Intelligence, Critical Thought Capacity, and Heridability

[Bruno Putzeys]

Regression to the mean, while often explained in terms of "pre" and "post" test does not require time or even order of occurrence. It is simply an effect that happens between any two variables that are not fully correlated, regardless of whether those variables are gathered in chronological succession or concurrently.

--

Racism is not so much trying to find an explanation for race disparities. There may or may not be an objective reason for a correlation. Studying this is a slippery slope but is not morally or logically flawed per se.

Racism is failing to take a person at his own worth, but pre-judging him based on statistics gathered on a group to which he happens to belong. Discrimination in general is treating individuals as representative of groups instead of who they are as individuals.

 

[Dave1001]

I think part of the problem is that I'm combining too many questions about regression to the mean and the heridability of adult IQ into each post.

So let me start with just one:

What exactly is the mean that that the adult IQ of of the offspring of adult IQ deviant parents regress to?

I take it that it's the mean adult IQ of the species as a whole? But why would it be, unless we're keyed to a specific adult IQ as a species, and randomness causes individual adult IQs to fall along a bell curve around it?

 

[Dave1001]

I'm just perusing the articles quoted throughout this thread, but I can't find any references to anyone trying to determine heritability based on parents' scores. Heritability is easiest (and hence most commonly) determined through twin and adoption studies. My guess is that judging how a certain trait is passed from parents to children requires figuring out which genes are involved.

How could they possibly know that regression to the mean occurs using twin studies? Wouldn't one have to know what the parents' adult IQ scores were? I would think one would have to control for a lot of environmental factors before determining that heritability difference between twins was strictly random, which I think would be one prerequisite before determining that regression to the mean would occur with adult IQ (unless the adult IQ enriching environment every was raised in was random, the adult IQ of the procreative partners one chose was random, etc.). Am I missing something here?

 

[Dave1001]

Edit

 

[Ivor the Engineer]

A persons IQ score is heavily affected by how the questions are put to that person. Also, what purpose other than racist is there for studying group IQ? What function in the real world does an IQ score have?

Most human endeavours are team based. If you've got high IQ's with no social skills then you’re not going to produce much.

What would the smartest person in the worlds score be on an IQ test if they couldn't read?

As for the idea of multi-dimensional analysis what a load of nonsense! In any set of data you will find correlations between variables. It's called data mining. A very popular way for woo organisations to demonstrate their claims.

Of the research I've read the correlation between the IQ’s of twins is higher than that of separated twins which is higher than that of non-twin siblings. Thus there is both a genetic component and an environmental (including the womb) component.

Most people are good at some things and worse at others. Better to spend time researching how to get people into jobs that are productive and they enjoy.

 

[Bruno Putzeys]

Until you properly understand what regression to the mean means, I don't think it makes sense to continue the discussion on any concrete topic. I can't explain it any simpler than I have. Regression to the mean is a purely mathematical phenomenon that is totally agnostic about where the data came from.

 

[Bruno Putzeys]

A persons IQ score is heavily affected by how the questions are put to that person. Also, what purpose other than racist is there for studying group IQ? What function in the real world does an IQ score have?

You're pretty deluded if you think IQ does not correlate with professional and social success. The correlation is not 100% but to claim it is zero verges on the religious.

What would the smartest person in the worlds score be on an IQ test if they couldn't read?

Conversely, what do you get when you teach someone with an IQ of 90 nuclear physics? At best someone who can recite from memory, but not a nuclear physicist.

Most people are good at some things and worse at others. Better to spend time researching how to get people into jobs that are productive and they enjoy.

People with high IQ, for instance, do well in jobs that require abstract thought.

As for the idea of multi-dimensional analysis what a load of nonsense! In any set of data you will find correlations between variables. It's called data mining. A very popular way for woo organisations to demonstrate their claims.

Nope. If the sample is large enough you can't find correlations that aren't there. If you think you've found one, check against sampling error. If you still believe it's there, posit it as a hypothesis, make a prediction based on it and collect new data.

That's not data mining.

Of the research I've read the correlation between the IQ’s of twins is higher than that of separated twins which is higher than that of non-twin siblings. Thus there is both a genetic component and an environmental (including the womb) component.

Identical twins tend to share a womb. This is why you need to cross-check with adoption studies.

 

[Ivor the Engineer]

You're pretty deluded if you think IQ does not correlate with professional and social success. The correlation is not 100% but to claim it is zero verges on the religious.


Conversely, what do you get when you teach someone with an IQ of 90 nuclear physics? At best someone who can recite from memory, but not a nuclear physicist.


People with high IQ, for instance, do well in jobs that require abstract thought.


Nope. If the sample is large enough you can't find correlations that aren't there. If you think you've found one, check against sampling error. If you still believe it's there, posit it as a hypothesis, make a prediction based on it and collect new data.

That's not data mining.


Identical twins tend to share a womb. This is why you need to cross-check with adoption studies.

To respond to your points:

I did not claim IQ is uncorrelated with any particular variable. Just that a person’s success is determined by lots of other things too. Look at the US president. Pretty successful but by most accounts not too bright. Measuring the one variable (IQ) does not help you or the person very much.

High IQ + no interest = little useful output.

So you agree that a person’s score on an IQ test is dependent on prior knowledge?

So High IQ's can think about abstract concepts. Why not use a test that tests that ability more precisely that IQ then? Call it the Abstract Thought Quotient. (ATQ).

But even if you find a solid correlation it is not the same as cause and effect. There's a correlation between watching violent TV and violent behaviour. Which causes which?

My original sentence included separated twins. Please re-read.

In summary IQ is a blunt tool that is (in my opinion) used way too much and valued far too highly.

 

[Bruno Putzeys]

There's a multiplicity of IQ tests. Most are indeed sensitive to learning (ie. if you do 'em twice you get a higher score). This is unavoidable. One of the things an IQ test tries to assess is how quickly one learns (sees a pattern). The result is that IQ scores for any individual can vary well over 10 points, dependent on the test methodology.
The best tests are the ones that minimise cultural content, for example by working with simple only.

At any rate, an IQ test simply tells you how powerful your engine is. Not how well you drive it. It does not even attempt to figure out the latter.

Feel free to develop a test to evaluate abstract thinking directly. One that will be free from controversy concerning applicability. In other words, one that has better correlation with abstract thinking capability.

Separated twins still share a womb, which means that environmental effects up to that point are difficult to separate from genetic effects.

 

[Ivor the Engineer]

There's a multiplicity of IQ tests. Most are indeed sensitive to learning (ie. if you do 'em twice you get a higher score). This is unavoidable. One of the things an IQ test tries to assess is how quickly one learns (sees a pattern). The result is that IQ scores for any individual can vary well over 10 points, dependent on the test methodology.
The best tests are the ones that minimise cultural content, for example by working with simple only.

At any rate, an IQ test simply tells you how powerful your engine is. Not how well you drive it. It does not even attempt to figure out the latter.

Feel free to develop a test to evaluate abstract thinking directly. One that will be free from controversy concerning applicability. In other words, one that has better correlation with abstract thinking capability.

Separated twins still share a womb, which means that environmental effects up to that point are difficult to separate from genetic effects.

But to carry on your analogy, aren't you driving while taking the IQ test?

What if your drunk, hung over, tired, a loved one has died, etc. etc. What if you don't understand a word or phrase used on the test? What if a concept used is alien to you? What if you’re not particularly motivated to take the test?

All these and more confounding variables are ignored and out pops a number.

On a good day you get the power of the person’s engine. On a bad day you get the standard of their driving! How from the single number do you decide which is which?

Surely it's better to test a person on tasks similar to those that you actually what them to perform along with an interview/trial period where you can see if a good/poor result on a test might be caused by something other than an "exceptional"/"insufficient" IQ?

I was not trying to specify any particular correlation between IQ and the womb, just indicating that it is another variable that needs to be considered when talking about the origins of a persons IQ.

As for personal success, I think there should be another measure called Screw Over Your Fellow Man Quotient (SOYFMQ). Bet most of the worlds leaders would score high on that one...

 

[Bruno Putzeys]

You can botch any psychometric test by refusing to cooperate.

 

[Dave1001]

Until you properly understand what regression to the mean means, I don't think it makes sense to continue the discussion on any concrete topic. I can't explain it any simpler than I have. Regression to the mean is a purely mathematical phenomenon that is totally agnostic about where the data came from.

Bruno, I think there's a disconnect here that doesn't have anything to do with your belief that I don't understand what regression to the mean means. I welcome other folks to tackle some of the questions I've asked about this, though.

 

[Dave1001]

A persons IQ score is heavily affected by how the questions are put to that person. Also, what purpose other than racist is there for studying group IQ? What function in the real world does an IQ score have?

Most human endeavours are team based. If you've got high IQ's with no social skills then you’re not going to produce much.

Well, I think that some folks, as discussed on the wikipedia pages on IQ, believe it's been demonstrated that social skills (social intelligence) in adults is an artifact of adult IQ, which they believe is measure of the capacity to think abstractly. However, I think this is probably counterintuitive, at least to a degree, in a lot of our real world experiences. Of course, for all I know, very socially intelligent people have far higher adult IQ's than do socially awkward and inept people with Ph.D.'s in science and engineering.

What would the smartest person in the worlds score be on an IQ test if they couldn't read?

Well, it's easy to botch the adult IQ of someone who would have the most positively deviant genetic component to adult IQ: simply deprive their brain of oxygen for a few minutes. Not sure what this contributes to our discussion though.

As for the idea of multi-dimensional analysis what a load of nonsense! In any set of data you will find correlations between variables. It's called data mining. A very popular way for woo organisations to demonstrate their claims.

Right. Why does that make multi-dimensional analysis a load of nonsense?

Of the research I've read the correlation between the IQ’s of twins is higher than that of separated twins which is higher than that of non-twin siblings. Thus there is both a genetic component and an environmental (including the womb) component.

Most people are good at some things and worse at others. Better to spend time researching how to get people into jobs that are productive and they enjoy.

I think this board is in part a leisure pursuit for many of us, not a philanthropic endeavor. So why should our every discussion have to first be justifiable as being the activity of maximum utilitarian value to the world that we can be engaged in?

 

[Dave1001]

To respond to your points:

I did not claim IQ is uncorrelated with any particular variable. Just that a person’s success is determined by lots of other things too. Look at the US president. Pretty successful but by most accounts not too bright. Measuring the one variable (IQ) does not help you or the person very much.

High IQ + no interest = little useful output.

So you agree that a person’s score on an IQ test is dependent on prior knowledge?

So High IQ's can think about abstract concepts. Why not use a test that tests that ability more precisely that IQ then? Call it the Abstract Thought Quotient. (ATQ).

But even if you find a solid correlation it is not the same as cause and effect. There's a correlation between watching violent TV and violent behaviour. Which causes which?

My original sentence included separated twins. Please re-read.

In summary IQ is a blunt tool that is (in my opinion) used way too much and valued far too highly.

I'm not sure that George W. Bush and other "successul" people that perform a public shtick of being of average intelligence really are only of average intelligence. In a populist culture such as ours, it may be to their advantage for us to think of them as of average intelligence or below, when in fact they maybe among the very smartest people in the world. Michael Kinsely wrote awhile ago of the danger of Democrats publicly attacking George W. Bush as being dumb, because it might add to his anti-elitist popularity and likeability.

 

[drkitten]

Can you (or someone else) explain regression to the mean in this context to me? There will probably be several follow-up questions. I think I understand regression to the mean as a general principle, but I'm not sure exactly how it operates in the case of IQ heritability.

Basically, IQ is known to have both genetic and environmental influences, and they interact in rather complex ways. A person with a very high measured IQ is therefore likely to have received both good genes and a good developmental environment, and probably a fair amount of luck as well. Similarly, a person with a very low measured IQ is likely to have been dealt a bad hand both from an environmental and a genetic standpoint.

"Regression to the mean" basicaly captures the idea that to be a winning football team, it's not enough to be a good team, but you also need to have the lucky breaks on your side; a bad but lucky team can still beat a good but unlucky one. So if you're at the top of the league tables in the middle of the season, that's likely to reflect both skill and luck -- and since luck is not stable or consistent, you're more likely to move down in the tables than you are up during the second half.

Similarly, if you have an IQ of 150, that represents both good genes and "good luck." Since your children, even your exact clones, are not likely to have the exact same environment as you, they're more likely to have worse luck than yours than better luck, which means that they're more likely to have lower IQs than you than higher. The exact opposite, of course, applies if you've got a very low IQ.

It's not just environment. It's the sum total of any factors that might cause your child to test differently than you did. If you scored really high, then most of the random factors were demonstrably in your favor, but you can't rely on that for your child.

 

[Dave1001]

Basically, IQ is known to have both genetic and environmental influences, and they interact in rather complex ways. A person with a very high measured IQ is therefore likely to have received both good genes and a good developmental environment, and probably a fair amount of luck as well. Similarly, a person with a very low measured IQ is likely to have been dealt a bad hand both from an environmental and a genetic standpoint.

"Regression to the mean" basicaly captures the idea that to be a winning football team, it's not enough to be a good team, but you also need to have the lucky breaks on your side; a bad but lucky team can still beat a good but unlucky one. So if you're at the top of the league tables in the middle of the season, that's likely to reflect both skill and luck -- and since luck is not stable or consistent, you're more likely to move down in the tables than you are up during the second half.

Similarly, if you have an IQ of 150, that represents both good genes and "good luck." Since your children, even your exact clones, are not likely to have the exact same environment as you, they're more likely to have worse luck than yours than better luck, which means that they're more likely to have lower IQs than you than higher. The exact opposite, of course, applies if you've got a very low IQ.

It's not just environment. It's the sum total of any factors that might cause your child to test differently than you did. If you scored really high, then most of the random factors were demonstrably in your favor, but you can't rely on that for your child.

drkitten,
Thanks, that's what I thought. You mentioned environmental luck and nonenvironmental luck as both being possible factors that can lead to regression to the mean.

1. I take it that nonenvironmental luck for the offspring would be the luck of how the genes of the parents combine in the offspring.

2. Doesn't saying that nonenvironmental luck (the luck of how the parental genes combine in the offspring) contributes to regression to the mean presume that human adult IQ is keyed to a certain mean as a species? Otherwise I would think luck would be equally likely to go either way: two parents with a 120 IQ would be equally likely to have nonenvironmental (genetic) luck result in them having an offspring with a higher IQ as a lower IQ if it wasn't the case that we'd be keyed to a mean IQ as a species.

3. I understand how even if we weren't keyed to a mean IQ as a species, how environmental luck could still be a factor that results to regression to the mean.

 

[Dave1001]

Edit

 

[drkitten]

1. I take it that nonenvironmental luck for the offspring would be the luck of how the genes of the parents combine in the offspring.

That's one factor. Another factor would simply be whether the offspring live in an environment where they can get a good night's sleep on a regular basis. When I was in college, I lived for a year across the street from the emergency room of a major hospital, and I don't think I got uninterrupted sleep the entire year. I'm sure that my tested IQ would have been at least five points lower during that year simply because I was always borderline sleep-deprived. That doesn't reflect anything significant about my cognitive capacity except that I wasn't smart enough to pick a quieter apartment.

"Regression to the mean" includes all the factors that make different people test differently or the same people test differently at different times.

Doesn't saying that nonenvironmental luck (the luck of how the parental genes combine in the offspring) contributes to regression to the mean presume that human adult IQ is keyed to a certain mean as a species?

I don't know what you mean by "keyed to" in this context. Human IQ has a mean (defined by the test scaling procedures as 100). But there's nothing special about that. Any data set has a mean.

And yes, human IQ does follow an approximately bell-shaped curve around the mean.

Otherwise I would think luck would be equally likely to go either way: two parents with a 120 IQ would be equally likely to have nonenvironmental (genetic) luck result in them having an offspring with a higher IQ as a lower IQ if it wasn't the case that we'd be keyed to a mean IQ as a species.

First, non-environmental != genetic.

But second, you're not looking at the full picture. We can think of it this way. Our hypothetical parent has a measured IQ of 120. Some of that component is genetic, some of it is not. If we could somehow isolate the "intelligence" genes (which we can't, even theoretically, because the nature/nuture debate doesn't run that way), we might find any of the following situations

* His "genetic" IQ should be 130, but he had a rough childhood and a bad morning on testing day, which lowered his score by 10 points.
* His "genetic" IQ should be 110, but he got really lucky with his second form teacher and was feeling his oats that morning, so he got 10 bonus points
* His "genetic" IQ should be 120 and he had an "average" life and day.

Which is most likely?

Well, given that IQ follows a bell-shaped curve, people with IQs of 110 are much more common than people with IQs of 130. So, just as if you pick a person at random on the streets of Tokyo, you're more likely to pick a person who speaks Japanese than Swahili, so is he more likely to have an IQ of 110 than 130. Which means that his children are likely to inherit his "genetic" IQ of 110 and then have a "normal" day on top of that -- his children will probably test at 110. Voila --- regression to the mean.

Notice that this applies even if his children are exact clones of him. We don't need to worry about the mother's contribution or possible mutations or wierd genetic effects at all to get this effect.

 

[Dave1001]

That's one factor. Another factor would simply be whether the offspring live in an environment where they can get a good night's sleep on a regular basis. When I was in college, I lived for a year across the street from the emergency room of a major hospital, and I don't think I got uninterrupted sleep the entire year. I'm sure that my tested IQ would have been at least five points lower during that year simply because I was always borderline sleep-deprived. That doesn't reflect anything significant about my cognitive capacity except that I wasn't smart enough to pick a quieter apartment.

Oh, I thought you were considering these factors as part of environmental factors, along with prenatal environment, early child developmental environment, etc.

"Regression to the mean" includes all the factors that make different people test differently or the same people test differently at different times.

Yes, I understand that.

I don't know what you mean by "keyed to" in this context. Human IQ has a mean (defined by the test scaling procedures as 100). But there's nothing special about that. Any data set has a mean.

And yes, human IQ does follow an approximately bell-shaped curve around the mean.

Let me clarify. Why else would the nonheritable, nonenvironmental (chance elements of how the parent's genes combined in the offspring) regress to the mean IQ of humans, if we weren't genetically keyed to that mean IQ?

Let me frame it differently. Let's say a catastrophe in 2007 wiped out everyone that currently had an adult IQ below 140 as measured in 2006. I understand that in 2007 the new mean would be called 100. But it would still be a population of very deviant adult IQ compared to the world's population in 2006. Now let's fast forward to the year 2066. The survivors of 2007 have all had kids that have grown up to be adults. If we are keyed to an IQ as a species, and 140 is very deviant from it, then the adult IQ of the entire population would be expected to regress towards the 2006 mean. Imagine we could control for socialization based environmental factors, perhaps by making Turing test passable robots that act like all the lower IQ people who were wiped out in 2006. If there is a regression to the 2006 mean in the 2066 offspring population, I think that would indicate that we're genetically keyed to a certain IQ, and that it's a nonheritable form of genetic luck when folks (controlling for environmental factors) are born that will develop a substantially deviant adult IQ. I don't understand logically why there would be a deviation only to the new 2007 mean in the 2066 population, because the 2007 mean would have been very deviant in 2006. Instead, I would think that nonenvironmental luck would make it equally likely for two parents with deviant IQ in 2007 to have an even more deviant offspring as a less deviant offspring. But then that same principle would be in operation now, and apparently it's not.

That's why I think the information provided on wikipedia and others indicates we're keyed to an IQ as a species. But there seems to be conflicting information regarding that, such as the Flyn Effect.

First, non-environmental != genetic.

But second, you're not looking at the full picture. We can think of it this way. Our hypothetical parent has a measured IQ of 120. Some of that component is genetic, some of it is not. If we could somehow isolate the "intelligence" genes (which we can't, even theoretically, because the nature/nuture debate doesn't run that way), we might find any of the following situations

* His "genetic" IQ should be 130, but he had a rough childhood and a bad morning on testing day, which lowered his score by 10 points.
* His "genetic" IQ should be 110, but he got really lucky with his second form teacher and was feeling his oats that morning, so he got 10 bonus points
* His "genetic" IQ should be 120 and he had an "average" life and day.


Which is most likely?

Well, given that IQ follows a bell-shaped curve, people with IQs of 110 are much more common than people with IQs of 130. So, just as if you pick a person at random on the streets of Tokyo, you're more likely to pick a person who speaks Japanese than Swahili, so is he more likely to have an IQ of 110 than 130. Which means that his children are likely to inherit his "genetic" IQ of 110 and then have a "normal" day on top of that -- his children will probably test at 110. Voila --- regression to the mean.

Notice that this applies even if his children are exact clones of him. We don't need to worry about the mother's contribution or possible mutations or wierd genetic effects at all to get this effect.

Yep, I understand all that. But I don't think it renders my questions moot. I lumped bad test days in with environmental factors, which I think clears up our disconnect on this topic.

 

[Bruno Putzeys]

Let me clarify. Why else would the nonheritable, nonenvironmental (chance elements of how the parent's genes combined in the offspring) regress to the mean IQ of humans, if we weren't genetically keyed to that mean IQ?

This is pretty strong evidence you have no idea what regression to the mean means. Puhleeze read up on the maths before confusing yourself further. It is maths pure and plain and has nothing to do with whichever subject matter it is applied to.