Accelerating Future Transhumanism, AI, nanotech, the Singularity, and extinction risk.


IQ: “Lonely Ice Floe” or Consensus Science?

Malcolm Gladwell calls those who accept the Mainstream Science on Intelligence statement "IQ fundamentalists", but the reality of g and the predictive validity of intelligence tests are widely accepted as consensus science by intelligence researchers, with some caveats. Reading Eurekalert and PhysOrg, I see press releases practically every day that analyze the correlation of intelligence with a variety of genetic and environmental factors. Here's one from yesterday:

Fit teenage boys are smarter
But muscle strength isn't the secret, study shows

In the first study to demonstrate a clear positive association between adolescent fitness and adult cognitive performance, Nancy Pedersen of the University of Southern California and colleagues in Sweden find that better cardiovascular health among teenage boys correlates to higher scores on a range of intelligence tests -- and more education and income later in life.

"During early adolescence and adulthood, the central nervous system displays considerable plasticity," said Pedersen, research professor of psychology at the USC College of Letters, Arts & Sciences. "Yet, the effect of exercise on cognition remains poorly understood."

Pedersen, lead author Maria Åberg of the University of Gothenburg and the research team looked at data for all 1.2 million Swedish men born between 1950 and 1976 who enlisted for mandatory military service at the age of 18.

In every measure of cognitive functioning they analyzed-- from verbal ability to logical performance to geometric perception to mechanical skills -- average test scores increased according to aerobic fitness.

However, scores on intelligence tests did not increase along with muscle strength, the researchers found.

"Positive associations with intelligence scores were restricted to cardiovascular fitness, not muscular strength," Pedersen explained, "supporting the notion that aerobic exercise improved cognition through the circulatory system influencing brain plasticity."

I support the consensus science on intelligence for the sake of promoting truth, but I also must admit that it especially concerns me that the modern denial of the reality of different intelligence levels will cause ethicists and the public to ignore the risks from human-equivalent artificial intelligence. After all, if all human beings are on the same general level of intelligence, plus or minus a few assorted strengths and weaknesses, then it becomes easy to deny that superintelligence is even theoretically possible.

Some people are just more intelligent than others in every possible way. (Though most people have strengths that others don't, such as through learning and talent.) This sounds unfair and politically incorrect, but that's what we see in the data. The modern neo-mystical pseudoscientific folk view of intelligence seems to indicate that if someone seems genuinely more intelligent at first, that intelligence must surely be accompanied by some major flaws, to "balance it out" on the cosmic scale. This may be true sometimes -- for instance, nerds tend to have poorer social skills than average -- but it doesn't always apply. Some people are just better at everything. This sort of talk is often considered forgivable when people mention it casually in real life in relation to a specific circumstance, but for some reason when it is put down in text in general terms, would-be egalitarians try to shoot holes in it with unscientific theories like Gardner's multiple intelligences concept.

Comments (20) Trackbacks (0)
  1. the reality of g and the predictive validity of intelligence tests are widely accepted as consensus science by intelligence researchers, with some caveats.

    Hmm… That made me think. Especially of another g. A big one this time. The reality and predictive validity of G, the gravitational constant, are also widely accepted… though most of the more serious physicists probably wouldn’t dare to boldly state the “reality” of a thing that has the dimension of m^3 kg^-1 s^-2… But at least their G has units.

    Had you asked Newton about the reality of G, he’d probably have laughed (or started rambling about god, knowing Newton…) since for him it was just a thing needed to fudge the calculations.
    Asking today’s physicists about the reality of big G would either give you a shrugging “ask a philosopher” or an interesting tour through to the beginnings of the universe and the origin of G. They know that Newton’s G was just an ugly and ill shaped crutch, hiding something more profound.

    Many psychologists don’t seem so think that way, but see their g as some kind of reality in and for itself, which is a problem.

    Physicists need G to describe the world, as psychologists need g. Though when asked about the origin, meaning and significance of their g/G the psychologists come off far worse.

    “g has something to do with solving problems and figuring out things and such.” (freely rephrased from point 1 of the only half endorsed “Mainstream science on intelligence” article) That does not really do it for me.

    Point 2 of said article is the one and only redeeming feature of g: “We measure something there!” (freely rephrased)
    We know a little bit on what it depends and what consequences it might have if that measurement is high or low.
    We just don’t know what it is.
    We don’t know where it comes from.
    We don’t know how to build it.

    There’s no need for poking holes. g itself is an ugly crutch, but since it’s the only thing about thinking we can reliably measure it fills a void.
    At the same time there is a need for theories like Geardner’s multiple intelligences. Were it true it would have enabled us to chop the big little g into more manageable pieces.
    In order to understand intelligence, understand what the heck g actually is, where it comes from and how it works, chopping down is the thing that must be done. Though probably not at Gardner’s presumed fault lines.

  2. Wolfgang, nice commentary. I think you’re spot on. IQ and g are definitely just temporary crutches. Hopefully one day we will look back on our ignorance and laugh. I agree that we do need alternate theories and researchers pulling at the flaws and loose ends of g. New research appears every day.

  3. Michael, I’ve always tried to avoid the term “politically incorrect”–it usually signified an unacademic bent.

  4. As I understand it, the debate has two facets:

    1) Those who believe there is little or no difference in intelligence between people (a position that is near impossible to defend in scientific terms.)


    2) Those who think the measurement of intelligence is flawed, that “Intelligence Quotient” is currently invalid (a position that is very rational.)

    These two groups should never be treated as if they are identical.

  5. I agree with Singularity7337 ever time I hear the term “politically incorrect” I cringe inside. I do not see how you can say that Gardner’s theory is any less scientific then g; not that I necessarily agree with either as defined. Do you even know what scientific means?

    Both of these groups believe in physicalism. One group just believes that creating an AI will be easier than the other. Develop some algorithm and instantiate a million copies.

    There are plenty of neuroscientists among others who would not touch this debate with a 20-foot pole; not because it’s “politically incorrect” but because they just do not know what the right answer is and do not want to be on the wrong side of history. This leaves the dregs to speculate.

  6. Really how can anyone possible prove g to be false? It is un-falsifiable, sounds unscientific to me.

  7. Thanks “jimmy dean”. I’ve long thought the whole “political correctness” term is to avoid thought–we can be above that. I happen to think intelligence is a real thing, but just wanted to point out the term.

  8. As Homer Simpson would say “hummmm I guess there are three kinds of people in the world those who believe in general intelligence, those who believe in specialized intelligence and those who believe in no intelligence at all”. ;)

  9. jimmy:
    To prove Gardner’s theory as correct would have falsified g in the same instant.

    So in an experiment, measuring the aptitude in a wide array of cognitive tasks, your findings either correlate, or they don’t, disproving either g or Gardner.
    As much as I personally hate g, the data seems to be on its side.

    Also Gardner’s theory in itself isn’t unscientific. Just probably wrong.
    Unscientific are the notions of “fundamental equality” that sprouted in the popular venues as an aftermath of Gardner, since they aren’t based on anything. Which is I suppose was, what was meant, though not clearly said ;)

    There is a reason why many neuroscientists don’t touch the topic. It’s not their field yet.
    What “neuroscience of intelligence” there is today, doesn’t deserve the name. There is no understanding of intelligence (whatever that is) on a neuronal level. So it’s primarily a debate among psychologists and that for a good reason. They have the tools and the understanding to tackle g, most neuroscientists don’t.

  10. I am not sure what data you are referring to IQ scores do seem to be correlated to increased academic achievement. The effect is limit after you account for confidence intervals the effect explains perhaps 30% of the variation in College grades. High school GPA is a far better predictors however at about 50%. Both are still what are known as weak predictors.

    Remember you have to make sure that you are really seeing a statistically significant difference, pure correlation does not account for this at all. Some people say that IQ has more effect later in life I do not know if this is true or not.

    As far as IQ testing proving Gardner wrong you cannot prove that because you cannot prove that IQ test are measuring general intelligence and not specialized intelligence it could be measuring several specialized intelligences at once.

    Gardner cannot be the null hypothesis for g because g was around long before Gardner. However, g can be the null for Gardner. In support of specialized intelligence, you have autistics, savants and brain damaged people. General intelligence labels these people as abnormal somehow, this explains everything.

  11. I do not agree with “fundamental equality”. Nor do I believe in specialized intelligence as outlined by Gardner. But I do think some kind of specialized intelligence is more likely then not. If you can even call it intelligence.

  12. Sorry, I seem to have expressed myself a little badly. It’s not IQ testing that can give you a decision between Gardner or g, it’s a different experimental setup.

    The only thing you need to measure in order to decide between Gardner or g is a correlation between the performance in different specialized cognitive tasks.
    So if you test musical, spatial, mathematical, kinesthetic, etc. ability in many individuals you get two different predictions, depending on whether it’s g or Gardner:

    Either there is a factor that is common to all those abilities. Then people would tend to be either good at everything or bad at everything. There would be a correlation between different special abilities.

    Or those abilities are independent. Then there would be no clustering at top or bottom, and you couldn’t predict the performance of somebody in any subject by just looking at how well the person performed in others.

    In the first case it makes sense to talk about a general factor that ties different cognitive abilities together. g wins.
    In the second it would be, as you suggest, nonsense, because your measurement would only be an average through several statistically independent abilities. That’s Gardner.

    To sum it up: There have been studies and afaik they have pretty much all shown that there is a strong correlation in individuals between different cognitive abilities. So it makes sense to talk about a factor that ties them all together. It doesn’t make sense to talk about independent special intelligences, because they aren’t independent.

    What would have been nice about Gardner’s theory, as you say, is that it gives a better framework to place the outliers in, i.e. the savants.
    But still it’s not so much better. Even in a different framework those people are, in a statistical sense, outliers, because we simply can’t explain what’s going on there with either theory. We are poking at brains with sticks. We don’t have the tools to make sense of it.

    This is getting long again so just some quick points towards the end: I have a big problem with using the term intelligence in all of those contexts. It’s imprecise and has a strong emotional charge attached. We don’t know if an IQ test is measuring exactly what we mean with intelligence in our everyday lives. At the same time we don’t really have a comprehensive definition to give it a specific meaning in a scientific context either (unless you count the operational one: Intelligence is what IQ tests measure).

    Since it’s already that long, some more wild musings in favor for specialized intelligence: It could still be the case that there are several specialized intelligences at work in the brain (savantism points towards it), just that Gardner divided them up the wrong way, along the lines of behavioral tasks instead of i.e. neuroanatomical (actually “neurosystemic” would be the better word, if it existed) criteria.
    But I’m crossing the line from scientific to speculative, since I can neither produce such a division, nor do I have a testing method for such a hypothesis. To test the performance of specific neuronal systems through fitting behavioral tasks is a kind of tough nut to crack…

    That’s why I better end it here ;)

  13. I always thought multiple intelligences as defined by Gardner was unlikely. It sounds much too high level. I did not know that Gardner was so totally discredited do you have a link to the study or its title?

    I think there is another difference between G and g. Gravity would be falsifiable if object fell up or did not accelerate at the same rate. It is still possible that people are elaborate automatons. I guess that could be g’s null hypothesis.

    As a computer programmer if the brain is anything like a computer program. It is likely that different parts of the brain are highly interconnected. They maybe specialized but also almost inseparable. You would also expect a high level of redundancy do to parallelism. The other thing is since we do not really have a break down how these components work we do not even know where the overlap is. Anyone component can possibly degrade the rest of the system. Long and short term memory and especially memory access time is going to determine the result of any kind of timed test.

    The kind of test I would be more interested in is one where the individual is force to learn something new and then tested on it. Allowed to access to any materials necessary. Without severer time pressure perhaps a day or two in the lab, something hard…….

    Can I back this up nope nor do I need to. That is the great thing about being a computer programmer you do not need to be constrained by reality. Of course, if g were true it would make the programmer’s job a lot easier.

  14. This “intelligence” fetishism is ridiculous and disgusting and sad. As if you couldn’t “invent” a civilization like ours with less “intelligence”. As if this species is “smarter” than any other. By the way, a human is not a “species”. Only animals and plants have species. A human is a human, in God’s image. Sorry to break the news to you, but “IQ” doesn’t exist. Never has, never will. There’s no such thing as “intelligence” just as there’s no such thing as “evolution”. It’s clear for all to see why we prosper: God works through us and with us, but not animals. Because animals aren’t like us. That’s why we eat them but they don’t eat us. See? God gave us this civilization, not people. People merely make things, but only when allowed by God, through God.

    Lord have Mercy on this “species”!

  15. American Psychological Association’s report Intelligence: Knowns and Unknowns reports that IQ can only account for 1/6 variance in income and if you use more advanced statistics this is reduced by another quarter. People treat any positive correlating as supporting g, which is kind of unfair.

    Gardner is also not alone you also have
    Fluid intelligence (Gf or gF) and crystallized intelligence (Gc or gC)
    Speed vs capacity
    Verbal, spatial, working memory, and processing speed.
    Triarchic Theory of Intelligence

    All of those expect maybe Triarchic Theory of Intelligence are more accepted then Gardner.

    Despite high school GPA being a better predictor then IQ people still, want a standardized test. They fear that only people whom attended more prestigious high schools will be admitted.

    IQ is highly over sold why can we not use 100% of our brain whenever we want. Why does only 6% of our gray matter light up during IQ tests “Human Intelligence Determined by Volume and Location of Gray Matter Tissue in Brain” and yet we use vastly more of our brain in any given day. Why do psychologists have to keep inventing better IQ test? Surely its can be nothing more than an elaborate exercise in curve fitting. Without a care in the world about what the test are measuring because “all intelligence is fundamentally the same”

    PS mk you are ignorant.

  16. hi jimmy
    This article would be the kind of experiment I described in the post before.

    Johnson, Wendy and Bouchard, Thomas J. Jr. (2005). “The structure of human intelligence: It is verbal, perceptual, and image rotation (VPR), not fluid and crystallized.” Intelligence, 33 393–416.

    Falsifying g is simple.
    Null hypothesis: There is no correlation between a wide range of different cognitive tasks in an individual.
    If one is found, there is a g. If not, there is not.
    To iterate it again: If there is a correlation in performance among a wide range of cognitive tasks, there is (I’d even say per definition) a g, a general factor that within some bounds can help to describe that performance.
    And that’s all I’m claiming.
    That there is evidence supporting the fact that it’s there.
    And that, since it’s probably there, it makes sense to talk about it.

    How, whether we are elaborate automatons or not, could be a null hypothesis to g is something you’d have to explain to me in a little more detail.

    The social realities and all the talk about “intelligence” and IQ are a totally different thing from the lab findings.
    I’m very doubtful about the roles g, IQ and intelligence play in our lives. Since IQ has been around for 80 years now and has significantly been shaping social realities since then (especially in the US with its “testing culture”), it’s difficult to make any reasonable comments about intelligence that aren’t totally muddled up by subjective bias and multiple social factors. Statistics can alleviate those failings, but still fail to control for everything (and per definition for the unexpected or even the unassumed).

    Yes, IQ is highly oversold. Especially in the US. We Europeans don’t care about IQ testing that much anyway.

    Still it’s no wonder that only a fraction of our brain lights up during IQ testing. Most of it is dedicated to sensory-motor-control functions and not the abstract tasks you are engaged in during an IQ test. That’s also the reason why brain/mass ratio is so much more important than pure brain size as an indicator for intelligence among different species of animals. You just need some amount of brain mass simply to control your body. The bigger the body (think of a whale) the more brain you need just to be able to control all its functions it. All that’s called intelligence just sits on top of it.

    Inventing better IQ tests seems to be making them more g-centric recently. Basically they try to capture as much of the general factor they want to measure in their tests as possible, which seems to be a good thing. If you accuse them of curve fitting you’d have to tell me which curves they fit and to what they fit them to because about that really have no good idea.

    Hi MK

    Just to make the terms clear: With “IQ doesn’t exist”, do you mean existence in an epistemological, purely ontological or merely physical sense?


    Asperger Syndrome is the link between “nerds” and poor social skills. AS makes people more persistent (which leads to being more successful in most things that interest them) and also less socially skillful.

    I’m sure that they will define dozens of other neurological differences which lead to skills in some areas and disadvantages in others.

    Such neurological differences don’t however prove the “multiple intelligences” idea, in most cases such people merely have tendencies towards broad areas of work or study. This means that their general intelligence and ability to persist in practice and study are the factors that determine success.

    As for measuring adults, that is far too late. If you spend 10 years practicing something then you will be better at it.

    I think that we need some tests on babies and long term studies starting at a young age.

  18. Are you trying to tell me psychiatrists are doing real science? Maybe they are really statisticians, or philosophers who use statistics? Ha Ha Ha I don’t believe it.

  19. Don’t you worry, since the psychiatrists are still only in charge of the clinical side of the matter, they aren’t doing that much science here.
    Philosophers using statistics? Sounds like an oxymoron…
    And since in Psychology you somehow have to wring some data out of that hellishly inconsistent and complex thing that people are, I’d say yes, they basically are statisticians…

Leave a comment

No trackbacks yet.