news and views

aged adults, the discrepancy had been to reflect age-related increases in The democracy of the of IQ. This hypothesis is supported by research on octogenarian published earlier this year6, as well as by the genes studies of other familial pairings7. Devlin et al. offer, and provide statistical support for, an alternative account of this Matt McGue anomaly. Specifically, they hypothesize that The genetic heritability of IQ remains highly contentious. A new analysis failure to consider the shared prenatal envi- shows that genetic influences may be weaker, and prenatal ronment of twins results in an overestima- environmental influences greater, than previously appreciated. tion of IQ heritability in studies of twins reared apart, but would not bias heritability ir ’s contributions in distinguish additive gene effects, which are estimates in studies of twins reared together, 1869 alone were enough to ensure that independent of genetic background and thus where shared prenatal effects are cancelled Shis scientific legacy would be a lasting shared by parents and their offspring, from out in the comparison of monozygotic one. In that year, he not only helped in non-additive gene effects, which are depen- and dizygotic similarity. When shared founding this journal, but also published the dent on genetic background and thus not prenatal environments were included in the first empirical investigation of the inheri- shared by parents and offspring. Narrow- statistical model that Devlin et al. fitted to tance of human achievementl, a study that is sense heritability is the proportion of a the IQ correlations, they accounted for 20% generally credited with framing the modern trait’s variance that is attributable to additive of IQ similarity among twins but only 5% nature–nurture debate. On page 468 of this gene effects; broad-sense heritability is the among non-twin siblings. The former esti- issue2, Devlin, Daniels and Roeder report a proportion attributable to both additive mate is particularly remarkable given that statistical analysis of more than 200 familial and non-additive gene effects. twins, and especially monozygotic twins, can IQ correlations, the most recent in a long Herrnstein and Murray4 based their argu- experience radically different intrauterine line of empirical investigations aimed at ments on an IQ heritability of 60%; Devlin et environments even though they share the resolving issues that were raised by Galton. al. report a broad-sense heritability estimate womb at the same time8. Because it is associated with a wide range of 48% and a narrow-sense heritability of Devlin and colleagues’ report supports of social effects, including educational and only 34%. Although a narrow-sense heri- the view that the main environmental influ- occupational success, and even tability of this magnitude does not preclude ences on IQ occur early in life. If indeed they delinquency3, IQ has long been of interest to dysgenic trends (albeit at a slower rate than do, improved cognitive functioning might behavioural and social scientists. But these Herrnstein and Murray projected), it is cer- be an unexpected benefit of public health associations have also challenged cherished tainly too low to support the establishment of initiatives aimed at improving maternal beliefs about the nature of social achieve- a high-IQ caste. Over the long term, unless nutrition and reducing prenatal exposure to ment. If social status is influenced by IQ, both the narrow-sense heritability and the toxins. Nonetheless, it is important to recog- which is in turn substantially inherited, then rate of intermarriage is very high, genes for nize that the evidence of Devlin et al., like social standing will, in part, be a function of quantitative traits such as IQ are essentially that supporting the influence of early intel- one’s genetic endowment. Hard work may be democratic; by the third or fourth genera- lectual stimulation on synapse formation no guarantee of success, unless one has also tion, descendants of gifted individuals are and subsequent human intellectual perfor- received a lucky draw in the genetic lottery. not much more likely to be gifted than are mance, is indirect and as yet unreplicated. The general uneasiness over IQ reached descendants of ordinary people. Caution is certainly warranted. In large- new heights with the publication of The Bell Although their paper is entitled “The scale studies where pre- or perinatal influ- Curve4, by Herrnstein and Murray, in 1994. heritability of IQ”, Devlin and colleagues’ ences on IQ have been assessed directly, little The authors described declines in the popu- most important finding probably concerns evidence for any strong effect has been lation’s genetic potential for high IQ (that is, the nurture rather than nature of IQ. Behav- found9. Perhaps this early work lacked dysgenic trends), because of higher fecun- ioural geneticists had previously pointed out precise assessments of prenatal exposures, or dity among the poorly educated relative to an inconsistency in the familial IQ correla- perhaps the effect of any single prenatal the well educated; and they also forecast the tions. From data on twins that have been factor is relatively minor and thus difficult to establishment of a caste-like cognitive élite, reared together, the heritability of IQ is detect. Even though it does not establish the which would be maintained through inter- estimated to be about 50%; but from twins effect of any specific prenatal factor, Devlin marriage and the (genetic) transmission of reared apart, it is estimated at about 70% (ref. and colleagues’ statistical analysis implies high IQs from parents to offspring. In short, 5). Because IQ correlations for twins reared that a relatively sizable portion of IQ vari- they argued that Western (and in particular together are based primarily on adolescents ability can be attributed to the aggregate US) society was becoming a two-class or children, whereas correlations for twins effect of the prenatal environment, and so system, where the cognitively limited masses reared apart are based primarily on middle- provides a rationale for reconsidering these would be ruled by a relatively small and earlier studies. reproductively isolated cognitive élite. That the IQ debate now centres on Devlin and colleagues’ findings will lead whether IQ is 50% or 70% heritable is a to a reconsideration of these most dire con- That the debate now remarkable indication of how the nature– clusions from . The likelihood nurture debate has shifted over the past two of dysgenic trends occurring, as well as the centres on whether IQ is decades. The anti-hereditarian position that ability of parents genetically to reproduce 50% or 70% heritable is a there are no genetic influences on IQ has desired traits in their children, both depend crumbled for want of any empirical data that on the strength of parent–offspring trans- remarkable indication of would support such a radical view. Equally mission, which in turn depends on the how the nature–nurture remarkable is the increasingly dominant narrow-sense, rather than the broad-sense view that the major environmental influ- (total) heritability. Quantitative geneticists question has shifted ences on IQ occur within the first few years of

NATURE VOL 388 31 JULY 1997 | | Nature © Macmillan Publishers Ltd 1997 417 news and views life, or in the womb, and directly affect the 1. Galton, F. Hereditary Genius: An Inquiry into its Laws and bonate production peaked in subpolar and development of the brain. Research on the Consequences (Macmillan, London, 1869). subtropical oceans (R. Schneider, Univ. 2. Devlin, B., Daniels, M. & Roeder, K. Nature 388, 468–471 nature and nurture of IQ is converging on (1997). Bremen), reflecting a temperature-related the view that human intellectual ability has 3. Neisser, U. et al. Am. Psychol. 51, 77–101 (1996). shift in plankton ecology from diatoms to a strong, but malleable, biological basis — a 4. Herrnstein, R. J. & Murray, C. The Bell Curve: and calcareous plankton. Class Structure in American Life (Free Press, New York, 1994). convergence that Galton would, no doubt, 5. Plomin, R. & Loehlin, J. C. Behav. Genet. 19, 331–342 (1989). In shallow-water environments, the have found quite congenial. 6. McClearn, G. E. et al. Science 276, 1560–1563 (1997). evolution of several major reef systems (such Matt McGue is in the Department of Psychology 7. McGue, M., Bouchard, T. J. Jr, Iacono, W. G. & Lykken, D. T. in Nature, Nurture and Psychology (eds Plomin, R. & McClearn, as the Great Barrier Reef) accompanied and the Institute of Human , 75 East River G. E.) 59–76 (Am. Psychol. Assoc., Washington DC, 1993). increased reef carbonate production (A. Road, University of Minnesota, Minneapolis, 8. Price, B. Am. J. Hum. Genet. 2, 293–352 (1950). Droxler, Rice Univ.; P. Davies, Sydney Univ.). 9. Broman, S. H., Nichols, P. L. & Kennedy, W. A. Preschool IQ: Minnesota 55455, USA. Prenatal and Early Developmental Correlates (Erlbaum, The production of carbonate in continental e-mail: [email protected] Hillsdale, NJ, 1975). shelves and in mid-latitude open-ocean environments may partly explain the Palaeoclimatology extreme dissolution of carbonate sediments during MIS 11 throughout ocean basins A warm future in the past such as the Indian and Pacific3. (In the oceans’ carbonate cycle, biological forma- William R. Howard tion of hard body parts from carbonate out- strips river input of the component ions, and o understand the impact of a possibly followed by early returns to glacial conditions dissolution from the seabed makes up the warmer future climate, geologists are (Fig. 1). difference. In other words, to produce more Tsearching the past for warmer-than- Many other aspects of the climate were carbonate in one place you must dissolve present interglacial intervals — examples of strange. In deep-sea records of MIS 11, sub- more elsewhere.) what we may expect in a world with more polar waters in both hemispheres were at Perhaps barrier-reef tracts were estab- greenhouse gases than ours1. A common tac- their warmest (L. Burckle, Lamont-Doherty lished during stage 11 because tropical con- tic is to look at the last interglacial (around Earth Obs.), following one of the coldest tinental shelves were flooded (A. Droxler; 120,000 years ago), but a period 423,000 to glacial stages of the past 500,000 years (Fig. C. Kievman, Kean Coll.). How high did the 362,000 years ago may fit the bill better, 1c, d). Deep-ocean carbon isotope data show seas rise? Beach deposits in Alaska, Bermuda because the Earth’s orbital geometry has a maximum in the production of North and the Bahamas (P. Hearty, Bahamas) and been similar during the Holocene (the pre- Atlantic Deep Water (D. Hodell, Univ. uplifted reef terraces in Indonesia4 seem to sent interglacial period) to what it was then. Florida). Tracers of downward particle flux imply that the ocean stood as much as twenty This period is known to palaeoclimatologists in the equatorial Pacific show that biological metres above its present level. But this is as stage 11 or ‘MIS 11’, according to a productivity was at a minimum in MIS 11 uncertain: estimating sea level for MIS 11 is numbering system for glacial advances and (R. Murray, Boston Univ.). Calcium car- complicated by the fact that dating beaches retreats marked in the marine oxygen iso- Figure 1 Climate and tope record (Fig. 1). At a recent symposium*, Oxygen isotope stages evidence emerged of extreme climatic varia- 13 56 7 8 9 10 11 12 chemistry of the past tion and peculiar interplays between ocean 500,000 years. All the 550 a temperature, thermohaline circulation, 530 climatic variables b to f N –2 plankton ecology, sea level and reef growth 510 ° (refs 2, 8–11, 490 W m respectively) are related during MIS 11, all of which may provide 470 450 June 65 insight into the response of the natural solar radiation to solar forcing, a (ref. carbon cycle to future climate change. b 12), determined by

O 1

Changes in the Earth’s orbital geometry 18 changes in the Earth’s Ȏ explain the strong 41,000-year and 23,000- 0 orbit (c, d and e relative -1 year rhythms in climate over the past five Ice volume to average over the Marine million years (due to changes in axial tilt 1 Holocene — the past (standard deviations) (standard c and orbital precession, respectively). During 0 10,000 years or so). The -1 C the past 600,000 years, however, the domi- ° -2 long, hot interglacial of nant rhythm has been a 100,000-year cycle. -3 stage 11 occurred under -4 Although this climate cycle is at a similar Subpolar Southern Ocean similar orbital 5 d frequency to orbital eccentricity variation, the conditions to those we 0

radiative impact of that orbital cycle is too C have now — so is it a 2 ° -5 small — at least in linear climate models — to Sea surface temperature model for our own time? -10 produce the large swings in the climate record. Subpolar North Atlantic Relative to Holocene Relative e

MIS 11 is the warm peak of one such ‘100k’ C 0 13 cycle, with two intriguing features. First, Ȏ climate change at the 12–11 transition is near- (‰)

-1 production deep water Benthic ly the most severe of the past half-million South Atlantic North Atlantic ) years (Fig. 1), yet it comes at a time of low f 3

–1 2 orbital eccentricity, thus of low precessional kyr amplitude. Second, interglacial conditions –2 1 last longer. Subsequent interglacials (stages 9, g cm Subpolar Southern Ocean

0 accumulation rate Carbonate (CaCO 7, and 5) show short periods of warmth 0 100 200 300 400 500 *Carbonate Marine System During Oxygen Isotope Stage 11, Age (thousands of years ago) American Geophysical Union Spring Meeting, Baltimore, Maryland, USA, 27–30 May 1997.

NATURE VOL 388 31 JULY 1997 418 Nature © Macmillan Publishers Ltd 1997 | |