I finally got around to reading Plomin et al's classic 1997 parent-offspring adoption study. Background: Way back in 1975, Plomin and co-authors launched the Colorado Adoption Project (CAP). They put together a sample of 245 adoptees, their biological mothers, their adoptive parents, plus a control sample of 245 non-adoptive children and their parents. Then they followed the children through life, administering various tests along the way.
If the only result from this study had been the "IQ is heritable," it would have been just another study. But its special methodology - studying adoptee's development from birth to adulthood - confirmed a shocking finding: As children grow up, the heritability of IQ rises, and the influence of family environment on IQ literally vanishes. Here are the custodial parent-child IQ correlations, by age, for adoptees and the biological control group:
We naturally think about the effects of family as cumulative: The longer you're in a family, the deeper the impression. At least for IQ, though, this "natural" thought turns out to be wrong. Family affects the very young, then fades out.
In hindsight, should this pattern really have been so surprising? Yes and no. Consider the parallel case of church attendance.
For a young child, family has near-absolute control over church attendance (unless you're Damien in The Omen, of course!): If your parents go, so do you; if they don't, you don't. As you get older, though, you gain some independence - and with it, a chance to show your true colors. By the time you're an adult, you only go to church if you want to. So it's not surprising that family matters less over time.
Even so, though, you would expect attending church to be at least somewhat habit-forming. Adults only go to church if they want to, but what they want has something to do with what they've experienced. The surprising thing about family influence on IQ is that the effect actually goes to zero. As you grow up, you find your own cognitive level, and the cognitive level you're looking for has nothing to do with the cognitive level you grew up with.