Brain and Mind

What Should We Ask About Intelligence

In this article Professor Sternberg argues that what he calls ‘the IQ revolutionaries’ are united by a belief that IQ represents too narrow a conception of intelligence and share the view that intelligence can be modified.

Robert J. Sternberg is IBM Professor of Psychology and Education, Yale University.
This article appeared in the Spring 1996 issue of The American Scholar, Vol. 65, No. 2. Copyright © 1996 by the author.

Alice was the admissions officer's dream. She had 800s on her boards, close to a 4.0 average, and glowing letters of recommendation from her teachers. She was the proverbial good student with paper credentials that just couldn't be beaten. She was accepted by every graduate program to which she applied. We were thrilled when she decided to matriculate in Yale's doctoral program in psychology.

The tests and other predictors of success were correct as far as they went: Alice excelled in her first year of course work, competing with just one other student for the highest academic average in an already highly selective program. Then something went wrong something big. By the time Alice was done with the program, she was roughly in the bottom 20 percent of her cohort. And it wasn't that she didn't try. On the contrary, she was highly motivated to succeed.

What went wrong with Alice is what has gone wrong with thousands and thousands of students: they are brilliant when it comes to remembering and even analyzing ideas, but they are dim when it comes to generating their own ideas. They may have 700s or even 800s on their boards, and often lQs of 140 and above, but they seem to lack even an ounce of creativity. In other terms, they are analytically, but not creatively, intelligent.

Contrast the fate of Alice to that of Barbara. Barbara applied to Yale's graduate program in psychology with good but not outstanding grades. More notable were her superb letters of recommendation from eminent people and her record of published work. The impressive creativity of this work was apparent to almost anyone who took the trouble to read it. But Barbara's test scores, although not awful, were modest. Barbara, unlike Alice, was rejected. Moreover, the Barbaras of the world tend to be rejected, not just by Yale, but by other highly competitive programs that can give a future scholar a head start in academic life, a future lawyer a stepping-stone to the world of law, or a future doctor an edge up on the best internships.

Barbara was one of the lucky ones. I hired her as a research associate. She demonstrated exceptional creative abilities, and two years later, when she reapplied to our graduate program, she was admitted as the top pick. She even received a special fellowship reserved for top applicants. But for every Barbara who gets a chance, there are unknown thousands like her who are consigned to the academic waste basket - they never get the chance that Barbara got. We never find out what happened to them, because we never give them the chance to show us what they might have done.

Paul might have seemed like the ideal admissions' candidate. He combined Alice's analytical ability with Barbara's creative ability. His professors were delighted with him and expected him to be a smash hit on the academic job market. Actually, the professors weren't the only ones impressed with Paul; Paul was too, and it showed.

When Paul went on the academic job market, he was asked to interview at every institution to which he applied-an enviable record. His hit rate for getting jobs wasn't quite so enviable, however: he was offered only one position, at the weakest department to which he applied. Clearly, he was far from a desirable commodity, his analytical and creative abilities notwithstanding.
Ironically, Sam, who had received no interview offers at all in the first round, was later offered many of the interview opportunities Paul had initially flubbed. Sam was offered several positions and within a few years had tenure, whereas Paul was out of a job. Sam's work was nothing special, but he knew what his department valued, and he delivered. What went wrong with Paul was straightforward: Paul was so lacking in common sense that he couldn't hide his arrogance even on the one day he needed to hide it-the day on which he had a job interview. And once hired, his arrogance led him quickly to become a pariah among his colleagues. Paul was analytically and creatively intelligent but lacking in practical intelligence. Sam, more modest in analytical and creative intelligence, was able to translate his practical intelligence into good, although perhaps not distinguished, career success.

The stories of Alice and Barbara and of Paul and Sam are all true, with only the names changed. They are also, in their themes, common stories in academe. But stories like these are what have led many psychologists, myself among them, to conclude that conventional notions of intelligence may be correct as far as they go but that they do not go far enough. These psychologists have suggested that conventional notions of intelligence (a) define intelligence too restrictively and (b) often provide reasonable answers, but to narrow questions. The problem is that the answers may be fine, but the questions are not.

Today, the field of intelligence is going through a heated, no holds-barred battle between adherents to a conventional paradigm that has its roots at the turn of the century and adherents to new paradigms that are attempting to turn the old paradigm on its head. The adherents to the old paradigm have reacted in various ways to the revolutionaries, all of these ways predicted in spirit by Thomas Kuhn in The Structure of Scientific Revolutions. The traditionalists' reactions are similar to those of any entrenched power structure: they ignore the revolutionaries, hoping they will go away or not be noticed; or they give them a glancing notice but try not to take them seriously; or they fight them head-on.

Some traditionalists choose to ignore the revolutionaries altogether. Other traditionalists, such as Richard Herrnstein and Charles Murray in The Bell Curve, briefly acknowledge the existence of the revolutionaries and then move on. But the revolutionaries are becoming harder to ignore. For example, every major college-level textbook in introductory psychology now prominently features two of the revolutionary theories, my own theory and that of Howard Gardner, a kindred spirit. Moreover, research as well as theory in the field of intelligence more and more is reflecting the revolutionary paradigms.

Still other traditionalists, among them Malcolm Ree, Jack Hunter, and Frank Schmidt, have joined the battle and chosen to take on the revolutionaries, responding to them directly in print. But what is the battle about, anyway? On what grounds is it being fought?

The grounds of the intellectual battle ought to be over "What should we ask about intelligence?" Consider a concrete example. The traditionalists take a battery of conventional tasks used to measure intelligence, such as the tasks on an IQ test, and ask: "What is the latent structure of intelligence underlying observable scores on conventional tasks used to measure intelligence?"

This is the question that Charles Spearman sought to answer in his 1904 analysis of intelligence-test performance, and it is the question that traditionalists have sought to answer ever since. Practically speaking, traditionalists are intellectual descendants of Edwin Boring, who in 1923 espoused the operationist dictum that intelligence is what intelligence tests test.

Although these traditionalists agree that intelligence tests measure intelligence adequately, they do not agree as to the latent structure underlying performance on the tests. Some theorists, like Spearman in the past as well as Arthur Jensen and Richard Herrnstein and Charles Murray in the present, have believed in a single, general factor of intelligence (called g), represented by a global IQ (intelligence-quotient) score; other theorists, like Cyril Burt and Philip E. Vernon in the past and John B. Carroll and Jan-Eric Gustafsson in the present, have believed in a hierarchy of abilities, with general ability at the top of the hierarchy, and successively more narrow levels of abilities below g. And other points of view have been presented as well.

The revolutionaries do not accept the answers of any of the traditionalists, because they do not accept their question. They believe that the kinds of tasks used in conventional tests of intelligence are largely arbitrary and lacking in any theoretical basis. The revolutionaries do not believe that conventional tests adequately sample the universe of tasks needed to assess intelligence. These researchers believe that the general factor is in part an artifact of the method used to analyze test scores-factor analysis. This method is mathematically designed to maximize the amount of variation that occurs in the first factor, thus yielding a general factor as a result of mathematical rather than psychological necessity. But more important, they believe that g is an artifact of the narrow range of kinds of tests conventionally used to measure intelligence. In their view, the general factor would disappear if the tests were more widely conceived and based on a broad, well-specified theory of intelligence.

Thus, these researchers believe that what traditionalists believe they know is a result of the traditionalists' asking too narrow a question. The revolutionaries see the so-called intelligence quotient-the IQ-as neither a quotient nor an indicator of intelligence, broadly defined. Indeed, IQs today are virtually never computed as quotients, but rather as scores derived from properties of normal statistical distributions. And they are based on a notion of intelligence that the revolutionaries believe the field should recognize as outdated.

The revolutionaries are uncomfortable with the fact that the two major tests used to measure intelligence, the Stanford-Binet and the Wechsler, are very old in their conception. The Binet tests date back to the first decade of the twentieth century; the Wechsler tests to the fourth. Successive editions of the Binet and the Wechsler tests differ largely in cosmetic ways from the original versions. Although the statistics are more sophisticated and the printing of the test booklets is better, the technology of intelligence testing has changed little in almost a century.

What kinds of abilities do the revolutionary theories of intelligence encompass? One such theory, Howard Gardner's theory of multiple intelligences, comprises seven abilities, which Gardner believes are distinct and relatively independent intelligences: (a) linguistic intelligence, used in reading a novel, writing a poem or an article such as this one, or generating an extemporaneous talk; (b) logical-mathematical intelligence, used in solving mathematical problems, proving logical theorems, or completing categorical or other forms of syllogisms; (c) spatial intelligence, used in finding one's way in unfamiliar terrain, figuring out how to fit suitcases into the trunk of a car, or figuring out where in a playing field a baseball batter's fly ball will land; (d) musical intelligence, used in remembering a tune, singing a song, or composing a sonata; (e) bodily-kinesthetic intelligence, used in dancing ballet, performing gymnastics, or playing tennis; (f) interpersonal intelligence, used in figuring out what other people mean from what they say, decoding what their facial expressions communicate, or deciding what is appropriate to say in an interaction with a superior; and (g) intrapersonal intelligence, used in understanding why one takes rejection so poorly, why one tends to be overconfident in certain instances, or why one has failed in achieving an important personal goal.

Gardner points out that conventional tests of intelligence measure only the first two, and sometimes the third, of the multiple intelligences. Moreover, their form of measurement is quite limited, often encompassing solely multiple-choice test items or test items that require only very restricted kinds of performances. Gardner has offered evidence from diverse sources to support his theory, including studies of brain damage, psychometric and experimental studies, and studies of psychological development, among other kinds of evidence.

Another such theory, my own triarchic theory of human intelligence, follows from observations such as those of Alice, Barbara, Paul, and Sam, as described earlier in the article. This theory holds that intelligence has three major aspects: analytical, creative, and practical. Conventional good test takers and good students tend to excel in analytical intelligence but not necessarily in the creative and practical aspects of intelligence. Underlying each of these aspects are various kinds of mental operations used to process information - for example, some operations that define what should be asked and other operations that then seek to answer what has been asked.

An implication of this theory is that the reason conventional intelligence tests predict school achievement as well as they do is that schools, like conventional tests, tend to emphasize analytical skills far more than they emphasize creative and practical skills. Indeed, the latter kinds of skills may even be punished, as when students who depart from a teacher's expectations or point of view find themselves graded down for having done so.

It is scarcely surprising that conventional intelligence tests should so well reflect the abilities required for conventional schooling. The first major intelligence test-that of Alfred Binet and Theodore Simon in turn-of-the-century France-was designed to distinguish students who were genuinely lacking in academic abilities from those who were behavior problems but not lacking in these abilities. Throughout their development, there has always been a close association between tests and academic performance, which is what the tests have been designed primarily to predict.

If tests predict academic performance fairly well, why do we need revolutionary conceptions of intelligence, or, indeed, any new conceptions at all? After all, the tests are doing fairly well what they were designed to do. Why not leave well enough alone?
Before we leave well enough alone, we need to ask, What is it about academic performance that the conventional tests predict so well? They do predict grades pretty well. But once we pass on to the college or especially the graduate level, just how important are grades? Who hasn't encountered the scholars whose brilliance seems to lie only in their ability to tear other people's ideas to shreds and not in their ability to come up with any significant ideas of their own? Who hasn't met the doctor, with a medical degree from a famous medical school, whose skills with patients seem not even to meet a minimal standard of human decency? How many major companies have not hired MBAs who graduated with honors but who quickly become lunch meat for hungry predators in the dog-eat-dog world of modern business?

In one study, a colleague and I sought to address the question of just what the tests predict by validating the Graduate Record Examination (GRE) as a predictor of various kinds of performance in our own graduate program in psychology at Yale. Together with Wendy Williams, I asked professors to rate all their primary advisees who were in the psychology doctoral program between 1980 and 1991. Forty professors, including some who had left during the time period, provided ratings on a I (low) to 7 (high) scale of 167 graduate students (68 males, 99 females) on five scales: (a) analytical abilities, (b) creative abilities, (c) practical abilities, (d) research abilities, and (e) teaching abilities. Means of these ratings did not differ significantly between the sexes.

We also had available students' scores on the GRE verbal, quantitative, and analytic sections, and on the advanced test (for 73 of the students, because this test is optional). We further obtained the students' GPAs during their first and second years of graduate training. We also had available ratings of each student's doctoral dissertation from three faculty readers (who did not include the primary advisor).

The range of GRE scores was considerable. For example, on the verbal section, the range was from 250 to 800, with a mean of 653 and a standard deviation of 96. (The standard deviation is a measure of the dispersion of scores, where roughly two-thirds of all scores fall between one standard deviation below the mean and one standard deviation above it.) Comparable data for the quantitative section ranged from 320 to 840, with a mean of 672 and a standard deviation of 78; and the analytic scores ranged from 410 to 810, with a mean of 656 and a standard deviation of 92. The range for advanced test scores was 490 to 850, with a mean of 690 and a standard deviation of 64.

We found that the GRE provided some prediction of first-year grades (correlations of 0.18 for the verbal section, 0.14 for the quantitative section, 0.17 for the analytical section, and most impressively, 0.37 for the advanced test). All of the correlations were statistically significant except that for the quantitative score. Thus, overall, the best predictor of future achievement was, perhaps not surprisingly, past achievement. But the GRE did not predict second-year grades.

The only consistent predictor of any of the more important measures of graduate performance (i.e., ratings of analytical, creative, practical, research, and teaching abilities, as well as of dissertation quality) was the GRE analytical section, and this prediction was for men only. Five of six correlations for this section were statistically significant for men, with a median correlation of 0.30, whereas none of the correlations were statistically significant for women, with a median correlation of 0.02. Thus, of eight sets of correlations (four GRE scores for each of the two sexes), only one set provided statistically significant results.

One could argue, of course, that the less than impressive predictive performance of the GRE reflects the restricted range of scores at Yale. But this explanation fails. For one thing, the standard deviations for the aptitude tests (96, 78, 92) were not far from the national average. For another, the best predictor of first-year grades, the advanced test, had the lowest standard deviation (64) of all the predictors. Moreover, the analytical test did in fact predict the ratings for men, suggesting that the ratings were not somehow simply inherently unreliable or otherwise unpredictable.

The message of the GRE study is largely negative: for the most part, a conventional test of intellectual abilities did not provide much prediction of interesting aspects of academic performance. There was some prediction, but it was only of first-year grades, and beyond that, the prediction was for men only, and only for the analytical test. Would we stand to gain anything by a broader theory and broader tests of intelligence?

Another study I was involved with, conducted over a five-year period in collaboration with Michel Ferrari, Pamela Clinkenbeard, and Elena Grigorenko, addressed this question. We sought to learn whether prediction of academic performance at the college level would improve if we used broader tests; more important, though, we asked whether students would perform better if they were taught and their achievement assessed in ways that reflected their patterns of abilities.

We invited high schools from around the country and abroad to nominate students for a summer college-level psychology course to be taught at Yale University. Nominated students were then encouraged to take a test of intelligence based on the triarchic theory of human intelligence. The test measured not only the analytical abilities assessed by conventional IQ tests, but also creative and practical abilities. It measured these abilities in the verbal, quantitative, and figural (geometric) domains, using both multiple-choice and essay test items. For example, in the creative essay test, students had to design and describe an ideal school; in the practical figural multiple-choice test, students had to use maps and diagrams to plan complex routes, as one would do in planning a set of errands in an unfamiliar terrain.

Eventually, 199 students became part of the program in the summer of 1993. The program was simultaneously a course and an experiment. Students were selected to fit into one of five categories: high (both with respect to other students and with respect to themselves) in analytical abilities only, similarly high in creative abilities only, similarly high in practical abilities only, relatively high in all three abilities, or relatively low in all three abilities.

Students who enrolled were then assigned at random to one of four types of course, identified by the type of instruction and skills emphasized: one type of section emphasized memory skills (the control condition), requiring students to recall and recognize; a second type emphasized analytical-thinking skills, requiring students to analyze, judge, compare and contrast, and evaluate; a third type emphasized creative-thinking skills, requiring students to generate, invent, create, imagine, and suppose; and a fourth type emphasized practical thinking skills, requiring students to use, implement, and apply concepts in their everyday lives. These sections, with their varied kinds of instruction, met in the afternoon and represented one component of the course. The basic instructional content, the text and the morning lectures, were the same for all four types of course.

Ideally, a single course would emphasize all four types of skills, so that students could learn both to capitalize on their strengths and to compensate for and correct their weaknesses. But part of our purpose in the study was explicitly to look at the effects of matching versus mismatching of abilities to instruction, and thus the instructional conditions were somewhat "purified" for purposes of the experiment.

All students were evaluated for memory as well as for analytical, creative, and practical achievements. The memory measures were standard multiple-choice, factual recall exams. Analytical measures required tasks such as comparing and contrasting two scientific theories or critiquing an experiment. Creative measures required students to generate their own theories and experiments. Practical measures required students to show how theories and experiments could be applied to specific case studies. What did we find?

First, with regard to our ability tests, the overall correlations between sections (analytical, creative, and practical) were low although statistically significant: 0.23 between the analytical and creative tests, 0.14 between the analytical and practical tests, and 0.15 between the creative and practical tests. Moreover, a factor analysis of the subtests revealed no general factor across all the subtests, but rather specific factors for these various subtests. In other words, the results tended to support the notion that the general factor typically obtained in conventional intelligence tests reflects, in part, the narrow scope of the tests. As the experience of many teachers suggests, students can be strong in some skills but weak in others.

Second, regardless of the measure of achievement (homework assignments, multiple-choice and essay examinations, independent projects) we used, at least two and in one case three of the kinds of abilities (analytical, creative, practical) significantly and substantially contributed to the prediction of course performance. At least one of these abilities was always analytical ability, as perhaps befits the traditional emphasis of our instruction. But the key point was that prediction of academic achievement was improved by adding the other abilities into the prediction equation.

Third, and most important, a number of different data analyses showed that students performed better when the kind of instruction they received was matched rather than mismatched to their pattern of abilities. In other words, students achieved at higher levels when they were taught in a way that recognized and encouraged their particular pattern of skills. If we teach and assess in ways that benefit primarily analytical students, we may indeed end up recognizing only these students as "smart." But if we teach and assess more broadly, we may be surprised to discover that many students are considerably more intelligent than we might have expected.

Casting a broader net becomes particularly important in a society that is as stratified as ours is, and, moreover, one whose stratification in part reflects racial and ethnic differences among its citizens. One of the interesting outcomes of our study was observed before the treatment even began. When we looked at the students who were identified as high-analytic, they looked like a typical group of "good students": mostly white, and from economically and socially privileged backgrounds. But students chosen for high scores in creative and practical skills looked quite different, in that they were much more diverse racially, ethnically, and economically. By creating a rather closed system that first rewards one type of student on ability tests and then rewards that same student in the classroom and on tests of achievement, we may be missing out on what a large segment of our population has to offer us as a society. Our tests may appear to be unbiased, not because they are genuinely unbiased, but because at least some of our criteria share the biases of the predictors. In other words, measures of achievement are biased in the same ways as are measures of abilities.

Suppose it is indeed the case that courses are taught in ways that primarily benefit analytical students, especially those with good memories. What follows is that students who are high in these abilities will look "smart" in the classroom, while those who are equally high in other abilities, such as creative or practical ones, may look quite ordinary, or even undesirable, as in the case of Barbara when she applied to our graduate program. What may result?

The failure to recognize these students' abilities may have quite serious implications for their careers. We have created a system of tests that values certain kinds of abilities, but not others. It is scarcely surprising that our society has formed what Herrnstein and Murray and others refer to as a "cognitive elite': in order to gain access to competitive colleges, as well as to competitive graduate programs, one has to test well. Students with lower test scores are often, and in many institutions, routinely, rejected. They are denied entrance to the access routes that would allow them to become distinguished doctors, lawyers, academics, executives, and so forth. The so-called cognitive elite is no fact of nature: it is something we have created, much as other societies (and our own in the past) have created elites based on the social class of one's birth. Moreover, it is based on a very limited kind of cognitive ability. The divine right of kings was neither divine nor a right: it was a creation, much like the cognitive elite. Were we to admit students to competitive colleges and graduate programs on the basis of their height, eventually we would find that individuals who are in highly regarded occupations are tall. We should never confuse something we have invented with something we have discovered.

But would the students denied the access routes to desirable occupations and economic success actually succeed in the occupations that are made so hard for them to enter? At one level, we cannot answer this question, because we never do find out what might have been. But at another level it seems self-evident that abilities other than analytical ones are key in the occupations that our society values.

Consider, for example, the sciences. To get high grades at the high school and often college level, students need primarily to memorize textbooks and lectures, and, sometimes, to solve problem sets at the backs of chapters and on tests. But how similar are those tasks to the tasks confronted by scientists? Hardly similar at all. As Harriet Zuckerman has pointed out, eminent scientists are those who have good taste in the problems they study - ones who ask good questions. They are the ones who, design broad, elegant, and usually empirically testable theories, or who design powerful experiments to test such theories. Excellent novelists as well as literary scholars also need to be creative, whether in the writing of novels, in the generation of literary theories, or in coming up with novel interpretations of authors' ideas. Indeed, outstanding experts in any field need to combine creative with analytical abilities.

In separate collaborations with Todd Lubart and Janet Davidson, I have done a series of studies on the creative aspects of intelligence. Studying individuals ranging in age from eight to over seventy years, we have found that the creative aspects of intelligence are relatively, although not totally, independent of analytical abilities. Perhaps more important, we have found that creative abilities are relatively domain specific: participants in our studies who were creative in producing artistic work, or in devising advertisements, or in solving scientific insight problems. There was correlation across domains, but it was generally only weak to moderate, about 0.4. Most relevant to schooling, we have found that the students identified as "gifted" are often not particularly high in creative insight skills and that many who are high in these skills are not identified as gifted.

Scholars as well as others also need the practical abilities to be able to communicate their work effectively and to be able to persuade people that their work is worthy of attention. Academics often tend to dismiss practical abilities at the same time that they know that such abilities are key to getting articles accepted by journals, grant proposals funded by government agencies, students to pay attention in classes, and the like.

In a series of collaborations with Richard Wagner, Wendy Williams, Joseph Horvath, and George Forsythe, I have found that practical abilities among adults show virtually no correlation with IQ-like, analytical abilities across a variety of domains. The participants in our studies have ranged widely, from business executives to salespeople to university teachers. Our measurements of practical intelligence bear little resemblance to conventional intelligence tests. They involve participants reading scenarios illustrating the kinds of problems professionals encounter in their everyday professional practice, and then rating various solutions to these problems for their soundness. Ratings are then validated against those of acknowledged experts.

A participant in one of our studies on college professors, for example, might be given a scenario describing all the things a particular professor would like to get done in the next three weeks that the professor doesn't have the time to complete. The participant would then be asked to rate the priority of each of the various things the professor is considering doing.

Although scores on tests of practical intelligence such as this one do not correlate with conventional ability measures, they do predict various criteria of job success - in the case of academics, such things as productivity, citation rates, attendance at professional meetings, and quality ratings of the institution at which one is teaching. This prediction is over and above that obtained from IQ tests.

The prediction is not only for academics. In a study of business executives conducted at the Center for Creative Leadership, Wagner and I found that the best predictor of performance on two managerial simulations was our own test of practical intelligence, followed by a conventional intelligence test, and then various personality measures. In a study of sales people, our test, but not a conventional intelligence test, predicted various measures of sales performance.

Work suggesting a separation of the more "academic" aspects of intelligence, as measured by conventional intelligence tests, and more practical aspects of intelligence is not limited to my own group at Yale. It is widespread and growing. For example, Steve Ceci at Cornell has found that men who consistently picked winners at a racetrack were generally of average IQ and that there was no correlation between their IQ and their ability to pick winners. Jean Lave at Berkeley found that housewives who could easily distinguish which of two products was a better buy had great difficulty on a paper-and-pencil test of mathematical operations. And Terezhina Nunes, now at the Institute of Education in London, found that Brazilian street children who were failing math in school could nevertheless do the mathematics to keep a successful street business thriving.

The message of these and similar studies is not that conventional views of intelligence are wrong but rather that they are highly incomplete. They deal only with a sliver of what revolutionary scientists now believe intelligence to comprise. Thus, when the traditionalists discuss what we know about intelligence, they are really discussing, from the revolutionary point of view, only a narrow part of intelligence. They are answering questions, perhaps, about IQ more than about intelligence, broadly conceived.

Traditional views of intelligence will not disappear overnight. I personally believe that they are much more likely to be subsumed than to disappear, because they constitute a subset of what the revolutionaries are talking about.

There is a revolutionary war going on, as there often tends to be in fields where the excitement is the greatest. The war has not yet been won by either side. Moreover, it is not, strictly speaking, a war with only two sides. The revolutionaries have divisions among themselves. For example, I do not accept all of the assertions of Gardner's theory of multiple intelligences, nor does Gardner accept all of the assertions of my triarchic theory. But we and other revolutionaries are united by our belief that IQ represents too narrow a conception of intelligence. We are also united in the view that intelligence can be modified and have done research supporting this view.

The traditionalists are holding strong in what they see as a solid fort that has withstood vigorous attacks. The revolutionaries, in contrast, see gaping and ever-expanding holes on all sides of this fort. As always, one set of data can lend itself to varied interpretations. Ultimately, perhaps, all parties to the battle will lose, as even newer and more appealing theories and tests of these theories come to the fore. But at the end of the day, that's what intellectual battles are about - the never-ending search for understanding, whether of intelligence or anything else.

I close by noting that, for me, the battle is a personal one. As a child, I did horribly on IQ tests. As a result, my teachers in the early years had low expectations, and I gave them what they wanted. They were happy and so was I. By fourth grade, I had a teacher who expected more, and I gave her more. I went from being a B-C student to being an A student. I attributed what success I came to have to going to bed early. I knew my IQ was low - what else could it be? Now, as a theorist and a researcher in the field of intelligence, I would like to think that there may have been more involved. More of what? Well, that's what I'm still trying to find out.

Source: 21st Century Learning Initiative