The Basis of Testing Intelligence

Psychometrics, the scientific study of quantitative psychology, began with Alfred Binet in late 19th century France. He discovered that children had "mental ages" that did not correspond directly to their physical ages. Rather, he found that the distribution of children's mental ages for a given physical age was a bell curve (or normal distribution) centered around the children's physical age.

Binet then created a metric called the "Intelligence Quotient" or IQ. A child's IQ is the ratio of their mental age to their physical age multiplied by 100. An IQ of 130, or a ten year old with the intelligence of a 13 year old, should occur in 2% of their population. Since the distribution of intelligence is symmetrical, an IQ of 70, or an adult man with the intelligence of a 13 year old, is expresses itself in 2% of an average population. People with IQs of 70 or lower are called mentally retarded, and Binet's test enabled educators to diagnose mental handicaps earlier and target educational resources more efficiently.

Later, intelligence tests were created for adults. But a score of 130 for an adult has a more abstract meaning than it does for a child. A child with an IQ of 130 can be seen as having the same mental development as a child 30% older than himself. But what does an above average IQ mean for an adult? Simply put, it means that the adult is smarter or more mentally developed than other adults. But what does it mean for one adult to be smarter than another? Does it mean that they have more memory? Does it mean that they have a mental development that is somehow "beyond" full mental development? These questions would be answered by the statistician Francis Galton and psychologist Charles Spearman.

General Intelligence

In the late 19th century, Galton discovered a mathematical tool called correlation. If two variables are strongly related to each other, we say that they are correlated. For example, the correlation between it raining and the ground being wet is 1.00. By that we mean that the two variables always coincide. If the correlation was 0.0, it would imply no relationship between the variables. If the correlation is negative, it implies an inverse relationship.

Spearman noticed that academic abilities were correlated. For example, you might have great grades in History but weaker math grades. But on average a person with stronger history grades will have stronger math grades then a person with weaker math grades. Spearman conjectured that there might be a "General Intelligence Factor", known as g, that represents the common overlap between all academic abilities. If something is very closely correlated with g, it is said to be g-loaded.

This new discovery dramatically increased the utility of IQ tests and made the concept of intelligence much more robust. If one wants to predict a child's academic future, they could give them something that is moderately g-loaded like a vocabulary test. But such a test might miss out on a more distant aptitude in mathematics that is not very closely correlated with a vocabulary test. It would be best to give an IQ test that is highly g-loaded, because it will partially predict the child's aptitude in a variety of subjects.

To think about g more closely, think of a few people you know and how they do in school. Of course every person has strengths and weaknesses, but if one student is setting the curve in their math classes, they will likely do better in their history classes than someone at the bottom of their math class. Most people have academic aptitudes that are somewhat even across different subjects. However there are a handful of people, known as Idiot Savants or more recently Autistic Savants who have a few mental gifts but are otherwise mentally handicapped. The existence of the general intelligence factor does not contradict the existence of autistic savants, rather it should be a rare condition.

Animal Intelligence

Some have pointed out that unintelligent animals can do very well on specific IQ test questions. For example gorillas can remember long strings of digits, and pigeons are very good at a specific on Raven's Progressive Matrices.

Environments and Genes

Intelligence is very heritable among humans in developed countries. Studies done with identical twins have shown that intelligence is determined by genes by as much as 80%. However we must be extremely careful with this number, because it only applies to the sample in which the study was conducted. Let us illustrate this with a simple thought experiment.

Suppose we give an English reading test to 100 American youth. We ought to expect most of the differences to be genetic and not environmental. We would expect that almost all of the variation to be due to genes. But then suppose we expand our sample to include Japanese children who have no familiarity with English, and give them the same test. In our expanded sample virtually 100% of the variation would then be environmental. This is because not knowing the English language was not an environmental variable that was included in the first study.

For these reasons we should look at international comparisons of IQ differently then we look at comparisons among people with similar environments. If we did a heritability study of IQ that had both the United States and Africa in the sample, we would expect the heritability to be small, although to my knowledge no such a test has been performed.

Why did I write this post?

I choose to write this rather lengthy blog post because of an atrocious article by David Shenk in The Atlantic Monthly. Although the standard for mainstream media articles written on IQ is very low, Shenk seems to lower the bar even further by showing a complete lack of knowledge or familiarity with the subject that he is writing about.

http://correspondents.theatlantic.com/
david_shenk/2009/07/the_truth_about_iq .php

This article reminds me of the Creationists who point out a fatal flaw in the theory of evolution - that no one has ever been able to photograph an animal while it is evolving. I'm dead serious, I have literally heard people say that, as if evolution suggests that animals deform into a blob of silver luminescence before emerging "evolved" as they do on the Pokemon television show.

Shenk writes:

"What, then, is IQ? Conventional wisdom says that IQ scores reveal our native intelligence. According to this view, IQ tests are different from school grades, different from SAT scores, different from any other test you will ever take"

Actually the correlation between the SATs and the ASVAB test is about 80%. The correlation between the SATs and the Raven's Progressive Matrices IQ Test is about 70%. I'd imagine that school grades are just as close to IQ if one controls for variable grading standards.

But one thing that's important to remember is that no test is a perfect measure of general intelligence. Raven's Progressive Matrices is close, but as I mentioned earlier there are some questions that can be mastered by Pigeons! But it is used because it reflects the intellectual potential of the severely autistic who appear retarded on other IQ tests.

"because they somehow reveal the core, innate abilities of each person's brain: your clock speed, your RAM, your absolute limit. "

The ability for a person to rapidly react to stimuli is only slightly correlated to the general intelligence factor, as is one's short term memory (what a computer would call RAM). If you test people on how many digits they can repeat back to you backwards you will get a test that correlates to g twice as much as if you had simply asked them to repeat back the digits without reversing them.

I suspect that this is the case because the brain naturally stores numbers in forward link lists, because has great biological value. To repeat numbers backwards, one has to actually store all of the numbers and keep them in a position. I suspect that this is very important for academic success where one has to organize ideas that are not directly related to one another.

IQ is none of the things that Shenk lists yet it is also all of them. It is the most general mental talent that we know of.