Various readers have been contributing comments and additional papers on the genetics of intelligence. The general theme is that Prof Rose’s Times Educational Supplement article did not present a balanced picture, and left out recent relevant work. The full TES article is available in a link given in a comment by Anonymous 27 January 2014 15:44. The same commenter gives a link to a more up to date, much larger sample study by Handscome et al (2012).
Here are some of the points which have come in by email:
“If twin studies are so flawed why do they yield similar results to adoption studies? Twin and adoption studies are used throughout the life sciences, not just for intelligence; and their results are reasonable throughout.”
“No one thinks any more that genetic effects are limited to the 2% of the genome in coding sequences. Current Genome Wide Association studies have been limited to additive effects of common SNPs — the focus now is on the vast majority of DNA sequence variation which is rare. Genome-wide complex trait analysis uses DNA alone in large samples of unrelated subjects, and by this means estimates substantial heritability and underlines that heritability of complex traits and common disorders (nothing specific to intelligence) is due to many genes of small effect that will be difficult to discover and more difficult to replicate.”
Twin researchers have told me that the numbers of monozygotic to dizygotic twins in the Turkheimer et al. paper are roughly what is always found, so is no problem. The confounding of race and SES is a problem. However, there has been quite a bit of subsequent work, some of it supporting the hypothesis that heritability is greater in higher SES families.
Power is a major concern in Genes x Environment studies. 300 pairs of twins has no power to detect reasonable differences in heritability between groups. The larger more recent studies tend not to replicate the SES finding of the Turkheimer paper.
As regards epigenetics, readers have pointed out that it is a mechanism for gene expression, and is not limited to parent-offspring transmission, which is what I had been concentrating on. Some genes get silenced by methylation. The reason why members of an MZ twin pair differ on measures of methylation is that methylation indexes environmental effects.
Here are some other readings which bring the story up to date:
Five Years of GWAS Discovery. Peter M. Visscher, Matthew A. Brown, Mark I. McCarthy, and Jian Yang. The American Journal of Human Genetics 90, 7–24, January 13, 2012
Childhood intelligence is heritable, highly polygenic and associated with FNBP1L. B Benyamin et al. (2013), Molecular Psychiatry 1–6.