 |
 |
 |
 |
| |
|||||
|
|
|
|
|
||||
Applying Neuroscience to Hard-to-Solve Problems Mark Happel 
t has been said that the 20th century was the heyday of physics. Advances in the understanding of the physical world yielded almost unimaginable military and economic power: supersonic aircraft, nuclear power and nuclear weapons, miniaturized electronics, digital computers and networks, orbiting reconnaissance satellites, and even the beginnings of manned space exploration. The United States has been successful at harnessing and managing this economic and military might. Yet today this nation faces major threats to its security and prosperity, and fighting them will require a different kind of power. For instance, the warfighter is now confronted by a battlefield in which distances and reaction times have been compressed by advanced weapons systems. Combating asymmetric threats posed by terrorists and transnational groups such as al Qaeda places a premium on the gathering and analysis of military and strategic intelligence. To prosper in the years ahead, many believe that the 21st century must be the heyday of neuroscience: the science of the human brain and mind. What exactly is neuroscience? Nobel laureate Eric Kandel of Columbia University has defined neuroscience as the study of "the biological basis of consciousness and the mental processes by which we perceive, act, learn, and remember." In dealing with such a broad charter, the neuroscience field has divided into a number of sub-disciplines, such as cellular neuroscience (focusing on individual nerve cells, or "neurons"), systems neuroscience (focusing on the structure and functions of brain subsystems), cognitive neuroscience (focusing on higher cognitive functions such as memory or decision making), social neuroscience (focusing on the biological basis of interpersonal interactions), and computational neuroscience (developing computational models of brain components or subsystems). MITRE is at the forefront of this movement, working to identify potential applications of neuroscientific research that could add value to our sponsors' work and to support those sponsors in the planning and execution of a number of innovative programs. MITRE is looking for ways to apply neuroscience research methods and findings to solve our sponsors' critical problems, particularly those that have long resisted solution by more traditional means. Our work can be divided into three major categories: understanding and enhancing human performance, bio-emulation, and neuroinformatics. Human Performance Advanced technology plays a major role in today's military systems and has taken the burden off troops when it comes to many tasks. However, the human operator will always be the most critical "component" of these systems, determining their success. By better understanding the cognitive capabilities and limitations of the warfighters, we can optimize overall system performance. Traditionally, such efforts have been limited to studies of the visual and tactile interfaces between humans and computer systems, but broader opportunities exist. For example, we have worked closely with cognitive neuroscientists at Harvard University to develop visual spatial skills performance tests that could help us better understand the perceptual capabilities required for success as an image analyst. (These analysts specialize in deriving intelligence information from aerial reconnaissance pictures.) By better understanding the perceptual skills required for imagery analysis, it should be possible to specify more appropriate tools for aiding analysts in the performance of their duties. In the broader sense, the study of the cognitive processes of intelligence analysts, such as memory retrieval, inference, and decision making, could lead to the development of a more systematic science of analysis. Building on the Harvard project team's research, we are currently developing
a novel training tool to improve the visual spatial skills of imagery
analysts. We are collaborating with researchers at Georgetown University
to conduct experiments in which participants will undergo functional brain
imaging while taking the visual spatial skills tests mentioned earlier.
(Brain imaging is a neuroscience research method in which images of the
brain are taken while the participant is performing a cognitive task to
show which portions of the brain are participating in the accomplishment
of that task.) By examining the effects of the new training tool directly
on the brain, it will be possible to more effectively assess the effectiveness
of this novel training method.
Bio-emulation Bio-emulation encompasses efforts to build models of neurobiological processes, components, or systems. These models could be intended as research tools, permitting the testing of theories and hypotheses when manipulations of the actual neurobiological substrates would be impractical or unethical, or they could be intended as practical applications systems, such as robotic controls. While these models are generally implemented as computer algorithms, they need not be. On the one hand, the models could be implemented as physical automata using specialized hardware (often called "neuromorphic" systems); alternatively, the models could be more abstract, such as the design for a business organization, including the requisite organizational structure and business procedures and practices, based on the substructures and patterns of information flow in the human brain. In one bio-emulation program, MITRE is assisting the National Geospatial-Intelligence Agency with research and development efforts that apply neuroscience knowledge about the human visual system to create automated image analysis systems. While research in computer vision traditionally relies on analyses of the physical processes underlying image formation and the application of sophisticated signal and image processing techniques, we have helped to put together a team of outstanding neuroscientists who specialize in studies of the human visual system. Because the capability to take reconnaissance photographs exceeds the capacity to manually extract usable intelligence from them, these neuroscience-enabled automated systems could someday play a key role in the nation's defense. MITRE is also working with the Defense Advanced Research Projects Agency's (DARPA's) Information Processing Tech-nology Office on an ambitious neuromorphic model. We are helping DARPA develop an information processing architecture for intelligent machines using the human brain as a model. The goal of the Biologically Inspired Computer Architectures program is to develop a neuroscience-based information processing architecture for intelligent machines that would be capable of solving some of the Department of Defense's most challenging problems. By breaking down the systems the human brain employs for memory, learning, executive processes, and language, we can begin plotting a process for exporting brain-based architectures into artificial cognitive systems. This work could provide revolutionary advances in artificial intelligence. Neuroinformatics Unlike the previous category, which described the application of neuroscience to information technology, neuroinformatics involves the reverse process: the application of information technology to neuroscience. The human brain, composed of as many as 100 billion neurons, is one of the most complex structures in the known universe. The high degree of variability among individual brains complicates attempts to understand the principles underlying normal brain function and development, making the identification of those mechanisms responsible for disease-related or age-related cognitive impairment that much more difficult. MITRE is working on the Human Brain Project, which is sponsored by the National Institute of Mental Health (NIMH). We are applying information processing techniques—such as data mining, database security, and probabilistic spatial mapping as a research tool—to unravel the complexity of the brain. One problem we're trying to solve is how to get the right data into the hands of researchers. There is a plethora of information, such as images of brain scans, in databases around the country. We have created the Content- Based Image Retrieval system to narrow queries more quickly and help researchers find relevant information. This project is part of a larger work program in support of the International Consortium for Brain Mapping (ICBM). We play a key technical role as part of the mixed academia/industry ICBM team (which is funded by NIMH). For more on this work, see Advanced Image Retrieval for Neuroinformatics. What's Next? As a not-for-profit organization made up of three federally funded research and development centers, MITRE's role is often to look ahead in areas of science and technology and figure out how new ideas may solve our sponsors' problems—today's and tomorrow's. We feel there is a great deal to learn in neuroscience that will be valuable to a variety of government agencies. After all, the workings of the human brain affect everything we attempt—every new idea we have. MITRE has built a solid area of expertise in neuroscience that continues to expand with the needs of our customers.
|
|
||||
| For more information, please contact Mark Happel using the employee directory. Page last updated: May 24, 2005 | Top of page |
|||||
Solutions That Make a Difference.® |
|
|
Download this issue [3 MB]