interest in the use of quantitative-genetics methods for untangling the effects of nature and nurture in humans was there all along


As an historian, intrigued by Peter’s question: “What happened historically when the methods of quantitative genetics were being transferred from the context of agricultural and laboratory breeding to analysis of human variation which allowed the restrictive conditions that hold in the former context not to be seen as a significant problem”?

Realized that although I’d written both on the history of the nature-nurture debate and the concept of heritability, had never asked when the first heritability estimates of human traits were made or how originally justified. I’d assumed, with Peter, that those methods were developed for the purposes of plant and animal breeding and later transferred to the human realm. When he asked when and how, I set out to try to answer the question.

Can’t answer it fully, or even to large degree, but I did learn something, and what I found was quite surprising. For it turns out that, as in the history of eugenics and statistics, interest in the use of quantitative-genetics methods for untangling the effects of nature and nurture in humans was there all along – or at least, nearly from the time that those methods were first developed. And the key figures in this story are Sewall Wright and especially Stanford psychologist Barbara Stoddard Burks (colleague of Lewis Terman).

In the early ‘20s, Wright developed the statistical method now known as path analysis (in an effort to understand the transmission of color patterns in guinea pigs; “Correlation and Causation,” Journal of Agricultural Research 20 (1921), 557-585). The first use of path analysis was by Burks, who in 1928 published a study that attempted to determine the relative effects of parental intelligence and environment in producing the IQ of children (“The Relative Influence of Nature and Nurture Upon Mental Development: A Comparative Study of Foster Parent-Foster Child Resemblance and True Parent-True Child Resemblance,” The Twenty-Seventh Yearbook of the National Society for the Study of Education 27 (1928), 219-316; see also Daniel J. Denis and Joanna Legerski, “Causal Modeling and the Origins of Path Analysis,” Theory & Science (2006), esp. pp. 8-10).

Her data was then employed by Wright in at least two influential papers published in the early 1930s: “Statistical Methods in Biology,” Journal of the American Statistical Association 26 (March 1931) 155-63 and “The Methods of Path Coefficients,” The Annals of Mathematical Statistics 5 (Sept. 1934) 161-215.

Wright, in turn, profoundly influenced Jay Lush (who apparently coined the term heritability in his 1937 book Animal Breeding Plans; (See William B. Provine, Sewall Wright and Evolutionary Biology (Chicago University Press, 1986, 322-6; Marcus W. Feldman, “Heritability: Some Theoretical Ambiguities,” in Keywords in Evolutionary Biology, EF Keller and LA Lloyd, eds., Harvard University Press,1992, 151-7). In his 1947 paper, “Family Merit and Individual Merit as Bases for Selection, Part II,” American Naturalist 81 (Sept.-Oct. 1947), 362-79), Lush incorporated data from a later (1938) study of Burks comparing mean differences in IQ between occupational classes.

So there is a route of transmission through Lush, but another that I think goes more directly from Burks. In the 1970s, she is cited by Arthur Jensen, Richard Herrnstein and others who argued for the value of studying the heritability of traits in human populations.

Haven’t tried to trace the pathways in any detail, though Google Scholar makes this much easier than it once was. But have learned enough to know that this is not a simple case of transference from an agricultural/laboratory to human context.

An aside on Jay Lush: In his paper, Peter raises some issues concerning the treatment of individuals based on their group membership (p. 13). In that connection, it’s interesting that, in his 1947 paper, Lush asks: when it comes to making choices between one human being and another, how much weight should we give to “family characteristics” (reputation and accomplishment of close relatives, social or professional class, schools attended, race, etc.)? He notes that in the case of humans, we are usually interested in the future performance of the individual rather than the quality of the offspring. Sometimes we have good predictive information (test scores, school grades, letters of recommendation, general reputation) and in such cases, there’s no need to consider group characteristics. But we often lack good evidence, e.g., when the person is too young for their abilities to be appraised. In these cases, we can increase the accuracy of our choices by taking into account the performance of close relatives or “the average characteristics of his group” (p. 372). Two pages on, Lush writes:

“The idea that each individual should be judged on his own characteristics alone, regardless of what or who his relatives or associates may be, has a strong sentimental appeal to our sense of justice and fair play. Yet justice and fair play for the employer whom the foreman will serve, or for the students…or patients whom the doctor will treat” also demand[s] that no relevant predictive information be ignored” (p. 374)].

(Diane Paul)