So far, I have used the word intelligence without exploring its meaning in any depth. I focused instead on a much more practical question: Does the highly intelligent person have an advantage in achieving success? On this, the data clearly answer, yes. We know, for example, that IQ scores predict success in school. In general, children and adults who have higher IQ scores learn more efficiently and find it easier to master complex material. IQ is also a good predictor of workplace performance, especially in jobs that are complex. In today’s economy, the workplace demands that people are able not only to learn, but also to think productively—to grasp complex problems and generate effective solutions. Now that we have established that intelligence is important, and increasingly so, we need to change directions. We need to confront the pressing question: What exactly is intelligence?
Let’s begin with some basic distinctions. A good starting point is to recognize that IQ and intelligence are not exactly the same thing. IQ is simply the quantified summary value of intelligence. As a two- or three-digit number, IQ is reductive in the extreme. Yet the compression of intelligence into a single number has practical value for predicting future success. We would be mistaken, though, to think that IQ is intelligence, or that an IQ score can fully and accurately summarize the intelligence of any person. Inevitably, the reduction of intelligence to a number skims over important information. To understand why, consider other forms of numeric rating. A baseball player’s batting average reveals something important, but necessarily omits much about that player’s strengths, weaknesses, and style.
IQ and intelligence are not equivalent for other reasons as well. IQ tests are logically limited by our best available conceptualizations of intelligence. Think about it: If our theories of intelligence are imperfect, which they inevitably are, then IQ tests will reflect those imperfections. Intelligence is, in many ways, deeply mysterious. Like all marvelously complex domains of nature—oceans, planets, quarks, and living ecosystems—human intelligence is an impossibly intricate puzzle that constantly awaits discovery and beckons ongoing exploration. Understanding is not achieved once and for all, but gradually and by degrees. Over the last 150 years, we have made a lot of progress in illuminating the nature of intelligence, but much more remains to be discovered. In a sense, then, an IQ score is a caricature of intelligence—a way of pinning a number on a human trait that is vast, complex, and still quite elusive.
One way to gain a foothold in understanding intelligence is to pose a very fundamental question: Is intelligence a unified entity, or is it instead a diverse collection of intellectual capabilities? Simply stated, is intelligence one thing or many? Historically, scholars of human intelligence have been divided on the question.
Around the year 1900, the British psychologist Charles Spearman argued insistently that intelligence is a unified entity. He believed there was a general intelligence factor, which he famously called g. Spearman believed every intellectual act draws upon g, at least to some extent. On the other side of the Atlantic Ocean, the American psychologist L. L. Thurstone took the opposite position. Thurstone believed that intelligence was not a unified ability, rather it consisted of an array of separate abilities, such as memory, mathematical reasoning, and verbal comprehension.1 He believed that each distinct ability worked in combination with the others to produce intelligent thought. If Thurstone was correct, then the term intelligence was a bit of a misnomer, possibly even baseless, because the work of the mind does not draw on a single unifying intelligence but rather operates through several independent forms of intelligence working in concert.
Spearman and Thurstone were not merely speculating from upholstered armchairs when they proposed their competing theories. Both were hard-nosed empiricists who subjected their theories to scientific confirmation based on objective data. To test their theories, they relied on scores from tests they administered to hundreds of people. Those tests were diverse, tapping an array of verbal, mathematical, and spatial skills, as well as various forms of reasoning, such as deduction, creativity, and problem solving. Because these tests measured such a broad span of mental activity, the correlations among the tests could, in principle, reveal patterns that illuminated the nature of intelligence—including, for example, whether it is unitary or multiple. Let’s now explore how this is possible.
As we saw in the previous chapter, a correlation (symbolized r) quantifies the strength of a relationship between any two variables. We might suspect, for example, that in any population there is a positive correlation between shoe size and glove size. A correlation value can test whether or not our suspicions are true as well as quantify the extent of the association. Correlations can be used to quantify the strength of associations between other variables as well, such as height and weight, education and income, or verbal and quantitative test scores. To test their theories, Spearman and Thurstone needed a correlation value for every pair of tests. Modern software makes it very easy to compute a correlation between any pair of test variables. In Spearman’s and Thurstone’s day, though, correlations had to be computed by hand in a very laborious process. The correlations among tests were so vital to theory testing that the resulting data tables were worth the considerable toil.
Now here’s the key: The pattern of correlations can provide evidence that favors one theory or another. One possible pattern is that high correlations appear virtually everywhere in the matrix. This tells us something: When each test is correlated with all other tests, it implies that Spearman’s g is at work. A single, unified intelligence appears to be influencing performance across very different kinds of mental activity. But a second possible pattern is that test correlations are clustered—some tests correlated highly, but others weakly or not at all. That pattern supports the view that intelligence is not unified but is instead a collection of independent mental abilities. This is the pattern that would support Thurstone’s theory of separate and distinct abilities.
Which theory wins out? The answer was unclear for several decades. Spearman’s analysis came first. To test his hypothesis of unified intelligence, Spearman created a diverse array of cognitive tests of language, music, perception, mental speed, and other abilities. When he analyzed the relationships among the tests he found positive correlations spread throughout the matrix. Examinees who did well on one test were likely to do well on others, even if the content of those tests differed markedly. There seemed to be a general mental quality, which Spearman called g, on which performance depended and on which people differed. The pattern of high correlations across diverse tests of mental ability was striking. Spearman gave a name to this pattern—the positive manifold—which was primary evidence that general intelligence, Spearman’s g, was real. Some theorists regard the discovery of the positive manifold as one of the most important findings in the history of psychology.2 It meant something important: Spearman’s g theory appeared to be vindicated.
The matter was not settled, however. Unlike Spearman, Thurstone did not find a uniform blanket of positive correlations across the matrix; instead, high correlations grouped together in clusters. This led Thurstone to the opposite conclusion: g does not exist. He stated this conclusion directly in 1938, declaring, “We have not found the general factor of Spearman.”3 In trying to account for why Thurstone obtained these results, it’s helpful to know that he focused his research on college students. In Thurstone’s day, colleges admitted only the highest achieving high school students. These students were drawn from a narrow range of IQ—that is, uniformly high. The limited variation of IQ results in what statisticians call restriction of range. Reduced variation makes it difficult to detect correlations. For example, you might believe that a family’s annual income is positively associated with its overall health, but to test the hypothesis you would want to study families with a range of incomes, not only the very wealthy. Thurstone may have had difficulty detecting the correlations that compose the positive manifold precisely because the students he studied were too similar to each other intellectually.
Even though Thurstone did not detect a strong and pervasive positive manifold, he did find clusters of correlations, each of which seemed to measure a distinct dimension or factor. Those factors included numerical ability, verbal comprehension, and memory. Other scholars of intelligence later found similar patterns, namely, multiple factors instead of a unitary intelligence. Depending on the particular theory, the number of clusters, or factors, ranged widely. At the high end, the American psychologist J. P. Guilford identified a whopping 150 factors.4 Yet even as new data and research methods became available to test theories of intelligence, the central paradox remained unresolved.
The first paradox of intelligence—one or many?—was surprisingly vexing. From the start, what seemed to be a simple question did not yield a simple answer. As we have seen, Charles Spearman and L. L. Thurstone staked out the original competing theories, with each man staunchly defending his own theoretical position. Later, though, the antagonists experienced a remarkable turn of events: Each man discovered data that supported the theory of his rival.
Thurstone modified his original claim on the basis of a mathematical procedure called factor analysis. Through factor analysis it became possible to extract complex patterns in a correlation matrix that cannot be identified by simple visual inspection. As we have seen, Thurstone used factor analysis to identify six factors. By altering the assumptions of his analysis, Thurstone was able to develop an alternative solution in which the factors themselves were correlated. It therefore became possible to do a second factor analysis—not on the correlations between tests but on the correlations between factors. When Thurstone conducted this “second-order” factor analysis, what emerged on top was a single general factor that looked a lot like Spearman’s g. Writing in 1941, Thurstone bravely admitted, “Our findings seem to support Spearman’s claim for a general intellective factor.”5
Like Thurstone, Charles Spearman was also willing to change his mind if the data so indicated. On the basis of further analysis, Spearman moved away from his insistence that intelligence is purely unitary. The positive manifold of correlations held true, but within that pattern some tests were more highly intercorrelated than others. In other words, Spearman’s g factor applied across all tests—that conclusion remained unchanged—but within his matrices Spearman could also identify narrower abilities that resemble the factors Thurstone described.6 It’s a credit to both Spearman and Thurstone that each could eventually admit that his rival was correct—or at least partly so.
Now it seemed that the question of whether intelligence is “one” or “many” did not have a simple answer after all. Instead, intelligence displays qualities of both unity and diversity. Intelligence is both one and many. When first acknowledged by Spearman and Thurstone, this conclusion seemed deeply paradoxical. It was largely up to other brilliant theorists to put these two realities together. Eventually the paradox was resolved. The unity and diversity of intelligence were reconciled by thinking about intelligence in a new way—not as “one” or “many,” but as a combination of the two. To see how this works, consider Figure 2.1.
The diagram shows Spearman’s model at the top and Thurstone’s at the bottom. When the two models are merged, the combined structure looks something like a pyramid. The pyramid is actually a hierarchy. Hierarchies are ubiquitous in the material world, or in the mind’s way of categorizing the world, depending on your perspective. Think about sports as an example. Is the category of “sports” meaningful? Of course it is. We have sports channels, sports pages, sports magazines, and sports writers. No one is confused by the general category of “sports.” But we also have particular sports—soccer, baseball, basketball, and many others. The two levels of categories fit together easily if we think of the general category, sports, at the top of the pyramid and the particular varieties of sports—soccer, baseball, basketball, and so on—underneath. No doubt you can imagine hierarchical structures for other broad categories, such as animals, cars, machines, art, music, and clothing. Hierarchies are everywhere. The abstract phenomenon of intelligence, like so many other entities in human experience, can also be organized hierarchically. At the top of the pyramid is general intelligence, or Spearman’s g, which extends its sweeping influence over the broad span of mental activity. Lower in the pyramid are broad abilities that resemble Thurstone’s identified factors.
Figure 2.1
Hierarchical Model of Abilities: Combining the Theories of Spearman’s g and Thurstone’s Primary Mental Abilities. Reprinted with permission from Martinez, M. E. (2000). Education as the cultivation of intelligence. Mahwah, NJ: Erlbaum.
The hierarchical model is widely regarded as capturing something essential and enduring about intelligence: It is neither one thing nor many, but both. General intelligence is a legitimate concept, but it does not fully explain the powers of intellect. That’s because intelligence also has multiple aspects or, more technically, factors, each of which expresses a particular dimension of the larger whole. To tell the whole “story” of intelligence, we needed the hierarchical model to combine its properties of unity and diversity. Reflecting on this remarkable synthesis, psychologist Raymond Cattell proclaimed that through the hierarchical structure, the ideas of Spearman and Thurstone were “reconcilable, and with mutual illumination.”7 The hierarchical model was a huge step forward. Its structure allowed for a more complex concept of intelligence, and one truer to its nature. Though established as early as 1950, the hierarchical structure of intelligence has proven to be remarkably durable. Near the end of the 20th century, prominent intelligence theorists affirmed that “the empirical evidence in favor of a hierarchical model is overwhelming.”8
Let’s be careful, though: Different theorists have proposed somewhat different hierarchical models. Which one is best? There is a very good candidate for the single best hierarchical model—the three-stratum model advanced by the psychometrician John Carroll.9 John Carroll’s model is compelling because it was built from a mountain of data. Instead of relying on a single data set, Carroll analyzed 460 data sets, many having historical importance reaching back decades. Carroll’s three-stratum model deserves our attention. Let’s have a look (Figure 2.2).
The three-stratum model has three layers, as the name suggests. At the top is general intelligence, which exerts influence over all the abilities lower in the hierarchy. In the middle layer are broad factors that bear some resemblance to Thurstone’s factors. At the bottom of the pyramid are narrow factors—specialized mental abilities that are useful for particular kinds of mental performance. Procedurally, Carroll’s data warranted factor extraction three times, while Thurstone required only two factor extractions to obtain a satisfactory solution. This procedural difference is what led to Carroll’s model having three levels in contrast to Thurstone’s model having only two.
Figure 2.2
Carroll’s Three-Stratum Model. Adapted with permission from Carroll, J. B. (1993). Human cognitive abilities: A survey of factor-analytic studies. New York: Cambridge University Press.
Note one peculiarity of the three-stratum model: The pyramid is not symmetrical, but is skewed to the left. That’s because connections between general intelligence at the top and the broad factors one level below are uneven. Shorter lines indicate stronger connections with general intelligence; longer lines mean that those connections are weaker. Clearly, two broad factors have especially strong connections with g—fluid and crystallized intelligence. These two factors further illuminate the nature of intelligence. We have already seen that intelligence is structured hierarchically, like a pyramid. Now we gain a second key insight, which is that two manifestations of intelligence—fluid and crystallized—show how a capable mind interacts with the world. Let’s consider why these two aspects of intelligence are so important.
In everyday speech, the word intelligence has two distinct meanings. The most familiar meaning is the ability to think and reason. An intelligent person can manage complex information quickly and accurately, as well as generate interesting ideas, effective strategies, and warranted conclusions. A second meaning of the word intelligence is a body of knowledge. We see the word intelligence used this way when describing the missions of the Central Intelligence Agency and the British Security Service, MI-5. To a significant degree, the distinction between fluid and crystallized intelligence parallels this difference in meaning.
Fluid intelligence refers to the mind’s ability to adapt to novel, complex, and challenging environments. Of these three descriptors, the word novel is probably the most important. Fluid intelligence is applied whenever a person must adapt to a new situation, such as an unfamiliar culture, a new job, or a perplexing problem. A person who excels in fluid intelligence has the mental resources to tackle a completely novel challenge and find a way to succeed. Crystallized intelligence, a counterpart form of intelligence, is manifest in the ability to master large bodies of information. Crystallized intelligence is also strongly associated with verbal ability, hinting that knowledge is most often learned, comprehended, and remembered in verbal form.
Fluid and crystallized intelligence complement each other beautifully. One way to think about their roles is that each represents a vital resource for the intelligent mind. Crystallized intelligence offers the invaluable resource of current knowledge—what is already known. This is more than the skillful use of memory so heavily relied on by school systems. The ability to acquire knowledge in ways that can be applied intelligently is much more complex and far-reaching. Crystallized intelligence represents the developed ability to master organized bodies of knowledge that have functional value in a complex, information-rich world. But if current knowledge is not enough, which is often the case, the gap is filled by fluid intelligence—the capacity to deal with the unknown. We draw upon both resources as we go about our lives. To think and act intelligently, knowledge is valuable and desirable, and more knowledge is almost always better. Inevitably, though, we must supplement our knowledge with the capacity to adapt flexibly to unfamiliar environments and the novel problems they present.
To some extent, the distinction between fluid and crystallized intelligence can help us understand the condition known as savant syndrome. Savants are people who are identified by low intellectual functioning except for a few particular, and often peculiar, skills in which they excel. In the movie Rain Man, the central character, Raymond Babbitt, played by Dustin Hoffman, could memorize a phone book and perform calculations at lightning speed. Babbitt was modeled after a real-life man named Kim Peek, who can perform numerical calculations with extreme speed; he is also a prodigious reader who is said to have memorized more than 7600 books.10 He was born with brain abnormalities, including the absence of the corpus callosum, the nerve structure that connects the brain’s left and right hemispheres. The absence of a corpus callosum may have caused Peek’s brain to adapt by producing other structures that underpin his extraordinary skills.
Other savants have skills that are different, but no less striking. Some can recall sports statistics with uncanny accuracy or estimate distance precisely by sight alone. One savant named Ellen has such a highly developed sense of time that, without the aid of a clock, she can note the precise time of day to the second. These “splinter skills” are impressive, but they lack the depth and flexibility that make them broadly useful to the savant. For example, one musical savant, NP, was able to reproduce Grieg’s “Melodie” on the piano after hearing it played only once. But when exposed to an unusual atonal composition, Bartok’s “Whole Tone Scale,” his attempts to reproduce the piece were slow and awkward. NP made errors that imposed more traditional melodic structure over the composition. Without question, NP’s ability to reproduce conventional tonal music was amazing, but its operation was confined to a specific framework of performance.11
Kim Peek and other savants are not known for their academic success, nor for their personal effectiveness in the world. About half of all savants are autistic, and so face severe challenges in adapting to the demands of everyday life. Although Kim Peek has been employed successfully as a payroll bookkeeper, the fictional character Raymond Babbitt was institutionalized because of cognitive and behavioral limitations, despite his extraordinary abilities to calculate and memorize. The drama of Rain Man is closer to the lived reality of many savants. These facts lead us to an important question: How do the focused talents of savants, however extraordinary, differ from what we are calling intelligence? First, it seems clear that the ability to memorize quickly and efficiently does not adequately characterize crystallized intelligence. A person who is high in crystallized intelligence typically has no difficulty in academic settings and has a large fund of knowledge at his or her disposal. But that knowledge is not fixed like the entries in a telephone book. Rather, it is flexible and interconnected; ideas are organically related such that one can trigger another. Crystallized intelligence does not consist in sheer volume of knowledge, but rather knowledge that can be drawn upon to support intelligent behavior.
One way to understand how crystallized and fluid intelligence fit together is to map them onto “potential” and “achievement.” In the investment theory of intelligence, a person high in raw intellectual ability (fluid intelligence) can “invest” that resource to develop an educated mind (crystallized intelligence).12 The investment theory corresponds to our everyday intuitions about a child’s capacity to learn. A child who has high ability, through opportunity and self-discipline, can capitalize on that resource to gain knowledge and academic success over a span of many years. When this happens, fluid intelligence is invested to yield an accruing dividend of crystallized intelligence. Alternatively, a child’s fluid intelligence might be squandered. For any number of reasons—poor motivation, lack of discipline, inadequate resources, or simple lack of opportunity—the child who holds the resources necessary for tremendous personal achievement and contributions to the world does not develop that gift. Like a blue-ribbon seed that never germinates, uninvested fluid intelligence represents lost potential.
Among some psychologists, however, the concept of “potential” is not very popular. The reason is that “potential” hints of an imaginary entity that might or might not exist in the future. Critics argue that we can measure only what exists in the present. Yet the concept of a child’s potential for future success is a very intuitive idea to most teachers, coaches, and parents. We recognize that some children display uncommon aptitudes for music, swimming, chess, mathematics, science, or art. The standout aptitude can be manifest as a keen and inquisitive mind, one that is distinguished from the crowd. Indeed, the entire field of gifted and talented education is premised on an assumption that some children are marked with an unusual aptitude—so much so that the standard curriculum is an impediment to the child’s development.
Many intelligence theorists, recognizing the fluid-crystallized distinction, have designed tests to report separate scores for fluid and crystallized intelligence. Other tests, such as the Wechsler, report a similar distinction between verbal and performance (nonverbal) IQ scales, which, over time, have been interpreted as measures of crystallized and fluid intelligence. These shifts in labeling and interpretation show that the fluid-crystallized distinction has had a major influence on intelligence testing. The influence has a theoretical side as well: By studying how people solve problems, especially on tests of fluid intelligence, our understanding of intelligence has become richer and more complete.
Fluid intelligence can be assessed in a remarkable variety of ways. Some fluid intelligence tests use abstract shapes, while others use words or numbers. All measures, though, have this in common: Tests of fluid intelligence require a person to perceive a complex pattern and to apply that pattern to solve a problem. Another key feature is novelty. It’s important that the problem or puzzle be new to the person’s experience. The novelty requirement is met by a classic test of fluid intelligence, Raven’s Matrices, originally developed by John Raven, a student of Charles Spearman. The task involves a 3 × 3 array of eight abstract shapes with a blank spot in the lower right corner. Figure 2.3 is an example of a Raven’s-like problem.
Figure 2.3
Raven’s Matrices Problems. Martinez, Michael E., Learning and Cognition: The Design of the Mind, ©2010. Printed and Electronically reproduced by permission of Pearson Education, Inc., Upper Saddle River, New Jersey.
To solve this problem, you must first examine the various elements—their shapes, orientations, number, and shading—and discover correspondences from row to row and column to column. Those rules of order dictate which pattern should fill the blank cell. A Raven’s Matrices test is organized so that it presents relatively easy problems at the beginning and steadily progresses to more difficult problems. The most difficult problems involve multiple simultaneous transformations from top to bottom and from right to left. To solve a complex Raven’s problem, a person must perceive each dimension of change in the abstract shapes, combine those dimensions mentally, and then select the ideal pattern from the options available. If any detail is overlooked, the chances of selecting the right answer are thereby reduced.
Let’s probe more deeply into the kinds of thinking evoked by Raven’s Matrices. The puzzles are unusual in several ways. Most important, Raven’s never explains the rules that guide how the puzzle is structured. The point is precisely that the rules must be figured out based on each incomplete matrix. Those rules change from one matrix to the next. The examinee must study the specific elements of each matrix and then infer the rules that explain the pattern. In the more difficult Raven’s problems, this is quite challenging: It requires careful and systematic comparison to be sure that the inferred rules are both precise and complete, enabling the prediction of the missing piece. This sequence of thought is a perfect example of inductive reasoning.
Inductive reasoning always proceeds this way: from the specific instance to a general pattern. Whenever we encounter a new situation, whether an abstract puzzle or an everyday problem, we must first understand it. That means seeing how the various parts fit together. Understanding the problem helps us to work effectively toward a solution. Inductive reasoning serves another extremely important function—it helps us to apply our understanding to a new second problem, then to a third problem, and so on indefinitely—as long as those problems are structured similarly. This portability is what allows a doctor to diagnose and treat a range of problems or a CEO to transfer effective leadership in one company to a different one.
Inductive reasoning is tremendously important. Stop to consider that scientific discovery draws heavily on inductive reasoning. Scientists work from specific data gathered in their laboratories to induce a rule, principle, or law whose validity extends beyond the specific context of discovery. It must generalize. The discovery qualifies as science precisely if it applies broadly to other laboratories and data sets, as well as across time. Inductive reasoning—finding the general pattern in specific data—is vital to the scientific enterprise. More generally, inductive reasoning is also an indispensable element of intelligent thought. It’s so important, in fact, that some theorists believe inductive reasoning lies at the very heart of fluid intelligence. The most daring expression of this hypothesis was advanced by the psychometrician Jan-Eric Gustafsson, who proposed that inductive reasoning is nearly identical to fluid intelligence. He went even further to argue that fluid intelligence is, in turn, nearly identical to general intelligence.13 Expressed compactly and without qualifiers, Gustafsson’s bold hypothesis is as follows:
Inductive reasoning = Fluid intelligence = General intelligence
Even if the Gustafsson hypothesis turns out to be incorrect, its mere plausibility is instructive. Whatever the panoply of cognitive skills that compose the repertoire we call general intelligence, the place of inductive reasoning ability within that repertoire is somewhere near the very core.
A second cognitive skill used to solve Raven’s Matrices is the ability to break down the larger problem into smaller problems. Rather than trying to understand the matrix in its entirety, a more flexible approach is to adopt a simpler subgoal.14 Research shows that the most successful examinees can focus on a single aspect of the problem and progressively use the information learned from the subproblem to generate a solution that works for the problem as a whole. This very ability—to divide a cumbersome task into smaller and more manageable subproblems—is perhaps the most powerful problem-solving strategy of all. In the case of Raven’s Matrices, people vary in their ability or tendency to form subgoals. Those who break down complex problems into more manageable pieces tend to have higher scores overall.
Performance on Raven’s Matrices draws on yet a third cognitive skill, the ability to hold several goals in mind at once and to track progress toward those goals.15 This is a bit like what we ordinarily call “multitasking,” but applied to a single complex problem rather than to different activities that overlap in time. Careful tracking is most important in the context of information overload, and when some of that information is relevant and some merely distracting. The ability to separate relevant from irrelevant information, and to hold relevant information in mind while solving a problem, is sometimes described as executive functioning. Like a capable business executive, the mind must sift pertinent data from distractions, formulate goals and strategies, and then initiate a coordinated plan of action. Executive functioning also includes the ability to monitor progress toward goals, and to switch or modify strategies if necessary. To ensure success, no piece of essential information can be allowed to slip from active consideration. This challenging set of requirements is clearly higher-order in nature. It also ties in quite directly to the known functions of the brain’s frontal lobe, the seat of higher-order thinking.
Let’s take stock, then, of what we know about Raven’s Matrices. First, we recognize Raven’s as an important tool in the measurement of fluid intelligence. For reasons that are still somewhat mysterious, Raven’s Matrices affords us a glimpse into the heart of intelligence. Because Raven’s works so well, psychologists have studied what people who obtain high scores actually do. Psychologists want to know: What patterns of thinking separate high performers from low performers? Any insight we gain can help us to understand how the cognitive functions that underlie performance on Raven’s Matrices might apply to non–test situations. Ultimately, the practical manifestation of intelligence is what we want to understand—the substance of real-life success, and more exactly the intellectual skill set that makes success possible.
From studying performance on Raven’s, we have learned that people differ in their ability to perform three kinds of intellectual activities: (1) finding order in complex patterns, (2) breaking problems down into subproblems, and (3) managing several goals at once without being overwhelmed. When we think of these three expressions of intellect, we can easily see how they can apply broadly to the sorts of problems we face every day. First, finding order in complex patterns is vital to navigating our data-packed world. We must learn to take in the complexity that surrounds us, understand it, and then apply that understanding to make smart choices. Often those decisions are not simple. Almost always, they entail choices about how to subdivide unwieldy problems into manageable subproblems. This ability to decompose a large problem in order to make it manageable is the second skill that distinguishes high performers on Raven’s. Finally, as we pursue our goals, we must do so without being overwhelmed. Our executive functions help us to keep track of our goals and pursue them in an order that makes sense, without letting any important piece of information slip from consideration.
Crystallized intelligence is the intellectual resource that corresponds to structured knowledge, along with the learning skills that facilitate building that knowledge base over time. Knowledge is fantastically important in supporting intelligent thought and behavior. Personal experience tells us that knowledge can take many different forms. Some of what we know is like pictures in a photo album: In our mind’s eye we can see familiar people, places, and objects. Some knowledge is maplike, which allows us to navigate familiar places, such as our hometown, easily and efficiently. Other forms of knowledge are mathematical in nature. We interpret our world with a sense of numerosity and make judgments about quantity in coarse terms—one or many, short or long, small or big—while knowing that quantity can also be measured precisely using the number system. The mind knows what it knows in many other ways as well. We recognize the smell of a pine forest, the taste of cheese pizza, and the texture of silk. We also know how to do things: walk, run, skip, shake hands, play a musical instrument, and sign our own name.
Among this vast array, one form of knowledge rises above all others in importance to the work of the mind—language. Although language has no strict monopoly on the way the mind stores knowledge, words and word meanings are so central to human cognition that some theorists treat language and thought almost as if they were two sides of the same coin. It should be no surprise, then, that what we call verbal ability has a special connection to intelligence, and in particular to crystallized intelligence. This does not mean that verbal ability and crystallized intelligence are identical. They are distinct constructs, but with significant overlap. We see a similar relationship between fluid intelligence and inductive reasoning. Fluid intelligence draws on the ability to see the order in complex problems, and to apply those inferred patterns to solve the problems.
Crystallized intelligence draws heavily on the ability to structure knowledge in verbal form, and to use language to make sense of the world. Language is important because words represent ideas and concepts that are the substance of thought. This means that language builds our understanding of the world not directly, but through the concepts that words represent. Given the connection between crystallized intelligence and language, it makes sense that the size of a person’s vocabulary is a good predictor of IQ. In exploring why this is so, let’s recognize two facts. First, children learn new words at a fantastically rapid rate, on average about 10 words each day.16 Second, children vary significantly around that average—some children learn new words at a faster rate, others slower. Over time, the different rates of learning new words produce tremendous variation in the size of children’s vocabulary. Adults, too, vary in the size of their vocabularies. In both children and adults, that variation correlates significantly with measured intelligence.
In the armamentarium of the human intellect, vocabulary seems to be very important—surprisingly so. This connection implies that we ought to build our functional vocabularies as much as possible. In the normal course of development, we learn new words in a variety of ways: by reading definitions in a dictionary, through instruction in school, or in the case of young children, by being taught by parents and other adults. These modes of word learning are direct and purposeful, but vocabulary growth cannot be accounted for solely by intentional teaching and learning. Word knowledge is not normally gained primarily through rote learning and memorization, but indirectly through encountering new words in the daily flow of conversation, during instruction by teachers, or through books or other media. Unfamiliar words are not usually defined outright. Instead, their meanings have to be inferred from context.
Let’s also be clear about what it means to learn a new word. It’s nothing like tossing a penny into a jar of coins. The mind’s growing knowledge base is not a particulate array of isolated bits of knowledge. It’s much more organic. Long-term memory is a massive system of interconnected concepts. We use that conceptual grid every day to interpret the elaborate world of people, things, and ideas, and to act intelligently within it. Vocabulary acquisition therefore entails much more than learning the definition of a new word. It must connect to our existing knowledge base, including other word meanings, and so becomes an element in a vast system to understand the world as well as a bridge to further learning. Language and language-based knowledge are not to be underestimated in their intellective potency. That’s why knowledge in verbal form is rightly considered a powerful form of intelligence—crystallized intelligence in particular.
Together, fluid and crystallized intelligence open a window into the deep structure of intelligence. We can dismiss the simplistic notion that there exists only one form of intelligence, pure and simple. The more complex and accurate picture is that human intelligence has different manifestations, two of which are fluid and crystallized. Both are vitally important, but are they equally so? Or is one more important than the other to living intelligently in the 21st century? The answer is not obvious. Without question, crystallized intelligence is crucial. After all, structured knowledge in specialized fields is indispensable to every complex society. Think about the technical knowledge needed to design a smart phone, perform heart surgery, or draft legal documents to support the merger of two companies. Specialized knowledge matters greatly, yet a compelling argument can be made that fluid intelligence is paramount. In today’s world, and pointedly in business, the best ideas—those that are most profitable and transformative—are breakthrough concepts and designs. The flashes of insight that change the way we think, act, and live are manifestations of fluid intelligence.
Crystallized and fluid intelligence are both critical. But provisions for their development in society, especially by our educational systems, are a bit lopsided. To make the point, consider which receives greater emphasis in schools. It’s quite obvious that crystallized intelligence is the intelligence factor that has received the higher billing. In a way, this is understandable: It’s easier to teach and test for highly structured crystallized knowledge than the capacity for adaptive thinking that constitutes fluid intelligence.17 Yet to admit that the preponderance of educational effort is channeled to the development of crystallized intelligence is not to say that a skewed curriculum is best or should be maintained. Quite the opposite: If we believe that fluid intelligence is at least equal to crystallized in promoting the survival and success of future generations, the imbalance must be redressed. This won’t be easy. Centuries of structuring curricula around traditional subject areas have established teaching practices that have formidable inertia. To build the vital intellectual capital that consists jointly of crystallized and fluid intelligence requires changes in the ways schools operate. Even more fundamentally, assumptions about the purpose of education must be challenged.
To be effective, every institution—large and small, public and private—needs to cultivate a strong base of fluid and crystallized intelligence among its members. With plentiful supply of both forms of intelligence, chances are good that an organization’s human capital will have both the expertise and the creative “mojo” to propel innovation in the years ahead. But how can this be accomplished? With intellectual firepower so important to modern society, and especially to a society’s capacity to innovate, what can be done to equip a nation, company, or educational institution with sufficient reserves of fluid and crystallized intelligence?
To understand how to increase fluid and crystallized intelligence, we must first consider that they are not static over the course of life, but shift in predictable ways.18 Figure 2.4 depicts the typical pattern of change through the life span.
Figure 2.4
Fluid and Crystallized Intelligence Over Life Span. Martinez, Michael E., Learning and Cognition: The Design of the Mind, ©2010. Printed and Electronically reproduced by permission of Pearson Education, Inc., Upper Saddle River, New Jersey.
To interpret the graph, let’s begin with fluid intelligence. The pattern of change over time can be stated succinctly: Starting in childhood, our capacity to adapt to challenging environments rises quickly. Through adolescence and into early adulthood, we make rapid gains in flexible thinking as applied to new problems. Unfortunately, that marvelous upward trend does not continue forever. Fluid intelligence tends to peak between the ages of 22 and 30—around the time that many young adults launch their careers or begin their most demanding periods of study in graduate or professional school.19 It’s also the age range of those who make major breakthroughs in certain technical fields, especially mathematics and physics.
For most people, learning that fluid intelligence peaks in early adulthood is no surprise. Conventional wisdom tells us that younger adults are more adaptable in thought and behavior than older adults. Other factors being equal, a young adult will display a greater degree of cognitive flexibility—the trait that we identify with fluid intelligence. It’s easy to think of unflattering attributions that are commonly applied to adults in middle or advanced years: We can become “set in our ways” and find it uncommonly hard to “learn new tricks.” Those clichés may contain a grain of truth.
It’s not all bad, though. The picture is more assuring when we consider the counterpart trend: Over the life span, we become more knowledgeable. Crystallized intelligence typically rises until the age of 50 or so, and then is maintained indefinitely. Beyond the age of 70, crystallized intelligence may decline, but seems to do so only in response to poor health. If good health continues, crystallized intelligence might continue to rise. In this expression of intelligence, age—or rather, time—seems to be an ally. Indeed, expert status in any technical field is most often accorded to experienced adults. Years of specialized work can build technical knowledge to impressive proportions. The resulting manifestation of crystallized intelligence as expertise is a tremendous asset.
Now let’s return to the trend for fluid intelligence. This is the trend that is cause for the most concern. It means that a person’s ability to respond to a complex environment rises into young adulthood, but after that, through middle age and beyond, the capacity to adapt begins a long, slow decline. No one really knows why fluid intelligence tapers off as we age. Pessimistically, it may be an inevitable consequence of brain deterioration. A more optimistic explanation is that the decline is a consequence of lifestyle decisions. The first two or three decades of life are marked by rapidly shifting demands in both intellectual and social realms, and these changes demand that the young person adapt, often in very significant ways.20 Starting in middle age, however, adults typically choose a more stable life path, one that requires less flexibility. With greater life stability we may come to rely more on routines and less on strategy shifts. The decline in fluid intelligence that accompanies aging may be a direct consequence of changes in intellectual habits.21
In theory, declines in intellectual flexibility could be offset by a conscious decision to resist a predictable lifestyle. This might mean taking up new areas of study, pursuing new career paths, learning a new language, or exploring the world through travel. The press to adapt, ever-present in youth and early adulthood, can be maintained through middle age and beyond with the hypothetical consequence of preserving fluid intelligence. Indeed, many older adults intuitively comprehend the connection between self-challenge and mental agility. They deliberately organize their lives to change up their daily experience, in effect forcing themselves not to rely on established routines. The hope is that this self-directed training will preserve cognitive flexibility and adaptability—the hallmarks of fluid intelligence. This hope is a reasonable expectation: Research supports the belief that engagement in an active lifestyle is associated with preservation, or even improvement, of cognitive abilities in the advanced years.22
Perhaps in the future we will come to know the full explanation for declines in fluid intelligence with age. For now, the important fact remains that older workers tend to lack the cognitive flexibility of younger workers. By itself, this fact may be more interesting than concerning, but if a nation or company happens to have an aging workforce, then serious implications might follow. The combination of these two trends—the rising need for fluid intelligence in the 21st century and the aging workforce characteristic of many countries around the world—spells bad news. Psychologist Earl Hunt faced the dilemma squarely in his book Will We Be Smart Enough?23 He pressed the question of whether aging populations pose a risk for economic viability and competitiveness. These trends might be especially worrisome for nations that are experiencing declining birthrates, such as Germany and Japan.
It’s an issue for corporate workforces, too. As personal investment portfolio values declined during the recession, many workers were forced to postpone retirement. Established companies, saddled with a heavy burden of retirees’ pensions, found it difficult to compete with younger companies that were less fiscally, and perhaps intellectually, encumbered. Around the world, governments must now consider whether to shift the normative retirement age upward. As the average age of workers drifts higher, organizations gain greater reserves of crystallized intelligence that derive from long years of experience. But the crucial resource of fluid intelligence—the engine of adaptability and innovation—declines on average. This down-tick may well put a company or nation at a competitive disadvantage.
There is no question: Workers in every setting—businesses, universities, nonprofits, hospitals, and government agencies—benefit from the potent resource of fluid intelligence. That’s why institutional leaders should consider how they can structure the experience of their workers to maintain, and possibly improve, their reserves of fluid intelligence—intelligence that helps secure a prosperous and profitable future. Yet it’s a mistake to downplay crystallized intelligence in favor of fluid intelligence. Both are necessary. The combination of fluid and crystallized intelligence is what makes people—students, workers, and citizens—most effective. Fluid intelligence yields the marvelous adaptability and innovative thinking that are so highly prized and that transform the way we live. Crystallized intelligence is hugely important, too. At higher levels, it manifests as the expertise that is foundational to nearly every great achievement.
Why choose between fluid and crystallized intelligence? Is it not ideal to have generous reserves of both? And rather than worrying excessively about birthrates and average age of a workforce, wouldn’t it be better to focus on raising the intelligence of everyone? If such a thing could be done, then the trend lines over the life span would retain interesting and important patterns, but would no longer suggest scenarios of doom. To enhance intelligence directly and intentionally is to gain tremendous freedom. This can become our vision for a bright and exciting future.
Psychologists tend to see intelligence as a manifestation of higher-level thinking. This was Binet’s perspective: He understood intelligence as a person’s ability to perceive order in a complex world, to understand that order, and to think and act adaptively within it. A person’s ability to solve problems, to think critically, and to understand complex subject matter are all signs of a highly effective intellect. But is there more to intelligence? Should human intelligence be understood only in terms of higher-order thought?
Not necessarily. Remember that even before Alfred Binet developed his first tests of intelligence, Francis Galton collected information about people’s basic perceptual acuity and reaction time. Galton believed that these lower-order biological factors underlay the manifest differences in people’s success in Victorian England. Early in this chapter, we considered a fundamental paradox in understanding human intelligence: Is it one thing or many? Now we face a second paradox. If we place Binet’s and Galton’s theories side by side, we confront that paradox squarely: Is the essence of intelligence complex reasoning, or is it rooted firmly in differences at the level of perception and reaction time, which in turn reflect differences at the neuronal level?
This second dilemma has parallels in computer functioning. Imagine a very powerful computer. It can sift through mountains of data to find patterns, perform calculations, and formulate projections—all at lightning speed. What seems obvious is that the computer’s ability to deal effectively with massive data inputs is a function of its sophisticated programming. Yet beneath this manifest performance is the computer’s basic design. This architecture sets the computer’s processing speed and its data storage capacity, which in turn are functions of its hardware and software parameters. To some extent, then, a computer’s ability to perform rapid, complex, and perhaps even intelligent work is a function of its fundamental architecture.
Now the question is: Can we detect an equivalent to a computer’s architecture in the expression of human intelligence? A fascinating body of research suggests important parallels between the two. Here, I refer to research on elementary cognitive processes. One of the best-known examples is an extremely basic visual task known as inspection time. In this task, a computer screen shows two parallel lines of slightly different lengths. The response is simply to indicate which one, right or left, is longer (Figure 2.5).
Figure 2.5
Inspection Time Comparison: Which Line Is Longer, Right or Left?
Respondents almost always select the correct line, and so the point of the task is not to determine whether the response is right or wrong. Instead, the parameter of interest is reaction time. How long an exposure is needed for a person to perceive a difference in the lengths of the two lines? The presentation of lines on a computer screen can be shortened to tiny fractions of a second—no more than a brief flash. If the exposure is gradually reduced, eventually it becomes so short that no viewer would be able to detect a difference. That minimal duration varies from person to person. The briefest exposure required to answer correctly is the person’s inspection time. Inspection times have surprisingly high correlations with IQ, about r = −.50.24 Here, the negative sign indicates that a high IQ corresponds to a low response time. People with high IQs tend to make perceptual judgments very quickly; they need only the briefest periods of exposure to detect a difference in length between the two lines. Fast inspection times correlate especially well with high scores on tests of fluid intelligence.25
The most common interpretation of inspection time data is that it reveals the biological limits of information processing. For reasons that may trace to the level of neurons, people vary in the time needed to distinguish between the lengths of two lines. It’s not completely clear, however, that personal limits in information processing are actually what is being measured. If the task taps some sort of fundamental limit, then we would expect inspection time to be highly stable. This turns out not to be the case. A person’s inspection time can vary from minute to minute, and performance can improve significantly with practice.26 The stable patterns that form when data are averaged are much more erratic than those averages themselves suggest. This means that the proper interpretation of inspection time research is not totally clear. Yet even if we don’t know why, we must acknowledge that what seems to be a ridiculously simple cognitive performance—judging which of two lines is longer—somehow predicts a span of cognitive capabilities.
Reaction times from other elementary cognitive tasks have likewise been correlated with IQ scores. In one experiment, participants use a console of lights and buttons. At the top is a row of five lights and five buttons, one button per light. Those lights flash in a random order—such as 5, 2, 3, 4. At the bottom of the console is a single “home” button, which the participant keeps pressed until the lights stop flashing. The correct response is to move a finger from the home button to press the buttons below each light in the same sequence—5, 2, 3, 4. Though not quite as simple as the inspection time task, this task is obviously not difficult. Participants make few errors, but average reaction times do vary from person to person. On this task, also, reaction times correlate surprisingly well with IQ.27
What do these correlations mean? They are based on tasks that seem incredibly basic. Their correlations with IQ appear to challenge the belief that intelligence is higher-order in nature, or at least challenge the assumption that intelligence is only higher-order. Elementary information processing in some way composes part of the complete picture of human intelligence.28 Certainly human intelligence includes higher-order thinking, but maybe intelligence is more truly a reflection of the nervous system—maybe it is more biological than cognitive in essence. That, at least, seems to be the standard interpretation of the data relating IQ to performance on simple tasks.29
These streams of research seem to link intelligence to basic brain processes. They imply that intelligence relies on the brain’s ability to analyze basic sensory stimuli and to respond efficiently. Apparently, differences in basic neural architecture give rise to differences in measured IQ. However, at least two cautions are in order. First, any inference that biological factors are fully dictated by DNA is too simple. After all, genes respond to environmental variation. One person may be genetically predisposed to be tall and another to be short, but their eventual heights will be affected by nutrition. Likewise, a man who is genetically predisposed to heart disease can establish a regimen of exercise and a low cholesterol diet that counteracts the risks set by his genes. Only quite specific traits, such as eye color, are strictly controlled by genotype. So while acknowledging the shaping forces of genetics on individual differences—both physiological and intellectual—we must understand that genes rarely dictate a trait. DNA is not destiny.
A second caution applies directly to the interpretation of data on elementary cognitive processes. This caution is somewhat technical, having to do with the difference between means and standard deviations. Recall that a statistical mean is simply the average, a computation any of us can do on a calculator. The standard deviation tells us something different: It measures the spread of the data points around the average. Now let’s apply this distinction—means and standard deviations—to the data on elementary cognitive tasks. It’s true that people with high IQs tend to have faster reaction times compared to those with lower IQs, and this is interesting. But equally interesting is that the two groups also differ in the standard deviations. Here the focus is on personal variation—how the reaction times of a single person vary from one trial to the next. This is the most interesting part: IQ is correlated with mean reaction times, but IQ is correlated much more strongly with standard deviations.30 Somehow, the standard deviations of reaction times are saying something important, but what?
For the explanation, consider the basic pattern: High-IQ participants tend to be consistent from trial to trial. Because their reaction times don’t vary much, the standard deviations around their mean times are small. Lower-IQ participants show the opposite pattern. On some trials they respond quickly, but on others they are slow. Their data points scatter widely, resulting in a larger personal standard deviation. This inconsistency may explain why their average reaction times are slow. Apparently, at play in the uneven performance are lapses in attention. Such lapses directly increase the standard deviations, but they also affect the means: Slower reaction times drag down the average.
This explanation implies that at least some of the gap between high- and low-IQ people is the inconsistency of participants with lower IQs. In fact, it may explain the entire correlation between reaction time in IQ. It’s not that high-IQ people are more capable of responding faster; rather, their more consistent responses mean that they have fewer slow responses. A similar pattern has been detected in studies of brain waves: The wave forms of high-IQ people appear to be much more consistent than those of lower-IQ subjects.31 What explains these differences in consistency? The answer seems to involve higher-level functions of attentional control—functions that are regulated by the brain’s frontal cortex, the seat of higher-level cognition.
If this line of reasoning holds, then our initial interpretation of reaction time data is now completely inverted. Reaction time, which on its face appears to be a lower-level cognitive function, involves the higher-level skills of monitoring and control. The same interpretation—that apparently simple processes might actually rely on higher-level control—may apply to other elementary tasks as well. This brings us back to the question of whether intelligence, at root, is an expression of basic biology or higher-order cognition. We have plenty of evidence to show that intelligence reflects higher-order functioning. As we have seen, one excellent test of fluid intelligence, Raven’s Matrices, draws upon a person’s ability to induce rules from a complex grid of patterns. Inductive reasoning, problem solving, and the management of complex goals are all known to underlie performance on IQ tests.32 We know too much about the nature of human intelligence to believe that higher-order skills don’t matter. They do. Yet even very simple perceptual tasks correlate significantly with IQ scores, so basic processes are also important. The intelligent mind relies on both lower-level and higher-level functions working in tandem.
This chapter explored the nature of intelligence by confronting two paradoxes. Each paradox concerned two dualities—two seemingly opposite qualities. Phrased as a question, the first paradox is simply this: Is intelligence one entity, or is it many? If the answer is many, then the word intelligence may be a misnomer. Although we use the singular word intelligence in everyday speech, perhaps there are only “intelligences,” distinct mental capabilities that coordinate to produce intelligent behavior. In a way, the question of whether intelligence is a unified whole or a diverse collection of mental abilities is the most fundamental theoretical question imaginable. To make any real progress in understanding what intelligence is, the early intelligence theorists needed an answer.
At first, the paradox of unity or diversity was only deepened by data that seemed to support both positions. The research of the British psychologist Charles Spearman confirmed the mathematical reality of general intelligence, which Spearman symbolized as g. The American researcher L. L. Thurstone arrived at the opposite conclusion: His data indicated a handful of distinct mental abilities rather than a unified general factor. Eventually the paradox was resolved—not by the vindication of either Spearman or Thurstone, but by each acknowledging that his rival was, at least to some extent, correct. Independently, each man concluded that intelligence has properties of unity and diversity.
Ultimately, the unity–diversity paradox was resolved in the hierarchical model of intelligence. At the apex of the pyramid-shaped model is general intelligence, akin to Spearman’s g, and below the pinnacle is a small set of broad factors that bear some resemblance to Thurstone’s primary mental abilities. Among the broad factors, two stand out as especially important—fluid intelligence and crystallized intelligence. At the base of the pyramid are narrow factors, such as perceptual speed, that identify quite specific aspects of mental proficiency. Articulated as a hierarchy, these three layers—general intelligence, broad factors, and narrow factors—present a coherent picture of intelligence.
A second paradox also emerged in the earliest days of scientific research on intelligence. It concerned whether intelligence consists essentially of elementary processes, such as reaction time and sensory acuity, or whether intelligence is at heart complex, expressed as the ability to process abstract and intricate ideas. The polarities of this paradox were identified by two pioneers in the study of human intelligence, Francis Galton and Alfred Binet. In the late 1800s, Galton poured considerable effort into testing his hypothesis that eminent achievers were biologically superior to their less remarkable peers. Galton focused on basic parameters of the senses, nerves, and motor control.
Alfred Binet’s assumptions about intelligence flowed in the opposite direction. He saw intelligence as reflecting higher-order thinking, including both knowledge about the world and the ability to reason. Binet’s assumptions were built into the design of the first intelligence tests and received immediate support through the ability of those tests to predict children’s success in schools. Galton’s assumptions, by contrast, were not immediately confirmed, but subsequent generations of researchers uncovered evidence for his ideas. Some very simple tasks, such as speed in comparing the length of two lines, correlate surprisingly well with IQ scores. These studies implied that intelligence reflects basic parameters of the nervous system, such as reaction time and nerve conduction velocity. This is not too different from the way Francis Galton thought about the sources of eminence in the mid-1800s. Alfred Binet’s view, that intelligence is tied to higher-level functions, has a great deal of support and is much more popular today.
Here, then, is the resolution to the second paradox: Research supports a view of intelligence as both lower-order and higher-order. Both Galton and Binet had valid insights into the nature of intelligence. This makes sense: The mind’s ability to engage in higher-level operations must in some way rest on a foundation of lower-level functions. This explanation maps nicely to a computational model of the mind. In such an interpretation, the efficiency of lower-level process such as neuron function and basic perceptual tasks are akin to a computer’s underlying hardware. Higher-level reasoning and problem solving draw, in turn, on the mind’s software—the accumulated content-rich knowledge acquired through experience.33 If the comparison between the mind and a computer holds up, we see the dualities converge and reconcile. The power of any computer to do complex work rests on its elementary design parameters, such as clock speed, cache memory, and bit register size. That human intelligence might likewise depend partly on the mind’s basic design parameters should not be too surprising.
We can sum up this way: Over many decades of research, progress on understanding human intelligence has entailed the exploration and eventual resolution of two fundamental paradoxes:
1. Is intelligence one or many?
2. Is intelligence simple or complex?
The history of research on intelligence reveals that the resolution of key paradoxes was not accomplished by one side winning out. Instead, the paradoxes were resolved through the emergence of new ways of thinking: Intelligence is both unified and diverse, and intelligence is both simple and complex. The answers were, in effect, more complex than the questions originally posed. This tells us something: We must understand intelligence as fantastically rich and complex, an object of inquiry that is worthy of our best efforts to illuminate it. But we should not expect simple answers. Progress will be difficult. Yet the nature of intelligence is discoverable, and its essence is still being revealed to this day. In the next chapter, we’ll examine some of the more recent forays into the frontier.
