Written by: Stephen Hsu
Primary Source: Information Processing
James Miller interviews Greg Cochran on a variety of topics.
Some comments on the early part of the interview (you might need to listen to it to make sense of what I write below):
1. The prediction I’ve made about the consequences of additive genetic variance in intelligence is not that we’ll be able to realize +30 SDs of cognitive ability. That would only be true if we could ignore pleiotropy, nonlinear corrections to the additive approximation, etc. What I claim is that because there are +30 SDs up for grabs in the first order approximation, it seems likely that at least a chunk of this will be realizable, leading to geniuses beyond those that have existed so far in human history (this is the actual claim). To doubt this conclusion one would have to argue that even, say, +8 or +10 SDs out of 30 are unrealizable, which is hard to believe since we have examples of healthy and robust individuals who are in the +6 or +7 range. (These numbers are poorly defined since the normal distribution fails to apply in the tails.)
### I could make further, more technical, arguments that originate from the fact that the genomic space is very high dimensional. These suggest that, given healthy/robust examples at +X, it is very unlikely that there is NO path in the high dimensional space to a phenotype value greater than X while holding “robustness” relatively fixed. ###
2. Greg comments on whether super smart people can have “normal” personalities. This is obviously not necessary for them to be viable contributors to civilization (and even less of an issue in a future civilization where everyone is quite a bit smarter on average). He posits that von Neumann might have been radically strange, but able to emulate an ordinary person when necessary. (The joke is that he was actually a Martian pretending to be human.) My impression from reading Ulam‘s autobiography, Adventures of a Mathematician (see also here), is that von Neumann was actually not that strange by the standards of mathematicians — he was sociable, had a good sense of humor, enjoyed interactions with others and with his family. He and Ulam were close and spent a lot of time together. I suspect Ulam’s portrait of vN is reasonably accurate.
### A masochist in the comments asked for the actual argument, so here it is: ###
Here’s a simple example which I think conveys the basic idea.
Suppose you have 10k variants and that individuals with 5.5k or more + variants are at the limit of cognitive ability yet seen in history (i.e., at the one in a million or billion or whatever level). Now suppose that each of the 10k + variants comes with some deleterious effect on some other trait(s) like general health, mental stability, etc. (This is actually too pessimistic — some will actually come with positive effects!) These deleterious effects are not uniform over the 10k variants — for some fixed number of + variants (i.e., 5.5k) there are many different individuals with different levels of overall health/robustness.
Let the number of distinct genotypes that lead to (nearly) “maximal historical” cognitive ability be n = (number of ways to distribute 5.5k +’s over 10k variants); this is a huge number. Now, we know of many actual examples of historical geniuses who were relatively healthy and robust. The probability that these specific individuals achieved the *minimum* level of negative or deleterious effects over all n possibilities is vanishingly small. But that means that there are genotypes with *more* than 5.5k + variants at the same level of general robustness. These correspond to individuals who are healthy/robust but have greater cognitive ability than any historical genius.
You can make this argument fully realistic by dropping the assumption that + effect sizes on cognitive ability are uniform, that effects on different traits are completely additive, etc. The point is that there are so many genotypes that realize [cognitive ability ~ historical max], that the ones produced so far are unlikely to maximize overall health/robustness given that constraint. But that means there are other genotypes (off the surface of constraint) with even higher cognitive ability, yet still healthy and robust.