Skip to Navigation
University of Pittsburgh
Print This Page Print this pages

October 9, 2008

Science 2008: Scanning the brain

Can magnetic resonance imaging (MRI) scans of the brain aid in developing instructional techniques? Can they teach us why adolescents engage in high-risk behavior? What do they reveal about the highly evolved human ability to recognize faces?

These were some of the questions that researchers addressed at “Brain Functional Evidence of Normative and Impaired Behavior,” one of the sessions at Pitt’s two-day “Science 2008: On Our Watch” symposium. The eighth annual symposium was designed to showcase the region’s academic strengths in science, engineering, medicine and computation and highlight some of the recent research in these areas.

Among the speakers at the Oct. 3 session were Julie A. Fiez, Pitt associate professor of psychology and of neuroscience; Beatriz Luna, Pitt associate professor of psychiatry, School of Medicine, and of psychology, and Marlene Behrmann, Carnegie Mellon professor of psychology.

Julie A. Fiez:

“Can Functional Brain Imaging Be Used to Inform Educational Practice?”

Fiez focused on the link between neuroscience and education, based on her brain-imaging research. “In pursuing this research, what struck me is how enthusiastic teachers were about this connection,” she said. “At the same time, many of them seemed very naive and not necessarily in line with current thinking in neuroscience.”

Popular questions from instructors, for example, included: Is it true we only use 10 percent of our brain? and How can I teach in a “right-brain style” for those learners who are more right-brained?

“So there is a gap between new knowledge” and what’s in the public arena, Fiez said.

“In 1997, John T. Bruer of the James S. McDonnell Foundation published a very influential article called ‘Education and the Brain: A Bridge Too Far,’” Fiez said. “His major thesis was that we know too little about the neuroscience of the brain to make a quick jump to inferences about what we should be doing in the classroom. In the past decade, however, we’ve learned a lot. Is it possible that we might now be able to make inroads on thinking about what current knowledge has to say about what works in a classroom and why?”

Fiez cited three approaches to addressing this question that are emerging in the scientific literature: bolstering basic knowledge by using cognitive neuroscience techniques that eventually will carry over to classroom instruction; using a disease model to study the brain basis for certain educational dysfunctions, such as dyslexia, and comparing those with normal brain patterns, and taking an “engineering approach,” which Fiez focused on.

“The idea is that perhaps we could think about what are the neural systems that are engaged when we are learning tasks. Then we might learn what neural systems we want to engage in order to achieve our educational goals,” she explained.

Fiez said her research focused on a reinforcement core learning system, that is, adding a tangible reward for positive outcomes in the learning environment in order to shape a student’s behavior.

“We started with a very simple task,” she said. Subjects, who were hooked up to brain-scanning equipment, were shown the back of a playing card with numbers in the 1-10 range. They were asked to guess whether the value of the card was higher or lower than 5.

“After a short delay to measure the blood flow and the changes in brain activity, we gave them the outcome. They received either a green arrow with a message saying, ‘Congratulations, you are correct. You just won $1’ or a red arrow that says, ‘Sorry, incorrect, you just lost 50 cents.’”

After a brief interval to allow the subjects’ brain activity to revert to normal, they repeated the process.

“What we were looking for was whether there are brain areas that respond differently depending upon the outcome, a positive rewarded outcome or a negative punishing event like losing money,” Fiez said. “What we found, and what since has been replicated across a number of studies, is that the pattern of response is different between the outcomes.”

Those differing patterns were registered in the areas of the brain that contain basal ganglia, a group of nuclei found in the cerebral cortex that are associated with a variety of brain functions including learning.

“From the initial studies, we went on show that these responses appear in situations where the individual believes he or she has some control about the outcome,” Fiez said. “We then began to be more interested in whether or not these were learning signals and, if so, what kinds of tasks might these signals be important for?”

The researchers then completed a more complicated experiment called a declaration memory task, which was designed to look closer at the cognitive end of the spectrum. Subjects were shown a set of 60 cards, each containing three words. The subjects were to read the first word — for example, “desk” — and then choose between two following words — in this case, “herd” and “soap.”

“Words were chosen intentionally that have no relationship to one another, and the ‘correct’ answer was randomly chosen beforehand and never changed,” Fiez said. “We also didn’t tell the subjects they had to learn the correct pair for future trials.”

Again, correct answers were rewarded and incorrect ones carried a penalty.

The success rate in round 1, as expected, was about 50 percent, because the answers were random, Fiez explained. “What we found is that, if we look at performance in the first of three rounds, changes in the basal ganglia were relatively flat whether the outcome was positive or negative.”

But, on round 2, changes in the brain did occur based on the outcomes. “On round 2, the subjects now have some previous experience. The minute a set of words popped up, they now have some faint memory to draw on. But because they’d only seen the set of 60 words one time, they were not very confident of their response,” she said.

The task is analogous to being introduced to a group of 60 people and, afterwards, being asked to recall their names, Fiez said.

“In round 2, though, they now have some measure of control. If their memory is correct, then it’s within their ability to get a positive outcome. If their memory is faulty, they’ll get an incorrect feedback after their choice,” she said.

In round 2, there was more activation of the basal ganglia signals for positive outcomes, and even more so in round 3. “So basal ganglia signals are important and it depends on which round we were engaged in for the amount of activation. We also noted that the more the basal ganglia was engaged, it was also playing an important role in behavior, because the subjects were faster in making a decision. Once subjects know there’s a reward, plus the fact that they are more confident, there is higher activation in the basal ganglia.”

The researchers also found a corresponding activation in the brain’s cortical areas, which fluctuated based on correct or incorrect answers.

“This is speculation at this point, but we think there’s a link between the activation of the basal ganglia area and the outcome-related responses in the cortical areas,” Fiez said. If that can be shown, the finding may have implications for classroom instructional design based on an outcome-response reward system, she said.

Beatriz Luna:

“The Young and the Restless: fMRI Studies of Adolescent Brain Development”

“The period of adolescence is a unique stage of development. It’s far different from what we see in childhood and what we see later on in adulthood,” said Luna.

Adolescence is the stage of development where disorders, such as schizophrenia and mood disorders, emerge. It is also the peak time for suicides, she said.

“Despite the fact this is the time when we’re in our best physical health, it is also the time when we see a peak in risk-taking behavior, which actually increases the mortality rate by 200-300 percent,” Luna said.

She examined these adolescent vulnerabilities with the goal of understanding the association between brain and behavior in a normal population of adolescents.

First, changes in risk-taking behavior between children and adolescents is not explained by brain size, she said. “If I were to show you the brain of an 8-year-old and the brain of a 20-year-old, you would not be able to tell the difference. But there are processes that continue to develop. The last areas to develop are around the brain, on the outside of the brain, where information is being associated,” Luna said.

In the first two years of life, the brain has the ability to support complicated computations that in turn support voluntary behavior. But the connections in the nervous system that are required for those computations are isolated due to underdeveloped neural transmission channels, she said.

“As we age, the brain can work more in a collaborative fashion, and the executive parts of the brain can guide behavior. At the same time we’re seeing this, there is evidence of continued improvement in our voluntary control,” Luna said.

The two essential aspects of voluntary control, or executive function, are “voluntary response inhibition,” which Luna focused on, and “working memory,” the part of the brain where calculations are made prior to behavior, which also continues to develop in adolescence.

“Voluntary response inhibition really refers to the fact that whenever we do a voluntary act, there are many different choices we can make. Some may not be the best choice toward a particular goal, and those need to be inhibited, to be seen as inappropriate responses,” Luna said.

To study the differences in the ability to access a voluntary response inhibition among young children, adolescents and adults, Luna and colleagues studied subjects performing what is called an oculometer task, which requires non-verbal, unwritten responses.

“The task is as follows: We ask subjects to look at [a designated area on the wall] in front of them. They are instructed that there will be light flashing from somewhere in the field of vision, and that when they see the light, they are to not look at the light and immediately to look to the other side,” Luna explained. “This is a model that allows us to tap into this association between brain and behavior for three reasons.”

First, it is a direct measure of cognitive control, with a visual stimulus and a corresponding visual response, requiring minimal strategy. Second, the areas of the brain where the neural responses take place have been well-documented by researchers.

“Finally, and this is very important in developmental studies, the instructions are very simple, so that we can separate out discrepancies in development that would be due to the inability to understand the instructions,” Luna said.

Children do poorly at this task. “They keep looking at the light and then correcting themselves,” she said. “However, children do have the ability, because they can complete the task a low percentage of the time. What this shows developmentally is that younger people are unable to stop their response, that is, to access their executive parts working to control the proper response.”

The proportion of error decreases with age, Luna said. Adolescents have success rates much closer to the adults than to the children.

“We looked at the brain function while the subjects were doing this task. What we found was wildly distributed circuitry in the brain being activated. The collection of the regions were more active in the adult,” she said.

“What this told us is, No. 1, the response operation is crucial for executive control, and, No. 2, the response operation might still be immature, or limited, in the adolescent. We found that the adolescents, instead, were showing increased activity in the prefrontal cortex, in a similar manner to what we would see in an adult study when you’ve increased the cognitive load.”

The findings indicate that the circuitry and systems that support the ability to have a single inhibited response are there, even in childhood, but what is developing is the ability to use this tool in a more consistent manner, which requires executive function.

“This system is there, it’s being tapped into, but there is increased effort when you do it correctly, which means, in effect, they were all doing the same thing,” Luna said. “Looking at the cerebral cortex we found that in order to make a correct response, a part of the brain needs to be shut down — it’s a default system — before the activation of the inhibited response can occur. In adults, we saw this robust activity in the areas of the brain that are known to monitor our performance.”

There was only minimal activity in those areas for adolescents, which suggests those areas likely affect the ability to perform inhibited response consistently, she said.

In conclusion, she said, adolescents do show the ability of an adult-level of cognitive control and behavior, but the brain mechanisms that support it are immature. “Adolescence is not a disease. It is in fact a crucial and necessary stage of development when brain circuitry and behavior are beginning to be established.”

Marlene Behrmann:

“Facing Facts About Face Recognition: Underlying Natural and Psychological Mechanisms”

“The goal of my talk is to examine the underlying neural substrate that gives rise to facial recognition. This discussion also has more general implications on how complex behavior or function emerge from the brain itself,” Behrmann said.

Face recognition may be the most developed of human sensual skills, she said, because so much information can be extracted: approximate age, gender, identity and states of emotions. What is the system that allows for accurate and efficient face recognition?

“There’s been a longstanding view in the literature that the solution to that complexity is simply to take the face recognition problem and assign it to one region of the brain, that there is a single dedicated part of the cortex whose sole job is to recognize faces. This has long been the dominant view,” Behrmann said. “This view I will challenge.”

Conventional wisdom says the dedicated region of the cortex, known as the fusiform face area (FFA), is the area of the brain that is activated during face recognition. FFA is located on the ventral surface of the brain’s temporal lobe within the cortex.

Cortical specialization is perhaps even specified in the genome, so that one doesn’t have to have experience with identifying faces in order to get this region of the brain working at its maximum capacity, Behrmann said.

In challenging the idea that there is this one-to-one structural correspondence, Behrmann studied people who have prosopagnosia, a perception disorder where the ability to recognize faces is impaired, while the ability to recognize other objects may remain relatively intact. About 2 percent of the population suffers from the disorder, although it varies in degrees.

Behrmann focused on a subdivision of the disorder known as congenital prosopagnosia (CP).

“For people with CP, it’s not that they don’t know who these people are that they’re seeing, the problem is in the perceptual domain,” Behrmann said. “These individuals have no brain damage, they have normal intelligence, and the disorder appears to run in families.”

She discussed a study of a group of people with CP measured against a control group of those with normal ability to recognize and distinguish faces.

First, she showed the two groups pictures of pairs of faces (including some famous people) and asked which were pictures of the same person and which were not. She measured the accuracy of the responses and the time of response, and the regions of the brain that were activated by these stimuli.

In the control group, brain scans indeed showed high activation in the FFA during the exercise, Behrmann said. “The premise would be that for those who do not perform well in face recognition there ought to be no uniform activation in the FFA. But that is wrong. They do show uniform activation. They performed poorly, yet there was evidence of normal activation.”

That suggests that activation of the FFA is necessary but not sufficient on its own to support face recognition, she said.

By performing a structural analysis, region by region of the brain, using high-resolution nuclear magnetic resonance spectroscopy, Behrmann determined that in the control subjects there was high activation in anterior regions outside the FFA, suggesting that an underlying distributed interactive network of neural fibers also is necessary for normal face recognition. “To do this highly developed skill, you’ve got to make contact with other regions of the brain,” she said. “It’s not enough to recognize eyes, or a nose. You can’t go to a local feature, you need to combine those local elements and to derive the interactions between the features.”

People with CP disorder, then, may be experiencing a failure to propagate information from the posterior regions to the anterior regions of the brain, Behrmann said. “You need to have this integrated circuit intact in order to put together pieces of the puzzle,” she said.

“It’s not enough to apply a single function to a single structure in the brain. This is an oversimplification of brain behavior. There are contexts underlying the surface. That’s true with face recognition, and it’s true with other functions as well.”

—Peter Hart

Filed under: Feature,Volume 41 Issue 4

Leave a Reply