, , ,

Here are observations on the ostracism of James Watson, stripped yesterday of all his titles and honors at Cold Spring Harbor Laboratory, the research lab he led for many years:

If the same word, racist, refers to a ninety-year-old scientist who draws incorrect conclusions from data, and neo-Nazis who want to kill Jews and people of color, I believe we need a new vocabulary. The word racist does not mean anything to me in this context.

Watson is not the first scientist to draw erroneous conclusions about race and intelligence, based on tests that supposedly yield an intelligence quotient, where quotient is the amount of a specified quality or characteristic. When a method yields results that are so plainly incorrect, as IQ tests do, why do we not question the method? Why do we not criticize Watson for citing preposterous, pseudo-scientific data, rather than criticize him for being a racist?

Intelligence does have a general meaning, which comes from the Latin, intelligere, or understand. So most would agree that when we say a person is intelligent, we refer to their faculty of understanding, or comprehension. We also routinely use the word as praise, a compliment, even an honorific. I will argue below that although the general definition is coherent enough, when we try to measure this quality, we discover it is neither measurable nor coherent.

Your score on a so-called intelligence test is just a score on a test. Reasoning that suggests you can draw conclusions about intelligence from a test score is mistaken: today we would say fake. How do we know the reasoning and conclusions are spurious? Because the conclusions are clearly impossible. The test pretends to measure inherent or inherited intelligence – note the common root, inhere – when you cannot possibly assess intelligence at birth. Any measurement of inherited intelligence after birth is clearly invalid. If we have no way to measure intelligence at birth, why would anyone want to draw conclusions about inherited intelligence, based on a test administered long after you are born?

Your score on a so-called intelligence test is just a score on a test.

At this point one might say to Watson, to support conclusions about his racism: if people of color score higher on intelligence tests than whites, you would pay less attention to these results. If test results contradict your beliefs, dismiss the results. If scores confirm beliefs you have already formed in other quarters of life, you pay more attention to them, and talk about them if we encourage you to do so. If one suspects racist beliefs, one can induce an elderly scientist to take one more step into permanent disgrace. Anyone can play the race-baiting game.

The fact remains – no matter what you think about Watson – any score distribution for these tests is merely that: a score distribution for a given test. You cannot draw conclusions about relative intelligence among groups based on score distributions, any more than you can draw conclusions about relative intelligence among individuals based on individual scores. Remember that the entire purpose of intelligence tests is to compare scores, first among individuals, then among groups. That purpose should make you question the validity of the test.

I could draw out a fairly long list of pseudo-scientific endeavors. They all share a common quality: incorrect reasoning from observations leads to outcomes that have no foundation in the original observations. Indeed, the outcomes have no foundation at all.

We observe differences among people, including differences in intelligence, and conclude we should increase the number of intelligent people in the population. Thus eugenics promotes sterilization programs. We observe that certain patterns of behavior cause difficulties in a family or in a community, and try to locate these problems in the troublemaker’s brain. Thus abnormal psychology promotes frontal lobotomies. We observe that liars sometimes experience stress while they tell a lie, so we develop a machine to detect stress. Thus forensic science promotes lie detector tests.

Shall I go on? In each case, we enthusiastically adopt methods that appear to yield both accurate measurements and desirable outcomes – when in fact the measurements are dubious or false, and the outcomes abominable: sterilized men and women who can never have families, lobotomized vegetables who will never again cause trouble, defendants sent to prison for crimes they did not commit. Judge the so-called science by its fruits.

We enthusiastically adopt methods that appear to yield both accurate measurements and desirable outcomes – when in fact the measurements are dubious or false, and the outcomes abominable.

If a multiple-choice test leads to invidious conclusions about intelligence among groups who have differing amounts of melanin in their skin, would you not recognize that as an absurd conclusion? Would you not say the conclusion invalidates the assessment method? To use today’s vocabulary once more, would we not call assessments that measure inherited intelligence a hoax? Instead we call people who fall for such trickery racists, and we do not question the method of measurement at all! Why would we not say, “You can think what you like about group intelligence, Mr. Watson, but you have to reexamine your reasoning. Why would you draw conclusions about intelligence in groups from a test that so clearly fails to measure what it claims to measure?”

In fact, a victim of bad reasoning, or someone confused about what conclusions you can or cannot draw from a given set of data, is not a racist. Unlike neo-Nazis in Charlottesville, Watson does not claim supremacy for his race. He does not argue that Western civilization is superior to other civilizations, or suggest you can evaluate civilizations based on predominant skin coloration. People who condemn and ostracize him, however, attribute exactly those arguments to him when he makes comments about intelligence and race. If you say one thing, x, you must believe these other things, y and z.

Watson may look around him and say, “Well, these intelligence testers have all the same credentials I do. They have doctorates, they publish in academic journals, they draw pay from established, prestigious research institutions. Why should I disbelieve their results?” One can answer with equal confidence, “Within the large framework of molecular biology, your double helix model explains how nucleic acids in DNA propagate inherited characteristics. Within the large framework of human behavior and neural networks, standardized tests that claim to measure differences in inherited intelligence among individuals and groups do nothing of the kind.”

Unlike neo-Nazis in Charlottesville, Watson does not claim supremacy for his race.

Ostracism is a serious sanction. Better to ask James Watson why he believes intelligence tests yield valid results. Ask why score distributions for these tests merit our attention. Most importantly, ask what we mean when we talk about intelligence. If you cannot give a coherent answer to that question, you have nothing to measure. Alfred Binet and Theodore Simon developed the first intelligence test in 1904. I am willing to guess, without reading the history of intelligence measurement, that the original Stanford-Binet test attempted to measure an ineffable quality not so different from the quality we try to measure with similar tests more than a hundred years later.

If you ask, many intelligence testers will define intelligence in terms of their test. “Don’t you see,” they would say, “the ability to recognize patterns and sequences, to translate these patterns from one form to another, these are marks of intelligence.” Yes, Bobby Fischer had a remarkable ability to recognize and analyze patterns; people judge that he had an IQ of about 175. His ability to play chess did not prevent him from being extraordinarily stupid in other areas of life. If you start and end with the assumption that good chess players are intelligent, there you have your definition of intelligence.

For the most part, intelligence testers cannot formulate a definition independent of what they measure when they administer their test, which is to say, independent of the skills required to mark answers that match the tester’s key. Here most people would pause and say, “Hold on, if you administer a test that measures one’s ability to do well on the test, you have nothing but a circular process. You want to evaluate some quality or variable external to the test.”

If you start and end with the assumption that good chess players are intelligent, there you have your definition of intelligence.

You would find quickly enough, if you explore the concept, that intelligence is a disjointed concept, and given both its cultural and its social implications, an essentially contested one. You cannot measure the amount of intelligence in a person, or in members of a group, when the concept itself has no agreed meaning. You can only measure one’s ability to mark correct answers for a particular set of questions. If you want to call that ability intelligence, you are welcome to it, but others will say your definition has little significance beyond the test.

To conclude, compare religious conceptions of divine intelligence with our Western, analytic definitions. John’s treatise on the nature of God begins, “In the beginning was the Word, and the Word was with God, and the Word was God.” We do not develop multiple choice tests to measure the Word’s presence in an individual. If you are a religious fanatic, you put people to death if they do not have the correct presence. You measure fidelity, and judge infidels not fit to live. Historically, religious fanatics do not claim to have science on their side, as Galileo learned. Here in the West, we do claim the authoritative weight of scientific reasoning, sometimes for conclusions that have nothing to do with science.

Related articles

Intelligence quotient

Race and intelligence