The Lying Stones of Marrakech (44 page)

The common source of these errors lies much deeper than any crude correlation to a political utility that most of us do not even recognize and would disavow if we did. The source lies, I believe, in a general view about causality that has either been beaten into us by a false philosophy about science and the natural world, or may even record an unfortunate foible in our brain's evolved mode of operation. We favor simple kinds of explanations that flow in one direction from small, independent, constituent atoms of being, to complex and messy interactions among large bodies or organizations. In other words, and to use the technical term, we prefer to be “reductionists” in our causal schemes—to explain the physical behavior of large objects as consequences of atoms in motion, or to explain the social behavior of large animals by biological atoms called genes.

But the world rarely matches our simplistic hopes, and the admittedly powerful methods of reductionism don't always apply. Wholes
can
be bigger than the sums of their parts, and interactions among objects cannot always be disaggregated into rules of action for each object considered separately. The rules and randomnesses of particular situations must often be inferred from direct and overt study of large objects and their interactions, not by reduction to constituent “atoms” and their fundamental properties. The three common errors of genetic explanation all share the same basic fallacy of reductionist assumptions.

1. We regard ourselves as sophisticated when we acknowledge that
both
genes and environment produce a given outcome, but we err in assuming that we can best express such a correct principle by assigning percentages and
stating, for example, that “behavior A is 40 percent genetic and 60 percent environmental.” Such reductionist expressions pass beyond the status of simple error into the even more negative domain of entirely meaningless statements. Genetics and environment do interact to build a totality, but we need to understand why resulting wholes are unbreakable and irreducible to separate components. Water cannot be explained as two-thirds of the separate properties of hydrogen gas mixed with one-third of oxygen's independent traits—just as wanderlust cannot be analyzed as 30 percent of a gene for feeble inhibition mixed with 70 percent of social circumstances that abet an urge to hit the road.

2. We think that we have reached some form of subtle accuracy in saying that many genes, not just a Davenportian unity, set the hereditary basis of complex behaviors. But we then take this correct statement and reintroduce the central error of reductionism by asserting that if 10 genes influence behavior A, and if the causes of A may be regarded as 50 percent genetic (the first error), then each gene must contribute roughly 5 percent to the totality of behavior A. But complex interactions cannot be calculated as the sum of independent parts considered separately. I cannot be understood as one-eighth of each of my great-grandparents (though my genetic composition may be roughly so determined); I am a unique product of my own interactive circumstances of social setting, heredity composition, and all the slings and arrows of individual and outrageous natural fortune.

3. We suppose that we have introduced sufficient caution in qualifying statements about “genes for” traits by admitting their only partial, and often quite small, contribution to an interactive totality. Thus, we imagine that we may legitimately talk of a “gay gene” so long as we add the proviso that only 15 percent of sexual preference records this cause. But we need to understand why such statements have no meaning and therefore become (as for the first argument above) worse than merely false. Many genes interact with several other factors to influence sexual preference, but no unitary and separable “gay gene” exists. When we talk about a “gene for” 10 percent of behavior A, we simply commit the old Davenportian fallacy on the “little bit pregnant” analogy.

As a concrete example of how a good and important study can be saddled with all these errors in public reporting (and also by less than optimally careful statements of some participating researchers),
The New York Times
greeted 1996 with a headline on the front page of its issue for January 2: “Variant Gene Tied to a Love of New Thrills.” The article discussed two studies published in the
January 1996 issue
of Nature Genetics
. Two independent groups of researchers, one working with 124 Ashkenazi and Sephardic Jews from Israel, the other with a largely male sample of 315 ethnically diverse Americans, both found a clearly significant, if weak, association between propensity for “novelty-seeking” behavior (as ascertained from standard survey questionnaires) and possession of a variant of a gene called the D4 dopamine receptor, located on the eleventh chromosome, and acting as one of at least five receptors known to influence the brain's response to dopamine.

This gene exists in several forms, defined by differing lengths recording the number (anywhere from two to ten) of repeated copies of a particular DNA sub-unit within the gene. Individuals with a high number of repeated copies (that is, with a longer gene) tended to manifest a greater tendency for “novelty-seeking” behavior—perhaps because the longer form of the gene somehow acts to enhance the brain's response to dopamine.

So far, so good—and very interesting. We can scarcely doubt that heredity influences broad and basic aspects of temperament—a bit of folk wisdom that surely falls into the category of “what every parent with more than one child knows.” No one should feel at all offended or threatened by the obvious fact that we are not all born entirely blank, or entirely the same, in our mixture of the broad behavioral propensities defining what we call “temperament.” Certain genes evidently influence particular aspects of brain chemistry; and brain chemistry surely affects our moods and behaviors. We know that basic and powerful neurotransmitters like dopamine strongly impact our moods and feelings (particularly, for dopamine, our sensations of pleasure). Differing forms of genes that affect the brain's response to dopamine may influence our behaviors—and a form that enhances the response may well incline a person toward “novelty-seeking” activities.

But the long form of the D4 receptor does not therefore become
the
(or even a) “novelty-seeking” gene, and these studies do not show that novelty seeking can be quantified and explained as a specified percent “genetic” in origin—although statements in this form dominated popular reports of these discoveries. Even the primary sources—the two original reports in
Nature Genetics
and the accompanying editorial feature entitled “Mapping Genes for Human Personality”—and the excellent
Times
story (representing the best of our serious press) managed, amidst their generally careful and accurate accounts, to propagate all three errors detailed above.

The
Times
reporter committed the first error of assigning separable percentages by writing “that about half of novelty-seeking behavior is attributable
to genes, the other half to as yet ill defined environmental circumstances.” Dr. R. P. Ebstein, principal author of one of the reports, then stated the second error of adding up effects without considering interactions when he argued that the long form of the D4 gene accounts for only about 10 percent of novelty-seeking behavior. If, by the first error, the totality of novelty seeking can be viewed as 50 percent genetic, and if D4 accounts for 10 percent of the totality, then we can infer that about four other genes must be involved (each contributing its 10 percent for the grand total of 50 percent genetic influence). Ebstein told the
Times
reporter: “If we assume that there are other genes out there that we haven't looked at yet, and that each gene exerts more or less the same influence as the D4 receptor, then we would expect maybe four or five genes are involved in the trait.”

But the most significant errors, as always, fall into the third category of mis-proclaiming “genes for” specific behaviors—as in the title of the technical report from
Nature Genetics
, previously cited: “Mapping Genes for Human Personality.” (If our professional journals so indulge and err, imagine what the popular press makes of “gay genes,” “thrill genes,” “stupidity genes,” and so on.) First of all, the D4 gene by itself exerts only a weak potential influence upon novelty-seeking behavior. How can a gene accounting for only 10 percent of the variance in a trait be proclaimed as a “gene for” the trait? If I decide that 10 percent of my weight gain originated from the calories in tofu (because I love the stuff and eat it by the ton), this item, generally regarded as nutritionally benign, does not become a “fatness food.”

More importantly, genes make enzymes, and enzymes control the rates of chemical processes. Genes do not make “novelty-seeking” or any other complex and overt behavior. Predisposition via a long chain of complex chemical reactions, mediated through a more complex series of life's circumstances, does not equal identification or even causation. At most, the long form of D4 induces a chemical reaction that can, among other possible effects, generate a mood leading some people to greater openness toward behaviors defined by some questionnaires as “novelty seeking.”

In fact, a further study, published in 1997, illustrated this error in a dramatic way by linking the same long form of D4 to greater propensity for heroin addiction. The original
Times
article of 1996 had celebrated the “first known report of a link between a specific gene and a specific normal personality trait.” But now the same gene—perhaps via the same route of enhanced dopamine response—also correlates with a severe pathology in other personalities. So what shall we call D4—a “novelty-seeking” gene in normal folk, or an “addiction”
gene in troubled people? We need instead to reform both our terminology and our concepts. The long form of D4 induces a chemical response. This response may correlate with many different overt behaviors in people with widely varying histories and genetic constitutions.

The deepest error of this third category lies in the reductionist, and really rather silly, notion that we can even define discrete, separable, specific traits within the complex continua of human behaviors. We encounter enough difficulty in trying to identify characters with clear links to particular genes in the much clearer and simpler features of human anatomy. I may be able to specify genes “for” eye color, but not for leg length or fatness. How then shall I parse the continuous and necessarily subjective categories of labile personalities? Is “novelty seeking” really a “thing” at all? Can I even talk in a meaningful way about “genes for” such nebulous categories? Have I not fallen right back into the errors of Davenport's search for the internal scarlet letter
W
of wanderlust?

I finally realized what had been troubling me so much about the literature on “genes for” behavior when I read the
Times's
account of C. R. Cloninger's theory of personality (Cloninger served as principal author of the
Nature Genetics
editorial commentary):

Novelty seeking is one of four aspects that Dr. Cloninger and many other psychologists propose as the basic bricks of normal temperament, the other three being avoidance of harm, reward dependence and persistence. All four humors are thought to be attributable in good part to one's genetic makeup.

The last line crystallized my distress—“all four humors”—for I grasped, with the emotional jolt that occurs when all the previously unconnected pieces of an argument fall suddenly into place, why the canny reporter (or the scientist himself) had used this old word. Consider the theory in outline: four independent components of temperament, properly in balance in “normal” folks, with each individual displaying subtly different proportions, thus determining our individual temperaments and building our distinct personalities. But if our body secretes too much, or too little, of any particular humor, then a pathology may develop.

But why four, and why these four? Why not five, or six, or six hundred? Why any specific number? Why try to parse such continua into definite independent “things” at all? I do understand the mathematical theories and procedures that lead to such identifications (see my book
The Mismeasure of Man)
, but
I regard the entire enterprise as a major philosophical error of our time (while I view the mathematical techniques, which I use extensively in my own research, as highly valuable when properly applied). Numerical clumps do not identify physical realities. A four-component model of temperament may act as a useful heuristic device, but I don't believe for a moment that four homunculi labeled
novelty seeking, avoidance of harm, reward dependence
, and
persistence
reside in my brain, either vying for dominance or cooperating in balance.

The logic of such a theory runs in uncanny parallel—hence the clever choice of “humor” as a descriptive term for the proposed modules of temperament—with the oldest and most venerable of gloriously wrong theories in the history of medicine. For more than a thousand years, from Galen to the dawn of modern medicine, prevailing wisdom regarded the human personality as a balance among four humors—blood, phlegm, choler, and melancholy.
Humor
, from the Latin word for “liquid” (a meaning still preserved in designating the fluids of the human eye as aqueous and vitreous humor), referred to the four liquids that supposedly formed the chyle, or digested food in the intestine just before it entered the body for nourishment. Since the chyle formed, on one hand, from a range of choices in the food we eat and, on the other hand, from constitutional differences in how various bodies digest this food, the totality recorded both innate and external factors—an exact equivalent to the modern claim that both genes and environment influence our behavior.

The four humors of the chyle correspond to the four possible categories of a double dichotomy—that is, to two axes of distinction based on warm–cold and wet–dry. The warm and wet humor forms blood; cold and wet generates phlegm; warm and dry makes choler; while cold and dry builds melancholy. I regard such a logically abstract scheme as a heuristic organizing device, much like Cloninger's quadripartite theory of personality. But we make a major error if we elevate such a scheme to a claim for real and distinct physical entities inside the body.