Genetics is a Process

“Life is a becoming, not a being,” my embryology professor said the first day of class. “Anyone who pays any attention to fetal development can see that.”


For all the Zen and the Art of Motorcycle Maintenance mysticism of the remark, it’s true. As Heraclitus, the pre-Socratic philosopher, is said to have said:

“Everything changes and nothing remains still … and … you cannot step twice into the same stream…We both step and do not step in the same rivers. We are and are not.”

Heraclitus in Raphael’s The School of Athens, supposedly modeled off of Michelangelo.


The adult who steps across a stream may be a different person when he returns a year later, but a three-year-old, or a seven-year-old will be a really different person a year later. Adults may be ever-changing, but children are really ever-changing, in ways moderated not just by their environment and who they were yesterday, but by genes.


One of the main ways in which, it seems to me, mainstream social scientists misunderstand (and underestimate) the effects of heredity–if they consider it at all–is by viewing it as an endowment.
For example one will typically read a passage that says, “we observed kids at point x. This initial baseline represents the effects of parental and neighborhood inputs including genetics, socioeconomic status, and pre-school experiences. Then, at point y we observed them again. The difference between scores at these two points can be attributed to the effects of their educational environment.” Or (for the more sophisticated), “The difference in differences in scores between two groups of students is a proxy for the variation in classroom quality, and while membership in disfavored groups can reduce the expected gain from year to year, the variation within groups can be attributed to variation in classroom quality.” This is, for example, the way Value Added Models are constructed, and also the way in which economists construct  measures of the long-term effects of teacher quality on wages and other outcomes.
There are a number of problems here.
a) If heredity acts like an endowment at the beginning of schooling, then we would see the effects of heredity attenuate over time, rather than becoming steadily stronger in shaping cognitive traits. Moreover, equality in educational inputs appears to cause a greater percentage of variation in outcomes to be explained by heredity rather than less.
 Change in Heritability
b) The change in test scores conditional on baseline is not, empirically, a random walk. A five year old’s parents’ IQ is a better predictor of their adult IQ than their own IQ measured at five years old. Even advocates of pre-K mention that their effects on cognitive skills rapidly fade out (as do advocates of VAM in discussing fade-out of teacher effects.) While advocates argue that the re-emergence of long-term effects is due to long-term effects of interventions on non-cognitive skills that persist despite the fade-out of impacts on cognitive traits, this fade-out nonetheless strongly suggests that the change in scores at one point in time is not a reliable measure of the long-term effects of educational interventions.
c) At a theoretical/biological level, the idea of heredity as an endowment doesn’t work because, obviously, genes are a guide to a continuing process of transcription and protein synthesis that continues throughout the lifespan rather than a package that is delivered in toto before school age and then is done.
d) At an econometric level, it seems to me that many of these methods confuse ex ante prediction with ex post attribution. That is, you observe a kid at 6 years old, and he gets +0.5 SD on a standardized test. Given this, you predict that he will get a 1050 on his SAT at age 16. When he instead gets a 1400 SAT, you attribute the difference between prediction and actual score to the effects of a series of “positive shocks” the child experienced during his education. He had a +1.5 SD quality education– give those principals and teachers a raise!  Or, the kid scores a 800 SAT: he had a -1.5 SD education– we should have fired his principals and teachers long ago!
But this is wrong. While the distribution of potential outcomes for any kid who scores +0.5 SD at age 6 may have its mean at 1050, with a broad range, the actual distribution of potential outcomes for *this particular kid who scored 1400* was almost certainly higher and with a narrower range. The converse is true for the kid who scored 800. Otherwise we wouldn’t see such concordance among identical twins raised apart.


James Heckman generally disagrees with this, or at least believes that gene expression is so moderated by environmental influences so as to be not usefully characterized as an independent process. For myself, I don’t understand structural equation modeling well enough to agree or disagree with his recent work on the technology of skill formation.

Technology of Skill Formation

Obviously, it’s at least partially correct that investments make a difference: my guess is that the marginal return on any investment is determined not just by the quality of the program or the kid who receives it but by that child’s particular developmental receptivity and that particular moment in time. Practically every teacher who sticks with it for a while has had the experience of a former student returning to them years later to tell them, “man, your class saved my life,” to which you as teacher ask yourself, “really? All I remember is you and Jason playing paper ball soccer with each other on the desk.” But to that kid, at that time, it made a difference. Perhaps any half-way competent teacher would have made the same difference; perhaps if it hadn’t been you it would have been someone else the following year.

All the kid knows is that he has returned to the river, and neither he nor the river is the same.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s