Shattering Intelligence: Implications for Education and Interventions

The concept of a general intelligence or g factor has proved enormously fruitful in two respects. On the level of individual differences, it captures the fact that if one person outperforms another on one kind of conceptually demanding task, that advantage is likely to persist over a whole range of other cognitive tasks. On the level of group differences, we find that the average Full Scale IQ of two groups on a good IQ test often predicts things like their occupational profiles. Various occupations have minimum IQ thresholds. If 50 percent of group A scores above say an IQ of 100, while only 16 percent of Group B do, then Group A will have a three to one ratio in its favor in terms of the proportion of its members who are professionals or technicians or managers.

An example of a good IQ test is the WISC (Wechsler Intelligence Scale for Children). The reason it is good is that its ten subtests have enough cognitive complexity so that a high IQ person tends to beat the average person by a handy margin on all ten. That is equivalent to saying that it is a good measure of g. A test that included subtests of minimal cognitive complexity, let us say tying your shoes, would be a bad IQ test. The task is so simple that unintelligent people would perform it as well as intelligent people.

Shattering general intelligence

Despite all the triumphs of the concept of general intelligence, I believe intelligence is like the atom: you have to know both why its parts cohere and why they sometimes fly apart. Americans made massive IQ gains on the WISC between 1947 and 2002 amounting to almost 18 points of Full Scale IQ. These gains ranged from only 2 points on the WISC subtest called Information to 24 points on the subtest called Similarites (what do dogs and rabbits have in common?), despite the fact that both have the cognitive complexity that makes them good measures of g.

A bright person tends to accumulate more general information than a dull person at any given time and also tends to better at classifying things (will say that dogs and rabbits are both mammals). But over time, we find that society can develop these conceptual skills quite independently of one another. Children may progress a lot over time in putting on scientific spectacles, which means that many more of them will say “both mammals” rather than say something like “I use my dog to hunt rabbits.” While thanks to the rise of a visual culture that discourages reading, the average child today may have no greater store of general information than children did 55 years ago.

Diagnosing how conceptual skills actually develop

The fact that various conceptual skills develop so independently over time has wide implications for education. At any given time, a person that beats another at arithmetical reasoning is likely also to do better on a test called Raven’s Progressive Matrices. You are given a pattern of shapes that make up a square and you have to see the logical pattern across the rows and down the columns. If row one across is X O O O. and row two across is X X O O, then you must deduce that row three should be X X X O. Because g indicates that the skills go together, American schools since 1990 have been teaching children Matrices problems on the theory that it will develop their mathematical skills. But cognitive gains over time second-guess g: between 1947 and 2002, gains on the Arithmetic subtest of the WISC were only 4 points and gains on Raven’s were about 27.5 points. Clearly there is no functional relationship here. After all, there is a functional relationship between having a good vocabulary and reading serious literature, so if the ability to do the latter increased over time, the former would have to increase.

Thanks to the illusion fostered by g, the schools are wasting their time. I suspect that only for mathematicians is math a logical enterprise where all is proven and linked by logic. For 99 percent of us, it is a strange separate world in which very different rules hold than in everyday life. Therefore, we should try to isolate the fundamental concepts of that world, like measurement from a zero point, how numbers are created, why things are equivalent across an = sign, and so forth, and slowly get children to feel comfortable in the world of mathematics.

Once you break intelligence down into its autonomous components, many things become clear. For example, the Nation’s Report Card shows that today’s children are ahead of their parents in reading at early ages and then the gains fade away by the age of 17. How is that possible? The children are doing much better on heavily g loaded IQ tests like the WISC at all ages. Should not brighter people be able to read adult novels better?

This mystery is solved when you look at IQ trends over time. Since 1972 (when the NRC began), the big IQ gains have been on certain subtests and not others. There have been virtually no gains in vocabulary and information. You cannot enjoy War and Peace very much if you have to run to the dictionary or encyclopedia every other paragraph. We are doing a better job of teaching children the mechanics of reading at early ages. But their parents had mastered the mechanics by age 17 and at that age, neither generation has an information or vocabulary advantage. So we have made no progress is teaching young people how to enjoy adult literature.

A blinding obsession

The response of many g-men to IQ gains over time is to say: “You grant that gains on various subtests do not correspond to how well each subtest measures g. Well, that shows that the gains are hollow, that is, since they are not g gains they are not real intelligence gains and lack significance.” As you can see, this is just a way of saying that if all complex cognitive skills do not move together — that if they have different trends — then the trends just cannot be significant. This is the saddest result of the obsession with g: it makes the limitations of the concept no longer a matter of evidence. Any evidence that challenges the supremacy of g is not good evidence because it challenges the supremacy of g and that is that.

Note that we would not reason in this way in other areas. There is a musical g in the sense that whoever is better than me on the piano will probably outdo me on the organ. But skills could improve on one and not the other, and that would be of great significance to the world of music. There is a moral g in the sense that good people tend to be both more tolerant and more generous than the average. But over time, white Americans may have tended to become more tolerant of other races and no more generous in giving to charities. No black American would say that unless all of the components of moral g moved together, the trends were not significant.

The transience of intelligence

General intelligence or g has something to do with brain quality, and good genes have a lot to do with having an above average brain. Therefore, there was a tendency in differential psychology to think that our genes-determined brain accompanies us throughout our lives and that environment, except in extreme conditions (living with wolves since infancy), merely leaves minor imprints on that brain. After all, twin studies showed that even when identical twins were separated at birth, they had IQs at adulthood that were much more similar than the IQs of randomly selected people. What better evidence did you need that genes/brain went though life and environment just did a bit of tinkering along the way?

But this created a dilemma: if genes were so dominant, how could IQ gains over time be so huge? Unless you thought that there was a large genetic upgrading from one generation to the next, large intelligence gains should be impossible. Yet they occurred, which implied that there were environmental factors of huge potency. How could environment be both so feeble and so potent?

The Dickens/Flynn model resolved this dilemma. Two twins raised apart, thanks to having slightly better genes than average, would both get into increasingly privileged environments. Both would get more teacher attention, would be encouraged to do more homework, would get into a top stream, and by adulthood, they would both be far above average. Moreover, thanks to their identical genes, their environmental histories would be very similar. Their identical genes were getting all of the credit for the combination of identical genes plus nearly identical enriching environmental factors! The environmental factors were not feeble at all: they just tended to be similar for identical twins when raised apart, which made them look feeble.

This means a huge shift in perspective. The g-man view was that environment made little difference throughout life because environment makes very little difference at any point in life. The Dickens/Flynn view is that environment makes a lot of difference, which meant we had to look elsewhere for why its effects seem so transient. Our conclusion was that present environment swamps past environment in terms of effect on your level of cognitive functioning.

Cognitive exercise

The first implication of the new perspective is the benefit of persisting in cognitive exercise throughout life. There is the dramatic case of Richard Wetherill. He played chess in retirement and could think eight moves ahead. In 2001, he was alarmed because he could only think four moves ahead but he continued an active mental life until his death in 2003. Autopsy showed that his brain was riddled with the plaques and tangles that are characteristic of Alzheimer’s. Most people would have been reduced to a state of total confusion. This does not mean that cognitive abilities fail to decline with age. After all, at any given age, an athlete is better off for training. But however hard you train, your times will get slower as you age.

The brain is much more like our muscles than we had thought, even in the sense that specialized exercise affects different parts of the brain. Autopsies show that the brains of London taxi-drivers are peculiar. They have an enlarged hippocampus, which is the brain area used for navigating three-dimensional space. Here we see spatial abilities being developed without comparable development of other cognitive skills. To develop a wide variety of cognitive skills you need a wide variety of cognitive exercises.


Interventions that may enhance IQ include the efforts of parents, programs that afford an enriched environment to children at risk, adoption, and university study. Beginning with the family, the Dickens/Flynn model posits a tug of war between two environments: the environment parents impose, which is not directly correlated with the child’s unique genetic endowment; and the environment the child creates by interacting with the world, which does tend to match the child’s unique genetic endowment. With each passing year, a child transcends parental influence and becomes an autonomous actor. Parents cannot prevent their child from rebelling against a teacher with whom there is little rapport or getting in with the wrong crowd.

Preschool interventions also impose an environment on children that is uncorrelated with their genes, usually a uniformly enriched one that includes stimulation through educational toys, books, contact with sub-professionals, and so forth. If these terminate as children enter school, the intervention is likely to lose the tug of war even earlier than a child’s parent do. After all, the parents retain some influence after the preschool years. Since the imposed environment was far more enriched than any available at school, the children will begin to match environments that get further and further below its quality.

The most radical form of environmental intervention is adoption into a privileged home. Adoptive parents often wonder why the adopted child loses ground on their natural children. If their own children inherit elite genes and the adopted child has average genes, then as parents slowly lose the ability to impose an equally enriched environment on both, the individual differences in genes begin to dominate.

Finally, note that university education is a partial attempt to impose an enriched environment on students regardless of their genetic differences – that is, it constitutes a quasi-environmental intervention on the adult level. It too will see its effects on IQ fade unless quality of environment is maintained, for example, unless thanks to a good university education a student of average ability qualifies for a cognitively demanding profession. Then the job takes over the university’s role of imposing duties that foster the intellect.

Note that none of this means that interventions have no lasting effect; it is just that their non-IQ effects are likely to be more permanent than their IQ effects. If parents encourage persistence, honesty, and self-esteem, their children have a good start in life that may prove far more important than their gaining a few jumps on the IQ hierarchy. Similar characterological enhancement has been claimed for intervention programs like Head Start. Contacts made at a good university may confer an enhanced income and socioeconomic status throughout life.

These comments about interventions may seem to imply that no one can really hope to improve on his or her genetic endowment. This pessimism is no more in order than pessimism about whether people can improve on their physical endowment for running. To do so, you must either have good luck or make your own luck. Either a happy chain of circumstances will force you to train throughout your life or you can develop a love for running and train without compulsion. Training will not override genes entirely, of course. There are runners I cannot beat even when I train more than they do. But I can run rings around every couch potato within 20 years of my age.

There is one way in which individuals can make their own luck. Internalize the goal of seeking challenging cognitive environments — seek intellectual challenges all the way from choosing the right leisure activities to wanting to marry someone who is intellectually stimulating. The best chance of enjoying enhanced cognitive skills is to fall in love with ideas, or intelligent conversation, or intelligent books, or some intellectual pursuit. If I do that, I create within my own mind a stimulating mental environment that accompanies me wherever I go. Then I am relatively free of needing good luck to enjoy a rich cognitive environment. I have constant and instant access to a portable gymnasium that exercises the mind. Books and ideas and analyzing things are possessions easier to access than even the local gym.

If only we who teach could make more of our “subjects” fall in love with ideas. Then we would have truly effective interventions.

Three levels and three concepts

All of this has implications for the theory of intelligence. There is nothing really the matter with the concept of g; it is just that we have misused it by making it the omnipresent concept in our study of cognitive abilities. Intelligence is important on three levels, namely, brain physiology, individual differences, and social trends (collectively, BIDS). The core of a BIDS approach to intelligence is that each of those levels has its own organizing concept, and it is a mistake to impose the architectonic concept of one level on another. We have to realize that intelligence can act like a highly correlated set of abilities on one level and act like a set of functionally independent abilities on other levels.

Take the brain. Highly localized neural clusters are developed differentially as a result of specialized cognitive exercise, but there are also important factors that affect all neural clusters such as blood supply, dopamine as a substance that renders synapses receptive to registering experience, and the input of the stress-response system. When we map the brain’s structure, we find a mixture of commonality and neural decentralization. The commonality may well give rise to g on the individual differences level, while the decentralization leads to the phenomenon of various cognitive skills developing independently over time.

As for individual differences, that is the proper kingdom of g. There is simply no doubt that performance differences between individuals on a wide variety of cognitive tasks are correlated primarily in terms of the cognitive complexity of the task or the posited cognitive complexity of the path toward mastery. However, we need to avoid the mistake of thinking that the interaction between genes and environment is less complex than the reality.

On the social level, it is also beyond doubt that various real-world cognitive skills show different trends over time as a result of shifting social priorities. The appropriate dominant concept on this level is not g but something like social utility.

In closing, I want to stress that the BIDS approach does not aim at the abolition of g. It merely endorses a separation of powers that gives each dominant construct the potency needed to rebuff the other two. The U.S. Constitution attempts to make the President, Congress, and Supreme Court dominant in the executive, legislative, and judicial areas, respectively. I want the same kind of separation of powers for the three levels of intelligence.

James R. Flynn is emeritus professor of political studies at the University of Otago in New Zealand and author of What is Intelligence: Beyond the Flynn Effect (Cambridge 2007), from which this essay is adapted.

Also from this issue

Lead Essay

  • In this month’s lead essay, famed intelligence researcher James R. Flynn draws on his new book, What Is Intelligence? Beyond the Flynn Effect, to help answer some of the puzzles surrounding the controversial issue of IQ. Flynn, who first discovered that IQ scores were rising generation after generation, notes that this improvement has been more dramatic on some parts of IQ tests than on others, challenging ideas of a unified general intelligence or “g.” Flynn argues that the fact cognitive abilities do not develop together has important implications for education. Additionally, Flynn contends that the evidence for the co-determining reciprocal influence of brain and environment should lead us to set aside simple ideas about the primacy of nurture and nature in intelligence. Once we grasp that “the brain is much more like our muscles than we had thought,” we can do more to improve cognitive performance by doing more to exercise the brain.

Response Essays

  • Linda Gottfredson, co-director of the Delaware-Johns Hopkins Project for the Study of Intelligence and Society, defends the unity of general intelligence, or g, against Flynn’s attempt to “unravel g into its component parts” by charging that “his core argument rests on logical fallacies that profoundly misinterpret the evidence.”

  • University of Virginia psychologist Eric Turkheimer contests even James Flynn’s modest accommodation with the construct of g, or general intelligence. Turkheimer argues that the “fundamental intuition” of g is that “universal positive relations among mental tests compels a single dominant explanatory construct,” but that “the fundamental intuition is wrong.” According to Turkheimer, g is not “discovered,” but is simply posited as a convenience, like the Prime Meridian. g may be useful for some purposes, but a multidimensional explanation of ability may still be correct. “The trick is not to get hooked on any particular way of dividing up the pie,” Turkheimer writes, “because it is a short step from there to trying to find the Greenwich Meridian at the bottom of the North Atlantic.”

  • Stephen J. Ceci, the Helen L. Carr Professor of Developmental Psychology at Cornell University, explains why Flynn’s work has been so important. According to Ceci, the received wisdom about the nature and effects of g, or general intelligence, led “some to argue that inequality in the distribution of wealth, prestige, and educational attainment is, in part, a consequence of unequal distribution of the intellectual capacity needed for high levels of functioning.” However, Ceci says, “[Flynn] has shown beyond doubt that general intelligence fluctuates systematically over time and this cannot be due to our having better genes than our grandparents,” and he goes on to explore the puzzles raised by this discovery.