Arthur R. Jensen (2000) Race Differences, g, and the "default Hypothesis". Psycoloquy: 11(004) Intelligence g Factor (24)

Volume: 11 (next, prev) Issue: 004 (next, prev) Article: 24 (next prev first) Alternate versions: ASCII Summary
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 11(004): Race Differences, g, and the "default Hypothesis"

Reply to Locurto on Jensen on Intelligence-g-Factor

Arthur R. Jensen
Educational Psychology
School of Education
University of California
Berkeley, CA 94720-1670


Locurto (1999) is in essential agreement with my position on the psychometric aspects of g, its diverse nonpsychometric correlates, and its practical predictive validity in education, employment and other real-world outcomes, but he questions my "default hypothesis" concerning the causal basis of racial differences in g. It is encouraging that he is willing to discuss this subject at all, as the discussion of racial differences in any behavioral traits has become generally taboo in academe.


behavior genetics, cognitive modelling, evoked potentials, evolutionary psychology, factor analysis, g factor, heritability, individual differences, intelligence, IQ, neurometrics, psychometrics, psychophyiology, skills, Spearman, statistics
1. Locurto (1999) fears that my book (Jensen, 1998, 1999) will be studiously ignored because it deals in part with racial differences in mental abilities, particularly g. I agree that, so far, this has happened to some extent, considering the relatively few published reviews (as compared with the number of reviews received within the same period of time in 1980 by my book "Bias in Mental Testing" ]Jensen, 1980a, b], which never mentioned race differences in IQ or g in connection with genetics).

2. The initial rejection of the present book (Jensen, 1998) by several publishers was very clearly because of Chapter 12, in which I rather fully discussed race and g in the context of genetics. I put forth a hypothesis that I think most parsimoniously explains the findings connected with the Black-White (B-W) group difference in g. According to the "default hypothesis," the distribution of B-W differences in g results from the same mix of genetic and nongenetic factors that cause the distribution of individual differences in g within each group. Thus group differences in g are viewed simply as aggregated individual differences. The total population can be divided into many different subpopulations in terms of various criteria without reference to g, and yet, for various reasons, the mean level of g in these aggregations of individuals will often differ significantly. But the mean differences are all made up of the "same stuff" as all the individual differences that went into the aggregates.

3. In studying representative samples of B and W populations, I find no unique differences; that is, there appear to be no differences BETWEEN the groups in the external environmental and achievement correlates of g that do not hold for individual differences WITHIN each population when the level of g is taken into account. In recent years this generalization no longer holds for certain outcome variables that are were targeted by the institution of Affirmative Action, such as selection in higher education and employment and promotion in higher-level occupations.

4. For example, in large samples of middle-aged men, it was found that the linear regressions of both job status and annual income on g factor scores showed that in the region of the bivariate distribution above a g score equivalent to about IQ 100 (i.e., the average IQ of the W population) Bs' job status and income are increasingly higher than that for Ws. Below that g cut-score at about IQ 100 (i.e., the cross-over point of the regression lines for B and W populations), Bs have slightly lower incomes (but not lower job status) than Ws at the same level of g (Nyborg & Jensen, in press).

5. In general, the whole nexus of social variables correlated with g is much the same for Bs and Ws. Thse who hold that there exist no genetic differences in g have to exaggerate the social-environmental differences in order to explain the relatively small average racial differences in g, which, within similar socioeconomic groups are about the same as the average difference between full siblings reared together in the same family. The problem is not the average difference per se, but the statistical consequence of differences in the degrees of overlap between the B and W bell curves, i.e., the difference in the proportions of each population that fall into different segments of the IQ scale that are statistically associated with desirable or undesirable real-life outcomes, such as in education, employment, job status, and income. It is troubling to contemplate the typical contrasts on these variables between individuals falling more than one standard deviation (SD) above and one SD below the overall population mean level of g, regardless of their ethnic or racial classification.

6. If, as Locurto writes, my book is racially tinged, what does that imply, other than that I take a careful look at the question of race differences in g? There are a number of medical books and articles (cited in Jensen, 1998) on race differences in disease rates, dosage responses, racial genetic factors in blood pressure, and the like. Are these investigations also racially tinged ? The study of genetic variation in human populations is not racism, whether applied to physical or behavioral variables. True, theories of the evolution of racial variation in most traits, including mental abilities and other behavioral traits, are quite speculative and new theories are still being suggested. Most of them, perhaps all, will remain speculative, as they are either exceedingly difficult or even impossible to test. That is why I consider evolutionary explanations of racial variation in g as much less liable to refutation than findings derived from the direct study of biological correlates of g in contemporary populations. In view of the tenor of our times, I sympathize with younger scientists for shunning my views on this topic or for not pursuing this topic for investigation themselves. As the worst that can possibly happen to me for doing so has already happened, I have nothing further to lose or fear about treating the subject empirically and objectively, just as I have done with every other behavioral phenomenon I have ever researched (Jensen, 1998b).

7. Extremely contrasted environments are mentioned (but then discounted) by Locurto as a possible candidate for a purely nongenetic account of the B-W difference in g. But does this conjecture not ignore Carnap's "total evidence rule"? How extreme are the environments of middle-class Bs and Ws, or the differences between the environments of W and B children adopted into upper-middle class homes? Yet the average B-W IQ differences in these conditions of rearing remain in the range of 0.7 to 1.0 standard deviations (reviewed in Jensen, 1998, Chapter 12). If one does not view all the items of evidence simultaneously, one too easily dismisses its full impact.

8. The brain-size correlation with IQ is indeed just a linear one (+.30 to +.40 in several MRI studies); but that's a limitation. If there are any significant nonlinear components present, taking them into account could only increase the overall correlation (i.e., multiple R) between brain and IQ. The linear correlation itself is far from trivial. However, a correlation of this magnitude means that within the normal distribution of IQ a larger brain size is neither necessary nor sufficient for higher levels of IQ. Adult midgets, with head size like that of pre-school children, have IQs completely on a par with average sized adults (Skoyles 1999). The midgets' head size measurements therefore greatly underestimate their IQs. The same regression of IQ on head size just does not apply to midgets and the rest of the population. What remains uncertain about head size and IQ in racial comparisons is not whether there is a mean difference, but whether one and the same regression line predicts IQ equally well within and between all the main racial groups.

9. It has already been shown that the overall brain size is less strongly correlated with IQ than are certain areas (or cross-sectional MRI slices) of the brain (Willerman et al., 1991), but I am not aware of any racial comparisons on this basis reported in the brain-imaging literature. The evidence for the brain-size/IQ correlation and the brain-size/race differences, in contrast, is now well established. At present, however, these two distinct items of evidence, although wholly consistent with the hypothesis that race differences in g have a biological component, can at present be regarded only as weak and tentative evidence for this hypothesis. It would gain greatly in importance if the density of neurons in those parts of the brain most highly correlated with g within groups were found not to differ between groups to such a degree as to offset the already established racial differences in brain size. In the case of the sex difference in brain size, the greater packing density of neurons in the female brain almost exactly offsets the sex difference in overall brain size. This evidence of a male-female difference in the packing density of neurons may account for the virtual absence of any g difference between males and females despite their average difference (about 80 cc) in brain size (Jensen, 1998, Chapter 13).

10. Locurto's pessimistic view of the possibility of demonstrating the correctness of any theory of the cause of racial difference in g is too much like the same attitude seen throughout the history of science, in connection with the heliocentric model, the speed of light, the conservation of energy, evolution, the particulate inheritance of traits, color-vision, relativity, and the big bang. Their predictions seemed impossible to demonstrate empirically until they actually were demonstrated through scientific ingenuity, and always much technical trial-and-error. At present I don't know of any laws of nature that should make it impossible to test the gist of the "default hypothesis" that I have suggested. I find more evidence that is consistent with it than evidence that contradicts it. It should be pursued further, by predicting its many possible consequences and determining whether they are borne out.

11. My latest empirical research in this vein concerns the congruence of a set of complex psychometric differences BETWEEN different racial populations with the same set of differences between older and younger age groups WITHIN each racial group. A high degree of congruence would pose no problem for my default hypothesis, but the opposing social-causation and cultural-difference hypotheses would have to invoke the purely ad hoc postulate that social and cultural differences between groups perfectly simulate developmental psychometric differences within the groups. Within-group differences in g represent individual differences in mental growth trajectories. The default hypothesis predicts that the B-W difference in the pattern of psychometric differences for same-age children should, on average, look the same as the pattern of differences between older and younger children of the same race.

12. The progressive scientific theory is evaluated by its predictable implications and whether they are borne out in a variety of contexts. A degenerating theory finally collapses from the burden of the increasing body of inconsistent ad hoc speculations needed to explain away its predictive failures. I believe it more likely than not that the strictly environmentalist and nonbiological theories of individual and group differences in g will eventually meet the fate of a degenerating theory, and future historians will note the social and political ideologies prevailing in the twentieth century that made it take so long.

13. After expressing his regrets about my position on the hypothesized genetic and environmental causes of the B-W difference in g, Locurto's paragraphs 13 through 22 are a succinct and cogent defense of most of the other key topics in my book. He understands my thinking on all these points most admirably, and I greatly appreciate finding such large areas of agreement from one who is expert in this field. It encourages me that further investigations, if unhindered by ideological prejudice, will eventually lead to resolution and consensus on the big but still unsettled questions concerning the nature and causes of individual and group differences in g.


Jensen, A.R. (1980a). Bias in mental testing. New York: Free Press.

Jensen, A.R. (1980b). Precis of "Bias in mental testing" Behavioral and Brain Sciences 3(3) 325-372.

Jensen, A.R. (1998). The g factor: The science of mental ability. Westport, CT: Praeger.

Jensen, A.R. (1998b). Jensen on "Jensenism." Intelligence, 26, 181-208.

Jensen, A.R. (1999). Precis of: The g Factor: The Science of Mental Ability PSYCOLOQUY 10(23). psyc.99.10.023.intelligence-g-factor.1.jensen

Locurto, C. (1999). A balance sheet on persistence. PSYCOLOQUY 10(059) /psyc.99.10.059.intelligence-g-factor.9.locurto htp://

Nyborg, H., & Jensen, A.R. (In press). Job status and income as a function of g. Intelligence.

Skoyles, J.R. (1999). Human evolution expanded brains to increase expertise capacity, not IQ. PSYCOLOQUY 10(002). psyc.99.10.002.brain-expertise.1.skoyles

Willerman, L, Rutledge, J.N., & Bigler, E.D. (1991). In vivo brain size and intelligence. Intelligence, 15, 223-228.

Volume: 11 (next, prev) Issue: 004 (next, prev) Article: 24 (next prev first) Alternate versions: ASCII Summary