It is just over 50 years ago that I wrote my first article on the topic of intelligence (Eysenck, 1939). I was reviewing Thurstone's famous monograph (Thurstone, 1938) in which he criticized Spearman's (1927) theory of intelligence as a single entity or concept, spreading over all cognitive activities; he postulated instead a number of ‘primary abilities’, separate and uncorrelated. Reanalysing his very extensive data, I found evidence both for a general factor of intelligence, very much as postulated by Spearman, and also for a number of primaries, very much as postulated by Thurstone. Both Spearman (Spearman and Jones, 1950) and Thurstone (Thurstone and Thurstone, 1941) finally agreed that a hierarchical system of description of the cognitive space incorporating both a general factor of intelligence and various group factors or special talents, was most in line with the available facts, and there is now considerable consensus on some such system (Vernon, 1979; Eysenck, 1979; Brody, 1992; Carroll, 1993).

While quite happy with such an account at the descriptive level, I felt that if there was a strong genetic basis for IQ, as there undoubtedly is (Woodworth, 1941; Eysenck, 1979; Plomin, De Fries and McClearn, 1990), there must be some physiological or hormonal intermediaries between DNA and behaviour – it is impossible for DNA to influence behaviour directly.