The Science of Everything Podcast - Episode 81: Intelligence Part 2

Episode Date: February 24, 2017

Continuing the series on intelligence, in this episode I discuss the genetic and neurological correlated of intelligences, the efficacy of measures to increase intelligence, and the causes and consequences of the Flynn effect. Also includes an analysis of the heritability of intelligence and how to properly interpret heritabilty research. Recommended pre-listening is Episode 80: Intelligence Part 1.

you're listening to The Science of Everything podcast episode 81. Intelligence Part 2. I'm your host, James Fodor. So this episode will be continuing from part one of the series on intelligence, so episode 80. I recommend you listen to that before listening to this one. And in this episode, we'll be building on some of the material that we discussed last time. So in the last episode, we talked about some definitions of intelligence, various methods of measuring intelligence.

And we're talking about the G factor, how that's defined, how that's determined, and various debates about the interpretation of the G factor, theories of multiple intelligence and so on. In this episode, what we're going to cover is some of the research relating to the biology of intelligence, including genetics and neuroscience of intelligence. We're going to look at some of the literature relating to interventions to increase intelligence. We'll also look at some of the research. on the heritability of intelligence and the sort of nature versus nurture debate

and we'll be talking about some of the methodological issues relating to some of that research. And finally, we'll also discuss the Flynn effect, the strange phenomena of increasing IQ scores over time. And some of the methodological discussions that we'll have, the issues that we'll discuss in this episode will be relevant to the third part of this series,

which will be focusing on group differences in intelligence. Okay, so with that groundwork, the framework of what we're going to be talking about, let's make us start talking about the biology of intelligence. So although from researching cognitive psychology, we're pretty sure that there is this thing called intelligence, which is some sort of, refers to some sort of underlying cognitive phenomenon, whereby some people are better, on average, at thinking and reasoning and problem solving and abstract thinking than others, and that this is measured by consistent performance across a wide variety of cognitive tasks and tests. So we're pretty sure that that's a real thing. That's a

psychological phenomenon. However, the underlying biology behind why that is the case, that is why some people do better at these tests than others, why some people are more intelligent than others, that's still quite poorly understood. So obviously, if we think that, the brain gives rise to the mind, then differences in the mind, that is psychological differences in performance and intelligence tests, must in some way be the result of differences in the brain, differences in neural architecture or functions. But we don't really know how or what or exactly what's going on there. Now, one theory of intelligence is that greater intelligence is at least partly the result of increased speeds of neural conduction. This is the neural

efficiency hypothesis. We talked a little bit about this in the last episode. Direct evidence in favor of this hypothesis from electrical studies of the properties of neurons is sort of mixed some evidence for and some negative results. So it's not clear whether people who are more intelligent literally just have faster neurons, that is, they propagate the electrical activity faster than others, that is than people who are less intelligent, although that may be a contributing factor. It's not so clear yet. As I noted in the previous episode, something that is well established is that more intelligent people are able to perform the same reasoning task with a lower energy expenditure, that is, a lower

glucose metabolic rate in their brain compared to less intelligent people. So they are more efficient in that sense. It's not clear whether that translates to increased speeds of neural conduction. Now, another very interesting result from neuroscience research is that brain size is in fact moderately well correlated with intelligence. Within each gender, it should be said. That's a very important caveat there because men on average have larger brains than women, but they're not on average more intelligent. But within each gender, there's a positive correlation between brain size and intelligence. According to one manner analysis, I found that 27 out of 28

structural MRI studies, so that looks at the structure of the brain tissue, found a positive correlation between IQ and total brain volume with an estimated correlation of 0.33. So that means that roughly 10% of the variation in intelligence can be explained by variation in total brain volume. Now that's a significant result. You know, that's 10%'s a reasonable amount. It's not nothing. But that means 90% is not explained by variation in brain size. So we certainly can't say that being big-brained means that you're intelligent. There's obviously a lot more going on too. Another correlation that's been found is the amount of gray matter.

So gray matter refers essentially to mostly neural bodies, that is the main body of the neuron, as well as dendrites and certain other things as well. Mainly it's excluding the axons, that is, the large output, if you like, that the large wires connecting different parts of the brain together, loosely speaking, or different neurons and different parts of the brain together. So that's white matter. Gray matter is mostly the cell body itself. Anyway, the amount of gray matter is also positively associated with intelligence. So one way they look at this is cortical thickness. So the gray matter, the neural bodies are concentrated on the surface of the brain in humans. That's called the cerebral cortex. And inside

or underneath that, I suppose, you could say, consists mostly of white matter, which is tracts, axonal tracts and insulation of those tracks connecting the different parts of the brain together, as well as ventricles and other sort of support structures. But the thickness of that outer layer, the thickness of the cortex of the cell body consisting of cell bodies, is also positively correlated with intelligence, a result that's been corroborated across a number of studies. Now, we also know that these factors like cortical thickness and brain size are at least partly genetically influenced. That is not really a surprise,

because we also know, as we'll discuss a bit later, that intelligence is genetically influenced, is heritable to some degree. However, that doesn't mean that brain structure is fixed. Now, this is a point that I think I've made before in episodes, but we'll make sort of repeatedly in this in the next episode, the fact that something is some biological structure or function or psychological behavior or tendency or whatever, the fact that something like that is found to

be, is found to have some sort of neurological or biological or genetic correlate, it does not mean that that behavior or characteristic is fixed and immutable and somehow sort of determined by the biology and therefore not able to be influenced or manipulated. In the case of brain structure, we know that this is just not true. Numerous neuroimaging and structural imaging studies have found that brain structure changes when people learn new tasks. For example, when trained on juggling particular regions of the brain will increase in thickness, increasing cortical thickness or increase in relative size. Increased hippocampal volumes, the hippocampus is a part of the brain that's associated

with long-term memory formation. Increased hippocampal volumes have been found in experienced taxi drivers, and there was a correlation between the length of time that they'd been working as a taxi driver and the volume of the hippocampus, indicating a process of learning with experience. Grey matter increases have also been found associated with motor, auditory, and visual spatial brain regions

when comparing professional musicians with a matched group of amateurs and non-musicians. So all of these types of evidence, and I'm just sort of very briefly going through some of them here, but all of these types of evidence point to the fact that training, that is, performing particular tasks repeatedly, leads to structural changes in the brain. And of course, this is what we'd expect, right?

Because if we think that training and learning produces new memories, and if we think that memories are ultimately instantiated in neural tissue, you know, in neural structures, then learning must have structural biological correlates, right? It just follows from that. So the fact that we can observe them in some of these neuroimaging studies is interesting, but it's fundamentally not really surprising. The important point is just to realize that just because you see a difference in brain structure,

it doesn't mean that that's somehow determined outside of experience or that it's rigid in biology and therefore can't be changed. Some people, I think, have sort of an odd idea about this. Now, in terms of functional studies as to which regions of the brain are most responsible for intelligence, it's fairly clear at this stage that there's no single region or there's no single narrowly defined area of the brain that's singularly and individually responsible for intelligent behavior or for reasoning and problem solving

and stuff like that. It's clear that intelligence, these sorts of intelligent behaviors, rely on distributed activity across the cortex. It does seem to be particularly strong activity in the frontal lobe, which is not surprising because that's long been thought to be associated with higher executive

and cognitive sort of functions. but it's not, these sort of activities are not isolated to single regions. So when we perform complex mental functions like this, you know, planning and decision making and so on, we're using a wide range of different areas of the brain and exactly which parts are doing what is not very well understood at this point. Now, aside from the neuroscience research, there's also been a great deal of investigation into the genetics of intelligence,

particularly people are looking for genes associated with intelligence. Now, in the few decades that people have been looking at this in detail, quite a number of genes have been found that associated with defects of intelligence, that is, mental disability. However, very few have panned out in terms of genes associated with greater intelligence, that is, positively associated with higher intelligence. There have been, I think, several hundred gene candidate studies of intelligence-associated genes. These are genes that are identified that are thought to be associated with intelligence. But pretty much none of them have panned out in terms of being able to explain any significant fraction of variation in intelligence across people.

So it's a quote from a study that I found, no individual genetic variants are conclusively related to intelligence or its change with age in healthy individuals, end quote. So there's certainly no such thing as an intelligence gene or, or anything along those lines. And if you hear people talking about such a thing, they're not saying sensible things. Because it's not really surprising that this finding shouldn't really come as a great surprise

because phenotypic traits like intelligence are so complicated that they're obviously going to be determined by many, many different genes interacting with each other. And this seems to be the overall finding that's coming out. That no individual gene explains more than a tiny fraction of 1% of the total variation. in intelligence outcomes, and the genetic determination of intelligence must therefore be the results of the interaction of many different genes. Exactly how that works is at this point not very well understood. Gene-wide association studies are a form of research which, instead of focusing on a single gene, it tries to find correlations between single-point mutations or sometimes a few-point mutations

across the genome and correlate them with each other. So this has the potential to find both a much larger range of mutations that could be associated with intelligence, but also correlations, so patterns of mutations that might go together, so not just a single mutation, but a few different mutations that together might be associated with intelligence. Several of these large gene-wide association studies are currently ongoing. So this is a fairly recent technology, which relies on relatively cheap genetic sequencing, which we've only really had for, you know, 10 years or so. So that's still ongoing and hopefully within the coming years we'll be able to see some results from that.

But it does seem from the research we have so far that there just isn't going to be any sort of smoking gun evidence of single genes or particular mutations in healthy people, particular allelic variants that are strongly correlated with higher intelligence outcomes. It's going to be more complicated than that. Some of the effects, apart from simply allelic variants that can result in differential outcomes, differential intelligence outcomes, include copy number effects, so the different numbers of copies of a gene in the genome. That can have an effect on, for example, the rate of transcription of that gene into protein, which of course can have an influence on phenotype. Another possibility is that there are complex interactions in regulatory networks in multiple genes.

We know this is the case generally, but we don't know specifically if it's relevant to intelligence. So it could be that you don't just need one variant of a particular gene. That's what an allele is. It's just a variation of a gene. You might need several different alleles in different places on the genome, which together form a regulatory network, which are responsible for upregulating or downregulating other genes, which in turn have an effect on intelligence. So it could be very much more indirect and complicated than just a simple, this one gene produces,

leads to an increased intelligence. So there's still a lot to learn there, but it doesn't seem there are going to be any simple answers about genetics and intelligence, which, as I said, is perhaps not particularly surprising. So that concludes the discussion of the biology of intelligence, with a focus on neuroscience and genetics. As you can see, there's still a lot that isn't known.

Now I want to talk a little bit about some of the research on increasing intelligence. So you may have come across on the Internet or elsewhere some of these sort of brain training games that claim to be able to increase your intelligence. There are various other interventions and psychological interventions and so on that people have proposed as well for training intelligence. Some people think that playing classical music to infants helps increase their intelligence. Many of these types of things are, well, borderline pseudoscience, really, that is, they're not really substantiated to actually do anything,

certainly have in terms of any long-term effects. In terms of measuring intelligence that's actually measurable and sustained increase in performance in intelligence tests, one thing to bear in mind that I noted in the previous episode is that it's very easy to increase performance on individual particular tests, simply by practicing those tests. and this occurs whether it's children or older adults. If you practice the test, you'll get better at it.

That's really not very surprising. However, the more you practice a single particular test, the less predictive your performance on that test becomes for other tests. So remember in the previous episode I said that people who do well on one intelligence test tend to do well in other intelligence tests. That relationship, however, breaks down if you do a lot of practice of a particular test. your performance there will become less and less predictive of your performance on other tests. And that's really not surprising, right?

Because the whole point of these intelligence tests is to test how good people are at thinking and reasoning and problem solving in novel tasks. That is, things that they haven't done very much of before, I haven't really seen. If you're testing how well people do on things that they've practiced a lot, that's not really testing intelligence. That's testing memory or skill that's being developed. And remember that those are distinct. your ability to complete specific tasks or remember specific information is not the same as a general abstract ability to problem solve and reason, which is what we're really interested in intelligence. And specifically this intelligence in that sense applies to reasoning in novel environments, because other things, skills in known environments can be learned by just training and repetition. That's different to intelligence.

So the more you practice a particular test, the less you're even using intelligence to complete the test, the more you're just using skills. skill or knowledge that you've trained. And that's, you know, that's a good thing. Like, that's how we get better at things. It's by relying less on the generalized intelligence that we use for novel tasks and relying more on the skill and the muscle memory or just the repetit, the fact that we've done something so many times. We just know how to do it reflexively. That's a good thing. That's how we get better at things. But it means that as you practice a test, it becomes, your performance on that test becomes less useful as a measure of intelligence. So the question is not whether people can increase their performance on intelligence tests or on particular tasks of intelligence tests.

If you practice all of the different tasks that are on intelligence tests, you can get better at all of them, and then you'd score a higher IQ as measured on the tests. But that really wouldn't be very useful, because all that means is that you've learned how to gain the IQ tests, and everyone knows, well, psychologists know that you can do this. You can game any tests by practicing it. The point is, can you actually increase the underlying problem-solving and reasoning ability that the tests are supposed to measure and do measure quite well as long as you don't try and game them by just practicing the tasks? So can we actually increase intelligence, intelligent behavior and thinking as opposed to increasing test performance, which we know we can? And this is the thing that a lot of those, especially online puzzle game type things, this is where they fall down, because certainly you can get better at things like that by practicing them. there's no question. The question is whether that increases your intelligence or whether it just increases your ability to do those tests. And I'm not saying that they can't increase intelligence,

but a lot of them haven't been tested rigorously in this way. So it's important to understand this difference. Now, there have been a few studies of attempts to increase intelligence. Many of them tend to focus on older adults, because the idea is that we want to develop interventions to try and reduce the onset or severity of neurodegenerative disorders in the elderly, especially because the aging population is becoming an increasing issue. And some of these studies have found some degree of success. So one study, which I just have to mention, it's one of my favorite studies that I've ever found out about. And the reason is because what they actually did is that these researchers took, they say, an off-the-shelf real-time strategy video game. So basically they just

bought some computer games or a specific computer game. But particularly the reason I love this is because the game that chose is Rise of Nations, which is a historical strategy game. This is one of my favorite games ever. It's a really wonderful game. I strongly recommend it, by the way, if you're a strategy fan. But anyway, they just got basically a bunch of old people to play this game. A training took place over a time period of four to five weeks and consisting of 15 training sessions each lasting 1.5 hours. So, I mean, that's a pretty awesome study. I think if you can get into it, just basically playing video games, strategy games.

But anyways, what they did is they used, I think, Ravens Progressive Matrices to assess intelligence before during and after the intervention. So what they're seeing is not, can people get better at playing Rise of Nations by practicing Rise of Nations? Obviously people can. That's not the question.

The question is, can people's intelligence be improved by playing a strategy game like Rise of Nations, as measured by their performance on Ravens Progressive Matricies, which is not the task that they're practicing. See, they're practicing a different task and then being tested using the Ravens Progressive Matrices. This is how to avoid this problem of just getting better at the single task. And they actually did find significant results after the period of training,

indicating that the training group increasingly improved over time on intelligence compared to the control group. So this, I think, is just a really cool study, but it's only interesting. It indicates that practice with thinking and reasoning and abstract problem solving, which is a lot of what you have to do in playing a strategy game, especially a real-time strategy game, which involves a lot of quick decisions. So it seems that practicing, making those sorts of abstract decisions can have some influence on our ability to perform other similar types of tasks.

So other types of interventions have included generic cognitive tasks, like solving literature or math problems, done by elderly persons, or intervention in providing voice and keyboard lessons to children. And so a number of these types of studies have found positive results, that is, improved cognitive performance as measured by IQ or intelligence tests as a result of these type of interventions. So the most straightforward interpretation of these findings is simply that

if you practice abstract reasoning and problem solving and thinking and these types of behaviors, you get better at doing them. If we harken back to the three-level model of the G-factor, which I discussed in the previous episode, where we recalled it at the bottom level, there's G, which is the underlying overall intelligence. Then at the level above that, there are domain-specific types of intelligence, like crystallized intelligence, fluid intelligence,

and maybe mathematical and musical intelligence and so on. And then at the level above that, again, there are performance on or ability to perform specific tasks or specific tests. So when you practice a particular task, you're certainly going to get better at that specific task, and that's at the highest level of the hierarchy. The question is whether practicing a specific task will also help increase things lower down the hierarchy. That is, will practicing a specific form of thinking or reasoning increase your ability to form other quite different forms of thinking and reasoning?

Or will it only help with that specific form? The answer seems to be yes, it will. That is, there is generalization. Practicing a particular type of cognitive task will certainly improve that task, but also, at least for some types of cognitive tasks, I should say, it also will help improve your general reasoning and thinking abilities. This doesn't mean that practicing any type of cognitive task will have the generalizing ability.

It seems likely that only certain types of tasks will do this. That is, tasks that are fairly integrative, that don't just use one narrow skill, but require a broader range of skills. So playing a strategy game, I would argue, incorporates quite a wide range of intelligence-type skills, because you have to, there's a lot of working memory stuff you have to do there, you have to remember what's going on in the game,

you have to strategize at sort of low level, that is what's going on in this particular region of the map, versus at a higher overarching strategic level, what are my goals? You have to plan from goals down to how am I going to achieve that, and sub-goals and so on. You have to manage all sorts of different things, resources and armies and so on. So there's a lot to do there, a lot of different tasks are integrated.

It seems that practicing something like that, or likewise writing essays about poetry or solving complex math problems as other studies have done, it seems that things like that are a lot more integrative than very specific, say, just practicing Ravens Progressive Matrices, for example, just practicing that very specific test. So it seems reasonable that the more integrative the task you practice is, the more likely it is you are actually going to be training overall intelligence rather than just getting better at that specific task. So I think that's probably the best way of interpreting the results so far, though it's still early days, and we still have a lot to learn about exactly what can increase intelligence and how much it can be increased. But I think certainly the evidence is that practicing intelligent behaviors can improve those behaviors, just as pretty much is the case for pretty much anything else. So that shouldn't be too surprising.

But it's good news because it sort of shatters this myth that intelligence is just fixed and there's nothing you can do about it. No, you can improve intelligence by practicing intelligent behaviours, thinking, reasoning, problem solving, etc. Okay, so moving on from studies of increasing intelligence, let's talk a bit about the heritability of intelligence. So this relates to the, as I mentioned before, the nature-nurture debate as it's sometimes described. And at the outset, I'll say that it's well-established that IQ or intelligence is substantially heritable. There's a large genetic component to intelligence. There's no question about that. Intelligent children tend to have intelligent parents and vice versa. Now, one thing, one type of research that's often done,

especially with intelligence, but with other psychological variables as well, is to look at the heritability, not as an abstract concept, but as a specific number to try and measure the heritability of intelligence. So these are often figures that you will see quoted, sometimes as fractions or sometimes as proportions or as percentages, relating to the heritability of particular properties or abilities or things like that. In this case, obviously, of intelligence. So you might hear things like the heritability proportion of intelligence is 0.5, or 50% of intelligence is genetically determined or statements along these sorts of lines. And this number refers to the heritability of a particular trait. So in this section, I'm going to explain a bit

about what these numbers mean and what they don't mean and look at some of the misinterpretations of what these numbers indicate, what they tell us, what they don't tell us. So the basic result is that something like 50%, maybe 60% of the variation in intelligence derives from variation in genes. This is equivalent to saying that the heritability of intelligence is somewhere around 0.5.6. Some studies give a range of 0.4 to 0.8, so I'm picking sort of middle values around there. The idea is a bit more than half, so maybe 60% or something like that, maybe even two-thirds of variation in intelligence seems to be genetically based, based on the most studies that are done. So this is sort of what most studies find. The rest of it, that is the rest of the

variation in intelligence, derives from environmental factors like parental wealth and home atmosphere, education, environment, and elderly childhood, things like that, which we'll talk more about later. So the way that this heritability estimate is determined, there are a few different ways. One is to use adoption studies, so that is study children who have been adopted into households by parents who they're, or by people who they're not related to. You compare the outcomes of adopted children to similar children, that is children raised by similar parents, but who were their biological parents.

So in those cases, if you imagine, say in the same household, to take a simple example, although this is also done across households, if you have two children in the same household, raised in the same environment, one of those children is a child of those parents, the other is adopted. That means one of the children, the biological child, will share, on average, 50% of the genetic material with each of their parents, whereas the adopted child will, on average, share none. So, whereas their environments will be roughly the same, at least their shared environment will be the same. So by comparing the differences in intelligence, as those two children, you can see, obviously not just of two children, you need to do it over a

large number of children, but by comparing the intelligence of adopted children versus biological children, you can see how much of variation in intelligence is due or is the result of genetic factors as opposed to environmental factors. So that's so-called an adoption study. Another type of study is a twin study. Now, the usual methodology here is to compare the outcomes across families, a similar way, but instead of using adopted versus biological children, it compares diszygotic with monozygotic twins. Monozygotic twins originate from a single zygot, which means that they are genetically identical. So identical twins or monosylogic twins share 100% of their genetic material,

and therefore any differences in outcomes between them will be obviously the results of environment because they're genetically identical. The method of twin studies then is to compare the outcome of the intelligence outcomes of monosylogic twins to dyszygotic twins who are, who derive from separate zygots, that is, so that they are not genetically identical. Dizogotic twins are no more similar to each other than any, in any sort of randomly paired siblings are. So, monozygotic twins are much more closely related to each other than disigotic twins.

So if you compare the intelligence outcomes of monosygotic versus dislikotic twins, then you can get an estimate of how much of the variation in intelligence outcome is due to genetics, since you've got a way of controlling the relationship coefficients of the twins there. So that's a so-called twin study. The final type of study, or the main third type of study, in these behavioral genetic studies, as they're so-called, because they look at behaviors and the genetic determinants thereof. So the third main type of behavioral genetic study is identical twins raised apart. And in some sense, this is a gold standard study because it takes identical twins, who are, as I said, genetically identical, and then looks at rare cases where identical twins have been raised in separate households or in separate environments.

In that case, you've got an unusual circumstance where genetics is the same in two cases, but the environment is different. So you can see by comparing the variation in outcomes of identical twins raised apart, and comparing that to, say, the general population, you can see how much of the variation in outcomes is due to genetics. So, for example, if I have identical twins living in Environment 1 and Environment 2, and if I compare their outcomes to just ordinary people of the same age and other factors, obviously, living in those same environments 1 and Environment 2, if the identical twins are, say, 10% different,

whereas the in intelligence outcomes, let's say one is 10% smarter than the other, whereas the ordinary people, the non-identical twins in those same two different environments, one is say 50% more intelligent than the other, then sort of crudely we can say that, well, 10% is due to the environment, but 40% is actually due to genetics, because when you control for the genetic difference, only 10% of the difference remains, that is, in the identical twins. Now, again, obviously, you don't just do that with two pairs of people,

you do it with large groups of people. But that's the basic idea. Using monosyogotic twins raised a pie to control for genetic variation and just look at environmental variation. So these three different types of studies, adoption studies, twin studies, and identical twins raised apart studies,

all allow us to get an estimate as to how much of the variation in outcomes here, intelligence, is determined by genetics as opposed to environment. I should say that, so what they try and do is describe the degree of the variation in outcome, the degree of phenotypic variation, so the phenotype in this case is intelligence, that is explained by particular causes. And the three causes

that are usually looked at are genetic variation, variation in shared environment, and variation in non-shared environment. Shared environment means things like children in the same household share in common, so like the parental, the household, for example, socio-economic status of the parents, the parental education and the amount that parents speak to them, things like that, things that are going to be roughly the same. Parenting style will be another one, personality of the parents. Non-shared environment are just everything else, idiosyncratic things, that even siblings living in the same household experience differently.

So, you know, different friends that they meet in the school playground or, you know, watching different shows or whatever. So the important thing here is just to remember that the three main variables that account for variance in these behavioral genetic studies. Genetic difference, shared environment and unshared environment or non-shared environment. So this allows us to

understand what the results of behavioral genetic studies mean. That is what these heritability proportions of 0.5 or 0.6, what they mean. And what they mean is that say if the heritability is 0.6, then 60% of the

variation in intelligence can be accounted for or explained by variation in genetics of the people that were sampled. Now, it's very important to understand what that means that we're only accounting for the variation in intelligence across people. We're not accounting for the total, in any sense, the total level or a total achievement of intelligence. That is, the 60% figure doesn't mean that 60% of intelligence is determined by genetics.

This is a common misinterpretation of what these heredibility. proportions mean, because we're not trying to explain the total achievement of intelligence. What we're trying to explain only is the variation in intelligence. To use an analogy that might help to understand the difference here, imagine we had a race where we had a bunch of athletes, and they're all pretty similar ability, and imagine that then we gave each athlete a slightly different variety of shoe, or running shoe to use. Some of these running shoes are a bit better than others, some were a bit worse, So imagine then we let them race, and we measure each of their time, the running time of each athlete.

And then we correlate that against the quality of the shoes that we gave them. And perhaps we say that we find that 90% of the variation in running time can be explained by variation in shoe quality. And what would that mean? It would not mean that 90% of the outcome as to how fast they ran was determined by the shoe. So it wouldn't mean that if you took away the shoe, then they would run, you know, they would run almost 10 times slower or something like that. Nor does it mean that if any old person were to put on the shoes that the athletes were wearing, that they could run almost, they could run like 90% as fast as the athletes.

It doesn't mean that because it's not a measurement of the total performance or the total contribution to the overall speed of the athletes. All it says is that the variation, that 90% of the variation in the times of the athletes is explained by variation in shoe quality. Remember that the athletes might only vary in their race times by maybe a few fractions of a second. So the total amount of variation that we have to explain might not be very much.

So in fact, the shoes might only make a tiny bit of a difference, a small fraction of a second between the fastest and the slowest runner. So the 90% figure here is very misleading. The shoes may actually make almost no difference at all. It's just that the tiny difference that they do make is very important relative to the already source. small difference between the runners. So hopefully that analogy helps a little bit. Heritability proportions don't tell you anything about how important genetics are sort of overall in determining

intelligence phenotype. All it does is tell you the proportion of the variation accounted for by genes as opposed to the environment. Now there are a number of corollaries of this fact, one of which is the very counterintuitive result that if we made the environment the same for everyone, that is if every person was raised in exactly the same environment, including shared and non-shared environment, the heritability proportion would be 100%, or 1, because 100% of the variation in outcomes of intelligence would be genetically determined, obviously, because we've just said that there aren't any environmental differences. So if you got rid of all the environmental differences, then all of the difference that remained would have to be

genetic, so the heritability proportion would be 1 or 100%. Now, a further corollary of this, is, and I think this is the most important thing to bear in mind about heritability proportions is that there just really isn't such a thing as the heritability proportion. There's no single number that you can meaningfully give here because the number that you get always depends on what population you look at and how much variation there is over that population. Because remember, the heritability proportion measures the proportion of the variation within that population that can be explained by genetics. How much of the variation can be explained by genetics? How much of the variation can be explained by genetics depends upon how much variation there is in the first place

in the population that you're measuring. So imagine that we did some of these behavioral genetic studies and that we mostly recruited for the studies, sort of middle or upper-middle-class persons, you know, relatively stable families, relatively high income or, you know, at least comfortable incomes. And then imagine that we get heritability proportions of, say, you know, 0.7 or 0.8 or something like that. Does that tell us that the overall heritability proportion of intelligences 0.7.8, that, you know, environment doesn't matter very much and it's mostly genetics? No, it doesn't tell us that. Why? Because there isn't much variation of environments, especially shared environment, across the population that we've sampled. We've only sampled relatively well-off people. So, you know, there's going to be some environmental difference, but not very much within this group. So obviously most of the difference is going to come from genetic difference.

since we've already gotten rid of most of the environmental difference. Now, you might ask, well, why would we sample only over middle or upper-middle-class income-type people? Well, one of the reasons is that a lot of these studies, it's just a lot easier to initially recruit and also to retain people from middle- or upper-middle-class families. It's much harder to recruit from people from much poorer families. I mean, you think about trying to recruit from, say, inner-city areas where there's a lot of gang violence and other... other problems, especially in places like America, or from, say, migrant communities who may not

speak English and things like that. So there are lots of reasons why it's practically more difficult to get people in these studies, especially if you're talking about things that are quite intrusive, like adoption studies and twin studies, require a lot of interaction with the researchers. It's much more difficult to get, to recruit people for these sorts of studies from lower economic status. So the result is that, in fact, a lot of these studies are only sampling from a relatively small range of environments, and therefore it's not surprising that they're going to find high heritability proportions. In order to find, in some sense, the true heritability proportion for the whole population, you need to conduct these studies with a representative sample of the entire

population. But of course, we don't have that. And in fact, there's direct evidence for this because some studies that do focus on parts of the wealth spectrum, there are some, once the focused on the low end of the wealth spectrum, only find about a 0.1 heritability proportion. That is only 10% of intelligence outcome is mediated by genetics. Whereas at the higher end, it was about 0.7. So that is, the more income, or the better off the family was, the less heritable intelligence was found to be. And that's not going to be as surprising, because the better off the families that you're

looking at, the better resource the environment is going to be. So this is consistent with a hypothesis that children from relatively impoverished or deprived environments don't reach their full genetic potential, whereas children from relatively well-off environments do reach their full genetic potential. So if you take a sample from relatively well-off environments, most of the difference in outcomes is going to be from people reaching their genetic potential. So it's going to be mostly from, most of the variation is going to be accounted for by genetic variation.

On the other hand, if you take people from low socioeconomic backgrounds, there'll be a much wider range of resources there because of idiosyncratic factors and some people just happening to have a better school than others. There's going to be less consistency owing to, across families, owing to lower overall resources. And therefore, you're going to have some people who are through luck or hard work or whatever are able to meet their genetic potential for intelligence, but many more who aren't going to as a result of their circumstances. And therefore, the overall amount of variation, particularly the environmental variation is going to be greater. and therefore the measured heritability proportion is going to be a lot smaller.

So the results that we get from these sort of behavioral genetic studies really have to be interpreted carefully because the heritability proportion very much depends upon who you're sampling, what the population is and how much environmental variation there is. If there's not much environmental variation, then you're just not going to find, then necessarily you're going to find a high heredability proportion. An analogy to this is that if you look at height,

something like height, that's a bit simpler, we know that height is almost completely genetically determined in environments with good nutrition and healthcare. In those environments, when children are able to reach their full height potential, because healthcare and nutrition and other such factors are sufficient. In environments where that's not the case, however, very large environmental effects in height can be manifest, because some people have access to enough food, some people don't, some people have better medical care, some don't.

In those cases, there will be a much larger environmental variation in height outcomes, and therefore a much lower measured genetic contribution. So it's similar to intelligence. People from a relatively more impoverished background have a larger range of environmental variation, and therefore a smaller genetic contribution compared to people from relatively prosperous environments. There are other sort of similar effects that are relevant also to identical twin and adopt. because most adopted children are adopted into relatively well-off families. So the environmental variation in adoptive households across households is in general,

not going to be very large. Therefore, most of the variation that you're going to find as a result of looking at adopted children or identical twins raised in separate households is for the most part, in most cases, going to be pretty small. The environmental variation is pretty small, therefore the genetic variation is going to be big. So again, there's There's a built-in bias in these sorts of studies to overestimate the genetic contribution because of underestimating the environmental variation by only, or at least disproportionately, sampling relatively prosperous households, relatively well-off, well-resourced households. There's yet another problem with this type of research that it doesn't even really make

sense to try and separate genetic and environmental effects in this way, and this is because genetics interacts with environment. Maybe as a very young child, I just sort of take my environment is given and respond to that reflexively. But certainly as we age, that's not the case. I respond to my environment, and in doing so, I shape my environment. I'm active

in the process of shaping my environment, and the way I shape my environment is going to be partly determined by my previous environmental exposure, but also by my genetics. That is, if I start off as being a bit more intelligent because of my genetics, then I'm more likely to put myself into, or develop an environment around me

that's conducive to further intellectual development. Whereas again, on the opposite side, if I'm less intelligent to begin with, for whatever reason, then I'm more likely to put myself in an environment and develop an environment around me, you know, associate with people and involve myself in certain activities, etc., which is less conducive to intellectual flourishing and therefore likely to reduce my intellectual achievements, my intelligence scores. So there's a dynamic interaction here. And one of the studies that I read explains this quite well, so I'll, I'll quote from one of these studies. If both twins, so this is talking in the context of a twin

studies, if both twins are taller and quicker than average, although raised in different cities, they are more likely to play basketball a lot, get selected for their school team, and get professional coaching. There's a dynamic interplay each step of the way, whereby better than average ability accesses an enriched environment, which also enhances the ability advantage, which accesses an even more enriched environment, and so forth. Similarly, if twins have slightly better brains than average, although raised apart, they are both more likely to enjoy school, getting into an honor stream, go to a top university, and so forth. Dickens and Flynn call this process the individual multiplier. Thanks to the individual multiplier, the fact that

separated twins have powerful environmental factors in common is masked, and their common genes get all of the credit for their highly similar IQs, end quote. So the basic idea is that some proportion of the commonality of, say, for example, the commonality in intelligence outcomes of identical twins is actually because they share environments that are more similar than you would think. But the reason they share environments that are more similar than you would think is because of their similar genetics. That is, their similar genetics leads them to experience and develop and put themselves into more similar environments, which in turn leads them to have more similar intelligence outcomes. So there's a dynamic interplay between the genetics and

the environment. And you can't just neatly separate them. in the way that people often think. So what can we take from these studies overall? I think overall we can say that, first of all, there is a significant genetic component to intelligence. There's no question about that. The question as to how much is much harder to answer because it depends on what environment you're talking about.

It seems that in well-resourced environments, where children are able to reach their genetic potential, most of the variation in individual outcomes comes from genetics. though still with a significant proportion coming from environment, even in those situations. However, in environments that are resource poor, relatively deprived, most of the variation seems to come from differences in that environment. That is, some people who have a bit better access to the resources they need to flourish than others. So that variation within the environment seems to be doing a lot more of the work in the lower ends than in the higher end case.

Another finding from this research also that I hadn't mentioned before is that the heritability of intelligence seems to decrease with age. That is, it seems that young children are more strongly influenced by environment than are older children. Not exactly clear why that should be the case, but that does seem to have been a finding of the research. Okay, so final thing that I want to talk about in this episode is called the Flynn Effect, which you may have heard of. I've mentioned it a couple of times in this series already. the Flynn effect refers to an empirical finding, which is that basically intelligence scores or IQ is increasing over time. In fact, it has been for many decades. It's named after Flynn, who was a researcher who first discovered this, using data that spanned from 1932 to 1978 across a wide range of different tests.

So this is not just a single test. This is a wide range of intelligence tests. And this original result by Flynn has now been replicated in many other studies in many different countries. most of the research tends to look at Britain and the US as well of some European countries. And the exact results do differ between countries, but the broad result that there have been increases in IQ over time over decades has held up. So across one meta-analysis that I looked at that examined 285 different studies in Britain and the US, with a total of about 14,000 different subjects, over a period of 1951 through 2010, so that's a 60-year period,

and of course incorporating a wide range of different intelligence tests. The overall average effect size was about two to three IQ points a decade. And that seems to be the general finding. It's something like two to three points a decade. Now, that might not sound like very much, given that we measure IQ out of 100. Well, it's not out of 100, but I mean the mean the mean is 100. However, that does mean that over a 60-year period, if you have an average of three points increase in a decade, that's 18 points.

That's more than a whole standard deviation. Now, it's a little bit controversialist to sort of how long the Flynn effect has been going on. It seems to have been going on at least since the 1930s, probably for longer than that, because it was probably going on before we were doing these sorts of measurements. In terms of whether it's still ongoing, there seems to be a bit of mixed evidence about this. It seems to have slowed or ended in some Scandinavian countries, possibly some other European countries as well. But it does seem to, the Flynn effect still seems to be going strong in the US.

It's also been found in many developing countries. we'll talk a bit more about group differences in intelligence in the next episode, but broadly speaking, people in developed countries score significantly higher, you know, like 20 points or so, 10 to 20 points higher than people in developing countries. This shouldn't be too surprising, given the improved environment, nutrition, and education, so on in developed countries. So the fact that we're seeing a Flynn effect both in developed and developing countries is quite interesting. That they're sort of at different levels, but they're both increasing.

So the existence of a Flynn effect, even if its exact distribution of the countries, is not completely agreed on. The existence is fairly well established. In terms of explaining why we see the Flynn effect, well, that's much less clear. There are a wide range of different proposals that have been put forward as to why we see an increase intelligence over time. A few things that stand out in terms of this literature. First of all, it's clearly not genetic. It can't possibly be genetic. The rate of increase is far too rapid to be genetic.

If you extrapolate a rate of about three IQ points a decade back 100 years, this indicates, roughly speaking, that about 100 years ago, half the population would be mentally disabled by current standards, because one common cut-off point for intellectual disability is an IQ of 70, that's two standard deviations below the mean. There's just no way at all that that rate of change could possibly be due to genetic changes. changes. Genetic can't possibly change in large populations at that rate. So it can't be genetic effect. It's got to be some sort of environmental effect. The question is what environmental effect? And we don't really know what environmental effect. Nutrition, improved nutrition has been suggested as a factor. From what I've seen, that seems unlikely, at least since roughly 19,

roughly the end of World War II, maybe 1950s, maybe in the early decades, particularly maybe after the Great Depression, that could have had some effect, and in developing countries today, that might be explaining some of the Flynn effect, but I think overall, particularly in recent decades, probably there hasn't been much of an improvement in nutrition, and that's probably not driving it. Improvements in healthcare, neonatal, or other forms of child healthcare, may be contributing some of it. Again, it's not clear that that's all of it. One common proposal is that changes in education with increased in recent decades, with increased emphasis on problem solving and abstract thinking over wrote learning has played a role.

That's possible, we don't really know. Another hypothesis, which I tend to think is the most likely, though this is somewhat speculative, is that there have been, over the past decades, increasing demands, both in entertainment and employment, in terms of cognitive activity. So in the last few decades, Western countries have become more service-oriented. There are more people working in so-called knowledge, technologies, knowledge industries, working with computers, working with numbers, working with communication, complex things that require thinking and abstract problem-solving. Many fewer people working in manufacturing jobs or agricultural jobs, which are more procedural, which doesn't mean they don't require skills, but it means they don't require the same level of abstract problem-solving and reasoning

that many modern service-based occupations do. So essentially people are practicing this stuff more is the idea, and therefore getting better at it. Similarly, it's thought that entertainment has become more cognitively demanding. I think video games are a good example of this. Video games, at least sometimes video games, require quite a lot of thinking and planning and strategizing and so on, and kids are playing more and more of this than they were obviously in the 70s or something.

So possibly this has also contributed to, a rise in intelligence. Other people have also pointed to just the increasing saturation of society, Western society at least, with advertising and print literature and TV and video and more recently the internet. There's just more and more of this all over the place and people are having to deal with the exposure to all sorts of information, sort of reading more and just attending to more of this sort of stuff on a daily basis, which is thought perhaps to be contributing to just practicing these sorts of filtering and thinking and cognitive tasks, even if we don't necessarily think of that sort of thing as work or academic, but nevertheless,

it may be contributing to increased performance and intelligence tests. So we don't really know, but it seems like some combination of these factors of health care, different education standards, increased complexity of work and entertainment demands. Also, another hypothesis is reduced environmental. toxins, certainly things like lead and certain other toxins have reduced concentrations in the environment in Western countries in the last few decades, so that might also be contributing. So likely it's a combination of factors, because the Flynn effect is an empirical finding. It doesn't say what the cause of it is. So there's no reason to think that there's only one

cause. It's likely that there's a wide range of causes contributing to this. And it might also explain why we see different rates of Flynn effects in different countries, because some of these factors are going to be more relevant to some countries than others. One interesting application, well, I guess disturbing application of the Flynn effect is that that's become relevant in recent years is the need to use up-to-date IQ tests, particularly when testing people relative to, say, Social Security benefits or particularly criminals. This is especially relevant in the US where the death penalty is used because in many states an IQ test is one of the factors used to,

determine whether someone is eligible for the death penalty. And if you use an old IQ test that was normed, that is, the mean was established decades ago, that's going to be out of date compared to present standards, and therefore it will overestimate someone's intelligence. Because there's been an overall increase in intelligence over time, if you compare your performance today to someone to tests that were developed a few decades ago, they're going to overestimate your intelligence relative to people today, because essentially the baseline is higher than it was a couple of decades back. So someone who scores 74 on an IQ test, for example, might actually score, say, 68 if they used an up-to-date IQ test. So that can be the difference between eligible for the death penalty and not

eligible for the death penalty, or eligible for special assistance and not eligible for special assistance in education or something along those lines. So this is becoming an increasingly important issue that the Flynn effector needs to be factored in to these sort of use of the intelligence intelligence tests. It's also something that people should bear in mind if you ever hear reports of people's IQ. Sometimes this is something that you hear talked about, or people will tell you what their IQ is. One thing is that different tests have slightly different measurements, so you should make sure what test they're using. But the other thing is how old or how recently was that test normed? That is how, you know, the mean set at 100 essentially.

If you used a test that was normed decades ago, then that's going to overestimate a person's intelligence. So that's all I wanted to cover in this episode. In the third part and final part to this series, I'll talk a bit about a group differences in intelligence, which will build upon some of the research that I've discussed and the methods discussed today relating to genetics and heritability of intelligence. Particularly there, we're going to focus on some of the controversy surrounding race differences in intelligence between blacks and whites in the US, which is where most of the research has focused on.

So if you enjoyed the episode or the podcast generally, feel free to leave a favorable review on iTunes or other aggregator of your choice. I always appreciate that. You can also send me an email at Fods12 at gmail.com, F-O-D-S-1-2 at gmail.com, where you can ask a question, leave some feedback, just let me know that you like the show.

I always enjoy hearing from listeners. Thanks again for listening, and I'll talk to you next time.