In the late nineteenth century, Sir Francis Galton collected information on accomplishments, physical traits and occupational status of members of families and began the correlational approach to behaviour genetics. In agreement with the prevailing sentiment of the British establishment, the finding that these characters aggregated in families was used as a biological justification for the social class structure of the time and even for the predominance of the British Empire. Kamin (1974) points out that similar reasoning by American psychologists responsible for the administration of intelligence tests to immigrants to the USA led to Congressional passage of the Immigration Act 1924 which restricted immigration to the USA from southern and eastern Europe. Thus over the past hundred years the issues involved in the relationship between genetics and behaviour have had serious political ramifications.
R.A.Fisher (1918) reconciled particulate Mendelian transmission genetics with the continously varying phenorypes that interested the biometricians. This paper led to the variance-analysis approach to familial data on continuously varying traits, in particular, behavioural characters. The idea is that overall phenotypic variance, P, can be partitioned into contributions due to variation in gene action, G, those due to environmental variation, E, and interactions between genes and environment. Later, animal breeders termed the ratio of G to P the ‘heritability’ of the trait. The sense in which animal breeders used heritability was as an index of amenability to selective breeding. Of course, this is inappropriate in the context of human behaviour, and in this context it unfortunately developed the connotation of an index of biological determination and refractivity to environmental intervention.
The use of heritability in situations where experimental controls are lacking has been criticized by geneticists, who prefer to think in terms of the ‘norm of reaction’. The norm of reaction for a given genotype is the graph of the phenotype against the environment. It emphasizes the dependence of gene action on the particular environment; a genotype that performs better than another in one environment may do worse in a second. An example is the human genetic disease phenylketonuria, PKU, in which sufferers accumulate toxic concentrations of phenylalanine resulting in extreme mental retardation. Under a diet that restricts intake of phenylalanine from birth, normal mental function occurs. The norm of reaction approach informs us that even if the heritability of IQ were 100 per cent, this would say nothing about its potential for environmental manipulation.
Fisher’s approach to the genetics of continuous variation produces expected values for correlations between relatives of all degrees which can then be compared to observed correlations and heritability estimated. In principle the most powerful data of this kind use adoptions and, in particular, identical twins reared apart. Such twins are extremely difficult to find and unfortunately the largest sample, that published by Burt in his analysis of the heritability of IQ, has been shown to be fraudulent. The remaining samples of this kind normally suffer from non-randomness in the adoption procedure. Nevertheless, until about 1970 the estimates obtained using data that included Burt’s produced the widely accepted statistic that genes accounted for about 80 per cent of variation in IQ.
Wright’s (1934) method of path analysis became the predominant one for estimating heritability. In 1974 this method produced an estimate of genetic heritability of 67 per cent. Path-analytic treatments make allowances for assortative mating and for the transmission of environments within families, that is, cultural transmission. Estimates by Rice et al. (1980) and Rao et al. (1982) suggest that genetic and cultural transmission each account for about one-third of the variance in IQ. As with Fisher’s variance analysis, the path-analysis approach is based on linear models of determination and has been criticized for that reason. Among the other criticisms are that most adoptions are not random, that the increase in mean IQ of adoptive children over that of their biological parents is ignored, and that the estimates of genetic and cultural heritability depend on how the environment is defined and its transmission modelled.
It has frequently been claimed that a high heritability of a trait within a group makes it more likely that average differences between groups, for example, races, are genetic. This is false since heritability is strictly a within-group measure strongly dependent on the range of environments in which it is measured.
In studies of the distribution of human behaviours within families, where the database is not as large as that used for IQ and where the trait in question is a clinically defined disorder, the twin method and the method of adoptions have been widely used. In the twin method the fraction of identical (monozygous or MZ) twin pairs in which both members are affected is compared to that in fraternal (dizygous or DZ) twin pairs. A significant margin in favour of the former is taken as evidence of genetic aetiology. Numerous studies of these concordance rates for criminality, neuroses, homosexuality, drinking habits, affective disorders such as manic depression, and schizophrenia have generally shown greater agreement among MZ than DZ twins. In these studies the twins are usually not reared apart and the role of special environmental influences, especially on MZ twins, cannot be discounted. Other problems such as the mode of ascertainment of the proband, heterogeneity in syndrome definition and variation among the concordance rates in different studies also raise doubts about the efficacy of the twin method.
In the adoption method, the incidence of a trait in the adoptive relatives of an affected adoptee is compared with that in the biological relatives. If the latter is higher than the former the inference is usually drawn that there is some genetic aetiology to the disease. Adoption studies of behavioural disorders from among those mentioned above have generally produced higher agreements between biological than between adoptive relatives. Again the interpretation of a genetic basis for the behavioural abnormality must be viewed with circumspection, since truly random adoption is extremely rare and frequently adoption occurs relatively late in childhood. In none of the behavioural disorders mentioned above has any biochemically distinguishable genetic variant been identified, although research directed at the role of variation in properties of catecholamine and indole metabolism still continues.
The evolution of social behaviour has provided something of a puzzle to natural historians since Darwin. This field of study was subsumed under ethology and behavioural ecology, and until the mid-1970s was relatively immune to the developments in evolutionary population genetics by Fisher, Wright (1934) and Haldane (1932; 1955). The year 1975 saw the publication of E.O.Wilson’s book Sociobiology in which he not only stressed that social behaviours were similar across the animal kingdom from termites to humans, but also claimed that these behaviours were genetically determined. Ethology and much of behavioural ecology were then subsumed under a new name, ‘sociobiology’.
J.B.S.Haldane had speculated in 1932 as to how alarm calls in birds might have evolved genetically but concluded that a simple genetic basis for the evolution of self-sacrifice might only apply to the social insects. In 1955 he foreshadowed sociobiology by asking for how many cousins should one’s self-sacrifice be equivalent to that for a brother. These speculations were formalized by Hamilton (1964), who modelled the evolution of a gene, one allele of which conferred altruistic behaviour on its carrier. Altruism here means that the fitness of the altruistic individuals is reduced by their performance of a behaviour which increases the fitness of others in the population. Hamilton arrived at conditions on the degree of relatedness between the donor and recipient of the behaviour that would enable altruistic genotypes to increase in the population. The condition is usually stated in the form βr>γ where β and γ measure the gain in fitness to recipients and loss in fitness to donors, and r is a coefficient of relatedness. Hamilton noted that in the haplodiploid insects like the hymenoptera his measure of relatedness between sisters is higher than for any other relationship. Since the above inequality is then easier to satisfy than in species where both sexes are diploid, this could explain on a simple genetic basis the evolution in the social hymenoptera of the social caste system with sterile workers. This theory of the evolution of the kin-directed behaviours is now called ‘kin selection’.
Hamilton’s theoretical analysis was made using a mathematical approximation that allows the allele frequency of the altruistic variant, the altruistic gene frequency, to play the central role. With this approximation the mathematics gives the impression that it is possible to add fitness contributions from all relatives affected by the altruism to produce the inclusive fitness of the allele. It has been shown that inclusive fitness is an unnecessary concept and that the theory of kin selection can be developed in terms of classical population genetic models with frequency-dependent Darwinian fitness differences. Hamilton’s formulation contains many assumptions, but when these are removed his theory remains qualitatively true: the closer is the degree of the relatedness between the donor and recipient of an individually disadvantageous behaviour (controlled by a single gene), the easier it is for that behaviour to evolve.
In Sociobiology E.O.Wilson extrapolated from Hamilton’s theory to posit that the evolution of social behaviour throughout the animal kingdom including Homo sapiens has followed these rules of kin selection. Of course this position ignores the general criticism that none of the social behaviours discussed have been shown to have a genetic basis and are certainly not under simple genetic control. Although many behavioural ecologists took Wilson’s position in the years immediately following the publication of his book, the difficulty of empirical measurement of relatedness, and fitness gains and losses, as well as technical criticism by population geneticists, have had a moderating effect. Kinship still plays a central role in behavioural ecology, but the explanatory limitations of the simple kin selection theory are now more widely appreciated.
In Homo sapiens the position of sociobiology is to minimize the role of learning and cultural transmission of social behaviours. In particular, the human sociobiologists have taken the position that such phenomena as aggression, incest taboos, sex-differentiated behaviours, sexual preferences, conformity and spite have largely genetic antecedents. There is clear political danger in acceptance of this assertion that such human behaviours have a genetic basis. We have the precedent of the politics of IQ based on erroneous inferences drawn from data of dubious quality. Sociobiology adopts a position of pan selectionism in which the terms adaptive and genetic are interchangeable. Cultural transmission, under which the properties of evolution are obviously different from those under genetic transmission, is ignored. Clearly sociobiology tried to claim too much: ‘Sooner or later, political science, law, economics, psychology, psychiatry and anthropology will all be branches of sociobiology’ (Trivers in Time, 1 August 1977). Fortunately, we are all biologists enough to tell the tail from the dog.
Marcus W.Feldman
Stanford University
References
Fisher, R.A. (1918) ‘The correlation between relatives on the supposition of Mendelian inheritance’, Transactions of the Royal Society 52.
Haldane, J.B.S. (1932) The Causes of Evolution, New York.
——(1955) ‘Population genetics’, New Biology 18.
Hamilton, W.D. (1964) ‘The genetical evolution of social behaviour, I and II’, Journal of Theoretical Biology 7.
Kamin, L. (1974) The Science and Politics of IQ, Hillsdale, NJ.
Rao, D.C., Morton, N.E., Lalouel, J.M. and Lew, R. (1982) ‘Path analysis under generalized assortative mating, II, American IQ’, Genetical Research Cambridge 39.
Rice, J., Cloninger, C.R. and Reich, T. (1980) ‘The analysis of behavioral traits in the presence of cultural transmission and assortative mating: application to IQ and SES’, Behavior Genetics 10.
Wilson, E.O. (1975) Sociobiology, Cambridge, MA.
Wright, S. (1934), ‘The method of path coefficients’, Annual of Mathematical Statistics 5.
Further reading
Cavalli-Sforza, L.L. and Feldman, M.W. (1978) ‘Darwinian selection and altruism’, Theoretical Population Biology 14.
——(1981) Cultural Transmission and Evolution, Princeton, NJ.
