Back to [Archive] Post-review discussions

[OBG] Sexual selection explains sex and country differences in fluid g
Sexual dimorpshism in fluid intelligence suggests that this phenotype is a sexually selected trait.
Sexual selection has the double effect of increasing average population phenotype and reducing genetic variation in sexually selected traits. Matching these predictions, the average country fluid g (estimated from PISA Creative Problem Solving) is positively correlated to sex dimorphism and the latter in turn is inversely correlated to variance in intelligence scores withn populations. Average country male height is negatively correlated to sex dimorphism in intelligence, supporting the notion of a trade-off between selection for brain and brawn.
Why did you only use one PISA subtest as a measure of country-level IQ?
Why did you only use one PISA subtest as a measure of country-level IQ?


Because I do not consider PISA Math, Reading and Science to be good measures of IQ. They are too much affected by scholastic skills.
What about the hundred years of research showing correlations upwards of 0.7 between scholastic skills and IQ, or the fact that scores on standard fluid intelligence tests are strongly affected by years of schooling?
What about the hundred years of research showing correlations upwards of 0.7 between scholastic skills and IQ, or the fact that scores on standard fluid intelligence tests are strongly affected by years of schooling?


Intelligence is novel problem solving. Thus, PISA Creative Problem Solving conforms better to the definition of intelligence than assessments of scholastic aptitudes.
You're not replying to my objections. Keep in mind, also, that the face validity of an IQ test is a poor guide to its "deep" validity, or what it "really" measures. Vocabulary seems fairly trivial and superficial, but it is actually among the best predictors of g, reaction time, scholastic performance, achievement, and so on.

Keep the analyses using only Creative Problem Solving if you like, but I insist that you re-run the study using the entire PISA, and PISA + nationalIQ (supplemented by Malloy's more accurate values).
You're not replying to my objections. Keep in mind, also, that the face validity of an IQ test is a poor guide to its "deep" validity, or what it "really" measures. Vocabulary seems fairly trivial and superficial, but it is actually among the best predictors of g, reaction time, scholastic performance, achievement, and so on.

Keep the analyses using only Creative Problem Solving if you like, but I insist that you re-run the study using the entire PISA, and PISA + nationalIQ (supplemented by Malloy's more accurate values).


I've already answered. PISA relies too much on scholastic knowledge, it's more a test of how much effort people put into studying and this varies considerably by sex, with women notoriously being more hard working in school settings. This would give them a non-g dependent advantage in PISA reading,math and science, making it a poor measure of sex differences in g. Moreover, PISA CPS has the advantage of testing the core of intelligence, and providing an estimate in a single number, whereas PISA reading, math and science has 3 different subscales which load on different subskills with big gender differences, as shown by the fact that girls greatly outperform boys in Reading whereas they underperform boys in Math. It would just make the analysis disorderly and cumbersome, without providing any additional benefit. I attach a table (that Gerhard sent me) showing correlations between PISA problem solving and IQ, PISA reading,math, science.
The PISA r x CPS is less than PISA r x Math, and they're all in the 0.8-0.9 range. As for sex dimorphism this could be measured using PISA factor scores, which would eliminate specific subtest variance. Look, if you really insist on leaving out the other PISA subscales fine (I'll still approve the paper, because of the high part-whole PISA subtest correlations), but you're weakening the paper by doing so.
The PISA r x CPS is less than PISA r x Math, and they're all in the 0.8-0.9 range. As for sex dimorphism this could be measured using PISA factor scores, which would eliminate specific subtest variance. Look, if you really insist on leaving out the other PISA subscales fine (I'll still approve the paper, because of the high part-whole PISA subtest correlations), but you're weakening the paper by doing so.


Philbrick, you are free to post your comments to any articles on OpenPsych but you do not have reviewer's privileges for OBG, because you were appointed as a reviewer only for ODP. Thanks for your "virtual" approval though :)
This is an interesting paper which I hope to eventually see published. However, a couple of suggestions. (1) I wonder if height is a good measure to use. These are often self-estimates and are thus not very reliable. As such, it might be better to use a number of different estimates to obviate the unreliability. For example, Lynn did a paper on penis length in different European countries. (2) It strikes me that the paper would be strengthened by using all of the PISA data as there is a high correlation between this and IQ.
This is an interesting paper which I hope to eventually see published. However, a couple of suggestions. (1) I wonder if height is a good measure to use. These are often self-estimates and are thus not very reliable. As such, it might be better to use a number of different estimates to obviate the unreliability. For example, Lynn did a paper on penis length in different European countries. (2) It strikes me that the paper would be strengthened by using all of the PISA data as there is a high correlation between this and IQ.


I expressely chose the height data not based on self report but on actual measurement.
There are several flaws with this paper:

1. Sexual dimorphisms are not necessarily due to sexual selection. Many are, in fact, due to differences in natural selection. Males and females differ physiologically for many reasons, often related to differences in diet or reproductive demands. Women, for instance, have a greater need for iron (because of their loss of blood during menstruation). This need constitutes a selection pressure, but we’re in the realm of natural selection, and not sexual selection.

2. Sexual selection results from too many of one sex competing for too few of the other sex. In most mammals, males have to compete for females. This is because males typically re-enter the mate market much sooner than females after mating. Females, by contrast, are absent from the mate market for longer periods because of the time required for pregnancy, lactation, and child care.

In humans, the situation has been more variable. In the tropics, particularly among tropical horticulturalists, women can meet their children’s needs year-round with limited support from their mates. Thus, the costs for men of taking a second wife are lower, and polygyny rates reach high levels (such as among the agricultural peoples of sub-Saharan Africa and Papua-New Guinea). Because these regions have the most intense competition among men for access to women, the pressure of sexual selection is much stronger on men than on women. In these regions, women tend to have a functional appearance with little development of ornamental features, e.g., length of head hair, diversification of hair and eye colors. In contrast, men tend to display traits that, in other mammals, result from intense male-male competition for females, i.e., increased muscle mass, higher bone density, longer penis size, and increased volume of ejaculate.

As one moves away from the tropics, the pressure of sexual selection shifts from males to females. Women depend more on men for food provisioning during the winter, when food gathering and farming are impossible. At the same time, male mortality tends to increase in relation to female mortality because mean hunting distance lengthens. As a result, women are the ones who must compete for mating opportunities.

All of this is contrary to the premises of this paper. The author argues that “sexual selection operated on males via intersexual competition (female choice or preference for smarter men) or intrasexual competition (competition between males for access to females).” In that case, we would expect mean IQ to be highest in the agricultural populations of sub-Saharan Africa and Papua-New-Guinea. Those are the areas where men have to compete the most for women.

3. Let’s put aside for a moment the issue of sexual dimorphism in IQ. If we look at anatomy, East Asians are among the least sexually dimorphic of all humans. Yet they also display some of the world’s highest mean IQ’s. How come?

I cannot recommend acceptance of this paper. The flaws in it seem to be fundamental, although they are not limited to the author. I have met many academics who assume that sex differences are due to sexual selection. I have also met many who idealize polygyny and think that the ideal society is one where horny old men can have as many women as they want.

Such societies do exist, but they are evolutionary dead-ends. A society of intense male-male competition is one where male-male solidarity is precarious at best, and where many complex economic activities become impossible. It’s the sort of society that promotes sexy “big men” who specialize in personality and not character, in bombastic speech and not intelligent debate. Is that the future you want?
There are several flaws with this paper:

1. Sexual dimorphisms are not necessarily due to sexual selection. Many are, in fact, due to differences in natural selection. Males and females differ physiologically for many reasons, often related to differences in diet or reproductive demands. Women, for instance, have a greater need for iron (because of their loss of blood during menstruation). This need constitutes a selection pressure, but we’re in the realm of natural selection, and not sexual selection.
In humans, the situation has been more variable. In the tropics, particularly among tropical horticulturalists, women can meet their children’s needs with limited support from their mates. Thus, the costs for men of taking a second wife are lower, and polygyny rates reach high levels (such as among the agricultural peoples of sub-Saharan Africa and Papua-New Guinea). Because these regions have the most intense competition among men for access to women, the pressure of sexual selection is much stronger on men than on women. In these regions, women tend to have a functional appearance with little development of ornamental features, e.g., length of head hair, diversification of hair and eye colors. In contrast, men tend to display traits that, in other mammals, result from intense male-male competition for females, i.e., increased muscle mass, higher bone density, longer penis size, and increased volume of ejaculate.

As one moves away from the tropics, the pressure of sexual selection shifts from males to females. Women depend more on men for food provisioning during the winter, when food gathering and farming are impossible At the same time, male mortality tends to increase in relation to female mortality because mean hunting distance lengthens. As a result, women are the ones who must compete for mating opportunities.

All of this is contrary to the premises of this paper. The author argues that “sexual selection operated on males via intersexual competition (female choice or preference for smarter men) or intrasexual competition (competition between males for access to females).” In that case, we would expect mean IQ to be highest in the agricultural populations of sub-Saharan Africa and Papua-New-Guinea. Those are the areas where men have to compete the most for women.

I cannot recommend acceptance of this paper. The flaws in it seem to be fundamental, although they are not limited to the author. I have met many academics who assume that sex differences are due to sexual selection. I have also met many who idealize polygyny and think that the ideal society is one where horny old men can have as many women as they want.

Such societies do exist, but they are evolutionary dead-ends. A society of intense male-male competition is one where male-male solidarity is precarious at best, and where many complex economic activities become impossible. It’s the sort of society that promotes sexy “big men” who specialize in personality and not character, in bombastic speech and not intelligent debate. Is that the future you want?


Your entire critique is flawed. You make the fatal logical flaw of assuming that sexual selection for beauty and for intelligence go hand in hand. You make the error of assuming that since in Africa selection is stronger on males, there should be more selection on IQ. But this is a bad mistake, as it's entirely possible that selection in Africa operated entirely on physical traits (see the stronger selection for height among lower IQ people) and less on intelligence. You naively assumed that the kinds of selection do not vary across species, which is the whole point of this paper, subsumed into the brain vs brawn selection.
Yes sexual selection might have been stronger in Africa, but if that was selection for height or strength and not for intelligence, this does not undermine the hypothesis put forward in my paper. Also I do not understand your last question "Is that the future you want?" I never made any recommendations for the society that I want, I am talking about what is and not what ought to be. I never advocated the desirability of a society that promotes sexy big men, as you seem to imply that I did.
Another example of how you completely misunderstood my argument
If we look at anatomy, East Asians are among the least sexually dimorphic of all humans. Yet they also display some of the world’s highest mean IQ’s. How come?
2. Sexual selection results from too many of one sex competing for too few of the other sex. In most mammals, males have to compete for females. This is because males typically re-enter the mate market much sooner than females after mating. Females, by contrast, are absent from the mate market for longer periods because of the time required for pregnancy, lactation, and child care.

This is exactly why I put forward the brawn vs brain model, which you utterly failed to understand. According to this model, East Asians are the most sexually selected for IQ but the least sexually selected for physical attractiveness.
Finally, males are the limiting sex in most species, humans included.There may be societies where men are more or less of the limiting sex compared to others but sexual selection is mostly on males.
The PISA r x CPS is less than PISA r x Math, and they're all in the 0.8-0.9 range. As for sex dimorphism this could be measured using PISA factor scores, which would eliminate specific subtest variance. Look, if you really insist on leaving out the other PISA subscales fine (I'll still approve the paper, because of the high part-whole PISA subtest correlations), but you're weakening the paper by doing so.


Philbrick, you are free to post your comments to any articles on OpenPsych but you do not have reviewer's privileges for OBG, because you were appointed as a reviewer only for ODP. Thanks for your "virtual" approval though :)


Oh, sorry. How embarrassing - I forgot about that completely.
The PISA r x CPS is less than PISA r x Math, and they're all in the 0.8-0.9 range. As for sex dimorphism this could be measured using PISA factor scores, which would eliminate specific subtest variance. Look, if you really insist on leaving out the other PISA subscales fine (I'll still approve the paper, because of the high part-whole PISA subtest correlations), but you're weakening the paper by doing so.


Philbrick, you are free to post your comments to any articles on OpenPsych but you do not have reviewer's privileges for OBG, because you were appointed as a reviewer only for ODP. Thanks for your "virtual" approval though :)


Oh, sorry. How embarrassing - I forgot about that completely.


That's ok
"You make the fatal logical flaw of assuming that sexual selection for beauty and for intelligence go hand in hand."

No, I'm not making that assumption. In fact, I don't believe that sexual selection of any kind has played a major role in increasing mean IQ anywhere (be it male-male competition for women or female-female competition for men). On the other hand, one could argue that male-male competition for women has prevented mean IQ from rising.

"it's entirely possible that selection in Africa operated entirely on physical traits (see the stronger selection for height among lower IQ people) and less on intelligence."

No, when male-male competition for females is intensified, there are certain general consequences, regardless of the species or geographic location. Males have to be physically stronger (to fight off rivals) and have to adopt a certain psychological profile (i.e., ability to intimidate through gestures, verbal bombast, etc.). Both consequences make State formation and social complexification difficult because of the large pool of restless single males. In Africa, this social contradiction is resolved by turning it outward: young men are encouraged to raid neighbouring tribes as a means to get women. Unfortunately, this strategy falls apart when the polity becomes too large. Beyond a certain territorial size, it becomes easier to raid communities within the polity than outside it. (it comes down to the mathematical problem that the external boundary of the polity increases more slowly than the area within the boundary). So State formation repeatedly aborts beyond a certain territorial size.

"East Asians are the most sexually selected for IQ but the least sexually selected for physical attractiveness"

It's impossible to judge the above statement unless you distinguish between sexual selection of women and sexual selection of men. Sexual selection of men has been weak among East Asians, as it has been among Europeans. On the other hand, sexual selection of women has been quite strong among East Asians, not as strong as among Europeans, but certainly above the human average. If we look at traits that seem to be sculpted by sexual selection of women, such as head hair length and smoothness of skin texture, East Asian women look very good.

I have no wish to come down hard on you. You seem to be perpetuating a long tradition among hereditarians, going back to Ernst Mayr, of believing that polygyny has been the main driving force in the steady increase of pre-human and human IQ. Yet if we look at modern human populations, the correlation between mean IQ and polygyny rate is negative, not positive. I don't believe that a high polygyny rate depresses mean IQ (although that argument could be made) but it certainly doesn't raise mean IQ. That experiment has been done, and the results have long been available.
"it's entirely possible that selection in Africa operated entirely on physical traits (see the stronger selection for height among lower IQ people) and less on intelligence."

No, when male-male competition for females is intensified, there are certain general consequences, regardless of the species or geographic location. Males have to be physically stronger (to fight off rivals) and have to adopt a certain psychological profile (i.e., ability to intimidate through gestures, verbal bombast, etc.).


Again you are making the same mistake. Stronger sexual selection is not equal to stronger sexual selection for IQ or height. It's entirely possible that populations with lower male sexual selection/polygny experience stronger sexual selection for height but weaker sexual selection for IQ. If on the other hand a population has lower polygyny/male sexual selection, it's possible that in this population sexual selection operates on IQ and not on height. I can easily see how this can be the case. In less polygynous populations, such as Europeans or East Asians, where there was less sexual selection on males, there was sexual selection on intelligence because men had to provide food through hunting and form stable bonds which required long term planning. But you cannot argue that among Europeans or East Asians there was no sexual selection on men. To reiterate, the mistake you made is that of disregarding the potential moderating effect of sexual selection strength. If sexual selection on men at intensity 10 produces some effects (higher stature, lower IQ) at intensity 1 it can produce opposite effects (lower stature, higher IQ) due to the different intellectual and physical requirements operating at different mating patterns and social conditions. All that matters is that sexual selection was at work. But your equation sexual selection= polygyny= dumb big males is too simplistic. My view is much more nuanced and requires an interaction effect.
I attach the paper with comments by GMeisenberg. I will reply to them soon.

His email: "Dear Davide,

Attached your paper with my comments and a few edits. This is an important subject. "
Here you should include a reference to Table 1. Also, better specify that higher male performance is mainly on non-verbal tests. One of the most consistent results in PISA is a large female advantage in reading in virtually all countries. Incidentally, this female advantage in reading casts doubt on the sexual selection hypothesis because it is more likely social and verbal skills that are sexually selected, not the kind of skills in which males excel. But perhaps our male ancestors used to get access to fertile females not by enticing them but by slamming stones or clubs on their competitors’ heads, which requires spatial rather than verbal ability.

Males historically have been hunters and warriors, two endeavours which require a great deal of visuo-spatial skills and strategic planning, which would require higher fluid intelligence and not much verbal abilities.

Why is this so? Because it is assumed that the amount of genetic polymorphisms that is expressed more or less equally in males and females is assumed to be larger than the amount of genetic variation whose phenotypic expression is sex-limited? This should be made explicit.

This is because the amount of genetic variation whose phenotypic expression is sex-limited is usually much smaller than the additive autosomal genetic variation.


Not only sexual selection but also other forms of natural selection tend to reduce genetic variation. Therefore low standard deviation cannot be taken as evidence for sexual selection. Sex differences in standard deviations can, in theory, but in practice such differences will most likely be too small to be above measurement error. Also, don’t forget population stratification by SES, ethnicity, an whatever that can introduce environmental noise which can easily swamp the genetic signal.


It’s true that also natural selection reduces genetic variation. However, the natural selection model has a worse fit to the data, because it does not account for the negative correlation between sex differences in IQ and SD (r=-.465; p= 0.01; N=44), which instead is better explained by a sexual selection model.
Unfortunately the problem of environmental noise is impossible to tackle effectively but I’ll add a note of caution regarding this in the discussion.

You conclude that there has been more selection for higher intelligence in males than females. Is it possible that there has been more selection for low intelligence in females than males, even before the demographic transition? Perhaps there have always been a few bright women who realized that they are better off remaining single than being married to a good-for-nothing husband and burdened with the enormous task of raising lots of children, in addition to the risks of childbirth. This level of cognitive development was certainly rare in the Palaeolithic, but at somewhat higher levels of cultural evolution it may have been sufficiently frequent to have evolutionary consequences.

It is certainly possible that there was selection for low intelligence in females than males, but in the absence of historical (pre-demographic transition) evidence I cannot discuss this hypothesis.
I attach an updated version of the manuscript, addressing GMeisenberg's comments. GMeisenberg had also spotted an error in the dataset (415 instead of 515 for Estonia), which I corrected. I re-ran the analysis but it's essentially unchanged. Corrected dataset attached.