EQ > IQ

Steelmanning the underrated intelligence

IQ has a reputation for being the hard-nosed psychometric option. Those who are willing to accept facts over feelings will admit its supremacy. But this is stolen valor! Emotional intelligence is fundamental to human evolution and the Good Life, but can’t be measured well. IQ is the opposite. It was designed to boil down information from our best psychometric instruments, but we don’t understand how it relates to intelligence. Ease of measurement should not be confused with importance. Particularly when the relation between IQ and intelligence is not understood, something its most ardent supporters admit.

Forms vs. measures

Traits exist on two levels. The first is something like their platonic form. An idealized version that exists as an abstraction. Think of the Pythagorean Rule: a^2 + b^2 = c^2. No matter how one positions the hypotenuse (c), if you measure the lengths of the sides, they always follow this formula. It is unsurprising that there are rules that generate observed data in geometry. The situation in psychology is messier, but there are still underlying rules. Ducklings imprint on a caregiver within 36 hours of hatching. Your degree of Extroversion predicts how you act at parties and on the company's Slack. The likes of Freud and Maslow looked for even deeper explanations for our behavior (e.g. Id, Ego, and Super-Ego); whether or not they correctly identified them is beside the point.

The conceit of psychometrics is that underlying traits can also be measured. Psychological instruments—typically questionnaires or tests—can map an individual onto some theorized axis, such as Extroversion. Naturally, this score is only an approximation of the form. But it allows researchers to deal in the world of statistics rather than verbally grappling with forms (though even quantitative research must be communicated in that messy domain).

The measurement process is incredibly lossy. Let’s say you want to measure someone’s Openness to Experience. To do this, you must develop a questionnaire that includes items you believe are relevant. This is ultimately down to human judgment. Confidence can be gained by comparing Openness to Experience scores to other potential correlates, such as brain connectivity or the number of stamps on a person’s passport. But in the end, the fidelity of the approximation is impossible to know. There is no ground truth to compare against.

Facts about instruments can lead to confusion about forms. Imagine if all our photos of Venus were from Galileo’s sketches, but we had photos of Mars from the Hubble. One might say, “Look at how crisp the edges of Mars are. Look at the red! Surely this is the best planet.” But planets exist outside what is measured by our instruments, as does intelligence. With that in mind, let’s define IQ and EQ.

General Factor of Personality (GFP)

Instead of talking about Emotional Intelligence, I will discuss the GFP. I hope you’ll excuse the bait and switch; however, it seems justified. In a meta-analysis, Van der Linden et al. find a 0.85 correlation between GFP and Emotional Intelligence calling the two “very similar, perhaps even synonymous.” Additionally, as we will see, the derivation of GFP and IQ are analogous.

The GFP has a couple of definitions. Let’s start with those that deal with its statistical nature. You can administer any broad personality test and then do dimensionality reduction to find the most significant “latent factor” in the data. This is the GFP. Holy wars are fought over the best statistical methods to use, but the idea is that surveys may include 100 items. This leaves you with 100 bits of information about someone, which is unwieldy. Dimensionality reduction finds a way to score the survey, which gives you one number that is maximally informative about how someone answered each question in the survey. Different questions will be weighted more or less, and by analyzing those weights, you can then describe what this latent factor is all about. It turns out that, qualitatively, this looks pretty similar if you do this exercise with any set of questions or any group of people. The dominant latent factor is always something like: “Honest, thoughtful, and kind. Someone you would want on your team.”¹ For the psychometric nerds, the same consistency holds (though less so) for the first five latent factors, AKA the Big Five.

Visualizing forms from language

I started this blog because of the untapped potential to deal more directly with the Platonic form of personality by using Natural Language Processing. The theoretical basis of the Big Five is the Lexical Hypothesis. This postulates that personality is really a character judgment, and millions of those are made every day. These in turn are communicated with language, and so the contours of these judgements should be reflected in the entire set of words we have available to judge one another. By quantifying this space—the contours of gossip writ large—we can find the latent factors of personality (the Big Five). Below are 100 personality adjectives plotted on the two most important latent factors.

Factors 1 and 2 are produced using PCA, a method to distill the personality information onto the smallest number of axes. For more information on how to get these from word vectors, see my paper Deep Lexical Hypothesis. You can also reproduce this result (and much more) using my code, which runs for free on Google Colab.

The words on this plot were forged over thousands of years of gossip. Modern language models absorb a substantial amount of the internet and most books ever written. Extracting the implied relationship between words gives us front-row seats to the world of forms. But this is just a visualization. The real prize is understanding what Factor 1 above represents—the rule that generated it².

At first glance, it may appear to be simply good vs bad, but we must think about how the data came to be. Language structure represents the view of society, the type of person that others like to deal with. As such, my contribution was to stare at the axis and relate it to the Golden Rule. Are you considerate, pleasant, and bright? Do you refrain from being abusive, intolerant, or uncooperative? At its core, that indicates a tendency to live the Golden Rule. Or at least that is my contention. There is considerable debate within the field.

Another definition is Social Effectiveness. For me, this fails the Hitler test. The Führer was certainly effective but not very nice. As per the words on the chart, the trait involves caring about others’ welfare, so it can’t just relate to efficacy. The implied goal is to lift others up, not look out for number 1.

It’s also much more than being Agreeable or pleasant. Notice that intelligent and knowledgeable are both associated with this factor. Applying the Golden Rule is, in fact, a subtle operation that requires modeling your own mind as well as that of others. Doing this allows individuals to reach win-win agreements (an act on which human society is built). This is not a simple task, and to whit, both gullible and naive are roughly neutral. (Though note that being naive is more forgivable. Fool me once, shame on you, fool me twice…) The negative words are also informative, with many variations of being intolerant or abusive—failing to consider others or considering them to cause pain.

So, going forward, when I say GFP, I mean the Golden Factor of Personality. A robust character trait that requires modeling your mind and that of others. You revel in their successes; turning the other cheek is optional³.

Darwin 🤝 Jesus

Because GFP is defined by language, there is already a theoretical connection to both the Big Five and gossip. Characterizing it as the Golden Rule also links it to evolution and morality. Consider Darwin’s understanding of the role of language in how we became human. From The Descent of Man:

After the power of language had been acquired, and the wishes of the community could be expressed, the common opinion of how each member ought to act for the public good, would naturally become in a paramount degree the guide to action.

He relates this process to the Golden Rule:

The moral sense perhaps affords the best and highest distinction between man and the lower animals; but I need say nothing on this head, as I have so lately endeavored to shew that the social instincts—the prime principle of man's moral constitution (The Thoughts of Marcus Aurelius,' p. 139.)—with the aid of active intellectual powers and the effects of habit, naturally lead to the golden rule, "As ye would that men should do to you, do ye to them likewise;" and this lies at the foundation of morality.

Once humans obtained language, there was selective pressure for moral behavior (enforced via gossip), which inexorably led to the Golden Rule. As the GFP is derived from gossip (personality adjectives), it would be surprising, therefore, if it were not shaped like the Golden Rule⁴. Darwin, once again, vindicated.

Nor is it surprising that Jesus condensed the 613 rules in the Old Testament to “do unto others as you would have them do unto you.” One can even think of it as a dimensionality reduction of the Old Testament. What is the one latent rule that generates all the others? Don’t kill or commit adultery are merely object-level implementations of this higher law, the Golden Rule. But importantly for Jesus, this is deeper than a commandment; it is a divine spiritual truth. Your soul was forged such that you cannot be at peace while you fail to treat others the way you would like to be treated. This is the Living Water that He offers. And it is strikingly close to Darwin’s claim that our minds must have been forged by this foundational tenant of morality.

Here a caveat is essential, for Darwin’s and Jesus’s version of the Golden Rule is more extreme than mine. Theirs requires one to “turn the other cheek” and to do good to those that despitefully use you. I find this hard to square with Darwin’s belief in the superiority of English culture, which did not conquer the world on this principle. Nor with the timeline of humanity. Language has existed for a long time and has been used to enforce morality; what was the game-theoretical basis of that morality⁵? If it was something like “treat others like you would like to be treated,” then that can explain why it seems to have such deep psychological hooks and is part of most religious traditions. It has been part of the fitness landscape for many millennia, forging our minds well before it was articulated.

At any rate, I am confident that all of these forms are deeply related, if not the same. The GFP is characterized by considering other people, which Jesus articulated as a spiritual truth, and Darwin as an evolutionary force. It is a formula that generates data in many domains. A rule that neatly cuts at the joints of human nature, evolution, and language.

A final note about the other definition of the GFP: social intelligence. This is also related to human evolution via Dunbar’s social brain hypothesis. So, even if one adopts an alternative definition of GFP, it is still fundamental to how we became human. (Though the connection to morality is less clear. See: Machiavelli.)

Measurement

The most common way to measure personality is self-report. Unsurprisingly, people are not good judges of whether they are a good person⁶. The other option is to characterize EQ as an ability that can be tested, such as whether someone can recognize emotion (eg. anger, sadness) based on facial expression (particularly cropped photos of eyes). The aforementioned Van der Linden meta-analysis found the GFP to correlate r = 0.85 with EQ_survey, but only r = 0.28 between GFP and EQ_ability (lower even than the 0.36 correlation between GFP and IQ). Such a yawning gap belies how much scores reflect instruments, and we only get a very noisy picture of EQ or GFP. Still, these scores are reasonably correlated with real-life outcomes like going to jail or keeping a job.

The general factor of intelligence

In 1904, Charles Spearman proposed that there was a single general factor of intelligence called the g-factor, or just g for short. As such, a single trait would account for much of the performance on many different types of tasks. Perhaps surprisingly, this turns out to be the case and is one of the most replicated findings in psychology. That is, someone’s vocabulary size correlates with their reaction time correlates with their ability to rotate shapes in their mind. To calculate g, simply perform dimensionality reduction on as many different intelligence-related tasks as possible. Like the GFP and personality surveys, g is the first latent factor. The difference is that g is calculated from True/False test questions, and the GFP is calculated from personality items or adjectives. Typically g captures a bit less than 50% of the variance in the data (i.e. individual variation), which is on par with the GFP⁷.

My previous job involved scoring VR tests to measure the degree of concussion in athletes or Alzheimer’s progression in seniors. These are two very different populations with very different problems, yet both include a marked drop in g. Indeed, it’s hard to make clinical decisions without knowing someone’s baseline IQ. More generally, IQ is the single best psychometric predictor of a host of positive life outcomes. For income, the relationship is r = 0.3:

The slope is noticeably positive. Still, it’s important to keep the nature of real-world relationships in perspective; there’s quite a bit of noise.

But what is IQ? Well, I will refer you to Arthur Jensen, who literally wrote the book The g Factor: The Science of Mental Ability. In it, he argues that g is a real, measurable phenomenon and the single most important factor in cognitive ability. Further, that g is largely hereditary and is not significantly influenced by environmental factors like upbringing or education. A decade later, he wrote the fantastic⁸ Clocking the Mind. He explains his interest in Reaction Time (RT) thus:

“The fact of a significant relationship between RT and psychometric intelligence has at least two immediate implications for theory and research on intelligence. First of all, it directly contradicts a widespread conception in contemporary psychology that our current standard tests of intelligence measure nothing but a particular class of specific knowledge and acquired cognitive skills or strategies for dealing with certain types of problems generally considered intellectual…Secondly, if there is a correlation between individual differences in RT and intelligence, it seems that research on the much simpler information processing phenomenon, RT, would lead more readily to an adequate theoretical account of it than would attempts to theorize directly about the much more complex phenomenon of intelligence.”

Jensen is perhaps the most informed true believer in g there is. Still, he spent years trying to establish a connection between g and RT so that we could start to understand the nature of g. We have no idea what it is in the world of forms!

Or consider another giant in the field, Ian Deary. He may be the most published psychometrician alive today. And, like Jensen, a fan of IQ⁹. As a neurosurgeon who studied psychiatry and then became an intelligence researcher, nobody is better poised to explain what IQ is. And yet, in an interview with René Mõttus (Editor in Chief of the European Journal of Personality), he says:

“If my skepticism about theory in psychology was an undertone in my book Looking Down on Human Intelligence, then I failed. Because it was meant to be an overtone. Even skepticism might be too mild. I’m just very critical of theory as it appears in psychology and specifically as it appears in individual differences and even more specifically in intelligence research. Now when one says that it’s easy for [detractors] to say “well god he’s just a boring dustball.” And I’m not. I think there are grown-up theories in other branches of science—not just the hard sciences, in biology as well. I am interested in intelligence and personality in the phenotypes, which of course involves a lot of data. I’m interested in their ability to predict variation in things that happen then subsequently—their predictive validity. I spent a lot of time as well looking at mechanisms. That’s what I meant “looking down on human intelligence.” I mean I’m really interested in trying to explain individual differences in cognitive test scores. So I’m interested in all the things that theorists say they’re interested in: prediction, phenotypic clarity, reductionism that is understanding things. If we think of names of theories [of intelligence] I think none of them deserve that label. Why is that? A theory would typically be a network of constructs which people are tying together in an original way to try and predict things in terms of working out a mechanism. I think the construct formulation is sometimes lacking. I think the empirical associations aren’t always there. Possibly most of all, tying them to real things, whether SI or units in biology is lacking as well. People when putting together theories often use sky hoops rather than cranes…a crane is genuinely something rooted on the ground with which you can do lifting, a sky hoop is just a promissory note. I suppose what I’m saying is you can’t just make stuff up. It’s got to be tied to real things.”

Deary is asking for something in the world of forms that can explain the statistical power of g. What is the fundamental rule that generates such a dominant latent factor? Synapse speed? Brain organization? Ability to accomplish one’s goals through abstract reasoning? Like Jensen, he believes there are no satisfactory accounts.

Plato? Aristotle? Socrates? Morons!

Never go in against a Sicilian when death is on the line

To my mind, the best indicator that g is tied to real things is that it is more related to nature than nurture. Unintuitively, this is evidence against g being fundamental to intelligence. Let me explain.

If a trait is fit, then it will increase every generation following the Breeder’s Equation,

\(Δz = h^2 * β,\)

where Δz is the change in the phenotype, h^2 is the narrow-sense heritability (additive genetic contribution), and β is the selection gradient. The h^2 of g is around 0.6¹⁰. Now to estimate the selection gradient. A common definition of intelligence is given by researcher Linda Gottfredson: “Intelligence is a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly, and learn from experience.” If this is substantially captured by g, then g should be correlated with fitness. Ostensibly having sex and the survival of one’s children are some of the goals that intelligence helps one accomplish.

In the days of yore, how many more surviving children would a 130-IQ man have than a 100-IQ man? As an intuition pump, let’s say that our institutions broke down, and it was every man for himself. Do you think the survivors would tend to be smarter? To a noticeable degree? Life in antiquity was a step in that direction. Let’s be conservative and say the relation between g and fitness was r = 0.1. Visually indistinguishable from noise, you wouldn’t notice the trend in day-to-day life:

Plugging these values in, we get Δz = 0.6 * 0.1 = 0.06 standard deviations per generation. Over 2,000 years (80 generations), the population average would increase by 4.8 standard deviations or 72 IQ points. Or projecting back in time, the Ancient Greeks should have an average IQ of 28. This follows from the estimate of 0.1. Maybe it was less, but this seems to be at odds with the characterizations of g as universally useful. One can tinker with the model, but the point is that it’s hard to claim IQ is noticeably fit and Plato, Aristotle, and Socrates were not morons (a la Vizzini)¹¹.

This leads me to the same disappointment as Deary. I find it troubling that g is a general factor of…something. Maybe test-taking? Seems like it has to be more, but it’s unclear. At least for now, it’s just a sky hoop, a construct defined statistically by reference to other correlates (test scores, income, trait impulsivity, etc.).

Selection on GFP

It is only fair to apply the same test to GFP. The narrow-sense heritability of GFP is a bit lower, and estimations are much more variable from study to study. For example, The genetics and evolution of the general factor of personality models all of the genetic contribution to GFP as non-additive (h^2 = 0). This is interpreted as evidence that it has been under recent natural selection. At any rate, it sidesteps the implication that our ancestors were deficient in the GFP.

Other studies find h^2 to be as high as 0.5¹². In that case, GFP maximalists also have to bite the bullet that the Greeks were grossly deficient in (emotional) intelligence. I’m willing to do roughly that. I have argued that because of social pressure to live the Golden Rule humans discovered introspection. This produced inner life, recursive thought, and the ability to perceive the world of forms. It was a phenomenological phase change before which there was no sapience.

As for the Greeks, I don’t think they were emotional morons. When estimating the selection gradient (β), one would not expect following the Golden Rule (or introspection) to be universally fit. Ghenghis Khan was a star performer, evolutionarily speaking. Further, the narrow-sense heritability is lower than for IQ, thus mitigating the change each generation. Still, I guess there has been selection in the last 2,000 years. We may be more prone to look inward.

The claim that EQ > IQ does not ride on the truth of my quixotic theory. But I wanted to demonstrate what even small amounts of selection would look like over millennia. The world starts looking cognitively foreign, even in recorded history¹³. Any choice of a selection gradient that implies IQ is useful also implies the Ancient Greeks would be legally retarded if adopted and raised in today’s society.

Conclusion

The derivations of GFP and g are analogous: the dominant latent variable in language and tests, respectively. Because language is more fundamental to what makes us human, I think the GFP is more fundamental than g. This is separate from the fact that g, by construction, is easier to measure. Returning to the telescope analogy, comparing the predictive power of g and GFP is like comparing Mars to blurry representations of Venus¹⁴. The traits must be compared on a theoretical basis.

The Golden Rule dances through the world of forms described by the greats spanning millennia. Jesus saw it as a spiritual truth, a law etched into every man’s soul. Two thousand years later, Darwin returned to the same abstraction when grappling with the effects of language on the evolution of our morality-seeking minds. In psychometrics, it appears repeatedly as a latent factor in personality tests as the GFP. Language filters untold social interactions through millions of minds. There, the Golden Rule emerges as the primary factor of character evaluations around the world. The Golden Rule is a formula governing our minds, evolution, and language.

g, on the other hand, has no theoretical basis. Its definition is statistical, supported by how well it captures test performance and correlates with real-world outcomes. This is not to say that we will never understand it or that it is fake or unimportant. In addition to clinical applications, g-loaded tests are an important tool in keeping institutions honest. As someone who has had to make my own way in life, I find the push to remove testing from university admissions deeply cynical. Similarly, we task police with enforcing the state monopoly on violence. It seems to me they should have to demonstrate they can rotate matrices at least well enough to tie their shoes¹⁵.

But g is not, fundamentally, intelligence. We don’t know how the two relate. What’s more, the way we use intelligent colloquially is closer to the GFP, as the word intelligent loads strongly on GFP. This is one more win for the Lexical Hypothesis, the wisdom of deferring to the crowds¹⁶. It’s also a win for the folk intuition that EQ is at least as important as IQ¹⁷, whatever the statistics say. In the words of Scott Alexander:

“I’ve been reading the biases and heuristics literature for fifteen years now and have developed the following heuristic: if a researcher finds that ordinary people are biased about how many marshmallows to take in a rigged experiment, this is probably an interesting and productive line of research. But if a researcher finds that ordinary people are biased about their most foundational real-life beliefs, probably those ordinary people are being completely sensible, and it’s the researcher who’s trying to shoehorn their reasoning into some mode it was never intended to address.”

The high heritability of IQ and, to a lesser extent, GFP have implications for our distant ancestors. How far back were they like us? I accept that there has been selection on the GFP, even if the implications are fantastic. Given the stronger claims that g-maximalists make about its evolutionary fitness, their model of the past should be more fantastic still.

Plato agrees with Homer and the ancients (to him!) that Athena represents “divine intelligence.” Together they cry: “This is she who has the mind of God.”¹⁸ And what is that mind? In Alcibiades II, Socrates (used by Plato as a character) converses with the ambitious and impetuous Alcibiades, who is preparing to make a public prayer. Socrates warns him to be careful about what he prays for, lest he inadvertently ask for something harmful. In the metaphor he chooses, there are echoes of the knowledge Eve introduced to Adam, moral discernment—the crux of the GFP. In the words of Plato:

Athena removed the mist from the eyes of Diomede, "That he might well discern both God and man," so you too must first have the mist removed which now enwraps your soul, and then the means may be given to you whereby you may distinguish between good and evil. For at present I do not think you could do so.

Call it EQ, the GFP, social intelligence, wisdom, or even Nous; these are closer to human intelligence than IQ. It is the lifelong process of understanding the divine spark within oneself and learning to revel in the successes of others. Many high decouplers pride themselves on being able to accept statistical evidence that intelligence can be significantly measured by an exam and that the trait is largely determined at birth. To be fair, the world is a gruesome place, and the notion can’t be rejected a priori. But, even on statistical grounds, there are many loose ends. And it flies in the face of thousands of years of tradition as well as common sense. At least the intellectual humility of Jensen, Deary, or Socrates is in order¹⁹.

Hermes and Athena, goddess of wisdom

1. This holds for extreme variations. In fact, this paper did dimensionality reduction on three very different exams. One was a regular personality survey, another measured psychopathy, and another measured personality disorders. The former asks people to rate themselves on how much they like to show off or are attentive to details. The latter two ask whether the person believes their legs belong to them. GFP correlates r = -0.90 with the general factor of personality disorders, which in turn correlates r = 0.92 with the general factor of psychopathology. Astounding how similar all these constructs are in the same population.

As the name implies, some conceive of the GFP as general. This means that every other personality factor exists under its umbrella, perhaps adding new aspects but always defined in relation to the GFP. Extraversion, therefore, may combine the GFP with energy and openness. In the Primary Factor of Personality, I explained why I prefer a more modest account and simply called it the primary (first and most important) factor of personality. This sidesteps statistical criticisms such as those made in The general factor of personality: A general critique.

2. See this post where you can practice this process on two mystery factors.

3. One reason I think the Golden Rule (at least identified in the GFP) does not require turning the other cheek is that I don’t think the data applies to extractionary relationships. This is made clear by the huge negative loading of words like abusive, but also because the purpose of gossip is to avoid lose-win and lose-lose relationships. Relationships that are not win-win are simply not very stable, especially in our evolutionary past when there was not such drastic social hierarchy. (There may be exceptions such as parent-child, but of course, those are a special case.)

4. And even more surprising if the factor was wholly a statistical artifact, as is commonly believed among personality psychologists.

5. Ostensibly, not the Golden Rule for Darwin because he also says:

“Nor is it probable that the primitive conscience would reproach a man for injuring his enemy; rather it would reproach him, if he had not revenged himself. To do good in return for evil, to love your enemy, is a height of morality to which it may be doubted whether the social instincts would, by themselves, have ever led us. It is necessary that these instincts, together with sympathy, should have been highly cultivated and extended by the aid of reason, instruction, and the love or fear of God, before any such golden rule would ever be thought of and obeyed.”

So what rule did language enforce earlier? Not clear to me why the rule must be articulated to be a force. Did Jesus uttering the Golden Rule suddenly change the fitness landscape? Very optimistic perspective on the morality of Christians ever since. No, I think it’s much more likely that we evolved to be considerate and Jesus articulated a very extreme version of that, where one should not consider one’s own well-being at all.

6. This actually contributed to a long debate on if this factor was anything more than respondent bias. My position is that word vectors address this argument.

7. I plot the eigenvalues of the GFP obtained via Natural Language Processing and traditional surveys in this piece, finding 23% and 35% respectively. Quite a bit less than 50%, but note that with different processing decisions (dimensionality reduction on the raw data instead of the correlation matrix of items), one obtains 80% on the surveys. Likewise, the NLP data reaches 80% if one does not enforce unit variance on each dimension of the word vectors before calculating the correlation matrix, a practice I am not sure is justified.

The General Factor of Personality: A General Critique review different data sets and report values from 29-50% for personality and 34-56% for cognitive ability (Table 2, column C1/N). Note that ability tests are scored right/wrong, whereas there is no correct answer for many personality questions. Given that reality, it is surprising that the first factor is so similar between the data sets.

General Factors of Personality in Six Datasets reports values from 27% to 63%.

8. I found this book most helpful when I had questions about measuring and interpreting reaction time for the concussion project. Turns out that individual reaction times are not that correlated with each other, even in the same individual. However, the average and variance of a person’s reaction time is quite predictive of mental function (with variance being a bit better).

9. Consider the abstract of his aptly named Intelligence:

“Some people are cleverer than others. I think it would be a good thing if more biologists began with that observation as the starting point for their research. Why? Because it is a prominent and consistent way in which people differ from each other; because the measurements we make of people’s cleverness produce scores that are correlated with important life outcomes; because it is interesting to discover the mechanisms that produce these individual differences; and because understanding these mechanisms might help to ameliorate those states in which cognitive function is low or declining.”

10. The reproduction of intelligence uses 0.5. Reconsidering the Heritability of Intelligence in Adulthood: Taking Assortative Mating and Cultural Transmission into Account finds 0.58.

11. This is especially true if your definition of g is similar to that of Van der Linden: “If the GFP can indeed influence a wide range of behaviors, then a subsequent question would be how to interpret such a construct. In the cognitive domain, the interpretation of g is straightforward: an individual’s ability to solve complex and novel problems. The interpretation of the GFP, however, seems less obvious.”

It’s hard to imagine “solving novel problems” not being substantially evolutionarily fit over the last 10,000 years. Life in this period has been a barrage of novel problems of increasing complexity. Further, it’s interesting that Van der Linden finds g to be understood but the GFP to be mysterious. I am, of course, arguing for the opposite.

12. Evidence for Shared Genetic Dominance Between the General Factor of Personality, Mental and Physical Health, and Life History Traits

A General Factor of Personality From Multitrait–Multimethod Data and Cross–National Twins

13. As far as I can tell, Gregory Cochran has done the most work on this in the sphere of IQ. The link for the Breeder’s Equation above is an article by Gregory Cochran, who explains, “And of course the breeder’s equation explains how average IQ potential is declining today, because of low fertility among highly educated women.” He finds this politically significant in that it will noticeably reduce abilities. This follows from high estimates of heritability and selection gradient. In his book The 10,000 Year Explosion: How Civilization Accelerated Human Evolution, he applies this principle to the past, making the provocative claim that working as money lenders increased the average Jewish IQ in the middle ages, which explains their 15-point advantage now. But if that increase can be had on the order of centuries, what about millennia? He is more modest about those effects:

“We suspect that increases in intelligence made agriculture possible, but the route may have been indirect. For example, the invention of better weapons and hunting techniques, combined with other technologies that let humans make better use of plant foods, could have led to lower numbers, or even extinction, of key game animals—which would have eliminated an attractive alternative to farming.”

And if there was any doubt as to what Cochran means by intelligence:

“Daniel Goleman has written of “emotional intelligence” and “social intelligence,” pointing out how they can help to predict job success and personal happiness. And other forms of intelligence have been proposed. In his 1993 book, Howard Gardner suggested that there are many types. But the data hardly support these attempts to complexify cognitive testing. The supposed special kinds of intelligence don’t predict anything useful or, when they do, predict only to the extent that they are correlated with general intelligence.”

As I have argued, it’s important not to confuse ease of measurement with the relative importance of traits. Further, I’m interested in the magnitude of selective pressure he imagines, or the average IQ of humans 10,000 years ago.

14. Though sometimes the two come out equal. See, for example, Revisiting meta-analytic estimates of validity in personnel selection: Addressing systematic overcorrection for restriction of range. This makes a lot of corrections when comparing different methods that predict job performance. IQ and EQ correlate with performance 0.31 and 0.30, respectively. (Relevant table clipped in this tweet.)

15. This also highlights the shortcomings of the GFP in the world of individual differences. Ostensibly, EQ or the tendency to live the Golden Rule should be more important than g in enforcing the state monopoly on violence. But those can’t be measured nearly as well, and certainly not in an adversarial setting where participants can lie on a survey. Further, g is even shown to be correlated with firearm proficiency.

16. The Lexical Hypothesis is often motived with this quote: "...our common stock of words embodies all the distinctions men have found worth drawing, and the connections they have found worth marking, in the lifetime of many generations: these surely are likely to be more numerous, more sound, since they have stood up to the long test of survival of the fittest, and more subtle, at least in all ordinary and reasonable practical matters, than any that you or I are likely to think up in our armchair of an afternoon—the most favourite alternative method." J.L. Austin, A Plea for Excuses

17. A survey of American workers found “73% say emotional quotient (EQ) is more important than intelligence quotient (IQ).”

18. “The ancients seem to have had the same belief about Athena as the interpreters of Homer have now; for most of these, in commenting on the poet, say that he represents Athena as mind and intellect; and the maker of names seems to have had a similar conception of her, and indeed he gives her the still higher title of "divine intelligence", seeming to say: This is she who has the mind of God”

19. Or Darwin, who left language (which he connects to morality) as a possible path to humans being different in kind rather than degree from animals:

Nevertheless, the difference in mind between man and the higher animals, great as it is, certainly is one of degree and not of kind. We have seen that the senses and intuitions, the various emotions and faculties, such as love, memory, attention, curiosity, imitation, reason, etc., of which man boasts, may be found in an incipient, or even sometimes in a well-developed condition, in the lower animals. They are also capable of some inherited improvement, as we see in the domestic dog compared with the wolf or jackal. If it could be proved that certain high mental powers, such as the formation of general concepts, self-consciousness, etc., were absolutely peculiar to man, which seems extremely doubtful, it is not improbable that these qualities are merely the incidental results of other highly-advanced intellectual faculties; and these again mainly the result of the continued use of a perfect language.