Wednesday, April 15, 2020

Book Review: Human Diversity

The full title of Charles Murray's recent book is Human Diversity: The Biology of Gender, Race and Class. It's a dense, tightly-reasoned, data-packed, expertly-written tome that will change the world. Or rather, the subject it discusses will change the world.

The major purpose of the book is to bring the educated layman up to date on the science of human genetics. Thanks to spectacular advances since 1980, and especially since the complete mapping of the genome in 2005, it is now possible to shed some serious light on the old nature vs. nurture question. While these results have until now stayed mostly within the specialist, scientific community, it is Mr. Murray's view that before the end of this decade, the consequences of this new knowledge will impact every aspect of our lives.

A short, oversimplified summary of the story goes like this: Back in the old days (before 1980) it was supposed that most human traits were the result of a gene or two, or perhaps maybe half a dozen. In this context we are using the word gene to refer to a bit of DNA that actually codes for a protein. In those days people supposed there were 100,000 or more genes.

This had numerous consequences. First, there was fear of genetically modified babies. If we could just change a specific gene, then we could make children much more intelligent. This possibility was never remotely realistic.

Then people assumed that evolution happened very slowly. Most mutations in actual genes are deleterious--resulting in birth defects or death. Only a very few are beneficial, and even those will need about 2000 generations (40,000 years) to spread from one individual to the entire human population.

Even then, it was understood that the bulk of our DNA was not part of a gene, i.e., didn't code for anything. Folks didn't know what that excess DNA did--it was often called junk DNA. Today we know different. It appears there are only about 20,000 actual genes in the entire genome. By far most DNA is in the non-coding portion. It appears that non-coding DNA regulates gene expression in some complex way.

While mutations in the genes are still usually deleterious, mutations in the non-coding parts of the genome tend to not have much effect. Therefore mutations accumulate there. Many of them spread across the human species: certain mutations that occur in more than 1% of individuals are called single nucleotide polymorphisms, or SNPs (pronounced snips). Wikipedia claims that more than 335 million SNPs have been found in our species, while any given individual will have between four and six million SNPs.

While an individual SNP has little effect, cumulatively the SNPs make a big difference. For example, it is now known that millions of SNPs help determine how tall you will be. With enough data and a big enough computer, one can isolate those SNPs that correlate with height.

So today we understand there isn't a "gene" for height, nor for personality/behavioral traits like intelligence, empathy, honesty, conscientiousness, etc. Yet we know that all these traits are significantly heritable--that is depend on our DNA. But instead of counting genes by the handful, we count SNPs by the thousands. Thus almost all traits (both physical and behavioral) are polygenic, in some cases nearly omnigenic (involving contributions from nearly all your DNA).

It's the SNPs that drive human evolution, and sufficiently fast that evolution can (and does) happen in historical time.

It is possible to count SNPs that enhance a certain trait, such as height or tendency toward schizophrenia. This leads to the polygenic score for that particular trait. Since your SNPs are fixed at birth, one can readily calculate your odds of being schizophrenic later in life. (It will never be 100%--environmental effects alter the odds significantly).

Likewise, there is a polygenic score for IQ. It cannot predict your IQ as an adult--for which the environment contributes about 40% of the variance--but it definitely narrows the field. If your child is born with a high polygenic score for IQ, then it might be worth saving money to send her to Harvard.

There are polygenic scores for things like criminality. The optimist will say that by judicious interventions we can prevent that outcome from occurring. The pessimist will fear that we'll be punishing people in advance of any crime--simply on the basis of their polygenic scores. I think they're both probably right--it is a brave new world coming on.

A second theme in Murray's book is an argument against the socialization theory. This theory--standard groupthink at American colleges--maintains that all differences between genders, races, and classes are merely social constructs, and have no underlying basis in biology. I'll mention that prior to 1980 this was a plausible idea, but since then, and especially since the rise of modern genomics, it becomes completely untenable.

Mr. Murray discusses at length the differences between men and women, who differ by an entire chromosome. There are genetic, physiological and psychometric differences between the sexes, compared to which "socialization" pales into insignificance. That said, it is not a simple switch--there is a continuum between male and female in nearly all traits. Few men have exclusively "male" traits, while few women are exclusively "female."

Murray uses facial features as an example. Consider the kinds of metrics facial recognition software might use: relative length of nose, relative distance between the ears, etc. Suppose there are twenty such metrics. It may be that women have slightly shorter noses (relatively) than men. But the effect is not very large--many women have noses longer than the average man, but most women don't.

So it is with all the other metrics. Women congregate slightly on one side of the spectrum, while men predominate mostly on the other, with a large overlap in the middle. Very few men are "masculine" on all twenty measures, and neither are many women "feminine" on all measures.

And yet it is easy to distinguish male faces from female faces. Even though each effect is small, the accumulated effects are big enough to see the difference. Only in a few cases is it truly ambiguous--I think of Rachel Maddow, for example. Two corollaries: facial recognition software can distinguish between male and female faces most of the time; it takes a skilled artist to draw an obviously female face, since most of the differences are so subtle.

Similarly, racial and ethnic differences show up in polygenic scores. That's how sites like 23&Me work their magic. Those very same polygenic scores imply ethnic differences in disease susceptibilities, physical attributes, and personality traits. Race is not just "skin deep" as the pre-1980 school would have it.

Class differences also depend on polygenic scores.

I'll close with some comments on how I think Murray's book will be received, especially by my former colleagues at Local State College.

First, most of the people who fancy themselves as experts on Race, Class and Gender are committed to the social construct theory and won't change their minds. But science progresses one retirement at a time, and these folks--tenured baby boomers--are not that long for the academic world. I don't think they're being replaced.

Second, those same self-proclaimed experts are typically in English and education departments. They know nothing about statistics. (Yes, I know--education faculty should know some stats, but I don't think they generally do.) One doesn't need a lot of statistics to read Murray's book, but a college-level introduction to the discipline strikes me as a prerequisite. (Sociologists have no excuse here.)

Finally, "Charles Murray" is a dirty word in those quarters. He should have written the book under a pseudonym. As the book has just come out, I can't blame them yet for not having read it, but they won't have read it five years from now, either.

So yes, while it will pass the academy by, I predict there will be a revolution in our understanding of human behavior. I'm not sure it will all be good. Lots of people (e.g., inhabitants of red states) will be classed as vaguely subhuman because of their average polygenic scores, as will many other groups of people.

There are some things this book should never change, concepts of morality, justice, and human worth among them. No polygenic score is an excuse to deny anybody their human rights. As Mr. Murray recently opined, religion becomes more important. Absent some kind of revival, the temptation to equate polygenic scores with human worth becomes irresistible.

You should definitely read Charles Murray's book. But whether you do or not, for better or worse it will definitely change your life.

Further Reading:

No comments:

Post a Comment