Sections include lectures by Dr. Michael Posner, Dr. Elizabeth Spelke, Dr. Brian Wandell, Dr. Ellen Winner, and Dr. Gottfried Schlaug
In 2004, the Dana Foundation began exploring whether training in the arts changed the brain in ways that transferred the benefits of arts training to other cognitive abilities. Dana established the Arts and Cognition Consortium—made up of nine investigators at seven major universities—to take largely anecdotal and correlative observations about the potential role of the arts in enhancing a child’s overall cognitive ability and subject these to rigorously designed neuroscientific studies.
Over the next three years, the researchers studied the brain’s response to early training in dance, drama, and music. In 2008, the consortium published its results: Learning Arts, and the Brain: The Dana Consortium Report on Arts and Cognition.
In the report’s opening remarks, Consortium Director Michael Gazzaniga, Ph.D., offered a measured, but ultimately optimistic, introduction:
Is it simply that smart people are drawn to ‘do’ art—to study and perform music, dance, drama—or does early arts training cause changes in the brain that enhance other important aspects of cognition?
The consortium can now report findings that allow for a deeper understanding of how to define and evaluate the possible causal relationships between arts training and the ability of the brain to learn in other cognitive domains.
Gazzaniga, director of the Sage Center for the Study of Mind at the University of California, Santa Barbara, summarized eight key highlights of the consortium’s findings:
- An interest in a performing art leads to a high state of motivation that produces the sustained attention necessary to improve performance and the training of attention that leads to improvement in other domains of cognition.
- Genetic studies have begun to yield candidate genes that may help explain individual differences in interest in the arts.
- Specific links exist between high levels of music training and the ability to manipulate information in both working and long-term memory; these links extend beyond the domain of music training.
- In children, there appear to be specific links between the practice of music and skills in geometrical representation, though not in other forms of numerical representation.
- Correlations exist between music training and both reading acquisition and sequence learning. One of the central predictors of early literacy, phonological awareness, is correlated with both music training and the development of a specific brain pathway.
- Training in acting appears to lead to memory improvement through the learning of general skills for manipulating semantic information.
- Adult self-reported interest in aesthetics is related to a temperamental factor of openness, which in turn is influenced by dopamine-related genes.
- Learning to dance by effective observation is closely related to learning by physical practice, both in the level of achievement and also the neural substrates that support the organization of complex actions. Effective observational learning may transfer to other cognitive skills.
While studies that measure cognitive changes before and after arts training can help determine whether the two are correlated, only through randomly assigning students to receive arts training or some other intervention can studies prove causation. Pragmatism, therefore, is the watchword of consortium researchers, who caution readers to avoid being carried away by the initial promise of the report’s findings. “These advances constitute a first round of a neuroscientific attack,” observes Gazzaniga, “on the question of whether arts training changes the brain to enhance general cognitive capacities. The question is of such wide interest that, as with some organic diseases, insupportable answers gain fast traction and then ultimately boomerang.”
The report has gained considerable notice since its debut in early 2008. In a year’s time, eight new scholarly articles cited research published in Learning, Arts, and the Brain, including A Federal Arts Agency at the Center of Reading Research: How We Got Here.1 From the April 2008 Neuroscience and Music conference (sponsored every three years by the Pierfranco and Luisa Mariani Foundation) to the Neuroscience Research in Education Summit at the Center for Learning and Memory at the University of California, Irvine in June 2009, consortium contributors met audiences eager for more information about their findings. Consortium and other researchers in the field gathered in May, following the summit, at the Learning and Brain Conference in Washington, DC, where the theme was “The Reactive Brain: Using Brain Research in Creativity and the Arts to Improve Learning.”
At the Hopkins summit, Dana Foundation Chairman William Safire declared the report’s essential findings now beyond dispute: practice in art forms changes cognition; genes and environment determine the efficiency of the neural networks involved in attention; and advances in neuroimaging allow an ever clearer view of these processes. His address underscored the relevance and timeliness of the consortium findings, calling for continued and concentrated research critical to identifying the causal relationships so eagerly anticipated by educators and scientists alike.
Consortium researchers Drs. Posner, Spelke, and Wandell, along with Drs. Winner and Schlaug, whose research had also been supported by the Dana Foundation, discussed their current work along with research that was published after Learning, Arts, and the Brain.
Edited Excerpts from the Research Presentations
Dr. Michael Posner
Neuroimaging has provided an analysis of many of the cognitive and emotional tasks that people perform. Using various types of imaging, researchers have been able to identify brain areas that are active as a person performs a specific task.
Today we’re especially interested in the brain networks involved in various forms of the arts. My Oregon undergraduates show interest in a particular art form more than in the arts as a whole. One may be interested in music, another in dance or theater. And their performance and observation of that art are highly correlated. If you’re interested in drawing, you’ll also be interested in observing fine arts and other people drawing.
Art forms involve distinct brain circuits, including, of course, sensory networks. For example, music engages the auditory system and the visual arts engage the visual system. Studies have provided a detailed analysis of the many brain areas involved in each of the art forms; these areas are quite distinct, although they may overlap in some cases. Dr. Daniel Levitin’s research identifies various parts of the experience of music (such as emotional, auditory, and so on) and then maps them onto particular areas of the brain. Levitin finds, rather surprisingly, that an area in the cerebellum is involved in some of the emotional aspects of music, probably because listening to and composing music involves movements, which are made precise by the cerebellum.
Research suggests that each art form involves some neural network, although this assertion is not without dispute and requires further study. But it’s more or less generally agreed that performance or practice of any art form strengthens the network involved in that art form. So on the question of whether the brain is plastic—can it change with experience—yes, it certainly can.
We know this from neuroimaging, which shows that the connections and activations within various parts of networks involved in specific art forms are changed with experience, with practice. Brain imaging has revealed a plausible process by which practicing an art influences cognition in general.
We found that our Oregon undergraduates were not only interested in particular art forms, but that their interest was related to a more general propensity to creativity and imagination. Interest in an art form is correlated with the degree to which that person feels interested in imaginative or creative acts. I think this openness to creativity in an art form is important in understanding how practicing an art actually produces changes in cognition.
So, these elements lead to a kind of theory of how the arts might be related to cognitive processes. First, there are neural networks for each specific art form. Second, there is a general factor of interest in the arts due to creativity, openness to that art form. If you choose an art form that a child is interested in and open to through a general factor of creativity, the child will be engaged when he practices that art form. Earlier today, Ellen Galinsky told us that when a child is engaged in the learning process, that’s when his or her attention is fully focused.
We now know that training preschool children and adults to focus their attention can produce improvements in general cognitive processes.
The network of neural areas involved in executive control or executive attention get exercised and strengthened such that the training will produce improvements in a large number of other cognitive tasks, including general intelligence.
Each neural network is associated with a specific neural transmitter—in the case of the executive attention network, the transmitter is dopamine— and therefore with particular genes involved in producing that transmitter and building the network of brain cells it uses to communicate.
We all have an executive attention network, but some of us have more efficient ones than others. These differences in efficiency are partly related to genetic factors and also to individual experience. We all have the same genes that build these neural networks, but there are alternative forms of these genes, termed different alleles.
In our longitudinal study of children who are followed from seven-months to four-years old, we have seen powerful interactions between their genes and experiences. In this particular case, the experience is the quality of parenting, which influences both a child’s behavior and how efficiently his executive attention network functions. For example, in children with one type of variation of a gene, parenting makes a huge difference in the child’s impulsivity and risk-taking. This is not the case in children without this specific allele, or gene variation.
In two-year olds with a particular allele, parenting makes a large difference in the child’s ability to attend to different visual locations. No such difference occurs in children without this gene variation. As these examples show, genes and environments interact to build the neural networks involved in attention.
There is a great amount of newly published research findings concerning various ways to train people to pay better attention. For example, we have trained children aged four to six over a period of five days by engaging them in tasks that exercise their executive attention network. Now, five days is not very long, but it’s crucial.
To establish that the training caused changes in the executive attention network, we randomly assigned the children to a group that underwent systematic attention training or to one that received another kind of training. At the end of those five days we found, by recording from small electrodes that are placed on the children’s scalps, that those who received the attention training showed changes in the underlying executive attention network. These changes not only produced better executive attention and executive attention tasks, but the improvements generalized to intelligence and therefore, we think, to other cognitive skills.
Another interesting finding has just come out from Dr. Ellen Bialystok. For a number of years, she has shown that children and adults who have learned multiple languages perform better on executive attention tasks. As state legislatures advocate that lessons be taught solely in English, research is showing that bilingualism leads to better overall executive attention and therefore increased intelligence.
In her most recent study, Bialystok demonstrates that, in addition to bilingualism, vocal and instrumental training also are correlated with improvements in executive attention. Because this is a correlational rather than a random assignment study, we cannot infer causation. But the research nonetheless shows us that changes in executive attention can occur with experiences that one is likely to have in the real world.
We have a plausible way of seeing how the arts may be able to influence cognition, including intelligence. If we are able to engage children in an art form that they are open to and for which their brain is prepared, then we can use it to train their attention, which seems to improve cognition in general.
Dr. Elizabeth Spelke
The first thing we’re apt to think of when we ask the question, “What’s special about the human mind?” is our extraordinary capacity to understand the world by developing formal systems, technologies, and also mathematics and sciences, activities that archeological and historical records show go back a long way.
But of course, those same archeological and historical records show that our propensities for artistic creation go back just as far. When we look around the world today, I think we see two things. First, that these kinds of activities are ubiquitous. Formal science may be a rather specialized thing, but the tendency to think systematically about the world, to transform it through technology, the visual arts, music, and so forth are characteristics of all living human groups, so in some way they come naturally to us.
The other thing that we see as we look around the world is great variety, both over space as we look from one human group to another and over time as we look within our own cultures. For example, we see enormous change in the technologies and art forms that our children are enjoying relative to those we found joy in as children. That change tells us that although the predisposition may exist for both science and art to be innate within us in some way—part of our human nature is to engage in these activities—the particular activities that we engage in are highly transformed by learning. The particular arts and sciences that we learn depend on the specific places we live and the activities of the people around us. In that context, it’s not surprising that education throughout the world has focused on mathematics and science but also on literature and reading and visual arts and dance and music and so forth.
Educators need to take diverse sets of human endeavors and present them to varied groups of students in a way that engages them, enables them to teach themselves, and allows their interest and knowledge to grow. The kind of work that I do, research in human cognitive development, will never give a direct answer to the question, “How can teachers better teach children?” But I do hope that as researchers, we can provide some insights that could become ideas for new directions in teaching that then could be pursued through further research, engaging researchers and teachers more directly.
One set of efforts, which might be useful to teachers, attempts to take all of the complex things that humans do, like formal mathematics or visual arts, and break them down into simpler component systems that emerge very early in the human mind and that children bring together as they start to master the complex products of our culture.
There have been three lines of research that have been useful in efforts to break down complex cognitive abilities. One compares the cognitive capacities of human infants to those of other animals, both those relatively close to us, like baby chimpanzees, and some more distantly related, like monkeys, other vertebrates, and even a few invertebrates.
This research asks two kinds of questions. First: What basic, evolutionarily ancient cognitive capacities are shared across broad ranges of animals, including humans? And second: What cognitive capacities are unique to us? What sets us apart from other animals on our distinctive paths of development?
That leads to the second line of research, comparing the cognitive capacities of infants to those of children from preschool throughout formal education, and to adults. Questions I think that are most useful to ask are: What are the basic cognitive capacities shared by people across all of these ages? And, what are the capacities that emerge later in development, and which processes lead to their emergence?
I began my research with a third set of comparisons, looking at mature forms of art and science across cultures, asking what’s universal and what’s variable from one place to another.
Much of my work has focused on developing knowledge of numbers and geometry, and has used these three research approaches to try to distinguish the basic cognitive systems that underlie these complex activities. We’ve found evidence for three systems that emerge at the beginning of human life, appearing in young infants, that appear to be foundational for the development of numbers, symbolic mathematics, and geometry.
One is a system for representing and reasoning small numbers of objects, for example, the difference between one object and two or three. The second is a system for representing and reasoning numerical magnitudes, a system that might let you, without counting, estimate that there are maybe 50 beads in a jar, with approximately equal numbers of reds and blues, and so forth. The third is a set of systems for representing and reasoning the shape of the surrounding environment, of forms, objects, and the large-scale spatial layout. Our capacities within each of these systems are limited, such as the number of objects that infants can keep track of at once, and the precision of numerical discrimination.
These limits allow us to track these systems over the course of human development. If we devise tasks for older children or adults that require them to make estimates about number or geometry without drawing on their high-level knowledge of mathematics, we find that they have the same abilities with roughly the same limits that we find in infants.
What’s more, research shows that schoolchildren draw on these systems when they learn further formal mathematics. We see this in two ways. First, as every mathematics teacher knows, some kinds of problems or principles are easy for kids to see while others are hard. Recent research, some from my lab, and some from the lab of Justin Halberda and Lisa Feigenson at Johns Hopkins, shows that there’s a tight relationship between one of these core systems, a system for representing approximate numerical magnitudes, and school achievement. If you separately assess children’s sensitivities to approximate numbers and then look at their symbolic mathematics achievement in school, you find relationships between these two abilities.
Finally, there’s a broad array of evidence showing that when adults engage in purely symbolic mathematical reasoning—for example, multiplying two-digit numbers in our heads—we engage these core cognitive processes that we share with human infants. Research involving special populations—for example, patients with brain conditions that damage the core systems—shows that they have corresponding impairments in the symbolic mathematical systems.
Before children begin school, they start bringing core systems together to master some of our culture-specific skills. There are three skills in particular that most children in our culture master somewhere between the ages of three and six. They bring together their representations of small numbers of objects and of large but approximate numerical magnitudes to construct representations of exact number, the system of natural number concepts. Children become highly skilled in the system when they master the mechanics and especially the logic of counting.
Before they get to school, children also begin to develop intuitions about measurement—the idea that numbers can be thought of as positions in space, points on a line. Clearly, most measurement skill is learned by children after they have started school. But intuitions about relations between numbers and space go back to infancy. Evidence suggests that this ability demonstrates that children spontaneously relate their intuitions about number to their intuitions about space.
Finally, we and other investigators find that preschool children are able to bring together their core understanding of space with that of objects to develop early symbolic abilities to use things like simple geometrical maps. In such maps, geometrical relationships among points on a page specify spatial relationships between objects in a real, three-dimensional environment.
I’ve taken you through this whirlwind tour of some of our work on mathematical development because this was about where we were in our research when Michael Gazzaniga and William Safire approached me with the form of the challenge: Do you really think mathematical development is only a matter of learning about numbers and points and lines? And what’s more, do you really think that if you understand mathematical development, or maybe the development of math in relation to science, that that will be enough to give you a real picture of the uniqueness of human nature and human cognition? What about all of those arts subjects that also characterize us as humans? How do they fit into the picture of the organization of cognitive systems in the mind?
I was, of course, struck by the longstanding suggestion that there’s a special tie between mathematics and music. Our research had shown that mathematical ability isn’t just a single special-purpose system in the human brain, but a process that comes together from multiple systems. What might this relationship between music and mathematics actually come down to? That’s the question my lab set out to answer as part of the Dana consortium.
We did three different studies aimed at three distinct populations of children: those in elementary, middle, and high school. All three studies asked whether children who received music training showed any associated advantage on the particular abilities underlying mathematics performance. In different studies, we looked at musical training at different levels of intensity, from extremely weak in the first study, to moderate in the second, and intense in the third. The third study focused on high-school students in a school for the arts; a particular art form was their primary academic interest.
We first assessed the functioning of children in all three groups on each of the three core systems that I described earlier. The first thing we found was that mild amounts of arts training had no effect at all. I think that’s probably because our measures weren’t sensitive enough. I wouldn’t draw any conclusions, positive or negative, from those findings.
But the children who received moderate or intensive music training showed significantly higher performance on tasks that tapped into just one of the three core abilities: there was a reliable difference in their representations of geometrical properties and relations.
Here’s the test that we used in this study: on any given problem, children were shown six different geometrical forms, five of which shared a particular geometrical property that the sixth did not. Across different problems, the particular geometrical property varied from one display to the next, as did the subtlety of the geometrical relationship. This task was hard enough that even the Harvard undergraduates, who were at one end of the age spectrum, were still making errors on some of these problems, but the difficulty was variable enough that three-year-old children were getting some of the problems right.
When we compared students who received music training to a matched group of students who received no special training in any art form, we saw a small but reliable association between music training and sensitivity to geometry.
At the high school for the arts, we compared performance sensitivity to geometry among students specializing in different arts disciplines. We found that intense training in visual arts, music, and dance was associated with better geometric sensitivity performance. Music and dance results looked indistinguishable from each other. Next, we looked at associations between music training and the core skills of counting, using number lines, and reading maps, and found associations between music training and the latter two skills, which tap spatial abilities. Children receiving moderate music training showed a small but reliable effect on the part of a map-reading task that relied purely on geometric skills. Students in the music and dance programs outperformed others on this geometric map task.
We’ve seen a consistent relationship between music training and three different measures of spatial performance. Many things could produce this relationship; all we have so far is a tight correlation. When you control for a number of other things, like motivation and verbal IQ, you still see this correlation. But the correlation doesn’t tell us what the source of this relationship is, which is what we’re trying to look at now.
To do that, I’m going back to my roots as an infant psychologist. The hypothesis that I’m exploring is the following: we know that, from birth, infants love to listen to melodies. A melody is a patterning of tones in time. There may be an inherent relationship between a melody’s temporal and tonal structures and representation of space.
Our hypothesis is that from infancy, when a child hears a long temporal interval between two notes, for example, they may spontaneously evoke a perception of a long spatial interval. When they hear notes going from low to high, they may spontaneously evoke a representation of a change in spatial position from low to high.
The first test that there’s a relationship between time, musical time, and space comes from a recently completed study from my colleague Susan Carey and her student Mahesh Srinivasan involving nine- or ten-month-old infants. They presented babies with worms to look at that were either short or long, accompanied by corresponding tones that were short or long in duration. In that situation, children learned the relationship between short objects and tones and long ones. To see whether this was a special relationship, they also tested a second group of infants, who saw exactly the same worms and heard exactly the same tones, but they were reverse-paired. The infants never learned that relationship, suggesting that there’s something special about visual length and auditory duration that could underlie a relationship between the experience of hearing sounds and the representation of space.
In our work with four-month-old infants, we were able to create sounds in different timbres. They were paired with objects of different heights. In some trials, a baby would hear a rising sequence of notes, while in other trials they would hear a falling sequence. But always, as in the case of the study I just described, the height of the object related to the height of the note: when the notes fell, so did the object.
In the second situation, we showed the infants the same objects and we presented the same sounds, but we reversed the pairing. Our findings were similar to those of Srinivasan. Four-month-old infants learned the pairing between tone and object heights when it was congruent but not when it was incongruent. As early as four months of age, babies seem to be sensitive to relationships between the two key properties of a melody and positions in space.
This finding motivates the following hypothesis, for which we do not yet have evidence: from the beginning of life, when an infant hears music, that music not only encourages melodic, but also spatial, processing. It may be that spontaneous spatial processing that gives rise to the relationships found later in life between music and mathematics.
We don’t know if that’s going to turn out to be the case, but I think it’s already enough of an active possibility that it gives us an additional reason for a flourishing arts curriculum in our schools. Connections across the arts and the sciences are rich and varied not only for adults, but also for young children.
Dr. Brian Wandell
I’m fascinated these days by our new ability to measure the connections in the developing human brain through imaging. It’s something that couldn’t have been done a decade ago.
The white matter of the human brain (bundles of brain cell axons that carry messages) connects different regions of cortex, a thin layer of gray matter (brain cells) that covers the surface of the brain where functional activity is measured. Connections between the parts of the cortex are just as important as the cortex itself. Some of these pathways, these white-matter connections, are essential if kids are going to learn how to read, and they’re also essential for learning certain mathematical skills.
It has been hard to get data about the white matter in the human brain to determine what is connected to what. You can’t pull apart bundles of white matter post-mortem without breaking the whole brain. But now with the development of magnetic resonance imaging techniques, we can measure how water moves around inside the brain in different directions. From these measurements, we estimate (using algorithms developed first by Tom Conturo at Wash. U., Susumu Mori at JHU, and Peter Basser at the NIH) where those major fiber bundles are headed to in the human brain. In diffusion tensor imaging, also called diffusion spectrum imaging, the goal is to learn where the white-matter connections are in the brain.
There are certain fibers that pass through the corpus callosum, the part of the brain connecting the two hemispheres. Looking at how effectively water diffuses in and around those fibers is quite predictive of how well children or young adults learn to read. Their reading capabilities and phonological awareness (the ability to manipulate speech sounds, which is predictive of reading fluency) are very highly correlated with the properties of these specific fiber tracts. The conclusion here is not yet that firm, but it’s roughly that the signals carried on these fibers between the two hemispheres are essential for learning the skills of phonological awareness, which is vital for learning the steps involved in reading.
As scientists, we think this is an important part of the learning pathway, and as we were conducting longitudinal studies in children over a period of time, Mike Gazzaniga and Bill Safire approached us. We began to consider whether or not we might make some measures of how much exposure these kids we were studying had to music or to visual arts. We took surveys of the parents, the kids, and so forth and made measurements.
One of the things that we found—that others have found also but that was quite striking in our study—was that in the children who had music training, the amount of this training they had in the first year of our study and over the three years of the study, was correlated with their reading skills. Music training explained 16 percent of the variance in the children’s reading abilities compared to those who did not have music training.
As a number of investigators have shown, this wasn’t an intervention—we didn’t randomly assignsome children to receive music lessons and others to receive some other kind of training. So we don’t know how this relationship between music training and reading fluency came about—whether children who had reading skills chose to learn music or the other way around. But now several of the investigators from this group, under the urging of the Dana Foundation, are doing controlled studies to see whether music training causes this effect in kids who hadn’t previously had music lessons.
We also discovered a modest correlation between visual arts and math skills, and this really surprised us. Jessica Tsang and Michal Ben-Shachar, collaborators in my Stanford lab, observed that visual-arts training was somewhat correlated with an ability to do certain kinds of mathematical reasoning (called a Woodcock-Johnson calculation). They then went to the literature to say, “Well, what could we as neuroscientists do to try and understand the basis of this?”
Based on work from Liz Spelke and Stan Dehaene on mental arithmetic skills and fibers connected to parts of the parietal lobe, Jessica got the kids in our study to come back and had them do various mental arithmetic tasks, either exact or approximate calculations. Jessica was an education student who started working in our lab because of our focus on the intersecting roles of education, cognitive science, and cognitive neuroscience. Her mom is a teacher in Oakland; one of the things her mom was stressed about was the pressure to teach kids to do approximate arithmetic. She didn’t know whether it was valuable or not, or how she should teach these things.
In our lab, Jessica looked for the parts of the brain that we might focus on for studies about visual-arts training and approximate mental calculation skills. She focused her measurements on how water flowed through a particular white matter tract, the arcuate fasciculus, based on work by Dehaene, and we saw that the correlation between this flow with approximate calculation skills was high. Then we said, “Well, maybe this correlation will be seen throughout the whole brain,” and so we looked at the adjacent white matter tract, but found no association with this math skill. The brain location that is associated with mental calculations of approximate math values really is quite specific, and is where studies can measure the effects of visual-arts training on this skill.
Brain region connections are another new thing that scientists can measure, even in very young children. We know that the healthy development of these connections is essential for cognition.
Dr. Ellen Winner
I’ll begin this talk, and Gottfried Schlaug will complete it, as we describe the study that we’ve been working on together for at least five years on the cognitive and brain consequences of music training in early childhood.
We’re first going to talk about the search for evidence that cognitive skills acquired from structured music-making transfer to other cognitive areas. We’re not going to be talking about the Mozart effect or music listening, only about children engaged in making music.
We are currently doing two prospective studies in other art forms. We are looking at the effect of theater on children’s ability to gain insight into other people’s mental states, empathy, and emotion regulation. That project is being led by my doctoral student, Thalia Goldstein, and was funded by the National Science Foundation, with Joan Straumanis as program officer. We are also looking at doing a prospective study of the effect of visual arts on spatial reasoning, which may improve geometric reasoning.
Today, we have two studies we’re going to talk about. The first is a correlational study; I’ll discuss the design and our cognitive findings. The second is a prospective study in which we followed children for a period of time. Gottfried Schlaug is then going to take over and talk about our brain findings and some interesting brain-behavior correlational findings.
I have been very skeptical of some of the extreme claims that have been made that when you introduce the arts into schools, test scores go up, attendance goes up, and everything improves. In 2000, with my colleague Lois Hetland, I reviewed all of the experimental causal studies looking at arts transfer published since the 1950s. We had to conclude that the claims exceeded the evidence.
There was some evidence that music-making improved spatial skills, but the results were mixed. Music was shown to improve spatial performance in some tests, such as the object-assembly test, which is basically a puzzle test, but not on the Ravens test, which is a matrices test that involves some spatial thinking.
In the verbal area, research showed that music training improved phonological awareness in children with dyslexia, but reading was not improved, though we’ve heard some new evidence since that review. Also, some published studies showed that verbal memory is improved in children and adults. Finally, we found six studies on music and math and conducted a meta-analysis of these; we found that the results were very mixed. It was not at all clear to us whether music would improve math.
Our correlational study asked whether learning to play a musical instrument is associated with higher cognitive skills in non-musical domains. We looked at 41 nine- to eleven-year-old children with three or more years of instrumental music training, along with 18 children in our control group who had no music training.
We had measures in near- and far-transfer domains. Near-transfer domains are those that are very closely related to music; the two we looked at were fine finger sequencing (we called that our motor-learning task) and melody/rhythm discrimination. We developed a task where children had to use their right and left hand separately on a computer keyboard; they had to learn a complicated sequence as fast and as accurately as possible. We also had a music task where children heard pairs of melodies, and had to decide whether the two melodies were the same or different. They also heard pairs of rhythmic patterns, and they had to decide whether the two patterns were the same or different.
For our far-transfer domains, we looked at spatial areas. We had three spatial tasks: the object-assembly task, which is like a jigsaw puzzle; a block-design task, where you’re shown a red and white geometrical design and you’re given a lot of blocks that are red and white and you have to copy the design with the blocks as quickly and accurately as possible; and the Ravens test, the matrices test.
We also looked at verbal measures, including the vocabulary test, which is often used as a proxy for verbal IQ. And we looked at a phonemic awareness test; we used the auditory-analysis test, in which children would be given a word, like “toothbrush,” and they would have to say it without the “r,” “tooth-bush.” They have to break the word apart into its phonemes and drop out one phoneme. That task is predictive of reading skill. We also looked at mathematics. We gave a standardized math test called the Key Math Test, which breaks down into many different areas of math.
On the two near-transfer domains (the motor-learning finger-sequencing task and the melody/ rhythm discrimination), we found that the instrumental children were significantly and reliably ahead of the control children. In our far-transfer tasks, we found that the instrumental children were ahead of the control children in verbal ability, as measured by the vocabulary test, and in nonverbal intelligence, as measured by the Ravens test. These scores were predicted by the duration of music training. However, we did not find any superiority in the music group in the block design, the object assembly—our spatial measures—the phonemic awareness, or in math.
In our prospective study, we asked: can we demonstrate near and far transfer from structured music-making in a causal intervention study? We gave our study participants a pre-test, had a four-year intervention with a music group and a control group, and then administered a post-test.
We did not randomly assign these children to the music instruction because we did not have the funds to do so, so we found children who were about to begin taking lessons on an instrument at age five, six, or seven, and we followed them for four years. We found other children who were not learning to play a music instrument, and we followed them as well.
The study started with 50 children between the ages of five and seven who were beginning piano or violin. They had 30 minutes a week of private instrumental lessons; we also measured how much time they practiced. Concurrently we followed 25 children of the same age who were not studying a musical instrument. By the 48th month—four years, a long time to keep kids in a study—we had 50 percent attrition, which is what often happens with longitudinal studies and why they are so difficult to do.
We tested every child at baseline, and we repeated assessments at 15 and 48 months; today we’re just going to talk about the 15-month analysis. Our measures were the same as those I described to you from the correlational study. Also, we used functional and structural brain imaging, and Gottfried’s going to present our results on that.
After 15 months, we had a small subset of children from the study, because we had to include only the children with usable MRI data. We had 15 children in our music group that were a little over six years old at baseline, and a control group of 16 children, who were about the same age at baseline and matched the children in the music group on several factors: verbal intelligence, as measured by the vocabulary part of the standard IQ test; gender distribution (we had the same number of boys and girls in each group); and interval length, the time between the first battery of cognitive and brain tests and the second battery, which was on average 15 months.
We found, first of all, no differences at baseline between our two groups on any measures. That’s good, because we weren’t able to randomly assign, but we want to be able to say that at baseline there were no preexisting differences.
Fifteen months later, on the near-transfer domains, we found that the instrumental group was reliably ahead of the control group on the finger motor-sequencing task with both their right and left hand, and they were also ahead in melody discrimination. However, on our far-transfer domain tasks, we had not found a superiority of the instrumental group over the control group after the first 15 months of the four-year study. Gottfried’s going to mention at the end of his talk why that might be, and he’s now going to present our brain findings.
Dr. Gottfried Schlaug
Those of us who play a musical instrument sometimes don’t think about what it all involves. Music-making is a multisensory motor experience, but it also involves attention networks, the motivation and reward system. I would challenge everybody to come up with another activity that engages as much real estate in the brain as music-making does.
What we have been exploring is whether or not the intense practice and early beginning of music-making lead to plastic changes in the brain, which in turn would support a nurture hypothesis, or whether professional musicians select themselves at a very early age because they have atypical brains to start out with, which make them predestined to become musicians. That’s one of the main questions that we have been trying to answer in this particular study.
I want to review with you just a few findings from the studies of adult musicians. One of our earliest studies looked at the corpus callosum, which is, as we have heard, a major fiber tract in the brain that connects the right and the left hemispheres. This major fiber tract is actually larger in adult musicians than in the matched non-musician.
We also found, in this adult group, that the earlier they began learning a musical instrument, and the more intensely they practiced, the more of a difference we actually saw. Some people take this as a marker in the absence of longitudinal studies that an early beginning and long duration of practice will actually lead to more brain differences and brain changes.
We’ve also looked at differences within musicians, to examine some of the hypotheses that suggest that this is really a selection bias, that musicians have brains that are conducive to making music. If you look at the motor region of the brain in keyboard players, you can see that it is enormously developed on both sides. Across the entire group, the motor region was more developed on the left compared to the right side of the brain, because fine motor control over the right hand is something that’s very important for a keyboard player. When you look at a string player, you see an opposite pattern, where the brain’s right side is much more developed than the left. Already we see a specialization in that part of the brain.
I want to provide a few other examples of differences that we have seen in some parts of the brain to demonstrate that the brain can really change not just in function but also in structure, and that these structural changes can be actually quite enormous.
These data were acquired from our studies in children, although we haven’t analyzed everything yet to see what kind of changes we might have over time. But I can tell you that our initial data indicate that there are very profound brain changes over time in relation to musical training.
First, I want to give you an example of functional changes in the brain before I discuss the results of our longitudinal study. In people with traumatic limb amputations, the brain’s motor region has been remodeled. This region has a little knob configuration—part of the precentral gyrus region—where the hand-movement region is localized. This knob configuration disappears on the affected side of the brain in people who have lost a limb, but remains on the unaffected side. There’s a complete remodeling of this normal anatomy on the affected side. It’s no longer disputed that the brain adapts in cases of injuries or loss of sensation. But whether or not a regular activity that one would do on almost a daily basis over many years would change the brain obviously needed to be proven.
As part of our longitudinal study, we asked our group of five- to seven-year-old children to do a rhythmic and melodic discrimination task. The children primarily activate temporal lobe regions when they do this task.
As we get older, the same task seems to activate more regions in the brain. When we look at our adult group performing the same simple task, they seem to be activating a lot of other regions, mostly multimodal, polysensory integration regions. Some of that greater activation could actually be related to performance, but it’s also that we’re using other regions of the brain to solve the same task.
We paid attention to some of the regions in the parietal lobe surrounding the intraparietal sulcus, which is one of the multisensory regions in the brain that integrates information coming from different domains. Closely related to what Liz Spelke was talking about, one of the theories we have is that the coactivation of some of these regions in the brain— and potentially the changes that music brings about in some of those regions—could be related to the association between music and math.
We don’t have direct proof of this, but we intend to follow up by determining whether or not the cognitive enhancements that we see are related to areas in the brain that are coactive or that have shared resources between different cognitive tasks.
When we look at our rhythmic-discrimination task, we didn’t see any significant change within the two groups at baseline. But by the time the second measurement was made, the activated areas seemed to be enlarged, not just in the temporal lobe, but also in different regions of the frontal lobe, and also in the cerebellum. We can actually see some of the changes that occurred between the two different time points in the temporal lobe, frontal lobe, and cerebellum when we do direct comparisons. We did not see these changes in our non-instrumental group.
Research results are increasing our understanding of the importance of the inferior frontal gyrus in various ways; this area might also be involved in some of the enhancements seen in other domains. Previously, many neuroscientists thought that the inferior frontal gyrus was mainly a region having something to do with speaking. But it’s actually much more complicated than this; it probably does tasks that we don’t fully understand yet in addition to playing a role in speaking. But its potential significance is broader than this.
We think it is a region that has something to do with mapping auditory sounds. Vanessa Sluming at the University of Liverpool showed that these regions located in the front of the brain seem to have more gray matter that is more fully preserved in musicians compared to non-musicians as they get older.
She was also able to show that this brain region is particularly active in musicians compared with non-musicians as they perform what we call a mental rotation task. Musicians seem to be using these regions to their full advantage, in some way affecting the sequential ordering of particular motor or sensory motor actions.
With regard to our morphometric findings, we developed a map of the differences that we found in adult professional musicians compared to adult non-musicians and amateur musicians. We found that primary motor regions, parietal regions, temporal lobe regions, and cerebellar regions all were different.
Those were the regions that we primarily concentrated on in our longitudinal study in kids.
Over time and across the entire brain, we saw differences in the motor regions and supplementary motor regions of the brain (more pronounced on the right side of the brain, which controls the left hand, because the left hand is the less skilled and needs more training). Within these regions, we found differences that were correlated with behavioral changes. So the better the instrumental children got in their hand tasks or their hand motor tasks, the more changes we actually saw.
We also found in a longitudinal study comparing instrumental kids with controls that the corpus callosum differentially changes. In particular regions that are located around areas where motor fibers, pre-motor fibers, and sensory fibers cross, we find the closest correlations at 29 months between the intensity of training and motor skills, and the relation of motor skills to the brain changes.
These changes are in the cerebellum and are related to auditory changes, or those involved in the melodic- and rhythmic-discrimination tasks, as well as to motor changes. Changes occurred in the brain’s auditory regions as well in the instrumentalist children compared to the control group, and the changes related to the behavioral gains that these instrumentalist kids showed in auditory- and rhythmic-discrimination tasks.
We consider this to be the first study that shows brain plasticity in young children as a function of instrumental music training. The amount of practice was related to the degree of changes, although the amount of practice was actually much less than we had initially anticipated. Nevertheless, we found that amount of practice was a predictor and that we have brain and behavioral changes that co-vary over time.
As Ellen already indicated, at 15 months we have not seen any clear far-transfer effects, the transfer to other academic domains. We have been debating why this is the case. Our tests may not be sensitive enough. There’s also quite a lot of variability in the testing. If we test kids in the afternoon or on the weekends, or if we test them while they are off in the summer break, the results are affected. While we were doing this longitudinal study we learned how one could potentially do it better, but I think some of the lack of far-transfer findings might potentially be related to the difficulty that these longitudinal studies impose on us.
Some of the brain changes were outside areas where we expected to find them—there were, for example, brain changes in areas typically associated with the attentional system. We would need to devise tasks that would specifically test or be related to some of these brain changes that we did not predict.
Last but not least, these additional brain changes in regions which are not currently correlated behavioral or cognitive factors obviously can be the basis for new research that we’re planning to do with these data.