A novel take on group differences in test scores. Part I

Adding culture to the genes vs environment debate about IQ provides a new way to see this issue.

Homo sapiens: The Cultural Primate

This is the first of a two-part post one the issue how IQ can show different average values between racial groups and be highly genetically heritable without genetics being responsible. The argument about this issue has revolved around genes versus environment and has ignored a third potentially very significant factor: culture. Genetic effects are assumed to be invariant, while environmental effects operate in real time. Culture is affected by past environments through a cultural evolutionary process that can be crudely represented by a model like that previously use for business culture. Culture can be inherited as are genes and is an important element in the definition of the groups under discussion. Hence it makes sense to bring it into the issue, which is what these posts seek to do.

IQ is a measure of intelligence which is typically seen as a relatively immutable individual characteristic, which heritable to a significant extant. It also differs between groups on average.¹ IQ research has identified something called general intelligence, or g, that correlates with a wide variety of mental tasks. Thus, a person who is good at a task like counting backwards will also be good at seemingly unrelated tests of vocabulary, pattern recognition or logic problems. An IQ score obtained by a consensus of many such tests can provide an estimate for g.  That is, IQ is directly related to g:

(Eq. 1) IQ = k * g

where k is some constant relating the two. Although g is correlated with all types of subtests, the degree of correlation with a particular test (called its g-loading) differs, being stronger with some kinds of tests than others. Tests also differ in how much cultural information is necessary in order to do them. For example, English and Spanish speakers should be able to perform a test of how fast they can turn off a blinking light equally well, while a test of analogies given in one language will bias the results towards fluent speakers of that language. Tests like the latter are culturally loaded because they presuppose pre-existing cultural knowledge, whereas tests like the former are not. Thus, different tests will have different degrees of cultural loading and g-loading.

A recent study has shown that, with both adults and children, tests showing a higher g-loading tend to be those with a higher cultural loading. The same study also showed that culturally-loaded tests with adults show greater heritability than unloaded tests. We can interpret these results in light of cultural evolution. Recall the previous discussion about the various channels through which people acquire potentially useful cultural information. Three of these (frequency, prestige, and identity biases) involve no cognitive effort to implement, our brains can employ these selection heuristics more or less instinctually. On the other hand, direct bias requires an assessment of the relative quality (i.e., propensity to lead to success) of various cultural models, which is enhanced by higher cognitive ability. High quality cultural knowledge will aid in the acquisition of symbolic prestige markers, which leads to increased cultural offspring via the prestige bias. As animals, we have an inborn drive to produce biological offspring. As Homo sapiens, the cultural primate, we have a desire to produce more cultural offspring by acquiring prestige. Thus, most people have a “success drive” as well as a sex drive. Higher intelligence can help achieve success by improving the application of direct bias.

From a cultural evolutionary sense, one’s ability to choose good models to copy and ideas to acquire in order to achieve success is largely what g is for. In this sense, we can think of g as the “smarts” needed to game the system in order to achieve an advantageous position in the social hierarchy, which allows for production of more cultural offspring (that is, be adaptive in the context of cultural evolution). To do this requires detailed knowledge of what is important to this endeavor, plus natural talent. Let us call this natural talent raw intelligence (I). General intelligence, g, is then a product of cultural knowledge on how to succeed in a given culture (C) and the ability (I) to use this knowledge:

(Eq. 2) g = C * I

I is purely genetic, how much a person has is what they were born with. C is specific to a given culture, how much C is possessed by an individual depends on both the culture and I. Thus, C is a culture-specific function of I, or C(I). Substituting equation 2 into equation 1 gives:

(Eq. 3) IQ = k*g = k*C*I                                                                                                  

IQ is a strong function of I because it is proportional to I and the C component is positively related to I. Non-culturally-loaded tests would be expected to measure I well, but not C. Culturally-loaded tests would measure both. This means g should be most strongly correlated with culturally-loaded tests, and such tests, having the most I-component, should have the greatest degree of heritability.

General intelligence will both affect genetic and cultural evolution and be affected by them. For most of human evolution, when humans lived in small groups with small hierarchies to navigate, it was mostly the natural environment affecting evolution. Reproductive success would largely reflect individual learning and assessment of the quality of other people’s knowledge. Both are aided by higher I. A rare genius might discover a bit of new useful cultural information (e.g. an improved way to straighten spears) through individual learning and use it to achieve hunting success, resulting in increased mating opportunities and more biological offspring. The new knowledge would simultaneously rapidly spread to the group via the prestige bias, and the entire group would thus become smarter (higher average C) and more successful relative to other groups. Thus, g was adaptive both genetically and culturally, and over time both I and C grew.

Over time, as group size increased, the environment in which humans evolved became mostly culturally constructed, and success became based more on interactions with society than with the natural environment. Genetic success became more associated with one’s inherited social status as opposed to one’s mastery of the social and natural worlds, and g ceased to be genetically adaptive, although it remained culturally adaptive.

A recent study of the evolutionary history of human brain size shows that it increased over most of human evolution until about 3000 years ago, after which it decreased. This finding suggests that the environment in which human IQ evolution occurred changed sometime around 1000 BCE, after which it no longer proceeded by biological mechanisms affecting brain size. Given that a substantial fraction of the world’s population was living in complex societies by that time, it is reasonable to suppose that culture had become the sole evolutionary environment affecting IQ evolution after that time, as reflected by the trend change in brain size.

After this shift, g continued to increase via its C component rather than I. Individuals with high I continued to employ individual learning to acquire new useful knowledge in the pursuit of success. Success led to the dissemination of this knowledge to the group through the prestige and frequency bias. It did not necessarily lead to more biological offspring, however. Thus, the inventor’s memes, rather than their genes, propagated throughout the population.

The phenomenon just described is a process known as group selection. Group selection works with cultural evolution, because the development of a cultural variant that increases an individual’s success will rapidly spread through the group through prestige bias, while the spread to other groups is suppressed by identity and frequency bias. Examples of resistance to outsider ideas would be the “not invented here” (NIH) syndrome in technology, or “gatekeeping” in popular culture fandoms. New memes appearing in a one group will tend to stay within that group long enough for them to be adopted by the entire group. Consequentially, the cultural inventory of the entire group evolves like the genome for a single organism. Entire populations (cultures) evolve in competition with other cultural groups, giving rise to group differences in the cultural component of g.

Group selection does not happen in genetic evolution among groups.² This is because a success-improving gene does not rapidly spread through the group in which it appeared. Long before the entire group can gain the success gene and use it to better compete against other group, it will “leak out” to other groups. Gene exchange does not seem to suffer from a sexual version of “not invented here” or gatekeeping mentality. Thus, advantageous genes will spread to other groups before they take over any one of them. Mutations will provide an adaptive advantage only to individuals, not to the group to which they belong. Over time, the more adaptive genes will become common variants throughout the human population and eventually dominant. Thus, genetically-evolved I will tend not to vary much among cultural groups (e.g. ethnicities), while culturally-evolved C will.

The processes just described explain why it is very likely that differences in IQ between groups are due to the cultural component C and not the genetic component I. With this, we divide out the constant I from equation 2 to obtain:

(Eq. 4) Avg. IQ = k g_G = k’ C_G

Here C_G is the average value of C(I) for a group. Equation 3 implies that group average IQ can be used as a proxy for the underlying variable C_G.

Figure 1 National C_G values versus per capita GDP

Over time, different peoples evolve different C functions based on their histories. Societies having more routes to success to navigate select for higher C_G than societies with fewer routes. A reasonable proxy for number of routes might be economic output per person. Richer societies will have a greater division of labor and more routes, implying higher C_G (and presumably, higher average IQ). To test this idea, Figure 1 plots average IQ scores (as a proxy for C_G) for various nations against GDP per capita (GPDpc). A rising trend in IQ with rising GDPpc can be seen up to a critical value, above which there seems to be no correlation. The data in the figure are presented relative to this critical value. An implication of this data is that as national economies grow and GDPpc increases over time, C_G (and average IQ) should rise. As it turns out, this is exactly what is observed. New Zealand intelligence researcher James Flynn has noted that population-average IQ rises over time by about three points per decade.³

To be continued in Part II.

References

1.      Encyclopedia of Diversity in Education. SAGE. 2012, 1209

2.      Boyd, Robert and Peter Richardson, Culture and the Evolutionary Process. (University of Chicago Press, 1985), 231.

3.      Flynn James R., Are We Getting Smarter? Rising IQ in the Twenty-First Century. (Cambridge University Press, 2012