We all know someone who is not as smart as we are—and someone who is smarter. At the same time, we all know people who are better or worse than we are in a particular area or task, say, remembering facts or performing rapid mental math calculations. These variations in abilities and talents presumably arise from differences among our brains, and many studies have linked certain very specific tasks with cerebral activity in localized areas. Answers about how the brain as a whole integrates activity among areas, however, have proved elusive. Just what does a “smart” brain look like?
Now, for the first time, intelligence researchers are beginning to put together a bigger picture. Imaging studies are uncovering clues to how neural structure and function give rise to individual differences in intelligence. The results so far are confirming a view many experts have had for decades: not all brains work in the same way. People with the same IQ may solve a problem with equal speed and accuracy, but using a different combination of brain areas. [For more on IQ and intelligence, see “Rational and Irrational Thought: The Thinking That IQ Tests Miss,” by Keith E. Stanovich.]
Men and women show group average differences on neuroimaging measures, as do older and younger groups, even at the same level of intelligence. But newer studies are demonstrating that individual differences in brain structure and function, as they relate to intelligence, are key—and the latest studies have exposed only the tip of the iceberg. These studies hint at a new definition of intelligence, based on the size of certain brain areas and the efficiency of information flow among them. Even more tantalizing, brain scans soon may be able to reveal an individual's aptitude for certain academic subjects or jobs, enabling accurate and useful education and career counseling. As we learn more about intelligence, we will better understand how to help individuals fulfill or perhaps enhance their intellectual potential and success.
For 100 years intelligence research relied on pencil-and-paper testing for metrics such as IQ. Psychologists used statistical methods to characterize the different components of intelligence and how they change over people's lifetimes. They determined that virtually all tests of mental ability, irrespective of content, are positively related to one another—that is, those who score high on one test tend to score high on the others. This fact implies that all tests share a common factor, which was dubbed g, a general factor of intelligence. The g factor is a powerful predictor of success and is the focus of many studies.
In addition to the g factor, psychologists also have established other primary components of intelligence, including spatial, numerical and verbal factors, reasoning abilities known as fluid intelligence, and knowledge of factual information, called crystallized intelligence. But the brain mechanisms and structures underlying g and the other factors could not be inferred from test scores or even from individuals with brain damage and thus remained hidden.
The advent of neuroscience techniques about 25 years ago finally offered a way forward. New methods, particularly neuroimaging, now allow a different approach to defining intelligence based on physical properties of the brain. In 1988 my colleagues and I at the University of California, Irvine, conducted one of the first studies to use such techniques. Using positron-emission tomography (PET), which produces images of metabolism in the brain by detecting the amount of low-level radioactive glucose used by neurons as they fire, we traced the brain's energy use while a small sample of volunteers solved nonverbal abstract reasoning problems on a test called the Raven's Advanced Progressive Matrices.
This test is known to be a good indicator of g, so we were hoping to answer the question of where general intelligence arises in the brain by determining which areas showed increased activation while solving the test problems. To our surprise, greater energy use (that is, increased glucose metabolism) was associated with poorer test performance. Smarter people were using less energy to solve the problems—their brains were more efficient.
The next obvious question was whether energy efficiency can arise through practice. In 1992 we used PET before and after subjects learned the computer game Tetris (a fast-paced visuospatial puzzle), and we found less energy use in several brain areas after 50 days of practice and increased skill. The data suggest that over time the brain learns what areas are not necessary for better performance, and activity in those areas diminishes—leading to greater overall efficiency. Moreover, the individuals in the study with high g showed more brain efficiency after practice than the people with lower g.
By the mid-1990s we were focusing on efficiency as a key concept for understanding intelligence. But then, in 1995, we discovered a difference in the way male and female brains work, giving us our first clue to what we know today: the concept of efficiency depends on the type and difficulty of tasks involved, and there are individual and group differences in brain function during problem solving, depending on who is doing the thinking. In the 1995 study we tested a specific mental ability—mathematical reasoning. We selected college students with either very high or average SATmath scores and used PET to investigate their brain function while they solved mathematical reasoning problems. Unlike the g studies, this study showed the people with high math ability using more brain energy in a certain region (the temporal lobes), but this was true only for the men and not for the women—even though both men and women performed at the same level on the test.
These observations have since been replicated by us and other researchers, especially in studies using advanced electroencephalographic mapping techniques. In addition to these experiments showing differences in brain function, brain structure also seems to play a role—studies have suggested that other gender differences in cognition, such as the tendency for men to have better visuospatial ability, may be rooted in brain architecture.
For example, in a series of papers published in NeuroImage starting in 2004, we used structural MRI scans to investigate correlations between gray and white matter volume and scores on intelligence tests. Gray matter, made up of neuron cell bodies, does the computational work of the brain. White matter enables communication among regions of gray matter via axons, brain cells' long, wirelike appendages. Our studies point to a network of areas distributed throughout the brain where more gray or white matter is related to higher IQ scores. The specific areas in this network are different in men and women, suggesting there are at least two different brain architectures that produce equivalent performance on IQ tests. In general, we found that in women more gray and white matter in frontal brain areas, especially those associated with language, was correlated with IQ scores; in men IQ scores correlated with gray matter in frontal areas and, especially, in posterior areas that integrate sensory information.
Children also show different developmental brain patterns related to IQ, depending on their gender. In a series of imaging studies with large samples, published from 2006 to 2008, neuroscientist Vincent J. Schmithorst, then at the Cincinnati Children's Hospital Medical Center, and his colleagues found that as girls age they show increasing organization—that is, well-defined paths between disparate brain regions—in their right hemisphere. Boys, in contrast, show this developmental trend in their left hemisphere. We do not yet know how these findings relate to behavioral or learning differences, but the research points the way for future studies to determine how brain development relates to boys' and girls' cognition and academic achievement.
A New Definition
Gender differences were merely the first indication that not all brains work the same way. In 2003 we investigated whether we could observe functional variations during passive mental activity without a task assigned. Again we used PET in two groups of volunteers selected for high or average scores on the Raven's test. Both groups watched the same videos passively with no problem solving or other task demands. The group with high test scores showed different brain activations in posterior visual-processing areas as compared with the average group. The data suggest that early stages of information processing are more engaged in individuals with higher intelligence, perhaps suggesting that the smarter people in the study were not watching the videos “passively” after all—they were actively processing what they were seeing.
Although more and more evidence shows that problem solving and even passive sensory processing do not look exactly the same in every brain, we still are able to identify a network of areas that seem to give rise to intelligence in general. In fact, defining the crucial regions and connections will help us delineate exactly how each person's brain works—every individual uses some combination of these areas in a unique way.
In 2007 neuropsychologist Rex E. Jung of the University of New Mexico and I reviewed the 37 neuroimaging studies on intelligence that existed at that point. In the journal Behavioral and Brain Sciences, we identified salient brain areas found in both structural and functional studies with some consistency. The 14 areas are distributed throughout the brain, refuting the long-held notion that the frontal lobes alone are the primary location for intelligence. In particular, parts of the parietal lobes, located under the crown of the head and known to be involved in sensory integration, play a significant role. Because areas in the parietal and frontal lobes were most represented across the studies we reviewed, we called our theory of intelligence based on this network the parietofrontal integration theory (P-FIT). The 14 P-FIT areas are involved in attention, memory, language and sensory processing.
Identifying the P-FIT network implies a new definition of general intelligence based on the brain's measurable characteristics. Both the amount of gray matter in certain P-FIT areas and the rate of information flow among these areas are likely to play key roles in intelligence. In 2009 studies at University Medical Center Utrecht in the Netherlands and the Chinese Academy of Sciences in Beijing used functional MRI to determine the efficiency of connections throughout the brain, pinpointing P-FIT areas where connectivity was especially associated with IQ scores. The findings support the idea that general intelligence not only arises from gray matter volume but also depends to a large extent on the white matter connections between crucial gray matter areas. More efficient connections allow information to flow faster—and quick processing times seem to go hand in hand with a high IQ.
Everyone Is Unique
But IQ scores do not tell the whole story—not even close. Intelligence seems to arise from varying combinations of the P-FIT brain areas in different people, which may explain each person's individual strengths and weaknesses. The challenges of identifying these patterns are well illustrated by the extremely rare cases of autistic savants. Daniel Tammet, for example, is a young adult with autism who has uncommonly high IQ scores. He sees numbers as colors and shapes, which allowed him to memorize the value of pi to 22,514 digits. He also learned to converse fluently in Icelandic after only seven days of instruction. Tammet leads an independent life and wrote a best-selling autobiography describing his extraordinary numerical and language ability. What would his “brain profile” show?
Although we are not currently able to deduce from a scan of Tammet's brain how his extraordinary abilities arise, the most recent wave of neuroimaging studies has given us clues to how we might one day do exactly that. New studies have found correlations between gray matter in certain areas and specific intelligence factors.
In March 2009 psychologist Roberto Colom of the Autonomous University of Madrid and his collaborators (including me) reported on the relation between gray matter volume and different intelligence factors in 100 young adults. Each person completed a battery of nine cognitive tests known to indicate different intelligence factors, including g, fluid intelligence, crystallized (factual) intelligence and a spatial factor. We found a positive correlation between scores on the g factor and the amount of gray matter in several areas predicted by P-FIT. And once we accounted for the common g factor, we found that gray matter volume in certain brain areas was related to the other specific intelligence factors.
One of the most tantalizing ideas to come out of this recent research is the possibility of matching an individual's gray and white matter pattern to his or her g and to other specific intelligence factors. In other words, the tissue in P-FIT areas may predict a person's unique pattern of cognitive strengths and weaknesses across a range of mental abilities. These differing brain profiles may explain why two people with an identical IQ score may show very different cognitive abilities. The data from Madrid illustrate this idea nicely. The person in our volunteer group with the highest g score showed far more gray matter than the group's average amount in several P-FIT areas—perhaps not surprisingly. But it is interesting to note that two people with identical g scores of 100, the average for the group tested in the study, exhibited different cognitive profiles, suggesting different cognitive strengths and weaknesses.
The idea that we all have our own pattern of variations in brain areas that contribute to different intelligence factors is underscored dramatically by a structural MRI study in March 2009 of 241 patients with brain lesions. Psychologist Jan Gläscher, then at the California Institute of Technology, and his colleagues showed that the site of each lesion was correlated with specific factor scores. For example, perceptual organization suffered—patients had trouble consciously understanding raw information from their senses—when their right parietal lobe was damaged.
A Smarter Future
These most recent studies suggest that neuroimaging could one day become a supplement or even a substitute for traditional paper-and-pencil intelligence testing. An individual brain profile could be valuable. In education, for example, a learning program could be tailored for an individual student, at any age, based on that student's brain characteristics. Perhaps vocational success could also be predicted—are there patterns of gray matter across some areas, for example, that make for the best teachers, fighter pilots, engineers or tennis players? People who are seeking a better life with vocational and career consultation certainly will want the choice of having a brain assessment if there are data to support its usefulness.
But it is worth remembering that, contrary to older dogma, the brain is not set in stone or in genetic immutability. Exactly the opposite is true. The brain is plastic—it changes. A brain profile detailing a person's strengths would offer a guide rather than a prescription—perhaps suggesting ways to practice skills or improve education so that a person could become better suited for the activities or careers he or she is most interested in. Fascinating recent studies show that learning to juggle increases the amount of gray matter in brain areas relevant to motor activity. When the training stops, the additional gray matter disappears. Because regional gray matter is related to intelligence, can training beyond conventional education approaches be directed at specific brain areas to increase intelligence? We do not yet know, but the prospect is exciting. Initial reports that memory training can increase intelligence were based on limited measures of intelligence, and, for the most part, they have not been replicated independently.
The next phase of neurointelligence research may include studies designed to answer such questions, including education experiments to determine whether different strategies produce specific brain changes and whether students selected on the basis of their individual brain characteristics are more likely to maximize learning in a particular subject with one educational strategy versus another. The goal would be to enhance current educational decision making by adding customized information about each student's brain. How any specific brain characteristic develops and how it may be influenced are critical, but separate, questions for research.
Whether everyone agrees on precisely the same definition of intelligence or not, progress in neuroscience is inexorable. We will continue to discover how the brain manages the complex information processing that undoubtedly underlies all notions of intelligence. Given the ravages of brain disease, aging, the technical needs of modern societies, the challenges of education and the joy of experiencing the world through intellect, there is some urgency to understand how smart brains work. It is not too early for discussion about the implications of the search for neurointelligence and our willingness to go where the data lead.