All right. Welcome to unit six, Office Hours. This week, we're talking about intelligence. As you remember, this is your Professor Matt McGue and I'm your TA Bridget Carey. So we'll just jump right into things and start off with a question from Steven. And he has a question about the Flynn effect. [SOUND] And he says, I accept we've got we've gotten better at doing IQ tests over the past 75 years. But I don't accept that we're intrinsically smarter. Evolution doesn't work at that pace. My laymen's explanation for the Flynn effect is that many of the kinds of problems used to test IQ manipulation of geometrical shapes, for instance, were unfamiliar 75 years ago, but have become very familiar now through the use of animation and visual media. So I guess where he's going with this is, can, do you have an explanation for the Flynn effect? >> So, first of all, you'll see [LAUGH] the students will find out that these, even though Steven's not a plant Bridget talks with me about the questions ahead of time. So I come in prepared. So, these aren't impromptu and, unfortunately the Flynn effect is one of my favorite phenomena in psychology. So I have a couple slides to make a couple points about the Flynn effect and responding to Steven's question. The first thing and I've actually posted this in one thread, I can't remember what thread it is but it, it appears that, at least in developed countries, the Flynn effect has probably come to an end. This is here data, you're not seeing it Bridget, but this is data from conscripts in Denmark and what you see here is the year of their testing. So they're being tested about 18 years after their birth. And by about 1998, so they had, had been bor, born in 1980. Actually it, it's pretty level off. And, in, in fact, it may have declined a little bit after that. There's data from other countries that would indicate that the Flynn effect isn't continuing a, again, in at least developed count, countries. In developing countries it probably continues. The second point gets more directly at Steven's question, is, well, why did it happen? >> Mm-hm. >> Why did IQs go up in the developed countries and continue to go up in developing countries? There are various explanations for that. Steven raises one hypothesis, which I think Is certainly entertained. And that is that we become more test sophisticated. >> Mm-hm. >> As we go up, we have more ex, exposure to the materials on the test. And therefore it's not a real increase in intelligence, but maybe just some artifact of the way we assess intelligence. I think that's probably part of the Flynn effect. I don't think it's all of the Flynn effect, and if I interpret Steven's question correctly, I don't think Steven believes that's all of it either. I think he, he says it's maybe part of it, but maybe there's also something real going on. What maybe, many people don't recognize, many young people, is that, if, if you look at the 20th Century, let's just take the country we're living in, the US, there were phenomenal increases in educational attainment. And I have a slide here for this as well. See, I told you I really came- >> You are prepared. >> This is just the average years of education, in the population of adults by year of, of them being assessed. Basically what it shows and then overlayed here is actually some graph showing the Flynn effect, increases an IQ. But what it shows is that it wasn't actually until 19, I think it was about 1970 that the average individual in the US had a high school education. That the majority of the population was graduating from, from high school so- >> That late huh? >> So well that's the av, average in of, of adults. >> Mm-hm. >> So there's been massive increases in educational attainment over the 20th century that happened at the same time that average IQs have increased. The other thing and the last thing on, on this point that, what, at least in the point of why it might occur comes from this study from a, a researcher. I forget, I'm forgetting his first name. I think it's David Baker, but I might have that wrong, so I apologize for that. And what he's done is actually quite interesting. He's gone back, we all have this notion, and we don't all have this notion. Many of us have the notion, that oh, when we went through school, it was the golden age of education, right. My education was so rigorous. Well, what Baker did, is he actually went back and he looked at text books. Going from the earliest 20th century, again in the US and, and only mathematics textbooks, right on through the end of the 20th century. And what's plotted here is essentially his measure of how complex, how demanding those elementary school arithmetic and mathematics books were. And what he's showing is that indeed these children, as they over time have been exposed to more demanding mathematics curriculum. So the, the second part of Steven's question is that I think probably educational exposure has contributed to the Flynn effect. The last point and I'll be very brief on this is, well is it real? Is there, have we gotten smarter what are the indicators of that? That's hard to tell. I think the evidence is on balance that we have gotten smarter. Our vocabularies are definitely bigger. That's not just test sophistication. People have more sophisticated vocabularies, we can document that, today than they had 50 years ago. We know more about the world today than we did 50 years. So probably, there really is some part of the Flynn effect that is real increases in intelligence. >> Interesting. Very thorough explanation. >> [LAUGH] Yeah, okay, well I was a little bit prepared for that one so- >> All right. Our next question is about realizing genetic potential. And it comes from Anna. >> [SOUND] >> And she says, as I understand, realizing genetic potential means that you inherit certain set of genes that allow you to have an IQ in the range from A to B and your IQ of X is high in this range, i.e close to B. >> Mm-hm. >> So she says, to check this, could we take the mean of IQ of the parents, stimate expected mean IQ of children using the Flynn effect. >> Mm-hm. >> And compare it to the actual mean IQ in the two groups of high SCS and low SCS. >> Mm-hm. >> So she wants to know would this essentially allow us to see if these groups are letting us reach our genetic potential? >> Her name is Anna? >> Anna. >> Anna, very clever question, Anna. So I, I think that's, it's a reasonable thought. And, basically what she's asking is. If you took the mid parent average IQ. Is that a good indicator of, of a child's genetic potential? And I guess the answer to that, I'd have to say is maybe not. Because, right, if, if it's the mid parent IQ in a working class family where they have not as many educational opportunities than a professional family, then that probably isn't a good indicator of what that child might be able to become. Children certainly do achieve beyond their parents if given certain opportunities. So I would say it's a, it's a good idea, but maybe in certain backgrounds, the mid parent IQ might be an underestimate of what the child could actually do in, in a very rich environment. I, I also don't want to I have a couple slides here, but I, I don't think I'm going to use them here Bridget. I, I did use the term genetic potential and although I think Anna used it in a little more sophisticated way than I probably introduced it in the lectures. So, thank you for that, Anna. I don't want to convey the thought, which would be inappropriate just to think that we're, you know, we're born stamped with some number. That that's as much IQ as we could, we could ever achieve. >> Mm-hm. >> Or that's how long we could live, or that's what our cholesterol could be. Whatever our genetic, whatever the really, realization of our genotypes are depends on an environment. And environments can change quite radically, so it's,it's the notion of genetic potential, is, is maybe a kind of a heuristic notion rather than something that exists in reality. >> Great. Our next question is about stability of IQ. And this is coming from the general forums instead of the office hours forum, but it was such a great question we decided to include it. [SOUND] I'm wondering if IQ changes, either increases or decreases throughout adulthood. Would I likely have the same score at 20, 40, and 60, assuming I had no accidents or diseases that would affect my mental capacity? Has this ever been studied? >> So, one thing that I should note here and, and actually this is a very good discussion in the forums. I, I, there are several good threads that are, I think. >> Okay. >> I want to thank people for contributing to them. And so basically what we're doing here is just making information available here to maybe a broader audience than, than those that have gone onto the forums, but I would encourage people to go on the forums. Right, one thing is I guess, before answering the, the question directly to note is that of course IQs are scaled such that at any age, the average IQ should be 100. >> Mm-hm. >> So the pop, the population IQ at any age should be the same. Basically by the way that the tests are constructed. In fact, the way they first determined that the Flynn effect was probably going on, is they recognized that if they didn't recalibrate the IQ test. The IQ test, IQs were creeping up above 100. There's a lot of research on just the stability of IQ, as you know. Psychomatricians call that reliability and from a psychometric perspective, IQ tests are extraordinarily stable over the short run. IQ tests if you take the same person or group of people and measure them on an IQ test. Let's say separated by less than a year anytime, one year less, the correlation's about .9. It's very, very highly stable. >> Mm-hm. >> Over longer periods of time IQs can shift. And there's two factors that underline how large a shift you might ob, observe. One is what was their initial age at testing? And the second is how long the interval? How long beyond one year when they're very stable? And what, I have two slides on this Bridget, but the first is from a, a famous study. That longitudinal study in the US called the Fells Longitudinal Study. >> Uh-huh. >> Fells is in, is in Ohio. I might be wrong. It could be in Indiana, but I'm pretty sure it's Ohio. >> I'll fact check that. >> And what they've done here, and, and again, you're not seeing this is, is they've, administered an IQ test to a reasonably large sample of, in this case, males and females. When they're children and adolescents, and then again when they're 40 years old. And what's plotted is the correlation between their IQ at that age and their IQ at 40. >> Mm-hm. >> When you're less than, let's say about age five, before you've gone to school, your IQ at those ages don't correlate. It doesn't correlate much with IQ at age 40. However once you get to about age ten or 11, then the correlations do begin to become quite substantial. And they, they, the, the correlation levels off a, across age, so by age 11 it's about 0.7. So, there's certainly opportunity for change there, but it becomes pretty stable. The second slide I have here is a really remarkable study, and you're familiar with this study. It's a Scottish study by Ian Deary at the University of Edinboro. And what they've done is they've taken a, a rather large sample of Scottish children who were tested for IQ at age 11. And then they've come back, 70 years later. Now they're 80 years old, I guess 69 years. >> That's incredible. >> And, and they've readministered the IQ test. And what you see here is a scatter plot of of their IQs at 11 and their IQs at age 80. And the correlation between the two is point 0.73. There's a very strong correlation even over a seven year period of time. Nonetheless there are people, if you look at this scatter plot, who, maybe people in here, where their IQs at age 11 were about 85, but now at age 80, their IQs are 110. Recall again that the IQ at agey, 80 is relative to other 80 year olds. >> Mm-hm. >> Nonetheless, some people are moving, can move quite a bit, even though overall IQ is stable. We might think that wow, those would be, and they are, very interesting people. Why did that person increase 20, 25, 30 IQ points over his or her lifetime. When psychologists have studied these people they, it's not like their life history gives any great hints as to what actually happened to underly that great increase in IQ. Maybe they just had an off day when they were taking the test at 11. Who knows. IQ is very stable, nonetheless even though it's very stable some people can change quite a bit. We don't know why they change this. >> Great, our next question is another one from the general forum. [SOUND] And it says in the professor's post, it's noted that IQ at preschool age does not correlate too well with adult IQ. And again you just reiterated that point. >> Okay. >> So it says, does this effect the interpretation of the preadoption and post adoption IQ measure from lecture four? Since the preadoption IQ test is taken on average, around age four? That's actually a great, a great question. It certainly does affect ones interpretation, I'm not sure exactly what he or shes, you know, in, in unit four. I, I should have gone back and looked at, at unit four where we were talking about heritability I think. But certainly it, it would affect. There's the notion that you can get, and actually this is going to come into this weeks- >> Mm-hm. >> Lectures, where we talk about development. That you can have a hot-house effect. That is, you can accelerate mental development, or maybe archly impede it, as well, any young person. But over the long term you may have less of an ability to affect the overall level that a person is able to achieve in their mental ability. So yeah, as far as people can be more advanced or more behind at age four, but eventually catch up or, or retreat back to the, to, to the mean because they just had certain experiences. And it's not until you get later in life that IQ stabilizes. So I think it does affect the interpretation. >> Great. Our next question is about epigenetics and it's coming from David. [SOUND] And he says fascinating subject Dr. Migue. So changes in DNA methylation patterns have been associated with the epigenetic inheritance of height. What has been discovered about the epigenetic inheritance of intelligence? >> The, so, actually I don't know of evidence on the epigenetics of height being affect. The inheritance of height being affected by epigenetics. Regardless, I mean this is going to be a quick and probably unsatisfactory answer to David's question. To my knowledge I don't know of any evidence of epigenetics right now, in terms of general cognitive ability. Might that change over the next couple years, I, I would imagine it might. But right now, I don't know of any evidence. On that account. >> Okay. Our next question comes from Michael. And he's curious about Head Start program. [SOUND] He says results reported in section six lectures suggest that Head Start programs should be effective, but reported Head Start results are negative. Do you have any explanation for this? >> So this is a little bit afield from behavioral genetics. But that's okay. It actually gets into another topic that I'm very interested in. Michael, was that the name? >> Mm-hm. >> Michael. So I'm not sure what Michael is alluding to when he talks about Head Start programs that failed. Head Start programs have multiple goals. They may fail on some, they may succeed on others. So, I'm not sure exactly what he's talking about, however there's one area, and I think maybe this is, it may be what he's getting at. One area where we might say early childhood educational programs did seem to fail. And that is, in this country, in the US, in the early 1960's, there were maybe 10 or 12 early childhood educational intervention programs. The typical study was often a randomized study. That is they randomized children to one condition in intervention versus the control. And, and the goal was to, and these were preschool children, so this was before they went to school. And the goal was to increase their intellectual capacity so that they did well in school. And this is actually, again, prepared ahead of time. But this is a slide from maybe the most famous of these studies. The study was done in Michigan. Called the Perry Preschool study. >> Mm-hm. >> And it shows, I think, results that were fairly typical of these early childhood intervention studies, and I should digress here to relate it back to Michael's point. Head Start is not the same as these early interventions. But usually people talk about them almost synonymously, and I'll do that here, as well. What happened in the Perry Preschool program. So the Perry Preschool program without getting into great detail was actually an intervention, that ended when the children were age five. And what you see here is the average IQs for the the children that had the intervention. And these are very poor children, and this goes back to the 1960s. This is a long time ago, 50 years ago, and the average IQs of the control children, randomized. And you can see at the end, end of the intervention there is actually a large effect of the intervention, about ten IQ points. But once the intervention stopped, and this is maybe what Michael's alluding to, the difference in IQ, actually, evaporates over time. So by the time they're eight, nine, or ten, there's really no difference in the IQs. It was thought, one day and this is the Perry program. But there are other programs like this that, that show the same thing. It was thought that these programs failed. They set out to increase intellectual capacity. They did that for a short period of time, as long as the treatment lasted. But after the treatment ended, the, the, unfortunately the intervention group regressed back to the control group's mean. Now that may be a consequence of putting the children back into the poor environment- >> Mm-hm. >> That they started out in. That may well be the case. But, nonetheless, as a booster, it didn't seem to have any enduring effects. The nice thing about, and, and the reason I'm interested in this particular story, is that a group of researchers in, and including the par, the original parasearchers, have even though they didn't find an effect on IQ. They continued to follow-up the participants in this study as those participants aged into adulthood. And what they found is that the, the treatment group actually did better than the control group. And I have one slide on this. The, the treatment group is in red here and the control group is in blue. The treatment group for example were more likely to complete high school. More likely to have a full time job when they're in middle age. More likely to earn a decent income from when this data was collected and they were less likely to be a, arrested. This is actually, opened up a whole new area of research within psychology that this observation from Perry. Because what it showed was that even though the treatment group did not have higher IQs they actually fared better in life. So it must be that the treatment had something, some effect other than on IQ. And the notion is what happened is that there must have been some effect on their personalities, and in particular they learned how to better control their behavior, so that they could achieve long-term goals. So, I guess I would take a little, maybe, quibb a little bit with Michael here. I don't know that Head Start has failed. If he means by Head Start failing, that Head Start-like programs like the Perry. >> Mm-hm. >> Didn't result in enduring increases in IQ, I can agree with that. If the claim is that it didn't have any benefit, then I would take issue with that, as the Perry results show. >> Really fascinating study. Okay, our last question is taking us back to unit five. And this about missing heritability. And the question is how about the non-additive effects. Could this explain the missing heritability? Has there been any research done on statistically significant interactions between snips and or copy number variance, variations or even on interactions of non-additive genetics and the non-shared environment? >> Okay. This is, who's is the question from? >> Marjaland. >> Oh. Marjaland. >> I apologize if I'm butchering that name. >> Okay. Okay. Well, I'm sure that's close. It's a good question. So the, to my mind, there's four major explanations that people are thinking about in terms of understanding the missing heritability. And maybe we should back up one and just state again for the students what missing heritability is. The missing heritability is the difference between the biometric estimates of heritability that come from twin studies and what is coming out of GWA studies. >> Mm-hm. >> You're right, there's a great disparity there. The GWA studies account for very small percent of the variance. Nonetheless the twin studies suggest that these phenotypes are moderately to highly heritable. The first hypothesis is the twin studies are wrong. That there's something about the twin studies that lead to an overestimate of their ability. >> Mm-hm. >> That may be the case and, and eventually I think we'll find out. If, if the twin studies are wrong, they're not only wrong for behavioral traits though, they're also wrong for physical traits and medical traits. I've given the student slide to show that missing heritability is not unique. That behavioral phenotypes is true generally, of all phenotypes. But that's one possibility. A second possibility is that we don't have large enough sample sizes to identify the effects on these snip arrays, the GWA arrays. And we need to get larger and larger sample sizes. Sure we have in some cases sample sizes of 100,000, but maybe we need a half million or a million, or two million to really identify the effects. That's certainly a reasonable possibility, it's almost certain to count for some of the missing heritability. How much I'm come back to in a second. The third possibility is non-additive genetic effects. >> Mm-hm. >> Which is what they're raising here. The difficulty with non-additivity is it's kind of mathematically right now intractable. If you take a million snips, that you've genotyped, which is, you know, routine in GWAs today And you're going to start asking about interactions among alleles at different loci. >> Hm. >> Just take every possible pair. Well, there's on the order of a million squared pairs. I mean, that's, the, the combinatorics make it completely unfeasible to. There, there are attempts using data mining techniques to do this but it's very, very hard. I think all, non-activity's almost certain to account for some. The question now is how best to get at that non-activity. And I think the approach the field is generally taking, is to say let's identify the variance that seem to have an effect using GWAs. That'll give us a more limited poll. And then we'll look for interactions among those variance. >> Mm-hm. >> So I would think in the next couple years, we'll begin to seeing, begin seeing studies of non-adrogenic variance of that type. With the phenotypes we're talking about and in a course like this. The fourth explanation we've al, already talked about on this course is that the, missing heritabit, heritability may owe to variance that aren't captured by GWAs platforms. >> Mm-hm. >> Rare variance and copy number variance. Those are almost certain to explain some variance. How much, we don't know. I want to go back to the second in the last couple minutes here, Bridget, because at least it allows me to introduce another methodology to, to students in, in the course. And that's a methodology that you're familiar with, in fact, we were just talking about it as we were coming across to the studio here. GCTA. And GCTA was developed by a group of researchers, but mostly Peter Fisher and his group from Queensland, Australia. And GTCA is an acronym for Genomide Complex Trade Analysis. And what I, I think the easiest way to think about GCTA is what is it's doing is, it's a mathematic, we don't need to go into the mathematics of it. But it's a mathematical way of estimating how much heritability you can account for the snip tube genotype, if you had an infinitely large sample. >> Mm-hm. >> If I kept increasing the sample, how much would I be able to account for by these variants? The way Fisher showed you can do this is by looking at genetic similarity among unrelated, quote unquote unrelated people. Now, it might be a sobering thought to some of the people in the audience here, but, but we're all genetically related, right? >> Mm-hm. >> Somewhere way back we have some, some common relatives. And that's, that information is with our genomes. So, what you can do is if you genotype a million snips, you can see how sim, some people are going to be more similar on those million snips than others. And this is a distribution. This is a slide from, actually, a study by Robert Plomin, who we've talked about. >> Mm. >> Already, in this course. An eminent behavioral geneticist at the University of London. And I just want to convey to the students that what he's measuring here. And he's setting up to do the analysis, this GCTA analysis. He's measuring the degree to which on it's about a million snips a genotype. How similar are unrelated people in his sample? Some are more similar. They're up at the right tail of the distribution than others. >> Mm-hm. >> The actual scale here is not important. What's important is that people differ as to how similar they are in the snips. >> Okay. >> Fisher takes advantage of that differential similarity on those snips. To use that to estimate heritability from the GWAs array, that heritability, which is called the GTCA heritability- >> Mm-hm. >> Is an estimate of how much you could account for by the genotypes snips if you had an infinitely large sample. The last slide I have is in this Plomin study the GCTA estimates versus the twin based estimates. So the biometric estimates. He happens to have a twin sample. And then the ratio of the two. And, and it shows it for cognitive ability but also for height and weight. And what you see is that roughly 40, 50, three quarters of the, of the overall heritability seems to be in principle, capturable by what's on those arrays. If you had a large enough sample. >> Mm-hm. >> There's still some missing. It's probably those other factors that are contributing to it. >> That's still much more than we're finding in [CROSSTALK] >> We're just samples. We need a lot of samples. >> We'll get there. >> Okay. >> All right, well that wraps up Unit Six Office Hours, and catch us next week for Unit Seven.
