Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Breeder's Equation

edge.org

Post reply on HN
Print view View original

61–70 of 76 posts

Re: The Breeder's Equation

#61

Post by bordercases »

ImaCake wrote: post #56
mynameishere wrote: post #23

Earlier quoted context omitted.

This kind of quibbling always comes up. He's just using professors and national merit finalists as examples. The truth is, you could probably take the top 10 percent highest-earning car salesmen (EDIT: and women), drop them on a desert island and, after two generations, get the highest-IQ society the world has ever known. But that's less obvious, so why use such an example?

But it’s not clear at all that intelligence is a single trait. It is incredibly messy and this means any analysis of it has to be very careful of confounders. In contrast, height is by many definitions a single trait, we know exactly what confounds it’s measurement (and those things are easy to measure!) and we have a pretty good estimate of it’s heritability.

Remapping intelligence to certain cognitive performance parameters, many of which can be measured in a very metrical and objective way, still tends to produce correlations. See Figure 3 of this article: [0] - there is a mix of simple perceptual tasks and more complex reasoning ones, which on the surface appear independent. The paper admits at the end that intelligence isn't one kind of task, but instead many kinds of systems used together. That still doesn't prevent their ability to construct a version of g-factor.

And very simple metrics, taken alone, seem to inter-correlate well with IQ scores, like reaction time.[1] This correlation increases with age, suggesting a biological factor in preserving brain integrity that would nullify educational effects of intelligence.

If your version of intelligence has nothing to do with cognitive activity I don't know what to tell you.

It's true that effectiveness in society might not be correlated with intelligence. That says more about what society rewards and the kind of labor required to get it. And I don't mean to say that society should reward people based off of test scores, that's dumb. Nor should we assortate people based off of pure test scores, also dumb.

But I see no real conceptual problems regarding intelligence that doesn't try to conflate intelligence entirely with general effectiveness. There are IQ realists like Steven Hsu that don't do this, placing it on equal importance with conscientiousness [4]. As with height, there are innate and behavioral/developmental components of general effectiveness (I could dope my child with growth hormone to make him taller, or mal-nourish them to go the other way). As with emotional intelligence, being a bastion of full-nurture theorists for behavior that look for non-academic modes of measuring effectiveness, we can get heritability correlates of 0.2-0.4[2], and for psychopathy up to 0.5[3], mediated by biological factors like inherent oxytocin concentrations.

It's hard to say you can succeed without without any ability to learn, but there are high paying jobs that involve mainly physical labor (while taking on some mortality risk which can be managed). And there are people who are smart assholes, and other people who are dumb yet compassionate. So I have no problem with accepting the cognitive version of intelligence as with the right attitude there is no anti-social conclusion to risk. Society takes all sorts of people to exist, even if to some extent, the more well-rounded the better.

[0] "Fractionating human intelligence", www.owenlab.uwo.ca/pdf/2012%20-%20Hampshire%20-%20Neuron.pdf

[1] "The relationship between intelligence and reaction time varies with age", https://www.sciencedirect.com/science/article/pii/S016028961...

[2] https://www.researchgate.net/publication/23302097_A_Behavior...

[3] https://www.geneticlifehacks.com/psychopath-genes-born-not-m...

[4] "Genetics and Intelligence", https://youtube.com/watch?v=62jZENi1ed8

Re: The Breeder's Equation

#62

Post by civilized »

civilized wrote: post #50
s28l wrote: post #48

Earlier quoted context omitted.

The breeder's equation described the expected change in a _population_ from one generation to the next due to selection pressures (natural or artificial). For example, say a cattle rancher with a large herd may want to increase the average weight. So if he only breeds those animals in the top half of weight, what is the expected change in weight from one generation to the next? For another example, say a particular s…

> He doesn't define what distinguishes the population of 120 IQ parents from the population as a whole. ...Surely what distinguishes them is that they have IQs of 120, whereas the parent population has average IQ of 100? > Further, he says that 120 IQ parents having children with mean IQ of 110 is an example of regression to the mean. I would say the exact opposite: the 110 IQ children _define the mean_ of the next g…

PS: if anyone's interested in a meaningful, serious bioethical discussion of eugenics, I can highly recommend this paper: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8321981/

Re: The Breeder's Equation

#63

Post by civilized »

ImaCake wrote: post #58

This is a pretty poor quality article. I say this as someone who works in a population genetics research group. If you like some actually thoughtful reading on genetics and eugenics I would point you towards gwern who, as far as I can tell, seems to at least get the science and math right. https://www.gwern.net/tags/genetics

Could we trouble you for a little detail about what's wrong with the article? It seems very clear and understandable for the layman, so it would be good to have an equally clear explanation of what's wrong with it.

Re: The Breeder's Equation

#64

Post by civilized »

ImaCake wrote: post #56
mynameishere wrote: post #23

Earlier quoted context omitted.

This kind of quibbling always comes up. He's just using professors and national merit finalists as examples. The truth is, you could probably take the top 10 percent highest-earning car salesmen (EDIT: and women), drop them on a desert island and, after two generations, get the highest-IQ society the world has ever known. But that's less obvious, so why use such an example?

But it’s not clear at all that intelligence is a single trait. It is incredibly messy and this means any analysis of it has to be very careful of confounders. In contrast, height is by many definitions a single trait, we know exactly what confounds it’s measurement (and those things are easy to measure!) and we have a pretty good estimate of it’s heritability.

Does it matter if intelligence is a "single trait"? IQ, at least, is a single trait and it's empirically heritable. There has been a great deal of work to address the confounders (e.g. twin and adoption studies) and shore up the construct validity of IQ. Nobody thinks IQ = intelligence, but it's a measurable trait that says a lot about intelligence.

IQ isn't on the same level of definiteness as height, but I don't think you have a bright line of "trait validity" with height on one side and IQ on the other.

Re: The Breeder's Equation

#65

Post by Causality1 »

bagels wrote: post #54
Causality1 wrote: post #18

Like most looming disasters, whether or not this is a real problem depends on quantifying the problem. Is this likely to significantly affect the population intelligence before we learn to engineer our children, say within the next 200 years?

I always find these statements interesting. Do you think many people 200 years ago based their actions on what effect they would have on those people 200 years hence?

Oh I'm not saying we will. I'm just saying that even taking the claimed effect at face value, it only has significance if it can hurt our genetic intelligence faster than engineering can improve it.

Re: The Breeder's Equation

#66

Post by bsder »

hirundo wrote: post #6

"And of course the breeder’s equation explains how average IQ potential is declining today, because of low fertility among highly educated women." In the Heinlein universe, the Howard Foundation runs a voluntary eugenics program by paying people with particularly long lived ancestors to mate with each other. Protagonist Lazarus Long is a product of this. The author is saying that this would work as expected, with off…

This wouldn't work for "longevity" because "longevity" isn't a bell curve.

If "longevity" were a bell curve, we would see some humans at the 150 years old mark. We do not. Human longevity has a "cliff" that it won't go past.

Selective breeding could get you arbitrarily close to whatever that cliff actually is, but no further.

To get further than that cliff, you would have to fix all the issues that cause it. There is no guarantee (and practically an anti-guarantee) that selective breeding would ever accomplish that.

Re: The Breeder's Equation

#67

Post by setr »

bsder wrote: post #66
hirundo wrote: post #6

"And of course the breeder’s equation explains how average IQ potential is declining today, because of low fertility among highly educated women." In the Heinlein universe, the Howard Foundation runs a voluntary eugenics program by paying people with particularly long lived ancestors to mate with each other. Protagonist Lazarus Long is a product of this. The author is saying that this would work as expected, with off…

This wouldn't work for "longevity" because "longevity" isn't a bell curve. If "longevity" were a bell curve, we would see some humans at the 150 years old mark. We do not. Human longevity has a "cliff" that it won't go past. Selective breeding could get you arbitrarily close to whatever that cliff actually is, but no further. To get further than that cliff, you would have to fix all the issues that cause it. There is…

> If "longevity" were a bell curve, we would see some humans at the 150 years old mark. We do not.

I agree with everything else (including that its not a bell curve), but this particular reasoning seems off — it could still follow a bell curve even if we never see the rarer items… because we’re only talking about approximating a bell curve. Just as height’s bell curve suggests a 30ft human could eventually occur, but we’d never actually expect it to actually happen in reality (because confounding factors will start to trigger) - but we still say height follows a bell curve.

Re: The Breeder's Equation

#68

Post by bsder »

setr wrote: post #67
bsder wrote: post #66

Earlier quoted context omitted.

This wouldn't work for "longevity" because "longevity" isn't a bell curve. If "longevity" were a bell curve, we would see some humans at the 150 years old mark. We do not. Human longevity has a "cliff" that it won't go past. Selective breeding could get you arbitrarily close to whatever that cliff actually is, but no further. To get further than that cliff, you would have to fix all the issues that cause it. There is…

> If "longevity" were a bell curve, we would see some humans at the 150 years old mark. We do not. I agree with everything else (including that its not a bell curve), but this particular reasoning seems off — it could still follow a bell curve even if we never see the rarer items… because we’re only talking about approximating a bell curve. Just as height’s bell curve suggests a 30ft human could eventually occur, but…

No, the bell curve for height does not suggest a 30ft human would ever occur. That would be somewhere around 90+ standard deviations out--that's so wildly improbable as to be effectively 0. A "mere" 7 standard deviations is effectively 1 in a billion. And that's somewhere around a 7 1/2 foot tall human (who we do see occasionally).

"Standard deviation of lifespan" (a somewhat erroneous idea since we're not actually on a bell curve) is somewhere around 10 years. 150 years old is 7 standard deviations--improbable but not impossible. If we're following a bell curve, 1 in a billion people should be 140 years+ and 1 in 10,000 should be 120 years+.

The fact that we don't see these probabilities is extremely strong evidence we're not on a bell curve. And not being on a bell curve immediately throws out huge chunks of statistical "rules".

Re: The Breeder's Equation

#69

Post by bayesian_horse »

daenz wrote: post #31
bayesian_horse wrote: post #20

The heritability of intelligence is overhyped, and in my opinion of no practical use. It's extremely hard to find data where it is even possible to distinguish between genetic and environmental influence. That would be the studies of the kind "twins raised apart". And that already excludes the reality that those twins will probably grow up in a similar background, similar health and educational system etc. The talk a…

>The talk about heritable IQ is pointless, even dangerous, because it leads to complacency around issues around structural racism and equality in education. Can you say more about this? Why would it lead to complacency? If there is scientific merit to some ideas like IQ heritability, don't you think we need to find a way to incorporate those ideas safely into our formulation of society, rather than suppress them as h…

You never heard about the trope of African people being called genetically inferior? That one is reverberating around time and space in a million variations.

And if intelligence is genetic, well, you don't need to improve their public schools to help them achieve equal grades, do you?

Re: The Breeder's Equation

#70

Post by bayesian_horse »

pjscott wrote: post #25
bayesian_horse wrote: post #20

The heritability of intelligence is overhyped, and in my opinion of no practical use. It's extremely hard to find data where it is even possible to distinguish between genetic and environmental influence. That would be the studies of the kind "twins raised apart". And that already excludes the reality that those twins will probably grow up in a similar background, similar health and educational system etc. The talk a…

The usual studies for disentangling genetics and environment involve twins raised together: they get basically the same shared environment -- this is a feature, not a bug -- but identical twins are more closely related than fraternal twins. For example, say you're trying to find out how much of adult height comes down to genetics (in a modern, non-malnourished environment). You measure the heights of a bunch of twins…

Yes. But then people take that street number from a "twins raised together" study and quote it as heritability of intelligence overall.

My problem with this: Yes, if you put extreme limits on the environmental factors, you are mostly left with genetic factors. But that is not the reality of Human existence.

Post reply on HN