The Science of Everything Podcast - Episode 80: Intelligence Part 1

Episode Date: February 12, 2017

The first in a three part series on human intelligence, covering debates about the definition of intelligence, how intelligence is measured, factor analysis and the origin of the g factor, and interpretations of the g factor as corresponding to intelligence. Also includes an analysis of Gardner's theory of multiple intelligences, and the cross-cultural validity of intelligence tests. Recommended pre-listening is Episode 79: Basic Concepts in Statistics. If you enjoyed the podcast please consider supporting the show by making a paypal donation or becoming a patreon supporter. https://www.patreon.com/jamesfodor https://www.paypal.me/ScienceofEverything

you're listening to The Science of Everything podcast, episode 80, Intelligence. I'm your host, James Fodor. So in this episode, or at least two episodes, possibly three, we'll see how we go. But in this series of episodes, we'll be looking at the subject of human intelligence, covering topics including how intelligence is measured, how it's defined and measured, although, as you'll see, measurement is actually in some sense easier than defining intelligence. debates about measuring intelligence, so the concept, debates about IQ and the G factor, which we'll talk about some of the biological research of intelligence, including genetics

and neuroscience research, a little bit of evidence regarding increasing intelligence, so interventions that have been developed to attempt to increase intelligence. We'll also discuss some of the literature on the heritability of intelligence, including twin studies, adoption studies and things like that. I'll talk about the Flynn effect, which is, a very interesting phenomenon that's been found relating to the increase of IQ over time. And finally, we'll conclude by looking at some of the research on group differences in intelligence, particularly with the focus on race differences.

So, there's quite a lot of material to get through. So let's make a start. So first of all, what do we mean by intelligence? Now, I should say that in these series of episodes, we're going to look at only human intelligence, so we're not going to talk about animal intelligence, and we're not going to talk about artificial intelligence. those will be different episodes, different topics. Obviously, there's commonalities there, but human intelligence is quite enough at the moment.

And basically, we're focusing on intelligence as it's studied in psychology and maybe also to some degree in sociology when it comes to some of the group differences stuff. So in that context, what do we mean by intelligence? First of all, it's important to understand that there is no generally accepted psychological definition of intelligence. They just isn't. There are many different definitions that psychological. will give and quite a lot of disagreement about sort of which is best. However, that itself doesn't invalidate the concept. And this is the second important thing to understand because

some people say that, well, you know, a psychologist can't agree on a definition of intelligence, therefore it's kind of a meaningless concept. Well, that doesn't follow because lots of concepts that we use, both in science and in everyday life, can't be given a precise, clear, single definition. Try defining a chair, for example, in those terms, or a game. You can't really do it. There's no single easy, descriptive criteria that you can give for those sorts of categories. And one of the reasons is because they're just not clear, precise things. They're sort of vague and fuzzy, but nevertheless meaningful. An idea from philosophy that you can use to describe is that intelligence is a concept describing family resemblances. That is a set of, it describes a set of things that are similar in important ways, but also kind of different.

and therefore you can't give a single necessary and sufficient condition or set of necessary and sufficient conditions to define what something, what counts is intelligent or what counts as unintelligent, but rather there are just collections of properties and phenomena that are more or less indicative of intelligence or representative of intelligence. So, that being said, one good definition that's sometimes quoted, and I think it's pretty useful, comes from an article called Mainstream Science on Intelligence, which was published in the New York Times back in the 90s, and was signed by a number of prominent psychologists at the time. So it's about as close as you can find to sort of representing the mainstream opinion of psychologists regarding intelligence. The definition given there was as follows. Quote, Intelligence is a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas,

learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings, catching on, making sense of things, or figuring out what to do. End quote.

So I think that encompasses the idea, well, the key properties and concepts that we think about when we use the word intelligence and really gives the sense of this family resemblance, because there's a lot of related things here that are different but sort of similar. And no one of them by itself defines intelligence. They're all related.

So in particular, intelligence relates to the ability to reason, to think well, to plan, to learn quickly, and competently and appropriately from experience, to comprehend, understand things, especially complicated things, to think in not just concrete, repetitive ways, but abstract ways and new ways as well. Some people think intelligence is essentially related to creativity.

And in particular, intelligence in psychology is not understood as knowledge. That is, in everyday language, we tend to say that someone is intelligent if they know lots of stuff, like lots of facts. And to some degree, that's relevant to the psychological definition as well, because some tests of intelligence do measure, for example, vocabulary knowledge. And ability to learn is part of intelligence. And, of course, if you have a greater ability to learn, then you're going to learn more things, more facts.

So they're certainly related. But the mere fact that you know lots of stuff does not mean that you're intelligent in the psychological sense. It just means you know a lot of stuff. It doesn't necessarily mean that you're good at thinking or understanding a lot of that stuff. And I think this is sort of consistent with our experience that there are people we know who have a lot of facts in their head but not necessarily are not necessarily very good at reasoning or solving problems or thinking abstractly about that knowledge.

The other aspect I want to highlight here is that intelligence is cognitive phenomenon. It involves thinking, reasoning, planning and deciding. It's not really about how good you are at achievement. things in the world as such, although, of course, the ability to reason and plan and learn is very helpful for doing things in the real world, but it's not sufficient because it doesn't say anything about whether you're, for example, good with people, social skills, or whether you have the ability to do things with your hands, whether you're good kinesthetically, things like that. So that's what we mean by saying intelligence is cognitive. It's primarily

about thinking, reasoning, in the mind, mental stuff. That doesn't, mean necessarily academic stuff. That's what is meant by its intelligence is not nearly book, not merely book learning. It's the ability to think well. Now, some theories of intelligence in psychology have emphasized the importance of working memory capacity and the speed of mental processing. I've talked a bit about working memory in some previous episodes, I think. Others, however, think that these distinct abilities. It's a bit hard to say because it's pretty hard to test for some, it's pretty hard to do any intelligence test that does not in some way involve working memory capacity and speed of processing. So it's very hard to teach those apart.

But needless to say, it's not clear whether intelligence is sort of the same thing as memory, that is being able to remember lots of things or having a good memory. This is how, again, intelligence differs from just knowing lots of facts. So people might have a very good memory. Some autistic savants are like this, for example, they can just memorize huge amounts of stuff. but it doesn't necessarily make them intelligent in the sense of the ability to solve problems, reason, think abstractly, comprehend complex ideas, etc., because merely memorizing something's not the same as understanding it or using it to reason. So there may be a connection between working memory and speed of processing and intelligence,

but they're not the same thing, and it's not clear that there's a necessary connection there, so that's a bit disputed. Okay, so that concludes our initial remarks on defining intelligence, and so that gives you a sense of what we're talking about here. So to recap, it's a set of cognitive abilities, a set of related and interconnecting cognitive abilities, that is relating to thinking and reasoning and planning. It's not the same thing as having a good memory.

It's not the same thing as memorizing lots of facts, and it's not the same thing as being able to do stuff in the world, like fix things or get along with people. That's what it means to say that it's a cognitive capacity. Okay, so now let's move on to talk about measuring intelligence, and this funny thing called the G factor, which you'll come across if you've ever read anything about intelligence or seen sort of psychological reports on this. Now, everyone's heard of IQ. IQ tests are probably the most common way of measuring intelligence, at least sort of in terms of the popular understanding.

There isn't really a sort of a clear definition of what an IQ test is, and psychologists aren't typically is interested in whether something is, from the perspective of measuring intelligence, whether something is sort of an IQ test or whether it's just similar to an IQ test. Really, there are lots of different tests that will measure different mental abilities that are more or less correlated with each other, that is, sort of, more or less measure the same underlying things. So IQ tests are just tests that are designed to measure intelligence, and generally they'll incorporate a number of sub-tests that look at different types of mental abilities. So typically they assess visiospacial, deductive, semantic, and, and

symbolic reasoning ability. So that's the ability, so for example, IQ tests will include solving some basic mathematical problems like fractional reasoning or multiplying, adding numbers, things like that. They'll often involve some memory tasks, so how many, you know, remembering a number and then repeating it after a short period of time or repeating it backwards. Visor spatial tasks, so one task that appears in IQ test is rotating mental rotation, so they'll show you a picture of some shape and then ask you which of these is a rotated version of the same shape. So you have to mentally rotate the object. Or you have to recognize differences between different images or other details of scenes,

things that are possible or impossible or whatever. A deductive reasoning, so they might give you puzzles of various sorts or problems that you have to figure out. Semantic reasoning, so some tests include, as I said, verbal components, so like linguistic knowledge. That one tests a slightly different. aspect of intelligence because it does relate to sort of how good you are at learning words, but nevertheless, learning is related to intelligence. Abstract problem solving is another important

part of intelligence test. So there's a particular test called Ravens Progressive Matrices, which you might come across at some point there. Each is basically a matrix of, I think it's four little boxes essentially with different shapes or markings inside of them. And essentially, you'll be given a few of these and then be asked to complete the sequence. So what comes next in the sequence of these matrices? And basically they come in a pattern, an abstract pattern. Maybe they're rotated in some way or some symbols are sort of in some sense increasing or changing in a regular pattern basically that you have to figure out and they get progressively harder. And Raven's progressive matrices are seen as a very good, largely culturally independent, not completely, which we'll get to, but largely culturally independent. That is it doesn't require any knowing any particularly. language, for example, way of assessing sort of abstract reasoning abilities. So these are the sorts of things that you'll find on IQ tests. There's quite a variation depending on exactly what the test is. Okay, so that's the basic idea of how intelligence is measured, that these different tests

are given. Tests vary widely depending on who you're giving them to, whether they're young children, older children, whether they're adults, whether they're mentally disabled or not, and also what culture they're from, whether they're from a literate or a pre-literate culture. IQ tests can be given to all of these populations, but they do need to be adjusted. So obviously, if people are from a pre-literate society, they can't be given a written test. So exactly what tests are given depend upon which group the test is designed for. And there are many, many, as I said, different IQ tests or related intelligence tests.

Now, at this point, you might be wondering, well, if there are so many different tests that can be given and so many different ways of measuring intelligence, then how can we meaningfully speak of a single thing, like a one thing called intelligence. Wouldn't it just be intelligence as measured by such and such a test or such and such a type of test? How can we speak of a single concept? And certainly that criticism has been raised in psychology, and we'll talk a bit more about that later. But here's perhaps one of the single most important findings, really in all of psychology, actually, I would say in my opinion, that relates to intelligence testing or measurements of intelligence, which have now been happening for, I think, around a hundred,

hundred years has been found overwhelmingly again and again in literally hundreds of studies, hundreds of different tests applied on different populations. This key fact, this key result that's been found in all of these studies is that people who do well on certain intelligence tests tend to do well on other intelligence tests. In other words, scores on intelligence tests are positively correlated with each other across a wide range of different tests and a wide range of different domains. Now, this very robust finding is usually taken to be evidence of the fact that there is some single underlying thing, or cognitive trait, if you like, that we can meaningfully refer to as intelligence. Now, just to explain how that reasoning works, if you imagine if you gave people,

say, a visuos spatial test and you gave other people, sorry, you gave the same people a vocabulary every test, and you gave other people a test of working memory and all sorts of other, you know, all sorts of other, you know, Ravens Progressive Matrices and all these other different types of tests that are given. And if what you found is that people who did well on one did okay on another and pretty badly on a third, and that there was no real relationship between how you did on one test and how you did it on another test, then we would conclude that there's no real single sort of intelligence, there's no underlying commonality or tendency here. Really, all there is is just different domain-specific competency, so some people are good

at some things and some people are bad at other things. Some people are good at the matrices, some people are good at vocabulary, some people are good at working memory, some people are good at certain types of problem solving, etc. But no underlying single intelligence. However, that is not what's been observed. And that's the key point. We don't observe that. We don't observe that there's just sort of no relationship to how people do in one task compared to another. Rather, with all of these cognitive tests, what we find is that there are, there's a positive correlation. If you do well in one test, it's likely that you would do well in another one of these tests and in another one of these tests, and really it's likely you'll do well in all of the

tests. Broadly, it seems from my reading that about half of the variance within a broad category of cognitive tests can be attributed to one single underlying factor. A factor is usually interpreted as intelligence. But let me explain a little bit as to how this is determined. The finding that a large proportion may be about half or more of the variance in the outcomes of these tests, these different intelligence tests, can be attributed to a single factor. That result comes from a methodology called factor analysis.

Now, basically, the way this works is that this factor analysis takes a bunch of data and looks at the variance in that data, so variability in terms of outcomes. Say it could be variability in people's heights and weights, for example. In this case, it's variability across different tests. So why do people score differently in one test compared to a third test, compared to a fourth test, etc, et cetera, et cetera. So the variations in people's scores across tests. Factor analysis looks at these variations and asks, is there any underlying commonalities, underlying causal factors that we could use to explain these variances, essentially? So why do we see this pattern of variations

that we do? And in particular, in the case of intelligence tests, the way it works is that you get a bunch of tests scores across the same individuals, but a bunch of different tests, right? and then work out the correlations in scores between the different tests. So remember correlation measures how the extent to which two variables go together. See the last episode, Episode 79, in the introduction to statistics, if that's a bit hazy for you. But if two variables are positively correlated, it means one goes up, the other is likely to go up as well.

Stronger the correlation, the more they move together. If the scores and two intelligence tests are positively correlated, it means people who score highly on one test tend to score highly on the other tests as well. Now, if you then imagine calculating those correlations for a bunch of different tests, not just two, but like dozens of tests or even hundreds, and then you put all that information in a matrix, just basically write it down in rows and columns, essentially. And then what is done is one conducts a factor analysis on that matrix.

Essentially, you ask, what are the common underlying patterns that can explain this pattern of correlations? So to get a sense of what's happening there, just think about it this way. if test one is correlated with test two and test two is correlated with test three and test three is correlated with test four then it's likely that test four is also correlated with test one and test two is correlated with test one and also test four and test three is correlated with test one and test two and test three and test four that is they're all intercorrelated with each other so you can look at all the different combinations of correlations say hey they're kind of each is correlated with all the other ones now how might that be Well, factor analysis attempts to answer that question, and the answer that it typically gives in the case of intelligence tests, is that one very simple way of explaining this pattern of correlations is that all of these tests actually are measuring, at least to some degree, the same underlying thing.

If all the tests are measuring the same underlying thing, then you'd expect them to give correlated results. imagine if I employed four different people to measure my height, or actually the height of a bunch of different people. So four different measurers using different instruments, say one is using a ruler and one is using a tape measure and one is using, I don't know, something else. They're all measuring the height of a bunch of people. Now, you wouldn't expect them to all get exactly the same measurement for each person because they're different people, their eyesight's a bit different, and they're using slightly different instruments to measure them, so they're all going to be a bit different. But you would expect them to be highly correlated. If someone's tall, you'd expect everyone to, all of my four measurers to, you know, measure that person is pretty tall, even if they get slightly different heights, likewise for short people and medium-sized people.

Now, if I did a factor analysis of these heights over my four different measures, I could conclude that there's one way of explaining this pattern of data is that there's a single underlying measure that actually is determining all of these four results, that is the person's height. And of course, in that case, we would know that it was the person's height because we know that people have a certain height. So it's the same thing in intelligence testing. If we find that all of these different tests are correlated with each other in terms of people who do well on one test, tending to do well on another test, then factor analysis, a factor analysis of the data,

shows that a simple way of explaining that would be if there is a single underlying thing, essentially, we don't know what it is yet, but a single underlying variable or thing phenomena, that all of these different tests are measuring, maybe in slightly different ways to slightly different degrees, but just as all of our four different measures, height measures, are measuring height in or being in different ways.

Likewise, our many different intelligence tests are all measuring intelligence just in slightly different ways and getting somewhat different results, but nevertheless correlated with each other. So that's what factor analysis essentially shows, that there's some underlying factor that accounts for a large proportion of all of the variants, maybe about 50% of the variance in scores and intelligence tests.

And this single underlying factor, the fact that there's one of them is very important because we want to identify this factor with intelligence, right? At least that's the idea. But that itself is controversial. The fact that there is a single factor that these tests are measuring just tells us that the test or measure something in common. It doesn't tell us what that thing is. Remember, in the case of the different people measuring heights, our height measure,

is we know that they're measuring height, because we tell them to do that. In the case of intelligence, it's kind of controversial exactly what the intelligence tests are measuring, because intelligence is a much fuzzier concept than height. So at this point, at least, we can't say that factor analysis tells us that there's a single one thing called intelligence, that all the different intelligence tests measure. All we can say is that there is a single common thing that they are all measuring, whatever it might be, whatever we want to call it. And so to avoid this sort of controversy about what to call this,

this single common factor. I think it might have been Pearson, a psychologist a long time ago, I forget exactly who and exactly when, but he called it just the G factor. So this is what the G factor is that I mentioned before and that you'll see over and over again in the literature about intelligence. It is a statistical construct. It's a number that comes out of doing this factor analysis of the relationship between scores on all of these different intelligence tests. And I call them intelligence tests, not because we know they measure intelligence, it's because we sort of think that they measure intelligence or that's what they're designed to measure. Maybe they don't actually measure intelligence. That's a separate question, but they try to measure

intelligence, right? So all of these different tests have, as I said, correlations between them. If you do well in one, you do well on another. If you do badly on one, you're likely to do bad on another. You do a factor analysis of that. You get an underlying variable that seems to be explaining a good chunk of all of the variation across the tests. This variable is called the G factor. So it itself is a statistical construct. It's not the same thing as intelligence. It might be a measure of intelligence. Maybe the G-factor is a measure of underlying intelligence. And that's what most psychologists think. Most psychologists think that this statistical thing, the G-factor, that we find commonalities between different intelligence tests,

they think that that is what we mean in natural language and also in cognitive psychological research when we say intelligence. It's actually this G-factor thing. Or in other words, the G-factor is a measure of intelligence. The G-factor is a statistical artifact. Intelligence is a thing in your head. It's important to keep those concepts distinct. But the reason I emphasize this difference is because that step is itself somewhat controversial. You can believe that the G factor is a thing, but say it's not really intelligence, it's something else. Maybe it's a combination of different cognitive abilities, for example, which some have said. Or maybe it's a combination of some cognitive abilities and say motivation. Maybe you don't think intelligence is the same thing

as memory, and maybe you think the G factor is measuring memory ability to some degree. So maybe you think that there's some sort of conflation there. So there's all the different possible positions you can take about what the G factor is. The most common one, and the mainstream one in cognitive psychology and study of intelligence, is that the G factor is a measure of, or corresponds to, intelligence in the way we normally understand the word. Remember, the definitions that I gave before. Okay, so hopefully that was reasonably clear. It's a little bit strange thinking about this factor analysis and G-factor intelligence and stuff, but the basic idea is simply that intelligence tests show that there is some common thing, some one thing, that

explains a good deal of the variation across people in test outcomes. Now, that thing is called the G factor. Whether you think that that is intelligence is going to depend on how you define intelligence and what else you think is going on in when people are taking these tests. But usually psychologists describe G factor as directly corresponding to intelligence. Now, at this point, I want to speak briefly about something that some of you may have heard of, which is a book by Stephen J. Gould. Stephen J. Gould is not a psychologist. He is an evolutionary biologist. But in the 80s, I think, he published a book called The Mis Measure of Man, which was a critique of IQ testing and definitions of intelligence and so on. This is an exemplar of a number of works like this,

which continue to this day. Usually not published by cognitive psychologists, but that's not a strict rule, some of them are, but what they have in common is they're all pretty critical of the very concept of IQ or intelligence testing. Some people say that there's no such thing as intelligence or that there's no such thing as a G factor or that there are multiple intelligences or that intelligence tests are biased against particularly people. There are various different claims and we don't have time to discuss all of them. But here I just want to discuss one criticism that Stephen Jay Gould raised in that book, which is that psychologists think or speak as if The G factor is like a thing that exists in the brain.

And, you know, it's certainly true that some people, perhaps more journalists than researchers, but nevertheless, some people have been cautiously sort of talked about the G factor as if it's like a substance or something like that. The word he uses, I think, is raification. That is making a, ratification means making an abstract thing as if it were a concrete real thing. And this is a very important critique to understand, because I don't think it's a good critique of the G factor as a concept. but it's a good critique of poor use or misunderstandings of the concept, which is why I raise it. So it's very important to understand what the G factor is. It's not a thing in the brain, right? It's not like it's a neuron or a particular region of the brain or a lobe of the frontal cortex or something like that, the frontal lobe of the cortex or something like that.

It's not a structure like that. It's not something you can point to. In fact, the G factor is just an abstraction found through factor analysis, right? It's a bunch of correlations, essentially, or an underlying factor that explains correlation patterns. So it's an abstract thing. It's the thing that exists in the world. The question is, why is the G factor found through these statistical methods? Why do we find a G factor? There's no reason a priori beforehand why we would necessarily find a G factor. If people did just differently, if some people did good at some tests and some people did poorly at others,

and there was no particular correlation, then there wouldn't be a difference. G factor. So the question is, why do we find one? The mainstream answer in cognitive psychology and intelligence measuring is that we find a G factor, which is the abstract thing, because there is a single cognitive ability or collection of related abilities that exists or is instantiated in people's minds, mind slash brains, because we think the mind comes from the brain, right? So there's a single set of abilities instantiated in people's minds called intelligence. And this ability, or set of abilities, leads people to do well at a wide range of different tests. And as a result, when we take these measurements of the tests, we find this pattern of correlations,

which then leads us to find that when we do a factor analysis, the G factor. So the idea is the intelligence, the fact of intelligence, which is sort of a fact about people's minds slash brains, leads to us finding the statistical abstraction called the G factor. But they're not the same thing. you can think of it as a G-factor as a measure of intelligence, loosely speaking. So Stephen Jekyll was correct in his criticism of IQ and IQ testing in the sense that it's important not to misinterpret the G-factor as if it's a thing that exists in the mind or in the brain. It isn't. It's a measure, quite an abstract measure of something that exists in the mind and the

something usually we think of as intelligence, but that is somewhat controversial. Now, I have said that usually the usual interpretation of the G factor is that it's intelligence, but why would we interpret it that way? Remember, the G factor, it just explains, literally it just explains the pattern of variation, would have to have further evidence to then say, well, why do we think it's intelligence? Why not something else? Some other cognitive ability or combination of abilities. Well, there are a few reasons why we say this.

One is simply, well, that the tests are designed to measure intelligence, so that if they're measuring something, it's presumably intelligent. right? I mean, that's what the tests are supposed to do. That's some evidence, perhaps, but it's not enough by itself. But there are other reasons to think that the G-factor is a measure of or corresponds to intelligence. One of them is that we see very similar correlation patterns with academic scores, that is, test scores and academic grades that people get in school and university show very similar correlation patterns and also correlations with achievements on intelligence tests as do, as we see across the intelligence tests. That is, if people do well on

intelligence test, they're likely to do well at their academic studies, and vice versa. And similarly, if people do well at one subject at school, they're likely to do well on other subjects. So it seems that there's a single underlying thing that explains a good deal of the variation, not only in intelligence tests, but also in terms of school subjects. And so, again, if we think that there's one underlying factor that determines or explains a large portion of variation of people's performance on school subjects, that sounds a lot like it's probably be going to be something related to intelligence, right? Because, again, we think that school subjects require, you know, good memory, abilities, abstract

thinking, problem solving, etc., etc. Another reason we think that the G-factor corresponds to intelligence is because performance on intelligence tests is reasonably constant over time, especially after early childhood. So IQ tends to fluctuate quite a lot in childhood, but through adolescence and once you get to adulthood, IQ, that is, performance on intelligence tests, is pretty steady. certain performances increase and certain performances decrease, which we talk about in a bit. But overall, there's a very high correlation between test performance in one year and test performance

the next year or even decades down the line. So, and studies have been done find high individual correlations over time. That is, it seems that whatever this thing is that intelligence tests measure, it's pretty constant overtime in adults. And again, that's something that seems to be consistent with intelligence. We think of intelligence as something that corresponds to it, or that an individual possesses, and it's fairly concerned over time. It's not like a mood, for example, where we think someone's moods sort of swings can swing quite a lot. And on some days you have good days and some days you have bad days. We don't think of intelligence is something that does that in quite the same way.

So that's, again, further evidence in support of the fact that the G-factor is measuring intelligence and not something else. Another reason we think that the G-factor is a measure of intelligence is because IQ, or other intelligence measures, have pretty strong correlations with a number of real-world outcomes, which we think relate to intelligence. For example, education is one of the biggest ones. So, according to one matter analysis I saw, there's about a 0.5 correlation with education,

which means that about one quarter, or just under one quarter, of the variation in education attainment across different people, can be explained by a difference in their intelligence test scores, which might not sound like a lot, because there's still three quarters that aren't explained by that. But remember that if we have a single factor

that explains such a line, proportion of a variation in education outcomes. Again, it's plausible that that's something to do with intelligence. Intelligence or IQ also correlates with occupation, that is, like, how prestigious occupation is, and also occupational performance. It also correlates with income, although the correlation is not as good as with education. In fact, intelligence correlates with a large range of positive life outcomes, including things like life expectancy, which you might not necessarily expect, and intelligence. So one study found that a large range of positive life outcomes, including and intelligence. So one study found that a single standard deviation advantage in intelligence,

that would mean if your IQ went from 100 to 115, say, that's a one standard deviation change, was associated with a 24% lower risk of death over a fall-up range of between 17 and 60 or 9 years, so that's a long time. So in other words, more intelligent people have a significantly lower, like quite a lot lower, 24% in this study risk of dying over a certain period of time. So that's actually a big difference. So intelligence certainly does have big real-world outcomes. So it does seem that whatever it is, it's a positive thing because people who have more of it do better in all sorts of ways, education, income, life expectancy and so on. And it seems to be something that's quite closely related to thinking and reasoning and scholarly achievement. So certainly it's sounding a lot like intelligence here. there have been a couple of studies that I found where psychologists have attempted to look at how psychologists, that is, academic psychologists define intelligence and compare that to how everyday people, lay persons define intelligence. And there seem to be sort of some commonalities and some differences. Many lay persons seem to make a distinction between academic intelligence or school intelligence, school smarts, and everyday intelligence or street smarts or some sort of practical intelligence. which is interesting. And it's not exactly clear what the distinction is, in part because, obviously, if these are lay persons, they're not going to have a very well. I'm likely to have a very clearly articulated sort of definition of what the difference is. But I'd probably say that the concept of practical intelligence is probably different to what psychologists would mean when they say intelligence, because practical intelligence might include things like knowing how to behave in certain situations or knowing how to get things done in the real world.

know how to work people or how to work the system or how to read a situation, stuff like that, what to do. And it seems largely that that's going to be, consist of a lot of facts you need to know, also a lot of experience and maybe combined with social skills, personality, cultural awareness, things like that. And certainly there's going to be some sort of problem solving or abstract thinking underlying some of that. But a lot of it is also, I think, going to be stuff that is what we might think of as not particularly cognitive, or at least not problem-solving related. That is just knowing facts or knowing what to do in a situation or also social skills and things like that.

And it's not quite clear that that's the same as the type of cognitive problem-solving, abstract reasoning that psychologists are really focusing on, although there's going to be overlaps as well. But it is interesting to sort of compare how lay persons and psychologists think about the concept of intelligence. Okay, so that's some information about measuring intelligence and the G-factor.

I've already talked about some of the debates surrounding IQ and the G factor, and I'll now talk a little bit about more, a little bit more about some of them, including this idea of multiple intelligences. First, though, it occurs to me that I haven't, there's a few key things about IQ that I haven't stated yet, so I should give these key facts. First of all is sort of just the numerics of how IQ works. So remember, I said there are lots of different intelligence tests and different ones incorporate different aspects of them. So psychologists often not too worried about exactly which tests you take because they all correlate with each other, right? This is the idea of the G factor. So different tests have what it's called a G-loading, which is loosely, it's basically

a correlation with this underlying G-factor. So some tests have a much higher gene loading than others. Raven's Progressive Matrices, that I mentioned before, has a very high G-loading, which essentially means that we think it measures this underlying single intelligence thing very well and doesn't measure too many other things. The lower the G-loading a given test has, the more it's measuring other stuff

than the single underlying G-fact or intelligence as we think it is. So that other stuff could be just knowledge, or it could be practical abilities with a pencil, or if it's some drawing activity, or just working memory unrelated to intelligence, all sorts of other things,

or just idiosyncratic individual differences. It just could be less reliable as a test, less precise. So the higher the G-loading of a test, the more closely and exactly it's measuring intelligence generally is the interpretation. Strictly, it's just the higher the G-loading, the more it correlates with the G-factor. But since we think that that's intelligence, we can interpret a high G-loading as more precise measurement of intelligence. So basically, if you know the G-loading of a test and you have information about the distribution of the scores of a test, you can effectively convert that into a sort of loose IQ rating.

It's not really an IQ rating if it's just one narrow test because IQ should incorporate a range of tests. But you can convert one test to another, basically, using these correlations as well as knowledge of the variance and mean of tests and other things like that. So that's why in many of these studies there's often not a deep concern of exactly what test was given. Sometimes that can be relevant, but often there's the ability to interconvert. So as long as they're a reasonable test of intelligence, then that's okay. Now, IQ though, the way it's typically reported, the term IQ stands for intelligence quotient, and a quotient is actually just one number divided by another. So 0.5 is a quotient. It's, you know, one divided by two. Simple quotient. The original idea of intelligence when it was proposed

is a way of measuring children's ability, particularly there was a desire to know whether some children are sort of advanced and some children needed extra help in terms of education. So the basic idea was that you compare the performance of, of a child on an intelligence test to children who were younger and older. And the intelligence quotient that was assigned to that child, or the IQ, is a quotient of their intellectual age, that is their performance on the test relative to people of different ages, to their actual age or biological age.

So that is, if a three-year-old performs as well on a test as a four- and six-month-year-old, then their intelligence age would be four and six months, while their biological age is only three years. So four years and six months divided by three years gives an intelligence quotient of 150. So that child would be bright. They're relatively advanced for their age. So they have an intelligence quotient of 150. That was the original idea of IQ. Now, that procedure of dividing intellectual age by biological age is no longer used, although some people still seem to talk about this,

but psychologists don't use this anymore. And so IQ isn't really accurate anymore because it's not actually a quotient of anything, that you don't divide anything like that. But the name is stuck, so we still talk about IQ. The way IQ tests are usually standardised these days is that the mean, that is the average, score that people get in a population,

is defined to be 100. So the tests are what is called normed, essentially scaled, so that the mean is 100. and as we'll get to the slenderfleck later on, that has to be adjusted periodically to keep the mean at 100. IQ tests are also usually adjusted so that boys and girls or males and females both get the same mean of 100

so that by definition can't be a gender difference there. So the mean of an IQ test is 100. The standard deviation is usually around 15. 15 is a standard, although sometimes some IQ test are a bit different. And effectively that means that if you have a standard deviation of 15 and a mean of 100, that means 95% of the population will score between 70, which is two standard deviations down, and 130, which is two standard deviations up. So IQs of more than, say, three standard deviations out from the mean, so that would be, what, 55 and 145, IQs outside that

are very rare, like, you know, one in a thousand or two in a thousand or something like that. So most people fall in that sort of 70 through 130-ish range. You have to be careful about IQ test, there are all sorts of essentially fake IQ tests on the internet or elsewhere that you might see. If you want a proper IQ test, you need to go to proper psychologists and it will take a couple of hours. It's, because it's a full-on test that will incorporate

a wide battery of different types of cognitive tests, of cognitive tasks. So, bear that in mind, so in various discussions of intelligence, IQ scores will be given. The mean of an IQ score is 100, and the standard deviation is usually 15.

Okay, so, that being said, Let's talk about some debates relating to IQ and the G factor and interpreting these things. So Howard Gardner is a figure who comes up here. He has proposed a theory of multiple independent intelligences that's attracted a fair bit of attention. So this is multiple intelligences theory, which you may have heard of. It seems to be quite popular in education, at least among some. The concept of the G factor or the general intelligence factor has been criticized on a number of bases, and one is that there's not just one thing.

There's multiple different types of intelligences that are independent of each other. This is what Howard Gardner thinks. The fact that they independent of each other is important because if we have a bunch of different intelligences which are correlated with each other, meaning that people who are more intelligent at one tend to be more intelligent at another, then, well, really, that's just a G-factor. And then we can say, well, the underlying thing causing those correlations is a single intelligence thing, or at least that's plausibly the case. So in order for this theory of multiple intelligences to be meaningfully different to the standard G-factor intelligence theory, it has to say that these different multiple

intelligences are independent of each other. So people who are good at one aren't necessarily any good or any bad, any better or any worse, at another of these independent intelligences. They're completely distinct and independent of each other. So these separate independent intelligences are in his categorization. Visiospacial, bodily kinesthetic, musical, interpersonal, intrapersonal, linguistic, and logical mathematical. Now, typical intelligence tests probably would only incorporate

the logical mathematical and some of the linguistic aspects of his multiple intelligences. Maybe some of the intrapersonal stuff, that's a little bit vague. He talks about it as understanding one's own goals and intuition and motivation, stuff like that. So some of that might be measured by intelligence tests,

I'm not sure, but certainly many of these other things are not, or at least not supposed to be measured by traditional intelligence tests and are not necessarily correlated with usual intelligence measures. So Garner's critiques of mainstream G-factor intelligence theory include that the G-factor

just measures things or particular abilities that are valued in school-type settings, so knowledge and abilities in those types of, you know, that you learn at school, like vocabulary and mathematics and logical puzzles, things like that, and that there are a broader range of intelligences or competencies that are not properly recognized by this single G-factor model of intelligence. This is also sometimes related to a distinct but related critique of intelligence,

which is that what is valued in terms of intelligence differs between cultures, and so the fact that in a Western culture, particularly literate culture, we value certain sets of skills and call that intelligent, doesn't necessarily mean that in different cultures, those same things are values, especially pre-literate cultures. and that if we ask them to develop intelligence tests, then we would probably get something very different. So these are some of the arguments that people have given against the G-factor theory,

or more specifically it's the intelligence interpretation of the G-factor, not necessarily disputing the G-factor finding, although some do that as well, but say, well, yeah, there's a G-factor, but it's not the same thing as intelligence. Intelligence is something broader than that. So mainstream psychologists, you know, that is sort of most intelligence, cognitive psychologist researchers,

generally not too sympathetic towards the multiple intelligence theory or these sorts of culturally based criticisms of IQ. There's certainly some validity to the culturally based criticisms and concern about making sure the tests are valid across cultures, and they're not just testing culturally specific things, because they're not supposed to be. However, I think there are some good reasons to think that there is a single underlying intelligence, and the multiple intelligence theory of Gardner is not, at the very least, sort of needs further development and clarification as to exactly what it's supposed to be explaining. So one argument against these sort of critiques is an evolutionary argument, essentially that it seems very clear that

humans differ from all other animals, especially non-primates, to a very significant degree, and that a large part of this difference is not anything that's necessarily an ability in terms of, say, ability to see better or greater dexterity or things like that, that it seems that these differences are largely cognitive, that is stuff that we would broadly define as, well, intelligence. So it seems clear that we do differ from animals on some spectrum like this,

and it seems also clear that these abilities must have developed for a reason, evolutionarily, that is, they must have offered some selective advantages in order to be selected for, some survival advantages and reproductive advantages. And it seems that those won't be purely culturally dependent, that there will be some universal application to those. Just purely evolutionarily speaking, we would expect that.

Another consideration is that intelligence is not the same as what a culture values, especially what sort of ordinary people lay persons in that culture value, because it's important to note that what lay persons in Western cultures value isn't really the same as intelligence either. Some aspects of it will be, but many other sort of practical skills and abilities, sort of common sense and street smarts, that sort of stuff. That's not really what psychologists think of as intelligence. So just sort of asking people or even imagining what people in different cultures would say,

it counts as intelligent. It's not clear that that actually tells us anything about what psychologists are interested in. That is, more abstract cognitive abilities. People in pretty much all cultures are interested in just sort of getting by on a day-to-day basis, doing what they need to do to survive, to get the paycheck, the bills paid, or the food hunted, whatever culture you're living. in, right? So for them, those particular skills that help for those things, whether it's managing the bureaucracy or milking the cows properly or handling the machinery, whatever type of job you have,

it's doing that sort of stuff that's typically most highly valued by people. That's what they consider to be most relevant. Someone who's really good at that, well, you know, they're switched on, they know what they're doing. They get stuff done, right? That's not really what psychologists are interested in when they're talking about intelligence. They're interested in abstract problem-solving reasoning abilities. So whether they're the same thing as what's valued in a culture when you ask people is sort of beside the point. And in particular, it doesn't prove that there isn't any relevance to these more abstract cognitive abilities that has cross-cultural relevance. The idea is not that intelligence tells us how to do particular things or gives us any particular knowledge.

It's rather a more generic underlying ability to develop skills and knowledge in a particular the cultural context. And the abstract abilities to do that is not really something that most people have access to. That is we don't see someone's ability to learn. We just see what they know. We don't see how well someone's problem solves in the abstract, in most cases, unless you're testing them specifically. We just see how good they are at doing particular tasks. So I think that this sort of what's valued culturally criticism is a little bit misconstrued in that it focuses on the wrong thing. The question is not what's valued in a culture or what people sort of see as being useful. It's rather what abstract, higher level, broader

sort of domain abilities underpin these specific functionings or specific adaptations in particular environments. And the hypothesis is that there is a single cognitive ability that helps with all those things, it helps with learning and problem solving and so on, and that we call that intelligence. Even if it's not quite the same, in psychology we call it intelligence, even it's not quite the same as layperson usage of intelligence in different cultures. Now, regarding Gardner's criticism that the G-factor simply measures school-type knowledge and abilities, I think this is largely disputed by mainstream intelligence researchers. So the reason for that is because there are various intelligence tests that don't rely on

anything that we would normally think of as sort of school-type knowledge and abilities in a Western setting. In particular, types of tests that rely on analyzing pictures or moving around blocks and things like that. And furthermore, there are some very strongly school-like tests that actually don't have very high G-loadings at all. That is, they're not closely strongly correlated with the G-factor. For example, tasks that use numbers and words to assess clerical aspects or perceptual speed and accuracy, things like writing letters, for example. So some things people do in schools and learn a lot in schools aren't really closely correlated with intelligence test scores, whereas other things that aren't are highly correlated.

So it does seem that there's a difference here, that intelligence tests aren't just measuring sort of test-taking ability in terms of like a school test. Likewise, it seems pretty clear that intelligence isn't just measuring aptitude for being a scholar in a Western university, because it correlates with all sorts of other things that are completely independent of that or completely outside of the academy, including, well, educational attainment, obviously that's relevant in the academy, but other things like occupation status, poverty and income, incarceration rates, chronic welfare use, and as we said before, life expectancy and rates of mortality. So across the

social realm, intelligence is a relevant variable. And as I said, it's a positive variable in the sense that you score higher on these tests. You're likely to do better at these other measures as well. So it's clearly not just measuring what's good in a Western education setting. There's something broader than that. And that's what we would expect to see if indeed it is measuring a broader reasoning, thinking, problem-solving ability, not just like test-taking ability. Another piece of evidence that many psychologists will cite in support of the interpretation of G-factor as intelligence is various biological correlates of G-factor or of intelligence test performance. So in particular, the G-factor correlates with certain biological variables like cerebral glucose metabolic rate and certain parameters of evoked potential, so essentially that's measuring electrical activity in the brain. It's also correlated with reaction time, inspection time, and certain other things.

basically the sorts of biological variables that we would expect to be correlated with. In particular, it's been found that people who score higher on intelligence tests use less energy in their brain, that is, lower glucose metabolic rate, to do a certain task. That is, for a given task of a given difficulty, more intelligent people use less brain power, that is literally less energy, less glucose metabolism, to do that task than a less intelligent person. Now, that result holds so long as the less intelligent person doesn't give up. That is, if the task is too hard, then the less intelligent person gives up. And, of course, then they don't use as much energy, because they're not trying to do the task.

But it does correlate with that. So, again, that's exactly what we'd expect if intelligence is measuring something, some mental ability that meaningfully derives from the underlying biology of the brain, that relates to mental processing. So for these sorts of reasons, many psychologists have said that Gardner's theory of multiple intelligence really aren't, they're not multiple intelligences. that really what he's got is a combination of some things that are, some abilities that correlate strongly with G, general intelligence, that would be including like the logical mathematics and the vocabulary one, certain special talents, like the bodily kinesthetic, for example, ability or the musical ability, and other things that are more like personality characteristics, such as the interpersonal ability, which seems to almost be measuring something like personal ability or extraversion or something like that. So there's a combination of cognitive, non-cognitive personality and domain-specific cognitive abilities

that his multiple intelligences seem to mix up. And it's not really clear that the particular, like Y-7 exactly, there's no clear justification for why they're split up in the way that he's done. Just because you can come up with a classification that seems a reason that doesn't mean that there's evidence that actually correlates or corresponds to something meaningful psychologically or biologically. Now, there is evidence that the G-Fact, is that all the evidence that I've cited before, the biological evidence, the factor analysis

evidence, the correlations with real world outcomes. These are all evidences in favor of there being a single underlying factor that is determining the outcomes here. Remember, the position is, the claim is not that it's the only factor. It only explains something like half of the variance, so there's a lot of other stuff going on as well, including domain-specific abilities. So there's not just one underlying cognitive ability, there are domain-specific abilities as well. So it's not just how smart I am, it's also how good I am at that particular task. But the point is that there is evidence for this underlying single cognitive ability. And it doesn't seem like there is similar evidence for Gardner's seven or six or seven different multiple intelligences.

And as one study I read has said that, well, actually, I'll just quote from them here, quote, many types of abilities can be thought of as intelligence of a kind, but it has proved hard to show that measures of emotional intelligence or social intelligence contribute to behavior we would want to call intelligent over and above their correlation with conventional IQ tests, end quote. So it's just not really clear what these multiple intelligence measures are doing in terms of measuring something that we want to call intelligence. They might be useful skills or abilities or personality traits, but intelligence is understood in psychology generally as a cognitive ability, thinking, reasoning, planning. It's just not clear that emotional intelligence or social intelligence or bodily kinesthetic intelligence is the same in that respect. That is not a value judgment about how useful or important those things are.

For example, I think extraversion actually does correlate strongly with some other variables like income and life expectancy, for example. I haven't checked that specifically, but I believe that's the case. And certainly it's reasonable to suppose that, say, measures of social and emotional intelligence might have real-world correlations with these sorts of variables. But that doesn't mean they're the same thing as intelligence, broadly speaking, or that they're like a different type of intelligence. I think they're just something else. So partly this debate comes down to how you want to use the word intelligence. Do you want to encompass a wide range of competencies, or do you want it to be specifically cognitive? That's, you know, thinking, reasoning, abstract planning, stuff like that.

Now, before I close out this first part of the episode, I wanted to talk about one last thing, which is this sort of, it's not exactly a model of multiple intelligences, but it's along vaguely comparable lines that a few theorists have developed. And this is the idea of, specifically the one that I'm looking at, the version of the theory that I'm looking at, is called a three strata model of intelligence. Now, according to the three strata model, at the bottom, I suppose, is the G factor, that is the single underlying intelligence variable. So this is, you know, the one underlying factor that's, you can measure in different people. Now, at the next level up from that, what are called broad abilities or domain-specific factors. So exactly how many of these there are.

It depends on who you ask. One source says 10, but this is still pretty controversial. But the point is there's certainly more than one, but not a huge number of these. So the two main ones that you here discussed are GF and GC. These are fluid intelligence and crystallized intelligence. Now, basically, fluid intelligence is supposed to be more like problem-solving ability, whereas crystallized intelligence is more like facts, you know, vocabulary, memory, stuff like that. So fluid intelligence is sort of more similar to,

to just intelligence that we've been talking about. And in that sense, some psychologists have said that intelligence tests, actually, they're more so measuring GF or fluid intelligence, and crystallized intelligence is kind of a separate thing. They're still controversy about that, because they tend to be correlated with each other. But according to this three strata model, there are sort of levels in the structure.

So there's the underlying intelligence, which is just a G factor. So it's present to some degree in all measures of cognitive ability. Then above that, on the second layer, the broad ability, so like fluid intelligence, crystallized intelligence. Some people say there's a mathematical one. Maybe there's a musical one. Broad domains. They're not independent of each other

because they all relate to this underlying G, this underlying intelligence factor, but they're somewhat distinct as well. And finally, at the very highest level, the third stratum, you will have fairly narrow abilities, like, for example, native language vocabulary knowledge, ability to perform specific

arithmetical operations, ability to discriminate musical pitches, stuff like that. So basically things that are specific individual test would test for. These narrow abilities are things that you can train, that you can practice doing fractions, or you can practice discriminating musical pitches. And you can get better at those individual tests of those narrow things. But this underlying G, it's less clear whether you can practice that. So the idea of this three strata model is that you sort of go from most fundamental and broadest to more specific and applied as you sort of go through the hierarchy.

So some people might be particularly good at one of these narrow. abilities, even though they don't have a very high underlying intelligence. And vice versa, someone who has a high intelligence might not be very good at this particular narrow ability. But nevertheless, overall, there will be correlation. So in general, people who are good at one narrow ability will tend to be good at other narrow abilities, but not as a hard and fast rule, as a general correlation. That's this three strata model, which I think is a useful way of thinking about variation here, because it allows us to understand that there is variation across the narrow abilities and across the domain,

or broad abilities, but there is also variability across the overall underlying single intelligence variable, or the G factor. So there's variation at different levels. And one application that this theory has been applied to understand is how IQ changes over a lifetime. So it's thought, for example, some test results have indicated that fluid intelligence decreases over the lifespan. Essentially, this means processing speeds, working memory and stuff like that gets a little bit slower. It deteriorates after, you know, mid-20s or something. But crystallized intelligence, so vocabulary knowledge or other sorts of things, increases over the age and keeps growing at least until you get to say the 60s or something like that.

And this sort of makes sense if you think about it. The sort of mental acuity might peak around the 20s, but sort of crystallized knowledge or things that you've learned, skills, even mental skills, tend to increase over time. But this is still somewhat controversial, so this is not firmly established. This is a theory that I think is, or hypothesis that I think is useful to think about. Okay, so I think that's a good place to stop it for the moment. So in this first part of the episode, we've talked about defining intelligence, measuring intelligence, the G factor, and debates about IQ and the G factor, and whether that measures intelligence or whether it's something else.

We've also talked about theories of multiple intelligences and some critiques of those. In the next one or two or two episodes, it's looking like this might be a three-part episode now. I'll talk about the biology of intelligence, methods to increase intelligence, the heritability of intelligence, we'll talk about the Flynn effect, increasing IQ scores over time, and then we'll talk about group differences in intelligence and some of the quite controversial research about that. So if you've enjoyed this episode, please send me an email at FOS12 at gmail.com. That's FODDS1.2 at gmail.com to give me some feedback, suggest ideas for future episodes, or just say hi, I always love to hear from listeners.

You could also leave a favorable review on iTunes or another podcast aggregator of your choice. iTunes seems to still be the most popular one, so that's particularly helpful, but anywhere is appreciated. Otherwise, thanks for listening, and I'll talk to you next time.