Developing Intelligence

Your ability to control thought and behavior relative to your peers – a set of capacities known as “executive functions” – is almost entirely genetic in origin, according to a newly in-press paper from Friedman et al. Over 560 twins completed tests to measure fundamental components of these executive functions, and the results were analyzed in terms of how similar identical twins performed to one another relative to fraternal twins (all twins in the study were reared together). Astonishingly, the results show that the variance common to all executive functions is correlated roughly twice as much between identical twins as between fraternal twins, and that individual variance in executive function falls directly in line with what would be expected from a perfectly heritable trait.

The components of executive function (as determined through previous latent variable analyses) can be loosely described as inhibition (the ability to resist habit), updating (the ability to quickly change the focus of attention or the contents of working memory), and shifting (the ability to quickly change goals and respond appropriately). Previous work demonstrated distinct patterns of individual performance on the tasks thought to measure each of these capacities, suggesting that they are partially dissociable but also that they contain substantial overlap with one another – thus there is both “unity” and “diversity” to the construct of executive function. This work requires the use of a statistical technique known as latent variable analysis, in which consistent covariation between various measures is extracted into a “latent” or “unobserved” variable which theoretically represents a purer measure of cognitive constructs that is relatively uncontaminated by measurement error. The same tasks and statistical techniques were used in the current study, in addition to a structural equation model known as “ACE” for partitioning variance into genetic, shared environment, and non-shared environmental influences.

This combination of statistical tools is powerful, because it allows the authors to fit the coefficients of a particular equation (the structural model) to predict the performance of individual twins on these tasks. For example, for identical twins, the “A” component refers to genetic influences and must be perfectly correlated between these twins (and thus is set to “1″), whereas among fraternal twins it’s set to .5 (since they share only half their genes). The “C” component refers to shared environmental influences, which is also set to “1″ for both kinds of twins, since they were all reared in the same household. Finally, the “e” component refers to unshared environmental factors, which by definition are uncorrelated between twins.

Once these structural models are in place, we can examine the extent to which inter-twin correlations differ between identical and fraternal twins. For example, if primarily environmental factors are involved, then correlations may be similar between identical and fraternal twins, and the “C” model parameter will be highly significant in the structural model. If genetic factors are mostly involved, then scores among fraternal twins should be correlated no more than half as strongly as correlations among identical twins, and the “A” parameter will be highly significant in the structural model.

The results from this approach are jaw-dropping: variance shared among each variety of executive function (inhibition, updating, and shifting) is nearly perfectly heritable: the contribution of the “A” component to those correlations is 99%. This heritable variance in the common executive function predicts nearly all of the genetic variance in the inhibition factor, consistent with the idea that those constructs are isomorphic from a heritability standpoint. Second, genetic influences on updating and shifting were roughly half due to the common executive function (43% and 44%, respectively) and half due to unique genetic influences (56% and 42%, respectively). Thus, the overall picture is that executive functions, in both their unity and diversity, are somewhere between 86 to 100% heritable.

Furthermore, Friedman et al. integrated measures of general intelligence (“g”, estimated through the WAIS IQ test) and perceptual speed (essentially the speed with which subjects can complete very simple tasks) to show that the genetic contribution to executive function is not completely explained by genetic contributions to those more commonly-studied abilities. This is consistent with previous work showing that IQ is only moderately heritable (with 50-70% of variance explained due to genetic factors, far short of the 99% explained here).

The authors conclude that environmental influences can not be “substantial contributors to the unity and diversity of executive functions” (i.e., the apparent structure of executive functions is due to different patterns of genetic contribution). In fact, this study shows that executive functions are among the most heritable cognitive traits ever discovered – nearly completely genetic in origin.

Yet Friedman et al are careful to point out that these results do not mean that environment simply doesn’t matter – environmental influences can change the mean value of any pair of twin’s score, but the extent to which two twins share that deviation from the mean is almost entirely attributable to genetic influences.

Of course, there are some limitations to the study:

- Gene-environment interactions are not accounted for here; in other words, genes may predispose individuals to seek out certain environments, which would be lumped into estimates of genetic variance in the current study

- Although this was a community-wide sample of twins, environmental influences might appear artificially small in samples which come from the same SES. So these estimates might differ if conducted in third-world countries, where environmental influences might curtail the full expression of genetic variance in executive functions

- Although it’s possible that identical and fraternal twins do differ slightly in the degree to which they share environments (e.g., if identical twins have their similarity exaggerated by dressing similarly, and if fraternal twins seek to differentiate each other), scientific estimates of these factors seem to have minimal influence on heritability estimates

One important future direction for this work is the determination of the specific genes which contribute to executive function and their mechanism of action. Such research could pave the way directly towards more accurate computational modeling and perhaps even direct pharmacological manipulation of executive functions.

Friedman, N. P., Miyake, A., Young, S. E., DeFries, J. C., Corley, R. P., & Hewitt, J. K. (in press). Individual differences in executive functions are almost entirely genetic in origin. Journal of Experimental Psychology: General.

Comments

  1. #1 Russell
    May 13, 2008

    That’s… stunning.

  2. #2 PENIX
    May 13, 2008

    GENETICS FTW! ETHNIC CLEANSING FTW!

  3. #3 Pen
    May 13, 2008

    I understood all of that except the last bit about press embargos. Could you explain, please?

  4. #4 CHCH
    May 13, 2008

    Pen is referring to something that I’ve removed; I posted this before this paper was available online, and sometimes journals will request that the information not be released to the public until they have issued a press release on a certain day. that’s all…

  5. #5 Boris Kazachenko
    May 14, 2008

    I suspect that a good part of “executive function” is simply a longer “intellectual horizon” & a corresponding preference to operate on a higher generalization level. I will also go on limb & suggest that it’s largely determined by the size of cortical minicolumns & the resulting range of intracortical connections, which are also genetically determined. This would make an executive function a complimentary of “autism phenotype” subset as a minicolumnar disorder, per Dr. Casanova: http://en.wikipedia.org/wiki/Manuel_Casanova. That’s a testable hypothesis, do know of any evidence? I have a related “Generalist vs Specialist neuroarchitectural bias” post on my blog http://scalable-intelligence.blogspot.com/2008/04/analytical-depth-vs-response-speed.html.

  6. #6 Hank Roberts
    May 14, 2008

    To make the “scalable-intelligence” link work, lose the final period, which confuses the Web:
    http://scalable-intelligence.blogspot.com/2008/04/analytical-depth-vs-response-speed.html

    I’m very curious to know if these same research tools give an idea of how _plastic_ or teachable the ‘executive function’ is, and whether the various new tools for keeping older people’s cognitive status from declining would apply to that.

    Please hurry (grin)

  7. #7 stewart
    May 14, 2008

    Interesting, but how does this square with Salthouse’s notion that ‘executive functioning’ is really just general intelligence?
    Your comment on the environment is important. Environmental effects on measures such as verbal fluency, concept formation, and task switching are profound, as demonstrated by the large Flynn effects found for these tasks, even after accounting for age and education.

  8. #8 Hank Roberts
    May 15, 2008

    That latter link is broken by the trailing period.

    Lose that and it will work:
    http://scalable-intelligence.blogspot.com/2008/04/analytical-depth-vs-response-speed.html

  9. #9 Simfish InquilineKea (simfish@gmail.com)
    May 16, 2008

    It’s possible that academic achievement is highly heritable for factors totally unrelated to intelligence. And executive function is *far* easier to improve than intelligence.

    And holy crap – I’ve just realized – a lot of measurement tests (that might be used for IQ tests) measure executive function (or maybe just indirectly measure them – does anyone know if ADD negatively affects IQ for reasons unrelated to intelligence?). It’s entirely possible that much of the correlation between IQ and genetics could be attributed to the heritability of executive function. This could be somewhat good news for those who don’t want their intelligence to be imprisoned by their genes (as executive function is easier to “correct” than intelligence).

  10. #10 Kapitano
    May 17, 2008

    individual variance in executive function falls directly in line with what would be expected from a perfectly heritable trait.

    When your statistical methods give results that are statistically impossible, you’ve learned one thing – that something, somewhere, is seriously wrong with your method.

    The last time twin studies produced numbers this neat, Cyril Burt was involved.

  11. #11 Boris Kazachenko
    May 20, 2008

    Thanks Hank!

    “Executive function” is a fuzzy concept, the way they define it seems to refer to a specific “operational” attention span. But if you want to improve focus and memory, that’s done by upregulating, respectively, dopaminergic and cholinergic function. Ask Chris :).

  12. #12 Charlie
    October 25, 2009

    To what extent are individual differences in intelligence determined by variability in executive function?

  13. #13 recette dukan
    September 15, 2011

    Apparently this web-site does not load very well using a Motorola Droid Pro. Are other folks having the same problem ?