Tuesday, July 31, 2007

Eugenics, schmgenics   posted by Razib @ 7/31/2007 12:43:00 PM

Ezra & Ross as still arguing about the definitions for eugenics and what not. Clearly there is a lot of baggage associated with the "e-word." In any case, in an email exchange with Armand Leroi about the use of the term "eugenics" to refer to selective abortions of individuals whose fitness verges upon zero (and so aren't going to have a long term impact on the gene pool anyhow) it seems clear that his own view of the term is more liberal than Ezra or mine. And when someone in John Brockman's stable talks, we should listen because these are the public intellectuals who have disproportionate impact on our understanding of scientific terms (another Brockmanite, Dawkins, was the one who introduced eugenics as one of his "Dangerous Ideas"). I suspect operationally more people would align with a broad usage of the term, though "eugenics" has too many negative associations for it to be resurrected I would bet. But in any case, as I have suggested the semantical argument is besides the point, no matter if x, y & z are instances of eugenics, x, y & z are already penetrating the domain of normalcy. As many of Ross & and Ezra's readers note part of Ross' objection surely has to do with the fact that he is opposed to abortion on principle, which is a proximate process via which selection for traits can occur. How would he feel about the screening of unfertilized gametes? One can imagine super-wealthy social conservatives going to the extent of not destroying life in such a manner. For Dune nerds you know that the Bene Tleilax perceive themselves to not be violating the injunctions of the Butlerian Jihad (which do include bans on particular genetic technologies), but they certainly violate the spirit of the law. Conversely, I assume that most pro-abortion rights liberals are not down with the creation of Aryan supermen, but genetic technology is going to be advanced enough soon that two parents who want blonde and blue-eyed children and have the genetic potentiality for such offspring can load the die. I have noted in email to friends that with the knowledge of the genetics of skin color many South Asian couples could now load the die so that their offspring would be selected from the lighter skinned range of the probability distribution (the extant variance of the South Asian genetic architecture naturally results in offspring that deviate from the expectation a lot). These are questions which I think are more interesting than the definition of eugenics.

Update: The Elf weighs in. I think she hits it about right, though this is a sprawling issue, made worse by the fact that there are disagreements about the term. The only qualification I would have is that some Lefty/progressives with a strong sympathy toward Deep Ecology and China's population policies might be the sliver of a connection that conservatives might be looking for between the past and the present (albeit, this is a very small group from what I can tell).

Related: Notes on Eugenics.


Malthusian me?   posted by Razib @ 7/31/2007 12:09:00 PM

In the comments below in regards to eugenics I made an argument that rationing is going to be inevitable in national health care systems as the information we have about the propensity (or inevitability) of diseases outruns the ability to treat those diseases. In particular, I believe that it may come to the point where though one can treat something in theory with medical technology the costs may simply be prohibitive. The argument I'm making smells a lot like Malthusianism, insofar as I believe that genomic diagnostic technologies will decrease in price in a manner that scales downward to a far greater extent then the treatments for those diagnosed probabilities for the total sample space of possibilities. But of course, we know how Malthusianism worked out when it came to the argument about population & food production, so I'm not sure about this. We (humans) have a tendency not to account for future innovation. In the near-term (less than 20 years) what do those who know about the costs of medicine think?


Living Neandertals?   posted by Razib @ 7/31/2007 10:11:00 AM

On my other weblog I posted about research which suggests Neandertal-human cohabitation in France. A reader pointed me to the visual proof of the hybridization event.


Monday, July 30, 2007

GWA for multiple sclerosis   posted by p-ter @ 7/30/2007 09:34:00 PM

The latest phenotype to get the scrutiny of a genome-wide association study is multiple sclerosis: three separate reports (ok, only one of them is genome-wide) point to variation in various immune system genes as predisposing to the disease. The effects of one of the variants seems to be non-additive-- one group reports that the heterozygotes for the "causal" allele seem to actually be protected, while the homozygotes have a higher risk.

There are a number of reason why this could be the case-- linkage disequilibrium patterns and the existence of multiple predisposing alleles can lead to odd patterns of risk, even flipping the apparent effect in some cases. Another possibility, of course, is that there's some interesting biology there. More research, as they say, is needed.

Labels: ,

Eugenics, what's in a term?   posted by Razib @ 7/30/2007 07:30:00 PM

Ross Douthat is concerned with the "New Eugenics." He linked to my summation of some of the data which Armand Leroi has collected on the rise of selective abortions & genetic screening. Ezra Klein isn't buying Ross' characterization of course; actually, like Ezra I think there are serious differences between the old eugenics which emerged from the biometrical school (which became quantitative genetics) and the new eugenics which is predicated on the ubiquity of genomic & fertility technology. Nevertheless, in the proximate sense, for example in our life spans, it might not make much of a difference whether the selective abortions are of heritable traits or those with would result in sharply reduced fitness anyhow (like Down Syndrome). It seems to me that to some extent the Left, which does not fear reproductive technology, is allergic to the term eugenics because of its historical resonances. I'm not going to argue over a word. But, I will offer that Richard Dawkins, no Ross Douthat, is willing to bring up the term in reference to the trends in genetics. I think that sentiments and dynamics reminiscent of the old eugenics is pretty much inevitable now that "we have the technology," it really doesn't matter what you call it.

P.S. And once we get government backed health care in the USA I'm pretty sure that the element of coercion will come into play (perhaps we'll call it "incentives for terminative preventative care" or something).


Religion promotes cooperation?   posted by Razib @ 7/30/2007 12:19:00 PM

Religious concepts promote cooperation:

Participants primed with religious concepts gave their partner an average of $4.22, compared with only $1.84 in the control group. But those who declared themselves religious before the study were no more generous than non-believers.

"The effect of the religious prime was both large and surprising, especially considering that during exit interviews the participants were unaware of having been religiously primed," says Shariff.

A second study introduced a third group, primed with words associated with civic responsibility such as "jury", contract", and "police." This group behaved almost identically to that primed with religious concepts.

You can read the full working paper for free. There were two groups. One consisted of 50 UBC students, and the second a somewhat larger and more diverse group from the Vancouver, BC, area. The basic finding was that "priming" subjects with religious terms seemed to elevate generosity during an un-iterated Ultimatum Game, where the 'rational actor' should just keep all the money. In the first sample there wasn't even a statistically significant difference between religious & irreligious students in how they reacted to the priming. The second study was more equivocal, and the authors in the discussion suggest that part of the reason that the irreligious tended to be less responsive toward religious priming was that the greater stringency of the test for 'atheism' filtered the individuals to a greater degree who were defined as non-religious, and a small number of subjects might simply even lack the implicit resonances of supernatural agents. Finally, the second study also showed that subjects could be primed toward generosity by exposing them to civic terminology.

First, the authors note the problems with their small and narrow sample sizes. Though statistically significant and powerful, the effects were derived from people from the Vancouver area, or, college students at UBC. I didn't see controlling for the fact that there is likely some correlation between ethnicity and religion in British Columbia. Specifically, a disproportionate number of secular British Columbians are likely to be Chinese origin. Second, cognition expresses and develops within a cultural context. In a society with less civic engagement and activity than Canada I would not be surprised if the effect of secular priming was trivial. Similarly, in a society that is extremely secular (Japan?) one might see far greater response to civic priming than the supernatural equivalent. Third, the authors suggest that the response of theistic and non-theistic individuals in the first group to supernatural concepts suggests an implicit association between religious concepts and altruistic behavior. I have suggested myself that the anthropomorphic bias which is a pillar of religiosity exists in many, or all, atheists. Rejection of a deity might be sincere on the explicit level, but the implicit mind might still be strongly shaped by early cultural conditioning. The secular individuals in the UBC sample were no doubt aware of the valences and power of religious beliefs and ideas, and it seems plausible that lifetime implicit associations would have been built up.

Overall, this study is good because as the researchers point out there is a lot of armchair bullshitting on this topic. I get plenty of it in the comments of my weblogs. This study shows supernatural agents can act as mediators of human action as posited by many. It also shows that secular institutions and values can trigger the same change in behavior. What does this tell us on the fundamental level? I'm not sure, after all, the typical modern human has been exposed to several thousand years of philosophical religion which has embedded within it an explicit moral/ethical dimension. Similarly, bureaucratic government and the ideologies of mass societies are "in the air," so to speak. In some "primitive" societies gods are seem as much more amoral creatures than in "advanced" cultures; they are mischievous agents who humans must placate and deceive. Additionally, they have no well developed theories of statecraft or a conception of law enforced by political fiat. It would be interesting to do this sort of study in a primitive society, though obviously the lack of literacy would cause problems with the priming the researchers used in this case.

Labels: ,

Pedo amygdala   posted by amnestic @ 7/30/2007 03:17:00 AM

I did a poor job asking pubmed for the paper Razib mentioned earlier, but this surely does look interesting:
Brain pathology in pedophilic offenders: evidence of volume reduction in the right amygdala and related diencephalic structures.
Kolja Schiltz, Joachim Witzel, Georg Northoff, Kathrin Zierhut, Udo Gubka, Hermann Fellmann, Jörn Kaufmann, Claus Tempelmann, Christine Wiebking, Bernhard Bogerts

CONTEXT: Pedophilic crime causes considerable public concern, but no causative factor of pedophilia has yet been pinpointed. In the past, etiological theories postulated a major impact of the environment, but recent studies increasingly emphasize the role of neurobiological factors, as well. However, the role of alterations in brain structures that are crucial in the development of sexual behavior has not yet been systematically studied in pedophilic subjects. OBJECTIVE: To examine whether pedophilic perpetrators show structural neuronal deficits in brain regions that are critical for sexual behavior and how these deficits relate to criminological characteristics. DESIGN: Amygdalar volume and gray matter of related structures that are critical for sexual development were compared in 15 nonviolent male pedophilic perpetrators (forensic inpatients) and 15 controls using complementary morphometric analyses (voxel-based morphometry and volumetry). Psychosocial adjustment and sexual offenses were also assessed. RESULTS: Pedophilic perpetrators showed a significant decrease of right amygdalar volume, compared with healthy controls (P = .001). We observed reduced gray matter in the right amygdala, hypothalamus (bilaterally), septal regions, substantia innominata, and bed nucleus of the striae terminalis. In 8 of the 15 perpetrators, enlargement of the anterior temporal horn of the right lateral ventricle that adjoins the amygdala could be recognized by routine qualitative clinical assessment. Smaller right amygdalar volumes were correlated with the propensity to commit uniform pedophilic sexual offenses exclusively (P = .006) but not with age (P = .89). CONCLUSIONS: Pedophilic perpetrators show structural impairments of brain regions critical for sexual development. These impairments are not related to age, and their extent predicts how focused the scope of sexual offenses is on uniform pedophilic activity. Subtle defects of the right amygdala and closely related structures might be implicated in the pathogenesis of pedophilia and might possibly reflect developmental disturbances or environmental insults at critical periods


Sunday, July 29, 2007

Selection on memory?   posted by p-ter @ 7/29/2007 08:30:00 PM

Razib recently mentioned a paper on a polymorphism regulating memory in Drosophila. As I often do when studies like these are published, I determined the human homologue of the gene in question (PRKG1, in this case), and checked out the plots of summary statistics available online. Sure enough, there are peaks in the significance of iHS, Tajima's D, and Fay and Wu's H in the area (all signs of recent selection), though strikingly these peaks are only evident in the Asian population. Hm.


Neuroscience, cancer/biology, math videos   posted by amnestic @ 7/29/2007 06:55:00 PM

Recently purchased a video iPod. I am the nerd sitting outside the coffeeshop groking brain network dynamics on my tiny screen. At least my case is stylish.

Here is a trove on that subject:
Conference on Brain Network Dynamics, 1/26/2007

That conference was in part a tribute to Walter Freeman. Here is more from him:
Poetry of Brains

Digging around more you can find several videos from the Redwood Center for Theoretical Neuroscience. For instance, Micro-circuits of Episodic Memory: Structure Matches Function in the Hippocampal System, and a debate entitled "Waves or words in cortex?" featuring Professor Freeman again.

On the mathy tip, you can download .mov's on some pretty interesting topics from the Mathematical Sciences Research Institute. If someone can point me to a converter so I can carry them around with me I would be muy grateful. I'm more interested in the ones that seem to have some relation to biology such as Frances Tong on Normalization of Western Blots.

Speaking of .movs to convert. I found an extensive discussion by Mark Ptashne on his book, A Genetic Switch. Presumably this coincided with the release of the new edition: A Genetic Switch, Lecture Series.

Saturday, July 28, 2007

Nerds   posted by Razib @ 7/28/2007 07:21:00 PM

Who's a Nerd, Anyway?:
But the nerds she has interviewed, mostly white kids, punctiliously adhere to Standard English. They often favor Greco-Latinate words over Germanic ones ("it's my observation" instead of "I think:), a preference that lends an air of scientific detachment. They're aware they speak distinctively, and they use language as a badge of membership in their cliques. One nerd girl Bucholtz observed performed a typically nerdy feat when asked to discuss "blood" as a slang term; she replied: "B-L-O-O-D. The word is blood" evoking the format of a spelling bee. She went on, "That's the stuff which is inside of your veins," humorously using a literal definition. Nerds are not simply victims of the prevailing social codes about what's appropriate and what's cool; they actively shape their own identities and put those codes in question.

I think the researcher interviewed is a bit too obsessed with straight-jacketing nerds into a racial identity (white). I speak as a brown nerd, though I doubt I'm as socially antagonistic toward colloquial slang and conventional mores as the archetypical nerds. Rather, I think the key to nerditutde is the lower emphasis on being accepted and so assimilating the normative tardish value system (who cares if your friends make fun of you for reading? Keep at it!). Of course, many more socially aware intrinsic nerds become adept at mimicking the tardish comportment during high school, only to show their "true colors" in college when they continue to focus academically and go on to professional jobs. Any true nerdologist has to grapple with the reality that the majority of nerds might actually be "passing."


Friday, July 27, 2007

Social networking, does it work?   posted by Razib @ 7/27/2007 10:34:00 PM

Just a quick question for readers: does social networking software help out in your professional life? I'm a very tepid user of the various sites, I accept invites and so on, but it isn't something I invest a lot of time on in building a large of number of friends/contacts or fleshing out my profile. I know that some research has shown that a professional contacts are often found through your second-tier relationships, that is, not close friends but good acquaintances and what not. People who you know, but whose contacts don't overlap much with yours. So I suppose that's the point of something like linkedin. But does it really work? I've receive much better contacts through the blog, email, e-lists and so on.

(note, I am not disputing that social networking software helps your personal or social life. I know many people who've gotten action through friendster or myspace, but no one who has made professional gains via linkedin)


True porn clerk stories   posted by Razib @ 7/27/2007 06:44:00 PM

Most of you have probably already seen/heard about this, but check out true porn clerk stories.


Homo amygdala?   posted by Razib @ 7/27/2007 04:54:00 PM

A Mind for Sociability:
The amygdala, a small, almond-shaped area deep within our brains, appears to be essential in helping us read the emotions of others. Research shows that the structure is crucial for detecting fear, but scientists have also found evidence that it can help spot a wide variety of mental states...scientists noted that the amygdalas of patients with autism, which is characterized by decreased social interaction and an inability to understanding the feelings of others, have fewer nerve cells, especially in a subdivision called the lateral nucleus.
In humans, however, the lateral nucleus occupied a bigger fraction of the amygdala, and was larger compared to overall brain size, than in the other species, the team reports online today in the American Journal of Physical Anthropology. Although the functions of the amygdala's subunits are unclear, the lateral nucleus makes more direct connections with the brain's temporal lobe--which is involved in social behavior and the processing of emotions--than other parts of the amygdala make, the researchers note.

In Grooming, Gossip, and the Evolution of Language Robin Dunbar argued for the critical selective pressure of social groups in driving up the size and complexity of the human brain (and obviously, the emergence of language). This might explain the gradual increase in brain size over the past few million years until about 200,000 years B.P., but what about the Great Leap Forward & expansion out of Africa ~50,000 years ago? Remember, behaviorally modern humans postdated anatomically modern humans (e.g., a form of H. sapien which was gracile, high cranial vault, etc., was extant in Africa before expanding to the rest of the world) by 150,000 years. In The Dawn of Human Culture Richard Klein suggests that there was a biological change, a reorganization of the brain (Dunbar offers this idea as well). Greg has suggested that Neandertal introgression & hybrid vigor might have been at work; remember that Neandertals had the largest cranial volume of any Homo species. In The Prehistory of the Mind: The Cognitive Origins of Art, Religion and Science Steven Mithen suggests that the breakdown of separation between domain specific intelligennces (e.g., social intelligence, theory of mind, intuitive physics, folkbiology, etc.) was the critical factor in triggering the cultural revolution which lead to modernity. Mithen argues that the use of analogy to map across the various domains, and apply insights from each domain to the others, might have resulted in a massive increase in cognitive flexibility and creativity. A neurobiological implication that our species' amygdala is more "hooked in" with our "higher cognitive functions" seems to lend some credence to that viewpoint.

Update: Kambiz has more.


The typical GNXP reader   posted by Razib @ 7/27/2007 02:11:00 AM

When you've blogged for a while, and with some frequency, you wonder what this is all about. I don't generally get too caught up in that, there's more interesting stuff to contemplate. But, check out this from Google Analytics for the past 30 days of traffic for this website:

I've long known that most GNXP "readers" are "one off" events. I have no problem with that, if they find what they're looking for then you've done some good. That being said, I was a little shocked (and pleased, frankly) to see that over 8,000 visitors have arrived over 200 times in the past 30 days! I don't know, or care, about the details of how Google calculates this, rather, I'm interested in the gestalt sense of what's going on. Obviously the same people don't come everyday, but our unique visitor traffic has been in the 2,000-4,000 range for years, so these data together suggest that many people skip days (perhaps refreshing the site a lot on the weekend, or during one particular post where they participated in the thread?).

(click image for larger view)


Sex & epistasis   posted by Razib @ 7/27/2007 01:44:00 AM

Since we're talking about sex & evolution I thought I would pass on this PNAS paper, Coevolution of robustness, epistasis, and recombination favors asexual reproduction (my emphasis). I covered statistical epistasis a few years ago, and at that point I was being told that synergistic epistasis was the critical cog that kept the species a rollin' and a rockin'. But perhaps not (the paper is Open Access).


Thursday, July 26, 2007

unSexy crayfish   posted by Razib @ 7/26/2007 11:18:00 AM

The Economist has a story about a mutant asexual crayfish lineage. It points out how this is a good test of the sex-is-good-against-disease thesis. The basic logic is that an allele which causes asexuality has a greater short term natural increase (because it has a 100% as opposed to 50% chance of being passed to the next generation). Nevertheless, because the conventional reshuffling that occurs during sexual reproduction through the synthesis of two haploid gametes into a diploid (as well as recombination) is not operative in a the clonal lineage, they are vulnerable to fast adapting pathogens which might wipe them out. One could conceptualize the modal genetic profile of the lineage as the selective pressure exerted upon the pathogens. In the case of an asexual lineage the the pathogen has a straight shot at a stationary adaptive peak (mutation of course will introduce variation over time). In contrast, sexual lineages with their wide distribution of of genetic architectures are like moving targets, as a population of pathogens shifts toward one peak, selection within the host population reshapes the adaptive landscape (remember the power of selection is proportional to extant heritable variation, which sexual lineages have in abundance vis-a-vis clonal ones). Imagine, if you will, a rugged adaptive landscape with many peaks always bubbling and morphing, as the pathogens race up the slopes the ground underneath them shifts and gives away and soon they find themselves having to traverse a radically altered surface. In contrast, an asexual population would exhibit one sharp peak of relative stability, and its lack of genetic variation means that the surface itself is relatively rigid and stable.

Note: Some researchers have proposed that the persistence of sex in complex organisms is a function of phylogenetic constraint. That is, once the lineages flipped from asexual to sexual they couldn't flip back, even if it was adaptive. This sort of view is rather diminishes the power of selection to overcome phylogenetic hurdles. To my mind crayfish are a rather complex organism. But to those of you who know organismic biology: what is the most complex animal which can reproduce asexually? (I'm assuming it is a reptile?) It is supposed that imprinting prevents the emergence of asexuality in mammals.


Wednesday, July 25, 2007

Open thread....   posted by Razib @ 7/25/2007 09:49:00 PM

People have bugged me about the "open thread" for a while, so by popular demand, it's back. It will be to the right indefinitely, though I'll purge it of old/tardish comments regularly. Please place interesting links/sites, etc. (this includes items of interest to young straight males, *hint*, *hint*).


Individualism & collectivism   posted by Razib @ 7/25/2007 11:30:00 AM

Self-centered cultures narrow your viewpoint:
Chinese students would immediately understand which wooden block to move - the one visible to both them and the director. Their US counterparts, however, did not always catch on.

"They would ask 'Which block?' or 'You mean the one on the right?", explains Keysar. "For me it was really stunning because all of the information is there. You don't need to ask," he adds.

While 65% of the American participants asked this type of question, only one of the 20 Chinese subjects did so, equating to just 5%.

This comes close to the classic "they did a study on that?" criticism of psychology. Of course we know that East Asians cultures emphasize a contextual perception of self and collectivist values vis-a-vis Western ones. But, it is nice to get a quantitative sense of the extent of this difference. You can read the full paper on the author's website. Here is an important point:
In fact, language can trigger a culture - bound representation of self...bicultural Chinese-born individuals tended to describe themselves in terms of their own attributes when writing in English, but to describe themselves in relation to other people when writing in Chinese.

This shouldn't surprise you if you read Geography of Thought. It seems people can be easily "trained" to change their vantage points (casting some light on the grand claims made by the author in the aforementioned book). The facultative nature of these extreme differences seems pretty obvious; especially given that extreme individualism manifest in English speaking peoples in particular, while continental Europeans tend to lay between the East Asian collectivism and Anglo-Saxon collectivism despite their far closer genetic affinity to the latter. Nevertheless, though the extremity of the differences in operation of Theory of Mind here is likely cultural, I can not be suspect that there might small, but significant, initial differences between populations. After all, Jerome Kagan has shown that personality differences exist between Asian and European infants at very young ages, as well as between blue-eyed and brown-eyed children (in both cases the former tend to be more withdrawn and inhibited than the latter). The question that comes to my mind is whether the cultural differences selected for different personality profiles, or whether there were initially difference personality profiles which resulted in different cultural outcomes. My own suspicion in the East Asian case is the former.


Small teeth & sexual dimorphism?   posted by Razib @ 7/25/2007 12:23:00 AM

A Hunk's Dental Downfall:
When males and females were about the same size, so were their teeth. But in species in which larger males evolved, tooth size increased relatively little. Thus, females ended up with larger chewing surfaces for their size than did males, the researchers report in the September issue of American Naturalist. The team concludes that teeth probably didn't grow at the same rate as body size because males can successfully compete for females only in their prime. Once teeth wear down, they become ineffective, and the animal gets weaker and more susceptible to disease or injury. But that doesn't matter to these males, as once they are too old to beat out rivals for mates, there's no need to live a long life. When it comes to how many offspring a male can father, "it seems that compared to body mass, tooth size is relatively unimportant," says Joanne Isaac, a mammalogist at James Cook University in Townsville, Australia, who was not part of the study team.

In highly polygynous species males in their prime are the fathers of a multitude. These species' males enter into a winner-take-all lottery game when it comes to reproduction. It makes sense that these males wouldn't live that long. It isn't likely that they could greatly increase the fitness of their numerous offspring through parental investment simply because there might be so many of them. Male investment in humans makes some sense in the case where a typical man may have only a few children who survive to their reproductive years. Nevertheless, there is some reproductive skew within our own species, and the extent of that skew varies from population to population and across historical epochs. The reproductive outcome for the total population may remain the same no matter if it is characterized by a equilibrium of low risk & low yield male strategies, or high risk and high yield strategies, but the dynamics within the society are likely going to be very different. I am not convinced that our current low risk & low yield strategy (i.e., monogamous pair-bonding) isn't just a metastable situation, highly susceptible to disruption.


Tuesday, July 24, 2007

Somatodendritic microRNAs   posted by amnestic @ 7/24/2007 10:54:00 PM

Kosik and colleagues used laser capture microdissection to get RNA populations from dendrites or cell bodies of cultured rat neurons. They optimized their technique so that mRNAs known to be enriched in dendrites, such as CaMKII and MAP2, showed about equal levels from soma and dendrite. They then performed multiplex PCR for several mRNAs and 187 miRNAs. The distribution of mRNAs and miRNAs is similar with a large somatic population and a gradient going out into the dendrites. Some small proportion of miRNAs have a little bit of dendritic enrichment. One point the authors are trying to get across is that there is no such thing as a 'dendritic RNA' because even the mRNAs and miRNAs that show some dendritic localization usually show just as much in the cell body.

Two nice things are that this paper validates a couple miRNA target prediction programs (Pictar and Targetscan) and that they provide a quantitative view of the miRNA copy number per cell. Both of these prediction programs suggested that miR-26a would target MAP2. This is convenient since both showed a somatodendritic distribution, meaning they hang out together even out at the farthest dendritic reaches. Inhibition of miR-26a with a synthetic oligonucleotide resulted in increases in MAP2 protein expresion, as one would expect from the classic miRNA-target relationship. As far as I am aware this brings the total of known dendrtici miRNA target pairs up to three, the other two being mir-268 and CaMKII (in drosophila) and miR-134 and LIM-Kinase. Quantification was achieved using PCR with known copy number standards. They knew how many cell bodies they captured, so they could get a copy number per cell estimate (probably a minor undershoot since even if they are awesome they probably couldn't save allllll the RNA from degradation). Anyway, they found... well I'll let them explain it:

rno-miR-124a is among the most abundant miRNAs in neurons and fell in the range of 10^4 copies per neuronal cell body. Despite its abundance, rno-miR-124a is enriched in cell bodies. rno-miR-26a and rno-miR-16 are less abundant miRNAs and fell in the range of 10^3 copies per neuronal cell body (Table 6). Because (delta)Ct of 2.61 +/- 0.39 describes the distribution of most miRNAs between the cell body and neurite, the number of copies of many miRNAs distributed along this gradient may be as low as in the hundreds of copies in the dendritic compartment. Even a one-order-of-magnitude error in this number is far below the number of synapses on the dendritic tree, and, therefore, the copy numbers of many miRNAs are likely to fall below one per synapse.

Delta Ct refers to the number of PCR cycles (i.e. doublings) it takes for the dendritic levels to reach the somatic levels. For instance, a delta Ct of 2.61 means that there are 2^2.61 (~6.1) times more somatic copies of the miRNA than there are dendritic copies.

I was particularly intrigued by this last sentence even though I have no idea what it means:

Stochasticity derived from the effects of miRNAs will contribute to the activation barrier for coherent responses, to the utilization of information provided by translational bursting, and to the flexibility needed by dendrites to sample alternative states (Kaern et al. 2005).

Guess I'll have to read Kaern et al. real quick.

Labels: , ,

Monday, July 23, 2007

The genetics of HIV infection   posted by p-ter @ 7/23/2007 08:16:00 PM

AIDS is obviously not a genetic disease-- if one were to make a list of risk factors predisposing to HIV infection, genetics would be a pretty low-ranking member (though still present on the list, of course). Yet genetics is still a useful tool for understanding the disease, as evidenced by this paper:
Understanding why some people establish and maintain effective control of HIV-1 and others do not is a priority in the effort to develop new treatments for HIV/AIDS. Using a whole-genome association strategy we identified polymorphisms that explain nearly 15% of the variation among individuals in viral load during the asymptomatic set point period of infection. One of these is found within an endogenous retroviral element and is associated with major histocompatibility allele HLA-B*5701, while a second is located near the HLA-C gene. An additional analysis of the time to HIV disease progression implicated a third locus encoding a RNA polymerase subunit. These findings emphasize the importance of studying human genetic variation as a guide to combating infectious agents.
People trained to think about disease from a "public health" (read: short-term) standpoint might be a little appalled by the amount of money spent on a study like this-- those hundreds of thousands of dollars could very well have been spent on far more effective ways to reduce AIDS prevalence.

However, the goal here is more long-term-- understanding the variation in how humans interact with pathogens will lead to more effective drug targeting and greater understanding of immunity down the road. Genome-wide association studies also, given the fact that they are largely hypothesis-free, also provide a way to generate novel hypotheses (or confirm old ones) about disease aetiology. In this study, for example, one of the major signals lies in an endogenous retrovirus-- that is, a virus that has incorporated itself into the genome. This raises the intriguing possibility that some of our immune response is mediated by viruses that previously spliced themselves into the genome (the authors mention that antisense transcripts would be a very plausible mechanism by which that could work).

The genetics of any phenotype you can think of will eventually be mapped, and this information will be useful not necessarily for its predictive value (though in some cases that will be the case), but also for the basic understanding of the phenotype that it carries with it. This site sometimes sees speculation as to the causes of variation in sexual orientation, for example-- genetic studies (assuming they're carried out) will severely restrict the plausible "hypothesis space" on that question.

Labels: , ,

Math and neuroscience   posted by Razib @ 7/23/2007 03:59:00 PM

I'm assuming that readers who know some neuroscience can make more sense of this paper, Monotonic Coding of Numerosity in Macaque Lateral Intraparietal Area (Neurons for Numerosity: As Quantities Increase, So Does the Neuronal Response, a summary for the general public). But this part of the abstract is what caught my eye:
The responses of these neurons resemble the outputs of "accumulator neurons" postulated in computational models of number processing. Numerical accumulator neurons may provide inputs to neurons encoding specific cardinal values, such as "4," that have been described in previous work. Our findings may explain the frequent association of visuospatial and numerical deficits following damage to parietal cortex in humans.

I'm sure most of you know know in psychometrics there is a correlation between visuospatial & mathematical aptitude. There is one group though which decouples these two traits: Ashkenazi Jews (who are weak on visuospatial tests in relation to their mathematical aptitude).


Sunday, July 22, 2007

Reporting genome-wide association studies   posted by p-ter @ 7/22/2007 08:56:00 PM

RPM points to a post from Mark Lieberman at Language Log on the reporting of genome-wide association (GWA) studies. His request (for the popular press; these things are always in the actual paper): report the allele frequency of the associated allele in cases, as well as the frequency in controls. I've often argued that people will get used to the complexity of "complex" disease once they're able to say "Oh, I know several people with the 'diabetes gene' that never got diabetes", but this is a more proactive measure towards that end, and I think it's a great idea.

Lieberman includes the numbers for the "restless leg syndrome"-associated allele I mentioned recently; the other disease I mentioned in that post was gallstone disease, in which the associated allele has a frequency of 10% in cases and 5% in controls. I'll try to remember to report those numbers every time I mention a GWA study from now on.

ADDENDUM: I also like the suggestion from the comments that posterior probabilities be given, as they are sometimes more intuitive. That is, if A is the disease allele and D is disease status, P(A|D) is less interesting, in some sense, than P(D|A). Unforunately, disease prevalances aren't always well-defined, and P(D) is necessary for the calculation. For gallstone disease, P(D) is about 0.15, so P(D|A) = P(A|D)P(D)/[P(A|D)P(D) + P(A|!D)P(!D)] = 26%. The corresponding posterior probability P(!D|A) is 74%. So someone carrying the A allele has a 26% chance of developing the disease, and a 74% chance of staying healthy.

Labels: ,

Saturday, July 21, 2007

In the name of a word   posted by Razib @ 7/21/2007 04:16:00 PM

The Man Who is Thursday has a long post titled Christocentrists: Mormons as Non-Christians. His basic argument is that there are some necessary preconditions assumed with the term "Christian" which Mormons violate, and so though they are followers of Jesus Christ, who they believe is the Son of God, they should not be considered Christians. The post points out my own main issue with the Mormon contention that they Christians: I do not think that other Christians, Muslims or Jews could consider them monotheists,1 so the fact that they revere Jesus Christ as their savior seems like a moot point. Of course, as someone who is not a Christian I don't have a strong opinion on this topic, I can grant Mormons their own self-definition as Christians and I can respect that other Christian groups do not consider them Christians. For me the primary issue is the mapping of the term Christian to a set of characteristics, after all the transmission of information is for me the primary role of terminology.

The Man Who is Thursday is a Christian (from the frozen north I gather), so the debate has more salience for him. What for me is an illustration of linguistic peculiarities and an anthropological curiosity is for him something which goes to the root of his beliefs about the ontology of the universe. This highlights the critical role that names play in the ecology of human social relations and the arc of history. I have made the point before that psychological and anthropological study has unmasked the reality that the vast majority of humans who subscribe with deep sincerity to "higher religions" do not truly conceptualize with any clarity the metaphysics to which they accede and profess as distinctive elements of their creed. In other words, the varieties of theology promoted by Muslims, Trinitarian Christians and devotional Hinduism have little relevance toward the psychological state of the individual believer when focused upon their deity of choice, the mental model of supernatural agency seems a human universal. But, despite this fundamental similarity the names ascribed to abstruse philosophical systems are essential toward coalescing groups which engage in conflict. The schisms of early Christianity, rooted in extremely fine and subtle philosophical distinctions, which sometimes resulted in persecutions and deaths, are illustrations of this principle. Similarly, the hostility of a substantial fraction of evangelical Christians to Mitt Romney despite his social conservatism, in principle (if of recent origin) and practice (in his personal life), also show the power of names.

But this is not restricted to religion. Consider the reluctance that many southern whites had in discarding their historical affiliation with the Democratic party. Or the latency that many who operationally change political orientation exhibit when it comes to shedding their old self-identification for a new (many neoconservatives remained Democrats for far longer than their political change would have implied). These latencies are important to consider because they have real world impact. I suspect that the tendency for many southerners to retain at least local Democratic affiliation slowed down the progress of the conservative revolution. Similarly, I would not be surprised if a Mormon nominee in the Republican party is not of much concern in a generation or two, but until then the latency of identification is going to have real world consequences (e.g., the nomination of Rudi Gulianni because conservatives won't coalesce around any one candidate?).

Addendum: In the current context the importance of names is pretty obvious in a place like Iraq. The emergence of a Arab Shia majority within the borders of what we call Iraq is probably only a fact of the past two hundred years. It seems that as modern irrigation techniques opened up vast swaths of southern Iraq to settlement by nomads who became farmers, there was a switch between a Sunni religious identity to a Shia one, in part due to the missionary activities of the residents of the Shia holy cities around which the farmlands lay. Over time there was a particular mapping between being Shia in Iraq (a southern peasant) and a Sunni Arab (a northerner, a nomad, or a part of the Ottoman era power structure). Fissures that emerge due to different interests of course can become crystallized as a "Shia vs. Sunni" conflict, though there are a host of other parameters at work (as illustrated by the fact that the Shia of southern Iraq ultimately repulsed the Iranian invasion during the 1980s). But the importance of a name should not be overemphasized, after all, Sunni Kurds have no qualms with pragmatically aligning with the Shia Arabs so as to marginalize their Sunni Arab co-religionists. The critical thing about names is their plasticity and manipulability, they are mental constructs and so extremely malleable after considerations of latency and cognitive friction are taken into account.

Update: From the comments, this makes my point much clearer:
....The associated metaphysics is secondary to the potentiation of collective action. Once a flag gets carried across a tribal border, be it a tribal flag, a national flag, a religious flag or whatever in the home context, across the tribal border it’s generally a de facto tribal flag.

I want to emphasize that this issue isn't limited to religion & metaphysics. After all, how many communists read Das Kapital front to back? Religious or political movements need the appropriate psychological "hooks" to have mass appeal, but they also seem to gain credibility through the generation of obscure intellectual justifications.

1 - There is some question among Jews and Muslims whether Trinitarian Christians are monotheists. But, the key is that there is a question; I do not think with Mormons there can be a question.


Evolution by mutations of large effect...or not?   posted by p-ter @ 7/21/2007 09:17:00 AM

The model of evolution that came out of the "Modern Synthesis" (see the figure on this site) predicts that evolution by natural selection occurs as a sort of stepwise optimization algorithm-- that is, a population will approach some fitness "peak" in a series of small steps, rather than all at once. This sounds rather reasonable-- making small changes to an organism is more "conservative", in some sense, than drastic restructuring with a single mutation, and thus less likely to knock out critical pathways. Yet there have seemed to be some examples of a single locus controlling major morphological differences between species. One example comes from Drosphila, where two species--D. melanogaster and D. sechellia--show rather different devlopmental patterns at a certain larval stage. This difference has been mapped (via inter-species crosses-- unfortunately not something that can be done in, say, humans and chimps) to a single locus referred to as shavenbaby (this being Drosophila, gene naming conventions are, well, non-existent).

However, it was unclear whether the changes in the regulation of shavenbaby were due to a single mutation of large effect, or several mutations of smaller effect. The latter would be predicted by the Fisherian model, and indeed, new research has found the precise genetic changes underlying the difference, showing that there are several mutations in different enhancers that individually generate smaller changes in shavenbaby expression.

The position of the gene in the developmental network is shown in the figure (from here), and the authors speculate that the reason for multiple regulatory changes at this single locus is due to the architecture of the network. In their words:
Given that laboratory-induced mutations in dozens of genes alter trichome patterns [the developmental pattern in question here], it is striking that multiple mutations at a single locus account for the entire evolved difference. Svb seems peculiar in the network of genetic interactions that establish the trichome pattern, because it sits at the nexus of the upstream patterning genes and the downstream effector genes. Although trichome pattern could be changed by altering any of several upstream genes, these changes would probably produce pleiotropic effects on other developmental processes. In contrast, none of the known downstream genes is sufficient on its own to prevent or promote trichome formation. Thus, changes at svb enhancers may provide the only available genetic mechanism to evolve trichome patterns without pleiotropic consequences.
Mutations cannot be fully studied in isolation-- they do not exist outside the context of genetic and biochemical networks.

Labels: ,

Thursday, July 19, 2007

The continuing success of genome-wide association studies   posted by p-ter @ 7/19/2007 08:05:00 PM

The first wave of genome-wide association studies has largely been confined to "big-name" diseases-- things like diabetes, heart disease, breast cancer, etc. There's a financial reason behind this, of course-- funding agencies like the NIH are most interested in diseases of major public health import, as are companies like DeCode. But in the next few years, there's no doubt it will become clear that any phenotype is amenable to this sort of genetic dissection. Genome-wide association studies are an important new tool in the biologist's toolkit, and it's worth noting that genetic data is (or will be) much richer in humans than in any other organism.

A couple new papers add towards what could eventually be a detailed understanding of the genetics of human phenotypic variation: first, a genome-wide association study of "restless leg syndrome", an ill-defined, heterogeneous disorder. The authors describe it thusly:
Nightwalkers, as individuals with RLS call themselves, are forced to move their legs during periods of rest especially in the evening and night to relieve uncomfortable or painful sensations in the deep calf. This diurnal variation leads to impaired sleep onset, and the periodic leg movements during sleep in the majority of patients contribute to sleep disruption and a reduced quality of life as a major consequence
The association study identified three risk factors near genes about which, as tends to be the case in these studies, very little is known. However, these are leads that will be immediately followed up, and likely with great impact.

Second, see this GWA study of "gallstone disease". Like restless leg syndrome (which has a prevalence of around 2-3%), this is hardly a rare phenotype-- the authors give the prevalence as between 10 and 20% of the population in industrialized nations. Though characterized as "diseases", both of these phenotypes lie within the range of normal human variation.

As prices on this sort of technology drops, the most interesting results will not necessarily come from the big genome centers, but rather from the people that choose to study interesting phenotypes. There's plenty of low-hanging fruit to be picked...

Labels: ,

Wednesday, July 18, 2007

New Pritchard paper, Adaptive evolution of conserved non-coding elements in mammals   posted by Razib @ 7/18/2007 10:10:00 AM

Jonathan Pritchard has a new provisional paper, Adaptive evolution of conserved non-coding elements in mammals, in PLOS Genetics:
Conserved non-coding elements (CNCs) are an abundant feature of vertebrate genomes. Some CNCs have been shown to act as cis-regulatory modules but the function of most CNCs remains unclear. To study the evolution of CNCs we have developed a statistical method called the 'shared rates test' (SRT) to identify CNCs that show significant variation in substitution rates across branches of a phylogenetic tree. We report an application of this method to alignments of 98,910 CNCs from the human, chimpanzee, dog, mouse and rat genomes. We find that 68% of CNCs evolve according to a null model where, for each CNC, a single parameter models the level of constraint acting throughout the phylogeny linking these five species. The remaining 32% of CNCs show departures from the basic model including speed-ups and slow-downs on particular branches and occasionally multiple rate-changes on different branches. We find that a subset of the significant CNCs have evolved significantly faster than the local neutral rate on a particular branch, providing strong evidence for adaptive evolution in these CNCs. The distribution of these signals on the phylogeny suggests that adaptive evolution of CNCs occurs in occasional short bursts of evolution. Our analyses suggest a large set of promising targets for future functional studies of adaptation.

Interestingly, there another paper another paper out in PNAS which speaks to the possibility of non-coding genomic regions in humans having functional significance.


Tuesday, July 17, 2007

The Elf vs. Cosma   posted by Razib @ 7/17/2007 09:47:00 PM

I just got an email that The Elf has decided to take aim at Cosma Shalizi's posts dismissing g & the heritability of IQ. Now, Cosma is an acquaintance of mine and we share some common interests, but here I hope the Elf and her kindred give him hell! Any GNXP readers with knowledge and spare time cycles should left the Elvenkind a helping hand.

Related: Tyler Cowen & Arnold Kling on IQ. And Michael Stansty too.


Two Books by John Emerson Published   posted by John Emerson @ 7/17/2007 06:43:00 PM

I've just self-published two books which might be of interest to a few people here. They mostly represent my literary side, one of them even being a book of poems I wrote 25+ years ago, but some might be interested in three sections of Substantific Marrow: "The Back Door of Europe" (on the Baltic-Black Sea corridor), "The State", and "Love or Money". Substantific Marrow can be bought for about $17, or $3.75 for an e-book. (YMMV because of state taxes and shipping costs).

More information here: http://www.idiocentrism.com/le%20real.htm

Books can be bought here: http://stores.lulu.com/emersonj

By Christmas I should have a third book out, about philosophy, economics, and temporality. Sometime next year my book on Inner Eurasian history should be out; this book should be of interest to many people here. I'm going to be spending the next several years gathering and finishing up stuff I've been working on since about 1985.

Steve Jones Article   posted by DavidB @ 7/17/2007 04:22:00 AM

Here is an article by Steve Jones in today's Daily Telegraph (the most 'conservative' of the quality daily newspapers in the UK). It is mainly about the application of evolutionary theory in engineering and medicine. Nothing very new, but may be useful as a reference.

Monday, July 16, 2007

Making sure good science doesn't go bad   posted by amnestic @ 7/16/2007 09:08:00 PM

I picked up a copy of Cerebrum 2007, a collection of essays related to neuroscience published by Dana Press. There were a couple good articles on stroke and pain (arguing that cancer patients have little danger of addiction even using infamous opioids like morphine), but I was disappointed by Henry Greely's scaaarrry article about the potential negative uses of neuroscience discoveries. He catalogs past misuses of scientific authority, perhaps the most egregious of which was the widespread use of lobotomy. But what to make of this:

Before eugenics disappeared in America after World War II, about 60,000 men and women were surgically sterilized by court order, for conditions such as feeblemindedness, alcoholism, insanity, epilepsy, and criminality, which have little or no genetic basis.

Alcoholism? Really? Somebody better tell the NIAAA quick.

In response to the overwhelming evidence from twin, family, and adoption studies for a major genetic influence on vulnerability to alcoholism, NIAAA has funded the Collaborative Studies on Genetics of Alcoholism (COGA) since 1989, with the goal of identifying the specific genes underlying this vulnerability.

When it comes time to argue about whether fMRI lie detection schemes should make their way into court, let's hope we can get our facts straight at least. As to the actual ethical questions raised, as in coerced fMRI or drug treatments I find my utilitarian standbys running up against my empathy. I suppose if you are doing nothing criminal and have no expectations of running up against the government you should have no concern about efficacious lie detection. On the other hand, I already think we have unjust laws on the books.

Clark's Survival of the Richest meets Mokyr's Industrial Enlightenment   posted by Herrick @ 7/16/2007 12:03:00 AM

Greg Clark's broadly Darwinian explanation for the Industrial Revolution, A Farewell to Alms, should be hitting bookstores in the next few weeks. As previously mentioned, Clark argues that in pre-Industrial Britain, the "survival of the richest" was the norm, and so today's Brits are largely the descendants of past elites. He says this process created a population that was more patient and better at abstract thought (he doesn't use the term "intelligence"). Clark has just enough data backing up his story to spur others to do more and better tests.

To be sure, Clark hedges his bets on how much of this evolution is cultural and how much genetic. But he clearly puts both possibilities on the table, a brave act for any social scientist. With just a little luck, the book will have a major impact.

So, how to prepare for all this? By reading some of Joel Mokyr's recent work, of course! His Lever of Riches is a wonderful, theoretically-informed-yet-historically driven tale of global (mostly European) technological progress. It's always a source of rich new insights.

But Mokyr's newest book, The Gifts of Athena: Historical Origins of the Knowledge Economy, speaks more to Clark's point. Mokyr's focus has always been on technology--not just on ideas per se, and not just on economic output per se, but on the interaction between the two. He argues that Western Europe's "Industrial Enlightenment"--the rise of methodical, scientific thinking many decades before the Industrial Revolution--was both a prerequisite for the Industrial Revolution and an elite phenomenon:

A century ago, historians of technology felt that individual inventors were the main actors that brought about the Industrial Revolution. Such heroic interpretations were discarded in favor of views that emphasized deeper economic and social factors....It seems, however, that the crucial elements were neither brilliant individuals nor the impersonal forces governing the masses, but a small group of at most a few thousand people who formed a creative community based on the exchange of knowledge. Engineers, mechanics, chemists, physicians, and natural philosophers formed circles in which access to knowledge was the primary objective.

Where did that few thousand come from? Why then? Why there? Of course, the temptation is to throw around words like "culture," a word that sounds to me a lot like "multiple equilibrium" or "tipping point," expressions that feel good but (usually) mean little. With some luck, Clark's semi-Darwinian explanation for the Rise of the West will give us some tools to figure out where the world's first scientific community actually came from.

Sunday, July 15, 2007

Most phenotypic evolution is neutral   posted by agnostic @ 7/15/2007 05:22:00 PM

Still not convinced? See below the fold for some (tame but NSFW) evidence that phenotypic evolution tends not to have to do with survival and reproduction.

Data gathered from fieldwork in Brazil, Turkey, and Iran. Comparable data could not be found for blue-eyed, blonde-haired Finland. That probably doesn't mean anything either.


Saturday, July 14, 2007

The New York Times: Willful Gullibility in Furtherance of Ideology   posted by TangoMan @ 7/14/2007 02:07:00 PM

You would think that the New York Times reporters on the education beat could write an article without blatantly proclaiming their gullibility or willingness to engage in obfuscation but I see no evidence of any desire to avoid such sins, especially in this article on how school districts are pursuing diversity agendas by parsing students according to socioeconomic status. The Times reports:
The most ambitious effort and the example most often cited as a success is in the city of Raleigh, N.C., and its suburbs.

For seven years the district has sought to cap the proportion of low-income students in each of the county's 143 schools at 40 percent.

To achieve a balance of low- and middle-income children, the district encourages and sometimes requires students to attend schools far from home. Suburban students are attracted to magnet schools in the city; children from the inner city are sometimes bused to middle-class schools at the outer edges of Raleigh and in the suburbs.

The achievement gains have been sharp, and school officials said economic integration was largely responsible. Only 40 percent of black students in grades three through eight in Wake County, where Raleigh is located, scored at grade level on state reading tests in 1995. By the spring of 2006, 82 percent did.

What's changed in the education landscape betweeen 1995 and 2006? We've seen the introduction of the NCLB which has created a massive incentive for States to game their achievement tests in order to meet Federal compliance thresholds. Is there any evidence that North Carolina has gamed their achievement tests in order to boost the proportion of students who met proficiency standards? You bet there is. Unlike the achievement tests that states devise the National Assessment of Educational Progress (NAEP) is a uniform measure not subject to "dumbing down." According to this national survey, North Carolina is ranked ninth in 4th grade reading score variance, and third in terms of 8th grade math score variance, between NAEP scores and state developed assessment tests.

If we go directly to the NAEP Reading statistics we see the results for the 4th graders in 1994 was 214 (212 national ave.) and that the 2005 scores were 217 (217 national ave.). The 8th graders 1998 scores were 262 (261 national ave.) and the 2005 score was 258 (260 national ave.) Here we see only miniscule gains across the years and a slippage against national trends. While it's certainly possible that the Raleigh schools are bucking the statewide trend the reporters don't verify that this is indeed the case and simply rely on meaningless statistics offered up in support of the educational fad du jour, that socioeconomic integration will produce the diversity that will finally close the achievement gap.

Instead of touting the success of the Raleigh schools in meeting proficiency standards because of economic integration shouldn't the reporters have verified that they weren't reporting phantom gains attributable to a gaming of the testing regime?

Now that's a buff dog   posted by Razib @ 7/14/2007 03:49:00 AM

OK, is a picture worth a thousand words? I don't know, but this photo is of a whippet with the condition p-ter talked about a few months back. We've addressed the possibility of this mutant in humans as well. The Daily Mail has more photos.


Giant Lion-Eating Chimps   posted by DavidB @ 7/14/2007 03:16:00 AM

Not an April Fool hoax, but apparently an entirely serious story in the Guardian. Actually, one sighting of a chimp nibbling the carcass of a leopard doesn't quite add up to 'lion-eating', but the facts as reported are impressive enough. Read it and make up your own mind.

Friday, July 13, 2007

Finnish geek culture   posted by agnostic @ 7/13/2007 12:26:00 PM

At the risk of boring readers with two Finn-baiting posts in one day, let me add two more data-points to the post below on losers from Finland. Some of the readers may have already known that Finland has the highest per capita membership for MENSA, the club that only lets in those with an IQ of about 132. Since they are not particularly smarter on average, this means that Finland is the country where smart people have absolutely nothing to do with their talents and instead waste their time attending MENSA gatherings. *

As another indication, the most elite guy at mastering insanely difficult stunts in the shooter video game Goldeneye is Ilari Pekkala, a Finn living in Sweden. Check out some of these videos in which the game has been pushed to maximal difficulty: one shot kills you, the enemies have 100% accuracy, they react speedily, and it requires 10 shots to the head to kill them (and more if you don't shoot their head). Just figuring out how to complete a level under these constraints requires lots of research, and actually accomplishing the stunt requires still more practice. Call him the video game version of the "White Death" sniper.

Looking over the rest of the above website, as well as this list of record-holders, it seems that Scandinavian and Germanic people are overrepresented among first-person shooter geeks. (And don't forget the most famous video of a gaming geek losing his temper is of a German kid.) I thought for sure there would be plenty of Northeast Asians, but there's nary a one to be found. Perhaps they're more interested in playing Tetris. Or maybe they've followed a more practical and less glory-seeking path by earning gold in World of Warcraft to sell to lazy Western players.

* While reading through a textbook on abstract algebra, I'm struck by how many Scandinavian names appear prominently -- Abel, Lie, Sylow -- since most math names tend to be German, French, Russian, or Ashkenazi Jewish. Although Scandinavians made a good showing among Fields Medalists early on -- a Swede living in Finland, a Swede, and a Norwegian -- the most recent winner won the medal in 1962. Get out of those damned MENSA meetings and video game competitions and do something new!


Thursday, July 12, 2007

Suomalaiset ovat losers   posted by Razib @ 7/12/2007 11:48:00 PM

Another loser from Finland:
Finnish filmmaker Aki Kaurismaki concludes his self-proclaimed "Loser's Trilogy" with the thoroughly deadpan drama "Lights in the Dusk." After "Drifting Clouds" and "The Man Without a Past," which focused on joblessness and homelessness, respectively, "Lights" takes on loneliness, an affliction that plagues a none-too-bright Helsinki security guard named Koistinen (Janne Hyytiainen).


Baron-Cohen on Autism   posted by DavidB @ 7/12/2007 05:01:00 AM

The UK Times today has an interview with Professor Simon Baron-Cohen on autism. Baron-Cohen is cautious about the supposed increase in the incidence of autism, and believes it could be largely due to 'diagnostic practice'.

Judging by the absurdly vague and loose criteria listed at the bottom of the article, I'm surprised anyone (except the jolliest of extroverts) escapes being diagnosed (and stigmatised) with some variety of autism. Thank Dawkins there was none of this nonsense when I was a kid. The whole thing looks to me like a monstrous boondoggle for the benefit of doctors and therapists.


Wednesday, July 11, 2007

Ultimatum elsewhere   posted by Razib @ 7/11/2007 01:08:00 PM

Jake & Kara @ Pure Pedantry have some interesting comments on the recent Ultimatum game & testosterone paper.


Tuesday, July 10, 2007

Judaism, a religion and a people   posted by Razib @ 7/10/2007 10:47:00 PM

Some of the comments below stumbled upon a topic which I think is important to highlight: that many Americans Jews who consider themselves religious wouldn't really be considered religious in a non-Jewish context. Most of you likely know that Judaism is both a religion and a nation (ethnicity). This isn't that exceptional. Some Zoroastrians present their own faith in the same way, as do many Hindus and some neo-Pagan reconstructionists (e.g., the Asatru). But in an America framed by Protestant assumptions about the centrality of religious profession, confession and adherence to a creed, when someone asserts that they are of a religion x the background schema of Americans is to assume that that entails adherence to a set of belief axioms. But check out some of the data from the American Jewish Identity Survey:
Belief that God performs miracles

Disagree strongly Disagree somewhat Agree somewhat Agree strongly
Jews by religion 11% 17% 31% 36%
No religion 7% 11% 31% 47%

I am contrasting Jews who have some religious affiliation (as opposed to avowedly secular Jews who have only an ethnic attachment) with those Americans who claim "No religion" (which can range from New Age theists all the way to materialist atheists). The point is that even religious Jews in the United States are in many ways more like the non-religious than they are like other denominations. This is reflected in their political liberalism as well as their hostility toward manifestations of "fundamentalist" religion (e.g., Creationism).


Evolutionary psychology and its "truths"   posted by Razib @ 7/10/2007 08:46:00 PM

Over the years I've become more skeptical of evolutionary psychology, or, more precisely the school of thought which follows in the wake of Leda Cosmides and John Tooby's collaborations with a host of researchers in the 1980s. Look over the Ten Politically Incorrect Truths About Human Nature. Lots of interesting hypotheses and conjectures, but some of the assertions they present as if they are established fact. There is unfortunately a reality to the "Just So" criticism of evolutionary psychology. The problem is, I think, a superficial comprehension of evolutionary biology. As an analogy: imagine that engineering was based upon conceptual physics as opposed to calculus based physics. You would have a "general sense" of the arc of a parabola, but you wouldn't have a mathematically precise picture so where the shot lands is still vague conjecture and intuition. Cosmides & Tooby made evolutionary psychology respectable again (after the sociobiology wars of the 1970s) by focusing as a matter of doctrine on human universals, but alas I fear that the ideological straight-jackets which the field started with (which one of the authors of the above piece, Satoshi Kanazawa, does not seem impacted by at least) resulted in a certain intellectual laziness. For a generation the criticisms of evolutionary psychology were directed from the corner of screaming ideologues; a natural response is to assume that most criticisms can be rejected as partisan and specious. But criticism and skepticism are essential parts of normal science.


The demographics of top law professors   posted by Razib @ 7/10/2007 04:08:00 PM

Via a reader I found a Imbler Volokh post on the demographics of top yong law professors. You can see his spreadsheet here. Volokh says:
If you thought the sex picture was hard to explain, try this: If you look at the same top 50 most-cited who entered law teaching since 1992, you also see that (by my rough count, and judging by likely ethnicity, not by religiosity) 19 are Jews [38%], a group that makes up 2% of the full-time working population. Part of this is the wild overrepresentation of Jews generally among the legal professoriate, a number that itself is hard to explain - Jim Lindgren's tentative survey from several years ago reported that 26% of law professors at top 100 law schools were Jews - but the numbers exceed even that.

Another 12 are Asians (meaning East or South Asians), a group that makes up 4% of the full-time working population. If you separate out South Asians (since in many ways it's just zany to lump Indians together with Chinese, or for that matter to lump together Chinese, Koreans, and Japanese), you'll find that 5 of the top 50 are South Asians, though South Asians make up 2/3 of 1% of the population. I don't recall precisely what fraction of the legal academy is Asian, but my recollection is that the fraction is no more than 5%, and thus far less than the 24% (or 10% for South Asians).

I generally track data about the medical and the scientific profession because I have friends in these fields (e.g., around 1 out of 20 to 1 out of 10 medical students in the USA currently is probably of South Asian ancestry). I didn't know much about law, it seems a field where cultural fluency matters much more than in medicine or the natural sciences, so I assumed that persons of Asian ancestry wouldn't be well represented.


Selection Within the Neutral Zone   posted by Matt McIntosh @ 7/10/2007 11:06:00 AM

(Achtung: I'm mostly thinking aloud here and haven't extensively explored these ideas. Just something I thought I'd throw out there.)

Stretches of DNA can get themselves replicated in one of two ways: They can help an organismal vehicle reproduce more effectively, or they can hijack an organism's transcriptional and replicative machinery to get themselves copied. In the latter case, if they travel autonomously they're called viruses and if they quietly hitch a ride in the DNA of some host organism they're called selfish DNA. Replicating inert pieces of selfish DNA takes up resources, but the cost of the marginal copy is low enough that quite a few can accumulate -- over 40% of the human genome, for instance, is estimated to be made up of retrotransposons.

But as long as the cost is non-zero, there'll still be a kind of fitness optimum for any bit of selfish DNA -- one that was too adept at getting copies of itself spliced into the genome would eventually get to the point where the host organism would start taking a fitness hit, and one that wasn't adept enough would get outcompeted by ones that did a better job.

How can selfish DNA elements be said to compete, if they're neutral by assumption and hence will just be blown around randomly by drift? While it's true that at any single locus the dynamics of selfish DNA elements will follow straightforward drift dynamics, retrotranspositional elements are special because identical copies of them can exist at many loci. To quantify this, at any single locus a string of selfish DNA's odds of sweeping are 1/N, and that goes for all its competitors too. But its expected score in the multi-locus game will be k*(1/N), where k is the number of loci where copies of it are present. It now becomes a game of "whoever has the most loci wins": if you're a retrotransposon with 500 identical copies of yourself on a genome, you've got an absolute advantage in the replication game over a different one that only has 100 copies of itself.

What retrotransposons are competing for in this case is the total excess carrying capacity of a population's genome -- i.e. the finite amount of room to expand your numbers before your hosts start getting hurt -- and this will vary from species to species. Note that all other things equal, a bit of selfish DNA that gets in early in the game will tend to profit more than a similar one that comes in late. But also note that all other things may not be equal: if population size varies over the course of the game, then the expected payoff for a new piece of selfish DNA that enters the game will be greater when N is lower, since its odds of sweeping to fixation are better.

But what happens when the carrying capacity has been filled? Since the dynamics at all the loci will be governed by mutation and drift, the equilibrium in this game is that nobody really wins because in the long run very few (none?) of the loci will be identical by state. [1] But in the long run we are all dead, and mutation+drift acts slowly enough that it would take a very long time to reach this equilibrium. A string of selfish DNA that controlled many loci would "outlast" one that only controlled a few, since even if another sweep occurred by chance at one of its loci it'd still be dominating a whole lot more.

In the mean time, the circumstances could change: Mutations at other loci or changes in the organism's environment could increase the carrying capacity of the genome (creating more room for expansion), or conversely the organism could evolve some mechanism for recognizing and snipping out (or suppressing replication of) the most common lineages of selfish DNA, clearing out room for an uncommon variant to take up the newly-freed space. [2] In this latter case we could hypothetically see a discontinuous sort of negative frequency-dependent selection, with variants that were "too successful" getting periodically wiped out.

I just thought this up as a way of demonstrating how a form of natural selection among replicators can occur even conditions where the loci involved are all effectively "neutral" from the organismal point of view. I have no idea whether something this has actually happened, but it seems quite possible in principle, unless I've gotten something terribly wrong somewhere.

[1] There is a way to escape this endgame, however: A bit of selfish DNA could mutate into something useful for its organismal vehicle, thereby biasing its own odds of being replicated. But in that case it's nolonger playing the drift game and can be modeled by the standard single-locus selective equations that we all know and love.

[2] Remember that the assumption here is that the marginal locus is neutral; but if the same string of selfish DNA was present in a large amount of loci, the total fitness hit to the organism could in principle make it worthwhile to evolve some mechanisms to keep the parasitic DNA under control.

Monday, July 09, 2007

Lead, Crime, and a decent newspaper article on methodology   posted by Herrick @ 7/09/2007 09:50:00 PM

Was the drop in crime causes by a drop in lead exposure? This Wash Post piece does a nice job reviewing a few recent studies on this issue, focusing mostly economist Rick Nevin's cross-country work. From the abstract to his key paper:

This study shows a very strong association between preschool blood lead and subsequent crime rate trends over several decades in the USA, Britain, Canada, France, Australia, Finland, Italy, West Germany, and New Zealand. The relationship is....consistent with neurobehavioral damage in the first year of life and the peak age of offending for index crime, burglary, and violent crime....Regression analysis of average 1985-1994 murder rates across USA cities suggests that murder could be especially associated with more severe cases of childhood lead poisoning.

I'm not qualified to address the medical angle, but the statistical discussion in the Post article is solid. I was pleasantly surprised to see a decent lit review that noted the inclusion of control variables and decent econometric identification methods in a few different studies. The key to Levin's statistical method, of course, is that the changes in lead exposure across time and across countries have to be uncorrelated with any omitted variables that might also impact crime. Any takers on the bet that some other variable can knock out lead's statistical significance?

For a short article, the journalist, Shankar Vendantam, does a great job. He even finds a hook to hang it all on: Giuliani's claim that his policies singlehandedly caused the drop in NYC's crime. If a political dig is the price of solid MSM science coverage, then so be it!

Oh, and for what it's worth, this line of research is just another example of how low IQ can have a social multiplier effect: Low IQ--> Low income--> Poor neighborhood--> Lead exposure--> Crime--> Poor neighborhood becoming bad neighborhood. Surely, there are (many) other links running from low IQ to bad social outcomes, but an anti-lead policy can weaken one of those (many) links.

Genetic determinants of "evolvability"   posted by p-ter @ 7/09/2007 06:43:00 PM

In comments on previous posts, I've alluded to the "mutational target size" of a phenotype-- that is, the number of locations in the genome that affect (or could, in theory, affect) the phenotype in question. If one imagines mutations as darts thrown randomly at a dartboard, the "target size" analogy is self-explanatory. There are a number of reasons why different phenotypes could have different target sizes-- the number of genes underlying the trait is likely an important parameter, and it seems intuitive to suggest that some phenotypes would be selected for robustness (ie. a smaller target size).

Data in this area is hard to come by. Luckily, expression microarrays have put at our disposition the ability to assay thousands of phenotypes-- gene expression profiles-- in parallel. A recent paper takes a look at this question through the use of mutation accumulation lines in yeast. I've mentioned MA lines before-- essentually, they're lines propogated with an extremely small effective population size, such that selection plays a nearly negligible role in the fate of new mutations.

The authors propogated a number of these lines for 4000 generations, then assayed gene expression to see which genes had diverged in expression, and by how much. The results are somewhat intuitive-- they find that gene expression evolvability is correlated with trans-mutational target size (as judged by expression profiling of knockout strains) and the presence of a particular promoter sequence. This is a small step in out understanding of regulatory evolution, but an important one. They conclude:
We show that not all genes are equally sensitive to the effects of random spontaneous mutations and identify structural properties (presence of a TATA box and trans-mutational target sizes) that greatly influence a gene's potential to undergo regulatory change. These determinants provide a mechanistic basis to serve as a foundation for more-realistic models of gene expression evolution that account for levels of polymorphism and divergence in cis and trans gene regulation.

Labels: ,

Sunday, July 08, 2007

Human variation and the ultimatum game   posted by Razib @ 7/08/2007 11:36:00 PM

A few days ago I posted on results which suggest a positive relationship between high testosterone and rejection of low offers in non-iterated "ultimatum games." I just noticed that the paper is open access, so I suggest you read it yourself. I will note two things:

1) The subjects were Harvard students who had taken at least two semesters of microeconomics. I don't know if offering this sample of students $40 is a good model for a realistic experiment (the typical Harvard student is rather affluent).

2) Though not statistically significant (the N was small) the authors did find that high testosterone males also offered more than low testosterone males. This might relate to the rather inexplicable generosity that alpha males in tribal bands exhibit when it comes to sharing out the proceeds of the hunt.


Talks and podcast   posted by amnestic @ 7/08/2007 07:12:00 PM

The Henry Stewart free 'talk of the month' in on vascular tissue engineering.
Talk Summary

Tissue engineering approaches – Different cell sources for a blood vessel substitute – Circulating endothelial progenitor cells: their characteristics, their use in a blood cell substitute and recruitment from the host circulating blood – Differentiation from ES into endothelial cells in vitro and their endothelial-like characterization – The use of different types of scaffolds – in vivo and in vitro remodeling – Engineering immune acceptance – Clinical applications: EC-seeded ePTFE grafts for peripheral applications and cell-seeded polymer scaffold grafts for pulmonary artery/Fontan procedure.

Cell has a new podcast.
June Podcast

In our third podcast appearing online on 14th June 2007, Dr Emilie Marcus talks to Dr Jonathan Weissman about exciting new technologies that are changing the research enterprise and we hear from Dr John Kuriyan about how kinases keep themselves switched off until they are needed. We also learn about some of the exciting research published in Cell in the last few months, including a study by Dr Wolf-Dieter Schubert that shows how to coax a human pathogen to invade a mouse.


Loss of Function is Adaptation   posted by Matt McIntosh @ 7/08/2007 01:30:00 PM

Over at his other blog Razib talks about the selective pressures that shaped the modern distribution of skin color. One thing he didn't emphasize but which I found illustrative is that loss of function plays a prominent role in this story as a special sort of adaptation: sometimes losing something can be good. But this raises a minor problem for interpreting some kinds of tests of selection: When you run, say, the McDonald-Kreitman test on MC1R in Europeans, the verdict returned is "neutral". But at the same time, the loss of function that turned this locus nearly-neutral is an adaptation!

The fact that a locus has faded into the nearly-neutral background isn't evidence against adaptation -- quite the opposite, in fact. Nothing is metabolically costless, so if the cost is significant and the benefit nolonger exists then there'll be selection for loss of function. Relaxation of constraint is just one way of shifting the fitness peak, and natural selection will respond to that as it always does. This is a convenient example of the general principle that you can't just take the outputs of canned statistical tests at face value: They require interperetation in the light of theory and history.

Mendel's Garden #16   posted by Razib @ 7/08/2007 11:47:00 AM

Eye on DNA has this week's Mendel's Garden.


Saturday, July 07, 2007

Straight from the mathematical mind   posted by p-ter @ 7/07/2007 10:15:00 PM

Terrence Tao, recipient of a 2006 Fields medal, has a blog. I found this post on compressed sensing and image compression to be rather illuminating.


The Aga Khan, the light of Islam?   posted by Razib @ 7/07/2007 09:13:00 PM

Since 9/11, and more broadly since the eruption of Islamic terror as a significant phenomenon in the past few decades, the media has attempted to impart to the general public information about Islam. Unfortunately, the media just plain sucks. Consider this article, Do Business and Islam Mix? Ask Him. "Him" refers to Aga Khan IV. The article suggests that he is the leader of the Ismaili Muslims, but more properly he is leader of about half the Ismaili Muslims in the world, the Nizari branch. For a detailed exposition of the multitudinous sects of Ismailis I recommend Mullahs on the Mainframe: Islam and Modernity Among the Daudi Bohras. Though focused on the Daudi Bohra sect, there is a detailed history in the first third of the book which relates how Ismaili sects differ from each other, other Shia, and Sunni Islam. The distinctiveness of Ismaili Islam is important to emphasize: to simplify they are the Shia of the Shia, a minority within the minority. Additionally, the Nizari Ismaili in the West have been very keen to acculturate to the norms of the surrounding population, even more so than other Ismaili groups, who remain more distinctive and recognizably "Muslim." So you have here a situation where the leader of a small sect of Muslims is held up as an Islamic exemplar from which we can learn about the faith. Though I do not deny that the Nizaris are Muslims, I think it must be emphasized that on a character-by-character basis of comparison they are very distinctive and peculiar as Muslims go. It is a fact that many Sunnis will even reject that Ismailis are Muslims at all (though with far less force than against the Ahmadiyya sect). As an analogy, imagine that the leader of the Seventh Day Adventists was being treated as a seminal and fundamental authority within Christianity. Finally, the article states that the Aga Khan is a "moderate Muslim." Moderate implies that someone is within the broad middle central tendency of a distribution, and that is simply not correct in the case of the Aga Khan and his followers. They lay at a tail. The mainstream media isn't really biased as much as it is mediocre, forgive them for they know not what they write, but also remember to be very cautious when they take any steps away from first order reportage.


Williams syndrome in The New York Times Magazine   posted by Razib @ 7/07/2007 06:50:00 PM

David Dobbs has an interesting article in The New York Times Magazine about Williams syndrome; a disorder characterized by verbosity and hypersociality in concert with abstraction capacities so attenuated that most suffers are mentally retarded. The piece juggles many phenomena, from general to domain specific intelligences and the interaction between environment and genetic biases which shape the mind's developmental arc.

Inverted: Hidden Smarts: Abstract thought trumps IQ scores in autism:
There's more to the intelligence of autistic people than meets the IQ. Unlike most individuals, children and adults diagnosed as autistic often score much higher on a challenging, nonverbal test of abstract reasoning than they do on a standard IQ test, say psychologist Laurent Mottron of Hopital Riviere-des-Prairies in Montreal and his colleagues.

Labels: ,

Genetic engineering in The New York Times   posted by Razib @ 7/07/2007 06:32:00 PM

Nick Wade has a short piece on "synthetic biology" (bioengineering) in The New York Times. When pieces such as these disappear from the science pages we'll know that biological engineering has arrived on the scene, the mundane and everyday is simply not newsworthy.


GNXP blocked by Websense   posted by Razib @ 7/07/2007 04:37:00 PM

From a reader:
I thought you might find this interesting and sad. I tried to log on to gnxp from work yesterday- web access was blocked by Websense (cisco) - reason- racial hatred. A pathetic sign of the times. Luckily I have a 3g capable smartphone so I was able to log on. But that really sucks. BTW instapundit was also blocked - reason - weapons. PC = damnation!


Jewish Genius, the follow up   posted by Razib @ 7/07/2007 03:37:00 PM

As most of you know Commentary has a series of letters & a response from Charles Murray following up his article titled "Jewish Genius." Most of the letters aren't particularly intelligent or sincere. Their general form is to dispute the possibility of intergroup differences and throw out all sorts of red-herring objections or insinuations of bad faith or ignorance, or alternatively to simply not understand evolutionary biology with any degree of depth so as not to be manifestly moronic. God knows how idiotic most of the letters must have been if these are the "good" ones. Greg & Henry have a response up to Charles' original article, and predictably I tend to agree with the basic outline. But, I feel I have to reiterate a point I made earlier in response to the original piece. Charles states:
As for the New Testament, unless one wants to argue that Jesus was not the source of his own teachings, and that Paul did not really play the role in formulating Christian theology that he appears to play, I do not see how one avoids concluding that the foundation of the New Testament's spiritual and intellectual power is Jewish.

The spiritual power is irrelevant to me, am I willing to grant that if there is a distinction to be made between pagan and Hebraic religiosity than the Christian religion has at its core a particular Hebraic sensibility. But I want to focus on the intellectual power, since that is the point of Charles' original article: systematic theology grounded in Greek philosophical traditions and methods was a Hellenic enterprise, not a Hebrew one. I say Hellenic specifically because the early Church theologians were generally culturally part of the Greek speaking intelligensia of the Roman Empire, not the Latinate administrative and legal class, let alone the Jewish minority. I've read the New Testament and broadly speaking there are theological elements to it, and certainly those were often exposited by Jews (whether the theology is fundamentally pagan or Hebraic in origin is irrelevant to me since Charles' contention is that Jewish intellectual brilliance is reflected in early Christian theology). But the shining intellectuals of the early Church tended to gentiles. For example, the Doctors of the Church are an eminent list recognized by the Roman Catholic Church. While the Latin intellectuals, such as Augustine or Ambrose, were eclectic and concrete in their thinking and intents, the Greeks such as Gregory of Nazianzus were more abstract and focused on the science of theology.

Myself, I don't find theology particularly interesting or insightful, but, I can't deny that a lot of intellectual firepower went into it and absorbing theological treatises require a level of close reading and attention which one does not associate with perusals of the Letters of Paul. If Charles means by theology the ideas exposited in the New Testament itself, I can grant that it is a very Jewish body of work in authorship (though not overwhelming). But I can not agree that the New Testament's theology is of high intellectual caliber, rather, its basic ideas are relatively accessible to all believers and the language is geared toward the modal intelligence (even the rhetorically elevated Gospel of John is intelligible to people of normal intelligence). If by theology one means the body of work which drew from Greek philosophical traditions, then I can grant that there is some intellectual heft and luminosity on display, but I can not grant that the authors are Jewish because we know the life backgrounds of many of the brightest lights and they were from gentile backgrounds, with a fair number converts from paganism (e.g., Augustine). As one would expect a priori from the division of labor during the Roman Imperial period, among the Christians the Greeks tended to focus on the more abstruse philosophical domains while the Latins were more worldly. This is all established history, so I can not see the point of controversy. Of course Charles can allude to the fact that the foundation of the systematic theology elaborated by the gentile intellectuals was Jewish. So what? Algebra and "Arabic" numerals were introduced into Europe from the Islamic Middle East, but we wouldn't discount the achievements of Europeans in abstract mathematics because the foundation of their abstractions was exogenous.


Sex-specific genetic networks   posted by p-ter @ 7/07/2007 02:43:00 PM

It's often assumed that the "wiring systems" of men and women, apart from the obvious differences in reproductive system, are essentially identical. And to a first approximation, that's of course true. But disparities in certain illnesses and quantitative traits suggest subtle differences in male versus female biology, even outside of obvious (i.e. reproduction) or controversial (i.e. neurobiology) areas.

Early last year, for example, a group of researchers took the simple step of dividing their genetic studies on a number of quantitative traits into males and females. Their results were striking: some traits are influenced by different genes in men and women. This is understandable if you consider a trait as being influenced by a network of interacting genes and proteins, along with the environment signals, as opposed to the simple Mendelian one gene-one trait framework. As men and women develop differently, these's no reason to expect that the "key" nodes in the network (with respect to the quantitative trait in question) will be the same in each. And in fact, this study demonstrates that for some traits, they're not.

A new study on liver cancer comes to a similar conclusion from another angle. The certain type of cancer they consider appears much more often in males than in females (in both humans and a mouse model). In their study (in mice, it should be noted) they find the application of the relevant carcinogen leads to increased levels of a certain inflammatory marker in both sexes, but that the increase is much more pronounced in males. Knockouts of the gene for the marker showed no sex differences in cancer rates, suggesting that the difference in response of a single gene in males and females could be responsible for the disparity.

How could such a differential response be mediated? It's hard to say, but I'd guess a subtle difference in transcription network architecture. Small differences in internal networks can lead to large differences in "higher" phenotypes.

Quick links   posted by p-ter @ 7/07/2007 01:43:00 PM

Robin Hanson at Overcoming Bias has a couple interesting posts up-- the first on a paper suggesting that conspicuous consumption in men and conspicuous "nice-ness" (my term) in women are "costly signals" in the sense of the handicap principle, and the second on a sociobiological theory for the demographic transition.

Labels: ,

Thursday, July 05, 2007

A top-down approach to genetic networks   posted by p-ter @ 7/05/2007 08:08:00 PM

One of the interesting findings to come from the recent burst of genome-wide association studies is that many seemingly disparate phenotypes share some genetic pathways. I don't think many people would have a priori considered the possibility of a genetic link between prostate cancer and type II diabetes, yet that's what the data suggest. Other links are somewhat more predictable-- type I diabetes and Crohn's disease both have an autoimmune component, so a genetic link might have been expected. And cornary heart disease and type II diabetes share some envronmental risk factors, so perhaps it's to be expected that similar genetic networks play a role in the two.

These genetic links have all been uncovered by classic reductionist methods-- find the molecular variation that predisposes to disease 1, find the molecular variation that predisposes to disease 2, and compare the two sets. It's simple, and it works.

However, a clever new paper takes a different approach:
Geneticists and epidemiologists often observe that certain hereditary disorders cooccur in individual patients significantly more (or significantly less) frequently than expected, suggesting there is a genetic variation that predisposes its bearer to multiple disorders, or that protects against some disorders while predisposing to others. We suggest that, by using a large number of phenotypic observations about multiple disorders and an appropriate statistical model, we can infer genetic overlaps between phenotypes. Our proof-of-concept analysis of 1.5 million patient records and 161 disorders indicates that disease phenotypes form a highly connected network of strong pairwise correlations. Our modeling approach, under appropriate assumptions, allows us to estimate from these correlations the size of putative genetic overlaps.
This is data mining at its finest.

I'll admit I didn't go through all of the hundreds of pages of supplementary material, but this is fascinating stuff. The authors seem to be particularly interested in autism, which they find correlates with a number of neurological disorders, but also bacterial and viral infections and autoimmune disease:
Our estimated significant overlap between autism and tuberculosis may indicate that both diseases are associated with genetic changes weakening the immune system
Or consider the overlap between bipolar disorder and breast cancer:
Although the competitive genetic overlap between bipolar disorder and female breast cancer has not been reported, there is recent indirect evidence that supports it: a well-established breast cancer drug, tamoxifen, was recently discovered to be effective in treating symptoms of bipolar disorder.
The amount of data generated by the medical community each day is staggering, and as genetic information gets cheaper, it will increasingly be a part of that data. Half the game is knowing what to look for.

Labels: ,

Final thoughts on adaptation   posted by p-ter @ 7/05/2007 06:44:00 PM

Larry Moran has responded again on adaptation, disapprovingly quoting Dawkins, who writes:
Natural selection is all-powerful with respect to those visible changes that affect survival and reproduction. Natural selection is the only explanation we know for the functional beauty and apparently "designed" complexity of living things. But if there are any changes that have no visible effect-changes that pass right under natural selection's radar-they can accumulate in the gene pool with impunity and may supply just what we need for an evolutionary clock.
Besides the possibly questionable use of the term "all-powerful", that sounds about right. The extraordinary power of the neutral theory to explain certain parts of molecular evolution cannot be denied (nor should anyone want to--it's quite elegant!). Yet the power of natural selection on the phenotypic level is becoming clear from molecular evidence, even if it wasn't already clear from simple observation (John Hawks provides a couple pictures to "frame" the debate, so to speak).

It's worth pointing out that evolution is stochastic-- most beneficial mutations are lost from the population immediately, and if we were to restart the evolutionary process multiple times, there would certainly be major differences in what shows up. But that's a very different process than genetic drift, which I think is perhaps the source of Larry's confusion.

In any case, Larry also has an interesting post up on mutation rates, in which he concludes:
With a population of 6 billion individuals on the planet, there will be 120 x 6 x 109 = 7.2 x 10^11 new mutations in the population every generation. This means that every single nucleotide in our genome will be mutated in the human population every 20 years or so.
Now that's something to think about.

Labels: ,

Variation as the ultimate   posted by Razib @ 7/05/2007 12:31:00 PM

Many of you have heard of the Ultimatum Game:
The ultimatum game is an experimental economics game in which two parties interact anonymously and only once, so reciprocation is not an issue. The first player proposes how to divide a sum of money with the second party. If the second player rejects this division, neither gets anything. If the second accepts, the first gets his demand and the second gets the rest.

In theory a "rational" player should accept whatever is offered when there isn't a repeated iteration. Reality is different. From The Economist:
...Those results recorded, Dr Burnham took saliva samples from all the students and compared the testosterone levels assessed from those samples with decisions made in the one-round game.

As he describes in the Proceedings of the Royal Society, the responders who rejected a low final offer had an average testosterone level more than 50% higher than the average of those who accepted. Five of the seven men with the highest testosterone levels in the study rejected a $5 ultimate offer but only one of the 19 others made the same decision.

What does this tell us? That physiological variables which are under biological (and ultimately genetic) control can affect the typical behavior a given individual exhibits, and, that that behavior can vary despite the same inputs across the population. There isn't any one H. economicus, there are many different ways humans interact and their propensity for a particular strategy might be conditional upon biological parameters.

But of course this doesn't mean that a given individual practices a fixed strategy even for the same inputs over time, just as strategies are mixed throughout the population so they are often mixed over time for any given individual. There is both population level and temporal variation which must be taken into account here; the flat uniform world of older economic imaginations were painted in shades of gray despite the multi-colored nature of reality.

Additionally, as I have noted before, even genetically close groups which are culturally distinct can exhibit wildly different modal responses to these various experimental economic games. This suggests that variation is not just extant on the biological level (e.g., tracking testosterone variation within the population), but also on the cultural level as the social parameters shift and reshape the landscape of gene-environment interaction. In other words, the behavioral economic biases can be likened to norms of response of particular genotypes in various cultural environments. Though the median value may shift, the distribution remains the same (e.g., if a particular individual is high testosterone it is likely that their response to the ultimatum game in one iteration will always lay at one end of the distribution across cultures though the range and shape of the distributions may vary quite a bit).

Reality is complex. I'm alluding here to the interaction of genetic parameters with various cultural norms. Additionally, the current work in relation to various small scale societies where the "nominal" sums offered by economists is non-trivial implies that analogical reasoning plays a strong role in determining how the typical individual will respond. It seems that most peoples don't conceive of utility maximization, they simply resort to analogies with transactions in their conventional life which can be mapped onto the games they are being forced to play. So there are innate parameters that result in a central tendency as well as variation, but there are also cultural parameters which modulate the range and constrain the scale, and, these often express themselves general intelligence operating through analogical (as opposed to deductive) reasoning. This turns rationality into a whole new beast altogether, not only is it bounded, it is nearly eviscerated as we understand it.

But why this variation in the first place? First, I am implying that the conditional responses that an individual gives has an expectation which is determined in large part by their genetic inheritance. Imagine for example that the ratio of "aggressive" to "passive" responses in a given game that an individual gives over time as a ratio, and that this ratio is placed upon a graph. I suspect that in many cases you would generate some sort of normal distribution (you might have to transform it though). There would be a median modal ratio; there would be those rare players who engage in "fixed" strategies where they were invariant. In this way you can re-conceptualize the behaviors documented in experimental economics as continuous quantitative traits. We know from population genetic theory that such traits have not been subject to powerful directional selection for long periods of time. Otherwise, the underlying genetic variation would have been exhausted as one behavioral morph comes to dominate the population of strategies (the range of basal testosterone should be very small and predominantly environmental/non-heritable). The reality of polymorphism might imply that the "rationality landscape" (to borrow a term) is characterized by multiple optima. Balancing selective forces such as frequency dependence and environmental variation might also result perpetuation of the mix. Layered on top of this evolutionary biological level is the flux of cultural inputs which serves as the background environment in which the predispositions develop into lifelong typical strategies. We've come a long way from reciprocal altruism.


Wednesday, July 04, 2007

Good Point?   posted by DavidB @ 7/04/2007 03:17:00 AM

Suppose that:

a) intelligence is partly inherited


b) less intelligent people on average have more children.

Does it follow that:

c) the average intelligence of the population will decline from one generation to the next?

A simple example, due to the statistician I. J. Good, shows that this is not necessarily the case. [Note 1]

Let H stand for High and L for Low intelligence (or any other trait). Let PG stand for the parental generation and O for offspring. (The parental generation should not be confused with parents. Not all individuals in the parental generation are necessarily parents.) Let the numbers of offspring in the categories H and L be as follows, with the numbers of the parental generation in square brackets:

[100 PGH] 28........18........ 46
[900 PGL] 72......882......954

This simple example satisfies assumptions (a) and (b), but not the conclusion (c). The proportions of H and L in the population do not change from one generation to the next. I will call any 2-class system with these properties a 'Good model'.

I draw attention to Good's little model because it does not seem to be widely known. For example, Richard Lynn has written an entire book, Dysgenics, on the threat of falling intelligence due to differential fertility, but does not refer to Good's article. Neither do Murray and Herrnstein in The Bell Curve. I came across it myself here, when I was searching for references to another article on the same theme.

That other article is by Samuel Preston and Cameron Campbell. [Note 2] Preston and Campbell present a far more elaborate and mathematically sophisticated model than Good. This is not because they are better mathematicians than Good (few people are!), but because they are aiming to give a realistic model of actual populations. Unfortunately it is so complicated that neither the authors nor their critics seem to understand what is going on. I think that Good's little model is more useful for analysing the essential issues. There is no advantage (unless the aim is obfuscation) in bringing in Markov chains, primitive matrices, eigenvectors, and the like, if the same points can be made with arithmetic and simple algebra.

Good's model shows that assumptions (a) and (b) do not imply conclusion (c). It does not follow that conclusion (c) is false. It may be that with additional reasonable assumptions, the conclusion can still be drawn. I discuss this in the continuation.

(Warning: this is a long piece aimed more at people searching for discussions of Preston and Campbell's theories than at casual browsers.)

[Added on 6 July: I should probably have said here, upfront, that ultimately I reject Preston and Campbell's model. I should also have emphasised that I. J. Good was only concerned to show that some loose arguments about intelligence are fallacious, whereas Preston and Campbell really believe that their 'equilibrium' model of intelligence is usually correct. But I should also have pointed out that there could be other ways of achieving equilibrium, so that rejecting Preston and Campbell does not necessarily mean rejecting equilibrium. Notably, many years ago the geneticist Lionel Penrose outlined a model in which heterozygote advantage maintains an equilibrium distribution of IQ despite greater fertility of the less intelligent. Roughly speaking, in his model the highly intelligent have low fertility, the less intelligent have high fertility, but the really stupid (who are homozygotic for 'stupidity genes') have very low fertility, which acts to purge 'stupidity genes' from the gene pool. (These are not Penrose's own terms!) As a twist to this which I have seen suggested somewhere, individuals who are homozygous for a lot of 'stupid' genes may have low fetal viability, so the 'purging' could be prenatal.)]

Good's Example Discussed More Fully

First, a few more comments on Good's own example. Good himself emphasises that his example is not biologically realistic, and his aim is only to show that certain loose arguments are invalid.

Good presents his figures in terms of fathers and sons, but this is not essential to the model, so I have referred more generally to parents and offspring. Reproduction within the model could be either sexual or asexual, and mating could be random, assortative, or disassortative. The only special problem arising with biparental inheritance is how to deal with offspring of one H and one L parent. The obvious solution in this case is to assign half of each offspring to the H parent and half to the L parent. It does not matter if this occasionally results in fractional numbers of offspring.

Good himself defines his two classes as 'Intelligent' and 'Unintelligent'. The conclusion that the distribution of intelligence is unchanged assumes either that these are uniform discrete values, or that each class covers a range of values with an unchanging mean.

Of course, a division of intelligence into just two categories, High and Low, is hardly realistic. Good himself describes it as 'much oversimplified' , but it would be superficial to reject the model on that account. The same essential features can be shown to apply in a model with any number of categories. (Effectively, this is what is done by Preston and Campbell.) There is no reason to doubt that at the limit it would apply to a continuous distribution.

In Good's figures the offspring of H parents are on average more intelligent than those of L parents, as required by assumption (a). For example, if we give H the value 1 and L the value 0, the average intelligence of the offspring of H parents is 28/46 = .61, while the intelligence of the offspring of L parents is far lower at 72/954 = .075. These values are based on the particular choice of H = 1, L = 0, but the rank order would be the same with any other values, provided H is greater than L. It is also interesting to see what would be the average intelligence of the offspring if H and L parents reproduced with equal fertility, which I will call the 'Equal Fertility Variant'. With Good's figures (assuming that the pattern of inheritance is otherwise the same) this comes out as .128. Setting out in order the mean intelligence of all relevant groups, we have:

High Parents: 1
Offspring of High Parents: .61
Equal Fertility Variant: .128*
Parental Generation: .1
All Offspring: .1
Offspring of Low Parents: .075
All Parents: .046**
Low Parents: 0

*this is the mean intelligence of all offspring on the assumption that H and L parents have equal fertility, and that the ratios of H and L among their offspring are as in Good's model.
**this is the mean intelligence of parents counting the parents once for each offspring. There are 46 offspring of H parents, with value 1, and 954 offspring of L parents, with value 0, so the average is 46/1000 = .046.

In Good's figures the offspring of both H and L parents are therefore more intelligent than the average of all parents, weighted by the number of offspring (not to be confused with the average of the parental generation), but those of H parents are more markedly so. I have calculated the correlation coefficient for intelligence between parents and offspring, and it comes out at the moderate figure of .37. [Note 3]

So far as fertility is concerned, L's in the parental generation are more fertile than H's, which satisfies assumption (b). Each individual in the PG L class has on average 954/900 = 1.06 offspring, while for PG H's the number is only 46/100 = .46, less than half that of L's. It is not possible to calculate a correlation between intelligence and fertility, since we do not know the variance of fertility within groups, but there is clearly a negative correlation; that is, high intelligence tends to go together with low fertility, and vice versa.

It will be noted that the offspring of H's are on average less intelligent and those of L's more intelligent than their own parents. This may be regarded as an instance of regression towards the mean. Using Good's figures I have calculated that the coefficient of regression of offspring on their parents is about .53. (This is higher than the correlation coefficient of .37 because the parents have a lower variance than the offspring.) The extent of regression towards the mean is paradoxically greater the lower the regression coefficient (one of the drawbacks of the established terminology), and a regression coefficient of .53 means that the offspring regress slightly less than halfway towards the mean. This is not an unreasonable figure.

It is more questionable that the average intelligence of all offspring (and even the offspring of L parents) is higher than that of all parents. It is also noteworthy that the average intelligence of offspring in the absence of differential fertility would be higher than either the average for actual parents or for the entire parental generation.

It is evident that if the same patterns of reproduction were to continue, generation after generation, then the proportions of H and L in the population would be unchanged, despite the continued greater fertility of L parents. The crux of the matter is whether it is biologically realistic to expect the same patterns of reproduction to continue indefinitely.

The General Properties of Good Models

But first I want to examine in more general terms the properties of Good models. There is nothing sacrosanct about Good's own figures. Here is a version with nice round numbers:

[500 PGH] 300....100.......400
[500 PGL] 200.....400.....600
Totals........500....500... 1000

This has the essential features that it is consistent with assumptions (a) and (b) but is a counterexample to the conclusion (c).

Using the values H = 1 and L = 0, the mean intelligences are as follows:
High Parents: 1
Offspring of High Parents: .75
Equal Fertility Variant: .542
Parental Generation: .5
All Offspring: .5
All Parents: .4
Offspring of Low Parents: .33
Low Parents: 0

With these figures the mean intelligence of the offspring of L parents is lower than that of all parents, unlike in Good's case. This shows that the contrary position is a peculiarity of Good's figures, not an essential feature of Good models.

It is natural to ask whether there is a systematic procedure for constructing Good models, and whether any general conclusions can be drawn about the range of possible parameters.

The essential structure of a Good model is a 2 x 3 table. There are 4 'cells' for the numbers of offspring and 2 for the numbers in the parental generation. In the absence of any other constraints each of the 6 numbers can be chosen independently of all the others. In technical terms the table therefore has 6 degrees of freedom.

Various other quantities can be derived from the 6 key numbers. Obvious examples are the row and column totals. Two other quantities of particular interest are the proportion of offspring who are in the same class as their parents, which I will call the Continuity Rate; and the ratio between the number of offspring and the number in the parental generation, which I will call the Fertility Rate. To satisfy assumption (b) the Fertility Rate of the L class must be higher than that of the H class. Good's own model also assumes that the total size of the population is constant. While this is not essential, it is convenient.

To construct a Good model we need to eliminate 6 degrees of freedom (DFs) and thereby fix values for the 6 key numbers. This could be done directly by inserting 6 numbers in the table, but usually we want to impose certain desired constraints, such as a given ratio between H and L. We therefore require 6 independent constraints to eliminate 6 DFs. The requirement of 'independence' can be tricky. For example, it might seem that if we insert all 4 row and column totals for the offspring, we have eliminated 4 DFs, which is sufficient to determine the 4 cell entries for the offspring. But in fact the row and column totals have only 3 DFs, since the sum of the row totals must equal the sum of the column totals. We therefore need to impose at least one additional constraint to obtain the cell entries.

Not all possible combinations of parameters have physically possible solutions. For example, if we stipulate that the Fertility Rate of the L class is 1.5, and that the Continuity Rate of L is .7, then the number of L offspring of L parents will be 1.5 x .7 = 1.05 times the total number of L's in the parental generation. It is therefore impossible to satisfy the equilibrium condition that the number of L's does not change from one generation to the next, since this could only be achieved if H parents had a negative number of L offspring, which is impossible. We cannot accept a solution in which any of the 6 key numbers in the table are negative.

Subject to these cautions, it is easy and mildly amusing to play around with different parameters. In each case we need to impose constraints in 6 steps to eliminate 6 DF's. In Note 4 I give four examples to illustrate some general points.

A Formula for Equilibrium

For those who do not want to play around with tables, the conditions for equilibrium can be summed up in a formula. I will use the following abbreviations:

N is the total population size, which we assume is constant.
H1 is the number of H individuals in the parental generation.
H2 is the number of H individuals in the offspring generation.
O'H is the number of offspring of H individuals.
R'H is the fertility rate of H, which we define as equal to O'H/H1.
A is the Continuity Rate of H (the proportion of the offspring of H individuals who are themselves H).
B is the proportion of the offspring of L individuals who are H (which equals 1 minus the Continuity Rate of L).

It follows from these definitions that the number of H individuals in the offspring generation is H2 = AR'H.H1 + B(N - R'H.H1). But the condition of equilibrium is that H2 = H1, so we have H1 = AR'H.H1 + B(N - R'H.H1). With a little rearrangement this can be expressed as H1 = BN/[1 - (A - B)R'H].

We therefore have a formula for calculating the equilibrium value of H1 for given Fertility Rates and Continuity Rates. From inspection of this formula it will be seen that the equilibrium level of H will generally be higher when the Fertility Rate of H is higher, and when the proportions of H among offspring are higher, though the balance between A and B is also significant.

General Conclusions about Good Models

To sum up some general conclusions about Good models:

a) equilibrium solutions exist for a wide range of parameters, provided they are not so extreme as to lead to negative numbers of offspring in any category.

b) if an equilibrium solution exists, it is unique (as there are no squares or higher powers in the equilibrium formula)

c) if the system is out of equilibrium, it will move towards equilibrium, provided fertility rates and continuity rates are constant from one generation to another. (See Example 4 in Note 4.)

d) it is a general feature of Good models that if the fertility rate of L is greater than that of H, then in equilibrium the average intelligence of offspring must be higher than that of their own parents. The reason is simply that if L parents are a higher proportion of parents than the proportion of L in the whole parental generation, then the average intelligence of parents is lower than that of the parental generation. But by the equilibrium assumption this is the same as that of all offspring, so the average intelligence of all offspring must be higher than that of their own parents.

e) If in equilibrium the fertility rate of L is greater than that of H, then the average intelligence of offspring under the Equal Fertility Variant would be higher than that of the parental generation. This is because under the EFV we would be increasing the proportion of H parents, whose offspring on average have higher intelligence. The average intelligence of the offspring will therefore be higher than under the scenario with differential fertility. But by assumption in this scenario the intelligence of the offspring is the same as that of the parental generation. Therefore under the EFV the average intelligence of offspring would be higher than that of the parental generation. The same principle can easily be extended to a system with more than two classes of intelligence.

Preston and Campbell

Having discussed Good's model at such length, I can deal much more quickly with Preston and Campbell's approach. (I have read their original article, but I will rely here mainly on a more accessible article by Preston [Note 5].) P & C's two main differences from Good are that they use a larger number of classes of intelligence (7 instead of just 2), and they allow explicitly for biparental inheritance. In principle, the father and mother can each come from any of the 7 classes, with a stated probability, so that there are altogether 7 x 7 = 49 possibilities to be dealt with. However, in practice P & C deal mainly with the simple case of assortative mating within the classes.

P & C appear to be unaware of Good's model, but Preston's article begins with a 2-class example which can easily be put into a Good-type table as follows (rounded to whole numbers):

[40 PGH] 22.....6.......28
[60 PGL] 18.... 54......72
Totals.....40.... 60....100

It will be seen that this has all the essential features of a Good model. I note in particular that, as in a Good model, the offspring under the EFV would have average intelligence higher than that of the parental generation.

In the more elaborate 7-class model P & C use the methods of matrix theory to reach some general theorems about the behaviour of the system over an indefinite number of generations. Their key assumptions are that fertility rates are fixed; that the probabilities of offspring falling into a given intelligence class depend only on the intelligence of their parents; and that these probabilities are the same in any generation. With these assumptions, they prove that the system will always result eventually in a constant equilibrium distribution. After our consideration of Good models, there should be nothing surprising in this conclusion.

Biological Objections

I now move on at last to consider whether a Good (or P-C) model is biologically plausible or even possible.

First, I deal briefly with a possible objection to such models based on a theorem of quantitative genetics known as the Breeders' Equation.

This can be expressed as:
R = h²S
where R is the response to selection, measured by the difference between the mean of the offspring of selected parents and the mean of the parental generation, h² is 'narrow heritability', as measured by the regression of offspring on parents, and S is the average deviation of the selected parents from the mean of the parental generation. [Note 6] If h² and S are both non-zero, then R cannot be zero. In a Good model it seems that h² and S are both non-zero, but that R is zero, since there is no difference between the mean of the offspring of selected parents and the mean of the parental generation. How can this be?

The short answer is that the derivation of the Breeders' Equation in its usual form makes various assumptions which are not satisfied in Good models. Notably, it assumes that the mean and variance of the offspring of unselected parents are the same as those of the parental generation. But in Good models this is not the case. If we want to use the Breeders' Equation for a Good model the response to selection needs to be measured in relation to what the mean of the offspring would have been in the absence of selection (in my terms, the EFV).

There remain several respects in which Good or Preston-Campbell models seem biologically implausible or problematic. I distinguish these as follows:

a) in equilibrium the intelligence of offspring is on average higher than that of their own parents;

b) the average intelligence of offspring under the EFV is higher than that of the entire parental generation;

c) the assumption, built into any model with a fixed number of values, that there are upper and lower limits to intelligence;

d) the assumption that the intelligence of offspring is influenced only by that of their parents, and not their more remote ancestry;

e) the assumption that the same patterns of inheritance can continue in equilibrium indefinitely.

I consider these aspects in turn.

a) As noted earlier, the offspring of H's are on average less intelligent and those of L's more intelligent than their own parents. This is not in itself unreasonable, as a result of regression towards the mean. In combination with the higher fertility of L parents, regression may well have the consequence that the average intelligence of offspring is higher than that of parents. For example, let us suppose that the offspring of L and H parents show the same average amount of regression towards the mean. It can be shown that if L parents are more fertile, the mean intelligence of the offspring will be higher than that of their parents, weighted by their number of offspring. To illustrate the process, suppose the population is divided equally into people 6 feet tall and 5 feet tall. Suppose that the offspring of the 6-footers are on average 5'10" tall, while those of the 5-footers are 5'2" tall. The extent of regression towards the mean, in units of height, is therefore 2 inches in both cases. Now suppose that the 5-footers have twice as many offspring as the 6-footers. For example, suppose there are 600 potential parents, of whom the 300 short individuals have 400 offspring, while the 300 tall individuals have 200 offspring. The mean height of the parents (weighted by the number of offspring) is [(400 x 5') + (200 x 6')]/600 = 3200'/600 = 5'4", while the mean height of the offspring is [(400 x 5'2") + (200 x 5'10")]/600 = 3233'/600 = 5'4.7". The average height of the offspring is therefore about three-quarters of an inch greater than that of the parents (counting each parent once for each child). Given the phenomenon of regression towards the mean, combined with differential fertility, it is therefore quite possible for point (a) to be satisfied. Of course, not every pattern of regression that is mathematically possible would be biologically realistic. However, I do not see anything biologically unreasonable about the patterns of regression required to satisfy point (a).

b) The fact that average intelligence of offspring under the EFV is higher than that of the entire parental generation seems to me a more serious difficulty with Good models (and the more elaborate Preston-Campbell versions). It is equivalent to saying that if everything else (all environmental and genetic conditions) were the same, but that each individual had one offspring (or each couple two offspring), then the average intelligence of the population would rise. This is surprising. Of course, if environmental conditions were to change, for example through improvements in nutrition or education, then a rise in average intelligence would be plausible, but the whole point of Good models is to show that the distribution of intelligence can be in equilibrium without such changes. It is remarkable that neither Good, Preston and Campbell, nor Preston and Campbell's various critics, seem to have noticed this paradoxical situation. In Preston and Campbell's case this is partly due to the complexity of their model, which obscures its underlying features, but also to the fact that they make use of empirical data derived from a study carried out in the 1960s. A 'Flynn Effect' increase in IQ is built into these figures, which makes them useless for exploring the general case where no such increase is in progress. I do not assert that point (b) is biologically impossible. There may be some scenarios in which, following cessation of selection, the average intelligence of the population would increase, for one or a few generations, until a new equilibrium is reached. The effects of genetic dominance and epistasis depend on the frequency of the relevant genes in the population, and on the systems of mating, so it is possible that a change in the frequency of certain genes or mating types, due to the ending of differential fertility, would raise average intelligence beyond that of the parental generation. But neither Good not Preston and Campbell appeal to any such phenomena, and I think the burden of proof is on the exponents of their models to come up with plausible genetic scenarios.

c) Julian Simon, in the note which draws attention to Good's little example, has stressed the role of a lower limit to intelligence in producing equilibrium. He points out that if there is such a lower limit, and if there is a non-zero probability that the offspring of low intelligence parents will have higher intelligence, then there will be an equilibrium level somewhat higher than the lower limit. I would agree that this is one way of producing an equilibrium, but I do not think it is the only way. While I have not worked out the details, I think it would be possible to have an equilibrium system, incorporating differential fertility, in which the intelligence of offspring is normally distributed (and therefore in principle with no upper or lower limits) around an appropriately regressing mean for each class of parents.

d) The assumption that the intelligence of offspring is influenced only by that of their parents, and not their more remote ancestry, cannot be strictly true in the case of dominant and recessive gene effects. Preston and Campbell recognise this as a weakness of their model but do not see it as a major problem. I am inclined to agree. While I don't have a proof, I think that the essential features of Good or P-C models could be found in a more complicated model where the intelligence of offspring is affected by more than two ancestors (i.e. the parents).

e) The most serious problem is the assumption that the same patterns of inheritance can continue unchanged indefinitely, despite continued differential fertility. If high and low intelligence were determined purely by environmental factors, this would be less surprising. It would be reasonable to suppose that the average quality of environments encountered by offspring is not affected by selection in previous generations. (I say 'reasonable', not certain, because one could imagine that environments deteriorate cumulatively as a consequence of declining parental intelligence.) In contrast, if high and low intelligence are determined partly by genetic factors, then they will be affected by the genetic composition of the population. This will change cumulatively as a result of selection. These changes will be permanent, in the absence of reversed selection (neglecting the small effects of mutation and genetic drift). If less intelligent parents are more fertile, then the frequency of genes associated with low intelligence will increase. Each such gene will contribute to a certain average expected value of intelligence (under constant environmental conditions). The effect of an increase in the frequency of genes for low intelligence should be to reduce the average expected intelligence of offspring to a lower level than would otherwise have been the case. One would therefore usually expect the average intelligence of the population to fall as a result of differential fertility, even if, as a result of regression towards the mean, the average intelligence of the offspring after selection is higher than that of their own parents. As discussed under point (b) it is conceivable [added: actually, I am not sure it is conceivable, but I concede the point for the sake of argument] that, due to some unusual genetic scenario, the average intelligence of the population may be sustained at its original level for a few generations, but this can hardly continue indefinitely if differential fertility also continues. There is an analogy with the heating of a liquid: as it reaches boiling point the temperature may remain constant for a certain period as heat is absorbed (so-called 'latent' heat), but if the heating continues the liquid will eventually begin to boil. Similarly, as the proportion of genes associated with low intelligence rises, individuals will be carrying an increasing number of genes with latent negative effects on intelligence. To take an example from eye colour, if parents with blue eyes (a recessive trait) have persistently higher fertility than parents with brown eyes, then the proportion of 'blue eye' genes in the population will steadily increase. Even though the offspring of blue-eyed parents will often have brown eyes (assuming random mating between blue and brown eyed parents), the probability of homozygosity for 'blue' genes will increase, and the probability that parents (whether blue or brown eyed) will have blue-eyed offspring will also rise. Initially the effect may be slight (if 'blue' genes are initially rare), but ultimately blue eyes would still drive out brown. The same principle would apply in the case of epistatic effects. The objection to Preston and Campbell's model is ultimately quite simple: by assuming constant patterns of inheritance, they are implicitly excluding the effect of changes in gene frequencies. They are therefore prejudging one of the main points at issue.

The surprising thing is not that Preston and Campbell's argument is flawed, but that the existence of some flaw was not obvious to the authors themselves. As Richard Lynn tersely commented, 'it [Preston and Campbell's article] attempts to prove mathematically that Darwin's theory of evolution does not work'. [Note 7] Why then did Preston and Campbell not themselves see that something must be wrong? I can only speculate about this. It may be partly due to an misplaced respect for mathematics: having shown by sophisticated methods that something is mathematically possible, Preston and Campbell gave too little attention to their underlying assumptions. But I suspect the main reason is in their disciplinary background, which is apparently in demography. Demographers have developed methods for analysing the transition between social or occupational classes. For example, the offspring of farmers are themselves often, but not always, farmers, and demographers have devised methods to analyse how occupational classes like farmers can be stable in size despite differences in fertility. Preston and Campbell have applied these methods to the case of intelligence. The assumptions they adopt may well be quite appropriate to demographic problems, where genetic inheritance is often either irrelevant or negligible. The same assumptions are misconceived when applied to traits where genetic heritability is (or may be) substantial.

Note 1: I. J. Good, "Fallacies, Statistical" in Encyclopedia of the Social Sciences, Vol. 5, p.292-300.

Note 2: Samuel H. Preston and Cameron Campbell, 'Differential Fertility and the Distribution of Traits: the case of IQ', American Journal of Sociology, 98, 1993, 997-1019, with comments by James S. Coleman and David Lam and reply by Preston and Campbell, pp.1020-43.

Note 3: to calculate a Pearson product-moment correlation coefficient, we may assign arbitrary values to H and L. It will be found that these cancel out in the final calculation, so the correlation is not affected by the choice of values. Some parents will have more than one offspring, in which case each parent-offspring pair should be counted in the covariance, and in the standard deviation of parents they should be counted once for each offspring.

Note 4:

Example 1

The key assumptions in this example are (a) equilibrium in intelligence levels (b) unequal numbers of H and L, and (c) the offspring of H and L have an equal probability of being in the same class as their parents. The main point of interest is to see what fertility rates arise from these assumptions.

Step 1: The total in the parental generation is 1000.
Step 2: The total of offspring is also 1000.
Step 3: There are 400 H's in the PG. Therefore there are 600 L's.
Step 4: As we are assuming equilibrium, there are 400 H's among the offspring. Therefore there are 600 L's.
Step 5: Among the offspring of H, H and L are in the ratio 7:3.
Step 6: Among the offspring of L, H and L are in the ratio 3:7. We can then calculate the row totals for the offspring. If we set the row total for H as X, the row total for L is (1000 - X). To obtain the value of X we can use the ratios stipulated in steps 5 and 6. These imply that .7X + .3(1000 - X) = 400, so we have X = 250. The rest of the table then follows easily, and we get:

[400 PGH] 175...... 75.......250
[600 PGL] 225.....525......750

The fertility rates come out as .625 for H and 1.25 for L.

Example 2

In this example we assume equal fertility for H and L, and explore what this implies for the proportion of offspring who are in the same class as their parents.

Steps 1 to 4: as in Example 1.
Step 5: With equal fertility, and a constant population size, the total number of offspring of H parents must be the same as the number of H parents, i.e. 400. Therefore the total number of offspring of L parents is 600. (Note that we are only losing one DF, not two.)
Step 6: It still remains to eliminate one DF. I choose to fix the proportion of offspring of H who are also H at .7.

This gives us the following table:

[400 PGH] 280....120......400
[600 PGL] 120.....480......600

The proportion of offspring of L who are also L is now 480/600 = .8, instead of .7 as previously.

Example 3

In this example I treat the numbers of H and L in the parental generation as unknowns, and derive them from the other parameters.

Steps 1 and 2: As in Example 1.
Step 3: The number of offspring who are H is to be the same as the number of H in the parental generation. Designate this number as X. The corresponding numbers of L are therefore 1000 - X
Step 4: I assume a fertility rate of 1.1 for L.
Step 5: I assume a fertility rate of .8 for H. With these fertility rates, and a total offspring number of 1000, X = 333.
Step 6: I assume that among the offspring of H, H and L are in the ratio 6:4.

These assumptions give the following table:

[333 PGH] 160.....106.....266
[667 PGL] 173......561.....734
Totals...... 333.....667... 1000

Example 4

Examples 1-3 all assume that the system is in equilibrium, with the numbers of L and H among the offspring equalling those in the parental generation. In Example 4 we explore what happens when the system is out of equilibrium. To do this I will take the figures of Example 1 and vary one of the parameters so that the system is no longer in equilibrium.

Steps 1 to 3: as in Example 1.
Step 4: We could vary any of the other parameters of Example 1. I choose to vary the fertility rates, and take the fertility rate for L as 1 instead of 1.25. It follows that the fertility rate for H is also 1.
Steps 5 - 6: As in Example 1.

The resulting table is as follows:

[400 PGH] 280.....120........400
[600 PGL] 180......420.......600

It will be seen that the numbers of H and L offspring no longer equal the numbers in the PG. If we insert the offspring numbers as the PG numbers in the next generation, and apply the same constraints as before, we get the following:

[460 PGH] 322........138.....460
[540 PGL] 162........420......540

It will be seen that the offspring numbers are still out of equilibrium, but they appear to be converging on 500 for both H and L. Given the symmetry of the fertility rates and proportions of H and L among the offspring, it is easy to see that the system will be in equilibrium if a 500:500 ratio is reached.

Note 5: Samuel H. Preston, 'Differential Fertility IQ and the IQ Distribution of a Population' in The Rising Curve, ed. Ulric Neisser, 1998, p.377-89.

Note 6: For details see D. S. Falconer, Introduction to Quantitative Genetics, 3rd edn., p.188-92.

Note 7: Richard Lynn, Dysgenics: Genetic Deterioration in Modern Populations, 1996, p200. Unfortunately Lynn, while seeing that something must be wrong with P & C's model, does not engage with it in detail.

Tuesday, July 03, 2007

More on adaptation   posted by p-ter @ 7/03/2007 12:49:00 PM

Larry Moran has a response to my post on adaptation. I say my piece in the comments on that post, and RPM has some further thoughts:
The modern selectionist does not invoke adatationist explanations for every evolutionary change. But he does not, by default, say "drift did it", either. Instead, he requires evidence for whichever conclusion he reaches. Maybe he's not really a selectionist; maybe he's just a population geneticist who understands how to detect selection.
This is the beauty of these here internets-- every time you start talking out your ass, there are enough people watching that most of the time someone will call you on it[1]. If I ever make a bone-headed remark about biochemistry, I certainly hope Larry Moran shows up and corrects me in the comments.

[1] I should be clear-- I'm not being facetious at all.

Labels: ,

Monday, July 02, 2007

Blowing off steam...   posted by DavidB @ 7/02/2007 03:35:00 AM

Excuse me while I blow off steam at a report in yesterday's New York Times (may require subscription) about the latest terrorist attacks in Britain. The authors casually refer to Britain's 'disenfranchised South Asian population'.

Disenfranchised??? According to my dictionary, 'disenfranchised' means 'deprived of the right to vote', or 'deprived of rights as citizen'. So in what sense, literal or metaphorical, are South Asians in Britain 'disenfranchised'?

Unless they are very young or very recent immigrants, they certainly have the vote (which in Britain depends on residence, not citizenship). Indeed, they may have more than one vote, as the South Asian communities tend to follow the old Ulster maxim 'vote early and vote often'. Nearly all the recent cases of electoral fraud in Britain have involved South Asians, especially in the abuse of proxy and postal votes. Never mind if your granny is in Pakistan (or dead), that is no reason why she shouldn't vote. And your local Mullah will probably be kind enough to deliver your postal vote for you - just leave the details blank!

Nor can it be said that South Asians are lacking in political influence. They have fewer Members of Parliament than their proportion in the population, but they make up for this in the influence they exert on other Members. This applies especially to Labour Members from the North and Midlands of England, where the Muslim vote is a powerful factor.

Of course, not all South Asians in Britain are Muslims. In fact the largest single group are Hindus of Indian origin. But they just do their work and make no fuss, so they are not a factor in politics.

Added on 3 July: This was written in haste, and there are a few errors. See the Comments page for corrections. They don't affect the point of substance that it is absurd to describe South Asians in Britain as 'disenfranchised'. But as it happens, the main suspects at present are not South Asians at all, but Arabs from comfortable backgrounds (doctors, etc.). Fortunately they also seem to have been wildly incompetent. Let's hope it stays that way.

Added on 6 July: According to the latest reports, 5 of the 8 key suspects are Arabs, and 3 are Indian Muslims (from Bangalore).