I've only taken an iq test once. Never trained for it. i came out 110. I've never had an issue with that. Being comfortable in your intelligence and what you know and don't know is liberating. I'm always the guy in the meetings asking questions when I think people are afraid to look stupid... I also meet people claiming to be 140. They must be tapping some serious G.
I have seen some estimates of the impact of training for IQ tests which suggested an impact of ~ half a standard deviation (~ 8 points) which is also consistent with training against g related tests - SAT/ACT. Now the populations in question may not be fully naive, so training may have somewhat higher impact for some populations. The bigger issue is g/IQ is a essentially capability that is valueless unless you use it. I have run across a lot of 'smart' people who seemed to have gotten out of the habit of using their intelligence.
If anything, in my own experience I’ve found that really intelligent people can be prone to using that intelligence to engage in reality-bending mental gymnastics, in order to confirm their preexisting biases.
Justification is far more common that reasoning, independent of intelligence. The smarter you are and the more you know, the more complex the justifications become. The intellectual humility to keep your beliefs as provisional and subject to correction via more data and reasoning is quite uncommon.
Probably the result of an imbalanced training regime. People tend to study verbal because economic incentives so they get verbal tilts. The real intelligence is given by the untrained subtest, usually spatial
It sounds like throughout this piece, you're using "g" to mean "measured performance" and "intelligence" to mean "latent g." Intelligence researchers already understand that an IQ score, even averaged across multiple administrations, is not a direct measure of what they refer to as g. They understand that averaging scores only eliminates random error, not non-g contamination. If g, the way they use the term, were trainable, we would observe stronger correlations across diverse tasks after training, not merely higher scores; in other words, practicing digit span would improve spatial rotation, and practicing vocabulary would improve matrix reasoning. But research has roundly debunked the brain-as-muscle metaphor. Gains from practicing tasks downstream of g do not propagate upstream to g and back downstream to very different cognitive tasks - that is what is meant by calling g "not trainable."
And an assumption that's doing a lot of work here (in service of your continued "wordcel = midwit" narrative) is that verbal subtests are trainable - and not only that, but the whole reason wordcels even score so high is because of such training. I'm skeptical of that. It's true that people who are underexposed to written language score lower than their genetic potential would allow. Mass schooling and the Internet have basically solved this problem in the modern West, Steve Sailer has argued. This has only made verbal testing more g-loaded because variation in innate verbal ability is all that is *left* to explain differences in performance. People clearly hit ceilings; once people reach a point of language saturation, there's no plausible amount of extra exposure that produces overperformance the way deprivation produces underperformance.
You speculate that people with verbal tilts probably trained with a verbal focus, but how? It's well established that verbal subtest performance (vocabulary and analogies) and verbal scores on the SAT and GRE are notoriously hard to improve. It's easier to get familiar with, say, the kinds of transformations you'll expect to see in Raven's matrices (such as XOR problems) or the math problems you'll expect to see on the Wonderlic (such as the "trains leaving at different speeds" problems) than to get familiar with enough of the English language to guarantee a high score. Reading widely and drilling vocabulary flash cards don't cut it, nor do "finding the main idea" and "making inferences" lessons.
I can't speak for everyone, but I got 99th percentile scores on the SAT and ACT with verbal tilts, and I can't say I ever thought, "Ah, that's a word I've seen in my vocabulary study" or "That strategy I learned really helped me find the main idea of that passage." Taking reading comprehension tests has always felt like, "Of course that's the right answer; how does anyone get this wrong?" My "training regimen" before I took those tests was... starting to read at age 3, to the surprise of my parents, and going on to read whatever captured my interest - popular middle grade book series, Wikipedia, forums, blogs, women's magazines, certainly no Tolstoy or Heidegger. I got the highest Lexile score my junior high school had ever seen, a "college level" score for someone who seldom read books beyond an eighth grade level! A case like this makes sense if verbal ability reflects the underlying capacity to absorb more information per exposure and to build and manipulate linguistic structure in working memory; it makes no sense if it's the product of training.
As you mentioned, the 12 year olds who demonstrated high school math skills on the SAT clearly needed to learn high school math concepts explicitly at some point, as they didn't derive the procedures themselves. But there's no clear equivalent of "learning high school math from a book" for those who ace verbal. Their advantage shows up automatically, early, and across contexts. (You could say adults' discussions and posts online were where I got my extra vocabulary and background knowledge from, but it makes more sense to say I was drawn to them *because* of my prior verbal ability than to call them "training" for tests.)
Verbal ability is bottlenecked by domain-general abstraction ability and working memory capacity, as my past two Substack posts discuss. If there were a way for readers to score well on reading tests despite their limited abstraction ability and poor WM capacity, well, public schools desperate to get their reading scores up would be teaching it. Spatial tasks are less g-loaded because the domain-specific variance in them isn't reflected as much in performance on other tasks, even other non-trained non-verbal tasks such as digit span. This isn't to say spatial ability is fake or irrelevant, just that a person can be excellent at mental rotation without being especially strong at abstract reasoning more broadly, and the cognitive resources recruited for mental rotation are not broadly recruited for non-spatial tasks.
A stereotype of verbal-tilts is that we emptily shuffle fancy words around, detached from understanding of the real world. In reality, the valuable ability to track whether there's actual structure and substance behind what sounds like highfalutin jargon *is* a matter of verbal intelligence. And just as sophisticated rhetoric built on faulty premises isn't smart, an elaborate numerical exercise built on faulty premises isn't as smart as it sounds either. Your "verbal tilt = trained midwit" argument must postulate a mechanism by which verbal scores can be inflated through training, separate from the practice effect inflating test performance overall. If that mechanism doesn't exist, does your view still hold up?
P.S. The papers you cite to indicate "differences in schooling, motivation, and need for cognition have small effects on g scores" indicate nothing of the sort. One says leisure activity is associated with cognitive ability in old age, with the authors concluding that they can't say leisure activity enhances ability, only that it might reflect greater preservation of earlier cognitive ability. The other says greater "need for cognition" as a personality trait is more correlated with greater fluid intelligence scores than with crystallized intelligence, while the reverse is true for "intellectual engagement" as a personality trait. This doesn't suggest that schooling or motivation *produce* one's Gf score; I would argue that being high in Gf produces a higher need for stimulation because someone with higher WM capacity needs more complex material to fully occupy their attention, as my last article touches on. I know you've said "correlation does not equal causation" is a "midwit line," and it sure is - but that doesn't give you license to make unsupported causal inferences and then call people midwits for pointing that out.
You associate IQ levels with how quantitatively one sees the world, yet your only reason for thinking training gives an inaccurately high measurement of IQ is that some people with high IQ scores qualitatively seem not as smart as their scores indicate. Isn't the more obvious interpretation that training in fact raises IQ?
There was a tall and a short man, and both were blind. The tall man could easily reach the top shelf, while the short man could not. One day the short man decided to train so he could reach the top shelf through a secret technique. Then he claimed to be as tall, or even taller than, the tall man. Many blind people believed it. One day, however, the two men were out and bumped into a date tree. "Why don't you grab a date for me", the tall man asked the short man, "since you are so tall, even taller than me, many say". The short man's technique failed; it only worked inside. The tall man easily grabbed a date and concluded that his friend was in fact short, and a try-hard.
Then you come along and say "that's crazy, Isn't the more obvious interpretation that training in fact raises height?"
The analogy doesn't work. Height, at least for adults, just doesn't respond to training. But this is a fact about height, not about human characteristics generally. Some human characteristics, like math skills, clearly do respond to training. You seem to be implicitly assuming that IQ is more like height, that it doesn't respond to training, and your whole argument hinges on this assumption. And I don't think this assumption is necessarily true. Why wouldn't IQ respond to training to at least some extent?
The analogy does work, because intelligence is like height. IQ and g do respond to training; that is the point of my article. IQ and g are like indoor reaching technique; intelligence is an underlying property of the brain that correlates with that technique.
Do you have any evidence to support that idea? Because right now all you have presented is a naked assertion, not an argument or any kind of evidence. And that is very unconvincing.
As I pointed out, the only "evidence" you mentioned is your qualitative, intuitive impression that some of the people you have interacted with don't seem as smart, or as quantitative, as their claimed IQ suggests. That might be because they are just lying or going off unreliable online IQ tests. Or that might be because you aren't as good at reading people as you seem to think you are. But it is certainly not something anyone should accept as a reason for thinking intelligence can't be trained.
I suspect what is actually going on here is a much simpler story - you want to feel good about yourself, you have failed in life by whatever metrics you care about, and you want to make yourself feel better, so you tell yourself you have this unalterable intelligence that makes you better than others despite your lack of success. It's less painful than realizing you might have lost some of the intelligence you once had.
I've only taken an iq test once. Never trained for it. i came out 110. I've never had an issue with that. Being comfortable in your intelligence and what you know and don't know is liberating. I'm always the guy in the meetings asking questions when I think people are afraid to look stupid... I also meet people claiming to be 140. They must be tapping some serious G.
I have seen some estimates of the impact of training for IQ tests which suggested an impact of ~ half a standard deviation (~ 8 points) which is also consistent with training against g related tests - SAT/ACT. Now the populations in question may not be fully naive, so training may have somewhat higher impact for some populations. The bigger issue is g/IQ is a essentially capability that is valueless unless you use it. I have run across a lot of 'smart' people who seemed to have gotten out of the habit of using their intelligence.
If anything, in my own experience I’ve found that really intelligent people can be prone to using that intelligence to engage in reality-bending mental gymnastics, in order to confirm their preexisting biases.
Justification is far more common that reasoning, independent of intelligence. The smarter you are and the more you know, the more complex the justifications become. The intellectual humility to keep your beliefs as provisional and subject to correction via more data and reasoning is quite uncommon.
What about spiky intelligence profiles? Is the phenomenon relevant at all or too rare?
Probably the result of an imbalanced training regime. People tend to study verbal because economic incentives so they get verbal tilts. The real intelligence is given by the untrained subtest, usually spatial
I was thinking of autism but I guess it applies either way
It sounds like throughout this piece, you're using "g" to mean "measured performance" and "intelligence" to mean "latent g." Intelligence researchers already understand that an IQ score, even averaged across multiple administrations, is not a direct measure of what they refer to as g. They understand that averaging scores only eliminates random error, not non-g contamination. If g, the way they use the term, were trainable, we would observe stronger correlations across diverse tasks after training, not merely higher scores; in other words, practicing digit span would improve spatial rotation, and practicing vocabulary would improve matrix reasoning. But research has roundly debunked the brain-as-muscle metaphor. Gains from practicing tasks downstream of g do not propagate upstream to g and back downstream to very different cognitive tasks - that is what is meant by calling g "not trainable."
And an assumption that's doing a lot of work here (in service of your continued "wordcel = midwit" narrative) is that verbal subtests are trainable - and not only that, but the whole reason wordcels even score so high is because of such training. I'm skeptical of that. It's true that people who are underexposed to written language score lower than their genetic potential would allow. Mass schooling and the Internet have basically solved this problem in the modern West, Steve Sailer has argued. This has only made verbal testing more g-loaded because variation in innate verbal ability is all that is *left* to explain differences in performance. People clearly hit ceilings; once people reach a point of language saturation, there's no plausible amount of extra exposure that produces overperformance the way deprivation produces underperformance.
You speculate that people with verbal tilts probably trained with a verbal focus, but how? It's well established that verbal subtest performance (vocabulary and analogies) and verbal scores on the SAT and GRE are notoriously hard to improve. It's easier to get familiar with, say, the kinds of transformations you'll expect to see in Raven's matrices (such as XOR problems) or the math problems you'll expect to see on the Wonderlic (such as the "trains leaving at different speeds" problems) than to get familiar with enough of the English language to guarantee a high score. Reading widely and drilling vocabulary flash cards don't cut it, nor do "finding the main idea" and "making inferences" lessons.
I can't speak for everyone, but I got 99th percentile scores on the SAT and ACT with verbal tilts, and I can't say I ever thought, "Ah, that's a word I've seen in my vocabulary study" or "That strategy I learned really helped me find the main idea of that passage." Taking reading comprehension tests has always felt like, "Of course that's the right answer; how does anyone get this wrong?" My "training regimen" before I took those tests was... starting to read at age 3, to the surprise of my parents, and going on to read whatever captured my interest - popular middle grade book series, Wikipedia, forums, blogs, women's magazines, certainly no Tolstoy or Heidegger. I got the highest Lexile score my junior high school had ever seen, a "college level" score for someone who seldom read books beyond an eighth grade level! A case like this makes sense if verbal ability reflects the underlying capacity to absorb more information per exposure and to build and manipulate linguistic structure in working memory; it makes no sense if it's the product of training.
As you mentioned, the 12 year olds who demonstrated high school math skills on the SAT clearly needed to learn high school math concepts explicitly at some point, as they didn't derive the procedures themselves. But there's no clear equivalent of "learning high school math from a book" for those who ace verbal. Their advantage shows up automatically, early, and across contexts. (You could say adults' discussions and posts online were where I got my extra vocabulary and background knowledge from, but it makes more sense to say I was drawn to them *because* of my prior verbal ability than to call them "training" for tests.)
Verbal ability is bottlenecked by domain-general abstraction ability and working memory capacity, as my past two Substack posts discuss. If there were a way for readers to score well on reading tests despite their limited abstraction ability and poor WM capacity, well, public schools desperate to get their reading scores up would be teaching it. Spatial tasks are less g-loaded because the domain-specific variance in them isn't reflected as much in performance on other tasks, even other non-trained non-verbal tasks such as digit span. This isn't to say spatial ability is fake or irrelevant, just that a person can be excellent at mental rotation without being especially strong at abstract reasoning more broadly, and the cognitive resources recruited for mental rotation are not broadly recruited for non-spatial tasks.
A stereotype of verbal-tilts is that we emptily shuffle fancy words around, detached from understanding of the real world. In reality, the valuable ability to track whether there's actual structure and substance behind what sounds like highfalutin jargon *is* a matter of verbal intelligence. And just as sophisticated rhetoric built on faulty premises isn't smart, an elaborate numerical exercise built on faulty premises isn't as smart as it sounds either. Your "verbal tilt = trained midwit" argument must postulate a mechanism by which verbal scores can be inflated through training, separate from the practice effect inflating test performance overall. If that mechanism doesn't exist, does your view still hold up?
P.S. The papers you cite to indicate "differences in schooling, motivation, and need for cognition have small effects on g scores" indicate nothing of the sort. One says leisure activity is associated with cognitive ability in old age, with the authors concluding that they can't say leisure activity enhances ability, only that it might reflect greater preservation of earlier cognitive ability. The other says greater "need for cognition" as a personality trait is more correlated with greater fluid intelligence scores than with crystallized intelligence, while the reverse is true for "intellectual engagement" as a personality trait. This doesn't suggest that schooling or motivation *produce* one's Gf score; I would argue that being high in Gf produces a higher need for stimulation because someone with higher WM capacity needs more complex material to fully occupy their attention, as my last article touches on. I know you've said "correlation does not equal causation" is a "midwit line," and it sure is - but that doesn't give you license to make unsupported causal inferences and then call people midwits for pointing that out.
You associate IQ levels with how quantitatively one sees the world, yet your only reason for thinking training gives an inaccurately high measurement of IQ is that some people with high IQ scores qualitatively seem not as smart as their scores indicate. Isn't the more obvious interpretation that training in fact raises IQ?
There was a tall and a short man, and both were blind. The tall man could easily reach the top shelf, while the short man could not. One day the short man decided to train so he could reach the top shelf through a secret technique. Then he claimed to be as tall, or even taller than, the tall man. Many blind people believed it. One day, however, the two men were out and bumped into a date tree. "Why don't you grab a date for me", the tall man asked the short man, "since you are so tall, even taller than me, many say". The short man's technique failed; it only worked inside. The tall man easily grabbed a date and concluded that his friend was in fact short, and a try-hard.
Then you come along and say "that's crazy, Isn't the more obvious interpretation that training in fact raises height?"
The analogy doesn't work. Height, at least for adults, just doesn't respond to training. But this is a fact about height, not about human characteristics generally. Some human characteristics, like math skills, clearly do respond to training. You seem to be implicitly assuming that IQ is more like height, that it doesn't respond to training, and your whole argument hinges on this assumption. And I don't think this assumption is necessarily true. Why wouldn't IQ respond to training to at least some extent?
The analogy does work, because intelligence is like height. IQ and g do respond to training; that is the point of my article. IQ and g are like indoor reaching technique; intelligence is an underlying property of the brain that correlates with that technique.
Do you have any evidence to support that idea? Because right now all you have presented is a naked assertion, not an argument or any kind of evidence. And that is very unconvincing.
There is evidence in the article. If you found that "unconvincing", it might not be for your current g level.
As I pointed out, the only "evidence" you mentioned is your qualitative, intuitive impression that some of the people you have interacted with don't seem as smart, or as quantitative, as their claimed IQ suggests. That might be because they are just lying or going off unreliable online IQ tests. Or that might be because you aren't as good at reading people as you seem to think you are. But it is certainly not something anyone should accept as a reason for thinking intelligence can't be trained.
I suspect what is actually going on here is a much simpler story - you want to feel good about yourself, you have failed in life by whatever metrics you care about, and you want to make yourself feel better, so you tell yourself you have this unalterable intelligence that makes you better than others despite your lack of success. It's less painful than realizing you might have lost some of the intelligence you once had.