2016 Personality Lecture 14: Openness and Intelligence

Okay, so today we're going to talk about openness. Now, you might say in some sense that openness is the last discovered of the big five traits, and it's also the one that comes up most weakly in the factor analysis. When Hans Eysenck formulated his trait representation, which was extraversion, neuroticism, and psychoticism, he felt that he encapsulated all of the fundamental variation in human personality, and he was loathe to include intelligence as a personality trait.

Now, the relationship between psychoticism and openness is rather complex. Eysenck basically felt that people could be extraverted, and that would be the positive dimension, and could be high in negative emotion, and that would be the negative emotion dimension, and that they could also have a predisposition to psychosis. He was thinking about the brain systems that underlie schizophrenia and that tendency to dissociate and develop hallucinations, but also maybe the systems in human beings that are affected by hallucinogenic drugs, which seem to affect human beings differently than they affect other creatures.

Eysenck's psychoticism eventually was broken down into low agreeableness and low conscientiousness; it didn't predict the predisposition to psychosis at all. And later, factor analysis found that you could load out intelligence and creativity as a personality dimension separately. Now, it's a bit tricky because you have IQ measures, and IQ measures are much more direct measures of intelligence than the personality measures. I mean, the personality measure is something like: do you think you're smart? You know, I'm being a bit flippant about that because it's more sophisticated than that, but when you talk about someone as a smart person, you're generally referring to the trait that we would describe as openness. You also are doing that when you refer to creativity as a predisposition.

And you can measure the smart part a lot better with IQ, and IQ is also a pretty good predictor of creativity. Well, so let's see how they're associated. You remember that you have your big two traits—stability and plasticity—and stability breaks down into conscientiousness, emotional stability, and agreeableness. And then you can see how each of the aspects are correlated—how highly they're correlated. So industriousness and orderliness at about .4, volatility and withdrawal in emotional stability—or neuroticism actually—about .6, and agreeableness and politeness at about .4.

And then the next trait is extraversion and openness—that makes plasticity—and people who are high in plasticity tend to be good entrepreneurs, by the way. And extraversion breaks into assertiveness and enthusiasm, correlated at about .5, and then openness and intelligence—the correlation there is lower—it's .35. The average correlation between the big 5 is .2, and the correlation between the big two is about .2. And so you have to remember, you might ask yourself, well is a correlation of .4 twice as large as a correlation of .2? And the answer to that is no because you have to square them to compare them arithmetically.

So a correlation of .2 gives you .04, 4% squared, and a correlation of .4 gives you 16% squared, so there's actually a difference of a factor of 4. And so you need to remember that when you're considering effect sizes so that you can understand exactly what they mean. Now, if you're accounting for 16% of the variance—which would be a correlation of .4—then that leaves 84% left unexplained. But that's a bit of an overstatement most of the time in psychology because our measures are rather imperfect.

And so, for example, if you give someone a personality test, and let's say that the validity of their self-report is about .6 or .7, which is probably what it is—maybe it might be a little bit higher than that—that means that their self-report is only picking up 50% of the variation in their personality. And if an R derived from that measure accounts for 16% of the variance, in some sense, it's twice as powerful as it appears to be because the measure is so error-ridden.

And a lot of the things that we measure in psychology are error-ridden, and what that means is that the correlations—generally what it means, although not invariably—is that the correlations are bigger than they look. In any case, the correlations that come up in personality research are certainly big enough to be meaningful, and I would say meaningful in terms of predicting major life outcomes and predicting important economic outcomes and that sort of thing.

One of the things you might ask yourself is, exactly what does intelligence measure? So we'll start with intelligence, and that would be equivalent to intellect, roughly speaking. So intellect in the big 5, the aspect seems to be associated with interest in ideas. Whereas the openness proper aspect seems to be associated with aesthetic sensibility and creativity. That's the best way to think about it, as far as we've been able to figure out.

And the biggest relationship is between intellect and intelligence, as measured psychometrically. Now, intelligence is an interesting concept. It's very controversial. In fact, one of the things that continually happens over and over again in psychology is that psychologists discover all sorts of new kinds of intelligence, and the biggest two proponents of that sort of thing, I suppose, were Bob Sternberg, who came up with this thing he called practical intelligence, and then Howard Gardner, who came up with the theory of multiple intelligences, which had a very large effect on educational psychology when the theory was first proposed.

And that's not saying much because, generally speaking, it's very difficult to find a discipline that's more susceptible to fads than educational psychology. As far as I can tell, generally speaking, each fad is worse than the previous one. So, Gardner posited that there were 7 or 8 different intelligences: linguistic, musical, logical, mathematical, spatial, body/kinaesthetic, intrapersonal, and interpersonal. And people talk about emotional intelligence, and they talk about practical intelligence.

As far as I'm concerned, all of that's complete rubbish, and there's technical reasons for that. I mean, there's technical and philosophical reasons for that. The first thing is, you can't just mess up a word. You know, the whole point of having a word is so that it defines some things, it describes some things, and doesn't describe other things, right. And so you can make the word intelligence account for whatever you want.

And so you can say that the ability to dance is a form of intelligence. But the problem with that is you blur out the word so badly, you can't tell what it means anymore. And so I would also point out: we had perfectly good words for those major intelligences. We called them talents. And so it's perfectly reasonable to make a distinction between a talent and intelligence.

Now you might say, well how the hell do you know the difference? If things look similar to some degree, then how do you know if they're the same or different? Well that's exactly what you do when you do the construct validity process. So let's say you rated a number of people on their dancing ability. And then you rated a number of people on their ability to multiply two two-digit numbers in their head quickly. Well then, you see technically, if both of those were intelligences, then the people who could dance better could multiply two-digit numbers faster in their head.

They would be slightly positively—coughs—see that's what happens when I don't have any diet coke. So anyways, if intelligence was the right word to subsume both of those, then what you'd see is there'd be a high correlation. The people who were good dancers would also be the ones who could multiply most rapidly—I mean multiply in their head. But you're not going to be able to extract out a single factor like that. It's just not the case that those things associate like that.

So for something to be one phenomena, the things that it—coughs—obviously I've talked too much this semester. All the things that are subsumed under that definition have to be correlated. And highly correlated. Because otherwise, they're not the same thing. It's the definition of the same thing. Now IQ is a very peculiar construct, a very unique construct from a psychological perspective.

But also, more generally, from a social science perspective because IQ has the most predictive validity of anything ever discovered in the social sciences, period. Now the other thing that is interesting about IQ is that a lot of the people who were interested early in the measure of intelligence were engineers turned psychologists. And the engineers have actually had quite a whopping effect on psychology.

Part of the reason that psychology has advanced methodology and advances statistical techniques is because we got invaded by engineers in the early part of the 20th century, and so the engineers made our measurements much more rigorous than they would have otherwise been, and it's been a huge advantage. I mean, one of the things that's happening to psychology is that it's starting to eat other disciplines. It's had a massive effect on economics and it's going to have a massive effect on political science.

And the reason for that is because we know how to measure things, and it actually turns out to matter that you know how to measure things. Obviously, I mean if you're going to be talking about something that's specific and well-defined, first of all, if you can't make it specific and well-defined, all you can really do is argue about it. And there are better and worse arguments, but the problem is that just because an argument is good, doesn't mean it's right.

And that's a big problem. It's also something to remember when you're arguing about people that you have a relationship with because if you happen to be better at structuring your arguments and you're more verbally fluent, and perhaps more assertive, and perhaps less agreeable, you're going to win the argument. But that doesn't mean you're right. Unfortunately it's not that simple.

And so as far as I can tell, the empirical process is restricting or eliminating the catastrophically detrimental effects of pure theory. Well, and our measurement techniques have been pretty good about that. Okay, so now what happens with IQ here? I can define IQ for you just so you know what it is. Imagine that you had a library of questions.

So here's some questions. There's millions of questions in this data bank. What's 2 times 68? What's the capital of Georgia? What's the definition of hypertrixanemia? What comes after in this sequence: 2, 4, 8, 16? Remember these numbers—so I'd tell you the numbers and you'd have to tell me them back—2, 4, 1, 3, 5, 7, 6, 12, 15, 14, 18, 20, 22. And then I might ask you, in fact, to remember them backwards, and that's a working memory test.

Okay, so those are all questions that require abstraction and mental operations; that's the key as far as I can tell. So you might say, exactly what constitutes an abstraction? And it looks to me like what an abstraction is—there's the capacity to formulate an abstraction and there's the capacity to manipulate them.

To formulate an abstraction, I think what you do is you take a phenomena and you represent it with something that's low resolution. That's more or less what an abstraction is. So, for example, if I imagine this classroom later, my imagined representation is going to be of much lower resolution than the actual classroom, but it's sufficient for certain purposes. Just like a thumbnail on a computer is sufficient for certain purposes.

And then once I simplify the phenomena in question, then I can also perform certain operations on it, say, in imagination or semantically, and the capacity to generate that abstraction and the ability to manipulate those seems to be the core element of intelligence. And I think part of the reason why intelligence, roughly speaking, is a human dimension—although obviously you can to some degree rank order animals in terms of their capacity to learn, but there's this whopping differentiation between human beings and animals in this area particularly—is because we can use abstractions.

Now, I heard a woman named Temple Grandin speak once and it was a very interesting talk. She's very autistic but very brilliant, which is not very common for people who are autistic because usually it's associated with a substantial amount of intellectual impairment. But she also had an extraordinarily devoted parent who spent almost all her time trying to pull her daughter out of her autistic isolation.

Now Grandin has gone on to become a University of Chicago professor—and she works in the agricultural area—and she has redesigned the chutes, the walkways basically, that take cattle to the slaughter house. And she's done that to make them much less stressful for the animals. And part of the reason she can do that, as far as she's concerned, is because she thinks like an animal.

And what that means, as far as she can tell, is that she cannot generate an abstraction—not easily. Now she can speak, so obviously she learned to do it to some degree, but she talked more about deficits in imagination or concretization of imagination. So one of the weird things that you see with autistic kids, not frequently but now and then—there's actually a famous example of this on the web.

There's a guy, I think he's from Britain, and you can take him in a helicopter and fly him over a city, and then he can stand in front of a piece of paper, like 12x8, start anywhere, and draw the entire city. Now and then you see autistic savants who are children who can draw perfectly realistically, and so maybe they'll draw a horse and they don't even start the horse like a kid would—like a normal kid would sketch the outline of the horse and put some eyes in. They put some eyes in and make the horse perfectly detailed.

And what Grandin claimed was that—she said okay, you think of "house" or think of "church." And she says, what comes to mind if you think of church? And maybe she'd say, well you get this little image of something that's sort of shaped like this with a steeped on top or a cross or something like that, and that's a church. And so maybe you think of house and it's like a little kid's house. It's got a rectangle and a parallelogram— that's not a parallelogram, whatever the hell you call that thing on the top—and then it's got a chimney with smoke coming out of it, right. Two windows and a door. House. Yeah, but there are no houses like that. Each individual house looks a lot different than that.

And so in order to have that conception of house, essentially what you have to do is you have to take that class of all possible houses and extract out the features that are common across all of them. And that's an abstraction. And so, as far as Grandin's concerned, she doesn't have abstractions like that. When you ask her to think of "church," she remembers "a church." It's always singular instances.

Now the thing about human beings is that we perceive singular instances to some degree, although I think our abstraction capacity often interferes with our ability to do that. But we also perceive abstractions. And you can kind of tell this if you're ever trying to learn to draw. Like if I say, draw a hand. You guys, most of you unless you have artistic training, you're going to draw what's essentially a hieroglyphic hand.

I think the drawings that children come up with aren't drawings; I think they're hieroglyphs. They're abstracted representations that are quasi-linguistic. They're not artwork, they're proto-language. And hieroglyphs, of course, were one of the earliest forms of written language. It makes sense, because if you want to communicate about something, why not draw an image? And eventually, the image turns into a word.

But I actually think that's the developmental progression of language among human beings. It's the abstracted image first and then it's the word. Anyways, if you drew a hand, you'd likely draw a hieroglyphic hand: 5 fingers or maybe even 4, right, because that's what animators do. You can't fit all those sausages on one circle. So they just simplify it down to 4 and no one even cares.

You watch an animated program, which is highly abstracted—I mean Homer Simpson looks nothing like an actual person. Like really tremendously unlike an actual person, and it's completely irrelevant. After two or three seconds it doesn't matter at all. Which is also why I think 3D movies—it's never a technology that's ever caught on. Because after the first 4 seconds, who the hell cares.

It's irrelevant whether you see in three dimensions because you don't need much information to actually structure your perception. Now if you want to see your hand—weirdly enough—so that you can draw it, you have to snap out of your normal mode of perception. I find often it's a lot easier to do that with one eye closed because it flattens it out. And then you have to stop looking at your hand as if it's a hand.

You have to look at it again like it's an object. And as soon as you do that, it's actually rather shocking because it's such a strange looking thing. If you put your hand in a non-canonical position gestures—that's a canonical position, I would say—if you put your hand in a non-canonical position and then look at it, it's the weirdest shaped thing. It's quite shocking to see it, you know.

And then you can draw it—once you can actually see it like that—but before that, you're pretty much stuck in "hand." So anyways, I think what happens with intelligence is that people are capable of abstracting and then they can manipulate abstractions. And so let's take another look at what abstractions might be. Now part of the reason that I think abstraction is so useful is because things are really complicated and it's going to be a lot more difficult to manipulate the actual thing than to manipulate some representative abstraction of it.

So if you could get an abstraction that captures the gist of the entity—now how we do that is not exactly obvious—but if you can, then you can expend a lot more resources concentrating on the relevant parts of the entity and a lot less time concentrating on the irrelevant parts. So for example if you ask a child to draw a house, what colour the house is is irrelevant.

Now why it's irrelevant is not exactly obvious. I guess it's because colour can vary across houses. And colour varies across all sorts of other things, and so colour turns out to be a non-canonical element of "house" and so you can just dispense with it when you're thinking about a house. You'd say, well it doesn't matter what colour the house is—it's still a house. So what exactly defines those features that make something a house rather than something else are not that easy to figure out.

You could say, well, a house has four walls and a roof. And those are the external walls, and I guess you could identify the minimal necessary components, but even that's not so simple. And you could say a house is also something that someone lives in, which allows you to put caves as houses because caves obviously share very little in common with the standard house.

My point is that it's difficult exactly to figure out what set of features we do use to determine whether something is a member of a class. There's this guy named Gigerenzer who’s thought about this a lot, and he thinks that they're a combination of objective and functional, basically. And that enables us to do things like come up with a category of "things you would take from your apartment if it was on fire."

Which is a really interesting category because most of the things that you're going to take bear very little resemblance—it’s hard to see why they make a category. Pets, family members—probably in that order—photo albums, identification, what else might you take? Your phone, I suppose. What else would you take out of an apartment during a fire? Well you guys would be out on the street stark naked, obviously.

Well, you get the point that those things that I just mentioned seem to share very little in terms of objective features in common, but there's some reason that they still cohere as a category. Alright, so why do you need to abstract? Well, because you can handle the complexity of the world better. Even your perception is a form of abstraction, you know, because when you look at someone, a lot of the abstraction is done for you because you just can't perceive very well.

But when you look at someone—first of all, you only see the surface of them that's facing you and not the other surface, so that's a big limitation. You only see them now in this split second, rather than extended across time. You don't see any of their microstructure beyond the general pattern of the physiology at this level of resolution.

You don't see anything beyond that. I can't see your cells or your organs or anything like that, which is actually a real problem often for us. And I can't see the systems that you're part of. So right off the bat, just to perceive gives you a low resolution representation, and a lot of that is limitations of your processing. Your eyes just can't zoom in that far. You don't have the cortical capacity to improve your vision.

Plus, how much do you really want to see at one time? No more than necessary, whatever that means. And evolution seems to take care of that problem. So your perception abstracts for you right away, but then I would say maybe you make a leap from perception to conception, something like that. And so that would be the leap from the perception of the actual object to the image of the object, and then a leap from the image of the object to the word.

So it looks to me that a word is a representation of an image which is a representation of an object or a situation—it's something like that. And you need those because you have to deal with a complex world. So here's one way of thinking about it. Look at the top left corner of that diagram. I call that the thing in itself, and what it is is an array of dots, roughly speaking, and—I think there's 350 dots there—and you might say, well—and then if it was a real world object you could imagine that would be in three dimensions, not two because that's two, and it would be transformed because it would be also extended across time.

And it would have microstructures that were far smaller than that and macrostructures that were far bigger, so even that is an abstraction, but I'm using a fairly complex abstraction to represent the thing in itself. That's the thing you cannot perceive, right. What something actually is. You just get an image. And then the next five things are ways you could look at that.

So you could say, well, it's a rectangle. You could say it's four rectangles, and forget about the dots. You could say that it's—I think that's 16 lines or, what is it, 32 lines? 24 lines. So if that thing in itself was an orchard and you were going to walk south to north through it, that would be a good representation. Those would be corridors, roughly speaking, and then object four is a combination of object one and object two.

And then object five is, well, all that's blurred out are the little groups of six dots. And so you might say, well what is that thing? And then the answer is, well it's any of those five things. And so that's an interesting way of thinking about it too because one of the things it shows you is that you can disagree about what something is right at the level of perception.

And then you might say, well which of those is the best representation? And the first answer might be, well, the most complex one: the one in the left-hand corner. But that's not necessarily the case—it might depend on what you want the representation for. Like imagine—it's weird, it's weird—because imagine you have a map, okay. And so you think, well I'm going to make the best map in the world, and so you make a map that's exactly the same size as the territory that it represents and has all the detail.

Well that's not any use at all, that thing, because you could just use the territory, right? Just because, as a map gets increasingly detailed, to some degree, it's increasingly useless as a map. And so what you need is this weird combination between accuracy and simplicity. And so you want your map to be no more complicated than is necessary than to take you from point A to point B, that's it.

Because otherwise, it wouldn't be simple to use and the additional complexity would limit the utility of the map. So you might say that what we're trying to do with our intelligence is to get what we need with the least amount of effort possible. And there's a sort of Piagetian idea that's lurking behind that, which is: you have limited resources, you don't want to expend more energy than is necessary, and you want to build representations of the world that suffice to keep you going now and then next week and next month, and then along and then in groups.

And so you build a representation of the world that has some concordance of the thing in itself, but has this functional simplicity so that you can actually use it in the real world. Alright, so back to this diagram. So I've got the object and five representations, and then there are ways of representing it linguistically at the bottom. What that is—the words are basically a simplification of something that's already simplified.

So intelligence seems to have something to do with the process of functional simplification. Now, back to the library of questions. So you have this database of questions and so then what is you is pull out a hundred questions, randomly. Alright, and you want to pick out a hundred because if you just pick out one—like if I asked you what the capital of Georgia was and I asked you what the capital of Georgia was, and you got it wrong and you got it right, I couldn't really say with any degree of certainty that that means that you have more intelligence.

Because the probability that I'd get an accurate measure with one question is quite low, but if I used three questions, well then, I would start to become—my measurement would start to become less error-ridden, and if I used 10 it would even be better, and if I used 20… For a personality questionnaire, the rule of thumb—and this is just a rule of thumb—is that, if you don't have 20 questions in the questionnaire, then you don't have enough questions to really generate a reliable measure.

So you'll still see single item measures used in psychology, but you can add additional accuracy by increasing the number of items. I designed an IQ test a while back, and we got a pretty good reliability and pretty good validity. Reliability is, you know, you test the same set of people twice, and the rank order of their performance stays the same. So if you got 90 on the previous test, you'd get 90, and if you got 95, because you're more intelligent in this example, if you got 95 on the last one you'd get 95 on this one. So that's reliability.

We could get a pretty reliable IQ test with nine items, but it was substantially improved with 17. And with increments, you get a, what would you call it, diminishing returns. So a hundred is plenty. Okay, so you pull out a hundred questions and then you take 100 people and you gave them those questions, and then you score them—right and wrong, that's all there is—and then you sum the scores.

Okay, so basically, the sum behaves like the average—let's say you average the scores just to make the argument a little bit more straightforward. You average the scores, and so each of you gets the average calculated across the times that you were right and wrong, so it's an approximation of the frequency of how right you were, and then we rank order you: 1 to 100. That's IQ. That's all there is to it.

Now people say IQ doesn't exist. Well, that's annoying because all they're doing is gerrymandering the definition of the word exist. You know, because you can do that: if you ask if A is equivalent to B, the answer is, it depends on what you mean by A and what you mean by B. It's almost always a foolish question because you can take B and A and bend them and twist them in such a way that you can make one thing another thing without too much problem.

People usually get around that by only using a word in a context, right—in a sentence or a phrase or a paragraph that defines the word so that you can't weasel around like that. But more specifically in psychology, we have methods to tell if something is real, and if IQ doesn't pass the test of reality for a psychological measure, then no other psychological measures pass the test because they're validated exactly the same way IQ is, and they don't work as well.

So you can say, well no, that's not real, but then you have to throw out—you probably have to throw out the social sciences completely insofar as they're actually sciences. Because the same statistics are used to generate their findings. So you can't just say, well this really powerful thing doesn't exist while a bunch of these weak things exist and they're measured the same way. You don't get to play that game.

Okay, so does it exist? Well, you can measure it reliably, so that's definitely something. You know, because you might say, well does your shadow exist? Well, generally speaking under normal conditions you could measure your shadow relatively reliably. Does it exist? Well, it passes one test of existence. You can't detect it with a number of senses—you can't touch it—for example, but whether or not something exists is actually a rather tricky question.

But you can measure it, and it's useful, and that's not a bad definition of "exist." So, what's it good for? Well, it predicts major life outcomes, and importantly. And we can talk about the power of intelligence momentarily. If you were actually going to calculate an IQ score per se, you would take the rank order that I just described, and you would correct if for age, because you can't really expect a 7 year old to know as much as a 15 year old or a 20 year old.

So generally speaking, if you take a technical IQ test, it will rank order you, but it will only rank order you with people the same age as you. So that's just another, you know, it's an additional correction; whether or not that actually gives you a more accurate estimate of IQ depends on what you mean by accurate. A very bright 7 year old generally doesn't know enough to do much useful in the way of work, but they can learn extremely quickly, and so if you compare 7 year olds to people of other ages purely in terms of their score on an IQ test, there's a high probability that you would underestimate their speed of learning.

By the same token though, by the time you're 40 you have a lot of accumulated knowledge, and so if you only measured learning speed, then you would overestimate the 7 year old's advantage over the 40 year old. IQ is basically a unitary phenomena, but you can fractionate it into...it's very very complicated, for reasons that I'll tell you about in a minute. If you had to divide IQ into two, the way you would divide it would be into crystallized and fluid intelligence.

And fluid intelligence would be, roughly, the ability to learn. And crystallized intelligence would be, roughly, a measure of how much you've learned. Now those are separate to some degree, and here's why they're separate: as you age, your capacity to learn decreases. It's quite ugly. So by the time you're about 24, you're starting to get stupider, from the perspective of being able to learn. And that's just pretty much a linear trip downhill until you depart the world.

So it's nasty to see because it's quite a steep generation. You can stave that off, by the way, with cardiovascular exercise and weightlifting—so that's the best way to keep your learning capacity intact. It's not to do cognitive exercises; it's to do physical exercises. And then as you age, the amount you know goes up, and what happens is those two things actually tend to average out.

Now you might say, well if I'm claiming that IQ is a unitary thing, how can I make the claim that you can divide it into two elements and that those elements are in some sense separate. And I think the reason is for this: your prefrontal cortex, roughly speaking—so that's the part of your brain here forward—you can think about in part, it's the part that does the programming of motor activity—voluntary motor activity—but it also seems to be...so let's say you're doing something new and you're pretty clunky and awkward at it, and then you do it more and more and you get smoother and smoother and smoother at it, and you automatize it.

What happens is you're using your prefrontal cortex and, to some degree, your whole right hemisphere, as you're first doing it, and then as you get better and better at it, the ability shifts to the left and then goes backwards. So what's happening, to some degree, is your prefrontal cortex is programming the rest of your brain. And you can think of fluid intelligence as a measure of the ability of your brain to program itself, and crystallized ability as a measure of how much programming is there.

So they're separate in that manner because they perform separate functions, but then you can also understand that the better your programmer is, the better your programming is going to be so that inflates the correlation. And so you can have to separate things functionally—that are even dissociable in terms of age—but because they work so tightly together—it's like, how fast can you dig a hole and how big is the pile of dirt? Well, obviously those two things are different in some important sense, but they're going to be highly correlated because the faster you can dig a hole, the bigger the pile of dirt.

So anyways, crystallized intelligence is also often measured as verbal intelligence, but the better measure fundamentally is crystallized intelligence. Now, you know how we looked at the personality hierarchy, right—that was this gestures to slide. And what you see is three strata, roughly speaking: aspect, trait, superfactor. And it looks like—the superfactors, I would say, the validity of those superfactors is not yet clear.

You can get a lot of useful prediction by looking at the traits, and there's more and more evidence—there's about 400 studies now using the aspects—there's more and more evidence that you can also derive additional useful predictive validity from differentiating the personality down to the aspect level. So that's the highest resolution personality level we have.

There are people who claim that there's something at the apex which you might call "good personality," but I don't think so. I think it's—I don't think so. You'd have to read the literature to understand the dispute—it's a statistical dispute, roughly speaking—but we've found in our research that stability and plasticity are only correlated at 0.2. They're pretty separate. So I'm going to stick with that for now even though it might be wrong.

Okay, so if you read the cognitive literature and the neuropsychological literature, you will find that neuropsychologists tend to assume that there are forms of cognitive ability that you cannot describe as IQ. And they make that claim because they'll take a test and they'll use the test to measure something, and then they'll correlate that with IQ, and the correlation will only be about 0.2 or 0.3.

And they'll say, well that's not a high enough correlation to justify saying that those things are identical—because the correlation should be more like 0.8 for them to be identical. But the problem is, let's say I had a measure of your IQ and your IQ and your IQ and your IQ, and then I asked you two questions, and you and you and you, and I summed your answers, whether they correct or incorrect.

Now what would happen is, there'd be some correlation between your total score on those two questions and your overall IQ. Maybe it would be 0.3. But that wouldn't justify me in claiming that the two-question IQ test is different from the full-scale IQ test—it's just worse. And you're going to run into this problem when you read psychological literature, and if you ever do any research because it's very difficult to determine whether you get what's called "discriminant validity," which is, this thing measures this thing, and this thing measures this thing, and they're actually different.

Or whether you just have a lousy measure of this and a lousy measure of this and so the correlation is low. Now here's a neuropsychological test, and you can make the case that this is measuring something different than IQ. So imagine that you see an array of 12 pictures—they're just drawings, drawings of a telephone and a booth and a chair—just common objects. So you see 12 of them. And then you have to click one, and then when you do that, they move.

And then the next time you have to click one you haven't already clicked. And then they move again and you click, and then they move again and you click. And you see by about the time that you get to about 8 of them, you're starting to run into trouble because it's difficult to remember more than 8 abstractions simultaneously—7 actually, and some people claim it's as low as 4—that's the capacity of working memory.

Which is why we kind of like phone numbers to be 7 digits long. A lot of the things we want to just remember are 7 units long, and if we want to remember more than that, we chunk. Although if I show you some dots draws dots how many dots? 3. You don't have to count them. You can see 3 dots. draws more dots You can see 4 dots. It's not obvious that you can see 5 dots. At that point you start to go "3 and 2."

And then if there's 6 dots, well that's how they would be arranged on a dice, right. The reason they'd be arranged on a dice that way is because you can see two sets of three and infer 6. You know, if they're sort of randomly dispersed like that, and I say well how many dots is that, well you have to count them. And what that shows you is that your working memory—which is really a function of fluid intelligence— is a very very narrow thing.

And it's like your vision. You know, remember we saw the gorilla video, right, and you can only pick up the thing that you're attending to. Well, working memory is like that, it's this narrow narrow form of attention and it sort of specifies unitary phenomena—I mean that's not quite unitary but it's close enough.

And I think the reason it does that—it's like, why's your attention so narrow—is because often what you're trying to do as you work in the world is to make binary choices. You know, you don't want to choose between 20 things, you want to choose between two things, because then when you choose between two things you can actually act. If it's still two things you can't act, if it's one thing you can act.

So imagine you go into a Chinese restaurant is a good example because Chinese restaurants virtually always have like 400 items on the menu. And you might ask yourself, well how the hell do you figure out what you want? And what you don't do is go through every single menu item and contemplate it because then you'd be there for a week. So what you seem to do is eradicate chunks.

So you might say, well I'm not in the mood for anything but fish. Okay, bang, three-quarters of the menu disappears. You know, and then you might go, well I don't want it breaded. Whack, and then another half is gone and now you're down to maybe 8 choices. And, you know, you reduce that to 4 and then to 2 perhaps. Well, I'd like this or I'd like that. But the waiter doesn't care about that. The waiter wants to know which one you want because until it's reduced to one you don't get to act.

And so the reduction, the reduced narrow strip that we use to perceive the world in partly is a consequence of processing inadequacy. But it's also partly because we really are trying to reduce things all the time to 0 and 1, yes and no. Okay, so back to the three-stratum theory of cognitive abilities. Now, you've got this array of 12 items and it keeps shifting.

And what I would do is score you by the number of duplications—if you pick the same item more than once, you lose a point. So the maximum points you can get is 12 and, you know, people usually make 3 or 4 mistakes. And so I can take your score on that test—maybe I'd give it to you 2 or 3 times just to make sure I have a reasonably stable indicator—and then I could correlate it with your full-scale fluid IQ.

And I'd get a correlation of about 0.25 or 0.3. Now, if you don't know anything about psychometrics you'd say, well that's obviously measuring something different than full-scale IQ because it only correlates at 0.25 or 0.3. But what it turns out is, no: it's more analogous to the problem that we were just talking about where I asked you three questions and tried to derive your entire IQ from that. It's just a bad measurement.

Now what happens is if you take the neuropsychological tests that people have developed, that theoretically assess the different parts of the brain, and you factor analyze them and you extract out a single factor, the single factor looks exactly like fluid intelligence. Now, is it exactly the same as fluid intelligence? The onus would be on the person who claims it's not to demonstrate the difference.

We did give university students a very extensive battery of pre-frontal-cortical tests and measured their IQ, and then we used the pre-frontal-cortical test average and the IQ to predict grades. And what we did find was the pre-frontal tests predicted over and above the IQ. But we thought maybe, if you took the pre-frontal average and IQ and you added them to predict grades, say—that's a regression analysis—they both added in.

So they predicted different parts of the variation. It added additional necessary information. However, if you took all the neuropsychological tests and all the individual IQ tests and you put them in one data set and then extracted out a single factor—so that would be fluid intelligence, roughly speaking—that predicted grades very well. And then what was left over didn't predict at all.

So what we concluded was that neuropsychological tests—pre-frontal-cortical tests—test intelligence, and they test a slightly different thing than fluid intelligence tests do, but that isn't because they're testing something different, it's because fluid intelligence tests, the way they're currently constituted—don't sample the universe of potential questions as well as they should. So it wasn't that we discovered something different; we discovered a measurement error, roughly speaking. That only took 5 years, by the way.

It took 5 years to figure that out. It was very annoying because when we started I was trying to predict grades, you know, and I didn't know the IQ literature very well at that point, because psychologists don't usually teach it very much because IQ bothers the hell out of everybody. And so, and the neuropsychologists who have their own field are motivated to make the claim that they've discovered something new, and I just took that on face value.

But then we tried to predict—like okay if I'm going to predict how well you do at your job, well I want to get a good cognitive measure because if it's a complex job it keeps changing and you have to learn to keep up, and if you can't do that, you're not going to do very well. So that would be bad for you and it would be bad for everyone around you. And so we were trying to pick up additional information that would enable the accuracy of the prediction to be increased, and looked at the neuropsych literature but, at most, it was an ambivalent success.

(00:49:25) Okay, so how does that relate to this? Well at the bottom tier, where it's the most differentiated, you have single tests. And they're only correlated with each other at about 0.2, maybe 0.3. So you might say, well, they're all measuring different things—they're not. They're all measuring the same thing badly. And then you might clump them. So let's say I ask you—oh god—maybe I give you a test that asks you—that tested your vocabulary with 6-letter words, and then I gave you another test that tested your vocabulary with 7-letter words, and then another one that tested your vocabulary with 8-letter words.

And I said, well those are all different tests. Well, they would correlate quite highly—not perfectly—but I could clump them together and make one sort of supertest out of them. And then I could take supertest A and supertest B and supertest C and look at the correlations between those, and those would climb to about 0.5 or 0.6. And then if I clumped the supertests together, which would be happening by the time we got to stratum 2 there, then the correlation between the tests would start to rise to about 0.7.

And then if I collapsed all the supertests into a single measure, then the correlation between any of those stratum 2 tests and G—because that would be the highest level thing—G is general intelligence, that's roughly speaking fluid intelligence—the correlation between any of the supertests and G would be about 0.8. So the point of all this—there's no way of discussing this sort of thing without doing it statistically and technically because they are statistical and technical processes.

So what happens is that, now let's say you have a measure of fluid intelligence. Okay, you've got this measure. And you take one of the stratum 2 sets of tests and you use both to predict academic achievement. What happens is that the superfactor, G, kills the rest of them. You're just not going to be able to find any of the tests in the bottom strata that are going to add to the prediction of what the general factor can give you.

Now with personality, that's just not the case. You can't pull out a single factor that predicts how personality is going to influence job performance across your life. If you add neuroticism—if you take a single factor and you add neuroticism, neuroticism is going to kick the single factor out. Conscientiousness is going to kick the single factor out, and so forth. Because there isn't a single factor.

It's not—the tests aren't highly correlated enough to say that they're all measuring the same thing to some degree. The personality traits really are different. You know, you can clump them into plasticity and stability, but even there you're maintaining a lot of validity. So if I took stability and used conscientiousness to predict job performance, conscientiousness would wipe out stability—it'd be a better measure. But you never get that—you virtually never get that with the lower tests in the IQ strata.

Okay, now mumbling. That's a good representation of how these things are associated. So at the bottom, you have the single tests and then they chunk into, say, sequential reasoning, quantitative reasoning, and induction on the fluid intelligence side, and general knowledge and language development on the crystallized intelligence side—that's the red—and then they're all subsumed underneath fluid intelligence, which is the measure that you really want to have.

Okay, so what good is that? Well, it basically correlates 0.5 with major life outcomes. And that's 25% of the variance, and maybe you can get it a bit higher than that, but we'll just stick with 0.5 as a representation. And so, let's say you take a bunch of students, 100 of them, and you want to sort them into those who will pass the class with a 50% failure rate. So you just pick them randomly—you're going to pass, you're going to fail, you're going to pass, you're going to fail.

So that's no knowledge. I have no predictive knowledge whatsoever. So my validity—the validity of my classification is basically going to be equivalent to chance: zero. Right, I'm only going to have a 50% chance of picking who passes and who fails. If I used IQs—if I had all your IQ scores—and I did the prediction, I'd get it right 75% of the time. So it's a huge difference. It cuts the failure rate in half by using IQ.

And IQ is a very good predictor of health; it's a predictor of longevity; it's a predictor of resistance to post-traumatic stress disorder; it's a predictor of—obviously—occupational status; it's a prediction of educational success; it's a predictor of income. It's a very powerful predictor. And here's another way of conceptualizing its effects. So imagine you could choose how you were going to be when you were born—this is in North America because it's going to vary by society, at least to some degree. You get to be born into a family that's at the 95th percentile for wealth, or you get to be born at the 95th percentile for intelligence.

Who's better off at the age of 40? And the answer is the person who picks being born with an IQ in the 95th percentile at birth. It's a more powerful predictor of long-term life outcome than familial wealth. And it shows up everywhere. One of the things we did recently was look at disgust sensitivity—we're going to talk a little bit about that when we talk about conscientiousness because orderly people seem to be more sensitive to disgust than disorderly people.

That seems to be why they’re orderly. But, the higher you are in IQ, the less disgust sensitive you are. Now we don't know exactly why that is. Maybe it's because, you know, maybe you could make the inference that IQ is related in some way to the physiological integrity of the cortex rather than the limbic system, which is the source of, say, emotions and motivations.

And the more powerful it is, the more inhibitory capacity it has over the more fundamental motivations and emotions. You could make that case. The problem is that intelligent people don't necessarily seem to be any less impulsive, so...and you can have a pretty vicious personality disorder that's characterized by extremely disorganized behaviour and a complete inability to put long-term plans into operation, and still have a high IQ.

So, one of the things we really can't figure out—it's just a hell of a thing to try to figure out. Like the relationship between IQ and industriousness is zero. And that just makes no sense to me because most of the brain models are predicated on the idea that your ability to engage in long-term planning is a factor that's associated with intelligence. But then industriousness people seem to not only engage in long-term planning, they seem to do it right so that if they're more industrious and they put their plans into operation, then the plans seem to actually work.

But it's not correlated with IQ. So then I can't figure out, well obviously the industrious person, in some way, is able to regulate their own behaviour. You know, well they're not procrastinating, for example. And you'd think that that—the ability to not procrastinate—would be a cognitive feature, but it doesn't seem to be. And we have no idea what makes people industrious, and we can't figure it out. So it's this incredibly potent predictor—it's just about as powerful as IQ—and we have no idea what it is.

So if you think you have a smart idea about that and you want to pursue it, feel free because there's a big mystery there that no one's been able to crack. And we've been at it for a long time and have had almost no success. You know, we had people do things like, we'd give them sentences of n’s, m's and u's—sort of randomly distributed—and then we'd have them count the u's. You know, like a whole page of sentences—they’re not sentences, they're just strings of letters. Count the u's—how useless! You'd think that someone industrious would do that better.

They don't. That's an IQ test. The people who can count the u's faster have higher IQs. Almost everything that you would do where it has to do with manipulations of abstractions of any sort—even something that basic—seems to be fundamentally associated with IQ. People with higher IQs have slightly bigger heads, if you control for body size. They have slightly bigger brains, if you control for body size.

The axons on their neurons are a bit thicker, so the electrical messages seem to travel a little bit more efficiently. They are slightly faster in simple reflex tests. So it goes right down to the level of...because a simple reflex—a light goes on, you push a button—there's not many neurons mediating that response. You know, chains of neurons. There's only a few neurons communicating so that you can do that.

But even at that relatively simple processing level, IQ is associated with speed. So there's a physiological component. You're less likely to develop Alzheimer's disease if you have a high IQ. And maybe that's just because your brain is more robust. You know, so you could sustain damage, say, of up to 50% of your brain and you wouldn't even show it if you have a sufficiently high IQ.

Whereas if you're on the bottom end of the IQ distribution, you're much more—you seem to be much more susceptible to physiological damage. Nutrition is a big predictor of IQ variation. And a lot of that's been hammered out of modern societies. You know, so 150 years ago people's IQ was pretty tightly associated with their nutritional status, but now there are very few people in North America who don't have enough to eat, even though there's some variation in the quality of diet.

And that's flattened out a lot of the cultural variability in intelligence. Okay, so this guy named Hemphill did this paper—an American psychologist a while back—and one of the things he was interested in doing—remember we talked about effect sizes, right. How big is an effect in psychology or in social sciences in general? And people have used rules of thumb that 0.5 was a moderate effect and 0.3 was a minimal effect and 0.2 was starting to diminish to the point of non-utility.

But what Hemphill did instead of guessing, because that was just a guess, was actually go and look at papers that were published in psychology to see what effect sizes were being reported. So it's an empirical analysis of how big the effects that psychologists and other social scientists report. And so what he found was that the highest third of papers or the third of papers that had the biggest effects reported effects from 0.35 to 0.78, the middle third from 0.21 to 0.33, and the lower third from 0.02 to 0.21.

mumbling If you look on the left side here you see an r of less than 0.15 characterizes the first quarter of published studies. An r of 0.15 to 0.35, the next quartile. So 50% of psychological studies report an r of less than 0.35. 90% report an r of less than 0.5, and only 7 to 8% report an r of 0.5 or above. So what that means is that the relationship between IQ and, say, academic performance or long-term life income is higher than that reported in all but about 5% of psychology studies.

Okay so here's an interesting—you might say, again, in order to determine that something exists, you have to be able to measure it and then it has to be good for something. So it has to be grippable and it has to be useable as a tool, we'll say for the sake of argument. What sort of IQ do you need for what kind of job? This was derived from a company called Wonderlic, and Wonderlic does employment testing. Psychologists usually don't use the Wonderlic test.

It's a relatively short IQ test, and it's a good test. I think psychologists don't use it because it was developed commercially, but we tested the Wonderlic out a lot and it has excellent psychometric properties. It's a very good IQ test. It probably errs a little bit in testing crystallized IQ rather than fluid IQ, but we won't argue about that for the time being. Alright so, if you have an IQ from 116 to 130, that puts you in the 86th to 95th percentile.

Right, so that means you're minimally—if there were 8 people in a room, on average you'd be the smartest person. I should tell you, just so you know, that if you're a state college student—we'll use the US as an example because there's more differentiation between the American universities in terms of quality than there is the Canadian universities—if you're in an average American post-secondary institution, your IQ is 115.

And so, you think, that's not all that bright. But that's brighter than 7 out of 8 people in the general population. And so there are as many people—if you look at the North American distribution...so you have the 50th percentile in the middle of the normal distribution, you have the 85th percentile here, and you have the 15th percentile here. So about 70% of people fall between the 15th and the 85th percentile.

At the 15th percentile, you're not really very literate. And so what that means is that in the United States there's just as many people who aren't literate as there are people who are in college. And what I would mean by literate is not so much—like here's a bare bones definition of literate: you can follow written instructions. And so what that means is it's kind of a two-fold issue, right. You can make out the words and the phrases and the sentences, but then you can translate them into actual action—you can use them as information.

And at 85, you're going to have not a very good time doing that. And so, when you're wondering things about the way society is structured and why people don't necessarily make the most intellectually sophisticated choices, one thing that's very much worth remembering is that there's just as many people who are functionally illiterate in North America as there are people who are in tier-one colleges.

It's quite shocking. Anyways, okay so you want to be an attorney, a research analyst, an editor, an advertising manager, a chemist, an engineer, an executive, a systems analyst, an auditor—so you're up in the professional levels there. You need an IQ of at least 116. And so 130 would be better. My suspicions are that the average IQ of people in this room is about 120 to 125.

It might be a little higher than that. It's hard to measure IQ in the U of T population because a lot of people have English as a second language and so that makes it harder to measure crystallized intelligence obviously using an English language test. But we've found, when we've done our studies at the U of T, that the average exceeds 120. So, you know, so you guys have the cognitive power to basically pursue professional level careers.

From 110 to 115: manager, supervisor, programmer, teacher, general manager, purchasing agent, registered nurse, sales account executive. These are actually empirically derived, by the way. So the Wonderlic company has tested a very large number of people. And so these are the actual averages of the professions that they are describing—it's not hypothetical.

103 to 108, so that's the 60 to the 70th percentile: clerk, customer service rep, computer operator, medical debt collector, secretary, accounting clerk, general sales, telephone sales, assistant manager, credit clerk, drafter, designer, bookkeeper. There's a lot of white-collar, like low level, entry-level, white-collar jobs in that particular category. So that would be the brighter people among the high school students who didn't go to college and university.

IQ, this is dead average, 50th to the 55th percentile: police officer, receptionist, cashier, general clerical staff, inside sales clerk, meter reader, printer, teller, data entry, electrical, helper, dispatcher, general office. So that's right at 100. 95 to 98, so that's the 42nd to the 45th percentile: quality control checker, claims clerk, security guard, or unskilled labourer.

So look, the unskilled labourers are up at 95th or 98, so that's—at the unskilled labour level you still have an IQ that's higher than 40% of the population. Ark welder, die setter, mechanic, medical dental assistant. IQ 87 to 93: messenger, factory production worker, assembler, food service worker, nurse's aid, warehouseman, custodian, janitor, material handler, and packer.

Okay, so now what you see happening—it looks to me like you can conceptualize jobs most simply with a two by two matrix. Okay, so there's managerial/administrative jobs and creative jobs—that's the first one—and then there's low complexity/high complexity. And so, what you're starting to fall into at the 21st to the 37th percentile—so that's still one person in three—are low complexity, managerial jobs.

And so what I mean by that is, if you're in a low complexity job, imagine you're responsible for 30 things. But roughly speaking, it's the same 30 things all the time. And roughly speaking, they have to be taken care of the same way each time. So you have a finite domain of responsibility and you can master it.

And IQ will predict how fast you master it, but once you master it, it won't predict how well you do. What happens then, in all likelihood, is that conscientiousness starts to predict, or maybe emotional stability starts to predict, or some of the other personality factors start to become more important. So, but at 87 to 93, you're down at the level of custodian and janitor.

And then that's that. And that's not so good because you have the bottom 15-20% of the population who don't have a cognitive ability at that level. And so one of the things that you're going to hear a lot about as you get older is reasons for unemployment. And one of the prime reasons for unemployment in the future is increasingly going to be that there is just nothing for someone of that level of intelligence to do.

And this is partly why people hate IQ because you think, that is a nasty nasty thing to conclude. Let me tell you a story. So, I had a client at one point who I suspected he had a verbal IQ of about 85 and a non-verbal IQ that was lower. He had other psychological issues, but we'll just concentrate on that one for the time being. And he had a hard time finding a job.

He was socially anxious and that didn't help, but even if he could overcome that, it was very difficult for him to master the basic skills that were involved in what you would consider even relatively straightforward tasks. I spent a lot of time trying to find him a job, which was insanely difficult. He went to the government first to get help, and they told him to type up his resume on a computer and distribute it—it's like, well that wasn't very helpful because he didn't know how to use a computer and he wasn't going to type up his damn resume.

It was, "if you want to get a job, go out and do the things that it would take to get a job." Well yeah, if you can do that then you don't have a problem. And so the government agencies were completely useless, and they were staffed by people who would assume that if you didn't do what you were advised to, then you just didn't want to have a job.

Instead of ever thinking that maybe you couldn't do it. It's not the easy to use a computer. And then I tried to get him a volunteer job—well you can bloody well forget about that because it's harder to get a volunteer job than it is to get a real job. And the reason for that is that you have to step through a number of complex bureaucratic hoops, including having a police check.

And you know, first of all that's intimidating as hell for people, even if they haven't done anything wrong. And second, it's not that straightforward. It involves maneuvering through a complicated bureaucracy, and that's not the only step. There's all sorts of other steps. So volunteer work, that's out man. That's just not going to happen, not generally speaking.

And then I found him a job helping a guy out who had a bicycle shop/bookstore—kind of a strange combination—but it was a government subsidized temporary position. And then he did a good job at it—he was reshelving books and that sort of thing—and so he was reliable. He could do that. But, you know, the subsidy program ended and of course, if you run a bike shop/bookstore it's not like you have additional money to hire someone because that's a store that's just not going to be generating maybe any income, but certainly not enough to hire someone.

So that fell apart. Then I made a bunch of personal contacts with charity organizations to see if I could get him a volunteer position. I finally got him a volunteer position. And then I went with him to train at the volunteer position, and that was so enlightening. So what he had to do was, he had a stack of pieces of paper here and then a stack of envelopes here, and in principle the pieces of paper—which were letters—were in the same order as the envelopes were.

And so what he had to do was fold the pieces of paper three times and stuff the folded piece of paper into the envelope and then seal the envelope and then put it here. But it was a bit more complicated than that because sometimes the letters were in French and sometimes they were in English, so what that meant was that he had to watch and see that the French envelopes went into this stack and the English envelopes went into that stack.

So that's one degree of complexity, right. But then it also turned out that those damned envelopes had to be run through an automatic letter sorting machine. Now you might think, well that doesn't matter, but it does matter because those things have very tightly defined tolerances. And if you make a mistake—imagine you fold a piece of paper and the first time you fold it you made a mistake of an eighth of an inch. And then the next time you fold it you make another eighth of an inch mistake.

That means your piece of paper is now one quarter inch out of truth. And then you stuff it in the envelope and fold it, but then you crinkle the envelope. And so then the envelope gets stuck in the automatic processing machine. So I taught him probably 20 hours, how to fold this piece of paper into three. You know, and he was a bit shaky, and he just didn't have the fine motor ability or the cognitive ability to do that.

And then—I’ll end with this—when he finally couldn't do it anymore, what happened was the letters came with photographs attached to them. And then he had to fold the pieces of paper around the photographs in a way that didn't crumple the photographs so that they could still fit into the envelopes. And each photo was stapled, maybe half an inch different, per piece of paper.

So then he had to fold them properly with a different folding technique for every piece of paper, and he had to sort them into French and English, and then also if the envelopes and the papers ever got out of sync—which they did now and then—he had to figure out how to sort the envelopes and the pieces of paper to make sure that they were matched with the proper envelope. It was like, no. That was the end of it.

And so I'll tell you the rest of this story later. So anyways, the moral of the story is if you're smart, you're privileged and, thank god, make use of it. And if you're on the low end of the IQ distribution, man, you've got one tough life ahead of you. It has nothing to do with willingness to work, or virtually nothing. And it's a good thing to know, even though it's horrible.