Against the Mutational Load Hypothesis: Part 3

Case studies and conclusions

This post is one part of a piece that was written as a single post that was split due to issues with the Substack editor (part 1, part 2). Read those first!

Case Study #1: Rome

The primary proponent of this hypothesis, Joseph Bronski, has claimed that the increasing leftism (feminism, sexual deviancy, mass immigration, decline of ethnocentrism, decline of religiosity) common to many declining empires, not just the modern West, are the result of mutational load. In particular, he has cited Rome as an example. 

There is good evidence that dysgenics played a role in Roman decline, backing up the historical record of low Roman elite fertility and mass Middle Eastern immigration. But there is absolutely no evidence mutational load played any role.

Result from here: (https://emilkirkegaard.dk/en/2023/07/fall-of-the-roman-empire-polygenic-score-edition/). This result shouldn’t be seen as definitive and may be revised if better samples are obtained, but matches what would be predicted from the historical record.

It is important to get the timeline right. Roman female emancipation began in the Republican period, before Augustus and before mass immigration to Rome. Furthermore, emancipation began in the upper stratum of society, before trickling down to the lower classes. 

From J. D Unwin’s Sex and Culture

This led to low elite fertility (which Augustus tried and failed to halt through legislation) and thus selection for lower intelligence. In other words, the social change, female emancipation, predates the genetic change, not the other way around. Later, this was compounded by mass immigration from the Middle East.

PCA showing the genetics of Rome at various points in time. Republic-era and modern Romans cluster with other Southern Europeans, but Imperial Rome is intermediate between Italy and the Middle East, confirming historical evidence of mass Middle Eastern immigration.

There is no evidence at all of mutational load accumulation in Rome during the Republican period (or after, until the modern day), and no reason to believe it happened. Rome had high infant mortality and generally poor diets, as you would expect from a premodern society.

An estimate of Roman infant mortality based on skeletons found it to be comparable to extremely poor preindustrial countries with high infant mortality.

An estimate of biological health (result of nutrition and disease) from skeletal data (https://www.sciencedirect.com/science/article/abs/pii/S1570677X18302600?via%3Dihub). Note that there is no great improvement that could potentially lower infant mortality in the Republican period.

As such, purifying selection remained strong, and so increasing mutational load is not a plausible contributing factor to the decline of the Roman Empire. The same is true of every pre-modern state. Genetics matters, but through dysgenic selection and mass immigration, which postdate the beginnings of “leftism” in the Roman Republic. The political (including geopolitics, in this case the conquest of the Middle East) and social shift precedes and causes the genetic shift.

Case Study #2: Quebec vs France

To have a good natural experiment, we need two populations, racially as similar as possible and with similar initial social structures, but isolated from each other politically and with differing levels of mutational load. Then we see if the one with higher mutational load became leftist faster than the other. This is a surprisingly difficult task; generally, ethnically similar populations have had very similar infant mortality trends over the past 200 years. But there is one significant exception¹. 

From 1670-1960 or so, Quebec had a very different population history from France. Not only was infant mortality lower due to comparatively abundant food and low disease, but paternal age in 17th century Quebec had a very low effect on reproductive success. On average, a decade of additional paternal age reduced the number of surviving children by 3%. A decade of paternal age adds ~20 de novo mutations, so this is a measure of how much mutational load affected reproductive success. This is identical to 20th century Sweden, a wealthy, industrialized welfare state, and less than half of the effect in 18th century Sweden (-7%) or Germany (-8.4%). Similarly, the negative effect of an additional decade of paternal age on infant survival was 1% in Quebec, only half the effect (2%) in historical Sweden or Germany (negligible in modern Sweden²). This means that additional mutational load had much less effect on both reproductive success in adulthood and on infant survival odds in Quebec compared to continental Europe and consequently not only do modern Quebecers have higher genetic load then their continental European counterparts, but this has been true since the late 17th century. This prediction is confirmed by direct measurement of mutational load in France and Quebec.

In fact, direct (though imprecise) estimates of the effects of 30 years of paternal age (or 60 de novo mutations) on fitness in Quebec, modern Sweden, and historical Sweden and Germany, show that mutational load had nearly identical effects on fitness in premodern Quebec as it did in 20th century Sweden.

These numbers show the estimated fitness costs (from a model built on baptism records) of 30 years of paternal age in 18th century Sweden, 18th century Germany, 17th century Quebec, and 20th century Sweden. This is a measurement of the fitness costs of mutational load. Note how close 20th century Sweden is to 17th century Quebec, and how the fitness costs there are clearly lower then in premodern Sweden and Germany (https://royalsocietypublishing.org/doi/full/10.1098/rspb.2017.1562).

This means that, in terms of accumulation of mutational load in modernity, Quebec has a ~200 year head start on continental Europe. So according to the MLH, in which mutational load accumulation is the dominant factor in modern political shifts, Quebec politics ought to be far, far to the left of continental Europe, including France. Even 19th century Quebec attitudes ought to be comparable to modern France. Were they? Obviously not. While it is difficult to directly compare historical levels of ethnocentrism or socialism due to their differing political statuses (great power empire vs province of a backwater part of someone else’s empire), we can look at their respective attitudes towards sex roles and religion. And we see very clearly that Quebec was far to the right of France until around 1960, with the Quiet Revolution.

Note how until the Baby Boom, Quebec has the highest fertility (an indicator of feminism and secularism) and France nearly the lowest.

A map of when birth rates first fell sharply in Europe. France was unusually secularist and feminist early on, and consequently was the first country to go through the modern demographic transition, well before the economically more advanced England.

Less quantitative measures of politics would give a similar result. Up until the Russian Revolution, France was approximately³ the most left-wing white state in the world, at the epicenter of socialism, secularism, feminism, and republicanism, while Quebec was conservative and extremely Catholic. Today, post-60s Cultural Revolution, Quebec and France have similar politics on everything from women’s roles to Islam to state intervention in the economy, rather than Quebec being far to France’s left as predicted by the MLH. 

A list of the changes of Quebec’s Quiet Revolution - note the timing, 1960 - 1970. From https://www.cremieux.xyz/p/the-fruits-of-philosophy.

You could try to save the theory by positing that Quebecers were incredibly strongly selected for rightism compared to Frenchmen (both as a founder effect and in the centuries after British conquest), and that as such even centuries of additional mutational load did not incline their politics to the left of France. But recall that the theory requires that the effects of mutational load on politics be much larger than that of selection, as they have to overwhelm generations of positive selection for rightism in the 20th century, so this is not plausible (nor is there any evidence this is the case⁴).

Case Study #3: Breeder Cults

These high-fertility groups are socially and culturally cut off from broader society, but they face the same low infant mortality fates (Amish, Haredim), relaxed selection, and consequently increasing mutational load as the societies they are embedded in. And yet their politics do not match the predictions of the MLH; these groups manage to maintain high fertility, high religiosity, and high ethnocentrism, and to the extent they are involved in politics, are overwhelmingly right-wing⁵.

Possible Explanations

So given that premise (1) is true, and premise (2) is plausible, why might the conclusion of the MLH be false? A few possibilities, in order of most to least likely:

1. Premise (2) is false, and there is no connection between mutational load and politics at all. I give this a ~60% chance of being true. The evidence is weakly favorable, but the null is no relationship.

2. Premises (1) and (2) are correct, but the magnitude is just too small to matter. If premise (2) is correct, this is almost certainly the case, as seen by Quebec vs France, the generally small size of paternal age effects, and the lack of impact of load accumulation on population intelligence.

3. Purifying selection isn’t 0. Direct estimates of purifying selection from 20th Sweden show that it is about half as effective as premodern Sweden. So accumulation may be too slow, or simply not reach a high enough level. According to the simplest model of mutation-selection equilibrium, halving selective pressure doubles load (though in practice it’s much more complicated), which, depending on initial levels of load, may not be very important.

4. Asymmetric mutation payoffs may make mutational load a nonissue in humans (at the population level).

Conclusions

1. Even assuming mutational load is directly connected to left-wing politics (which is still not clear), this does not imply mutational load accumulation played a nontrivial role in Western political shifts. The magnitude of the shift is far larger than the change in a similar trait, intelligence caused by load. 

2. Mutational load played no role in Roman (or any other premodern empire) decline, and to the extent that modern Western decline parallels the decline of older empires, this is evidence against the MLH, not for it.

3. The best natural experiment we have, Quebec vs France, gives the opposite result as predicted by the MLH.

4. Genetic change causing political change is overwhelmingly a function of immigration (strongly inclining politics to the left). With the (important) exception of the Anglosphere settler colonies, which promoted immigration for economic and security reasons in the 19th and early 20th centuries, mass immigration is the result of previous leftwards shifts in politics. It is a positive feedback loop, but the initial impetus is politics, not genetics. Selection (weakly inclining politics to the right) is secondary. There is no detectable signal of mutational load. The overwhelming majority of political change is not due to genetic changes.

The MLH is an intriguing conjecture. But it’s wrong.

1. Arguably two, since you could make a similar comparison between Anglo-American whites and the British. Much like the Quebecers, colonial Americans saw massive population expansion thanks to abundant land and low disease burden. Anglo-Americans are consistently among the most right-wing American white ethnic groups, while 20th century British politics were generally to the left of their American counterpart (and this without a significant leftwing nonwhite presence or the leftward shift caused by non-Anglo European immigration to the US, particularly Catholic and Jewish). But this comparison is heavily confounded by intermarriage with other European ethnic groups and by large and continued English immigration to the US after English and American infant mortality rates converged (while French immigration to Quebec was almost totally cut off between the British conquest and the mid-20th century, and other immigration was minimal), and as far as I know there are no published estimates of Anglo-American vs British mutational load, so I relegate this comparison to a footnote.

2. Parenthetically, this also puts an upper bound on the effect of lowered infant mortality on mutational load. Texts such as chapter 6 of Modernity and Cultural Decline (page 201 footnote) that posit large mutational load increases due to lower infant mortality use hypothetical examples of ~50% differences in infant mortality due to an additional decade of paternal age to illustrate their point, however the true, measured value is only ~2%, in Malthusian preindustrial societies (please note that Modernity and Cultural Decline does not claim that a decade of additional infant mortality has such a large effect; this is clearly noted to be a hypothetical).

3. It is difficult to map the 19th century political spectrum onto the 20th and 21st, so this comparison must necessarily remain qualitative.

4. While there is no GSS data for Quebec, left/right fertility differences in the US are minor pre-1960s, probably due to hegemonic religiosity and lack of 2nd wave feminism. The same is likely true of pre-1960s Quebec, and there is absolutely no evidence of unusually strong selection for conservatism there.

5. It is possible that selective attrition happens to remove genetic load in such a way as to prevent accumulation despite near-zero infant mortality, but there’s no evidence this is the case and attrition is too low for the numbers to work out anyways (around 10% among Israeli Ashkenazi Haredim and  around 5% in the even more conservative Hasid subgroup, compared to >30% premodern infant mortality rates).