news and views

the possibility of mental processes that are not strictly reducible to brain processes. Symmetry as In between are the colonizers from cogni- tive psychology, who betray their insecurity by adopting the term ‘cognitive scientists’. destiny — taking a They propose to explain away — rather than to explain — the brute facts of variabili- ty in cognitive capacity. They supply the

algorithms for solving problems which keep balanced view of IQ the sky pilots on their toes. For about 825 years they have been promising to account for variability in IQ in terms of cognitive- Steve Blinkhorn processing models. (The first positive results Genetic vulnerability during development may be responsible for are still eagerly awaited.) environmental effects on . Taken with a study showing that Results, processes, substratum — the the of IQ does not change with age, could these results mean three foci of interest that divide those who that nature and nurture are no longer in opposition? study . But somewhere in there is a mind–brain barrier, a conceptual he study of human intelligence has of should become proportionately and philosophical divide that stands in the always had its feet in biology and its less. But if the Swedish study is to be believed, way of progress. The artificial-intelligence Thead in the clouds — the recent environmental influences on IQ have their and psychometric researchers may set prob- encounter between Gary Kasparov and effects early. This is hardly controversial lems and recognize the results as showing the Deeper Blue showed what can be achieved in stuff and, indeed, potential factors (such as external hallmarks of intelligence, but they confronting human intelligence with meth- atmospheric lead, maternal nutrition and have little to say about how to make people ods that have no plausible claim to mimic its fetal alcohol syndrome), which indicate a more intelligent, or overcome handicaps to cognitive processes. Two studies at the other biological rather than a strictly psychological achieve their potential. Cognitive scientists end of the spectrum now serve as a reminder influence, will always focus on the earliest have shown how styles of problem that were that intelligence operates on a substrate of stages of life. once to require poorly understood wetware, not silicon. Unfortunately, having relatively stable human capacities can be reduced to com- In the 6 June issue of Science, McClearn estimates of the heritability of IQ advances putational algorithms. Yet they have contri- et al.1 described a study of octogenarian our rather little. Because we have buted essentially nothing to our knowledge Swedish , in which they found that the no sensible systems by which to measure of how some people, when presented with a heritability of psychometric intelligence either the variability of environments or the mass of confusing data, can clearly see the remains much the same in old age as in variability of genetic influences, we are left underlying structure of a new problem, youth. And in this week’s Proceedings of the with a ratio of two unknowns. The only whereas others cannot. And the biological Royal Society, Furlow et al.2 show that there is conclusions that are left to be drawn are determinists can point to over 100 years of a modest, but replicated, association socio-political — heritability of IQ is so consistent findings, but not a lot of hope on between IQ and asymmetry in bilateral great (or small) that educational or social offer. human physical traits. These observations programmes will have insignificant (or This is where the paper by Furlow et al.2 lead to nicely paradoxical conclusions, relat- enormous) effects. seems to break new ground. They have stud- ing environmental influences to the genetic The problem is, opinion is split three ied fluctuating asymmetry — an asymmetry basis of differences in IQ. ways as to how to handle the raw observation of usually symmetrical bilateral traits, which Research into heritability and the bio- that observed intellectual capacity varies occurs as a result of biological stress during logical correlates of measured intelligence widely. Those with their heads in the clouds development. Fluctuating asymmetry is tends, unfortunately, to descend into a kind — artificial-intelligence researchers and assessed by an index of variability in the of cognitive Calvinism. Heritability esti- constructors of psychometric tests (sky dimensions of feet, fingers, wrists, elbows mates take on the mantle of constants of pilots for short) — mark their progress by and ears. The authors claim that fluctuating nature, social pessimism holds sway, and all results. Intelligence is about handling diffi- asymmetry correlates between 0.20 and 0.25 concerned resign themselves to predestina- cult tasks successfully, and doing well when with psychometric intelligence. Moreover, tion due to unknown genes. Because there you started off not knowing what to do. this is probably an underestimate — values has long been evidence for the heritability These sky pilots aim to keep at least two steps of 0.30 to 0.35 might be expected in experi- of IQ, the idea took hold that there must, ahead of those who find algorithmic solu- ments using a better IQ test criterion and therefore, be genes for intelligence — just as tions to cognitive tasks, and they provide subjects from the full range of IQ, rather than theories now circulating suggest that there measures which act as criterion variables in the undergraduates from the University of may be genes for homosexuality. Of course, studies of heritability and the biological New Mexico who were studied by Furlow ideas about the heritability of intelligence correlates of intelligence. and colleagues. pre-date the notion of a gene, and they go The second group — the biological deter- What form of stress might cause fluctuat- back certainly as far as . minists or chiropodists (podiatrists if you ing asymmetry, and in what sort of sub- McClearn et al.1 have found that esti- must) — have their feet firmly planted in optimal development might it manifest mates of the heritability of IQ amongst octo- biology. They, at worst, suppose that the itself? Furlow et al. cover all possibilities, genarians are of much the same magnitude most we can do is to observe biological deter- including effects on any level of structure in as estimates from younger age-groups (of the minism, and then look for simple models to the central nervous system. This is anything order of 0.60). This result is surprising only explain why variation in intelligence reflects but satisfactory from a theoretical point of because common sense suggests that, with biological variation. Such models include view, but quite in keeping with their report, the passage of time, the effects of experience differences in neural efficiency, speed of which is essentially a data paper. will become increasingly important in deter- conduction, or error rates. But whatever The authors propose that people are mining cognitive abilities. So, the influence you do, don’t confuse them by pointing out differentially susceptible to environmental

NATURE | VOL 387 | 26 JUNE 1997 849 Nature © Macmillan Publishers Ltd 1998 news and views influences on their IQ, because of their moment correlation coefficient. They have a of the Universe is a powerful clue to the genetic endowment. They believe that these couple of tentative theoretical accounts as timescales over which galaxies form. The susceptibilities are likely to have their effects to why they should have got the results they most remarkable aspect of Fig. 2 is the rapid early in life (for instance, prenatally). That is, report. These include structural develop- decline at high redshift: this indicates that they suppose that there is a real, common, mental imperfections, leading to less current surveys, by probing out to redshifts causal link between bodily asymmetry and efficient neural processing, and a differential near 5, are now seeing almost all of the galac- lowered IQ. Indeed, they are prepared to esti- use of energy budgets in people who have tic history of the Universe. The epoch of peak mate that anything between 17 and 50 per suffered varying degrees of developmental star formation is around z =1.5. cent of the variability in IQ is attributable to stress. But neither of these is more than an The curve supports a hierarchical picture such causes. If the upper estimate proves to indicative hypothesis for the time being. of galaxy formation. Hierarchical mode8ls be accurate and replicable, then fluctuating Expect many attempts at replication — “treat galaxy formation as a process, not an asymmetry could account for almost all low-life demonstrations that blacks/half- event’’ (S. White, Max Planck Inst. Astro- heritable sources of variability in IQ. But it breeds/less-favoured races suffer more from phys.): the idea is that small, amorphous is hard to believe that such a long-standing fluctuating asymmetry than white, middle- proto-galaxies form first, eventually settling conundrum as the relative contributions of class golden youth. Also expect, in general, a into disk galaxies such as spirals, which can nature and nurture to IQ could be turned failure to recognize that Furlow and his then also merge to form ellipticals. Rival into no more than a neat paradox — colleagues may just have glimpsed a way of theories suggest that giant galaxies form via nurture’s influence depends on genetic reconciling the long-standing antagonism the collapse of a single massive gas cloud at predisposition — in one go. between chiropodists and sky pilots. high redshift, with the nature of this initial The really neat thing is that Furlow et al. Steve Blinkhorn is at Psychometric Research and collapse, rather than subsequent galaxy have used a method that is potentially within Development Limited, Brewmaster House, The mergers, determining the form of the the reach of any psychology undergraduate Maltings, St Albans, Hertfordshire AL1 3HT, UK. galaxy. — a method, in fact, that was almost within 1. McClearn, G. E. et al. Science 276, 1560–1563 (1997). Hierarchical models predict star-forma- the reach of Francis Galton — requiring 2. Furlow, F. B., Armijo-Prewitt, T., Gangestad, S. W. & Thornhill, tion histories that agree at least qualitatively no more than callipers and the product– R. Proc. R. Soc. Lond. B 264, 823–830 (1997). with those in Fig. 2. There are a few theoreti- cal embarrassments which refuse to dis- Galaxy evolution appear from the simulations — such as egregiously small galactic disks with low The end of the beginnings angular momenta (G. Efstathiou, Univ. Oxford; M. Steinmetz, Steward Obs.) — but Roberto Abraham even so there is considerable confidence in these models, and so much of the theoretical

n eerie feeling has overtaken recent focus is now on the details of the different SCI gatherings of observational and theo- galaxy types6–8 that contribute light at differ- ST Aretical cosmologists* — a sense of ent redshifts. For example, is the light at the impending (if not quite actual) consensus. highest redshifts coming from proto-ellipti- A broad outline of the processes by which cal galaxies (M. Giavalisco, Carnegie Obs.), galaxies form and evolve, in rough agree- or do ellipticals form continuously over a ment with the predictions of theory, seems to range of redshifts (G. Kauffmann, Max have emerged from a number of recent Planck Inst. Astrophys.)? galaxy surveys1–4. But there is one unresolved question We classify galaxies by their distribution about the overall star-formation rate. Some of stars, so galaxy evolution can be quantified of it must be missing from Fig. 2 because by changes in star-formation rates. These of dust contamination. Dust scatters and can be measured by looking back to galaxies absorbs light most strongly at short wave- with increasing redshifts (thus further from lengths, and, because of the expansion of the us, and earlier in the Universe; Fig. 1), at dif- Universe and the corresponding redshift, the ferent wavelengths, to see how the average more distant the galaxy, the shorter was the brightness and spectrum change. In the past Figure 1 Back in time — the Hubble Deep Field, original wavelength of the light we see. So it two years the accessible redshift range has of which this is a view, has helped to revolutionize may be that distant, actively star-forming but expanded up to z < 4. This is remarkable understanding of the Universe at high redshift. very dusty ‘starburst’ galaxies contribute progress: until recently, studies of ‘normal’- much of the total brightness of the Universe galaxy populations were restricted to red- probed by galaxy surveys. That is compara- but are nonetheless missing from optical shifts z < 1. However, the discovery4 that the ble to the fraction sampled by the relatively surveys. Sensitive infrared surveys will be so-called ‘Lyman dropout’ galaxies (which rare quasars and radio galaxies. able to detect such systems, if they exist. appear anomalously faint through blue fil- The current observational picture is The question is: how much light are we ters because a prominent ultraviolet spectral summarized by the star-formation history missing? Nearby starburst galaxies appear to break has been redshifted into optical wave- curve (Fig. 2) (P. Madau, Space Telescope Sci. have spectra that curve upwards from the lengths) are at redshifts greater than two, has Inst.). By adding together the luminosity ultraviolet to the infrared, and if dust- increased enormously the number of known contributed by all the galaxies in a series of enshrouded starbursts in the distant Uni- high-redshift galaxies. Around 80–90% of redshift intervals, this curve maps out a pre- verse (at z >1) have similar properties, they the total age of the Universe is now being cipitous rise and fall in the production of could contribute more than ten times as stars and ‘metals’ (elements heavier than much energy as is accounted for in the star- *The Hubble Deep Field: Space Telescope Science Institute May helium) in a given volume of the Universe as formation history curve (G. Meurer and T. Symposium, Johns Hopkins University, Baltimore, USA, 6–9 May a function of redshift5. Heckman, Johns Hopkins). That agrees with 1997; and The Ultraviolet Universe at Low and High Redshift: Probing the Progress of Galaxy Formation, College Park, Maryland, USA, 2–4 The volume-averaged star-formation a preliminary analysis of images obtained May 1997. rate over such a large fraction of the total age by the Infrared Space Observatory satellite,

850 NATURE | VOL 387 | 26 JUNE 1997 Nature © Macmillan Publishers Ltd 1998