IQ and the Health of States Kanazawa*

IQHSBI and the Health of States

SatoshiIQ and the Health of States Kanazawa*

Department of Management, London School of Economics and Political Science; Department of Psychology, University College London; Department of Psychology, Birkbeck College, University of London

ABSTRACT: Epidemiologists contend that income inequality reduces the health and life expectancy of the whole population, but this argument does not make sense within its own evolutionary framework. Recent evolutionary psychological theory suggests that the human brain, adapted to the ancestral environment, has difficulty comprehending and dealing with entities and situations that did not exist in the ancestral environment, and that general intelligence evolved as a domain-specific adaptation to solve evolutionarily novel problems. Since most dangers to health in the contemporary society are evolutionarily novel, it follows that more intelligent individuals are better able to recognize and deal with such dangers and live longer. Consistent with the theory, and replicating an earlier study of cross-national data, income inequality has no effect on the health and longevity of the population across the American states, when the racial composition (percent black) is controlled, but the average intelligence of the population (state IQ) has a significant effect. The data presented here and in the earlier study challenge the conclusion that income inequality reduces the health of the population.

INTRODUCTION damage health. Wilkinson’s conclusion is succinctly captured in the blurb on his In a series of articles and books, Richard book Mind the Gap, “Inequality kills. Peo- Wilkinson argues that economic inequal- ple die younger in countries with greater ity reduces the health of the population inequalities in income.” While Wilkinson’s and lowers its life expectancy (Marmot original observation on the negative asso- & Wilkinson, 1999; Wilkinson, 1992, ciation between income inequality and 2000). He claims that humans and other health involves cross-national comparisons primates have an evolved physiological of nations (Wilkinson, 1992), others have mechanism whereby their cortisol level since documented the same negative asso- goes up when they are under attack or ciation for subnational units, such as the otherwise in submissive situations. American states (Ben Shlomo, White, & Heightened levels of cortisol and other Marmot, 1996; Kaplan et al., 1996; Kawachi stress hormones allow the individuals to & Kennedy, 1997; Kennedy, Kawachi, & deal with the short-term emergencies but Prothrow-Smith, 1996; Lynch et al., 1998; at the cost of long-term health. When the Ross et al., 2000). The negative conse- submissive status prolongs, the continu- quences of income inequality for popula- ously high levels of stress and anxiety tion health and longevity has become part of the established knowledge of epidemiol- *Address correspondence to: Satoshi Kanazawa, ogy and public health. Interdisciplinary Institute of Management, London Wilkinson’s theory, however, does not School of Economics and Political Science, Houghton Street, London WC2A 2AE, United Kingdom. E-mail: make sense within its own evolutionary [email protected] framework. Given that prolonged stress

200 IQ and the Health of States 201

and anxiety lead to health problems and African savanna during the Pleistocene early death, with no apparent compensating Epoch). It may therefore have difficulty benefits in reproductive success,1 any comprehending and dealing with entities genetic mutation that allows its carrier not and situations that did not exist in the to experience stress in the face of perma- ancestral environment (Kanazawa, 2002, nently low status from which it cannot 2004a). On the other hand, an evolution- escape (the “Que sera sera” gene?) will be ary psychological theory of the evolution selected. Chronic low status, being at the of general intelligence2 proposes that bottom of a social hierarchy for life, general intelligence may have evolved as seems dramatically different from the a domain-specific adaptation to solve acute emergency of imminent physical evolutionarily novel problems, for which attack, with which cortisol and other there are no predesigned psychological stress hormones are designed to deal. Any adaptations (Kanazawa, 2004b, 2008a). genotype which makes a distinction The synthesis of these two theories, the between short-term emergencies of physi- Savanna-IQ Interaction Hypothesis, cal attack and long-term chronic low sta- implies that the human brain’s difficulty tus, and which responds differently, with with evolutionarily novel stimuli may a surge of cortisol, prolactin and other interact with general intelligence, such stress hormones to the former, but not to that more intelligent individuals have less the latter, will be favored by natural selec- difficulty with evolutionarily novel stim- tion over another genotype which does uli than the less intelligent individuals. In not make such a distinction and responds contrast, general intelligence may not similarly to both. Since all primate societ- affect individuals’ ability to comprehend ies are hierarchical and there are chroni- and deal with evolutionarily familiar enti- cally high- and low-status individuals in ties and situations. them, suggesting that the origins of Some critics (including one anony- human social hierarchies may go back at mous reviewer) contend that general least 15–20 million years in evolutionary intelligence could not be an adaptation history, it appears that there should have because it is an individual difference vari- been enough time for such a genetic able. Adaptations are universal and con- mutation to emerge and spread. What is stant features of a species shared by all its the ultimate function of stress and anxiety members; in contrast, there are obviously in the face of low status? If there are heritable individual differences in general none, why hasn’t natural selection elimi- intelligence and some individuals are nated it? more intelligent than others. These critics Recent evolutionary psychological argue that adaptations and heritable indi- theory suggests an entirely different vidual differences are mutually exclusive. determinant of health and life expectancy I strongly disagree with these critics. in contemporary society. On the one I believe that a trait could simultaneously hand, evolutionary psychology (Craw- be an evolved adaptation and an individ- ford, 1993; Symons, 1990; Tooby & ual difference variable. Full-time bipe- Cosmides, 1990) posits that the human dalism is a uniquely human adaptation brain, just like any other organ, is among mammals, yet some individuals designed for and adapted to the conditions run faster than others. The eye is a complex of the ancestral environment (roughly the adaptation, yet some individuals have 202 Kanazawa better vision than others. Language is an general intelligence is an adaptation and adaptation, yet some individuals have has zero heritability (in the sense that all greater linguistic facility than others. humans are highly intelligent), even Individual differences in general intelli- though the exact level of an individual’s gence is what Tooby and Cosmides intelligence is not an adaptation and is (1990) call “random quantitative variation highly heritable. And Tooby and on a monomorphic design.” “Because the Cosmides (1990, p. 57) contend that elaborate functional design of individuals “nonadaptive, random fluctuations in [in this case, general intelligence as a the monomorphic design of a mental domain-specific adaptation] is largely organ can give rise to heritable individ- monomorphic [shared by all members of ual differences in nearly every manifest a species], our adaptations do not vary in feature of human psychology.” We their architecture from individual to indi- would therefore expect individual dif- vidual (except quantitatively)” (Tooby & ferences in general intelligence as a Cosmides, 1990, p. 37, emphasis added). domain-specific adaptation. Intraspecific (interindividual) differences As it happens, there has been accumu- in such traits pale in comparison to inter- lating evidence for the Savanna-IQ Inter- specific differences. Carl Lewis and I run action Hypothesis. First, individuals’ at a virtually identical speed compared to tendency to respond to TV characters as if cheetahs or sloths. Similarly, Einstein and they were real friends, first discovered by I have virtually identical intelligence Kanazawa (2002), is limited to those with compared to cheetahs or sloths. I there- below-median intelligence (Kanazawa, fore believe it is possible for a trait to be 2006a); individuals with above-median universal and species-typical (exhibiting intelligence do not become more satisfied virtually no variation in a cross-species with their friendships by watching more comparison) and manifest vast individual television. differences among members of a single Second, less intelligent individuals species. I believe general intelligence is have more children than more intelligent one such trait. individuals, even though they do not want Tooby and Cosmides (1990, pp. 38–39) to, possibly because they have greater make this exact point, using “a complex difficulty effectively employing evolu- psychological mechanism regulating tionarily novel means of modern contra- aggression” as their example. They con- ception (Kanazawa, 2005). Another tend that it is an adaptation, even though indication that less intelligent individuals there are heritable individual differences may have greater difficulty employing in the mechanism’s threshold of activa- modern contraception effectively is the tion, whether one has a “short fuse” or fact that the correlation between the life- not. Tooby and Cosmides suggest that a time number of sex partners and the num- complex psychological mechanism reg- ber of children is positive among the less ulating aggression “is (by hypothesis) intelligent but negative among the more universal and therefore has zero herita- intelligent. The more sex partners less bility,” even though “the variations in intelligent individuals have, the more the exact level at which the threshold of children they have; the more sex partners activation is set are probably not adapta- more intelligent individuals have, the tions.” I would similarly propose that fewer children they have. IQ and the Health of States 203

Third, net of age, sex, race, education, each other, if they could get away with it. income, and religion, more intelligent Second, the institutions that control, individuals are more likely to acquire evo- detect, and punish criminal behavior in lutionarily novel values and preferences society today -- the police, the courts, and such as liberalism, atheism, and, for men, the prisons -- are all evolutionarily novel; sexual exclusivity, than less intelligent there was probably very little formal individuals (Kanazawa, 2008b). Consis- third-party enforcement of norms in the tent with the Hypothesis, however, gen- ancestral environment, only second-party eral intelligence has no effect on the enforcement by victims and their kin and acquisition and espousal of evolutionarily allies. Thus it makes sense from the per- familiar values for children, marriage, spective of the Savanna-IQ Interaction family, and friends. Hypothesis that men with low intelligence Fourth, net of age, race, sex, education, are more likely to resort to evolutionarily income, religion, marital status, number familiar means of competition for of children, and social class, more intelli- resources (theft rather than full-time gent individuals are more likely to prefer employment) and mating opportunities evolutionarily novel purely instrumental (forcible rape rather than computer dat- music (such as classical, big band, and ing) and not to comprehend fully the con- easy listening) than less intelligent indi- sequences of criminal behavior imposed viduals, while general intelligence has no by evolutionarily novel entities of law effect on individuals’ preference for evolu- enforcement. tionarily familiar vocal music (Kanazawa & Most dangers to health and life today Perina, 2008). In addition, the Hypothesis (cigarettes, alcohol, junk food, sedentary can potentially explain why general intel- life, automobiles, guns) are evolutionarily ligence is positively correlated with open- novel; very few individuals in the United ness to experience (i.e. novelty) across States today die of being mauled by wild individuals (Ackerman & Heggestad, 1997; animals or falling from a cliff. While our Chamorro-Premuzic & Furnham, 2006). ancestors in the African savanna may Finally, criminologists have long have partaken in psychotropic drugs and known that criminals on average have intoxicating substances, since their use is lower intelligence than the general popu- known among contemporary hunter-gath- lation (Wilson and Herrnstein 1985; Her- erers, they certainly did not have anything rnstein and Murray 1994). From the nearly as potent and thus potentially dan- perspective of the Savanna-IQ Interaction gerous as crack cocaine or vodka. The Hypothesis, there are two important Savanna-IQ Interaction Hypothesis would points to note (Kanazawa, Forthcoming-a, therefore predict that more intelligent Forthcoming-b). First, much of what we individuals can better recognize evolu- call interpersonal crime today, such as tionarily novel health dangers and risks, murder, assault, robbery, and theft, were deal with them appropriately, and remain routine means of intrasexual male compe- healthier and live longer. tition in the ancestral environment. This is Consistent with this prediction, Deary how men competed for status, resources, et al.’s (2004) longitudinal analysis of the and mating opportunities for much of Scottish Mental Surveys of 1932 and human evolutionary history; they beat up 1947 reveals that childhood IQ has a posi- and killed each other, and they stole from tive effect on longevity; more intelligent 204 Kanazawa

Scots live longer than less intelligent an evolutionarily familiar environment Scots. Elias et al. (2003) demonstrate that (such as hunter-gatherer tribes in sub- obese men (but not obese women) have Saharan Africa). lower intelligence than their nonobese The current analysis empirically tests counterparts. Kanazawa’s (2006b) analy- the Hypothesis at the macro level across sis of cross-national data shows that the 50 American states and the District of average intelligence of a population has a Columbia. If more intelligent individu- very large and significant effect on life als stay healthier and live longer than expectancy at birth, infant mortality rate, less intelligent individuals, as the and age-specific mortality rate across Hypothesis suggests, then it logically 126 nations. Neither economic develop- follows that, at the higher, macrosocial ment (GDP per capita) nor income level of aggregation, the average intelli- inequality (Gini coefficient) has any effect gence of state populations (“state IQ”; on health and longevity net of national Kanazawa, 2006c) should have a posi- IQ. Further consistent with the Savanna- tive effect on the health and longevity of IQ Interaction Hypothesis, national IQ the population. The present study aims does not have any effect on health and to replicate Kanazawa’s (2006b) cross- longevity in evolutionarily more familiar national results within the United States, sub-Saharan Africa, where only income to examine whether state IQs signifi- inequality has a significant effect. cantly influence their health and longevity This paper seeks to make a contribu- of state populations. tion to the emerging field of cognitive epidemiology (Deary & Der, 2005; Lubinski DATA & Humphreys, 1997). There have been RESPONSE VARIABLES numerous studies which show that more intelligent individuals stay healthier and I use four measures of health and lon- live longer, even net of education and gevity which have often been used in the social class (Deary et. al, 2004; Gottfred- past studies of the relationship between son, 2004; Whalley & Deary, 2001). income inequality and health: Life However, exactly why this is so is not expectancy at birth (the mean number of clear. Deary and Der (2005, p. 67) note years of life left remaining to a person at that the mechanism behind the correlation birth), age-adjusted death rate (the number between IQ and longevity is “as yet of deaths from all causes per 100,000 unknown”; Gottfredson and Deary (2004) population, adjusted for age by applying ask, in the title of their article, “Intelli- the age-specific rates to standardized age gence predicts health and longevity, but distribution in order to eliminate the why?” The Savanna-IQ Interaction effect of differences in population age Hypothesis can potentially offer one pos- structure), infant mortality rate (the num- sible mechanism by which more intelli- ber of deaths of infants under one year of gent individuals stay healthier and live age per 1,000 live births),3 and percent longer in the (evolutionarily novel) obese (the proportion of the state popula- industrial and postindustrial societies. tion that are classified as obese with body The Hypothesis simultaneously pre- mass index greater than 30). Data on life dicts that more intelligent individuals expectancy at birth in the 50 states and the do not stay healthier and live longer in District of Columbia in 2000 are available IQ and the Health of States 205

from the U.S. Census Bureau (http:// high school and not all high school www.census.gov/population/projections/ seniors take the SAT. SAT is a better MethTab2.xls). Data on age-adjusted measure of general intelligence than death rate and on infant mortality rate for either ACT or NAEP (National Assess- 2000–2002 are available from the ment of Educational Progress) (Frey & National Center for Health Statistics Detterman, 2004). Both ACT and NAEP (2005, pp. 168–169, Table 28, and p. 160, purport to measure knowledge acquired Table 23, respectively). Data on percent in the school curriculum (“crystallized obese in 2006 are available from the Cen- intelligence”), whereas SAT purports to ter for Disease Control (http://apps.nccd. measure reasoning ability (“fluid intelli- cdc.gov/brfss/list.asp?cat=OB&yr=2006& gence”), hence its original name, Scho- qkey=4409&state=All). lastic Aptitude Test (Kanazawa, 2006c, p. 594). While Koenig, Frey, and EXPLANATORY VARIABLES Detterman (2008) show that ACT may Income Inequality. I use Gini coeffi- also be used as a measure of general cient to measure the extent of income intelligence, SAT scores’ correlation inequality in each state. Gini coefficient with general intelligence is higher than is calculated with the formula: G = ACT scores’ (.87 vs. .77). See Kanazawa 1 m m (2006c) for the details of the estimation ∑∑ xx− , where x procedure, and the actual state IQ x ij i 2*mm *( -1)* i=1 j=1 estimates. is the units of income (e.g. dollar) that the individual i earns, and m is the total Racial Composition. The states vary number of individuals in the economy in the proportion of the population which (Foster, 1985, pp. 61–65). There are a is black, from the low of 0.4% in Montana total of r units of income in the economy to the high of 58.9% in the District of Columbia. For a variety of genetic, hor- ().∑ xi = r G = 0.0 indicates perfect monal, and developmental reasons, blacks income equality, where each individual are more susceptible to certain diseases earns the same income (r ), and G = 1.0 (Ellis & Nyborg, 1992; Polednak, 1989; m Ross et al., 1986), which contributes to indicates maximum income inequality, their higher mortality rate and shorter where one individual earns all r income in life expectancy (Rushton, 1995). The the economy. Data on Gini coefficients for white-black gap in infant mortality the states in 1999 are available from the exists even when both parents are U.S. Census Bureau (http://www.census. university-educated (Schoendorf et al., gov/hhes/www/income/histinc/state/state4. 1992), so it is not entirely a function of html). social class. It is therefore important to State IQ. State IQ is estimated from control for percent black in the popula- the SAT scores, using Johnson and tion in any aggregate comparison of Thomopoulos’ (2002) work on the char- health and life expectancy. Data on the acteristics of the left-truncated standard proportion of the state population normal distribution to correct for the which is black in 2003 are available left-censoring problem that not every- from the U.S. Census Bureau (2005, one in a birth cohort graduates from p. 24, Table 21). 206 Kanazawa

RESULTS state IQ has a large and significant effect on measures of health and longevity. Table 1, Column (1), shows that, when State IQ has a significantly positive effect entered alone, state Gini coefficient has a on life expectancy at birth (b = .03, beta = large and significantly negative effect on .23, p < .01), a significantly negative life expectancy at birth across the 50 states effect on age-adjusted death rate (b = −1.77, and the District of Columbia (b = −34.01, beta = −.27, p < .01), on infant mortality standardized beta = −.57, p < .0001). State rate (b = −.03, beta = −.29, p < .001), Gini coefficient has a similarly large and and on proportion obese (b = −.11, beta = significant effect on age-adjusted death −.49, p < .0001). State IQ alone explains rate (Column (4): b = 1670.01, beta = .51, nearly 30% of the total variance in pro- p < .001) and infant mortality rate (Column portion obese (partial r = −.53, controlling (7): b = 27.50, beta = .48, p < .001). This for income inequality and percent black). is consistent with Wilkinson’s claim that While state IQ is significantly correlated “inequality kills,” and replicates all the with economic development (Kanazawa, studies on the American states, cited 2006c), controlling for gross state product above, which show that income inequality per capita does not alter the substantive has adverse effects on the health and conclusions above (results available upon longevity of the population. However, as request). Column (10) shows, state Gini coefficient is not at all correlated with percent obese in DISCUSSION the state population (b = 8.67, beta = .08, ns). The conclusion that income inequality Figures 1-3 show the partial effect on lowers population health, however, appears life expectancy at birth of income inequal- premature. Table 1, Columns (2), (5), and ity (Figure 1), percent black (Figure 2), (8), shows that, net of percent black in the and state IQ (Figure 3). They collectively population, the state Gini coefficient has demonstrate that income inequality has absolutely no effect on life expectancy at virtually no effect on life expectancy at birth (b = −4.30, beta = −.07, ns), age- birth, net of the racial composition of the adjusted death rate (b = 226.92, beta = .07, state population and state IQ, while the ns), and infant mortality rate (b = −8.77, latter two factors have significant effects beta = −.15, ns). In contrast, percent black even net of income inequality. has a large and significant effect on life Given that nations with higher national expectancy (b = −.09, beta = −.72, p < IQs have less income inequality (lower .0001), age-adjusted death rate (b = 4.50, Gini coefficients) than nations with lower beta = .63, p < .001), and infant mortality national IQ (Kanazawa, 2008b), one might rate (b = .11, beta = .91, p < .0001). Column suggest that the effect of state IQ on (11) shows that percent black also has a health across American states is mediated significantly positive effect on percent by income inequality. This does not appear obese (b = .14, beta = .55, p < .01) while to be the case, however, as state IQ and income inequality doesn’t (b = −35.67, Gini coefficients are not significantly cor- beta = −.31, ns). related across states (r = .07, ns, n = 51). Table 1, Columns (3), (6), (9), and The analyses presented above in this (12), shows that, even when income ine- paper replicate Kanazawa’s (2006b) quality and percent black are controlled, earlier findings in a cross-national study .49 .11**** .26 - - - 29.68 - , 14** .13** .31 ATES - 35.67 R - ORTALITY M EONATAL .29 .03*** .12 .08 7.00 8.67 - - - N - AND

.1534 NFANT 8.77 - - . I IRTH B AT

VARIABLE

XPECTANCY BESE .27 E 1.77** - O EPENDENT - IFE D L ON ERCENT TABLE 1 1 TABLE P IQ IQ AND TATE S AND

LACK B ERCENT s (beta weights). , P .23 .70 .63 .61 .91 .89 .55 .52 .09**** 4.50*** 4.39**** .11**** .11**** . .10 .51 .07 .10 .48 (.01) (.65) (.01) (.03) 5.78 1670.01*** 226.92 322.47 27.50*** - - - - NEQUALITY ession coefficient ession I < .0001 p .72 .09**** .07 (.02) (.02) (1.07) (1.01) (.01) (.01) (.05) (.04) 4.30 NCOME - - - - I ized regression coefficients. regression ized OF

Life expectancy at birth at expectancy Life rate death Age-adjusted rate mortality Infant obese Percent < .001 **** p (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) FFECTS .57 .33 .59 .65 .26 .46 .53 .23 .65 .74 .01 .16 .40 E - (3.12) (3.31) (3.19) (182.67) (212.07) (204.37) (3.22) (2.94) (2.65) (7.38) (9.21) (8.08) (6.95) (7.66) (7.23) (406.95) (491.16) (462.68) (7.17) (6.80) (5.99) (16.44) (21.33) (18.29) 51 51 51 51 51 51 51 51 51 51 51 51 < .01 *** HE p T < .05 ** p * (Numbers parentheses in are standard errors). regr italics are standardized in Numbers Note: Main entries are unstandard are entries Main Note: 2 Constant 92.18 79.91 77.98 112.60 708.68 833.12 -5.19 9.79 12.10 21.18 39.49 47.31 R State IQ .03** Percent black n Gini coefficient 34.01**** 207 208 Kanazawa

FIG. 1.—Partial relationship between income inequality and life expectancy at birth, controlling for percent black and state IQ.

FIG. 2.—Partial relationship between percent black and life expectancy at birth, controlling for income inequality and state IQ. IQ and the Health of States 209

FIG. 3.—Partial relationship between state IQ and life expectancy at birth, controlling for income inequality and percent black.

at the subnational level of the American inequality lowers the health and longevity states. The earlier study shows that of the population appears to have been pre- neither economic development (GDP per mature. Variations in health and longevity capita) nor income inequality (Gini coef- across American states appear primarily a ficient) has any significant effect on life function of state racial composition (per- expectancy at birth, age-specific mortality cent black) and secondarily of state rate, and infant mortality rate, across IQ, and not at all of income inequality 126 nations. The present study shows (Gini coefficient). Contrary to Wilkinson’s that, net of percent black, income inequal- claim, inequality does not kill; lower ity has no effect on life expectancy at birth, intelligence does. age-adjusted death rate, infant mortality The results are consistent with the rate, and percent obese. Further, even recent evolutionary psychological theory after controlling for income inequality that the human brain has difficulty com- and state racial composition, average prehending and dealing with entities and intelligence of the state population has situations that did not exist in the ancestral a significant effect on the measures of environment (Crawford, 1993; Kanazawa, health and longevity. The unanimous con- 2002, 2004a; Symons, 1990; Tooby & clusion of past epidemiological studies Cosmides, 1990), and that this difficulty cited in the Introduction that income interacts with general intelligence, such 210 Kanazawa that less intelligent individuals have affects health and longevity only in so far greater difficulty with evolutionarily as the latter are affected by individual novel stimuli than more intelligent indi- choices and behavior, via the more intelli- viduals (Kanazawa, 2005, 2006a, 2008b). gent individuals’ ability to recognize and Because most health risks and hazards in deal properly with evolutionarily novel the contemporary societies are evolution- health risks and hazards. The Hypothesis arily novel, this reasoning predicts, con- equally predicts that general intelligence sistent with the Savanna-IQ Interaction does not affect mortality when it does not Hypothesis, that the more intelligent indi- involve individual choice and behavior, viduals are better able than less intelligent and instead caused by diseases of purely individuals to recognize such health genetic origin, such as Huntington’s dis- hazards, deal with them appropriately, ease. A crucial test to adjudicate between and, as a result, stay healthier and live the Savanna-IQ Interaction Hypothesis longer. In conjunction with an earlier and the “body system integrity” view is to finding that, even after controlling for examine whether general intelligence age, sex, race, marital status, education, reduces death from all causes regardless occupational prestige, and income, intelli- of the degree of individual choice gence has a significantly positive effect on involved. reported health of Americans (Kanazawa, There are other alternative explana- 2006b), the present paper provides empir- tions for the empirical results presented in ical support for the recent evolutionary this paper. For example, Gottfredson psychological theory on the nature and (2007) argues that individuals with low function of general intelligence. The anal- intelligence are disproportionately more ysis provided above replicates at the likely to have “fatal accidents” and thus macro level a large number of recent stud- lower life expectancies, without assum- ies that show that, even net of education, ing that general intelligence is a domain- social class and other social and economic specific adaptation or that the effect of factors, more intelligent individuals stay general intelligence on health and life healthier and live longer than less intelli- expectancy was markedly different in the gent individuals (Deary & Der, 2005; ancestral environment, as Kanazawa Deary et al., 2004; Gottfredson, 2004; (2004b) does. The domain-specificity or Gottfredson & Deary, 2004; Whalley & generality of general intelligence remains Deary, 2001). an open question in evolutionary psychol- One alternative explanation for the ogy and intelligence research (Roberts, effect of general intelligence on health 2007). and longevity is that both intelligence and While the Savanna-IQ Interaction health reflect the underlying “body Hypothesis, among others, can account system integrity” (Deary & Der, 2005; for the effect of state IQ on measures of Whalley & Deary, 2001). In this view, health and life expectancy, it cannot genetically and developmentally healthier explain the effect of racial composition. individuals are simultaneously more Rushton’s (1995) differential K theory of likely to have higher intelligence and race differences can. Rushton argues that remain healthy and live longer. In con- individuals of different races have trast, the Savanna-IQ Interaction Hypoth- slightly different life history strategies, esis suggests that general intelligence whereby Africans tend to start reproducing IQ and the Health of States 211 earlier, produce a larger number of off- life expectancies and higher infant and spring with less parental investment in adult mortality rates. More theoretical each, and end their reproductive careers and empirical research is clearly neces- earlier than other races. His theory can sary to investigate the effect of general therefore explain why states which have intelligence and race on health and larger percentages of blacks have lower longevity.

NOTES

1. Some researchers have proposed just such various cognitive (“IQ”) tests. What I mean by benefits, however. For example, Korte et al. (2005) “general intelligence,” however, is the ability to propose that vulnerability to stress may be an indi- think and reason, deductively or inductively, think vidual difference variable and may have evolved as abstractly, use analogies, synthesize information, and a behavioral strategy of low-aggression doves to apply knowledge to new domains. They are therefore environmental threats. Allen and Badcock (2003) not exact synonyms. The factor-analytic g is a mea- suggest that depression may have evolved as an sure of general intelligence, not general intelligence adaptive response to the situation of chronic low sta- itself. My theory is about general intelligence as an tus, in order to reduce risks in social interaction, evolved psychological mechanism, and not at all when one’s value to the group relative to one’s about the g as the latent variable in a factor analysis. burden on it is particularly low. However, this 3. Of course, infants under one cannot choose to explanation once again raises the question of why or not to avoid evolutionarily novel hazards to the capacity to be depressed did not evolve without health. It is the parents’ and other adult caretakers’ the negative health consequences. general intelligence that matters to infant mortality 2. In my 2004 Psychological Review article and that the state IQ measures. More intelligent (Kanazawa, 2004b), I use the phrases “general intel- parents and caretakers are expected to engage in ligence” and “the g factor” synonymously and inter- behavior which promotes the health of the infant, by changeably. Technically, however, the g factor is a recognizing and properly dealing with evolutionarily latent variable which emerges in a factor analysis of novel risks and threats.

REFERENCES

ACKERMAN, P. L. and E. D. HEGGESTAD. 1997. Intelli- ELIAS, M. F., P. K. ELIAS, L. M. SULLIVAN, P. A. gence, personality, and interests: Evidence for WOLF, and P. B. D’AGOSTINO, 2003. Lower cog- overlapping traits. Psychological Bulletin, 121: nitive function in the presence of obesity and 219–245. hypertension: The Framingham heart study. ALLEN, N. B. and P. B. BADCOCK. 2003. The social International Journal of Obesity. 27:260–268. risk hypothesis of depressed mood: Evolution- ELLIS, L. and H. NYBORG. 1992. Racial/ethnic varia- ary, psychosocial, and neurobiological perspec- tions in male testosterone levels: A probable con- tives. Psychological Bulletin, 129:887–913. tributor to group differences in health. Steroids, BEN SHLOMO, Y., I. R. WHITE, and M., MARMOT, 57:72–75. 1996. Does the variation in the socioeconomic FOSTER, J. E. (1985). Inequality measurement. In characteristics of an are affect mortality? British YOUNG, P. (ed.), Fair Allocation, Proceedings of Medical Journal, 312:1013–1014. Symposia in Applied Mathematics, Volume 33 CHAMORRO-PREMUZIC, T. and A. FURNHAM. 2006. (pp. 31–68). Providence: American Mathemati- Intellectual competence and the intelligent per- cal Society. sonality: A third way in differential psychology. FREY, M. C. and D. K. DETTERMAN, 2004. Scholastic Review of General Psychology, 10:251–267. assessment or g?: The relationship between the CRAWFORD, C. B. 1993. The future of sociobiology: scholastic assessment test and general cognitive Counting babies or proximate mechanisms? ability. Psychological Science, 15:373–378. Trends in Ecology and Evolution, 8:183–186. GOTTFREDSON, L. S. 2004. Intelligence: Is it the epi- DEARY, I. J. and G. DER. 2005. Reaction time demiologists’ elusive “fundamental cause” of explains IQ’s association with death. Psycholog- social class inequalities in health? Journal of ical Science, 16:64–69. Personality and Social Psychology, 86:174–199. DEARY, I. J., M. C. WHITEMAN, J. M. STARR, L. J. GOTTFREDSON, L. S. (2007). Innovation, fatal acci- WHALLEY, and H. C., FOX, 2004. The impact of dents, and the evolution of general intelligence. childhood intelligence on later life: Following In Roberts, M. J. (ed.), Integrating the mind: up the Scottish Mental Surveys of 1932 and Domain general versus domain specific pro- 1947. Journal of Personality and Social Psychol- cesses in higher cognition (pp. 387–425). Hove: ogy, 86:130–147. Psychology Press. 212 Kanazawa

GOTTFREDSON, L. S. and I. J. DEARY. 2004. Intelligence KOENIG, K. A., M. C. FREY, and D. K. DETTERMAN, predicts health and longevity, but why? Current 2008. ACT and general cognitive ability. Intelli- Directions in Psychological Science, 13:1–4. gence, 36:153–160. HERRNSTEIN, R. J. and C. MURRAY. (1994). The bell KORTE, S. M., J. M. KOOLHAAS, J. C. WINGFIELD, and curve: Intelligence and class structure in American B. S. MCEWEN, 2005. The Darwinian concept life. New York: Free Press. of stress: Benefits of allostasis and costs of JOHNSON, A. C. and N. T. THOMOPOULOS. 2002. Char- allostatic load and the trade-offs in health and acteristics and tables of the left-truncated normal disease. Neuroscience and Biobehavioral Reviews, distribution. Proceedings of the Midwest Deci- 29:3–38. sion Sciences Institute, 133–139. LUBINSKI, D. and L. G. HUMPHREYS. 1997. Incorporating KANAZAWA, S. 2002. Bowling with our imaginary general intelligence into epidemiology and the friends. Evolution and Human Behavior, 23: social sciences. Intelligence, 24:159–201. 167–171. LUBINSKI, D., C. P. BENBOW, R. M. WEBB, and A. KANAZAWA, S. 2004a. The Savanna Principle. Mana- BLESKE-RECHEK, 2006. Tracking exceptional gerial and Decision Economics, 25:41–54. human capital over two decades. Psychological KANAZAWA, S. 2004b. General intelligence as a Science, 17:194–199. domain-specific adaptation. Psychological Review, LYNCH, J. W., G. A. KAPLAN, E. R. PAMUK, R. D. 111:512–523. COHEN, K. E. HECK, J. L. BALFOUR, and I. H. YEN, KANAZAWA, S. 2005. An empirical test of a possible (1998). Income inequality and mortality in met- solution to “the central theoretical problem of ropolitan areas of the United States. American human sociobiology.” Journal of Cultural and Journal of Public Health, 88, 1074–1080. Evolutionary Psychology, 3:249–260. MARMOT, M., and R. G., WILKINSON (Eds.) (1999). KANAZAWA, S. 2006a. Why the less intelligent may Social determinants of health. Oxford: Oxford enjoy television more than the more intelligent. University Press. Journal of Cultural and Evolutionary Psychol- National Center for Health Statistics. (2005). Health, ogy, 4:27–36. United States, 2005, with chartbook on trends in KANAZAWA, S. 2006b. Mind the gap . . . in intelli- the health of Americans. Washington D.C.: U.S. gence: Reexamining the relationship between Government Printing Office. inequality and health. British Journal of Health POLEDNAK, A. P. (1989). Racial and ethnic differ- Psychology, 11:623–642. ences in disease. Oxford: Oxford University KANAZAWA, S. 2006c. IQ and the wealth of states. Press. Intelligence, 34:593–600. ROBERTS, M. J. (ed.) (2007). Integrating the mind: KANAZAWA, S. 2008a. Temperature and evolutionary Domain general versus domain specific pro- novelty as forces behind the evolution of general cesses in higher cognition. Hove: Psychology intelligence. Intelligence, 36:99–108. Press. KANAZAWA, S. (2008b). De gustibus est disputandum ROSS, R., L., H. BERNSTEIN, R. JUDD, M. HANISCH, II: Why liberals and atheists are more intelligent. PIKE, and B. HENDERSON, 1986. Serum testoster- Interdisciplinary Institute of Management. Lon- one levels in healthy young black and white don School of Economics and Political Science. men. Journal of the National Cancer Institute, KANAZAWA, S. (Forthcoming-a). Evolutionary psy- 76:513–532. chology and crime. In Walsh, A., and Beaver, K. ROSS, N. A., M. C. WOLFSON, J. R. DUNN, J-M. (eds.), Contemporary Biosocial Criminology. BERTHELOT, G. A. KAPLAN, and J. W. LYNCH, New York: Routledge. 2000. Relation between income inequality and KANAZAWA, S. (Forthcoming-b). Theft. In Duntley, J., mortality in Canada and in the United States: and Shackelford, T. K. (eds.), Evolutionary Cross sectional assessment using census data Forensic Psychology. New York: Oxford Uni- and vital statistics. British Medical Journal, versity Press. 320:898–902. KANAZAWA, S. and K. PERINA. 2008. De gustibus est RUSHTON, J. P. (1995). Race, evolution, and behav- disputandum III: Why more intelligent individuals ior: A life history perspective. New Brunswick: like classical music. Interdisciplinary Institute of Transaction. Management. London School of Economics and SCHOENDORF, K. C., M. P. H. CAROL, C. J. R. HOGUE, Political Science. J. C. KLEINMAN, and D. ROWLEY, 1992. Mortality KAPLAN, G. A., E. R. PAMUK, J. W., LYNCH, R. D., among infants of black as compared with white COHEN, and J. L., BALFOUR, 1996. Inequality in college-educated parents. New England Journal income and mortality in the United States. British of Medicine, 326:1522–1526. Medical Journal, 312:999–1003. SYMONS, D. 1990. Adaptiveness and adaptation. KAWACHI, I. and B. P. KENNEDY. 1997. The relation- Ethology and Sociobiology, 11:427–444. ship of income inequality to mortality: Does the TOOBY, J. and L. COSMIDES. 1990. The past explains choice of indicator matter? Social Science and the present: Emotional adaptations and the struc- Medicine, 45:1121–1127. ture of ancestral environments. Ethology and KENNEDY, B. P., I. KAWACHI, and D., PROTHROW- Sociobiology, 11:375–424. SMITH, 1996. Income distribution and mortality. TOOBY, J. and L. COSMIDES. 1990. On the universality British Medical Journal, 312:1253. of human nature and the uniqueness of the IQ and the Health of States 213

individual: The role of genetics and adaptation. WILKINSON, R. G. 1992. Income inequality and life Journal of Personality, 58:17–67. expectancy. British Medical Journal, 304:165–168. U.S. CENSUS BUREAU. 2005. Statistical abstract of the WILKINSON, R. (2000). Mind the gap: Hierarchies, United States: 2006. Washington, D.C.: U.S. health and human evolution. London: Weiden- Government Printing Office. feld & Nicolson. WHALLEY, L. J. and I. J. DEARY. 2001. Longitudinal WILSON, J. Q. and R. J. HERRNSTEIN. (1985). Crime cohort study of childhood IQ and survival up to and human nature: The definitive study of the age 76. British Medical Journal, 322:1–5. causes of crime. New York: Touchstone.