Showing posts with label factor analysis. Show all posts
Showing posts with label factor analysis. Show all posts

Friday, May 15, 2020

Why IQ Tests Don't Measure Intelligence

Consider a surgeon, a composer, an artist, and a physicist. Each is the world's best. Each therefore is highly intelligent as the Oxford English Dictionary defines intelligence:
the ability to acquire and apply knowledge and skills,
And as they are all highly intelligent, then, according to the psychologist's understanding of the term intelligence, they must all have really high IQ's. But do they? And even were it so, what would that mean?

The surgeon for all his astonishing deftness with a scalpel, and his knowledge of human anatomy, may be dumb as a brick at math, and not especially articulate either. The composer probably can't draw, and the artist may be unable to carry a tune. And the physicist,  may, like the great Paul Dirac,  be linguistically monosyllabic. So to say that these highly intelligent people are equally intelligent, even in the unlikely case that they scored equally on an IQ test, would be meaningless. They are not equally intelligent at all. They are each vastly more intelligent, than all the rest in their own particular way.

What's hard to understand about that? Everyone knows that that is how it is. Intelligence is not one thing, one faculty, one gift: it is an array of abilities, each person having intellectual strengths and weaknesses. How then did psychologists come to adopt the seemingly nonsensical idea that one number, the so-called Intelligence Quotient, could measure a persons intellectual worth?

The answer to that question is two-fold. First, IQ testing is like the busted science of phrenology — the reading of cranial bumps: it allows the practitioner to claim to know a person's intellectual standing in the world, and hence their future potential. It is this claimed ability to judge the worth of a man that places the psychologist in a position of authority.

The second reason that psychologists proclaim the power of IQ tests to measure mental horse-power is a piece of mathematical data manipulation invented by a Victorian eugenicist and statistician, Karl Pearson. The math in question is called factor analysis, and what factor analysis does is to examine the relationships among various characteristics of some class of things, for example, towns, countries, planets, acorns, or apples, thereby to discover whether there is a common factor or factors underlying these relationships.

An underlying factor, is exactly what the psychometricians — as the quantifiers of mental capacity call themselves — found when they examined variation from individual to individual in mental abilities. Those who score high on mathematical logic, tend to do relatively well on other tests such as verbal reasoning, or the interpretation of geometric puzzles. This factor, quantified by Pearson's factor analysis, they named "G" for "General Intelligence."

Having gone thus far, it was a small step to the conclusion that a person's mean score on the various components of a test of cognitive abilities was a valid way to estimate General Intelligence, hence the results of tests of multiple cognitive tasks was called an "Intelligence Quotient" or IQ, and the test itself, an IQ test.

This view led to a general belief that intelligence is just one thing, underlain by some as yet unidentified common factor, which determines a person's overall intellectual capacity.

But the correlations among cognitive capacities are low. In fact, mostly very low. See here, for example, where correlation coefficients (r values) among a large number of tests averaged less than 0.3. That means that, on average, less than 10% (r squared values) of the variation in any one ability is explained by variation in any other ability.

So yes, mental capacities share at least one common underlying factor, but its effect is weak, meaning that an individual's relative ability at one kind of mental activity will rarely be an accurate guide to their ability at another type of mental task.

It is the weakness of G, and the dependence upon it of their conception of IQ as a measure of intelligence, that psychologists have been loathe to admit. Thus has emerged a widely held idea that G as estimated from an IQ test score measures the essence of intelligence, just as chip speed measures the power of a computer central processing unit. What this is widely understood to imply is that intelligent mental activity depends on either a common mechanism, or a common feature of nerve cells that dictates the scope and power mental activity. However, the slightest awareness of brain anatomy and physiology would disabuse one of the notion that the brain has anything equivalent to a CPU, or even a uniform functional cellular capacity. On the contrary, different mental activities depend on different neural lobes, networks and ganglia, or on hierarchies of lobes, networks and ganglia.

Moreover, these components of the brain are far from identical in physiology and structure. There are many types of nerve cells or neurons and their supporting glial cells, as there are many different signalling methods within the brain, these involving at least eighty neurotransmitters. Despite the subjective unity of mind, the brain is thus a collection of many neurological machines, each with its own genetic determinants and its own history of past experience.

Thus, G or general intelligence, far from representing a fundamental component of intelligence, reflects only the dependence of the functioning of the entire brain on either other organs or some weakly influential characteristic of all brain tissue.

For example, brains without oxygen die within seconds, which means that brain function depends on lungs, heart and the vascular system. Moreover, without a continuous supply of glucose brains cannot function, which means a dependence on the liver and on the endocrine system that controls blood sugar. And without means to dispose of waste products, the brain is rapidly poisoned, meaning dependence on the kidneys. As to general properties of brain tissue, effective function depends on  many general features such as mitochondria, ribosomes, microtubules, and much else, all of which must function properly or the brain will function poorly or not at all.

Thus the mystery of G is revealed. It is a reflection, simply, of the brain's dependence on the rest of the body and on the cellular machinery common to all nerve cells. If all supporting systems and sub-cellular components are in the highest working condition, then the multiple components of the brain can all function at the peak capacity. But all defects or limitations in the performance of supporting systems and cellular organelles limit mental performance. Thus, beside variation in relative power of the various components of the brain there will be variation from brain to brain due the functionality of the brains support systems and components.

Thus just as a large town will tend to have more crime, traffic congestion and air pollution than a small town, so those with the best overall health and the best cellular machinery, will tend to have higher IQ's than those whose mental function dependent on defective support systems or cellular machinery. But still, among individuals, the big differences in intellect, are on specific tasks not on overall performance, or G, as assessed by a so-called IQ test.

So, yes, IQ-ism is largely bunk and the sooner we're rid of it the sooner will psychologists be able to study intelligence more intelligently.