The Intelligence Paradox: Why the Intelligent Choice Isn't Always the Smart One - Satoshi Kanazawa 2012
The Savanna-IQ Interaction Hypothesis
When Intelligence Matters (and When It Doesn’t)
The logical intersection of the Savanna Principle—discussed in Chapter 2—and the theory of the evolution of general intelligence—discussed in Chapter 3—suggests a qualification of the Savanna Principle. If general intelligence evolved to deal with evolutionarily novel problems, then the human brain's difficulty in comprehending and dealing with entities and situations that did not exist in the ancestral environment (proposed in the Savanna Principle) should interact with general intelligence. In other words, the Savanna Principle should hold stronger among less intelligent individuals than among more intelligent individuals. More intelligent individuals should be better able to comprehend and deal with evolutionarily novel (but not evolutionarily familiar) entities and situations than less intelligent individuals.
So I now propose the Savanna-IQ Interaction Hypothesis1 (or “the Hypothesis”). The Hypothesis qualifies and elaborates on the Savanna Principle by introducing intelligence and how it modifies the operation of the Savanna Principle.
The Savanna-IQ Interaction Hypothesis: Less intelligent individuals have greater difficulty comprehending and dealing with evolutionarily novel entities and situations that did not exist in the ancestral environment than more intelligent individuals. In contrast, general intelligence does not affect individuals’ ability to comprehend and deal with evolutionarily familiar entities and situations that existed in the ancestral environment.
Recall the definition of “comprehension” from Chapter 2 as the true, logical, and scientifically and empirically accurate understanding of how something works. Now I am going to review the empirical evidence for the Savanna-IQ Interaction Hypothesis in many different domains of life.
Back to TV Friends
In Chapter 2, I discuss my 2002 study,2 which suggests that people may have some implicit difficulty distinguishing their “TV friends”—characters that they repeatedly see on TV—from their real friends. The more they watch certain types of TV shows, the more satisfied they become with their friendships, just as they do if they have more friends or socialize with them more frequently. This makes perfect sense from the perspective of the Savanna Principle. Because there were no realistic electronic (or photographic) images of other humans in the ancestral environment, the human brain has difficulty with such images. Since all realistic images of other humans in the ancestral environment were other humans, the human brain implicitly assumes that any such images of other humans, who don't attempt to kill or maim them (which very few TV characters do), are their friends.
In 2006, after I formulated the initial ideas behind the Hypothesis, I reanalyzed the same data from the General Social Surveys to see if individuals’ IQ was related to their implicit tendency to confuse “TV friends” and real friends.3 And it was. This tendency, which I initially thought (according to the Savanna Principle) was a universal human trait in 2002, appears to be limited to men and women below median intelligence (consistent with the Savanna-IQ Interaction Hypothesis). Those who are above median in intelligence do not report greater satisfaction with friendships as a function of watching more TV; only those below median intelligence do.
This seems to suggest that the evolutionary constraints on the brain predicted by the Savanna Principle, whereby individuals have implicit difficulty with recognizing realistic electronic images on TV for what they are, appear to be weaker or altogether absent among more intelligent individuals. Since truly enjoying the experience of watching TV requires suspension of disbelief and not really understanding that characters repeatedly seen on the screen are highly paid actors hired to play scripted roles, this new finding can potentially explain why less intelligent individuals tend to enjoy the experience of watching TV more than more intelligent individuals do.
Of course, just like everything else I say in this book, the negative association between general intelligence and the enjoyment of TV is an empirical generalization, for which there are many exceptions. I personally happen to love watching TV myself. However, among my highly intelligent academic colleagues, there are many who do not watch television at all. Some don't even own a television set. I'm often frustrated with them, because I have nothing to talk to them about if they don't watch television at all. I share no common points of reference with such people. Like the comedy writer Tina Fey, I believe that American television is one of the greatest things about the country, and I feel sorry for people who cannot appreciate its many wonderful programs. Nevertheless, the Savanna-IQ Interaction Hypothesis can explain why many intelligent people do not watch television at all.
We all know people who have a tendency to talk back and speak to characters they see while they are watching TV or movies. I believe this habit also stems from their implicit confusion of “TV friends” and real friends, the confusion of realistic electronic images of other human beings on the screen and real human beings. I would therefore predict that less intelligent individuals are more likely to talk back and speak to TV and movie screens, and further that these are precisely the people who enjoy watching TV and movies more than others do. Less intelligent individuals are less likely truly to comprehend electronic images on the screen and more likely to forget unconsciously that they are not real live human beings who can hear us when we talk back to them.
Back to Pornography
In Chapter 2, I explain that men's and women's brains cannot truly comprehend pornography—realistic photographic and videographic images of sexually aroused men and women—because no such thing existed in the ancestral environment. The Savanna Principle can explain why men and women confuse porn stars with real potential sex partners, just as they confuse “TV friends” with real friends. Now the Hypothesis would predict that such confusion of porn stars with real potential sex partners might interact with general intelligence, just as does the confusion of “TV friends” with real friends.
In the first independent empirical confirmation of the Hypothesis, conducted by researchers other than myself, Gorge A. Romero and Aaron T. Goetz, two young evolutionary psychologists at California State University-Fullerton, demonstrate that this indeed is the case.
In their survey, Romero and Goetz measure three things from their male respondents: (1) their perception of women's sexuality and sexual behavior; (2) their consumption of pornography; and (3) their general intelligence. Their analysis demonstrates that there is a positive association between men's consumption of pornography and how sexually promiscuous they think women are, in other words, how likely they think women in real life behave like porn stars. The more frequently they watch porn, the more they believe that women are sexually more promiscuous (have a large number of sex partners and one-night stands), and enjoy having casual sex, giving oral sex, receiving anal sex, and having threesomes, just as women typically do in porn movies. However, this confusion of real women and porn stars happens only if the men are less intelligent (at least one standard deviation below the mean), not if they are of average or above-average intelligence (at least one standard deviation above the mean).
This is a brilliant empirical demonstration of the Hypothesis in operation. Men's brains confuse real women and porn stars—thereby believing that real women act like porn stars by being sexually promiscuous and enjoying unusual sexual acts—but only if they are of below-average intelligence. Romero and Goetz further demonstrate that, once they control for general intelligence, the consumption of pornography alone does not increase men's tendency toward this confusion. One must simultaneously have below-average intelligence and consume a large quantity of pornography in order to confuse real women with porn stars.
The Failures of the Truly Gifted
Perhaps nothing illustrates the operation of the Hypothesis, and the sharp distinction between the evolutionarily novel and evolutionarily familiar domains of life, more clearly than the profile of the truly gifted. The Study of Mathematically Precocious Youth tracks the lives of more than 5,000 individuals who have been identified as truly gifted in the SAT talent search.5 Most people take the SAT in the last year of high school at age 17. Participants in the talent search take it in the 7th or 8th grade, before the age of 13. If they score within the top .01% (top 1 in 10,000) for their age, by scoring either more than 700 (out of 800) on the SAT mathematical reasoning ability or more then 630 on the SAT verbal reasoning ability, they are included in the study. Their IQ is therefore higher than 155.
The SAT is considered to be a reasonable IQ test, a test of reasoning ability, not of acquired knowledge, while the ACT is more an achievement test of acquired knowledge.6 According to the ACT's own website, “The ACT is an achievement test, measuring what a student has learned in school. The SAT is more of an aptitude test, testing reasoning and verbal abilities.”7
As you might expect, these individuals of extraordinarily high intelligence achieve equally extraordinary success in the evolutionarily novel domains of formal education and paid employment in the capitalist economy. More than half of them (51.7% of men and 54.3% of women) have earned a doctorate (Ph.D., M.D., or J.D.), compared to the population baseline in the US of 1%. An additional 5.3% of them have earned an MBA, all but one of them in the top 10 US programs. Nearly half (45.8%) of them are university professors, engineers, or scientists; an additional 13.6% are in medicine or law. More than a fifth (21.7%) of those in tenure-track positions in the top 50 US universities are already full professors in their early 30s. (It is virtually unheard of for someone to achieve the rank of full professor in their early 30s in any university, let alone in a top 50 US university.) More than a third of the men and about a fifth of the women earn more than $100,000 a year in 2003—2004 in their early 30s. (Yes, in their early 30s!) Additionally, 17.8% of the men and 4.3% of the women have earned patents, compared to the population baseline in the US of 1%.
No matter how you slice it, there is no question that these individuals in the Study of Mathematically Precocious Youth go on to experience tremendous success in life, measured by their educational achievement, professional careers, income, and creativity. All of these, however, are evolutionarily novel areas of life, which our ancestors did not have more than 10,000 years ago. How do the same individuals, with IQs higher than 155, fare in evolutionarily familiar areas of life?
Mating and parenting are eminently evolutionarily familiar domains of life. Despite the cumbersome interventions of modern inventions (condoms, sperm banks, internet porn), we still mate, pretty much the same way as our ancestors did 10,000 years ago. Sexual courtship today still involves initial visual and chemical attraction, verbal and physical interaction, mutual mate choice based on social status, physical attractiveness, and moral character as clues to good genes and parental abilities, foreplay, copulation, positive or negative reaction to the mate choice by friends and family, etc. And we still have children as our ancestors did then. Children today, as then, are raised by pair-bonded couples, single mothers and their kin, biological mothers and stepfathers, etc. Few other domains of life today are as evolutionarily familiar as marriage and parenting, so the Hypothesis would predict that more intelligent individuals do not fare better than less intelligent individuals in the domains of marriage and parenting.
This indeed appears to be the case. In stark contrast to their stellar successes in education and employment, the participants in the Study of Mathematically Precocious Youth do not do very well in the evolutionarily familiar domains of marriage and parenting. For example, 64.9% of the men and 69.0% of the women remain childless at age 33, compared to the population baseline of 26.4% in the age group 30—34. The majority of parents only have one child. As a result, the mean number of children is .61 for men and .44 for women, compared to the population baseline of 1.59 for women in the age group 30—34. Despite their extraordinarily high general intelligence, these men and women seem to lag behind everyone else in the evolutionarily familiar domains of marriage and parenting.
The fact that truly gifted individuals have no particular advantage (and often disadvantage) in such evolutionarily familiar domains as mating is illustrated by the following exchange between Stephen Hawking and Larry King. Hawking appeared on Larry King Live Weekend on Christmas Day 1999, on the eve of the new millennium, when the following exchange took place:
Larry King: What, Professor, puzzles you the most? What do you think about the most?
Stephen Hawking: Women.
Larry King: Welcome aboard.
The Hypothesis would indeed predict that the Lucasian Professor of Mathematics at the University of Cambridge (a post once held by Isaac Newton), who is putatively the most intelligent person in the United Kingdom today and can figure out the origin and the destiny of the universe, has no particular advantage in evolutionarily familiar domains of life such as mating, over someone like Larry King, who has only a high school education, and, incidentally, has had seven wives and five children.
What about the Rest of Us?
The sharp contrast between great success in evolutionarily novel domains and great failures in evolutionarily familiar domains for the very intelligent is not limited to the truly gifted in the Study of Mathematically Precious Youth. General intelligence affects life outcomes in virtually every area and throughout the entire life.8 From schooling to employment to crime and welfare dependency to civility and citizenship, not only do more intelligent individuals achieve more desirable outcomes, but general intelligence almost always has a linear positive effect on the desirability of the life outcomes. The more intelligent the individuals, the more desirable the life outcomes.
Marriage and parenting are among the very few exceptions to this pattern in a comprehensive review of American life. In fact, “very bright” individuals are the least likely to marry of all the cognitive classes. (Recall their definition of cognitive classes at the end of Chapter 3.) Only 67% of these “very bright” white Americans marry before the age of 30, whereas between 72% and 81% of those in other cognitive classes marry before 30.9 The mean age of first marriage among the “very bright” whites is 25.4, whereas it is 21.3 among the “very dull” individuals and 21.5 among the “dull” individuals. The more intelligent you are, the later you marry.
The pattern is similar in parenting. For example, general intelligence does not confer advantages in giving birth to healthy babies. For example, 5% of white babies born to “very bright” mothers suffer from low birth weight, compared to 1.6% of those born to “bright” mothers and 3.2% of those born to “normal” mothers. Only babies born to “dull” mothers (7.2%) and “very dull” mothers (5.7%) fare worse.10
The lack of IQ advantage continues later in the childhood. “Very bright” mothers are more likely to have children who are behind in motor and social development or have the worst behavioral problems. Specifically, 10% of children born to “very bright” white mothers are in the bottom 10% of the motor and social development index, compared to 5% of those born to “bright” mothers and 6% of those born to “normal” mothers. Similarly, 11% of children born to “very bright” mothers find themselves in the bottom 10% of the behavioral problems index, compared to 6% of those born to “bright” mothers and 10% of those born to “normal” mothers.11 It is important to note that the problems suffered by children born to “very bright” mothers are not just social and behavioral—for which there might be varying and changing cultural definitions of what constitutes “normal”—but are also physical, such as birth weight and motor development, for which the criteria of normal development are objective and invariant.
Now, since “very bright” white women marry later, and thus give birth to their babies at older ages compared to other mothers, perhaps some of these physical and behavioral problems of their children may be attributable to their older maternal age at birth or the greater possibility of harmful mutations in the sperm of older men.12 But this is precisely my point. Women with higher intelligence are not using their intelligence to marry early and have healthier children, which are the direct means toward achieving reproductive success. More intelligent individuals are more likely to receive more education and earn more money, both of which are evolutionarily novel, but they are certainly not more likely to achieve the evolutionarily familiar goals of marriage and reproduction. This has led some scholars to muse, “Can mothers be too smart for their own good?”13
The Exception that Proves the Rule: Evolutionarily Novel Elements in Mating and Parenting
This is not to argue, however, that intelligent people are not better mates or parents in general today. Intelligent individuals do make better mates and parents in some ways in the current (evolutionarily novel) environment. For one thing, more intelligent individuals universally attain more desirable outcomes in all evolutionarily novel domains, such as education, economy, criminal justice, even health and longevity in contemporary society.14 More educated, wealthier, healthier parents who avoid trouble with the law undoubtedly do make better parents today.
One need go no farther than to recall the news story from many years ago of an illiterate teenage mother whose baby died of dehydration because the mother could not read the instructions for how to make the baby formula and instead fed the dry powder to the baby as is, without first dissolving it in water. Note, however, that this tragedy happened precisely because it involved written instructions for making a baby formula—an evolutionarily novel stimulus about an evolutionarily novel product. My contention is that even this mother may have done fine raising her children in the ancestral environment, where childrearing most likely did not require general intelligence, as all the answers necessary for parenting would have been provided by other evolved psychological mechanisms. In the ancestral environment, everyone was illiterate.
Modern means of contraception are another evolutionarily novel element in the evolutionarily familiar domain of mating and parenting. In the ancestral environment, our ancestors probably mated all the time, with pregnancy and lactation (lactational amenorrhea) serving as the only natural means of contraception, besides abstinence. As a result, our ancestors invariably produced a larger number of offspring than we do today, but many of them died in infancy due to infectious diseases, malnutrition, and other natural causes (including predation by humans and other animals).15 The average number of offspring surviving to sexual maturity in the ancestral environment might not have been much larger than it is today. So while mating and parenting are evolutionarily familiar, voluntary control of fertility through contraception (such as condoms or the pill) is evolutionarily novel. So the Hypothesis would predict that more intelligent individuals are better able to control their fertility voluntarily through artificial means of contraception than less intelligent individuals.
This indeed appears to be the case.16 Among contemporary Americans in the GSS data, the association between the lifetime number of sex partners and the number of children is positive among less intelligent individuals (who are below the median in verbal intelligence), but negative among more intelligent individuals (who are above the median in verbal intelligence). The more sex partners less intelligent individuals have, the more children they have, as a natural consequence of greater sexual activity with more partners. In sharp contrast, the more sex partners more intelligent individuals have, the fewer children they have. You cannot have fewer children on average by having more sex partners, unless you employ effective contraception.
Figure 4.1 Partial association between lifetime number of sex partners and number of children among the less intelligent
Figures 4.1 shows the partial association, after controlling for age, race, sex, education, marital status, and religion, between the lifetime number of sex partners and the number of children, among individuals who are below the median in verbal intelligence. As you can see, the relationship is positive, as indicated by the regression line with an incline.
Figure 4.2 Partial association between lifetime number of sex partners and number of children among the more intelligent
Figure 4.2 presents the same relationship among individuals who are above the median in verbal intelligence. Here the relationship between the lifetime number of sex partners and the number of children is negative, as indicated by the regression line with a decline.
The contrast between these two graphs suggests that more intelligent Americans are indeed more efficient in employing (evolutionarily novel) modern means of contraception than their less intelligent counterparts.
Intelligence and Interpersonal Relationships
Interpersonal relationships are an eminently evolutionarily familiar domain of life. Even in the ancestral environment, our ancestors had friends, allies, and enemies that they had to deal with. They also had parents, children, siblings, and other relatives. There is nothing evolutionarily novel about interacting with these categories of people.
In addition, it was very important in the ancestral environment (as it is now) to maintain good relations with these categories of people (except, perhaps, for enemies). Reliable friends and allies are crucial in survival and reproductive success,17 and investing in kin is a very important means of increasing reproductive success.18 So the Hypothesis would predict that general intelligence does not have any effect on individuals’ ability to maintain interpersonal relationships with these evolutionarily familiar categories of people.
Survey data from the United States support this prediction of the Hypothesis.19 While more intelligent Americans socialize with their friends significantly more frequently than their less intelligent counterparts, intelligence does not seem to improve interpersonal relationships with other evolutionarily familiar categories of people. In fact, more intelligent individuals socialize with neighbors, siblings, and other relatives significantly less frequently than less intelligent individuals. Investing in kin is one of the important means of increasing reproductive success, yet more intelligent American seem less able to do so than less intelligent individuals.
Intelligence and Wayfinding
In the hunter-gatherer life of our ancestors on the African savanna, navigation and wayfinding was an essential skill, on which their very survival depended. After a long hunting or gathering trip, which could sometimes last for days, our ancestors had to find their way home without relying on maps, street signs, artificial landmarks, and the satellite navigation devices. Those who could not find their way home from their trips probably faced certain death. I would therefore expect navigation and wayfinding to be an evolutionarily familiar task, for which there is an evolved psychological mechanism, and the Hypothesis would predict that general intelligence is independent of wayfinding abilities.
A couple of studies support this prediction of the Hypothesis. In a highly ingenious experiment,20 researchers at York University in Canada took participants on a meandering journey through a wooded area, without any visible landmarks or maps, and asked them, at predetermined locations, to point to the direction of the origin. The participants must then lead the researchers back to the origin. In this study, the participants’ wayfinding ability had no correlation at all with their general intelligence, measured by Raven's Progressive Matrices (which, as discussed in Chapter 3, is the best single measure of general intelligence).
Researchers at the University of Arizona replicated the Canadian study in virtual reality.21 Their participants navigated in computer-generated “rooms” displayed on a computer screen by way of a joystick, and had to find an invisible target placed somewhere in the room on the floor. They got a beep when they (initially unknowingly) “walked” over the invisible target, and then must find it again and again in the same room by navigating to the same location in the room. The researchers’ data showed that the participants’ general intelligence had no effect on their ability to learn their way around the rooms and return to the invisible target. They concluded that the individual's ability at spatial navigation and general intelligence were largely independent. Given that more intelligent people tend to do virtually everything better than less intelligent people, their distinct lack of advantage in wayfinding is noteworthy.
Intelligence and Exercise
As hunter-gatherers, our ancestors engaged in constant physical activities. Their active, nomadic lifestyle, combined with their limited caloric intake, meant that obesity was probably very rare in the ancestral environment, and most of our ancestors maintained a healthy, active lifestyle by our contemporary standards. It means that regular exercise for its own sake, that is, exercising in order to stay healthy and control weight, is probably evolutionarily novel, and the Hypothesis would therefore predict that more intelligent individuals are more likely to engage in regular exercise today than less intelligent individuals.22
A couple of recent studies show that this indeed appears to be the case.23 In one study, the frequency of exercise is significantly positively associated with general intelligence.24 In another, individuals who can successfully maintain a regular exercise schedule are more intelligent than those who are unsuccessful by more than one full standard deviation (122.50 vs. 106.25). And this is not because intelligence is associated with conscientiousness.25 It appears that more intelligent individuals are more likely to adopt the evolutionarily novel lifestyle of regular exercise for its own sake.
Of course, more intelligent individuals are more likely to have white-collar desk jobs that are more physically sedentary, whereas less intelligent individuals are more likely to have blue-collar jobs that are physically more active. Thus more intelligent individuals may need to exercise more than less intelligent individuals. However, in the first study, education, income, and work status are all statistically controlled, and the participants in the second study are all college students (86.3% of whom are female, who are less likely to have blue-collar jobs). So the effect of intelligence on voluntary physical exercise appears to be genuine, but more research is necessary to arrive at a more firm conclusion.