January 25, 2017

Multiple Intelligences, Emotional Intelligence, Creativity and “g”

  1. Description of Multiple Intelligences vs. General Intelligence (“g”)

The dominant paradigm for intelligence in psychology is the “general intelligence factor” along with individual intelligences. In a sense, the dominant viewpoint already embraces “multiple intelligences” to a degree. The dominant view does not reject specific domains.

Howard Gardner, in 1983, put forward a theory of “multiple intelligences” which included 7 “intelligence modalities” – musical, visual, verbal, logical, bodily-kinesthetic, interpersonal and intrapersonal (self-reflective).

In 1995, he added “naturalistic intelligence”, and in 1999 he added “spiritual / existential intelligence”.

In an interview, Gardner describes his theory thusly:

The theory is a critique of the standard psychological view of intellect: that there is a single intelligence, adequately measured by IQ or other short answer tests. Instead, on the basis of evidence from disparate sources, I claim that human beings have a number of relatively discrete intellectual capacities. IQ tests assess linguistic and logical-mathematical intelligence, and sometimes spatial intelligence; and they are a reasonably good predictor of who will do well in a 20th (note: NOT 21st) century secular school.”Belief in multiple intelligences theory implies that human beings possess several relatively independent computers; strength in one computer does not predict strength (or weakness) with other computers. Put concretely, one might have high (or low) spatial intelligence and yet that does not predict whether one will have high (or low) musical or interpersonal intelligence.”

Now first off Gardner incorrectly describes the standard view. Nobody, not Jensen, Murray, Herrnstein, Gottfriedson, has claimed “intelligence” was a single thing, just that there exists a “general intelligence factor” underlying all of the individual domains. This is a very important distinction that Gardner confuses.

He’s also just factually incorrect about his claim that IQ doesn’t predict school grades past the year 2000.

But those two falsehoods aside, this throws down his disagreements: Gardner *basically* denies any general intelligence factor, whereas mainstream intelligence researchers contend that intelligence is BOTH general and specialized.

  1. Difficulty in assessing the theory

One of the major difficulties in assessing “multiple intelligences” is that Gardner is opposed to psychometric testing – and so we have no way to measure “multiple intelligences”.

Gardner admitted that “MI theory has few enthusiasts among psychometricians or others of a traditional psychological background” because they require “psychometric or experimental evidence that allows one to prove the existence of the several intelligences.”

The Textbook “Introducing Neuroeducational Research: Neuroscience, Education and the Brain from Contexts to Practice” classifies multiple intelligences as a pseudoscience because it is unfalsifiable (there’s no way to show it’s wrong) and doesn’t make any predictions.

For example, how can one say that “musical ability” operates totally independently from “mathematical ability” when Gardner opposes any standardized testing of those things?

  1. Neuroimaging evidence supports mainstream psychology

The 2010 paper entitled “Gray matter correlates of cognitive ability tests used for vocational guidance” looked at brain activation in people when performing various cognitive tasks. And in it the researchers found that some tasks correlated with specific regions of the brain, whereas others did not correlate with specific regions and can then be said to be more generalized:

These observations are based on qualitative comparisons but they illustrate the potential value of examining separate indices of performance. There arealso meaningful results at the levels of group factors and g, and so it appears that analysis of all three levels is important for understanding brain/cognitive relationships [28]. The results for g are consistent with other findings, especially the P-FIT model of brain areas hypothesized to underlie general intelligence [9] and are detailed elsewhere [11]. Specifically for the other factors, in this sample, Speed of Reasoning and Memory showed relatively strong gray matter correlates. The two individual tests for the Speed of Reasoning factor showed different patterns and both contributed to the factor pattern. For the Memory factor, both tests showed similar results. One test in the Spatial factor was informative (PF) and the other not so much. Neither test in the Numerical factor showed informative gray matter correlates.”

The paper also had some images of the brain activation:

And so the neuroimaging data brings up some very regionally specific tasks, some tasks so generalized that there’s almost no neural correlate to be found, and also a theory about a set of regions of the brain that aren’t task specific, but underly a general intelligence – i.e. specific regions that have a big impact on general intelligence.

Gardner’s ideas about completely separate intelligences isn’t even mentioned. It doesn’t even come into it.

Lynn Waterhouse also mentioned neuroimaging research in response to “multiple intelligences”:

Taken together the evidence for the intercorrelations of subskills of IQ measures; the evidence for a shared set of genes associated with mathematics, reading, and g; and the evidence for shared and overlapping “What is it?” and “Where is it?” neural processing pathways and shared neural pathways for language, music, motor skills, and emotions suggest that it is unlikely that that each of Gardner’s intelligences could operate “via a different set of neural mechanisms” (Gardner, 1999, p. 99)”

Now just because the brain physiologically is both generalized and specific doesn’t *necessarily* mean function is also both generalized and specific. But it probably does.

  1. IQ subtests correlate with each other

In addition to physical generalization of intelligence in the brain as seen from brain scans, there is also the fact that IQ subtests correlate with each other. In fact, one of the points made by Arthur Jensen in “The g Factor” was that proficiency at almost any task that takes some thinking is correlated with proficiency at all other tasks that take some thinking.

Table 19.1 from the paper “Cognitive and Neurobiological Mechanisms of the Law of General Intelligence”, one of the things Christopher Chabris did was show the correlations between subtest scores for individuals taking the Wechsler Adult Intelligence Scale – Revised (WAIS-R):

Vocab. 0.83
Similarities 0.67 0.8
Information 0.72 0.59 0.8
Comprehension 0.7 0.58 0.59 0.75
Picture Arrangement 0.51 0.53 0.5 0.42 0.7
Block Design 0.45 0.46 0.45 0.39 0.43 0.7
Arithmetic 0.48 0.43 0.55 0.45 0.41 0.44 0.68


0.49 0.52 0.52 0.46 0.48 0.45 0.3 0.68
Digit Span 0.46 0.4 0.36 0.36 0.31 0.32 0.47 0.23 0.56


0.32 0.4 0.32 0.29 0.36 0.58 0.33 0.41 0.14 0.56
Digit Symbol 0.32 0.33 0.26 0.3 0.28 0.36 0.28 0.26 0.27 0.25 0.48

The paper isn’t even about this, it’s just mentioning this fact in relation to something else the author is talking about.

Now the question of HOW generalized intelligence is, or HOW specific and independent the individual domains are from each other – what does “g” mean – that is another topic.

Gardner doesn’t deny the statistical reality of “g”, but merely hand-waves it away, saying:

We do not really understand what is measured by ‘g’—it could be anything from sheer intellect to motivation to skill in following instructions to the ability to shift facilely from one kind of problem to another.

Of course this criticism could be used to deny the results of literally any psychometric test, as one could always just claim “effort is variable”.

And as shown previously, IQ is a better predictor of occupation and income than either grades, parental education or parental income. So does Gardner think that the people who put in the effort to get good grades DON’T put in the effort to get higher IQs? Given that job and income are going to be directly influenced by grades and effort on the job that produce grades – why then would we see people with higher IQs who get worse grades – do they just so happen to put in effort on the IQ test but not in school elsewhere?

The problem is that Gardner exists in a state of superposition, and it’s hard to pin down his opposition to “g”. He’s an intelligence researcher – so does he deny the correlation with life outcomes? Does he deny the heritability of IQ? Does he deny the correlation with perceptions of intelligence?

So I have arguments here that attack his position as if he doesn’t believe in “g”, but then he’ll say he does believe in “g” and that it could be innate. It’s difficult to argue against Gardner because he’s so vague:

Those involved in standard psychometrics are almost always critical of the theory; among those psychologists who are not psychometricians, there is openness to the expansion of the concept and measurement of intelligence.”

Gardner is opposed to measuring intelligence, practically speaking. He gives some theoretical situation in which he would approve of an intelligence test, but admits that those requirements would almost never be met. And as for “expansion of the concept” – naturalistic intelligence? Spiritual intelligence? Moral intelligence? Why not just describe everything humans do and call it an “intelligence”? How about “color intelligence” or “persuasion intelligence” or “teaching intelligence”?

Sure, you’re “expanding” your conception of intelligence by including things that aren’t measured, can’t be tested and don’t predict anything.

Still, psychologists like neat measures of their constructs and there is frustration that the “new” intelligences are not as readily measured as the standard ones”

Forget “neat measures” – how about ANY measure!

Conclusion on Multiple Intelligence

Really the point of this post is for people who cite “multiple intelligences” as an argument against IQ. A lot of them hear “multiple intelligences” and imagine that this is in opposition to the other idea which is that intelligence is just one thing. Nobody believes that, the fact that there exist tests on which “g” is less than 100% of the score is evidence of this fact. In the analysis above, “g” only correlates with each subtest at .685, not 1.000.

A1. Emotional Intelligence

Emotional Intelligence is not much more than big-five personality and IQ. Its correlation with an IQ test was found to be .454. However when scores on the “Big Five” personality traits (Nuroticism, Extroversion, Openness to Experience, Agreeableness, Conscientiousness) are used as well, the correlation between Emotional Intelligence and (Big 5 + IQ) was found to be 0.617.

However, the no test is completely reliable. Reliability simply refers to the correlation between a person’s test score at two different times. The test-retest correlation for IQ tests is about .87, which means it’s difficult to ever get a correlation higher than .87 in reality. If you were to take multiple groups of people and give them IQ tests at two different times, the correlation between the scores at the two times would be around .87 – sometimes higher, sometimes lower.

Anyway, when the unreliability of both the Emotional Intelligence score and the Big 5 / IQ composite were controlled for, the correlation rose to .806. To the extent intelligence is generalized, we would also expect emotional intelligence to be partly a function of general intelligence.

So sure, an “emotional intelligence” scale, test and score does exist (far more than can be said for Gardner’s ideas), but it doesn’t deviate much from already existing scales.

2010 meta-analysis which, among other things, compared EQ to job performance, showed that EQ only .052 independent predictive power from IQ, which had .328. Just one of the Big 5 measures, conscientiousness, had a predictive power of .051.

In addition, a 1999 paper found that one’s EQ-i score had almost no impact on grades.

EQ’s job performance predictive power is impressive AS AN IQ SUBTEST. Which is to say, if EQ is treated as just another kind of intelligence not measured by IQ tests. But to say it’s more important than IQ is silly both because EQ is in part a function of “g”, and because it matter of factly matters a lot less.

But the real reason EQ is brought up is to deny the importance of IQ and group differences. This won’t work, because not only do whites score better on EI than blacks, but since EQ is largely based on self-report, and blacks lie on surveys more than whites do, the “real EQ” gap is probably larger than the 0.34 SD in that paper anyway.

A2. Creativity

One odd argument is that IQ doesn’t measure creativity. Intuitively, you’d think it would. In order to create something novel and useful, you have to understand how the world works and how other things work. People who invent things, or design new engines, new ways to drill for oil, new ways to manufacture things, how to manage traffic or parking – these people all SEEM to have high IQs.

Given the hazards of measuring creativity, one study attempted to do that. What they did was give the subjects some objects, such as a can, a knife and a hairdryer, and asked them how they would use these objects to do various things, like “what can make noise?” and “what could one use for quicker locomotion?”.

The responses were then rated for their creativity by a panel of four students, whose answers correlated with each other at 0.80.

It was found that the rated originality of the top two answers a person gave was correlated with their IQ at 0.36, the average originality of their answers correlated at 0.35, and the “ideational fluency” correlated with IQ at 0.22.

I personally don’t know what to make of any of this, since “creativity” is so subjective. I would be interested in having an “objective” creativity test like “find your way out of this box with a paperclip and a string”, as it would test one’s ability to solve a real problem in a novel way – as opposed to “I can make noises with a knife like a harmonica” and the rater going “oh wow that’s really creative!”.

Now you could call a modern artist, or a blogger writing about their troubles as a black woman, or a child’s stick figures “creative”. Or define creativity as “new connections made by that person”. But there’s nothing real to base any of that on. Because those things don’t try to solve problems, we can’t say if they are correct or incorrect.

For example, if someone tries to make a blender that generates more torque with less electricity, but fails, okay, that’s a failed attempt at being creative. Someone who tries to do that, and succeeds, that’s a successful attempt at being creative. He was more creative.

We can’t quantify the precise relation between IQ and the unmeasurable essence of creativity, but we do see that the people who actually CREATE things that correctly deal with a problem do in fact have higher IQs. Moreover, the high IQ creators of things can very easily do the “creative” things that low IQ people do, but not the other way around. I.e. an engineer can create modern art or write about privilege and oppression, or babble about brains in vats and chinese rooms, but the opposite is not true.

And in reality people who have higher IQs are probably, on average, just as much more creative than everyone else as their IQ is higher than everyone else’s; and this distinction between intelligence and creativity is baseless.

A3. IQ only measures one’s score on an IQ test / The brain is complicated

“Math tests only measure one’s score on a math test” – see how stupid that sounds? Obviously a math test tests your ability to do math.

Another argument against IQ is “the brain is complicated” argument. It usually involves a word salad referencing specific regions of the brain, how neurons work, and saying “how can you reduce all of that to a single number!!??”.

Now nobody makes this argument when teams make athletes perform a series of lifts and agility drills to quantify how “strong” and “fast” they are. Sure, there are ~650 skeletal muscles in the human body, and they use multiple metabolic pathways, and are influenced by hormones.

And yet, with a series of lifts, that can all be reduced to a single number. And if you want to know more, you can look into each subtest – the squat, the dead lift, the power clean, the push-jerk. Just like with IQ you can look into each subtest to see how well they did there.

And, as it happens, IQ tests predict grades and job performance quite well.

That is the importance of “validity”. For an athletic combine test to be valid, it has to predict performance in the game somewhat. Of course it’s never a perfect predictor, but in general, football players who can push more iron and run faster are better than those who can’t. And if you can’t push a certain amount of iron, or are just slow, no amount of “game sense” or “x factor” will make you an elite athlete.

And clearly there are things related to grades and job performance other than IQ – but a higher IQ is still massively better than a lower one. And if your IQ is 80, you probably won’t be a chemical engineer. There is some “wiggle room” where people are able to do things beyond what their IQs would predict, just as some football players play at a level way above their combine stats – but there is a limit to that and the general trend holds.

So think of an IQ test as a series of mental lifts.


Now for some this is depressing, and that’s a product of having unrealistic expectations of personal development. You can develop knowledge, and you can create things, and learn to think scientifically and such, but for now, you’re not going to increase your base intelligence.

It’s like articles about a new way to diet – if there was a proven, easy way, it would be generally known by now. Same with raising “base intelligence”, if it was known, that knowledge would be ubiquitous and applied universally.

I talk about IQ all the time, but I don’t think about mine at all. Why would I? I can’t *really* increase it.

Of course I can learn to game the test and get my score up – but that’s not really raising the underlying intelligence, just test-taking skill. There is so much more upside in creating THINGS, in leaving behind WORKS.

Another problem with the denial of relatively fixed intelligence is that teachers believe the mythology that pure effort can make anyone an engineer. This causes teachers to push kids past their capacities, or at least to their limits where they are miserable, and to treat failure as a product of laziness.

Facebook Comments

    I remember being told this stuff in high school and it being immediately obvious that it was more about self-esteem building than it was about describing anything useful to understanding psychometrics.

    The psychological appeal of the Multiple Intelligences model is obvious.
    Having a large number of intelligences with vague underlying structures, and no way of verifying them objectively seems like a perfect way of mentally protecting one’s ego.

  • Emil Kirkegaard

    The meta-analysis cited for EQ shows that EQ personality/mixed does have incremental validity above FFM+IQ, betas of .253 and .326 compared with IQ’s .642 and .659. So, it does provide non-negligible amounts of incremental validity. EQ ability was fairly useless and perhaps just a poor IQ test.