Monday, June 06, 2005

Intelligent Diseased Ashkenazim

The NYT article about intelligent diseased Ashkenazim is being widely discussed. People may want to read the actual study rather than the NYT writeup. Unsurprisingly, there are those who just don't like it (I don't know whether "it" is the actual study or the NYT writeup). What I don't understand is the demand to see a "mathematical-type model of the strength of pressures necessary to maintain mutations with these adverese effects based solely on the advatages the allegedly confer in terms of survival (though such models are themselves highly speculative, i think)" in order to credit the study. We are discussing a hypothesis that is relatively simple to test empirically.

5 comments:

miriam said...

(my initial post was a half-articulated response toward population-biology-speculation as a whole as much as a response to this particular thesis, but now that you've called me out i went back and read the original article in an attempt to actually make an argument. anyway, what follows is my first attempt at an such...)


"We are discussing a hypothesis that is relatively simple to test empirically."

as far as i can tell, there are two separate hypotheses involved.

H1: the ashkenazi genetic "disorders" are also linked to increased intelligence for carriers.

this seems to be what you think is "simple to test empirically," and the authors seem to have some convincing evidence.

H2: H1 is the reason these disorders persisted at such high levels in the population.

"testing" evolutionary hypotheses like this generally involves not only showing that they are based on plausible biological mechanisms (shared pathways, etc, as in H1) but that the extent of the pressures and the time involved are enough for the proposed mechanism to produce the observed effect. this is where "mathematical type models" come in, and where I think this study, and many like it on different topics, may be weak.


in population genetics, people do models to predict gene frequency after a certain number of generations given a certain effect on fitness. alleles that are only deleterious are supposed to reach a low (or zero) equilibrium level over time. how much time can be determined mathematically(allegedly) based on how deleterious they are... (this is why the persistence of ashkenazi disorders is cause for study in the first place.) on the other hand, alleles that increase fitness would be expected to rise in frequency, and how quickly they become how common can (at least ideally) also be determined mathematically, based on how strong an increase in fitness they confer.

the problems i see with the way these authors apply this reasoning to this case are:

A - they "assume" (p. 5, for one) the level of the selective advantage confered by intelligence, and then use this to show that the selective pressures stemming from this advantage would have been enough to produce the observed characteristics of the ashkenazi population.
this is the initial reason i said i "think all of this is silly" - it's really hard to quantify how beneficial something like "intelligence" is, and therefore hard to analyze its effects using equations like "R = h^2S." Anyway, this may be a necessary evil of scientific speculations on these topics for the time being, it just makes me unhappy, is all.

B - as far as i could tell, the authors don't do the same math they did to show that an intelligence-boosting mutation would produce the observed situation once they take into account the fact that these mutations have serious downsides. they take their value for the selective advangtage of intelligence without modifying it for the fact that the _same_ mutations have selective disadvatages. in my semi-thorough reading, i didn't see them show that the pressures for increased allele frequency can not only outweigh the pressures against, but can do so to such an extent that they can be responsible for the observed high frequencies.

miriam said...

p.s.

Shmuli said: "Unsurprisingly, there are those who just don't like it (I don't know whether "it" is the actual study or the NYT writeup)"

"it" was, more than anything else, any excerise in speculative evolutionary exaplanations. I usually do like them, actually, as intellectual curiosities, I just think they are more fun (or "silly") than "hard-science."*

(*I mean hard science in whatever way it applies to serious biology, not in some abstract philosophical way that will get me into trouble with my ghost in the HIPS office...)

Shmuli said...

To the degree that you're dismissing speculative reasoning, fair enough. There is, of course, much else that can be dismissed as similarly "silly". What I liked about this specific exercise, though, is the empirical component. If the empirical predictions are actually verified, it significantly increases the likelihood that their explanation for the widespread existence of the genetic disorders is mostly correct.

You are correct to point out that the authors did not do similar math for the downsides of the genetic defects. Perhaps this is because the same downsides should affect the competing theory. Either that math was worked out by a bottleneck theorist or our authors felt that because the bottleneck people didn't grapple with it they have no advantage from doing so.

miriam said...

"What I liked about this specific exercise, though, is the empirical component."

by "empirical component," do you mean that one can observe a relationship between these mutations and inelligence?

"If the empirical predictions are actually verified, it significantly increases the likelihood that their explanation for the widespread existence of the genetic disorders is mostly correct. "

not sure. it still assumes a lot about the relationships between survival and intelligence. evolution doesn't mean "genomes keep getting better" - if something makes you smarter, but being smarter doesn't help you pass on your genes, it has little/no evolutionary significance. i would say that the "empirical" part doesn't "increase the likelihood" - it just allows this theory to contend at all. but i'm not sure this disagreement matters so much.

"Perhaps this is because the same downsides should affect the competing theory."

actually, i think the whole point of the competing theory is that it explains strangely high prevalences based on random genetic factors. in this case, from what I can tell, the "bottleneck" or "founder" theories say that the ashkenazi population descended directly from a small group that happened to have certain traits, and that because of this artifically small starting point and low rates of genetic exchange from outside the community, those traits persist despite their costs.

From the critique in the Utah article (the NYT analysis being ultimately pretty usueless) it seems that such a random effect can explain the prevalence of these traits, it just doesn't jibe with the analysis of _other_ genetic markers.

(i'm not an expert in all this technical stuff at all, though...)

Shmuli said...

From the article - "Slatkin’s model assumes that the heterozygous carriers of Tay-Sachs suffer no fitness impairment and that the gene frequency is low enough that selection against homozygotes can be ignored, i.e. that neutral theory can be applied."

Moreover it is irrelevant because if you assume selection works against these defects, it should affect both theories. The difference between the two theories is the bottleneck people assume a very narrow set of demographic histories versus the Utah people assuming 1)IQ advantage and 2)consequent selection advantage. If there is significant adverse selection then both theories are pushed into narrower assumptions. There, of course, is always the possibility of convergence. We might have to assume both if adverse selection is assumed to be strong. If we accept the Utah crew's genetic evidence against the bottleneck theory, however, that seems to leave some selection mechanism as a necessary component of an explanation of these genetic defects.