Thursday 9 April 2015

Original paper (part 1) Emil Kirkegaard on brain size and poverty


When Ezra Pound was planning to write the primer which eventually became The ABC of Reading he intended to give no more than a collection of key readings without any further comment. Pound thought that his readers would get the drift of what good literature was about, and that explanation was superfluous. In that spirit, why not read Emil Kirkegaard’s publication just out out today:

The general brain factor, working memory, parental income and education, and racial admixture



Famously, Ezra Pound resiled from his minimalist position, and consented to make a comment or two about the exhibited texts, thus writing a classic of literary tutelage.

What I like about Emil’s paper is that despite being a very bright young guy he goes through data sets in the ancient and systematic way that AE Maxwell always championed: he looks at the data in detail, and gets to know what is there: the good, the bad, and the ugly. If, as in any real data set, there are lapses, quirks, anomalies and blank spaces it is well to know about them in advance, not jump to the towering castles of fancy statistics before surveying the whole terrain, and seeking secure foundations. I can follow his thought processes, even though I am not running as fast as he is.

Emil gives an erudite introduction, and draws out the relationships between the variables. He then deals with missing variables; the problems of adjusting for sex differences in brain size; the benefits of adding in cubed age to deal with residuals; the problem of there being no size measures for the children with which brain/body comparisons could be computed (Encephalization Quotient); the reasons for factor analyzing the many brain surface measures; lists the known association between brain measures and cognitive ability; notes that 11 regions were studied in the left hemisphere versus 8 in the right (which might unbalance the analysis of results); finds that brain size appears to increase till age 10 and then falls (?); then slogs through many factor solutions; showing that 4 factors offer by far the best, congruent solution; looks at the results for the Flanker distraction task (I would not have given this measure priority, because it is of less significance than mainstream intelligence), and that some children simply don’t understand it; looks at the working memory test; shows the memory test is more correlated with brain size; then looks at the core measures of education and income and finds slightly higher correlations for boys than girls; works through the racial admixture data; finds difference between different scanners…….. and then it is night in Denmark, and also night in England.

More is on its way, but this an argosy of treasures. I await his next findings.

Before I leave you, I would normally have said: “This paper should be published in Nature”, but we can’t really say now, can we?

I hope you will be impressed, as I am, by the company I keep. 


  1. “This paper should be published in Nature” could indeed be construed as an insult. I'm so old that I can remember when becoming an FRS was universally recognised as a very big pat of the back, rather than being admission to a criminal crew. Or at least an inattentive crew led by con men and shakedown artists.

    I'm so old that I use "construe" rather than misuse "parse". But I'd rather be young again.

  2. I have no intention of publishing in Nature. I will publish with Winnower. Quick and easy and I get a DOI and get indexed by Google Scholar.

    I will read the comments here for feedback. Otherwise just start a thread on at the forum:

    I am still trying to deal with the question of how to deal with some missing data (for age) and how to control for age and scanner.

    The odd pattern is that age is correlated with African ancestry (the black children were older). This results in a positive correlation between African% and working memory, opposite of expectation (even for people who subscribe to environment-only models). It is because there is a very strong relation between age and WM r=.67.

    The results vary somewhat depending on how one does it and how one deals with the missing data. Sample size isn't large enough for certain conclusions, but hopefully the method will be sound, so it can quickly be used on a future dataset. I hope we can find some more data and use the cumulative dataset to get more statistical power.

  3. I apologize for going a bit off topic here -- I came from Steve Sailer's blog -- but may I ask Emil, are you descended from the famous Søren? Are the Kierkegaards as a family still very active in the arts and sciences?

  4. When a child first starts schooling there is not much to learn and remember. As the child progresses to a higher class the studies increase too. The memory of a person tends to weaken as they grow older. By the time a student reaches the college level, there is a lot of information that needs to be retained. College students use the aid of memory boosters or Nootropics to help them perform better. Nootropics the cognitive enhancers supplement