A collection of essays by Bill (website@ccjj.info) accompanied by feedback from his friends.

Wednesday, April 14, 2010

Book Review: The Genius in All of Us, by David Shenk

The following post was published on The Gotham Skeptic on April 7, 2010.

On Saturday, March 13, David Shenk, the author if "The Genius in All of Us" delivered a lecture to the New York City Skeptics. The book's press release promised a lot, saying Shenk would give us reason to "Forget everything you think you know about genes, talent, and intelligence.". Shenk said there was a "mountain of evidence" for a level of "talent abundance" that we had previously not known about. The talk totally failed to deliver on these promises, as did his book.

I am reminded of the claim that "we all use only 10% of our brains" that was discussed in the latest issue of "Skeptic" magazine, the implication being being that there should be some way to tap into that other 90%. How does one confirm or refute such a claim? But the burden of proof is on Shenk, since he did promised us a "mountain of evidence". Shenk mentions one experiment with a sample size of one, where a college student with average IQ was taught, over the course of a year, to memorize sequences several dozen digits long, but it is questionable whether this is a useful skill, let alone a meaningful increase in "intelligence". Shenk never gives clear evidence that the average human can be made into a genius, in fact the general thrust of his book is to dismiss evidence rather than provide any, unless you're willing to be persuaded by anecdotes, which a skeptic shouldn't be.

To begin with, what, precisely, is the thesis of the book? Shenk gives (mostly anecdotal) evidence, 99% of the time, for the idea that talent and intelligence have nothing to do with genes. He does discuss a concept that intelligence is not a matter of "G + E" (genes plus environment), as many experimenters have analyzed it, but rather "GxE", or a complex interaction of genes and environment (though in his prose, Shenk interprets this as mostly environment). But where is he taking us with this? Is he providing us with another function that fits the experimental data better? No, he just says "it's complex", doing everything he can to render the problem intractable, manufacturing an absence of evidence that will leave him free to draw any conclusion he wants, supported by anecdotal evidence alone. However, he realizes there is strong scientific evidence in favor of a substantial genetic influence, so he occasionally covers himself with statements like "The blank slate is dead. Genetic differences do matter" that contradict what he's saying most of the time. So, in a nutshell, his point is "Genes don't matter! Genes don't matter! Genes don't matter! (Except they do.)". This double talk, this not really admitting what he's obviously out to prove, is shifty and infuriating.

I don't think this "GxE" business comes as news to genetics researchers. Everyone has known all along that the interaction between genes and environment is complex. But when you're doing research and taking measurements, you have to start somewhere, make approximations, and fit data to simple curves. Shenk wants none of that, data and statistics are his enemies, he wants to live in the world of the purely anecdotal, because you can support pretty much any conclusion you want that way.
More from Shenk on statistics: "knowing the average lifespan doesn't tell me how long my life will be". Of course it doesn't tell me exactly how long my life will be, but that doesn't mean it's a useless bit of information. If I am trying to assess how long I will live, say in order to plan for retirement, the average life expectancy for someone of my age and gender is one of the first things I will look at. What is Shenk offering us as a replacement for the evils of statistics? One anecdote after another -- hardly an improvement.

Shenk spends only 10 pages discussing twin studies, which is way too little given that twin and adoption studies provide most of the evidence he wants to refute. Much of these 10 pages are wasted on anecdotes. He does, accurately, report that twin studies have determined the heritability of 60% of IQ, 60% of personality, 40-66% of motor skills, and 21% of creativity.
He then launches into a diatribe about how he doesn't like the term "heritable", pointing out that height is 90% heritable but that doesn't mean that "90% of my height comes from my genes and 10% from my food", which I agree with, but where is he taking us with this? He does not enlighten us with another, more meaningful way to interpret the word, rather his motivation is to give us an excuse to ignore it altogether. The fact that the IQ of adults is at least 40% heritable given a wide range of environments tells us a lot. It is a very meaningful fact.
Shenk attributes some of the observed similarities of identical twins reared apart to "hidden dissimilarities", that is, researchers reporting coincidental similarities while not recording all the dissimilarities, and "coordination and exaggeration", where twins who were raised apart but had met prior to the study would coordinate their scores on tests to be similar. Most twins reared apart and studied as adults had met each other years before being studied.
Regarding the "hidden dissimilarities" I agree that may be a point, and my response to it is to ignore all the stupid anecdotal coincidences that most sources (including Shenk) bring up when discussing twin studies. Regarding the "coordination and exaggeration", let me describe the methodology of the study where Thomas Bouchard at the University of Minnesota studied over 100 pairs of twins reared apart:

"Participants complete approximately 50 hours of medical and psychological assessment. Two or more test instruments are used in each major domain of psychological assessment to ensure adequate coverage (for example, four personality trait inventories, three occupational interest inventories, and two mental ability batteries). ... Separate examiners administer the IQ test, life history interview, psychiatric interview, and sexual life history interview. ... The twins also complete questionnaires independently, under the constant supervision of a staff member."

For twins to "coordinate" scores on IQ tests that they take independently and under supervision would be quite difficult. They would have to somehow assess not only which of the two of them was smarter, but also how much smarter, and have that person deliberately do more poorly than they could have on the test, but not
so poorly that they do worse than their twin. A substantial fraction of the test sample would have to be doing this, the whole effort never being discovered by the researchers. This is so farfetched that it's just not a credible explanation for the high correlation of twin IQ test scores, across multiple studies.Another argument Shenk makes is that early environment, both pre-natal and after birth prior to separation, is shared by twins, so that the similarity observed by twin studies may be caused by environment rather than genes. This would easily be examined by comparing the similarity of fraternal twins versus the similarity of non-twin siblings, I don't know of any studies that did this, nor does Shenk bring any up. I'll bet it has been done.
Another study mentioned by Shenk was in The Skeptical Inquirer; it tested the similarity of identical twins versus the similarity of strangers. The study found, not surprisingly, that when they assembled 25 pairs of same-sex, roughly same age strangers on a college campus, along with 13 pairs of identical twins, they found that some of the strangers had "uncanny" anecdotal similarities to rival those of the twins. Since I try to ignore anecdotes anyway, this doesn't have much impact on me. What Shenk didn't mention was that when the study subjected everybody to systematic, independently taken personality tests, they found the pairs of identical twins were more similar to each other than the pairs of strangers were, despite the small sample size.
Shenk points out one study by Turkheimer (more on it later, and see link at end of article), that studied children raised at or near the poverty line, and which found the genetic influence on IQ was near zero, which shows that a really bad environment can make a big difference, overwhelming the genetic influence.
For all of Shenk's criticism of twin studies, he doesn't suggest how a better study would be done. He is interested in dismissing evidence, not in exploring evidence that could either support or undermine his position.

There is a lot of evidence that Shenk does not discuss. The evidence for genetic heritability of IQ is not limited to studies of identical twins reared apart, there are also studies which compare the IQ of adopted non-twins to (a) the parents who raised them and (b) their biological parents. There are studies that compare correlations between identical twins (0.86), between non-twin siblings (0.47), between half siblings (0.31), between cousins (0.15), and so forth. Shenk never mentions that these studies exist, except that he mentions Turkheimer's study (which he thinks supports his thesis) without explaining that it did not involve adoption. A couple of things that arise from all these studies, neither of which Shenk ever mentions, are that

  • as the child gets older, the genetic influence on IQ increases rather than decreases, which surprised everybody.
  • the correlation of adult IQ between two people raised in the same family is very small, in some studies no more than two equally-related people raised in different families. The total non-genetic effect on IQ is pretty large, but the majority of this variation still seems random to researchers. Factors that would go along with the adoptive family, such as school quality, IQ of parents in the home, parenting style, and quantity and quality of books in the home, seem to have little influence.
The tendency of the effects of genetics to increase, rather than decrease, with age is explained by some by the fact that, as people create their environments as they get older, the genetic influence makes itself felt in their choice of environments. I am reminded of something Eddie Van Halen said: "If you want to be a great guitarist, you've got to like playing the guitar.". Similarly, if you want to be a genius, you've got to like thinking. And whether you like thinking may be genetic. Most people aren't that crazy about thinking, as a glance at the reading material near the checkout counter of any supermarket will tell you. So the people with genes that make them like to think will exercise and develop their minds more than would other people. Young children are in school whether they like it or not, being forced to think whether they like it or not, so the genetic influence on IQ at those ages is less.

One weakness of adoption studies is that poor people are ineligible to adopt, so that few adoptive households are really bad. Turkheimer's study got around this and measured poor children by studying children with varying degrees of genetic commonality raised by biological relatives. Adoptive families range from solid working class to extreme upper class, and over this range, adoptive homes are observed to make little difference. The feeling is that while a really bad environment can make a big difference, any environment at or above what we would consider "adequate" (the bar being pretty low) will make little difference. Note that Turkheimer's study only measured children at age 7, when the genetic influence is known to be weaker.

One interesting study discussed by Shenk involved rats running mazes. In 1958, Rod Cooper and John Zubek deliberately bred one group of rats to be smart at running mazes, and another group to be dumb at it. With time, they were able to observe a large genetic difference in performance between the two strains. Then they tried raising both the smart and dumb strains in a very unstimulating, limited environment. Like the humans in the Turkheimer study, the rats all did poorly and almost no difference in performance was seen. Then they tried raising both smart and dumb rats in an especially enriched and stimulating environment. Now the dumb rats were doing almost as well as the smart rats, again the difference in performance due to genetics was greatly diminished.
It should be noted that the data was shown in a plot. But it wasn't a plot of the actual data -- it was an artist's depiction of the data. I've seen research papers where they give actual data and artists depictions, and the depictions are usually wildly skewed in favor of the point the paper is trying to make.

I would have liked to read more detail about the study, but I couldn't find it on the web. The ramifications for humans raised in poverty were clear and not very surprising given the other studies we've looked at. But what about the "enriched" environment? Twin and adoption studies include children adopted into wealthy homes, and the effect of those homes is observed to be small. Could there be some way of "enriching" our environments to radically improve most people's intellects, like those of the mice? Thousands of educators, some of them very well-funded at the best schools, have been trying for a long time. Some major breakthrough may be possible, you never know. We should certainly keep trying, for the same reason we should keep doing nuclear fusion research -- though the odds are long, the potential payoff is huge.

All this talk about IQ leads me to ask the question, what about other outcomes? What about annual income, marital status, number of divorces, and number of offspring? It is possible that IQ tests have succeeded in measuring a largely genetic quality, but other outcomes might be more affected by adoptive parents. For example, one twin might major in art and wind up being a waiter at $20K per year, while the other twin's adoptive parents might refuse to pay their college tuition unless they do something more practical, so that they wind up being an accountant at $120K per year. Neither Shenk nor anything else that I read on twin studies discusses this.

Genetics denialism has been with us for a long time. During the sixties the consensus was that genes were pretty unimportant, but studies kept showing otherwise. There are a few reasons people want to deny the influence of genes on the human mind:

  1. Fear of Eugenics: Memories of the excesses of the Eugenics movement in the early 20th century, and the horrors of the Nazis, lead many people to fear that misapplication of genetic theories could be a Very Bad Thing.
  2. Social Plans: Many people have plans for the betterment of human society that rely on society's under performers being capable of better things under more favorable circumstances, and when science casts doubt on these pet plans, these people attack the science.
  3. Mysticism: People want to believe there is something "magic" about the human mind, and breaking down the influences on it to things as vulgar as neurotransmitters and genes is just so -- unflattering. For the same reason that many resist the belief that humans are descended from apes, many who accept the evolution of our bodies try to assert that it "stops at the neck", and our intellect is somehow above all that sordid business.
None of these are legitimate reasons to ignore or suppress the truth. As for the first: if you don't want to commit atrocities, then don't commit atrocities. It is my instinct that people in denial are more likely to do horrible things than people who aren't. Regarding the second, social plans that rely on flawed science in order to work aren't going to work -- time to develop new social plans. And the third reason is just inexcusable mush-mindedness.

There is much in the book I haven't covered here, some of it quite interesting, but this review is pretty long as it is. I have focused on the evidence that Shenk has to refute to make his point, and it is there that Shenk is at his worst. I have been a bit vehement, because of Shenk's taste for lengthy anecdotes, his attempts to not admit what his thesis really is, and his dismissal of the whole concept of statistical evidence.

If you want to read a book promoting genetics denialism, there is another, better book I recommend: "Intelligence and How to Get It" by Richard Nisbett. Nisbett claims to prove that normal variation in human genetics has little to do with intelligence. I wasn't persuaded by his arguments, and, like Shenk, he doesn't bring up all the evidence against him, but at least there isn't the double talk, the obsession with anecdotes, and the disdain for statistical evidence in general.


  • Book: "The Genius in All of Us" -- David Shenk
  • One of dozens of papers on Bouchard's Minnesota Twin Study: here
  • Book: "Born That Way - Genes, Behavior, Personality" -- William Wright
  • Book: "Intelligence and How to Get It" -- Richard Nisbett
  • "The Search For Intelligence" - Carl Zimmer, Scientific American, October, 2008
  • "Trends in Behavioral Genetics: Eugenics Revisited" - John Horgan, Scientific American, June 1993
  • Book: "The Blank Slate" -- Steven Pinker
  • "Natural Levels of Similarities Between Identical Twins and Between Unrelated People" - Joseph Wyatt, The Skeptical Inquirer, Vol 9, Fall 1984
  • Turkheimer's Study on Socioeconomic Status Affecting the Heritability of IQ: here
  • Statement by 52 scholars on intelligence published in "The Wall Street Journal", December, 1994: here
  • Longer statement by 11 scholars on intelligence published in "American Psychologist" in February, 1996: here
  • Wikipedia on Heritability of IQ: here
  • Wikipedia on Heritability of Personality: here

No comments:

Post a Comment