Saturday, 25 October 2014

More sex, cousin?


Regular readers will remember this topic, but here is the link to a previous post:

Interesting work on the link between cousin marriage and intelligence has been published by Farid and Afzal who have collected Wechsler test scores on inbred children in Muslim areas of India. They chart a very substantial effect of genetically driven cognitive decline, with consequentially higher rates of mental retardation in the inbred families. The differences are enormous: 25 IQ points of retardation on average, and a massive, mind-bending 37 IQ points for the most inbred. Can it really be as bad as that? Has inbreeding caused different groups to diverge so much that they are an average 1.67 standard deviations apart? Think of the heated debates and the social and political ramifications of a 1 standard deviation difference between African Americans and European Americans in the USA. Intellectual differences of close to 2 standard deviations will be of greater consequence in the Muslim world and beyond.

Mohd Fareed and Mohammad Afzal. Estimating the Inbreeding Depression on Cognitive Behavior: A Population Based Study of Child Cohort. Published: October 14, 2014. DOI: 10.1371/journal.pone.0109585

They say: A cohort of 408 children (6 to 15 years of age) was selected from inbred and non-inbred families of five Muslim populations of Jammu region. The Wechsler Intelligence Scales for Children (WISC) was used to measure the verbal IQ (VIQ), performance IQ (PIQ) and full scale IQ (FSIQ). Family pedigrees were drawn to access the family history and children’s inbred status in terms of coefficient of inbreeding (F).

They summarise their results: We found significant decline in child cognitive abilities due to inbreeding and high frequency of mental retardation among offspring from inbred families. The mean differences (95% C.I.) were reported for the VIQ, being −22.00 (−24.82, −19.17), PIQ −26.92 (−29.96, −23.87) and FSIQ −24.47 (−27.35, −21.59) for inbred as compared to non-inbred children (p>0.001). The higher risk of being mentally retarded was found to be more obvious among inbred categories corresponding to the degree of inbreeding and the same accounts least for non-inbred children (p<0.0001). We observed an increase in the difference in mean values for VIQ, PIQ and FSIQ with the increase of inbreeding coefficient and these were found to be statistically significant (p<0.05). The regression analysis showed a fitness decline (depression) for VIQ (R2 = 0.436), PIQ (R2 = 0.468) and FSIQ (R2 = 0.464) with increasing inbreeding coefficients (p<0.01).

Inbreeding (consanguineous marriages among humans) produces homozygous offspring, since the mating of pairs occurs between genetically closely related individuals. The phenomenon of inbreeding or endogamy, increases the level of homozygotes for autosomal recessive genetic disorders and generally leads to decreased fitness of a population known as inbreeding depression which provides a major focus in clinical studies [21].

Parental consanguinity has been associated with increased risk of adverse prenatal outcomes including stillbirths, low birth weight, preterm delivery, abortion, infant and child mortality, congenital birth defects, cognitive impairments, malformations and many other complex disorders [12][32].

A study has revealed that the overall incidence of congenital malformations was 2.5 times higher amongst the children of inbred families when compared to that of non-inbred families [33]. Consanguinity has been associated with significant decline in mean values for height, weight and body mass index (BMI) and the subsequent depression on children growth, much influenced in proportion to their inbreeding coefficients with least variation for non-genetic factors [21].

Genealogical information (family pedigrees) up to five generations back (volunteered by the parents) helped in ascertaining the consanguinity persuasion of their marriage and thus, child’s inbreeding status was determined. The information provided by the parents was cross checked by seeking help from the elder members of the family. In case of ambiguity such data were omitted.

Wright’s path relationship method for calculating the coefficient of inbreeding (F) was used for all mating types following the order: double first cousin (F = 0.125) > first cousin (F = 0.0625) > first cousin once removed (F = 0.03125) > second cousin (F = 0.0156). In non-consanguineous families or for the distant marriages, the coefficient of inbreeding was effectively zero (F = 0.000) [21].

The authors did full Wechsler tests, so this is a gold-standard assessment, and somewhat unusual in standard publications on intelligence with large sample sizes. In another unusual move in contemporary publications, they have plotted out their data so as to show the distributions. Respect. I am almost reaching out to give them the, as yet mythical, Thompson Prize for Plain Statistics. The intelligence scores are massively shifted to the left (lower intelligence) in the inbred group.

The mean difference in VIQ, PIQ and FSIQ showed a significant inflation with the increase of the degree of inbreeding (p<0.05, using post-hoc tests) and followed the order: double first cousin > first cousin > first cousin once removed > second cousin > non-inbred. This indicates the risk for cognitive impairments or ID was more common among the subjects having greater value of inbreeding coefficients.

The Table below shows the massive drop in Verbal, Performance and Full Scale intelligence associated with consanguinity measured by the F inbreeding coefficient. For the guidance of those about to marry, the worst effects are when two double cousins marry (see list above for verbal descriptions) and check the figures below for the F coefficient categories.



The Full Scale results are drawn out here.

Figure 5

Actually, I may have to hold back the Thompson prize, because they have now reversed the direction of IQ scores, so that high scores are on the left, against usual convention. Pity the poor reader. Nonetheless, the wind blowing through these intellects is clear: the Double First Cousins peak in severe mental handicap, the Non-inbred are at virtually normal scores.

This is all pretty dramatic stuff. The authors have done well. Good sized sample, excellent quality intelligence testing, plainly described results. 

However, I am not one to let any researcher get away lightly. The authors do not give the final group IQ, nor do the answer the obvious question: if 45% of this Muslim sample are inbred, what would their IQs look like if they stopped inbreeding?

For general guidance, the overall group IQ would be interesting to compare with other measures of Middle East Muslim intelligence, and with Indian intelligence estimates. Lynn’s mean value for India is about 82, Pakistan 84. Specifically, in this sample it would appear that Muslims who avoid cousin marriage have IQs of 96.5 which is relatively close to Greenwich Mean IQs. Seen in this light, one bad cultural habit has damaged the intellects of normally able people.

I would like to see the sample sizes for each level of inbreeding, so that I can do a few more back-of-the-envelope calculations on what might happen over a few generations if we let loose that great Western invention: Romantic Love. This provides the perfect excuse to ditch your responsibilities to your pushy parents, up sticks and strike out into the blue yonder looking for True Love, far away from your cousins. Time for Bollywood to triumph over tradition.


  1. Extraordinary stuff. With potentially enormous implications for understanding current geopolitical events.

  2. (1) Very good to finally see a comprehensive study that so clearly illustrates the massive IQ loss due to inbreeding in the Muslim world. Another positive thing, it just struck me, is that the authors - judging from their names, at any rate - are subcontinental Muslims themselves, so there's less scope for criticizing them on the grounds of "privilege" or whatever.

    (2) As was pointed out in a few Twitter discussions with hbdchick, Jayman, & Co., it is likely that the least inbred groups were also the higher status ones, so their "true" IQ (minus the inbreeding depression) could be expected to be higher than average. Consequently, the difference due to inbreeding, while huge, is nonetheless smaller than the figures presented. This makes sense, after all, Pakistan is your quintessential Third World country - much poorer than, say, Latin America - with some major nutritional and parasitic load factors bringing down IQ. It would be strange if stamping out cousin marriage were just by itself to bring them up to a Greece-like 97.

    1. Thanks for your comment. Originally I had put in a sentence to the effect that adjusting for living conditions would bring the outbred group to the Greenwich Mean, and then took it out, realising that I was making an assumption without knowing the facts. So, I don't know what the nutrition and parasitic load would be, though I assume far lower than sub-saharan Africa. Also, I am unsure about the "high status" argument. It is not an absolute scale, after all. You can be high status in a low ability area. IQ 96 will make you a king in a world where IQ80 is the average, because you will be 1 sd above the mean. It can be argued either way, but I favour the interpretation that IQ 96 is what you could expect without inbreeding. Still, worth resolving the issue with an experiment. Perhaps some local activist (probably a woman) could lead a revolt in at least one village, and then we would have testable IQs within about 7 years.

  3. AnonymousForGoodReason25 October 2014 at 16:22

    Easily the most mindblowing result I've seen on your blog.

    That being said, as AKarlin comments, it is difficult to account for pop strat.

  4. "However, I am not one to let any researcher get away lightly. The authors do not give the final group IQ, nor do the answer the obvious question: if 45% of this Muslim sample are inbred, what would their IQs look like if they stopped inbreeding?"

    That's the kicker, as Anatoly hits. How can we necessarily attribute these observed IQ differences to the inbreeding itself? Is there anything saying that their potential "outbred" IQ would be all that different?

    We'd need a few outbred (as in, the result of one generation of outbreeding) individuals from these inbred families to know.

  5. Mohd Fareed adds that there is a lack of awareness (about inbreeding) and a lack of genetic counselling in the region, and there is a need for health policy makers to train and make people familiar with its harmful effects.

    1. Maybe, but it is very hard to get people to give up traditions that are hundreds of years old. Is it really necessary to kick the hornet's nest? Pakistan is already unstable enough as it is. I can see heath policies having an effect in Punjab and Sindh provinces. No way in hell any health official will leave with his head from Khyber Pakhtunkhwa interfering with local affairs.

    2. This comment has been removed by the author.

  6. Thanks for your concern. Firstly, this study from Indian Northern State.
    Secondly, It is not only a problem of Pakistan or Muslim communities. We should treat as a genetic concern, not too much cultural or religious. If you have had read the history of cousin marriages, you will definitely come to know about its existence in every religion or race. European royal dynasties (Queen Victoria and Prince Albert were a preeminent example). Even some great geneticists like Charles Darwin and his wife Emma were first cousins and so on.....!
    Now we have to see: to what extent inbreeding is harmful?
    How many generations are responsible for causing such depression?
    There are some advantages also, so it is necessary to check it at genome level.
    New advances in genome sequencing enable us to identify the risk of inheritance of disease/traits. Runs of homozygosity (ROH), based on single nucleotide polymorphisms (SNPs) is very useful for such inbreeding studies. Exome sequencing, candidate gene approaches have advantages.
    Most importantly, genetic counselling, health policy, education and awareness needed to overcome this problem.

  7. This comment has been removed by the author.

  8. Dear Dr. James Thompson,
    I have seen your comment regarding FSIQ scores in figure 5 (reversed the order?).
    I wish to tell here, that it was done just to be easily understood by the general public and to create an attractive look as well. If we have done in the direction as you have pointed out, that will become more complex to understand. So, we deliberately did this and also there is no specific plotting rule for IQ scores in a given direction.

  9. Dear Mohd, I don't understand your argument. The differential between the plotted lines is entirely unaltered by a left right inversion.The position of the plotted scores would not change. As to the usual direction of scores, standard convention in every Wechsler manual, including the results summary sheet shows lower scores on the left. Reverse it immediately and you will reach a wider audience ! By the way, if Hollywood or Bollywood call you, please let me know.

    1. Dear Sir,
      Though the IQ scores (number) is decreasing from left to right, seems unusual. But in terms of intellectual disability, these scores represent the severity or mental illness, which is being increasing from left to right. In clinical studies, we always take normal (control one) in left side to compare with other groups. Same is the case here, higher scores on the left represent the group of normal (high IQs) with later increasing mental illness.

  10. I'm a product of a first cousin marriage. My parents are muslims and first cousins originally from Pakistan but living in Britain as I am.
    But, I'm an engineering student who achieved 3 A*'s in A level Maths, Physics and Chemistry.
    What you got to say about that?

    1. Congratulations on your achievements.
      If as many children of first cousin marriages achieve good scholastic results as everyone else that would invalidate the findings of this study. One simply needs lots of those individual cases, and certainly more than a few.