he Swiss group used electron microscopy to see what appear to be new physical connections being generated between nerve cells. Where there had been one physical link between neurons, there now appeared to be two. "One of our current hypotheses is that the brain is so effective because it is ‘programmable’ at the level of individual connections,"says Whitaker investigator Kwabena Boahen, Ph.D., of the University of Pennsylvania.
The same does not hold true for most computers at least not yet. Boahen’s laboratory is one of many that are exploring new ways of building computers modeled after the human brain.
One strategy is to construct virtual connections that can be rerouted on the fly, like the Internet, which operates without dedicated point-to-point wiring.
"In the brain, new connections sprout while old ones are pruned," he says."We are using this to model memory storage in the hippocampus."
His laboratory is also investigating synaptic organization for vision, spike timing for hearing, and firing modes for attention.
The research cuts both ways: It will aid in developing better computers and will yield new insights into the workings of the brain.
The Whitaker Foundation has supported a wide range of brain research through its Biomedical Engineering Research Grants and various awards for institutional, program, and student support. A large number of grants support brain imaging studies. These awards are unraveling some of the mysteries underlying basic brain function.
Others directly support work on such topics as learning, hearing, artificial vision, the effects of whiplash and other trauma, and specific diseases such as epilepsy and bipolar disorder.
In biomedical engineering, studies of the brain cover such areas as neuroengineering; neuromodeling and mapping; imaging; neural coding and signal processing; and neural sensors and stimulation.
Each of these areas is contributing to a deeper understanding of the brain in health and disease.
Neuroengineering attempts to mimic brain functions using engineering and computational approaches. The process yields applications for medicine and science and tells more about how the brain works.
Some neuroengineering studies look at how the brain perceives light and sound and try to apply this new understanding to restore sight to the blind and hearing to the deaf. Much of current prosthetic vision research is focused on one of four sites in the visual system: the surface of the retina, behind the retina, the optic nerve, and the brain’s visual cortex.
As you read this sentence, your brain transforms images from your eyes into words and sentences in a blink of time, perhaps assigning some of these perceptions to memory.
This starts with the retina at the back of the eye, a tissue-thin surface weighing less than a staple and using 0.1 watt of power. The retina is dotted with light receptors connected to ganglion cells that interpret small parts of the visual scene, compress this information by 100 times, and pass it to the brain.
"Ultimately building neural prostheses requires us to match the efficiency, and not just the performance, of the brain," Boahen says."Existing prostheses do not even attempt to replicate the physiological function."
His laboratory has developed a retinal chip that computes 1,000 times more efficiently than a computer using a small fraction of the energy demand, which would otherwise be dissipated as heat. Although it falls three orders of magnitude below the retina’s energy efficiency, the device can be mounted flat against the retina and stimulates up to 4,000 receptor sites, possibly enough to let a blind person read.
Neuromodeling and Mapping.
How do neurons know which other neurons they should seek out during development? How is one cluster of neuron firing related to another? What rules govern the formation of new connections?
These are some of the questions being investigated as researchers develop mathematical and physical models of the brain.
Whitaker investigator Christine Schmidt, Ph.D., of the University of Texas at Austin has studied how electrical stimulation enhances nerve growth. Knowledge of how this works can be used to develop models of neurite outgrowth and regeneration.
Some modeling studies have direct clinical implications. Whitaker investigator Stelios Kyriacou, Ph.D., of The Johns Hopkins University is developing a geometrical model of the human brain that can be used to predict the strains generated by neurosurgical instruments. The goal is to minimize tissue damage during brain surgery and improve surgical outcomes.
Mapping the surface of the brain has been likened to charting the Earth’s surface. David van Essen, Ph.D., of The Washington University School of Medicine leads a team that is assembling computerized atlases of the cerebral cortex, the center of thought, learning, emotion, perception, sensation and movement, and the cerebellar cortex, which relays signals for thought and movement.
"Our present level of understanding of cortical organization and function remains fragmentary," he says. "It is similar to the rudimentary understanding held by 17th century cartographers of the Earth’s geographic and political subdivisions."
A major advance in brain science in recent years is the ability to see the living brain at work. This is achieved with technologies such as functional magnetic resonance imaging (fMRI), brain mapping with electroencephalography (EEG), positron emission tomography (PET), computed tomography (CT), MRI spectroscopy, and single photon emission computed tomography (SPECT).
These technologies produce detailed computer pictures of brain structures and reveal neurochemical changes in the brain as it processes information or responds to stimuli, such as drugs.
By imaging the brain, researchers have identified niches that respond to such stimuli as pain, pleasure, fear, memory recall, touching, and hearing.
These images have formed the basis for refining drug treatments, improving surgical outcomes, and developing new insights into brain structure and function.
Each imaging technology has its own advantages and disadvantages because it provides different information about brain structure and function. A PET scan, for example, can reveal activity on the molecular level, while an MRI is good for showing brain structures where the activity is located.
To get more complete and accurate picture of the brain at work, researchers are now interested in combining two or more imaging techniques at a time.
Whitaker investigator John Belliveau, Ph.D., of the Massachusetts General Hospital has combined fMRI with EEG to see millisecond changes in brain firing. The approach provides a unique map of both the location and timing of the brain’s neural signals. He is testing the approach in the visual system before moving on to other applications in sleep research and epilepsy studies.
The system is expected to show changes in functional networks within the brain that cannot be observed otherwise.
Brain imaging generally requires the subject to be immobilized, which limits the types of activity that can be studied. A new optical topography system from Hitachi takes functional brain images while the subject is writing, talking and moving. The subject wears a cap that records brain impulses correlated with these specific activities. A long-term goal is to image the developmental stages of a child and use the information gained as a guide for improving the way children are taught in school.
Neural Coding and Signal Processing.
Communication among neurons appears to occur in at least two ways. In one, messages are sent by a few dozen neurons firing single spikes. In the other, half a dozen or so neurons generate brief bursts.
These crescendos occur at any time, with activity synchronized as tightly as a thousandth of a second. Time differences of a millionth of a second between the two ears are perceptible. This gives the barn owl all it needs to find the exact location of a mouse in the dark.
Whitaker investigator Xiaoquin Wang, Ph.D., of Johns Hopkins has been studying the communication code of neurons as it applies to human speech. He wants to explain how the brain manages to separate the thread of a lucid conversation from the din of a cocktail party.
Other laboratories have mapped out sound processing regions in the brain that overlap with the visual system.This hones our ability to detect the location of sounds and the motion of objects.
Neurons code not only for the rhythms of sound, but also for the rhythms of movement. Understanding motor control signals can help researchers design bionic prostheses for people who have lost the use of their limbs.
These bioengineering studies draw on robotics, control-theory, and neuroscience to understand the biological signals that control movements. By recording motor control signals in the brain and correlating them with specific limb movements, researchers can use computers to reproduce those movements by means of the control signals alone.
Andrew Schwartz, Ph.D., of the University of Pittsburgh has been studying the basic signals that control arm movement and has found a large population of cells in the frontal cortex that codes for the trajectories of the arm as it travels through space and time.
"We have been looking at drawing movements that generate large and continuous changes in the arm's direction and speed," he says.
His interpretation of the neural code is accurate enough to predict the shape being drawn.This degree of accuracy is also being applied in experiments in which monkeys move robot arms with their thoughts