movies about intelligence: the limitations of g Copyright © 2003 American Psychological Society CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 95 References Akins, C.K., Klein, E.D., & Zentall, T.R. (2002). Im- itative learning in Japanese quail (Coturnix japonica) using the bidirectional control proce- dure. Animal Learning & Behavior, 30, 275–281. Akins, C.K., & Zentall, T.R. (1996). Imitative learn- ing in male Japanese quail (Coturnix japonica) using the two-action method. Journal of Com- parative Psychology, 110, 316–320. Akins, C.K., & Zentall, T.R. (1998). Imitation in Japanese quail: The role of reinforcement of demonstrator responding. Psychonomic Bulletin & Review, 5, 694–697. Bandura, A. (1969). Social learning theory of iden- tificatory processes. In D.A. Goslin (Ed.), Hand- book of socialization theory and research (pp. 213– 262). Chicago: Rand-McNally. Custance, D.M., Whiten, A., & Bard, K.A. (1995). Can young chimpanzees imitate arbitrary ac- tions? Hayes and Hayes revisited. Behaviour, 132, 839–858. Dorrance, B.R., & Zentall, T.R. (2001). Imitative learning in Japanese quail depends on the mo- tivational state of the observer at the time of observation. Journal of Comparative Psychology, 115, 62–67. Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G. (1996). Action recognition in the premotor cor- tex. Brain, 119, 593–609. Heyes, C.M., & Dawson, G.R. (1990). A demon- stration of observational learning in rats using a bidirectional control. Quarterly Journal of Ex- perimental Psychology, 42B, 59–71. Heyes, C.M., & Ray, E.D. (2000). What is the signif- icance of imitation in animals? In P.J.B. Slater, J.S. Rosenblatt, C.T. Snowdon, & T.J. Roper (Eds.), Advances in the study of behavior, Vol. 29 (pp. 215–244). New York: Academic Press. Klein, E.D., & Zentall, T.R. (in press). Imitation and affordance learning by pigeons. Journal of Comparative Psychology. Meltzoff, A.N., & Moore, M.K. (1989). Imitation in newborn infants: Exploring the range of ges- tures imitated and the underlying mecha- nisms. Developmental Psychology, 25, 954–962. Mitchell, C.J., Heyes, C.M., Gardner, M.R., & Daw- son, G.R. (1999). Limitations of a bidirectional control procedure for the investigation of imita- tion in rats: Odour cues on the manipulandum. Quarterly Journal of Experimental Psychology, 52, 193–202. Mitchell, R.W. (2002). Kinesthetic-visual matching, imitation, and self-recognition. In M. Bekoff, C. Allen, & G.M. Burghardt (Eds.), The cogni- tive animal (pp. 345–351). Cambridge, MA: MIT Press. Piaget, J. (1962). Play, dreams, and imitation in child- hood (C. Gallegno & F.M. Hodgson, Trans.). New York: Norton. Zentall, T.R., Sutton, J., & Sherburne, L.M. (1996). True imitative learning in pigeons. Psychologi- cal Science, 7, 343–346. Movies About Intelligence: The Limitations of g James R. Flynn1 Department of Political Studies, The University of Otago, Dunedin, New Zealand Abstract There is a strong tendency for the same people to do better or worse on a wide variety of IQ tests. On this basis, some psy- chologists posit the concept of g, or a general intelligence factor. Does g show that performance on a wide range of cognitive tasks is influenced by individual differences in brain quality? It may, but if so, g lacks a sociolog- ical dimension and cannot ex- plain cognitive trends over time or assess their significance. It also encourages a paradox about nature versus nurture and over- simplifies the causes of the Black-White IQ gap. Keywords g; intelligence; IQ gains; race No matter whether mental tests feature vocabulary, general informa- tion, verbal oddities, scrambled sen- tences, logical reasoning, number series, pictorial oddities, spatial anal- ogies, or completion of matrices, the same people tend to do better or worse. Statistical analysis of the data suggests that a single factor accounts for much of this tendency toward consistent performance, and that common factor is what psychologists call g. It functions like a correlation coefficient with a value of .65. For ex- ample, assume a correlation between height and basketball performance of .65. Given this correlation, if we selected a sample of people at the 84th percentile for height, they would average at the 74th percen- tile for basketball performance. Sim- ilarly, if we knew people’s g scores, we could predict how well they would perform when using a huge range of cognitive skills. Cognitive performance in every- day life is influenced by g. Siblings who are superior to their co-siblings for g tend to enjoy greater academic success; making an omelet is a more cognitively complex task than scram- bling eggs and therefore has a higher g loading. Jensen believes he knows why g influences cognitive perfor- mance: “Some property (or proper- ties) of the brain . . . has cognitive manifestations that result in the emer- gence of g” (Jensen, 2002, p. 153). In other words, Jensen believes that g measures the influence of brain qual- ity. And a better brain gives you an advantage in school, on the job, wher- ever cognitive skills are relevant. The g we calculate by the tech- niques available today may not be a pure measure of brain quality; that is, it may be diluted by picking up the influence of nonphysiological factors like individual differences in motiva- tion. However, throughout this arti- cle, I pretend that Jensen’s ideal of a purely physiological g has been real- ized. I do this in order to show that the closer we approach that ideal, the more g suffers from a peculiar limita- tion: It becomes sociologically blind. The symptoms are most evident when we take a look at the evolution of cognitive skills over time. g CANNOT DETECT SOCIOLOGICAL FACTORS Since 1950, the populations of The Netherlands, Belgium, Israel, 96 VOLUME 12, NUMBER 3, JUNE 2003 Published by Blackwell Publishing Inc. and Argentina have shown gains of about 18 IQ points per genera- tion (30 years) on a test of cognitive ability called Raven’s Progressive Matrices. Gains on Wechsler tests (e.g., the Wechsler Intelligence Scale for Children, or WISC), which are often used to measure IQ, have averaged at least 9 points per gen- eration. These IQ gains have been influenced by sociological factors, sometimes almost entirely. There- fore, a g that refers to brain physi- ology can do little to explain them. The main candidates for physio- logical factors that might increase IQ scores are nutrition, advances in obstetrics, and increased outbreed- ing (hybrid vigor). Improved nutri- tion has been important at certain times and places, but not at others. In America, recent IQ gains have been as large at the middle and top of the curve as at the bottom. Be- cause one would expect improved nutrition to affect primarily the most deprived, and produce dis- proportionate gains at low IQ lev- els, nutrition does not seem to have played an important role in causing U.S. gains, at least not since 1950. Similarly, post-1950 improve- ments in obstetric and neonatal care have probably had no net effect. For every child who has escaped mental impairment, another who would have died without modern tech- niques has been saved. As for hybrid vigor, inbreeding within a small group has a nega- tive effect on IQ. If American his- tory was a story of little isolated communities being replaced by a highly mobile society, that might help explain the massive IQ gains America has made throughout the 20th century. However, Americans n e v e r d i d l i v e i n s m a l l i n b r e d groups. There was always a huge influx of migrants who settled in both urban and rural areas. There were huge population shifts dur- ing settlement of the West, after the Civil War, and during the World Wars. The growth of mobility has b e e n m o d e s t : I n 1 8 7 0 , 2 3 % o f Americans were living in a state other than the one of their birth; in 1970, the figure was 32%. Sociological explanations of IQ gains seem more promising. Be- t w e e n 1 9 4 8 a n d 1 9 8 9 , A m e r i c a gained the equivalent of 20 IQ points on the WISC subtest called Similarities. Similarities asks ques- tions like, “What do dawn and dusk have in common?” Answer- ing such questions demands solv- ing problems on the spot without a learned method for doing so. Dur- ing the same period of time, gains on WISC subtests like Arithmetic, I n f o r m a t i o n , V o c a b u l a r y , a n d R e a d i n g C o m p r e h e n s i o n w e r e comparatively modest or nil. The content of these tests is very close to school-taught subjects. How can we explain this puzzling pattern, that over the years we have be- come more mentally agile but learn no better at school? America’s post-1950 affluence brought smaller families in which c h i l d r e n ’ s “ w h y s ” w e r e t a k e n more seriously. More leisure made it possible to enjoy cognitive chal- lenges ranging from chess to video games (Greenfield, 1998). And more professional work roles de- manded independent thinking on the job (Schooler, 1998). People be- came more disposed to invest men- tal energy into problem solving for its own sake, or at least problem solving on the spot. At the same time, homework was resented and too much focus on basics thought boring. Americans became unwill- ing to see formal schooling become more cognitively demanding. The result of these two attitude shifts was that score gains accelerated on the Similarities subtest, and score gains faltered on the school-related Wechsler subtests. Even if I am mistaken in detail, sociological factors of some sort were the dominant cause of U.S. IQ gains. They would have been im- portant even in nations where nu- trition made a contribution. Jensen’s g simply does not provide a con- ceptual framework for identifying those factors: It puts its eggs in the physiological basket. g CANNOT ASSESS SOCIAL SIGNIFICANCE It is natural to ask whether IQ gains are g gains. Jensen answers t h i s q u e s t i o n b y u s i n g w h a t i s called the method of correlated vec- tors. For example, you rank the 10 subtests of the WISC in terms of the size of their IQ gains (Arith- metic at the bottom with nil gains and Similarities at the top with huge gains), you rank the same 10 in terms of their g loadings (once you have extracted g from a set of tests, you can see how much scores on each test correlate with g itself), and then you see if the two rank- ings are positively correlated. I believe this method has severe limitations. But setting those aside, if IQ gains were not g gains, would that drain them of social signifi- cance? IQ gains reflect fascinating trends in American intellectual life. The fact that we are better at on- the-spot problem solving is indica- tive of real-world cognitive gains. We appear better than we used to be at leisure activities that are cog- nitively demanding. The quality of play in chess tournaments has es- calated, and this trend may extend to games like bridge (Howard, 1999; Nunn, 1999). The level of po- litical debate has been enhanced (Rosenau & Fagan, 1997). The fact that more people can think inde- pendently helps fill the large num- ber of professional work roles the industrial revolution demands (Schooler, 1998). In addition, the fact that Americans have no greater arithmetic skills, nonspecialized vo- cabulary, or knowledge of general information has profound social significance. Copyright © 2003 American Psychological Society CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 97 S o d e t e r m i n i n g w h e t h e r I Q gains are g gains offers no criterion for assessing their significance— except that g gains would mean that brain physiology had been en- hanced. However, that is not a pre- requisite for progress. There is no reason to believe that the glory that was Greece, the grandeur that was Rome, or the Italian Renaissance was due to anything but cultural change. Fortunately, society can exploit unrealized brain potential, and fortunately, society can ignore the fact that all cognitive skills are correlated at a particular time and place. Over time, it can pick and choose among skills for emphasis. It can encourage on-the-spot prob- lem solving without performing the (perhaps impossible) task of getting us to do more homework. g CANNOT DETECT MULTIPLIERS Some 30 years ago, Jensen (1973) noted that twin studies gave a low correlation between IQ and envi- ronment (about .33). He concluded that for environment to cause two groups to differ in IQ by one stan- dard deviation, their environments would have to differ by three stan- dard deviations. In effect, the envi- ronment of virtually everyone in the higher-IQ group would have to be better than the average environ- ment of the lower-IQ group. This creates a paradox: IQ gains have of- ten been more than one standard deviation per generation, yet posit- ing three or four standard devia- tions of environmental progress over one generation seems absurd. The twin studies seem to show that what is known to be true (IQ gains are caused by the environment) cannot be true. How can twin stud- ies show environmental effects to be so feeble and IQ gains show them to be so potent? Let us see what sociol- ogy can do to answer this question. I d e n t i c a l t w i n s s e p a r a t e d a t birth and raised apart grow up w i t h s i m i l a r I Q s . T h e o b v i o u s cause is their identical genes. But things are not that simple. Identical genes tend to get matched with very similar environments and thereby co-opt the potency of pow- erful environmental factors. Con- sider a sport analogy. Your basket- ball genes are slightly better than average, and you are born a bit taller and quicker than average. You live in the basketball-mad state of Indiana. When you go to school, you are a bit better at bas- ketball than your classmates, so you are picked more often to play, practice more than most of them, make your school team, and get professional coaching. In contrast, people whose genes make them a bit shorter and stodgier than aver- age will get matched with a much worse basketball environment. In Indiana, if identical twins are g e n e t i c a l l y p r o g r a m m e d t o b e taller and quicker than average to the same degree, then despite be- ing raised apart, they will tend to get matched to basketball environ- ments of about the same degree of superiority. What would a twin study show? Very similar basket- ball skills, for which their identical genes would get all the credit. The fact that both twins benefited from more practice than their peers, making a school team, and getting professional coaching would be overlooked. Now for IQ. If John is born with a bit better brain than James, who will like school, get praised for his schoolwork, haunt the library, and get into advanced classes? And if John has a separated identical twin, who enjoys much the same school experience, what will really ac- count for their similar adult IQs? Not identical genes alone. Rather, the ability of those identical genes to co-opt environments of similar quality will be the missed piece of the puzzle. Within a generation, genes profit from seizing control of a powerful instrument that multiplies their causal potency. A gene-caused abil- ity advantage upgrades the school environment by more homework be- ing done, which upgrades the ability advantage, which upgrades the en- vironment by entry into a top aca- demic track, which upgrades ability further. Each feedback loop acts as a potent multiplier. Could some per- sistent environmental factor have been at work between generations, seizing control of a multiplier pow- erful enough to have caused the massive IQ gains of recent decades? Then our paradox would be solved. There would be huge environmen- tal effects on the average IQ differ- ence between generations—effects quite consistent with genetic domi- nation of individual IQ differences within each generation. The persistent environmental factor that has been at work is the industrial revolution with its social trends, smaller families, more cog- nitively demanding leisure, and more cognitively demanding work roles. As for the powerful multi- plier these trends have used, we might call it the “social multiplier.” Its essence is that rising average performance becomes a potent causal factor in its own right. Back to sports. About 1950, the advent of television sparked much greater and keener participation in basketball. This raised the general skill level; you had to shoot accu- rately to be better than most other players. Then you had to be able to p a s s w i t h e i t h e r h a n d , t h e n t o shoot with either hand. In other words, every escalation of the aver- age performance in the general population meant every individ- ual had to improve to keep up, which escalated the average per- formance further, which meant a new challenge to each individual— so the multiplier produced a huge escalation of skills in a single gen- eration. The same thing happened 98 VOLUME 12, NUMBER 3, JUNE 2003 Published by Blackwell Publishing Inc. after 1948 for on-the-spot problem solving. Society made new de- mands on the cognitive content of conversation, leisure, and work. This raised the average perfor- mance, and then everyone had to respond to keep up, which raised the average performance further, producing a huge escalation of skills in a single generation. So all is clear: Twin studies re- flect situations in which genes drive powerful multipliers; mas- sive IQ gains occur when environ- mental trends seize control of pow- erful multipliers. It all depends on whether genes or the environment is in the driver’s seat. Sociology can solve our paradox, and a g-ocentric view of intelligence cannot. If you focus primarily on g and the fact that g differences between individ- uals are genetically influenced, the paradox simply makes you want to find that IQ gains are caused by some genetic factor like hybrid vigor. Jensen said that twin studies show how improbable it is that the IQ gap between Black and White Americans is environmental. After all, that IQ gap amounts to a full standard deviation. Who could ar- gue that the average Black environ- ment is three standard deviations below the average White environ- ment? Such an analysis begs the question: Are Blacks like individu- als within White society who on average have inferior genes for mental abilities, or instead are Black-White differences in IQ more like generational differences due to persistent environmental factors? M c W o r t e r ( 2 0 0 0 ) b e l i e v e s t h a t Black Americans have a sense of victimhood that makes them shun mainstream American culture and see school achievement as selling out to White culture. If so, they would be ambivalent about match- ing average school performance. Therefore, the social multiplier would spiral the average down- ward rather than upward! HOW g UNRAVELS: TWO KINDS OF COMPETITION We use our cognitive abilities to compete with one another. Compe- tition to win creates g, and compe- tition to keep up destroys g. These two kinds of competition refer not to different motives (one always wants to win), but to different con- texts. Competition to win is a static competition. At a particular time, each person’s cognitive perfor- mance is measured against the cog- nitive performance of others. That kind of competition tends to pro- duce a single pecking order. If all players have a level playing field, who wins at a given place and time i s i n f l u e n c e d b y d i f f e r e n c e s i n brain quality. That common factor crosses the boundaries between various cognitive skills and weaves them together into g. Competition to keep up is dy- n a m i c . I t o p e r a t e s o v e r l o n g stretches of time and combines progress with anarchy and simply unravels g. The mean of on-the- spot problem-solving skills begins to rise and people compete to keep up, while the mean of boring old arithmetic skills is immobile and people relax. Every day, the enor- mous potency of an active versus an inactive social multiplier wid- ens the gap between Similarities and Arithmetic scores without any regard to their g loadings. g’S VANISHING ACT: A TRIP TO THE MOVIES Consider two films. The first is about the life history of individu- als. Some become doctors; others cannot pass the chemistry course to get into medical school. Their per- formance over a wide range of ar- eas is influenced by better or worse brains—and there is g. The second film is about American society since 1950. Shifting social priori- ties raise cognitive skills in one area, stall them in another, with no pattern except that set by the prior- ities themselves. Average brain quality neither improves nor de- clines—and g vanishes. Individual differences in brain quality still ex- ist, of course. But because they have no influence on the pattern of skill gains, and because g is a mea- sure of their influence, skill gains simply will not evidence g. You can get a look at g when- ever you want, but you have to go to the right film. To lament that you cannot see it when differences in brain physiology do not count is t o f o r g e t w h a t J e n s e n ’ s g i s a l l about. That g, ideally at least, is a pure measure of the brain’s influ- ence on cognitive performance. What society would prioritize cognitive abilities according to which were most influenced by brain physiology? If a society were bizarre enough to do that, you would get IQ gains on various WISC subtests that matched the g loadings of those subtests. But that would not be true g but g mim- icked. It would be a mere likeness painted by some mad dictator who had decided to encourage (or dis- courage) cognitive skills in terms of how much they were brain influ- enced rather than in terms of social priorities. It would be like a society that refused to allow people to choose between improving their basketball and baseball skills, but rather, imposed rewards and pen- alties in favor of baseball on the grounds that it was the sport in which performance was most af- fected by human physiology. WHAT IS TO BE DONE? What research might advance our understanding of human cog- nition, individual and group differ- Copyright © 2003 American Psychological Society CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 99 ences, and how to enhance cogni- tive skills? We need to identify the brain processes that influence cogni- tion. Jensen has found correla- tions between g and elementary cognitive tasks (mental process- ing speed), the brain’s electrical response to stimuli, and how quickly an injection of glucose is absorbed by the brain. Hope for further advance in this area lies in new techniques of viewing what brain centers are active when different cognitive tasks are being done. We should learn more about so- cial multipliers. Boozer and Cac- ciola (2001) showed that when reduced class size raises aca- demic performance, peer inter- action multiplies that rise and accounts for virtually all of the long-term gains. The relative potency of Whites’ and Blacks’ social multipliers should be compared. Although teaching children “how to think” is desirable, we should recognize that this will not neces- sarily enhance numeracy and lit- e r a c y . T h e f o c u s m u s t b e o n • • • • teaching reading and arithmetic skills. And note that if we really want to enhance those skills, there will have to be an attitude shift, so that Americans welcome core subjects that make greater cognitive demands. If all parents and children were like Chinese Americans, the “nation’s report card” would improve dramati- cally. Above all, we must go beyond g to develop a theory of intelli- gence with a sociological dimen- sion. In this theory, g will still play an important role. Within every generation, people com- pete to win, and, therefore, g will always help explain why some people excel across so many cog- nitive skills. Recommended Reading Deary, I.J. (2001). Intelligence: A very short introduction. Oxford, En- gland: Oxford University Press. Dickens, W.T., & Flynn, J.R. (2001, April 21). Great leap forward. New Scientist, 170, 44–47. Jensen, A.R. (1998). The g factor: The science of mental ability. Westport, CT: Praeger. • Note 1. Address correspondence to J.R. Flynn, POLS, University of Otago, Box 56, Dunedin, New Zealand; e-mail: jim.flynn@stonebow.otago.ac.nz. References Boozer, M., & Cacciola, S.E. (2001). Inside the black box of Project STAR: Estimation of peer effects us- ing experimental data (Center Discussion Paper No. 832). New Haven, CT: Yale University Economic Growth Center. Greenfield, P. (1998). The cultural evolution of IQ. In U. Neisser (Ed.), The rising curve: Long-term gains in IQ and related measures (pp. 81–123). Washington, DC: American Psychological As- sociation. Howard, R.W. (1999). Preliminary real-world evi- dence that average intelligence really is rising. Intelligence, 27, 235–250. Jensen, A.R. (1973). Educability and group differ- ences. New York: Harper and Row. Jensen, A.R. (2002). Galton’s legacy to research on in- telligence. Journal of Biosocial Science, 34, 145–172. McWorter, J. (2000). Losing the race: Self-sabotage in Black America. New York: Free Press. Nunn, J. (1999). John Nunn’s chess puzzle book. Lon- don: Gambit Publications. Rosenau, J.N., & Fagan, W.M. (1997). A new dyna- mism in world politics: Increasingly skilled in- dividuals? International Studies Quarterly, 41, 655–686. Schooler, C. (1998). Environmental complexity and the Flynn effect. In U. Neisser (Ed.), The rising curve: Long-term gains in IQ and related measures (pp. 67–79). Washington, DC: Ameri- can Psychological Association. Spanking Children: Evidence and Issues Alan E. Kazdin1 and Corina Benjet Child Study Center, Yale University School of Medicine, New Haven, Connecticut (A.E.K.), and National Institute of Psychiatry, Mexico City, Mexico (C.B.) Abstract W h e t h e r o r n o t t o s p a n k children as a discipline practice is controversial among lay and professional audiences alike. This article highlights different views of spanking, key conclu- s i o n s a b o u t i t s e f f e c t s , a n d methodological limitations of the research and the resulting ambiguities that fuel the current debate and plague interpreta- tion. We propose an expanded research agenda to address ques- tions about the goals of parental discipline; the role, if any, that punishment plays in achieving these goals; the effects and side effects of alternative discipline practices; and the impact of punishment on underlying de- velopmental processes. Keywords spanking children; punish- ment; parent discipline Spanking as a way of disciplining children is a topic of broad interest to people involved in the care and edu- cation of children (e.g., parents, teach- ers), as well as to the many profes- sions involved with children, parents, and families (e.g., pediatrics, psychi- atry, psychology, and social work). Hitting children is intertwined with religious beliefs, cultural views, gov- ernment, law, and social policy and