Saturday, May 3, 2014

How Genes Affect Learning

Genetic influences on behavior, and learning, are two things that interest me deeply so I picked up G is for Genes, ( 2014) by a behavioral geneticist from King's College, London, and a psychologist from the University of York, U.K.  The book addresses this question: how do genes affect learning?

This is not at all about eugenics or I wouldn't have read it.  The authors are interested in understanding exactly how genes affect learning so that we can narrow the learning-gap more skillfully.  I’ll try to peel out some of the most surprising, interesting, and useful bits, and I’ll draw on some courses I’ve taken recently too – Philosophy of the Mind by Patrick Grim of the State University of New York and Understanding the Mysteries of Human Behavior, by Mark Leary of Duke.

Height is heritable of course.  If your parents are both tall you're likely to be tall too.  In the U.S. , for example, for white males, it's about 80% that way;  the rest is due either to the shared environment (like a particular family or school) or a non-shared environment (experiences unique to the individual).  In the case of height, being raised in an affluent home is a shared experience and being stricken by a childhood illness is non-shared.  Both might affect your height.

 But in some parts of the world height is environmental – genes account for no more than half.   Why?  Not because the genes matter less, it’s that the environment matters more – there’s more variation.  In Somalia some people don’t have enough to eat, and some have plenty.

The same thing is true about education, of course.  Look at these findings. Eighty percent of reading differences in Australian children after kindergarten are genetic.  In Colorado it is just 66% and in Scandinavia, 30% of reading skills are genetic.  What's going on?  Australian children at the age of 5 go to school from 9 to 3 p.m.  Colorado requires just 3-4 hours a day, and in Sweden and Norway compulsory schooling begins at 7.  When some go to to preschool and some don’t, some kids have parents who read to them and some don’t -- the environment really matters.  By the age of 10 reading levels are 80% genetic in all three countries.

So is there a “reading gene?”  No, like most things, many many genes interact to affect many many things.  Genes are generalists, the authors say.  The environment is a specialist.  But how do we know that 80 percent of reading ability is due to DNA?  The answer to these sorts of questions usually involve twins:  identical twins raised apart or adopted children in the same shared environment.  Fraternal twins who are more similar than sequential siblings tell us something about the shared environment.  Steven Pinker says we should throw out almost all studies on parenting unless they control for heritability – maybe Johnny is a pistol because he had angry parents who spanked him, but then again maybe anger just runs in the family.  You need lots of twins to sort that out.

I’ve read many times that much of personality is genetic -- approximately half, depending on the trait. Here is a typical breakdown for personality, based on twin studies of course:

Back to learning.  When it comes to IQ the percent heritable, as you might expect, changes with aging – probably because the environment has the most impact early on, then less throughout most of life.  

Whatever it is that is measured by the IQ test is largely carried in DNA.  The authors of G is for Genes explored many nuances about IQ but didn’t really describe the measure itself as thoroughly as I had hoped and expected, so let me draw primarily on Dr. Grim’s lecture 14: Intelligence and IQ:

The test was designed in the early 1900s by Alfred Binet, a Frenchman, who had recently abandoned an attempt to correlate academic prowess with skull size (First, he couldn’t get a stable measure, and then he found no correlation).  His new test, made up of a battery of questions (he said the questions didn’t matter much, there just had to be numerous) intending to assess a child’s mental age.  He was attempting to characterize lagging students as “slow” rather than “sick," and thereby keep them out of the asylum.  Soon someone divided his score by chronological age, hence the “quotient” began.  Much later standard deviations above and below average were used as they still are today.  But Benet had explicitly stated that his test was not a single measure of ability, just a predictor of achievement, it should not be used for ranking normal children, and what it measured was neither innate nor immutable.  On the contrary, the whole purpose was to help slower children do better.

It was co-opted a few years later by the American H. H. Goddard, a eugenics enthusiast, and he used the measure in exactly the ways Binet had warned against.   Goddard thought the IQ was a measure of the whole person, he considered intelligence innate, and he thought it was immutable.  Many people still do.  He used these now-common terms for those at the lower end of the scale: idiot, imbecile, and moron. And it was on the basis of his work that Justice Oliver Wendell Holmes Jr. upheld a Virginia state law in 1920 forcing sterilization on those scoring poorly on this exam.  Holmes' infamous quote:“three generations of imbeciles is enough.”

Now the test is known to be sensitive to culture, reading skills, age, level of development, and practice – one can easily improve their IQ score through study.  And many now argue that there are different sorts of intelligence, pointing as evidence to savants who may be amazing at one skill and abysmal at another – or at brain injuries which impair one ability specifically. The following intelligences are often suggested: linguistic, logical/mathematical, spatial, musical, kinesthetic, interpersonal, intrapersonal, naturalist, and emotional.  Darwin is an example of naturalist intelligence, Einstein was logical/mathematical.  Mozart was musical. Buddha and Ghandi were intrapersonal.

G is for Genes would have benefited from such a closer look at IQ, which Asbury and Plomin refer to simply as a “g score,” but they do point out some interesting and unexpected things about it. 
“We asked thousands of children, parents, and teachers about class sizes, school buildings, resources like books and computers, chaos in classrooms, and a whole host of other oft-cited factors and yet, when we fed their ideas into genetically sensitive studies, these factors … accounted for almost none of the differences between our children in terms of their achievement.  … The environment within the school, it appeared, had no impact on children’s academic performance.”  (p 115-16)

Then what did? According to the authors, this startling finding was the reason they wrote the book.  After considerable effort they found that the answer might be in the interpersonal relationships students have in school -- the non-shared experiences such as interactions with peers and teachers.

They also found that while IQ or g is the greatest predictor of academic performance at later years, but it’s less important for the very young for whom shared environmental factors are more important (just like height in poor regions).  If you remove the effect of IQ altogether, achievement is still 40% heritable for other reasons.   One of these is confidence, they say.   It boosts academic achievement and what’s more, it's is at least as heritable as IQ itself -- self confidence is, in large part, genetic.  Socioeconomic status also has a bearing on achievement, and surprisingly it is heritable too.  It breaks down this way: 50% of educational attainment is heritable, 40% of occupational status is, and 30% of income is related, somehow, to DNA.  These results are from twin studies, I assume, although the book was not as well referenced as I would like.

Gender differences were discussed as well.  Two thirds of math ability is genetically influenced in both genders, and boys and girls have similar average abilities in the hard sciences.  However, there are two reasons why males might dominate the sciences.  First, although the averages are similar, there is more variability among boys.  That is, girls tend to be clustered near average in scientific aptitude, where more boys are extra good, and more are very bad.   The best boys are better than the best girls (and the worst worse than the worst).

Second, adults tend to study what they like, not what they’re good at; aptitude does not necessarily predict choice.  A recent study showed that children who were good at math and science but also had strong verbal skills were less likely to go into the STEM disciplines.  Women were more likely than men to be good at both.

So what comes from all this?  The authors devote the last part of the book to policy proposals, but honestly their plans are ideal, with a lot of individual contact, personalized assessment and customized programs of study.  It would be prohibitively expensive.  It draws to mind a bit of wisdom I once heard about innovation: it’s very easy to be innovative if you pretend money is not an issue. 

The biggest thing I drew from the book was that when it comes to learning, genes really really matter -- they matter about half – and this has to do with innate intelligence(s), confidence, socioeconomic status gender and more. 
In the other half, there is still a ton of wiggle room.  

No comments:

Post a Comment

I've allowed comments without login.