More on The Forbidden Experiment January 27, 2007Posted by Johan in Behavioural Genetics, Cognition, Developmental Psychology.
While looking around for post-grad opportunities, I somehow stumbled across an article by Rebecca Saxe on feral children. It’s a review of a book by Benzaquen, but only in the loosest sense of the word. One very interesting quote on the forbidden experiment:
But here’s the catch: the forbidden experiment may belong to a smaller group of experimental problems that persistently seem meaningful but are not. Intuitively, we expect that while human nature interacts with human society in a typical child’s development, the natural and the social are in principle independent and distinguishable. If this intuition is wrong, the forbidden experiment is incoherent. In fact, the social and the natural may be irretrievably entangled in development. In part this is because a social environment that includes other human beings is inevitably more natural for a human infant than any wholly artificial environment that could be constructed to replace it. Even the unfolding of innately determined human traits relies on a social environment. For example, virtually every human infant is exposed to a language and learns it; an infant who was never exposed to any language could not possibly speak one. Yet it is the children who do learn a language—through social interactions—who illustrate the natural human capacity.
This is incidentally part of Behavioural Genetics 101 – it is impossible to estimate the relative contributions of genes and environment to a trait in a single individual. The heritability statistic (i.e., the proportion of phenotypic variance that is attributable to genotypic variance) only works on a group level.
It’s quite hard to grok this on a conceptual level: if a trait has a heritability of .9 for a given population at a given time, this does not mean that each individual’s performance was 90 % genes and 10 % how momma treated him (in fact, it gets worse when we consider that environmental variance is typically of the nonshared kind). Nor does it mean that for 9 participants, genes ruled, and for 1 participant, the environment did.
So really, the forbidden experiment would have to use a sample of feral children. Feral twin study, anyone?