Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode 11: Intelligence
18 Plays5 years ago
We talk about the history of research on intelligence. Is intelligence a real thing? What does it actually refer to, and can it be measured? Joe and Rolf discuss.
Recommended
Transcript
Introduction to Cognation Podcast
00:00:06
Speaker
This is Cognation, the podcast about cognitive psychology, neuroscience, philosophy, technology, the future of the human experience and other stuff we like. It's hosted by me, Joe Hardy. And by me, Rolf Nelson. Welcome to the show.
Exploring Intelligence: History and Definitions
00:00:23
Speaker
This week, we're going to be talking about intelligence. Intelligence, obviously, is a topic that is widely talked about and thought about in society and in psychology as well.
00:00:35
Speaker
going to be talking really a little bit about the history of the study of intelligence, how it's been defined and measured over time. And we're going to be using a chapter from Robert J. Sternberg as the basis or foundation for the discussion. Yeah, so this is more of a this is from the Handbook of Psychology written a couple years ago, but still gives a really nice overview of
00:01:03
Speaker
the concept of intelligence and how it's been treated over the years, especially in the early years of research into this area, which was in the early 1900s and sort of how the conceptions of intelligence have changed since then. The study of intelligence is something that's been around in a rigorous way for well over 140 years.
00:01:30
Speaker
which is interesting in the sense that it's like one of the oldest areas of experimental psychology with the sort of deepest and richest history and has been continuously controversial for the entire period of time. So it's an interesting topic of conversation, I think.
00:01:50
Speaker
Yeah, it gets and it gets super authority when you start attaching any sort of value associations with intelligence that might not necessarily be there with other psychological concepts. I think you can measure something like how much empathy someone has or or any kind of psychological variable. But when you start talking about intelligence, you get into
00:02:14
Speaker
something that is incredibly value-laden, I think, where people attach certain meanings to it. And it's obviously good to have more intelligence. So if you're measuring individual differences among people, you can get into some tricky territory. Absolutely, absolutely. And I think for me, though, rather than pointing out that intelligence is especially value-laden, which is totally true, absolutely true.
00:02:40
Speaker
I think what's more interesting to me is how much the study of intelligence and the history of the study of intelligence points out how value laden all of psychology can be. All right. I mean, so many of the things that we talk about in psychology are have some implicit value attached to them in different ways, whether it be, you know, as you say, a concept like empathy or a framework like depression or
00:03:06
Speaker
whenever we talk about different abilities or learning or even perception or biases in perception, all of these things are, in some sense, value laden implicitly, if not explicitly. And I think understanding that and appreciating that is, I think, helpful and important. Yeah, I think that's the absolute first place to start with this.
00:03:32
Speaker
Should we talk a little bit about the history of some of these concepts starting with some conceptions of this in the 1800s? Now, obviously, there's been people have thought about intelligence for a lot longer, but we're thinking of this in terms of the birth of experimental psychology and then the application of things like intelligence tests to various fields. Right.
00:04:02
Speaker
The idea of intelligence, I think, for me, in some sense, it's important and useful to start from the perspective of just what do we think about when we think about the word intelligence just as an English language speaker. Okay, so what do you think about? What do you think about when you think of intelligence? Well, for me, when I use the term intelligence colloquially, it's completely equivalent to smart.
00:04:30
Speaker
you know, smartness. So being smart and being intelligent are the same. That's the one-to-one, the same concept. And also just it's universally a good thing. You never want less intelligence. Right. I think maybe in terms of how the term is used in normal speech, intelligence might be
00:04:55
Speaker
sometimes used in a way that's a little bit slightly derogatory, just in the sense of maybe book learning versus smart is always good. Being smart is always good. Well, unless you're a smart aleck, I guess you could be a smart aleck or being too smart in that sense. But yeah, I guess both are generally good. They're not, they're used slightly differently, but they generally mean that you are able to
00:05:22
Speaker
say things and do things to advance your own position some way that are somehow clever, somehow solving some sort of problem or thinking quickly. Yeah, and I think there's, I think one, often when you think of intelligence, you kind of think of it as contrasted to wisdom.
00:05:50
Speaker
So intelligence just kind of being the raw horsepower, you've just got this intellectual smarts to follow whatever goals you choose and wisdom being something different than intelligence, where intelligence just kind of feels like brains and wisdom feels like judgment or good judgment to choose the right sorts of goals. Right, exactly.
Early Approaches to Measuring Intelligence
00:06:15
Speaker
And what's so interesting about it is
00:06:19
Speaker
We'll get into the history here, but right away, I think it's useful to bring up the idea that this is a cultural thing. This construct is a culturally specific construct, and the way that we use it in English in the Anglo world is very distinct and somehow different than, say, for example, in, say, Mandarin. There is no exact word in Mandarin for
00:06:49
Speaker
the concept of intelligence the way that you're talking about it. But rather, there are terms that are more akin to the idea of wisdom. And I think that's an interesting thing to consider as we start to talk about the history and also the value judgments associated with. It is contextualized in a more Western and contemporary culture that you wouldn't
00:07:11
Speaker
necessarily see intelligence talked about as a one of the highest values or one of the one of the things that people should be aspiring to or something that should be admired in the same way as it probably exists now. Right. And I think, you know, as we talk about things like
00:07:29
Speaker
So much now when we're talking about intelligence, we're talking about some secondary type of intelligence, whether it be artificial intelligence, emotional intelligence, interpersonal intelligence, these different kinds of uses of this term. We're building this conceptual structure, but the base or the foundation is not defined adequately.
00:07:56
Speaker
Yeah, and if it were, maybe advances in artificial intelligence would come a little easier. I think this is particularly relevant. I mean, as much as on this show, we talk about futuristic technologies and where fields are headed. I think intelligence is tricky in humans, and it is tricky to figure out how you would make it artificially, how you would
00:08:23
Speaker
how you would apply the way that humans think to machines. We've probably been enriched in understanding human intelligence by understanding how other kinds of artificial intelligence fails. Right. No, I mean, I think it's and I think, you know, understanding what is artificial intelligence or what would be defined as artificial intelligence is is challenged because we're constantly getting caught up in the discussion of
00:08:51
Speaker
Well, what is that really intelligence? Well, that depends on what you think intelligence is. Right. Right. At the end of the day. But yeah, I think you're right. I mean, if we could understand better what human beings are doing to process information, make decisions effectively and efficiently, then then we could advance artificial intelligence more efficiently as well. But let's maybe get into the history a little bit. Yeah, let's start. Let's just quit dancing around this issue and get into it. Yeah, yeah.
00:09:20
Speaker
The chapter here really starts talking about Galton. The idea here is, you know, Sir Francis Galton was probably, well, he was certainly one of the first people to try to measure this idea of intelligence. It points to the fact that as a psychological concept, intelligence was around long before any kind of study of it, you know, rigorously or systematically.
00:09:46
Speaker
But so yeah, so Galton, he was inspired by the work of Darwin. And just thinking of how our capabilities, human capabilities were continuous with other animals, other species that we, you know, that we've evolved from or, or, you know, continuously with. And so that's, that's that part is true, actually. So it is useful in some ways to compare
00:10:13
Speaker
human beings stack up against other animals and what we can do. And there's a whole discussion to be had there about what that means for how we think about intelligence. But the way he took it was to say, really, intelligence is pretty simple. There's really just two kind of pieces to the puzzle. One is energy, which he defined as the capacity for labor. And so Galton believed that
00:10:36
Speaker
intellectually gifted individuals in a variety of fields are characterized by remarkable levels of energy. So basically people who just are active and bring energy to the table are going to be more successful and they're more intelligent. And the second general quality was sensitivity. And he really thought this is sensitivity in the sense of the senses.
00:11:01
Speaker
sight, hearing, touch, you know, smell, etc. So he went about trying to measure intelligence by measuring people's sensitivity to different stimuli, which really seems like a strange idea about intelligence. But I mean, this isn't this is sort of an early way to operationalize or have something that's testable and measurable.
00:11:27
Speaker
Maybe more intelligent people just have more sensitivity to things just in their everyday senses. So, for example, they could discriminate two different weights that are slightly off. They just have a little bit keener senses and that's something that's going to play heavily in how we consider them intelligent. That's right. And, you know,
00:11:48
Speaker
You know, as a place to start, it's interesting. And he spent, he actually collected a lot of data. He actually spent time in his laboratory where visitors would come into this museum and they would basically take these tests. And so he would actually test people to see how they performed on these various different sensory discrimination tests. So for example, yeah, as you say, measuring like how, whether it's one weight was heavier or lighter than another.
00:12:17
Speaker
And, uh, you know, for example, what's the, he had a whistle that he was able to see what was the highest frequency that someone could hear. And the idea being that if you were more sensitive, you have more capacity for, for intelligence. And there, there's something about.
00:12:33
Speaker
I think some of this has probably been passed down to us in a way. So the idea of energy, I guess, gets passed down to us in some of the research on the amount of expertise that's actually developed just from pure practice. So if you work at something hard enough,
00:12:52
Speaker
that idea of the 10,000 hour idea from Anders Ericsson and then popularized by Malcolm Gladwell, if you work at something hard enough, then you're going to become proficient at it. So in a way, that might be something that Galton was getting at. And then sensitivity, just kind of this. Well, sensitivity, if you think about it, is really like sensory discrimination types of experiments, like the things that I would do in my lab to understand how color
00:13:20
Speaker
was processed in the brain. So color, chromatic contrast sensitivity, for example, or the ability to discriminate, find details in different color types, or different shades of color. I think that's a direct descendant in some way of the types of research that he was doing. Some of the methods that he was deploying were primitive forms of research that would be conducted in psychophysics labs even today.
00:13:51
Speaker
And I think this is the aspect that just did not hold up. I think there's not really much to this idea that pure sensory discrimination is an indicator of a generalized kind of intelligence. That's right. And I think we would say today that probably these are distinct domains. So your ability to be highly sensitive to things that you hear or see or touch or smell.
00:14:16
Speaker
There are separate abilities or faculties that are somehow distinct from what we think about as intelligence today. They're not correlated directly to a very high degree. There may be some correlations there. But for example, one really good example here is that by all accounts, Helen Keller was highly intelligent. She was both blind and deaf. So from that perspective, on its face, the argument kind of falls apart.
00:14:44
Speaker
And one of the interesting findings that Galton did have, which I think was attributable to him, was that sensitivity to high frequency sounds declines as we get older. And that's an interesting finding in aging psychology that is very true. It's a super robust finding. And he actually wrote about it. So we can attribute some actual
00:15:09
Speaker
correct science to to to Galton in that in that domain, although not much else. Well, I think I mean, you know, there's all the at least one of the ideas of some cognitive training is that if you can increase the perceptual discriminability, and this is like Mike Murzenek stuff, right, that if you increase some perceptual discriminability, at least your
00:15:35
Speaker
allowing things to get in there better. If your contrast sensitivity is better, you can start processing things better. Whatever happens to them down the line, at least they're getting in. Right, exactly. I mean, that's sort of the idea of the signal-to-noise ratio of the stimuli that you're taking in. If you're processing
00:15:54
Speaker
Say, for example, if you're in an environment where it's noisy and you're trying to hear someone and it's difficult to hear what they're saying because of all the background noise, you may actually take in the words one at a time, but your comprehension for the conversation and your ability to remember the words will be degraded. And a theory there is that you're requiring more of your cognitive resources to process that sound information, so less of it is available for storage and other operations.
00:16:20
Speaker
but that certainly doesn't seem like a central, it doesn't seem as central to the idea of intelligence. But I think it's also, his idea was not ridiculous on his face. But the thing with Galton, why people don't like him anymore is because he was a racist, et cetera. So, for example,
00:16:46
Speaker
he said, he said really, you know, nasty things about people who he considered to be of lower intelligence, for whatever reasons. And, you know, his theories became, you know, as so happens with so many of these ideas and theories is the theories sort of circle back on themselves and become, you know, impossible to disprove, because you make the determination that a certain person or a certain group of people are less intelligence, and then you define intelligence by what separates those, those people.
00:17:16
Speaker
Yeah, and I guess this is this is the basis of a lot of a lot of subsequent racist activity, too. Yeah, no, absolutely. Absolutely.
IQ Development and Educational Implications
00:17:26
Speaker
Right. So the next the next sort of push forward in the theory came out of the laboratory of Raymond Cattell.
00:17:35
Speaker
And he actually also did some more work along the lines of trying to be more precise in the measurement of sensory discrimination. One of the criticisms that came out of that work was that, back to this idea of the highest frequency that one can perceive, the idea that it turns out that cats and dogs can perceive higher frequency sounds
00:18:05
Speaker
than humans. So if you're defining intelligence by this ability, then you've encountered a pretty obvious problem. Right. And we don't seem to get smarter by being able to amplify different sorts of signals. And if you could use a device to see infrared, that doesn't necessarily make you smarter. That's right. That's right. And so then the next step forward, where we start to get into things that sound more like,
00:18:35
Speaker
modern theories of intelligence really start with Benet. Benet and Simon developed a theory of intelligence that was published in 1916. In their view, whereas
00:18:57
Speaker
previous theories like the theories of Galton were that intelligence was very simple and very reducible to simple processes, but then Simon thought that things were a little bit more complex and they emphasized abilities that were more oriented towards the complexity of handling information and decision making that were in complex environments.
00:19:22
Speaker
And this is where we start getting intelligence tests that are recognizable now. Simon Benet intelligence test that's in an updated form still used today. And they were originally charged with constructing a test that they could use for a practical purpose. So instead of just pure research,
00:19:44
Speaker
tests that they can use to place people in different areas in society, I guess. So it's very practically oriented. Yes, exactly. I mean, they were really starting to get at trying to develop tests that got into the idea about logical thinking and reasoning and the ability to think about more complex types of operations. And they also wanted to build tests that could be taken by students at different ages, starting at
00:20:15
Speaker
you know, by like eight years old or older, maybe even some tests for even younger children. So, you know, the tests would include things like vocabulary, but also, you know, similarities and differences. And we start getting into things like comprehension. So these are the kinds of assessments that would be familiar to people looking at what they would consider an intelligence test today, for example.
00:20:40
Speaker
Yeah. And this is where the concept of an intelligence quotient started coming from, if I'm not wrong here. Is that right? Yeah, that's right. Exactly. And, you know, the idea of Stanford, of Benet and Simon is that you can use these tests to place children in different programs. So depending on your intelligence level, you may be accessing different programs in school, for example.
00:21:07
Speaker
And the idea is you make a test where the average or the center of the test is at 100 and then your intelligence, if it's higher than that based on the standard deviations, it would be above 100. And if it's less than that, it'd be below 100. And I think the general consensus has been regarding these kinds of intelligence tests is that they're pretty good at
00:21:35
Speaker
at lower IQs, so below 100, it's good at sort of placing you in relation to other people. And it's okay above 100, but after a standard deviation or two, it doesn't really make much difference. There's not much variability past that point. There's not much explanatory power past that point, I would say. The difference between 130 and 140 IQ
00:22:06
Speaker
Yeah, it's tough to know exactly what that distinction means. That's where you might get people that are just good at test taking that might make up that difference. Right, and I think at that point it's interesting to think about that for Benet, he was really thinking about this as having a protective function. So for example, for students who are perhaps not in a position to be ready for an advanced
00:22:35
Speaker
class, for example, that they would be given some additional instruction, or a different, more customized form of instruction that would be appropriate for where they were at in their development. So socially, maybe a little more, maybe coming from a better place, that it's more about helping out people who are having issues rather than labeling people as more or less intelligent. Right. And Benet also believed that
00:23:03
Speaker
Intelligence was not fixed, but was malleable and that you could actually exercise and improve intelligence through the proper training in school. And so I think from that perspective, it's pretty benign in a way and actually maybe even could be.
00:23:23
Speaker
perceived to be a positive thing. But as with so many things in the world of psychology and also in the world of schooling and society and everything else, these types of tests have been used in ways that can be considered a little bit less positive, let's say. For example, if you said,
00:23:47
Speaker
someone took a test and they were not considered to be high IQ, they might not get access to programs for gifted and talented students, for example. And that's where you start to get into questions of equity. And
00:24:07
Speaker
I think that those types of questions of equity are super important when you're talking about the use of intelligence tests whenever they're applied, whether it be in school or for work, et cetera, because we do see that there are important cultural factors in performance on these types of tests. So for, yeah. Yeah. And that's something that I think gets highlighted a little bit later. When is it like not till the
00:24:37
Speaker
later in the century, at least, that people start noticing that there are some cultural biases in these tests. Edward Boring's quote, when asked what intelligence is, he says, intelligence is what the tests of intelligence tests measure. Right, exactly. So it's a circular argument. We perceive something in the world that we think is intelligence. So there has some what we would call construct validity.
00:25:07
Speaker
you know, or sorry, I guess this would be face validity in that case. So it has some face validity, and then we develop a construct around that, what we think of as, you know, being intelligence, we build some tests, and they all correlate together. And we say that's intelligence. And so it ends up being that what intelligence is, is whatever the intelligence tests measure, and it becomes circular in that way. And as it turns out that
00:25:35
Speaker
people who come from different backgrounds perform differently on those tests. And that can be used in arguments that are essentially racist, or classist, or both, and have been in art all the time. And this is an active conversation today. So I mean, I don't know if you know this guy, Charles Murray. Yeah, yeah, sure. So he's the author of this book from the 90s that was pretty controversial, called The Bell Curve.
00:26:05
Speaker
Yeah, some controversial ideas in it. One of the reasons why I think this is interesting is because it, you know, it's plotted as, you know, should we use science or not to investigate issues of intelligence? And is there something really there? I don't know. I mean, these are, these are interesting questions that come up. Anyway, yeah, so still rages as a debate as of today.
00:26:31
Speaker
Right. And just to make it explicit, the bell curve basically argues that the full title of the bell curve is the bell curve intelligence and class structure in American life. And Richard Heronstein and Charles Murray, and it was published in 1996. The argument is that, essentially, if you look at the bell curve or the normal distribution of intelligence test scores, you see that if you drew a bell curve for
00:27:01
Speaker
say Caucasian Americans, and one for African Americans, the the curve of performance on intelligence tests for African Americans would be shifted to the left, relative to that of Caucasian Americans. Their argument is basically that this is this is the cause of the inequality that we see in our society rather than the consequence of the inequality.
00:27:28
Speaker
And a lot of people who are saying you shouldn't investigate this stuff in the first place, I don't know if that's necessarily the best way to approach it. I think what's interesting about it is the way that you interpret this and the way that this gets used for anything further. There's plenty written about this stuff, so I don't know how much to just jump in this debate. We'd be remiss not to touch upon it.
Controversies and Biases in Intelligence Testing
00:27:53
Speaker
To me, it's central to the whole concept really of intelligence and intelligence testing, because to your point at the beginning, the concept is value laden. And so it would be surprising if the people who produce the test did not perform well on the test. Yes, true. So old white dudes at universities who make the test obviously score well on the test. And so you might imagine that there was somehow like
00:28:23
Speaker
some very pure way of measuring some objective thing, which is intelligence that has some real meaning. If the argument is that people who score well on this test get good jobs at universities, I don't think that solves the problem. I don't think that takes you out of the circle based on how our society is set up and how our educational system is set up.
00:28:48
Speaker
who's got access to these jobs and who's got access to this power, I don't think it solves the problem. No, it does end up being more divisive than inclusive. Let's just drill down and think for a second, because there's a really super interesting example in this paper, in this chapter, that is just incredible from this perspective. It's an analogy, right? And these types of analogies are used
00:29:18
Speaker
in intelligence tests, okay? White is to black, as good is to... The correct answer is bad. Come on. How can you not see that? How can you be the author of this paper and not see that? It's incredible. It tries to jump off the page. I mean, I read the analogy, I said,
00:29:46
Speaker
The analogy, white is to black as good as to racist. I mean, all I could see is racism in that statement. I couldn't see anything else. I couldn't even answer the question. It was so obvious it jumped off the page to me. I mean, some of that comes from understanding some of this history and where it comes from, but I just literally couldn't see anything else there. And it's just amazing that that, and he uses the analogy, that specific analogy throughout. How can that even be the right answer?
00:30:15
Speaker
Well, I think white is the black as good as the bad. That's even even thinking about this stuff is I think it's a good illumination of the way that scientific research is often biased and researchers don't really you can't become aware of something that you're not aware of in the first place. You can you can try to you can try to be on the lookout for it. And that's what good scientists should be doing. Yeah, I mean, all science is laden with biases and and not being able to see
00:30:45
Speaker
Clearly exactly what is they're doing that you're doing because you know You're on the you're on a conceptual edge of things and you're trying to explore things that are unknown So you're you you may be missing out a lot as you're performing science and a lot of this stuff may be visible in retrospect and I think you know the test of this is how people correct course as they start understanding what intelligence tests are measuring or what they should be used for and
00:31:09
Speaker
Absolutely, absolutely. Yeah, I mean, so I think that's just yeah, I think really where we started the conversation, right, which is that this discussion is value laden. And all of psychology is every aspect of it is and I think if you don't think about it, you're going to miss something.
00:31:30
Speaker
like in our work of basically saying that that cognitive abilities can be improved. Just just making that statement. Forget, you know, the the exact results or whatever. If you just if you start from that perspective, you're just going to see everything a little bit differently than if you're saying, no, you know, cognitive abilities are fixed. And, you know, some people just are born with more of these and they're going to do better in the world and life is going to be great for them and everyone else is going to be a little bit screwed.
00:32:00
Speaker
you're going to see everything a little bit differently. You're going to interpret the same results that you're going to interpret the same experiment in a different way. So I think it's just important to think about what you're bringing to the discussion when you have that discussion. Yeah, and I think one of the
00:32:16
Speaker
Good points in this paper too is the idea that no area of psychology or no area of testing in psychology has been so influential on practical everyday matters as the idea of intelligence testing because it's used widely all over the place and has been for a while for
00:32:37
Speaker
placements and by employers looking to hire the best people. Their motive is generally you want to get the most intelligent person. It's just a given. You want to get the most intelligent person for a job. To that end, you use intelligence testing and you take it as a given that the researchers that have constructed this thing have constructed something that genuinely measures what you want as an employer.
00:33:05
Speaker
Yeah, and yeah, I mean, there's so much to say about that. There's so many worms in this can. Yeah, it really is. It's wild. But one of the things that makes me think of right away is that aptitude tests for college, the SAT, is not marketed as an intelligence test.
00:33:26
Speaker
But if you think about it conceptually, it really is the same concept. It's the idea that there's an aptitude that you have for performance at school. It's not that you know stuff. I mean, they're asking questions about things that you know or don't know, like math stuff, but the idea that it's a scholastic aptitude test. It's not an interesting thing that so it's.
00:33:51
Speaker
I think the abbreviation now is Scholastic Assessment Test. So it's okay. I mean, this is an area where they're trying to scrub it. They're trying to scrub it a little bit. I mean, it's the same exact test. It's the same thing. Yeah. It's the same framework. And everyone understands it the same way. This person has got a good score on the SAT. They have 1400 or whatever. I don't know what the numbers are even now. They got a good score on the SAT. They're smart. They're gonna do well.
00:34:16
Speaker
life is going to be good for them going forward. They're 16, 17 years old, we're making this judgment about their whole life course, based on their aptitude, which is meant to be somehow an inherent characteristic of that person.
00:34:30
Speaker
Well, at least in part or at least at least a decent predictor of other measurable successes in life like income and grades. It's not a perfect predictor. Not at all. It's a better predictor of performance in school than it is, you know, subsequent income or performance in other other domains. But I'm going to like I'm going to try to pull it back to like the discussion of the historical thing. We can dive back into some of the other
00:34:59
Speaker
social components. It's probably a whole series of discussions around that. Maybe we should have some guests on to discuss these specific things. It's a topic obviously that I'm passionate about, but I don't want to derail the entire discussion here because I think there's some interesting stuff to get out of this history lesson here. I was just starting to say one last thing about this, and this is not just to get the last word in,
00:35:27
Speaker
I feel like one of the difficulties here is if you're an admissions director for a school or if you're an employer, you may have a fair understanding of what's wrong with intelligence tests or SAT scores. But if you look in the literature, it's hard to find anything that has more predictive power than that.
00:35:50
Speaker
You know, as an employer or an admissions person, you want as much predictive power as you can get. And, you know, at least say it was 10% predictive. That's better than no percent predictive. Right. So it's just it's sort of like what we have. I mean, I guess what we have is not great, but it's hard to know what alternatives are in some cases. But but part of the problem is that it's predictive, not for the reason that we want it to be predictive.
00:36:20
Speaker
It's not that people are somehow inherently more capable of cognition and therefore they score better on the test. It's because the system is designed in such a way that performance on this test correlates with your performance in the system. I'd buy that. That's a hypothesis, but I mean, that's my hypothesis. Let's just say that. I have a lot to say on that topic,
00:36:50
Speaker
Maybe we should take a break and we can come back and maybe then talk about some of these other aspects like factor analysis and Spearman's two-factor theory. Good times. All right, welcome
Theories of Intelligence: Factor Analysis and Beyond
00:37:09
Speaker
back. We are going to jump into talking a little bit about
00:37:13
Speaker
factor analysis and how this math of factor analysis features in the discussion of intelligence theory. And I know people can't get enough of factor analysis. So this is going to be a hot topic. I mean, it is a little bit. It's a little bit of a challenging topic, but
00:37:34
Speaker
The cool thing about factor analysis is that it relates to so many things. Yeah. And conceptually, it's not that it's not that complicated of an idea to get. Not at all. Not at all. And I think I think probably that level is the right level for the discussion, which is sort of the conceptual level, which is basically factor analysis is a statistical model that tries to take a bunch of different variables and based on the correlation between those variables, reduce those
00:38:03
Speaker
into a smaller number of dimensions. In intelligence testing you might have 10 tests and you look at the correlations on those tests and you find out that well actually five of these people who perform well on those tests tend to perform well on all five and the other five
00:38:26
Speaker
there's no relationship between the first five, but people who perform well in the second five all perform well on all of them. So if you perform well on one of those tests, you perform well on all of them. So you might say there's actually, you're not measuring 10 different things, you're really only measuring two different things. So we're going to call those two different things, different factors. And this concept of dimensionality reduction is used in machine learning and data mining all the time. You know, factor analysis, exploratory factor analysis is one component
00:38:55
Speaker
where principal components analysis is highly related to factor analysis. Independent components analysis is like the next level up of complexity or mathematical endeavor in this direction. But they're all related to the basic concept of you're trying to take a lot of different pieces or variables that look like they're taken from different sources or different measures
00:39:22
Speaker
but are actually measuring the same thing and to basically say, well, let's reduce this into a more manageable number of dimensions.
00:39:31
Speaker
In intelligence tests, Spearman was the guy who really started this approach in the early 1900s. His first paper on the topic was I think 1904. What he was noticing was that if you look at the correlation matrix, so you just look at all of these different tests and you look at how they're correlated, certain groups or clusters pop out. But what he also noticed was that people who are good at
00:40:01
Speaker
any of these tests on the intelligence tests tend to be good at all of them. And people who are not good at one of them tend to be not good at all of them. So there's a high intercorrelation between performance on the different tests and an intelligence test. Just to sort of create an analogy, if you were hypothesizing something like how good at being a handyman are you?
00:40:28
Speaker
You know, we don't know and we don't say fictitiously. We don't know anything about what handyman capability is, but we measure you on how well you can drill. We measure you on how well you can saw. We measure you on how well you can put up drywall. And then we see to what extent those things are correlated. So if handyman is a good way of describing this construct, those people who are good at drilling should also be good at sawing and should also be good at putting up drywall.
00:40:57
Speaker
So likewise with intelligence, we can take a look at this construct that we've called intelligence and see how what we're thinking of components of it interrelate to each other. And so he out of this analysis proposed his two factor theory of intelligence. And this theory is still extremely influential today. In some senses is involved in every discussion of intelligence that is a statistical or rigorous discussion today.
00:41:26
Speaker
the two-factor theory says that there's a general factor called G that is common to all tasks requiring intelligence and one specific factor S, which is unique to each different type of task. So statistically, you could break out the variance in performance across these tasks into a G factor and an S factor, general factor and a specific factor. Yeah, and this is a
00:41:54
Speaker
a massive debate. I mean, all of these things kind of continue on in some form to this day. So to what extent is intelligence captured by just one single number that cuts across all of these things that we think of as being intelligent behavior? If it's true, this is a massively simplifying concept. You know, if you can explain all different kinds of
00:42:20
Speaker
performance based on one single number, you've got something that's massively explanatory. So it's a big idea. Yeah, it's a very big idea. And the reason why it's so influential is that the statistics are incontrovertible. Anytime you create something that is considered to be an intelligence test, you will find that there are inter correlations within the test. And
00:42:46
Speaker
you can find some domains where there are people that tend to be good at some of the tests, but not others. But that is always sort of overwhelmed by the overall impression, both qualitatively and quantitatively, that people who are good at the test tend to be good at the tests. If you're good at one of these tests, you're going to be good at most of the tests. Why is this a problem? Or to what extent may this not be true? Is this something that's influenced by the statistics?
00:43:13
Speaker
Yeah, I mean, there's a lot to be said here, because again, it all relates back to values and also measurement, a problem in theory and there's a problem in measurement. On the theory side, there are different ways to explain the underlying results. It is the case, as I say, having looked at a lot of these experiments, a lot of these different setups,
00:43:40
Speaker
These correlation matrices are impressive, which is to say that if you're good at one thing, you're good at other things. That's right on these tests. Then, of course, you pull this out and you say, well, people who get those tests also do better in school or more successful in careers, et cetera.
00:44:02
Speaker
it's tempting then to say really are measuring some general factor that is a capacity for intelligent thinking that is just generally beneficial. There are, however, other ways to explain the data. For example, Sir Godfrey Thompson in 1939 argued that it's possible to have this pattern of results and not have an underlying single factor or capacity or ability
00:44:32
Speaker
He basically suggested that there are these ideas of what he called bonds, which is a little confusing, but really it's a little easier to understand that if you think that each of these tests is measuring multiple abilities and each test measures these multiple abilities in different proportions, but it's the same underlying, say five or six abilities that are being measured by all the different tests, then you could get that there are in fact five or six different abilities.
00:44:59
Speaker
that are separable but are nevertheless inter-correlated because each test is touching on each of them to some extent, to a greater or lesser extent. Does that make sense? If you're talking about the handyman example, you could have different reasons why an ability to drill and an ability to saw are correlated with each other. That doesn't necessarily suggest that there is a handyman ability.
00:45:28
Speaker
Right. Just hand-eye coordination or some sort of skill with your fingers, I guess. Yeah. And particularly what I found super interesting about this part is this one is where it kind of gets the closest to my, what I would call informal hypothesis about what's going on here, which is.
00:45:51
Speaker
Although Thompson did not attempt to specify exactly what the bonds might be, for example, they might include understanding the problems and responding to them. What does that mean? In other words, do you understand what the questions on the test are asking and what you're expected to do in the response to the questions? I feel like that is a two parter. Part one, do you understand that
00:46:20
Speaker
social, political, and inter-level context of the test that you're taking? Why are people asking these questions in this way? If you're coming from a certain perspective where this is the world that you've grown up with, these questions make some sense. Or if you're coming from the outside from a different perspective, they make less sense. And then are you motivated to answer these questions in a way that people want you to answer them?
00:46:48
Speaker
I feel like that understanding the question and then motivation to answer the question are two components that I think are probably the driving factors of why performance on these types of tests are correlated with success in our culture. And the tricky thing is- It's also an economical system. Yeah, and I think the tricky thing in these sorts of statements is that
00:47:18
Speaker
It could be due to things like intelligence, or it could be due to things like practice and cultural competency. And it's just hard to figure out which one it is. It's possible that it's likely that both of them are true in some sense. Yeah, no, absolutely. And I mean, I think
00:47:47
Speaker
In some sense, it's a little bit like what kind of touch points do you have with these tests? What kind of points of connection do you have? If you've worked on similar problems in the past or you've been exposed to similar conversations or thought experiments in the past as they're being asked, like the analogies or the types of reasoning that are presented in Raven's Progressive Matrices example, then you're going to do better on the tests. Yeah. I mean, this is why Kaplan exists.
00:48:17
Speaker
you can have places that specifically help you with recognizing what types of things are on the test and practice for those. Right. I think it makes sense to get into, now that we're talking about G, I think it makes sense to talk a little bit about some of the other models that are broken off in terms of thinking about different types of intelligences.
00:48:44
Speaker
One of the areas that has been developed that remains incredibly influential and important in the field is Cattell's theory of fluid and crystallized abilities. Yes. So in the fluid and crystallized abilities discussion, essentially it's a hierarchical theory. So the idea is that there is a highest level G factor that is, you know,
00:49:14
Speaker
related to your overall performance in intelligence tasks and anything related to intelligence. But then below that, there's a fluid intelligence ability, GF, and a crystallized ability or GC. And fluid ability is the ability to think flexibly and to reason abstractly. And crystallized ability is the accumulation of knowledge over the course of one's life.
00:49:39
Speaker
And what's appealing about this approach is that it sort of tries to separate out these factors of what we might think of as like an underlying capacity, this GF, and just knowledge. Because these things get really conflated in a lot of discussions of intelligence, whether you know something or don't know something. If you never were exposed to it, never had an opportunity to learn it, you're not gonna know it. And so knowledge,
00:50:09
Speaker
is something that we think of as being distinct from intelligence, but yet it's also clear that we don't think of someone as being intelligent if they don't know anything. If you're making an intelligence questionnaire, you don't want to ask something like, who was the 14th president of the United States? That would be something that's purely crystallized intelligence. But like you say, yeah, you want people to have some sort of framework of knowledge that they can put new facts in, I suppose.
00:50:37
Speaker
That's right. And, you know, one of the, you know, the sort of the canonical test of crystallized ability or GC would be a vocabulary test. The canonical test of the fluid ability or GF would be the Raven's progressive matrices. And so I don't probably most people don't know what the Raven's progressive matrices are. So describe that.
00:51:02
Speaker
Basically, the idea here is that it's a matrix in the sense that it's like a visual test. There are different types like this, but in this particular exact one, you might have nine cells. It's like a three by three matrix. Think about like a tic-tac-toe board.
00:51:25
Speaker
In a very simple form of the test, you need to basically say what would be in the bottom right-hand corner cell. So the bottom right-hand corner cell is blank, and the top row would be circles. In the first column, there's one circle. In the second column, there's two circles. And in the third column, there's three circles. In the second row, you've got diamonds. There's one diamond, two diamonds, three diamonds. And then the bottom row, you've got one triangle, two triangles, and then a blank.
00:51:55
Speaker
I think we can all guess what's going to be in that bottom right hand corner. It's going to be three triangles. And so the correct answer is three triangles. And this is an attempt to get at something that everyone can do. So at least the attempt here is that it's relatively free of biases from language. It's a reasoning kind of task that you have to extrapolate what comes next.
00:52:24
Speaker
Exactly. And so the idea is it's sort of pattern recognition and reasoning, and it's supposed to be culture free. At least as different from crystallized intelligence, a list of facts as you can attempt to do. Yeah, exactly. And so obviously the example that I gave was very easy, but you can imagine all kinds of conjunctions of different factors.
00:52:52
Speaker
different color, form, direction, orientation, et cetera, that make this really quite complex. It's meant to be a measure of your flexibility and ability to solve adaptively problems. But you sound skeptical. I'm going to guess it's your skeptical. Well, you could always just say that right off the top. You don't even have to ask me a question. I'm going to be skeptical, whatever it is.
00:53:21
Speaker
But you're particularly skeptical. I'm extremely skeptical when it comes to this one because I worked on this one for a while because I wanted to build my own version of the test because it's useful for measuring this representation of intelligence. Really at the time that I was working on it, you know, it was five years ago.
00:53:42
Speaker
everyone would agree that this is the measure of fluid intelligence. Everyone was pointing to this. I think it still is, really, if you think about it, like the benchmark for fluid intelligence. If you can improve someone's performance on this Raven's progressive matrices, then you have an argument to make that you've improved fluid intelligence. Now, the problem is you can figure this test out.
00:54:11
Speaker
I got to the point where I can solve all of these problems. I can get 100% on any of these tests. So you have an infinite IQ? Yeah, I'm perfect. My IQ is perfect. I'm infinitely intelligent. That's great. And I can solve any problem. But guess what?
00:54:29
Speaker
It doesn't work like that. It does not work like that at all. I'm actually no smarter for having solved this test than I was before I figured out how the test works. I mean, it's really just, there are some rules. There are four or five underlying patterns or rule sets. And once you've learned what those are, then it's perfectly possible to solve every one. Except for some, occasionally there's like a weird one where you're like,
00:54:59
Speaker
And then when there's a weird one, there's an exception, you're like, well, why is that the answer? Who gets to decide that that's the answer to this? Well, you have to be pretty smart to make them, I guess, right? You have to be smarter than anyone else if you're gonna decide that that's the right answer. Exactly, exactly. So however you decide that the answer is, is by definition the most intelligent thing. So yeah, I mean, it gets back to this thing of, again, who makes the test and what are they going for? I mean, there's a whole another question of like,
00:55:29
Speaker
What are you learning in your cultural experience or your scholastic experience that allows you to perform this test better or worse? But you're definitely learning something that's allowing you to do this test. There are things that you do in school that are sort of like this, analogies and other, the way that math problems are set out. Yeah, you see familiar patterns that get repeated.
00:55:57
Speaker
So it's definitely something that's learned. I'm 100% sure. But it's meant to be the most pure basic form of assessment of fluid intelligence, which is really meant to be the most pure measure of innate ability. Yeah. And it seems like something that, again,
00:56:19
Speaker
a general conception of how intelligence tests work, that they may work better at the bottom of the scale than at the top of the scale, that after a certain point, especially if you get people that have noticed them and seen them a lot before, that you're not testing intelligence anymore. Yeah, exactly. Exactly. It's a thorny issue. But then I think as we move from
Multiple Intelligences and Diverse Capabilities
00:56:46
Speaker
from generalized G-factor type intelligence. I think it's also interesting to talk about Gardner's theory of multiple intelligences. So this is the one that if you're aware of theories of intelligence, this is another one that people will have heard of just or at least been exposed to at some point. So this is in the 80s when this stuff first came up. So Howard Gardner
00:57:12
Speaker
proposed this idea that a lot of people liked and a lot of people really took to that we shouldn't think of intelligence, it's really in opposition to G, in opposition to having one single thing that explains all intelligence, that it's really, we should divide it into several different ways. Now he divided it in a couple different ways. Let's see, what's the most
00:57:36
Speaker
Standard one that he's divided this into yeah, I think the most standard one is like this nine nine version. Okay, so what are the nine? so like There's like the verbal linguistic learning through spoken and written words mathematical logical learning through reasoning and problem-solving musical visual spatial bodily kinesthetic intrapersonal learning through feelings values and attitudes and
00:58:04
Speaker
interpersonal learning through interactions with others, naturalist learning through classification categories and hierarchies, and existential learning by seeing the big picture. So it's really meant to be also viewed in the context of a learning context. So the idea that people learn differently. Yes. The big draw to it, I think, for a lot of people is that
00:58:32
Speaker
it's maybe the values are broadened a little bit. So it's not just the kind of intelligence that people initially think of, kind of book learning being smart in a certain sort of way, that you can be intelligent in a spatial way, or you can be intelligent in a musical way. And that also counts as intelligence, bodily kinesthetic, that you can actually have a kind of body intelligence. And
00:59:01
Speaker
You know, there's I guess there's two ways of looking at this. One is that maybe he's getting at something real and maybe maybe we place too high of a value on a certain sort of intelligence when we can we can have intelligence all these different sorts of ways. Another way of looking at it is saying, well, are these other things really intelligence or should we talk about them as different kinds of competencies? People can be good at
00:59:30
Speaker
they can have good spatial abilities or good bodily kinesthetic abilities. But is that what we mean by intelligence? Right, right. And I think one of the things that obviously this points to is that maybe there are different ways to teach, depending on people's abilities, underlying abilities, and how they learn.
00:59:51
Speaker
It's kind of related to that of learning styles. I'm not as much of an expert in this domain, so I won't say too much about the evidence, but I understand that the evidence here is also quite controversial, as it is everywhere else, whenever the word intelligence comes up in the context of people. So there's some question as to whether or not there's evidence that the different techniques for teaching to people with different types of intelligences are better or worse.
01:00:20
Speaker
And Sternberg points out as of when this was published, which again is a few years ago, I think was this 2003. So after the theory of multiple intelligences is about 17, 18 years old, there had been no comprehensive empirical tests of whether or not this is actually the case. So I guess the idea here is that once you get into
01:00:50
Speaker
proposing a lot of different sorts of intelligences, the testing gets much more complicated. You have to test so many more different things and you have to cross a higher threshold to convince people that these are the right ways to think about intelligence. And I don't know what's happened since this paper. So I'm not exactly sure if people are still following this idea, but I guess that's maybe the impression I had is that it's a good idea, but it's hard to verify.
01:01:20
Speaker
Oh, it's it's it continues to be a hot topic. It continues to be a hot topic for sure. The reason why it is so difficult to get out of these eddies in in currents in the science of this is that all of these arguments statistically are based on correlations and correlation is not causation. But it's easy to get lost in that and get confused. I mean, that's a given example. So give an example of
01:01:49
Speaker
maybe something more concrete, where you might see a correlation and it might not be a causal relationship. Yeah, I mean, I think as a general principle, you know, you have all of these, you know, spurious correlations, they're just they just happenstance. Sometimes, you know, things just happen to be correlated. But that's not really what we're talking about here. What we're really talking about here is that things that are correlated, but often for a third factor. So there's a third reason
01:02:18
Speaker
two things are correlated. For example, someone might be successful in school and they might make a lot of money later on, but the underlying cause, you could argue, is some third factor like privilege, for example. Yeah, like if your parents have a lot of money, they send you to private school to prepare you for testing and also
01:02:46
Speaker
get you a connection at a job so you can get a nice high paying job after college. That's right. That's right. Yeah. But I mean, there's all kinds of just, you know, funny, like things on the internet with like the spurious correlations that are just unrelated things that prove the fact that correlation is not causation. But whenever you overlay two related things that are correlated, it's tempting to make them be
01:03:11
Speaker
somehow causally related in your mind. Especially if you have this feeling that they should be, or if it seems like a natural thing already. Right, exactly. But there's this one graph that I'm looking at right now, which is divorce rate in Maine versus the per capita consumption of margarine in the US. And these two things are perfectly correlated from 2000 to 2009. So that means that logically,
01:03:41
Speaker
People get divorced in Maine, and therefore they're eating less margarine. Or maybe the lack of margarine is really keeping them together in their marriage. That's a great example. I like that. That's one where I'm having a hard time thinking of something that would make any sort of logical sense for that. No, but the correlation is extremely strong. But yeah, I mean, in terms of these tests, as I said at the beginning, the correlation
01:04:10
Speaker
between the tests is high. And so it's hard to pull apart things that are not correlated together. But then within that, you can rotate the axes to be whatever you want them to be. So there's this whole industry within intelligence research to basically rotate these factor axes to match more directly to something that feels like it makes sense.
Challenges in Interpreting Intelligence Data
01:04:38
Speaker
So an axis might cut across two different tests, for example. So these two tests are, one is like a verbal list learning and one is matrix reasoning. And one factor might load both equally on those two tests. But you're like, well, I think that makes more sense for there to be like a verbal learning ability and a fluid reasoning ability. So I'm just going to rotate the axis
01:05:08
Speaker
And now the math works out the same because it's a coordinate system, just like if you took the north, south, east, west axes of the earth and you just said, all right, actually, they're just going to be 45 degrees rotated. So all the distances between the locations on the earth will still be able to be measured in the same axis in this coordinate system. But, you know, now you've got, you know, instead of the north, east, south and west, you've got like
01:05:38
Speaker
nerf or whatever, you know? And so, you know, different, it's just like you're calling it something else. And this is true of these axes in general is that there's no, there's no inherent truth to an axis and an axis and what lies along it. It's just sort of found truth that you've got these things that are correlated and then you label it with something that
01:06:06
Speaker
seems to make sense that captures what lies along that axis. What are two things that would correlate strongly in intelligence? Certainly, any of the analogies and matrix reasoning would correlate strongly. Verbal list learning and digit span correlate strongly.
01:06:33
Speaker
So you put those together and then after the fact you say this represents some kind of larger factor that we will call X. Right, exactly. Exactly. So you can play with the math, but the underlying, it's difficult to get out of it because the underlying structure of the argument is this correlation. So you don't know why these things are correlated. In order to have a causal
01:07:03
Speaker
mechanism, I feel like there's different ways to go about it. One, you could knock out a part of the brain, for example, and if that eliminated one of these abilities, you would have an argument that that part of the brain sort of in some sense was causal and its relationship to that, to that ability, or you could, if you could improve one ability and not another ability, then I think you would have an argument, again, to the structure of the cognitive space, or the intelligence space. The problem is that
01:07:33
Speaker
we have not been successful at making an argument that we can improve a cognitive ability that gets out of the loop of, well, okay, you got better at this test, but this test is actually just the same test as this other thing. It's measuring the same thing in a slightly different way. You haven't improved the ability, you just got better at the test. And I mean, I think that that argument has been overplayed
01:08:03
Speaker
to a large extent. I think that if you do one activity and you get better at a different activity, then I think you have to say that you've improved on some ability in some way. But yeah, there's no consensus there. I wonder if we can make an analogy to something else to help make sense of this, too. So imagine that you're talking about someone's improving their skill at baseball. And then you say, OK, what they're
01:08:34
Speaker
They're not actually getting better at baseball. What they're really doing is they're just getting better at hitting a ball and catching a ball, which is not generalized baseball, like a generalized baseball ability. They've only improved on these aspects of baseball. So we can't really say that there's a real construct called baseball that they're improving at. I feel like that's the sort of linguistic trap that you might get into.
01:09:03
Speaker
Right. And it feels absurd when you use it for baseball, because obviously then and you are getting by definition, if you're better at hitting and you're better at catching and you're better at, you know, right. Or even if you're better, any one of those, any one of those things, you have improved your baseball ability skills, even though you could you could more accurately say, no, what you're actually doing is you're just increasing your hand eye coordination or, you know, you could pick any arbitrary thing that you're improving. Right.
01:09:28
Speaker
or, but you might, or you might play some value judgment saying, well, actually it's not that important that you got better at baseball. However, if you improved, uh, if you improved your hand, eye coordination, now that's useful because that generalizes that generalizes. So now you can use that in football and basketball. And now that's useful. Just getting better baseball is not useful, but I don't agree with, but you know, that's the, that's the, that's the form of the argument that's made against,
01:09:56
Speaker
the ability to train certain elements of cognitive capacity or intelligence. It's the kind of argument that I think you could use against just about anything that has a conceptual basis or a construct that you're labeling something. And this is just an issue that people in cognitive psychology deal with all the time because
01:10:19
Speaker
Nothing in cognitive psychology is observed directly. You need a test for everything that you're trying to measure. If you want to measure reasoning ability, there's no pure reasoning ability that's independent from the test that you use for it. I mean, every kind of cognitive structure that we talk about is dependent on the way that you measure it, and it isn't something that we see directly. Nevertheless, it's still important that we know something about
01:10:48
Speaker
our cognitive abilities. But it is a generalized problem that people may specifically apply to intelligence more so than other concepts that they use in everyday life. Yeah, I totally agree. I think that might be a good place to wrap up the discussion for today. I think the discussion has touched on a number of interesting elements about the challenges of
01:11:11
Speaker
of cognitive psychology broadly? Is there a way to be useful and have impact and say something about the real world without
01:11:21
Speaker
causing more harm than good. And I think this historical overview of intelligence is probably a good way to start a discussion, and I'm sure to be continued at some point. Thank you all for listening. Oh, and a last little plug, if you can let other people know about the podcast if you like it. You can rate us on the iTunes store. You can also tweet JL Hardy PhD on Twitter. And I am something on Twitter. I don't think he's what. I think he's Ralph Nelson. We're going to get this right in one of these days.