IQ Is Not Intelligence

Once again “IQ“. We continue our quest to disabuse ourselves of the notions that “IQ” is intelligence and that one-number of summaries of intelligence are adequate. Both these contentions will be proved. This is a small thing, and the battle has already been won, because most who look into these things already agree with the criticisms below. All I hope for, likely in vain, is for us to say intelligence instead of IQ when we mean intelligence.

IQ is a score on tests that measures, however crudely or accurately, some but not all aspects of intelligence. Scores on a test are not intelligence: intelligence goes toward producing scores. Single-number scores cannot capture all there is to intelligence.

“IQ”, I repeat, is not intelligence. The Deadly Sin of Reification has struck every person who speaks of somebody “having” a low or high IQ. Unless, which is rare, they mean the score on some test the person has actually taken. Speaking of countries or regions or peoples has “having” a low IQ is to be damned by the Sin.

What people have, or lack, is intelligence. All experience through all history proves some are more intelligent, and some less. Not all people can be brought to the same level of intelligence, just as not all people can be brought to the same height, though the analogy suffers because height is usually thought of as one-dimensional. Small intelligence “gaps” between some persons, like in height by attending to superior early nutrition, might be closed or lessened in some cases, but not all. Large differences cannot be eradicated by good will, money, or lying. Some groups on average are less intelligent than others, and, it follows, some are more: but this is a statistical summary of groups and each man is his own self. The stereotypes you hold on intelligence, even those you won’t or can’t speak aloud, are generally true. Equality is a lie.

Any and all—as in all—questions of what IQ correlates with or predicts, such as this or that set of genes or success or failure in life, are here irrelevant. Mention any and you are out. Height positively correlates with driving ability: those under two feet drive far less well than those over five. But it does not follow that the tallest are the best drivers. Height is not a perfect measure of driving ability, though it can capture some aspect of it. Height is not driving ability.

Now let’s think about IQ scoring. Take any set of positively correlated numbers and submit them to a principal components analysis. A quite limited and error-prone method using only simplistic linear correlations, and ignoring all departures from linear correlations. Please read this for a discussion of the weaknesses of this method.

In this PCA you will always—as in always—find a first component, which we can call Gee!, that positively correlates with every—as in every—measure put into the analysis. Here is a list you might try: the daily high temperature in May in northern Michigan (but not this year), the population of the United States the last 31 years (to keep the same N for all), the body weight of pigs the first 31 days of their life, the number of components in a Boeing 737 from its 1st to 31st day of construction. Obviously, each of these will positively correlate (in a linear fashion) with each other.

The PCA of these will reveal a first component, our Gee!, that if when correlated to all the other measures has you saying in wonderment “Gee! There must be some underlying driving causal force behind all those measures”, then you have followed the same path as “IQ”. Gee! is, of course, a statistical artifact, and we would never make the mistake of ascribing an overall underlying cause for items in a list like this. But we do when it comes to IQ tests. The problem in intelligence research is that while it is obvious intelligence exists, how much the ‘g’ of IQ tests reveals aspects of true intelligence and how much of it is a statistical artifact cannot be got from the PCA (mere correlations) itself.

I urge you to read John Horn’s 2006 paper “Understanding Human Intelligence: Where Have We Come Since
Spearman?“, which reviews the history of Spearman’s ‘g’ and puts forth many cogent criticisms. Like:

The Thomson-Thorndike theory. Positive intercorrelations among all of a set of variables can be indicative of two or three or four or more common factors; it need not be indicative a single higher-order common factor. Indeed, Thomson argued that positive intercorrelations among abilities is just as compatible with a theory of many common factors as it is with a theory of one common factor….[an example follows]…Continuing in this way Thomson showed that all the devices of a battery of measurement devices can be positively correlated and not involve a single common factor.

The ‘g’ is IQ is far from well defined (with my emphasis):

…Carroll’s (1993) tour de force re-analysis of some 477 batteries of ability measuring devices. In Table 15.4 of his monograph Carroll identified “…53 factors, in 146 datasets, classified as measuring “general intelligence” or possibly Spearman’s factor g.” (p. 591). Inspection of these results indicates that often the general factors of different analyses contained no measurement devices (no abilities) that are the same; the factors were thus “general” in respect to entirely different sets of variables…Even when batteries were made up of the same measurement devices, the order of the correlations with the general factor were notably different: the factors did not pass the test of metric invariance (Meredith and Horn, 2001).

Horn has other critiques aplenty, including of those papers which purportedly discovered a ‘g’.

So all discussion of ‘g’ contains great uncertainty, and so is out as the definition of intelligence. All discussion of what intelligence is and how it might—I stress might—be measured is in. Just as you cannot quantify with precision how happy you are, though crude numbers have some rough-but-error-prone utility, so might we not be able to quantify with precision all aspects of intelligence.

Indeed, all—as in all—attempts to quantify the unquantifiable must fail at some point or level. You view a vista which you call resplendent. How resplendent is it on a scale from -17.2 to 142? You must be able to define a perfect number for this such that the distances on the scale between numbers are invariant and identical for all: they need not be linear, but each “step” must signify the same level and same increase or decrease in resplendentness. Or happiness, sadness, pain, perspicacity of all stripe, mental agility, inductive judgements, and on and on.

I have written on this in more depth before, but take the most intelligent man alive. He will (conceivably) score the best on any IQ test any other man can write. This best will not be a number arbitrarily high. None of those other men can out-think our man, and so they cannot devise a test to check for kinds and strength of intelligence greater than they themselves possess, which is less than that of our man. The same will hold for our man with respect to others. None of us, obviously, can think of what we cannot think of. Our man cannot out-think those to come (or already here, like angels) more intelligent than he. Well, there is certainly no evidence the most intelligent men (or man) wrote our current IQ tests, so they are necessarily limited, even if IQ was all there was to intelligence.

You must also avoid the temptation to equate intelligence with moral worth. We haven’t the space for a discussion here, but recall that the most intelligent being ever created is also the evilest. Even if you don’t believe that, that notion is there for a solid reason. The gods, far above us, were not all they should have been.

Numbers on tests attempting to quantify the unquantifiable can, of course, be put into PCA and other statistical analyses, but almost immediately, given the glory of the math, it will be forgotten the numbers cannot do the job asked of them to any great precision. Again, it is not—as in not—that these numbers won’t have some utility, they will, but that if they are taken too seriously over-certainty will abound. The measures will become, in the minds that use them, the thing attempted to be measured. This is the Deadly Sin.

That’s the end of the preliminaries. Now to intelligence itself.

Here, stealing from Wokepedia, as a basis for discussion, is the summary of the Cattell–Horn–Carroll theory of intelligence.

The broad abilities [of intelligence] are:^[11]

• Comprehension-Knowledge (Gc): includes the breadth and depth of a person’s acquired knowledge, the ability to communicate one’s knowledge, and the ability to reason using previously learned experiences or procedures.

• Fluid reasoning (Gf): includes the broad ability to reason, form concepts, and solve problems using unfamiliar information or novel procedures.

• Quantitative knowledge (Gq): is the ability to comprehend quantitative concepts and relationships and to manipulate numerical symbols.^[11]

• Reading & Writing Ability (Grw): includes basic reading and writing skills.

• Short-Term Memory (Gsm): is the ability to apprehend and hold information in immediate awareness and then use it within a few seconds.

• Long-Term Storage and Retrieval (Glr): is the ability to store information and fluently retrieve it later in the process of thinking.

• Visual Processing (Gv): is the ability to perceive, analyze, synthesize, and think with visual patterns, including the ability to store and recall visual representations.

• Auditory Processing (Ga): is the ability to analyze, synthesize, and discriminate auditory stimuli, including the ability to process and discriminate speech sounds that may be presented under distorted conditions.^[11]

• Processing Speed (Gs): is the ability to perform automatic cognitive tasks, particularly when measured under pressure to maintain focused attention.

A tenth ability, Decision/Reaction Time/Speed (Gt), is considered part of the theory, but is not currently assessed by any major intellectual ability test, although it can be assessed with a supplemental measure such as a continuous performance test.^[12]

• Decision/Reaction Time/Speed (Gt): reflects the immediacy with which an individual can react to stimuli or a task (typically measured in seconds or fractions of seconds; not to be confused with Gs, which typically is measured in intervals of 2–3 minutes).

Besides these, they also describe a range of “narrow abilities” such as mathematical knowledge, reading-decoding, general verbal information, others, and inductive reasoning. We’ll come back to the last.

Accepting as true their 10 aspects, then it follows intelligence is better measured and communicated with a ten-dimensional score, and not a single-dimension score, if numerical measures can be found at all. And that, if these 10 aspects are correct, a single-dimension score would in its averaging over the 10 necessarily blur these aspects. This single measure might have utility is various ways, and correlate with this and that, but that is beside the point. The single number would not be intelligence. It would committing the Deadly Sin of Reification to say that it is.

Worse, all measures downstream of this incorrect single number would themselves be wrong in this fundamental way (though they still will have some use). Heritability is one such measure. How much of intelligence is biological and how much cultural, and indeed how much is non-material, can be both exaggerated and underestimated, depending on which aspect of intelligence is meant, if only IQ is used. This limitation will be especially glaring when thinking in terms of groups.

Think back to those groups you considered most and least intelligent, on average. Do not use any examples before, say, the year of Our Lord 1900. If you search your mind like this, you’ll find some groups are consistently high, some only latterly high, and some always low. To avoid hurt feelings or boasting or (most importantly) distractions, I shall not here name them. But you will also find different aspects of intelligence prized in different times. For instance, musical ability is now little evinced—there is no Beethoven among us—though the least is praised the highest (rap). Now we enjoy praising certain analytical abilities in the sciences above all; yet we must all agree superior abilities here do not always, and increasingly not even often, lead to truth. Something is lacking.

I am not defending the Horn list per se. There may be modifications, extraneous measures or omissions, and indeed I think there are. It may be that, after further thought and experimentation, we discover any two of these are really aspects of the same underlying thing. However the list also seems to me incomplete. I, following Groake and Aristotle, say there is not one form of inductive reasoning, as Horn et al. do; there are at least five. Inductive ability is surely an aspect of intelligence, and I think the highest and purest sense of it.

Briefly (read the link just above for more) quoting myself quoting Groake (and with new emphasis):

Induction-intellection is “induction proper” or “strict induction”. It is that which takes “data” from our limited, finite senses and provides “the most basic principles of reason.”…Induction-intellection “Operates through infallible exercise of [nous], through the activity of intellection, understanding, comprehension, insight.” It produces “Abstraction of necessary concepts, definitions, essences, necessary attributes, first principles, natural facts, moral principles.”…

Induction-intuition is similar to induction-intellection. It “operates through cleverness, a general power of discernment of shrewdness” and provides knowledge of “any likeness or similitude, the general notion of belonging to a class, any discernment of sameness or unity.” Axioms arise from this form of induction…

Induction-argument, given by inductive syllogisms, is the “most rigorous form of inductive inference” and provides knowledge of “Essential or necessary properties or principles (including moral knowledge)”. The forecaster on television announces E = “It’s probably going to rain tomorrow” and we’re interested in the conclusion Y = “It will rain tomorrow”…

Induction-analogy is the least rigorous but most familiar (in daily life) form of induction and provides knowledge of “What is plausible, contingent or accidental; knowledge relating to convention, human affairs.” This form of induction explains lawyer jokes…

Induction-probability of course is the subject of most of this book [and my work]. It provides knowledge of “Accidental features, frequency of properties, correlations in populations” and the like. It is, as is well known by anybody reading these words, the most prone to error. But the error usually comes not in failing to see correlations and confusing accidental properties with essences, but in misascribing causes, in mistaking correlation for causation even though everybody knows this admonition…

These are not equal nor the same abilities (which is argued at the link and in Groake’s book, etc.). We have all met those who excel at some rather than others of these, including some who do not seem to be able to come to terms with the deepest aspects of induction. Equality, I repeat, has never been observed, and we induce that it is false.

It’s observed to be true that, generally, by which I mean on average, those excelling or failing at one aspect of intelligence also excel or fail at the others. More or less. That was the impetus behind ‘g’. How much more or less is a fascinating question, per individual, but in order to get to this we’d need the settle the aspect list itself. How much above or below in any one corresponds to the level in any other will surely have biological, cultural, and, yes, spiritual dimensions, too. That is, there won’t be one ‘g’; there might be three, four or five, or possibly even none.

My contentions are simple: there are multiple aspects and not a single aspect to intelligence, that whether there are many or one any can only be quantified to some incomplete level, because not everything is quantifiable, and that therefore IQ is not intelligence. That it is “close enough” is only true in some but not all cases. That it is useful is again only true in some but not all cases.

All I would like is for us to use intelligence every time we are tempted to use IQ (unless discussing a particular score on a specific test). It doesn’t seem much to ask. Alas, I know there is little hope. For one, IQ has only two syllables and intelligence four, and four is greater than two, and who likes to do more work? For another, IQ has become so reified that it is impossible for some to separate from intelligence.

There’s where the damage lies. In you go hunting for the snark, and have a grant to do so, you will find one. Or what you will call one. If you have thoroughly reified IQ, then you will not look into the questions of intelligence too deeply, about what separates the intellect from the will, and about all other such matters. You will turn into an academic.

Subscribe or donate to support this site and its wholly independent host using credit card click here. Or use the paid subscription at Substack. Cash App: $WilliamMBriggs. For Zelle, use my email: matt@wmbriggs.com, and please include yours so I know who to thank. BUY ME A COFFEE.