The Economist doesn't understand evolution

“Evolution and religion: In the beginning” from The Economist

One time could be an accident:

In the second camp are those, including some high up in the Vatican bureaucracy, who feel that Catholic scientists like Father Coyne have gone too far in accepting the world-view of their secular colleagues. This camp stresses that Darwinian science should not seduce people into believing that man evolved purely as the result of a process of random selection. While rejecting American-style intelligent design, some authoritative Catholic thinkers claim to see God’s hand in “convergence”: the apparent fact that, as they put it, similar processes and structures are present in organisms that have evolved separately.

Twice is a serious error:

But Benedict XVI apparently wants to lay down an even stronger line on the status of man as a species produced by divine ordinance, not just random selection. “Man is the only creature on earth that God willed for his own sake,” says a document issued under Pope John Paul II and approved by the then Cardinal Ratzinger.

Let’s be clear, “random selection” is not a short-hand for “random mutation and natural selection”. If anything, “random selection” is a description of neutral evolution.

Thus, as written, I have to join the camp that believes “that Darwinian science should not seduce people into believing that man evolved purely as the result of a process of random selection” and that “the status of man as a species [is] produced by … not just random selection”. Amen!

So WTF is wrong with the editorial staff at The Economist? They don’t seem to actually understand evolution. You can send them an email and explain it to them.

Correlation and Aggregation

There are many pitfalls in the interpretation of correlation coefficients. One relatively familiar one is the problem of restriction of range. To use a common illustrative example, if we take a sample of professional basketball players and calculate the correlation between their height and some measure of basketball performance, we will probably find that the correlation is weak. This does not mean that height is not an advantage in basketball: the reason for the weak correlation is that professional basketball players are a group with a restricted range of both height and basketball performance. The influence of height is therefore less conspicuous than if we took a random sample of the general population and tried them out at basketball.

The peril I want to highlight here is less familiar, and, unlike restriction of range, it tends to increase correlations. In general terms the problem arises from the aggregation of data. If data are aggregated and averaged, in some non-random way, the correlation between the resulting average values will often be higher than for the original disaggregated data, and may well increase with the level of aggregation. Suppose we take some trait in which there is a modest correlation at the individual level – say, between education and life expectancy. If now instead of individuals we take the averages for groups of individuals (for example the inhabitants of different towns, or even entire nations), and calculate the correlation between these averages, it is common for the correlation to be higher than for the individuals. It is also likely (though not certain) that as we take the averages of larger or more wide-ranging groups, the correlation will continue to increase. Correlation is essentially a way of measuring the proportion of the variability in a certain trait that can be accounted for by its association with some other trait. If we calculate the average values of the traits for wider and wider groups, the variability (technically, the variance) of the averages themselves will be reduced, because most of the random or localised influences on the data will cancel out as the groups are widened. But as the overall variance decreases, the proportion of the remaining variance explained by more general influences (which do not cancel out) is likely to increase. Technically speaking, the covariance may remain steady while the variance to be explained declines, or the covariance may decline, but the variance declines even faster.

This is all very abstract, so I will give a practical example from George Udny Yule’s Introduction to the Theory of Statistics…

Yule notes that in England, according to the official agricultural statistics, there is a modest positive correlation between the yield per acre of wheat and potatoes [Note 1]. Land that is good for growing wheat is also, on average, good for growing potatoes. If we calculate the correlation between wheat and potato yields at a fairly low level of geographical aggregation – the 48 counties of England – the correlation is .2189. If we then group the counties into 24 neighbouring pairs, the correlation increases to .2963. If we further aggregate them into 12 neighbouring groups, the correlation increases dramatically (it nearly doubles) to .5757; for 6 groups the correlation is .7649, and for 3 (which Yule describes as ‘the bitter end’) it rises to .9902. (Though Yule does not say so, if we actually reduced the number to 2, the correlation would be 1 or -1, as an algebraic identity. Of course in this case calculating the correlation would be pointless). [Note 2]

This does not necessarily mean that other factors correlated with wheat or potato yields are disappearing as the correlation between them increases. It is possible that if we measured, say, the correlation between wheat yield and annual rainfall, the correlation (positive or negative) would similarly increase as wider areas were aggregated. When all random or local factors have been eliminated by averaging, any remaining general factors may be correlated among themselves. For example there will probably be a strong negative correlation between annual rainfall and sunshine.

Unfortunately there seems to be no general rule to predict whether, or by how much, the aggregation of data will affect correlations (except that if groups of data pairs are selected at random from the entire set of data, the correlation between the group means should be approximately the same as between individuals – see Note 3). Yule gives another example of a study where a (negative) correlation increased steadily from -.502 to -.763 when 252 geographical areas were aggregated by stages into 25, but in the same study another correlation changed relatively little, and more erratically, during the same process of aggregation.

I will not attempt to draw any practical conclusions, except to say that it is a problem that deserves to be more widely known and taken into account, especially when we find very high correlations. As Yule advises: ‘What explanation we seek in individual cases depends on the individual circumstances. We can only leave the reader with the warning to watch very carefully the possibility of grouping effects, particularly in economic investigations’. Nor is the problem confined to geographical aggregation: it could apply also to grouping by social class, occupation, educational level, or any other criteria. It should not be inferred that that the correlations resulting from grouping are invalid or meaningless, just that the value of a correlation may be relative to the method of grouping. Group correlations derived at different levels of aggregation therefore cannot safely be compared. The problem is presumably well known to statisticians, and has been discussed from time to time since the 1950s, especially in economics and sociology, under the somewhat misleading heading of ‘ecological correlation‘, but it still does not seem to be sufficiently publicised. For example, in psychometrics it could be relevant to the correlation between test results, or between test results and other criteria, such as income. This is mentioned very briefly by J. P. Guilford, Fundamental Statistics in Psychology and Education, 1950, p.355, but on a cursory search I have not found any other reference to the subject in the psychometric literature.

Note 1: examples are taken from G. U. Yule and M. Kendall, Introduction to the Theory of Statistics, 14th edition, 1950, pp.310-315. Yule distinguishes two different but related forms of the problem. In one form, which he calls the ‘Modifiable Unit’, the items to be correlated are inherently variable in scope, for example because they involve a measurement over a geographical area. In the other form, which he calls the ‘Attenuation Effect’, the ultimate units of analysis are discrete items, but the problem arises when we choose to average them and then correlate the averages.

Note 2: I assume that correlation is measured by the Pearson product-moment correlation coefficient. The correlation then equals ∑(x – Mx)(y – My)/Nσx.σy, where N is the number of paired values, and Mx and My are the means of the x and y values respectively. If there are only two x values, a and b, and two corresponding y values, c and d, the correlation comes out as (a – b)(c – d)/√(a – b)²√(c – d)². Taking positive values for the two square roots (since they represent standard deviations) the correlation will be either 1 or – 1, depending on whether (a – b)(c – d) has a positive or negative value.

Note 3: This is stated without proof by Guilford, p.216. The explanation for the result is evidently that with purely random grouping, the numerator and denominator of the correlation coefficient are both reduced in the same ratio. Suppose we divide the N individual
pairs of correlated data at random into N/n groups of n pairs each. If we designate the deviation values (relative to the population mean) of one variable as a, b, c…, and the corresponding values of the other variable as A, B, C…, then the group means will be of the form (a + b + c…)/n and (A + B + C…)/n. In calculating the correlation between group means, each such pair of means must be multiplied together to give a product of the form (a + b + c…)(A + B + C…)/n². Expanding this gives a sum (aA + bB + cC… + aB + aC… + bA + bC…)/n². But except for the terms aA, bB, cC, etc, which are the products of the individual correlated pairs, the terms in this sum will total approximately to zero, since the factors in each product term are uncorrelated (having been chosen randomly). Since there are N/n groups, the total covariance will be ∑(aA + bB +cC…)/Nn, where (aA + bB + cC…) includes the contribution of all groups. This is 1/n times the covariance of the N individual pairs of data. But by a similar analysis the product of the standard deviations of the group means is 1/n times the product of the standard deviations of the individual values. The numerator and denominator in the correlation coefficient are therefore reduced in the same ratio, and the correlation itself is unaffected. There will of course be some fluctuation due to sampling error.

Intercourse and Intelligence

Tyler Cowen quotes from a new study testing the relationship between grades and delayed sexual activity.

Last December I passed a paper along to Razib showing that high-school age adolescents with higher IQs and extremely low IQs were less likely to have had first intercourse than those with average to below average intelligence. (i.e. for males with IQs under 70, 63.3% were still virgins, for those with IQs between 70-90 only 50.2% were virgin, 58.6% were virgins with IQs between 90-110, and 70.3% with IQs over 110 were virgins)

In fact, a more detailed study from 2000 is devoted strictly to this topic, and finds the same thing: Smart Teens Don’t Have Sex (or Kiss Much Either).

The team looked at 1000s of representative teens grades 7-12 in the National Longitudinal Study of Adolescent Health and The Biosocial Factors in Adolescent Development datasets, both of which include an IQ test, and include detailed sexual experience questions ranging from hand-holding to intercourse. As with the other study there was a curvilinear relationship: students with IQs above 100 and below 70 were significantly less likely to have had intercourse than those in between. Also like the other study, they found teens with IQs ranging from 75 to 90 had the lowest probability of virginity (the authors note this is also the same IQ range where propensity towards crime peaks).

Depending on the specific age and gender, an adolescent with an IQ of 100 was 1.5 to 5 times more likely to have had intercourse than a teen with a score of 120 or 130. Each additional point of IQ increased the odds of virginity by 2.7% for males and 1.7% for females. But higher IQ had a similar relationship across the entire range of romantic/sexual interactions, decreasing the odds that teens had ever kissed or even held hands with a member of the opposite sex at each age.

While these authors leave off at grade 12th, it would seem plausible to expect that this relationship extends beyond high school. To explore this, plenty of interesting facts come from a 2001 campus sex survey by the joint MIT/Wellesley college magazine Counterpoint (PDF). Looking within and between colleges, IQ appears to delay sexual activity on into young adulthood.

By the age of 19, 80% of US males and 75% of women have lost their virginity, and 87% of college students have had sex. But this number appears to be much lower at elite (i.e. more intelligent) colleges. According to the article, only 56% of Princeton undergraduates have had intercourse. At Harvard 59% of the undergraduates are non-virgins, and at MIT, only a slight majority, 51%, have had intercourse. Further, only 65% of MIT graduate students have had sex.

The student surveys at MIT and Wellesley also compared virginity by academic major. The chart for Wellesley displayed below shows that 0% of studio art majors were virgins, but 72% of biology majors were virgins, and 83% of biochem and math majors were virgins! Similarly, at MIT 20% of ‘humanities’ majors were virgins, but 73% of biology majors. (Apparently those most likely to read Darwin are also the least Darwinian!)

Looking at this chart it would strongly appear that higher complexity majors contain more virgins than majors with lower cognitive demand. This paper provides me with GRE scores by academic discipline, and, in fact, the correlation between the percentage of virgins in each Wellesley major and the average ‘Analytical’ GRE score associated with the discipline is 0.60.

One reason we might guess that smarter people in high school, or in more challenging colleges or majors, delay their sexual debuts is because they are delaying gratification in expectation of future reward. Sexual behavior (or at least the investment needed to procure a partner or sustain one) may compete with time/resources required for other goals, and intelligent people may have more demanding goals. James Watson even hinted at this in a recent Esquire magazine piece:

If I had been married earlier in life, I wouldn’t have seen the double helix. I would have been taking care of the kids on Saturday. On the other hand, I was lonely a lot of the time.

While sex may not be marriage, it may still require effort that intelligent people prefer to invest elsewhere. This would fit Aldus Huxley’s alleged definition of an intellectual as a person who’s found one thing that’s more interesting than sex.

Another idea is that smarter people are more risk averse, and delaying these activities is a byproduct of enhanced concerns about unwanted pregnancy and disease. While not avoiding sexual behaviors, per se, they are just less likely to seek it out or consent to it for fear of the potential consequences.

Another idea is that smarter people are more religious or more ethically conservative, and are trying harder to wait for marriage to have sex.

Another idea, consistent with popular media portrayals of geeks and nerds (males at least), is that intelligent people actually want to have sex, but are simply less likely or unable to obtain willing partners because they are disproportionately viewed as unattractive or undesirable as partners.

Another idea is that intelligent people have lower general sex drives. This shouldn’t be confused with the first theory, where their sex drives would be normal and they have greater self-restraint.

Some insightful digging by blogger Half Sigma into the General Social Survey, which also includes an abbreviated intelligence test, has turned up a number of associations that speak to these theories. The relationship between sexual activity and intelligence found across adolescence and young adulthood appears to continue on into adulthood proper.

Not only do intelligent people have a delayed onset of sexual behavior, Half Sigma found that they also have a lower number of premarital sex partners throughout adulthood (18-39). While this is consistent with the above theory that high IQ people are more religious and conservative, this is, of course, not true. Religiousness correlates with lower IQ, and as HS shows in the same post, intelligent people were also more likely to say that premarital sex was not immoral. (Leaving those who did think it was immoral to participate in the bulk of it!) Most of the other theories are still consistent with this finding though.

Perhaps more revealing, HS, also showed that intelligence correlates with less sex within marriage for the same age range. While still consistent with pregnancy fears and competing interests, lower sex drive seems like a better fit. In fact anothe
r revealing finding from the Counterpoint survey was that while 95% of US men and 70% of women masturbate, this number is only 68% of men and 20% of women at MIT!

Also the idea that more intelligent people are too busy for the opposite sex not just in 7th grade to college, but throughout adulthood and for their own spouse, seems unrealistic. In fact the GSS also shows (PDF) that smarter people spend more time socializing with their friends, indicating their hours aren’t spent as uniquely isolated and narrowly channeled as the theory would require.

But lower sex drive and anxiety about sex’s consequences can’t be the whole story either. Half Sigma also showed that the smartest men in the GSS (approx. IQ >120) were also more likely to visit a prostitute. (Hardly indicative of cautiousness) This may suggest intelligent men are less able to find willing sex partners. Are smart men less attractive to women? Perhaps in some ways. For instance HS found that smart men were less likely to be athletic, and this paper shows, unathletic men and women have fewer sex partners. Athletic men, with more willing sexual partners are also less likely to visit a prostitute. Athletic activity gives men more masculine bodies, which are more attractive to women. A more masculine physique correlates with (PDF) an increased number of sex partners.

So intelligent people have lower libidos and less masculine physiques. What hormone is responsible for both sex drive and masculine builds? That’s right: testosterone.

And two new papers suggest that testosterone may depress IQ. One team found that salivary testosterone levels were lower for preadolescent boys with IQs above 130 and below 70. (the same two groups most likely to be virgins in adolescence)

Another paper suggests that a gene responsible for androgen sensitivity and higher sperm counts may also create a tradeoff for intelligence.

Genetics of speciation

RPM points out that the most recent issue of Heredity tackles the issue of the genetics of speciation. Here’s an interesting thing I’ve noted, there are two ways to look at species questions. First, there are the taxonomists, who have been strongly influenced by the cladist revolution. They take a big picture philosophical view, and are obviously greatly concerned with process in terms of classification and demarcation. In contrast, there are the evolutionary geneticists who tend to be less interested in species qua species, as opposed to the process of genetic differentiation. In other words, for the latter camp species discussions are simply an ends toward elucidating the evolutionary dynamics of populations. The taxonomists in contrast are focused on species as the ends for generating their systems of evolutionary relationships. The Neandertal introgression story should make it clear I’m interested in the dynamics of evolutionary processes, not any rigorous species classification.

Addendum: Check out this review of Henry Gee’s In Search of Deep Time: Beyond the Fossil Record to a New History of Life, to see what I mean about the taxonomic sensibility. A friend of mine recalls observing a woman in her lab being upbraided by a cladist at an entomological conference for practicing “un-Popperian” science.

Hitch on Drugs

Via Drug WarRant, Hitchens is playin’ my tune lately:

An excerpt from his 21 Solutions to Save the World:

The largest single change for the better in U.S. foreign policy, and one that could be accomplished simply by an act of political will, would be the abandonment of the socalled War on Drugs. This last relic of the Nixon era has long been a laughingstock within the borders of the United States itself (where narcotics are freely available to anybody who wants them and where the only guarantee is that all the money goes straight into criminal hands). But the same diminishing returns are now having a deplorable effect on America’s international efforts.

The team draft

Echoes of Terror Case Haunt California Pakistanis. From the article:

Lodi, a city of 62,000 people 72 miles east of San Francisco, is something of an anomaly among Pakistani immigrants. Most come to the United States to pursue professional careers, to become doctors or academics in large cities. But mainly rural peasants started coming to Lodi around 1920, and residents say 80 percent of the town’s 2,500 Muslims are Pakistanis.

They came as agricultural laborers and never really assimilated, preserving their traditional ways by dispatching the young back home for arranged marriages.

I don’t really need to offer further comment on this. The culture described in the article seems more reminiscent of European Islam than American Islam, except that the Pakistanis in Lodi seem more gainfully employed. In much of Europe the local Muslim community consists of a monoculture derived from peasant immigrants (e.g., the Turks in German y, the Pakistanis of northern England). In the United States Muslims are ethnically diverse and subject to selective immigration policies which skews the migration stream toward professional elites (PDF). Lodi seems to resemble Europe….

Related: You can watch the Frontline special The Enemy Within on the web. It covers the Lodi case.

Posted in Uncategorized

Consciousness Catch-22

I was listening to a lecture by Christof Koch this morning, and I had a thought that I haven’t come across before regarding consciousness studies. There may be a very high technical bar for ethical studies into the nature of consciousness, higher than I expected before.My issue may just be the stretchiness of the term ‘consciousness’, but there might be something of substance in here too. Can we study anything besides humans in order to understand consciousness? If not, then we have to have safe temporary ways to manipulate consciousness in humans, which would mean safe, temporary ways to control neurons in humans. We won’t have that for a long time. By the way, I’m not talking about being awake or asleep. I’m talking about a more important type of consciousness. This type:

Are the “pains” that usefully prevent us from allowing our limbs to assume awkward, joint-damaging positions while we sleep experiences that require a “subject” (McGinn, 1995), or might they be properly called unconscious pains? Do they have moral significance in any case? Such body-protecting states of the nervous system might be called “sentient” states without thereby implying that they were the experiences of any self, any ego, any subject. For such states to matter — whether or not we call them pains or conscious states or experiences — there must be an enduring, complex subject to whom they matter because they are a source of suffering.

Akbar Ganji says suffering is the source of rights. I think that’s simplistic, but it might be some part. My basic point is this: If non-human primates can do consciousness-like things that are worth studying, then wouldn’t it be unethical to experiment on them? If consciousness is the substrate for suffering and suffering is the source of rights, wouldn’t they have rights like us? I’m not saying that they do have the capacity for suffering, but if they don’t, then they might not be worth studying for consciousness anyway.

In other parts of that Dennett essay I quoted above, he notes that outward behavior isn’t a good indicator of suffering in the morally meaningful sense. We can see an earthworm writhe and be unconcerned. So if we can’t tell if anything is capable of suffering, maybe its not the best line to use for choosing ethically in animal research. A lot of times when I’m thinking about people trying to free chimps or things in that vein I think “That chimp would rip you limb from limb if it got the chance.” I really don’t think a chimp would have any qualms about drilling a hole in your head and inserting electrodes if it new how. I guess what I’m thinking is that maybe its okay not to show so much consideration because they certainly wouldn’t show you any. If we were going to grant chimps human rights, wouldn’t they need to take responsibility and act like they deserve it?

Just a couple things I was thinking about. I thought it might be really interesting if a consciousness researcher managed to prove that the research he was doing was unethical.

How the Sabians saved civilization?

Reading The Later Roman Empire, 284-602: A Social, Economic, and Administrative Survey I stumbled upon this on page 939:

…In some places paganism survived the Arab conquest. in 830 the people of Carrhae [modern day Harran], a city always notorious for its devotion to the old gods, were threatened with massacre by the Caliph unless they abandoned their religion for Islam or one of the tolerated faiths and only saved themselves by profession themselves to be Sabians. To this day the heretical sect of the Nusairi in the mountains between the upper Ortones and the sea profess doctrines which clearly derive from the Neo-Platonic paganism of the later empire….

The “Nusairi” refers to the Alawites, a heterodox marginally Islamic sect whose claim to fame lay in its control of the modern day state of Syria. I had long known of the late paganism of Carrhae. In other cities where paganism was vital and dominated civic life during the 6th century the Byzantine Emperors employed the force of arms to destroy the temples and crush public sponsorship of non-Christian piety (Gaza, Heliopolis, etc.); the religious life of Carrhae was protected in part by its closeness to the Persian Empire. Some of the philosophers scattered after the closing of the Academy eventually settled in Carrhae, which in some ways resembled a time capsule that had preserved the sensibilities of pagan late antiquity, where the astral cults reigned supreme over a latitudinarian religious culture.

In any case, reading up on the Sabians I am not convinced of a direct connection between this group and the Alawites. Though we must classify and demarcate religious sentiments as if they stand alone, we intuitively understand that a system of beliefs are affected by the Zeitgeist. Carrhae was dominated by Sabians until 1050, when Muslims took over the city (the Sabians were found in nearby areas for several centuries until the Mongol invasions, their folkways are attested by Maimonides). This is approximately the period when many of heterodox Islamic and post-Islamic sects arose, the various Ismaili groups, the Druze and the Alawites & Yezidis. It stands to reason that the Sabians might have influenced the thinking of many of their neighbors because they were a prominent community. Similarly, the Sabians themselves emerged out of the substrate of the northern Levant and upper Mesopotamia, so the similarities between modern groups like the Alawites and the medieval Sabians might simply be due to the fact that they share the same mix of cultural preconditions.1

But my interest in the possibility that the Alawites descend from the last pagan remnants of antiquity in the east disappeared when I read about Thabit ibn Qurra, the most prominent of the Sabians. He was an “Arab” astronomer and mathematician, and one of the heads of the famous House of Wisdom. Some cursory searching on Google Books implies that he was not an anomaly, the Sabians were well represented amongst the translators who mediated aspects of Greek learning and made it accessible to the Arab Islamic world.

Why is this relevant? One of the historical myths of our era is that the Arab Muslim saved the Greek achievement for Western civilization. The argument is that there runs a line of tradition starting during the Greek Classical period down to the modern post-Enlightenment era which was preserved by the efforts of the House of Wisdom. This is false insofar as the Byzantines also transmitted Greek works to the West, and the refugees who washed up on the shores of Italy during the late medieval period as Constantinople fell before the Turks helped spark the Italian Renaissance. But the Byzantine role is not sexy because it doesn’t serve a multicultural narrative (before the contemporary period the emphasis placed upon Islamic civilization’s role in preserving Greek learning was used as a cudgel against Western Christianity). And yet an important fact about the House of Wisdom is that it was a multicultural affair, and that during the early phases most of the work was in the hands of multilingual dhimmis, who were after all in a position to know Greek and Arabic. Though I had known of the role of Nestorian Christians, the Sabians’ part was somewhat of a surprise (I was to understand that some of the translators were pagans, but I had not known that that was a synonym for Sabians from Carrhae). Now, unlike Christians or Muslims, I think one might contend that the Sabians of Carrhae had less ambivalence toward the Greek pagan heritage, after all, their culture was a descendant of one that had sheltered the last of the Neo-Platonic philosophers. I am therefore inclined to wonder if the Sabians in particular were a vector for preserving and promoting the rich intellectual tradition which stretched back to the pre-Socratics? I will have to look into this hypothesis (I’m skeptical actually).

On a broader theoretical level I am curious about the role that small cultures like the Sabians play in the dynamics of cultural and civilizational change. Carrhae remained a pagan stronghold because of an accident of geography, its strategic position near the border with Persia and the protection offered by the Shah resulted in the preservations of its peculiar civic paganism in the face of an aggressively Christianizing empire. Though a man of Carrhae could never hope to be great in imperial service without baptism, if one wanted to be a man of standing and influence within one’s own community then pagan profession was necessary so that one could partake of the communal sacraments. The forcible destruction of these sacraments in other pagan cities destabilized this social equilibrium and the result was inevitable Christianization as local elites defected from a religious cult which no longer accrued prestige but was a universal liability.

But though this was the proximate dynamic which led to Carrhae preserving its pagan character, I am offering here the possibility that this might have had a long term ultimate impact of serving as a major conduit for the thought of late antiquity down to the Islamic period. If Carrhae had not preserved its unique culture no doubt the Nestorian scholars of the House of Wisdom would have done their fair bit of translation, but one wonders what the Muslims might have overlooked? This is not to say that Carrhae was a font of rationality and wisdom, many would characterize late Neo-Platonism as a debased supernatural cult with only the faintest philosophical touches. But, just as Hinduism has under its broad umbrella primitive devotionalisms and rarified Avaita Vedanta, so late classical paganism spanned the gamut. In contrast, one might contend that the rise of Christianity and Islam resulted in a constraining of the avowed beliefs of the elite, a homogenization of the complexities of the late antique intellectual landscape. During the centuries after the rise of Islam perhaps Carrhae served as a reservoir of intellectual diversity? Do microcultures play the same role within the matrix of other homogenizing macrocultures?

1 – This region was the meeting place of Greek, Arab, non-Arab Semite, Armenian, Persian and Kurd, to name a few. There were also variations within this region, Syria had a far stronger Greek presence than northern Mesopotamia, which had an elite Syraic speaking culture. In any case, the presence of deep rooted Astral cults seems universal. I once read that the Ottoman sultan once made progress through a Kurdish town where the residents worshiped the sun. The sultan was angered by this paganism, and eventually the residents were taken under the wing of the local Jacobite bishop. This is very similar to the story of caliph Al-Ma’mum forcing the residents of Carrhae to choose a protected religion, so I am not sure if these incidents are necessarily true, as opposed to repeating a common motif.

Voles getting around

A follow-up on the vole/monogamy/vasopressin story has just been published in Trends in Genetics. A quick summary of the relevant information:

1. Prarie voles are socially monogomous (note the qualifier “socially”. Genetics suggests a certain amount of “infidelity”, if one can call it that in voles). Meadow voles are not.

2. There is a regulatory region upstream of the vasopressin receptor that carries a “short” allele in meadow voles and a “long” allele in prarie voles.

3. In vitro, the long and short alleles have higher and lower expression, respectively, of vasopressin.

4. When the prarie vole allele is transformed into meadow voles, there is as marked increase in pair bonding[cite]

5. Within prairie voles, different allele lengths are associated with pair bonding and vasopressin receptor distribution [cite]

This is a fascinating story, and many people, including the authors of the above papers, hypothesized a sort of single “switch” for monogamy in mammals. However,

6. Many non-monogamous species have “long” alleles [cite]

This, then, refutes the possibility of a single genetic switch controlling monogamy in mammals. The response just published in TIG accepts that, but argues against rejecting a role for vasopressin recepter variability in pair-bonding and other social behavior. I agree. The previous studies are quite convincing that such a role does exist (see in particular the experiments in the papers cited in points 4 and 5), and more detailed study of the vasopressin receptor locus is certainly justified.