In response to:
How to Inherit IQ: An Exchange from the March 15, 2007 issue
To the Editors:
Just a dash of neuroscience into the exchange concerning heritability of IQ [“How to Inherit IQ,” NYR, March 15]:
The most important period for the wiring of the human nervous system occurs during fetal development. During this period, the fetal environment is a critical variable. Since identical twins usually have a more similar local fetal environment than fraternal twins, to the degree that fetal environment is important, future behavioral traits are expected to be more highly correlated between identical twins than between fraternal twins. How large is the fetal environment component in intelligence? We don’t know. There have been no studies of the heritability of human intelligence that have
controls for fetal environment, so that so far effects of fetal environment variables cannot be separated from genetic variables. We do need to recognize that the nine months of fetal development occur in an “environment”—an environment more critical for the nervous system than any later environment. Without considering fetal environments, the controversy about the inheritance of intelligence is really vacuous.
Emeritus Associate Professor of Molecular Genetics and Cell Biology
University of Chicago
Frank J. Sulloway replies:
As Professor Agin reminds us, the ways in which fetal environments contribute to human development are important and insufficiently understood. He is mistaken, however, in his assertion that previous research on human intelligence has not considered the question. In fact, scores of studies over the last three decades have dealt with this issue, either by analyzing the differences in fetal development between twins that share the same placenta and those that do not (“chorion control studies”) or by studying post-conception differences in identical twins (“co-twin control studies”).1
Derived from the Greek word for “skin,” the chorion is the outermost membrane that envelopes the growing embryo and its placenta, the organ from which the fetus draws its blood supply. Twins differ in whether they share the same chorion and placenta. All fraternal (or dizygotic) twins, and approximately one third of identical (or monozygotic) twins, have separate chorions and placentas. As a consequence, these dichorionic twins do not exchange fetal blood during development. By contrast, approximately two thirds of identical twins develop within the same chorion, causing them to share the same placental blood supply. Such single-chorion twins are also exposed to the same blood-borne neural growth factors and hormones, because the placenta itself is an endocrine gland, producing estrogen, progesterone, and gonadotrophin.
Single-chorion twins, who share 100 percent of their genes, differ in another important way from fraternal twins and from identical twins who have separate chorions and placentas. Since single-chorion twins compete for the same placental blood supply, one twin may receive significantly more nourishment than the other. Approximately 15–20 percent of identical twins suffer from twin-transfusion syndrome, in which one twin loses some of his or her blood to the other twin, causing discrepancies in birth weight and other physical attributes, and in some cases leading to fetal death.
The disparate fetal environments of twins who share a chorion and those who do not are associated with two different hypotheses about fetal development. Because monochorionic twins always share the same blood supply, some researchers have argued that they should be more similar than dichorionic twins, thus inflating estimates of genetic influences on subsequent development.2 Other researchers have argued that identical twins who share the same chorion and placenta ought to be more different than otherwise would be the case because they are competing for their common blood supply.3
Fortunately, more than two dozen studies of identical twins allow us to arbitrate between these two alternative hypotheses. About a dozen of these studies have addressed the relationship between chorionicity and cognitive abilities, including IQ scores. Thus far, the results reveal a reasonably consistent picture when appropriate statistical corrections are made for multiple testing. For example, the largest among these studies (135 twin pairs) showed no difference in IQ by chorionicity, and the collective results from the more than seven hundred twin pairs studied to date suggest that any differences in overall IQ linked to chorionicity are so small as to be difficult to detect.4 The possibility still remains that modest differences exist on certain subtests of particular cognitive and linguistic abilities—differences that appear to cancel themselves out in overall measures of IQ.
In contrast to these collective findings for intellectual ability, significant differences according to chorionicity have been documented for a variety of personality and physical traits. For example, monochorionic identical twins have been found to be more similar than dichorionic identical twins in measures of Type A behavior (hostility) and for various aspects of temperament (irritability, impulsivity, and hyperactivity).5 At the same time, monochorionic twins tend to show greater differences than dichorionic twins in birth weight and in some physical attributes—differences that persist into adulthood.6
What this research suggests is that sharing the same chorion and placenta causes identical twins to be more similar in some traits, but more different in others, depending on how the attributes in question are influenced by specific aspects of gestation. Shared hormones, for instance, appear to increase resemblances in personality, whereas a shared blood supply appears to create disparities in fetal growth and various correlated traits. With regard to overall intelligence and most other cognitive measures, such contrasting influences appear to counterbalance one another, resulting in few demonstrable differences between monochorionic and dichorionic twins in IQ (and hence in the calculation of the extent to which IQ is inherited).
A large and impressive meta-analysis of 212 previous twin studies has nevertheless reached a different conclusion. Conducted by Bernard Devlin, Michael Daniels, and Kathryn Roeder, this study sought to determine the contribution of maternal (prenatal) environments to IQ by comparing the IQ scores of identical twins, fraternal twins, siblings, and parents and offspring.7 Based on statistical models—which posited that differences in IQ relate to disparities in shared genes as well as in fetal environments—the authors attributed 20 percent of the similarities between twins, and 5 percent between siblings, to shared fetal environments. Unfortunately, these researchers did not attempt to distinguish monochorionic twins from dichorionic twins. Moreover, even with their substantial estimate for the influence of prenatal environments on cognitive development, the meta-analytic models still allotted a substantial 48 percent of the variance in IQ to heritability.
There is nevertheless strong evidence that the approach used by Devlin, Daniel, and Roeder underestimates the heritability of IQ. Heritability can be calculated from data on twins in direct and indirect ways.8 Direct measures are obtained by comparing the IQs of identical twins raised apart, which typically produces correlations in IQ of .70 or higher. By contrast, indirect measures are obtained by comparing test results for identical twins raised together and fraternal twins raised together, and typically yield heritability estimates of about .50. It is this 20-percent discrepancy between the direct and indirect estimates that the meta-analytic models of Devlin et al. exploited, since the models attributed the difference to the fact that fraternal twins do not share the same chorions and placentas, whereas identical twins do so (at least about two thirds of the time).
Nevertheless, a substantial portion of the 20-percent discrepancy may have a very different explanation than the one Devlin et al. have proposed. Most studies of twins reared together are conducted on children and adolescents, whereas most studies of twins reared apart are conducted on subjects who have been reunited and tested only in adulthood. Measures of the heritability of intelligence tend to increase with the age of subjects, apparently because the influence of home environments during childhood is displaced, over time, by the cumulative consequences of genetic potential.9 Given the systematic increase in heritability estimates with age, it seems likely that some of the 20-percent variance that Devlin et al. attributed to shared prenatal environments owes itself, more correctly, to the different ages of twins being compared in their study. Although Devlin et al. did address this important issue, their models did not attempt to include age as a covariate along with prenatal environments. The inclusion of such a covariate would presumably have reduced—perhaps by as much as half—their estimates for the influence of prenatal environments, thereby increasing their estimates for the heritability of IQ.
In conclusion, while Professor Agin is absolutely correct to assert that fetal environments are potentially important for understanding heritability, his claim that “the controversy about the inheritance of intelligence is really vacuous” fails to take into account the results of the many twin studies on the subject. This accumulated evidence strongly suggests that the heritability of IQ is somewhere between a low estimate of about .48, provided by Devlin et al.’s impressive but nevertheless still inconclusive meta-analysis, and the more usual estimates of .60–.70 derived from classic twin methods. It is testimony to the remarkable progress of research that the question of prenatal influences on IQ can be narrowed down to a debate over 20 percent, and probably much less, of the overall variance.
J.A. Phelps, J.O. Davis, and K.M. Schwartz, “Nature, Nurture, and Twin Research Strategies,” Current Directions in Psychological Science, Vol. 6, No. 5 (October 1997), pp. 117–121. ↩
D.I.W. Phillips, “Twin Studies in Medical Research: Can They Tell Us Whether Diseases Are Genetically Determined?” The Lancet, Vol. 341 (April 17, 1993), pp. 1008– 1009; K. Stromswold, “Why Aren’t Identical Twins Linguistically Identical? Genetic, Prenatal and Postnatal Factors,” Cognition, Vol. 101, No. 2 (September 2006), pp. 333–384. ↩
T.J. Bouchard Jr., R.D. Arvey, L.M. Keller, and N.L. Segal, “Genetic Influences on Job Satisfaction: A Reply to Cropanzano and James,” Journal of Applied Psychology, Vol. 77, No. 1 (February 1992), pp. 89– 93; N.L. Segal, Entwined Lives: Twins and What They Tell Us about Human Behavior (Dutton, 1999), p. 319. ↩
See, for example, N. Jacobs, S. Van Gestel, C. Derom, E. Thiery, P. Vernon, R. Derom, and R. Vlietinck, “Heritability Estimates of Intelligence in Twins: Effect of Chorion Type,” Behavior Genetics, Vol. 31, No. 2 (March 2001), pp. 209–217; T. Reed, D. Carmelli, and R.H. Rosenman, “Effects of Placentation in Selected Type A Behaviors in Adult Males in the National Heart, Lung, and Blood Institute (NHLBI) Twin Study,” Behavior Genetics, Vol. 21, No. 1 (January 1991), pp. 9–19; and D.K. Sokol, C.A. Moore, R.J. Rose, C. J. Williams, T. Reed, and J.C. Christian, “Intrapair Differences in Personality and Cognitive Ability among Young Monozygotic Twins Distinguished by Chorion Type,” Behavior Genetics, Vol. 25, No. 5 (September 1995), pp. 457–466. ↩
Sokol et al., “Intrapair Differences,” p. 458. ↩
J.P. Race, G.C. Townsend, and T.E. Hughes, “Chorion Type, Birthweight Discordance and Tooth-Size Variability in Australian Monozygotic Twins,” Twin Research and Human Genetics, Vol. 9, No. 2 (April 2006), pp. 285–291. ↩
B. Devlin, M. Daniels, and K. Roeder, “The Heritability of IQ,” Nature, Vol. 388 (July 31, 1997), pp. 468–471. ↩
R. Plomin and J.C. Loehlin, “Direct and Indirect IQ Heritability Estimates: A Puzzle,” Behavior Genetics, Vol. 19, No. 3 (May 1989), pp. 331–342. ↩
K. McCartney, M.J. Harris, and F. Bernieri, “Growing Up and Growing Apart: A Developmental Meta-analysis of Twin Studies,” Psychological Bulletin, Vol. 107, No. 2 (March 1990), pp. 226–237. ↩