Friday, November 15, 2013

Reading, Ciphering and Hereditability

Arun has been interested in a study which apparently shows that literacy and numeracy are more hereditable than, g (AKA IQ).  Should that be surprising?  Afterall, literacy and numeracy are learned skills.

The answer, I think, is no.  Consider an analog, skill at basketball.  Nobody becomes a really good basketball player without long and intense practice, but after an equal amount of practice different players will have hugely different levels of accomplishment.  Some reasons are obvious - strength, height, jumping ability, agility, and running speed, for example.  Others may be more subtle, like hand size, peripheral vision, muscle precision and so on.  All these are highly variable in the wild population, and strongly dependent on genetics.  Only persons with big constellations of these talents are likely to be able to play at the top level.

Having a parent with such a constellation, or two parents with big parts of the constellation will greatly increase one's chances of winning the bball constellation lottery.

By comparison, consider one component of the constellation, agility.  Other things being equal, this is almost certainly a generally useful trait.  Consequently, it's likely to be less variable than traits of more ambiguous value, like height or hand size, and hence less heritable.  Traits very strongly selected for, like having two arms, will show almost no heritability.

Very plausibly, numeracy and literacy depend on constellations of talents some of which have not been selected for in evolution, writing and arithmetic not having been that useful to our stone age ancestors.  General problem solving skills though, were certainly useful for them.  If IQ is a good proxy for those, it's quite natural that it would be less heritable than numeracy or literacy.