Does the use of "regression to the mean" here confuse or enlighten? I'm used to thinking of Regression To The Mean as a purely statistical phenomenon. If you - say - one day make your best score ever at the dartboard, then statistically your performance the following day is likely to be worse - closer to the mean.
The issue here is not whether there is _regression_ taking place in "IQ", but whether there is an optimal IQ range, such that any large deviation in either direction is likely to be maladaptive - just as elephants don't keep getting heavier, or giraffes taller. I don't see any sign looking around that ultra-smart people are better at reproducing than anybody else, so perhaps if we want to answer the question we should look at just why that is the case. Surely it can't just be because we're all maladapted nerds who spend all our time debating the meaning of "regression to the mean" on substack when we should be out there breeeding?
Since IQ gets rebalanced to be 100, there will always be low IQ folk.
There should be a 2000g (or 2001g? Which becomes 1g) fixed IQ exam that stops being balanced, the IQ of 2001. With future 2024 IQ tests, which avg 100, might avg 105 1g points, or only 95.
Having a fixed standard g-general intelligence test that isn’t rebalanced will become important as AI changes what kinds of intelligence gets rewarded and increased thru school maturation of cognitive ability. More college educated folk should take more IQ tests, like every 10 years.
I see a fourth possibility to add to Friedman's list: High IQ may lead to children with lower propensity to reproduce. I have the impression (I'm not sure if I've seen hard data) that two high-IQ parents are more likely than average to have autistic children. If this is true, then high-IQ genes would be somewhat self-limiting.
A fifth possibility: a smarter brain may make someone less desirable as a mate (or at least be a less important trait than other things like physical attractiveness or social skills). If these traits are not correlated with IQ, especially at the high end of the distribution, any selection for IQ would be weaker than selection for attractiveness or social skills. If social skills are negatively correlated with IQ, particularly at the high end of the distribution, then very high IQ might be a negative factor for reproduction.
One slightly prosaic possibility is that high-intelligence trades off against getting stuck in the birth canal! There is a correlation between IQ and head size, and babies with large heads are more at risk during birth.
It is only required that there were downsides to higher intelligence in the paleolithic environment. Whatever was keeping intelligence at its actual level may not be doing anything today. It is also only required that high intelligence reduce reproductive success. It may not reduce quality of life, lifespan, happiness, or anything else we care about.
'I would speculate that for many generations, evolution did not care much about your IQ. In the past few hundred years, with modern culture, IQ began to matter more.'
I would say you have this entirely backwards. Evolution cared very, very much about IQ for a very long time, but cares much less for it now in modern culture. The answer to Friedman's question of 'why are there stupid people' is, there essentially aren't stupid people. Chimpanzees are very intelligent for animals and their average IQ estimates come in the range of 30, and very smart chimps as high as 70-80. An average human then is much further above a chimp in terms of SDs than they are behind the smartest humans ever. Even humans with severe genetic disorders like Downs Syndrome can use language and have vocabularies greater than highly selected and trained chimpanzees.
The modern era has reduced selection for IQ because we actively protect and accommodate for people with low IQs, who would just have died in early childhood in previous years.
While there are a lot of good points in both the article and the comments, I think the most likely answer is Thomas L. Hutcheson’s in the comments here. “Stupid” is relative, and we were a lot more stupid in the past. Intelligence has been selected for, but not everywhere all the time, due to all the different factors folks here have mentioned. Evolution is usually a slog.
It may also be that there's no feasible way for new mutations to improve the average or reduce the variance of human intelligence much, or to do so without introducing other non-adaptive traits. One can't always assume that every trait exists solely because it's adaptive compared with other imagined possibilities. To reiterate an old joke about this aspect of evolution: there's a reason why deer don't have built-in laser cannons to protect them from predators, and it isn't because that wouldn't be adaptive.
Or it may be that greater intelligence can only be achieved by (combinations of) genes that have adverse maladaptive side effects. For example, genes for greater intelligence may correlate with risk of relatives developing disabilities. Highly intelligent people are more likely than average to have autistic siblings, this is pretty well confirmed.
There are two effective strategies to solving problems. Create your own solution (rare), or copy or borrow one from somewhere else (vast majority aka culture). It doesn’t take much upstairs to copy or mimic or use tools or processes created by others. However, if just about everyone is copying the majority, this creates a competitive niche strategy of creating novel solutions. But this requires more intelligence to use imagination, consider tradeoffs, avoid potential new risks, and to combine ideas or transfer similar solutions from one domain to another.
This tracks closely to the hypothesis that there must be costs to intelligence (energy or such), but adds that the cost may actually be less copying and more wasted time energy and risk with novelty.
Thus the dominant strategy would be cultural copying, which is efficient and safe and doesn’t require much thinking, and the niche strategy would be to explore/create novelty which requires brain power.
What's the selection pressure against stupid people having babies?
I can imagine, perhaps, a future environment where smart people are more likely to survive than stupid people but that environment is not now and I would guess that people with low IQ are more likely to have more babies in the current environment, no?
What you say is probably mostly true in the general case. I doubt my opinion is any more true in that regard but I think your piece misses an extremely important factor specific to humans today. Less food doesn't just mean shorter, less cerebral people have an advantage in one way, it also makes offspring shorter (and less intelligent?). We have been getting taller for ~200 years and we don't know when that will stop. I suspect the same is happening regarding intelligence. So while the answer to why some people lack intelligence, and especially for severe low intelligence, is sometimes due to pregnancy issues, environmental issues like lead poisoning or mother's drug addiction, poor infant care, etc., it is more commonly a family history of insufficient calorie and nutrient intake.
Re: Friedman: I will hop on an unpopular soap box and highlight that per Darwin, evolution is the long and unconscious process of biological change in response to natural (e.g. non-human) pressures, whereas breeding is something conscious, chosen, selective, and unnatural. The two processes ought not to be conflated.
For social-political reasons, we have trouble speaking about humans in the same way that we do livestock. But the same forces of conscious selection and breeding are at play in humans as there are in livestock. You do not always want intelligent livestock for the same reason you do not always want intelligent people. We understand a great deal more, practically, about breeding livestock than we do people because many animals have faster breeding cycles and it is easier to define "success" with the animals. A good meat chicken breed is one that is ready for slaughter within weeks of specialized feeding and is resistant to some kinds of disease; a good laying chicken is good at producing eggs while also being resistant to disease within certain conditions. When you narrow down the traits you want, it becomes more feasible to create the kind of breed that you want. It is perhaps more challenging to select for effective hybrids than it is to choose specialists: in the books and catalogs of chicken breeds, it's easy to get into the weeds with the relative virtues of the hybrids than it is to evaluate the suitability of specialists.
Much as recent and concentrated selective breeding is more relevant when analyzing commercial chicken breeds than an analysis of evolution, the same goes for people. So if you are trying to breed economics professors that have stronger lecturing traits, you will select for height, beauty, longevity, glibness, and perhaps some Machiavellian cunning for dealing with administrators and grant committees. Physical traits you can see, but internal traits are much tougher to evaluate. But the match between Rose and Milton Friedman was perhaps less optimized for externals, and for Mrs. Friedman, marriage to a giant would have perhaps been impractical.
Due to survival of the fittest everyone can’t have a high IQ that would be bad for the world and the economy if everyone can excel at everything.
Does the use of "regression to the mean" here confuse or enlighten? I'm used to thinking of Regression To The Mean as a purely statistical phenomenon. If you - say - one day make your best score ever at the dartboard, then statistically your performance the following day is likely to be worse - closer to the mean.
The issue here is not whether there is _regression_ taking place in "IQ", but whether there is an optimal IQ range, such that any large deviation in either direction is likely to be maladaptive - just as elephants don't keep getting heavier, or giraffes taller. I don't see any sign looking around that ultra-smart people are better at reproducing than anybody else, so perhaps if we want to answer the question we should look at just why that is the case. Surely it can't just be because we're all maladapted nerds who spend all our time debating the meaning of "regression to the mean" on substack when we should be out there breeeding?
Since IQ gets rebalanced to be 100, there will always be low IQ folk.
There should be a 2000g (or 2001g? Which becomes 1g) fixed IQ exam that stops being balanced, the IQ of 2001. With future 2024 IQ tests, which avg 100, might avg 105 1g points, or only 95.
Having a fixed standard g-general intelligence test that isn’t rebalanced will become important as AI changes what kinds of intelligence gets rewarded and increased thru school maturation of cognitive ability. More college educated folk should take more IQ tests, like every 10 years.
I see a fourth possibility to add to Friedman's list: High IQ may lead to children with lower propensity to reproduce. I have the impression (I'm not sure if I've seen hard data) that two high-IQ parents are more likely than average to have autistic children. If this is true, then high-IQ genes would be somewhat self-limiting.
A fifth possibility: a smarter brain may make someone less desirable as a mate (or at least be a less important trait than other things like physical attractiveness or social skills). If these traits are not correlated with IQ, especially at the high end of the distribution, any selection for IQ would be weaker than selection for attractiveness or social skills. If social skills are negatively correlated with IQ, particularly at the high end of the distribution, then very high IQ might be a negative factor for reproduction.
One slightly prosaic possibility is that high-intelligence trades off against getting stuck in the birth canal! There is a correlation between IQ and head size, and babies with large heads are more at risk during birth.
It is only required that there were downsides to higher intelligence in the paleolithic environment. Whatever was keeping intelligence at its actual level may not be doing anything today. It is also only required that high intelligence reduce reproductive success. It may not reduce quality of life, lifespan, happiness, or anything else we care about.
'I would speculate that for many generations, evolution did not care much about your IQ. In the past few hundred years, with modern culture, IQ began to matter more.'
I would say you have this entirely backwards. Evolution cared very, very much about IQ for a very long time, but cares much less for it now in modern culture. The answer to Friedman's question of 'why are there stupid people' is, there essentially aren't stupid people. Chimpanzees are very intelligent for animals and their average IQ estimates come in the range of 30, and very smart chimps as high as 70-80. An average human then is much further above a chimp in terms of SDs than they are behind the smartest humans ever. Even humans with severe genetic disorders like Downs Syndrome can use language and have vocabularies greater than highly selected and trained chimpanzees.
The modern era has reduced selection for IQ because we actively protect and accommodate for people with low IQs, who would just have died in early childhood in previous years.
Choices. We all have choices to make.
This speaks to a lot of my thoughts on this subject.
"If I'm so intelligent why am I not more successful?", type thoughts.
Or how you want to learn about the experience of being more intelligent and even asking questions is offensive to those with lower IQs.
Arnold
Lamark and Lysenko were certain acquired characteristics can be inherited.
Lysenko and his ideas created starvation that killed millions.
Adopting similar explanations seems . . . foolish.
No mater how popular.
Thanks
Clay
While there are a lot of good points in both the article and the comments, I think the most likely answer is Thomas L. Hutcheson’s in the comments here. “Stupid” is relative, and we were a lot more stupid in the past. Intelligence has been selected for, but not everywhere all the time, due to all the different factors folks here have mentioned. Evolution is usually a slog.
It may also be that there's no feasible way for new mutations to improve the average or reduce the variance of human intelligence much, or to do so without introducing other non-adaptive traits. One can't always assume that every trait exists solely because it's adaptive compared with other imagined possibilities. To reiterate an old joke about this aspect of evolution: there's a reason why deer don't have built-in laser cannons to protect them from predators, and it isn't because that wouldn't be adaptive.
Or it may be that greater intelligence can only be achieved by (combinations of) genes that have adverse maladaptive side effects. For example, genes for greater intelligence may correlate with risk of relatives developing disabilities. Highly intelligent people are more likely than average to have autistic siblings, this is pretty well confirmed.
Another hypothesis.
There are two effective strategies to solving problems. Create your own solution (rare), or copy or borrow one from somewhere else (vast majority aka culture). It doesn’t take much upstairs to copy or mimic or use tools or processes created by others. However, if just about everyone is copying the majority, this creates a competitive niche strategy of creating novel solutions. But this requires more intelligence to use imagination, consider tradeoffs, avoid potential new risks, and to combine ideas or transfer similar solutions from one domain to another.
This tracks closely to the hypothesis that there must be costs to intelligence (energy or such), but adds that the cost may actually be less copying and more wasted time energy and risk with novelty.
Thus the dominant strategy would be cultural copying, which is efficient and safe and doesn’t require much thinking, and the niche strategy would be to explore/create novelty which requires brain power.
What's the selection pressure against stupid people having babies?
I can imagine, perhaps, a future environment where smart people are more likely to survive than stupid people but that environment is not now and I would guess that people with low IQ are more likely to have more babies in the current environment, no?
Are tall people smarter?
What you say is probably mostly true in the general case. I doubt my opinion is any more true in that regard but I think your piece misses an extremely important factor specific to humans today. Less food doesn't just mean shorter, less cerebral people have an advantage in one way, it also makes offspring shorter (and less intelligent?). We have been getting taller for ~200 years and we don't know when that will stop. I suspect the same is happening regarding intelligence. So while the answer to why some people lack intelligence, and especially for severe low intelligence, is sometimes due to pregnancy issues, environmental issues like lead poisoning or mother's drug addiction, poor infant care, etc., it is more commonly a family history of insufficient calorie and nutrient intake.
https://australian.museum/learn/science/human-evolution/how-have-we-changed-since-our-species-first-appeared/#:~:text=We%20are%20now%20generally%20shorter,height%20has%20started%20to%20increase.
Maybe intelligence HAS been drifting up over the eons.
Re: Friedman: I will hop on an unpopular soap box and highlight that per Darwin, evolution is the long and unconscious process of biological change in response to natural (e.g. non-human) pressures, whereas breeding is something conscious, chosen, selective, and unnatural. The two processes ought not to be conflated.
For social-political reasons, we have trouble speaking about humans in the same way that we do livestock. But the same forces of conscious selection and breeding are at play in humans as there are in livestock. You do not always want intelligent livestock for the same reason you do not always want intelligent people. We understand a great deal more, practically, about breeding livestock than we do people because many animals have faster breeding cycles and it is easier to define "success" with the animals. A good meat chicken breed is one that is ready for slaughter within weeks of specialized feeding and is resistant to some kinds of disease; a good laying chicken is good at producing eggs while also being resistant to disease within certain conditions. When you narrow down the traits you want, it becomes more feasible to create the kind of breed that you want. It is perhaps more challenging to select for effective hybrids than it is to choose specialists: in the books and catalogs of chicken breeds, it's easy to get into the weeds with the relative virtues of the hybrids than it is to evaluate the suitability of specialists.
Much as recent and concentrated selective breeding is more relevant when analyzing commercial chicken breeds than an analysis of evolution, the same goes for people. So if you are trying to breed economics professors that have stronger lecturing traits, you will select for height, beauty, longevity, glibness, and perhaps some Machiavellian cunning for dealing with administrators and grant committees. Physical traits you can see, but internal traits are much tougher to evaluate. But the match between Rose and Milton Friedman was perhaps less optimized for externals, and for Mrs. Friedman, marriage to a giant would have perhaps been impractical.