive generation.
Let's say that inheriting a mutation at a locus from your father but not your mother is equal to you getting half a variant worth of mutation. Each parent gives you 8 of these halves, and so you get '8 variants' total, or 16 halves. Each of these 16 halves will have a 50% chance of being inherited by your children, and so you pass on 8 halves, or 4 variants to your child, but you married another person to produce that child, and your spouse gave your child another 8 halves, and so your child has 16 halves ‘8 variants’ in total, right back where we began.“Indeed, modern day Europeans have more deleterious mutations than Africans despite the fact that they also have less overall genetic diversity due to the out of Africa bottleneck [20]” [1, ‘Conservatism VS Mutation’].
As it turns out, whether this is true depends on what molecular measure of mutational load you choose to use, and the better + more appropriate operationalizations find that the modern French aren’t anything unusual when compared to Africans, Han Chinese, and ancient hominids [2, figure 3]:
We can take things further than these molecular genetic comparisons however.
Let’s ask ourselves: Is it even true that that modern medicine enables the children of older parents to be more reproductively succesful than they used to be able to? As it turns out, “In historical Sweden, paternal age had a slight positive effect in m1 before using sibling comparisons, in the other populations the effect was negative in all models” [3, pp.4-5]. So, at the correlational level, men at higher paternal ages were having superior selection outcomes (implying that mutation should have been increasing with every consecutive generation while this is no longer the case), but when sibling controls were done to get closer at the causal effects of mutation itself irrespective of whatever other age-correlated confounds might’ve been causal for reproductive success, it was found that a decade of paternal age also had a more deleterious effect on fitness than it does today. Our current estimates of the number of de novo mutations that happen every generation come from simply counting how many variants a child has that couldn’t have come from either parent, but there’s no reason to think that parental age is the only thing that determines the number that accrue, and there’s no reason to think that the number is a universal constant throughout history. One could reasonably speculate that at a given level of parental age, better nutrition + more access to food + less life stress + more time indoors away from the sun + etc means less mutations per generation than in the past.
Regarding the sibling control effects, there were also two other historic populations that Swedes were compared with (Germans and early French settlers of Quebec), and the Swedes weren’t anything out of the ordinary [3, figure 2]:
Note that Figure 2 outlines the *sibling-control* results.
In addition to the relative effects of parental ages on selection outcomes, average paternal and maternal ages in Sweden were on the decline from the 1930s to 1970s before increasing again from the 1970s to 2010, but average maternal + paternal ages of Swedes in 2010 were still lower than those of historic Swedes [3, p.8].
Finally, are deleterious de novo mutations even causal for shitlib political views? Ideally, we’d want a sibling-control study. As the next best thing however, we can look at birth order controlled for number of siblings to get mostly the same signal. There was an N = 1,945 GSS study [4] which did just this, and it doesn’t look like parental age has much effect (I don’t take environmental effects of birth order seriously, especially in the modern United States. If you disagree, I don’t care, this is just what suffices for convincing me). The relevant thing to look at is model 2 from table 1 [4]:
Model 2 imo has the relevant controls but minimal irrelevant controls that needlessly throw away statistical power. Not that the different models and addition/removal of controls ever really changed anything. As we can see, if anything, it looks like the causal effect of deleterious mutational load is to slightly increase the prevalence of conservative beliefs. Maybe increased purifying selection is why Sweden is so much more Liberal than it once was? (No, this is a null result).
EDIT:
Bronski has his own study doing the same thing as the GSS study with a sample which is similarly powered. His headline analysis was paternal age controlled for sibling order, which I don’t care about, but his data supports the following reanalysis:
This paternal age controlled for older_n + younger_n is actually better than sib order controlled for sibship size, so this is the preferable thing to listen to. The coefficient for paternal_age_n is roughly what he wants, but the sample doesn’t quite have the power such that p=.695, and he’s recruiting more to account for this.
In any case, even though we’ve never had molecular verification of things like left-handedness or attractiveness correlating with mutation at the individual level, if forced to guess, I would say that mutation has a causal effect of left-wing views. But the societal-level results are still a big problem. Possibly a satisfying solution would be bioleninism. A good analogy might be the effect of education on income; the signalling effects of education affects how an individual ranks in income within his society, but it does not affect societal-level GDP/capita [5]. In the same way, mutation may make individuals less attractive than average and make them more sympathetic to equalitarian views by that mechanism, but it’s a consequence of how averages work that half of the people in any society will be below average.
EDIT 2:
Since people are still getting this wrong, I’m posting it again:
“In historical Sweden, paternal age had a slight positive effect in m1 before using sibling comparisons, in the other populations the effect was negative in all models” [3, pp.4-5].
This should clarify even further:
"(b) Statistical approach: ...We analysed reproductive success for all offspring, including those who died in childhood or never married."
This is why “number of children” appears in the chart twice. Infant mortality is taken into account unless specified otherwise: To be more specific, “number of children” looks at the children produced by old fathers and sees if these children produce a different amount of children of their own, and if a child dies in infancy, they count as giving their parent zero grandchildren. Purifying selection isn’t reducible to infant mortality.
And again, the chart shows the *sibling-control effect*. What matters for quantifying purifying selection is the non-controlled correlational result, and on this matter, the opposite of purifying selection existed in pre-industrial Sweden before flipping to be more normal in modern Sweden. The point of the sibling control analysis was as an argument that a decade of paternal age doesn’t have as great of a deleterious effect as it once did; this is the weakest of the three arguments the paper supports, so it’s not a big deal to dismiss. However it’s interesting that the point estimates of m3 are always interpreted as reduced selection pressure when the chart is misinterpreted, until suddenly, the overlapping confidence intervals prove no difference when people realize their interpretive blunder.
Technical protip: overlapping confidence intervals are a nice heuristic, but they don’t strictly imply insignificance. Significance is strictly implied when intervals don’t overlap, but insignificance does not strictly imply overlapping confidence intervals.
EDIT 3:
I don’t even think Bronski’s paper is the superior design anymore since family size mostly isn’t the mechanism by which you’d expect leftist genes to correlate with the age at which somebody has kids; if leftism causes you to have kids at a later age, you’d get a paternal age correlation even with zero mutation effect. I initially understood this and somehow forgot it.
Sauce:
Angleson, C. (2023). Mind Viruses Are Fake, Propaganda Isn't. half-baked thoughts. Retrieved from https://werkat.substack.com/p/mind-viruses-are-fake-propaganda
Simons, Y. B., & Sella, G. (2016). The impact of recent population history on the deleterious mutation load in humans and close evolutionary relatives. Current opinion in genetics & development, 41, 150-158. Retrieved from https://sci-hub.ru/https://doi.org/10.1016/j.gde.2016.09.006
Arslan, R. C., Willführ, K. P., Frans, E. M., Verweij, K. J., Bürkner, P. C., Myrskylä, M., ... & Penke, L. (2017). Older fathers' children have lower evolutionary fitness across four centuries and in four populations. Proceedings of the Royal Society B: Biological Sciences, 284(1862), 20171562. Retrieved from https://sci-hub.ru/https://doi.org/10.1098/rspb.2017.1562
Powell, B., Freese, J., & Steelman, L. C. (1999). Rebel without a cause or effect: Birth order and social attitudes. American Sociological Review, 64(2), 207-231. Retrieved from https://sci-hub.ru/https://doi.org/10.2307/2657528
Angleson, C. (2022). Book Review: Bryan Caplan’s “The Case Against Education”. half-baked thoughts. Retrieved from https://werkat.substack.com/p/book-review-bryan-caplans-the-case
I think the disagreement comes down to this. Bronski wants to show that leftism is associated with mutational load. He does this by showing a correlation between paternal age as a proxy for mutational load and leftism. This does not mean that the primary reason (or even a particularly significant reason) that mutational load is increasing in modern populations is due to increased paternal age, but because of lower infant mortality. So i think Bronski is right that this article is largely irrelevant as a counter to his thesis.
1. the first chart is not informative on mutational load theory wrt leftism, which is about the average effect a new mutation has on leftism. Measure new genes wrt some other species or distant ancestor are completely irrelevant to this. You are confused about what the theory is if you made this mistake. The terms are similar but these are not the same thing.
2. your second chart shows increased infant mortality with increased paternal age at all ages. The rest of the things are conditions on not dying in infancy, and none of them measure the average effect of a new mutation on leftism, so they are not informative in this
3. You third chart doesn't show paternal age or a good measure of leftism. Instead you're using birth order. I have replicated those effect sizes of birth order on leftism and shown that the paternal age correlation is robust to birth order controls.