The human brain is a fascinating, complex system whose mysteries are becoming increasingly accessible to the tools of modern science. Neuroscientists have amassed a sizable body of knowledge about the structure of the brain and its specific functions, which has improved our ability to treat a variety of mental and neurological diseases. Many other diseases are less tractable, however, and effective treatments will require major advances in both basic and clinical neuroscience. Underlying these advances will be an explosion of experimental data, whose magnitude poses serious problems for information management and communication.
Effective access to existing neuroscience information is critical to the enterprise of discovery: such information forms the basis of new hypotheses, drives the search for improved methodologies, and, ultimately, leads to insights applicable to human disease. New strategies must be developed to enhance integration of this information and to facilitate new discoveries about the brain. Within the range of potentially beneficial strategies, the greater use of computer and information technology in neuroscience research holds particular promise.
This report synthesizes the deliberations of the Institute of Medicine 's Committee on a National Neural Circuitry Database, which was formed at the request of the National Institute of Mental Health (NIMH), the National Institute on Drug Abuse (NIDA), and the National Science Foundation (NSF). The committee's task was to consider the desirability, feasibility, and possible ways of implementing a family of resources, both electronic (e.g., computer networks) and digital (e.g., databases),
for the enhancement of neuroscience research. The incorporation of computer and information technology into diverse scientific fields is often accomplished in the absence of coordinated policy through trial and error by individual scientists capitalizing on rapidly developing technological innovation. These efforts have often led to far-reaching changes in the style of scientific research that call for new governmental policies regarding the conduct of research in this country and abroad. Within this context, and with an overall purpose of increasing the resources available to the neuroscience community, this committee has sought to
formulate a position on the requirements for and appropriateness of establishing a family of electronic and digital resources for basic and clinical neuroscience that would allow optimal data communication and sharing among investigators;
consider the scope and elements of data that would constitute these resources and determine strategies for representing the diverse data types generated by neuroscientists;
consider data storage, retrieval, and sharing schemes of existing national databases to identify successful strategies and potential pitfalls for the possible establishment of computerized resources for neuroscience;
describe the optimal organization of a family of computerized resources so that it can be efficiently established and coordinated for research use by the neuroscience research community, clinical investigators, clinicians, students, and educators; and
provide recommendations to NIMH, NIDA, and NSF on future directions in program development with respect to establishing such electronic and digital resources.
Advancing Neuroscience in the Decade of the Brain
The major advances in neuroscience in the past two decades have generated the opportunity that now exists to achieve an integrated understanding of the brain's structure and functions. These advances, which are producing new data on brain activity, include
the identification of complex anatomical connections;
capabilities for understanding the biochemical, molecular, and genetic mechanisms that control brain structure and functions;
the ability to measure and visualize human brain functioning during mental activity; and
the ability to monitor neural activity simultaneously in complicated networks of neurons.
These and other advances have occurred primarily through the efforts of individual investigators working in small groups on highly specific projects. Despite the piecemeal quality of these efforts, the information derived from them is so extensive that it is extremely difficult to coordinate it all and produce a meaningful picture of how the brain functions. This traditional method of neuroscience research — individual investigators, highly specific projects—is similar to a group of surveyors, each of whom has been assigned to chart a different geographical region. To chart an entire continent, however, the work of each surveyor must somehow be coordinated with the work of all other surveyors. Based on input obtained through a variety of means, as well as on its own deliberations, the committee considers it necessary to establish a formal Brain Mapping Initiative to coordinate the valuable efforts of individual neuroscientists in such a way that new discoveries occur with greater speed and efficiency. The Brain Mapping Initiative is meant to subsume all the proposed aspects of a National Neural Circuitry Database. In addition, it expresses explicitly the overarching goals of the proposed effort and reflects more adequately the complex of electronic and digital resources that will be required.
Complexity drives the need for information management
The brain is more complicated than any other part of the human body. To understand the brain, scientists must measure and analyze the rapid changes in neuronal activity that occur throughout the brain 's many intricate neural networks and pathways. The scope of such an endeavor is daunting. Neuroscience research proceeds along a multilevel hierarchy, from behavior and emotion to molecular interactions and genetic expression. At each level, researchers use numerous techniques that are specifically designed to collect information appropriate to that level. But information from one level may, and most often does, have important implications for research and knowledge at other levels. Therefore, new methods for assembling and integrating the pieces of the “brain puzzle” can be as important as the individual discoveries themselves.
The goal of information management, which is distinct from simply acquiring data, is to realize the greatest possible benefit from the data that already exist. The field of neuroscience research has two major information management needs: (1) databases, which relate diverse data types systematically and efficiently, and (2) visualization of structures in three dimensions to capture the discoverable architecture of the brain and relate this architecture to brain functioning.
Examples of the Value of Integrating Knowledge to Solve Problems
It is useful to examine how the diverse neural hierarchy relates to real biological processes and human disease. The examples in this section provide a glimpse of the breadth of neuroscience research and suggest how that research, coupled with better information management, can have positive effects on human health.
Vision: How we see affects how we think
Interpretation, complex perception, and even the appreciation of beauty often begins with visual information. The neural basis of vision involves billions of neurons and more than 300 separate but interrelated pathways in the cerebral cortex alone. As in all neural systems, visual functioning relies on the coordinated activity of neurons that communicate with each other, employing hundreds of different molecules for the generation of specific electrical and chemical signals. Many of these molecules are arranged in distinctive patterns and located in specific brain regions. The combination of complex physiological processes, anatomical pathways, and molecular and chemical interactions creates a system that functions extremely well but that is exceedingly difficult to decipher. Notwithstanding all of the information currently available, the fundamental nature of visual perception remains a mystery. To understand it, we need more experimental data and new ways of assembling the diverse kinds of related data into an integrated whole.
Among the resources necessary to understand the basis of pattern recognition and to gain insight into existing control and regulatory mechanisms are methods for overlaying anatomical maps with chemical and physiological maps. Computer simulations would also be of great value for understanding the signal processing of neural sets and networks. (Such simulations could also be used outside of neuroscience to construct new kinds of sensors and signal processors for robots and other automated devices.) Enhancing research into vision with these electronic tools promises to help ameliorate visual deficits caused by injury and to increase the range of treatments available for a number of diseases, including glaucoma, diabetic retinopathy, and inherited retinal degeneration, as well as blindness from other causes.
Substance abuse: The search for the biology of self-destruction
A large fraction of the U.S. population uses substances that are injurious to health. These substances include legally approved drugs,
such as nicotine and alcohol, and illegal drugs, such as cocaine and heroin, that are accompanied by the societal burdens of violent crime and increased infant mortality, among others. The key to substance abuse lies in the brain. One of the clear successes of neuroscience has been the discovery that certain molecules on the surface of neurons, called receptors, specifically bind many drugs, including nicotine, heroin and other opiates, and even benzodiazepines, such as Valium ®. This finding has led to the search for synthetic compounds that can block the actions of an injurious drug, yet still satisfy a person 's craving for that drug. Neuroscientists have also begun to pay close attention to those areas of the brain that mediate not only the normal pleasurable experiences of eating and socialization but the motivating, often pleasurable, drug effects that can lead to psychological addiction. The complexity of these brain areas, however, greatly complicates the investigations.
Unlike the visual and motor systems of the brain, the so-called reward system of brain nuclei and cortical areas has not been clearly defined. Two kinds of computerized databases could help researchers construct a framework for this system. One would depict the distribution patterns of important receptors and compare these patterns to the known anatomical and neurochemical circuitry of the brain. The other would contain information specific to those areas of the brain known to be involved in addictive processes. These resources would allow investigators faced with integrating data on anatomy, neurochemistry, pharmacology, and behavior to benefit from the research of other subspecialties in the field.
Pain: Sometimes a warning, sometimes a curse
Pain is a ubiquitous reality of life, and we need it to a certain degree to recognize sprained ankles, overstressed back muscles, kidney stones, infections, and many other problems. Yet for tens of thousands of people who suffer from abnormal or pathological pain, any benefit from pain is hard to identify. For these individuals, pain is an intractable barrier to a happy, productive life. The neural basis of pain involves almost every region of the brain, spinal cord, and peripheral nerves. Often, pain affects other body systems, such as the immune and endocrine systems. The complexity of pain perception helps to explain the variety in the pain people experience and the treatments that have been developed. Much has been learned in recent years about the basic neural mechanisms of pain, but substantial gaps in knowledge remain. For example, we still do not understand pathological pain—the severe pain often experienced in an am-
putated body part (despite its absence, so-called phantom limb pain) or the terrible pain that can occur following seemingly minor injuries or after surgical procedures.
There is a pressing need to transfer the information gained through basic research about pain to clinical practice. At present, widely divergent strategies are used to treat pain and to use it diagnostically. A broad range of professionals are often involved; thus, the responsibility for pain management may shift as a patient is moved from the operating room, to recovery, to a postsurgery ward. A database that related clinical observations to an integrated picture of relevant basic scientific data would be of great value in pain management, especially if it were combined with a repository of treatment and diagnostic strategies and their documented outcomes. Indeed, better ways to integrate data from basic pain research could contribute significantly to the alleviation of a major cause of human suffering.
Schizophrenia: Broken minds, shattered dreams
The symptoms of schizophrenia often emerge in adolescence or early adulthood, just as young people are beginning to plan their futures. In this country the lives of more than 2 million people have been devastated by schizophrenia and its various manifestations, which include hallucinations, delusions, blunted emotions, cognitive deficits, and an inability to maintain meaningful relationships. Neuroscience has focused on three major research areas in its search for understanding: neurochemical abnormalities, structural and functional brain abnormalities, and potential genetic and environmental causes.
It has long been thought that the effects of the neurotransmitter dopamine are greater than normal in the brains of those with schizophrenia. Not all of the symptoms of schizophrenia are alleviated by depression of dopamine activity, however, and evidence is mounting that other neurotransmitters are involved. The structural abnormalities identified thus far in some individuals are enlarged cerebral ventricles and a thinning of the cerebral cortex in certain areas. Although there seems to be a concomitant decrease in functioning in the frontal cortex, neuroscientists do not know how these structural and functional abnormalities contribute to the causes or symptoms of schizophrenia.
Advances such as gene mapping and positron emission tomography (PET) scanning may lead to increased understanding of the neurobiology of schizophrenia, but greater integration of the available information is needed. The kinds of digital resources that would be of most use are maps of the distribution of dopamine, dopamine receptors,
and other neuroregulators in the brain, databases of the existing brain imaging data with detailed descriptions of each patient's history and specific constellation of symptoms, and databases containing information about the genes that are most likely to confer susceptibility to schizophrenia. For those whose ravaged lives testify to the burden of this disease, and for their families, needed answers are long overdue.
The Growth of Neuroscience
Neuroscience research has grown in response to critical problems
The preceding examples reflect only a small portion of the overall cost of mental and neurological diseases. These diseases, combined with drug abuse, constitute an immense financial burden to our population every year in direct care expenses and lost wages. Nearly 23 million Americans suffer from head and spinal cord injuries, hearing and speech disorders, or infectious diseases of the nervous system. More than 3.5 million people suffer from Alzheimer's, Huntington's, or Parkinson's disease, or from other degenerative disorders, including multiple sclerosis and amyotrophic lateral sclerosis. More than 60 million people suffer from mental illnesses, including schizophrenia and depression, and more than 20 million abuse alcohol or drugs. Each of these problems clamors for resolution.
In response, neuroscience has made steady progress in a number of areas. Researchers have applied many new technologies, from the first oscilloscope to modern computer graphics, to the study of the brain. These technologies, combined with insightful, painstaking research, have led to the important breakthroughs witnessed in the past two or three decades and have enlarged considerably our understanding of the biological basis of disease. In addition, many talented young scientists have entered the field of neuroscience: the membership of the Society for Neuroscience has risen from 500 in 1969 to more than 17,500 in 1990. Because we now possess vast amounts of data and thousands of bright, dedicated scientists, the opportunities for successfully addressing the remaining questions about the brain have never been more promising.
Neuroscience is a national priority
Another source of the existing opportunities in neuroscience is the high priority the United States places on research aimed at alleviating mental and neurological disorders. Research support has come from various government bodies, including the National Institutes of
Health (NIH), the Alcohol, Drug Abuse, and Mental Health Administration (ADAMHA), the Department of Veterans Affairs, the National Science Foundation, and other government agencies; private agencies and foundations, including the Howard Hughes Medical Institute, the MacArthur Foundation, and the Pew Charitable Trusts, have also provided funds. It is estimated that these groups invest more than $1.5 billion annually in neuroscience research, the majority of which comes from NIH and ADAMHA. Yet however impressive the total, this investment represents a very small fraction of the overall societal costs of neurological and mental diseases. In times of competing needs and fiscal constraints, it is important to find ways to derive the most value from the investments that are made. The committee believes that a Brain Mapping Initiative, by enhancing the process of discovery and the communication of new insights in neuroscience, can help to maximize the benefits gained from the present investment of national resources.
Computer and Information Technology in Biomedical and Neuroscience Research
In the committee's opinion, a complex of electronic resources that will enhance neuroscience research is an attainable goal. Current trends in computer and information sciences clearly point to an unprecedented opportunity to incorporate technologies that will enable neuroscientists to expand their use of valuable, hard-won data and to communicate these data more effectively to other scientists. In addition, the sheer mass of neuroscience information accumulated to date, and the accelerating rate at which new results are being obtained and reported, are becoming major driving forces for the kind of organization, structure, and accessibility that electronic and digital resources can provide. Increasing sophistication in computer technology, coupled with decreasing costs, holds out the promise of enhanced research capabilities for many fields of scientific endeavor—but in particular, for neuroscience, owing to its inherently visual, hierarchical nature.
Three areas of computer science are especially important for biomedical research: computer graphics, database technology, and electronic networking. The use of computer graphics, so pervasive in such fields as earth mapping and space sciences, has only recently emerged as a resource in biomedical research, a direct result of the rapidly decreasing costs of computer memory capacities. One of the most successful applications of computer graphics has been the modeling of molecular structures using data derived from x-ray crystallography. No longer are scientists confined to trying to visualize a dyna-
mic molecule from a ball-and-stick depiction; computer models can be rotated and manipulated to simulate the molecule's actual functioning. These models will soon help to predict which drugs stop certain viruses, how genes are turned on and off, and how two molecules may interact with one another. In neuroscience, the use of computer graphics has led to visualization of the activity of the human brain through PET scanning and magnetic resonance imaging (MRI). Such graphic depictions have proved to be so useful that greater and greater attention is being paid to the concept of visualization computing in biomedical sciences. One example of this increased attention is the greater priority given to biomedical computing by leading universities and government laboratories.
Databases, the second computer science area that is important to biomedical research, allow digitized data to be stored and organized in a way that makes the data easily accessible. Databases can be word, number, image, or sound oriented, and can be either public or private. The use of databases, including databases of biomedical information, has increased substantially over the past 15 years. The leader in the development of biomedical databases has been the National Library of Medicine (NLM), which currently maintains a number of on-line systems, including bibliographic databases of scientific literature, registries of chemicals and their toxic effects, and medical information related to cancer and other diseases. Other prominent scientific databases are the protein sequence and genome databases developed by a variety of institutions and individuals, which are being used in the efforts to map the human genome. The need for databases for neuroscience information is great, and the systems are beginning to emerge. Neuroscience database developers can learn much from the experiences of the NLM and protein and genome database originators.
The purpose of computer networks is to help create a communication environment that is as free of barriers as possible. In science the communication of data and ideas is as important to the growth of knowledge as the data themselves; consequently, the use of computer networks holds great promise for neuroscience. Since 1969, networks connecting research laboratories and universities have grown so rapidly that today more than 5,000 interconnected networks in 35 countries connect more than 300,000 computers. Further efforts, such as the National Research and Education Network (NREN), are under way to upgrade the transmission rates of U.S. scientific and educational computer networks, including the National Science Foundation's NSFNET, to permit them to handle more data and data of greater complexity (e.g., image data, which currently cannot be efficiently transmitted
through computer networks). If computer networks are to be a viable method for communication of neuroscience data, the planned increase in transmission rates is a necessity. Also needed are strategies for standard data formats to facilitate communication of digitized information among investigators.
Building Consensus, Identifying Needs
The committee sponsored a number of consultative activities to seek out the opinions and advice of the neuroscience community at large. These activities included written requests for opinions (published in various journals and solicited directly from the leadership of the Society for Neuroscience); formation of four task forces; and sponsorship of three symposia and open hearings, which were held in Washington, D.C., Chicago, and San Francisco. In addition, the committee commissioned two background papers: one traced the development and current uses of genome and related scientific databases, and the other investigated the Defense Mapping Agency's experience in converting cartographic data to digital formats. Participants in these activities reflected a wide range of backgrounds and expertise including library management, scientific database administration and design, computer science, and neuroscience research. Additionally, participants came from academic departments, government laboratories, and private industry.
The complexity of neuroscience dictates that the scope of data included in any group of electronic and digital resources must eventually be quite broad. The majority of the individuals providing input to the committee's deliberations envisioned the proposed complex of resources as necessarily containing more kinds of information than are already or can be contained in library reference materials and published journals. Participants also endorsed using the neuroanatomy of the brain as the principal organizing arrangement for the resources. Information on neuroanatomy could function as the skeleton or frame-work on which data from multiple levels of the brain's hierarchy, including information on functions, could be displayed. The information would include neural pathways, cell types, neurochemistry and identification of neurotransmitters, protein and gene sequences and gene mapping, receptor types, electrophysiological responses, and data regarding behavioral relationships, such as memory.
To realize such a database, the computerized resources must include a range of capabilities. Some of the features participants deemed necessary were the ability to
transform data into three-dimensional images,
browse through various kinds of data,
extract arbitrarily defined subsets of data, and
compare different brain images with each other by precise overlaying or co-registration of the images.
There was overwhelming consensus among all participants that a single National Neural Circuitry Database was not a workable plan. Rather, a complex of electronic and digital resources should be developed to include separate databases with varied levels of accessibility. This complex should be composed of reference databases, data banks, informal databases, national and international registries, research collaboration databases, and specialty databases.
To implement this complex of resources, however, significant advances will be required. The committee identified three areas deserving of attention: databases, networks, and imaging technologies. Only imaging is sufficiently advanced to be immediately applied to neuroscience research. Database management technology presents one of the most difficult barriers to implementation. Currently available database technology cannot handle images easily, although improved capabilities are under development. Database developers should consider incorporating these advances as soon as they become available. In addition, present mechanisms to interlink different databases will need to be improved substantially for application to neuroscience. Human–computer interfaces also require improvement so that they are uncomplicated for users yet powerful enough to enable the user to extract needed information. Underlying both database and user interface design issues is the challenge of developing software that allows the data to be accessible and usable. Finally, as noted earlier, computer network upgrading will be necessary for the transmission of complex image data.
The final technical topic covered by the task forces was the development of standards for the exchange of data. Four areas of technical concern were addressed:
Data representation and standard data formats are needed for textual and numerical data and for the generation of images and graphics.
Mechanisms are needed for conveying new algorithms for a variety of applications.
Standard human–computer interface packages should be explored to reduce barriers to the actual use of electronic resources.
Standard communication protocols are needed to accommodate the dynamic range of data accessibility required for research-oriented databases.
In addition to the technological changes that will be required to implement the proposed resources, the task forces identified sociological patterns and issues specific to the scientific community that must also be addressed. The development of standard data formats presents important sociological questions. Participants expressed strong views that standards should evolve out of the needs and perspectives of users and not be imposed from outside. A careful balance must be sought between the need for technical standards to facilitate communication and the barriers that can result from overly strict standard formats.
Data sharing is another major sociological issue. In some neuroscience specialties, sharing of unpublished and “raw” data is commonplace, whereas in other specialties such openness is rare. Although the committee places no value judgment on either end of the data-sharing continuum —restrictive to open—it recognizes the existence of a continuum and concludes that potential data-sharing mechanisms must be carefully considered as part of any initiative. For example, priority should be given to formulating methods to ensure that proper credit is assigned for those contributing data and that the privacy of human subjects is protected. In addition, the level of certification (peer-reviewed, verified, unverified, etc.) of all data must be clearly stated, although preliminary data probably would be included in special databases. Clear identification of different levels of certainty, including unambiguous labeling of preliminary data, will also need to be incorporated into the various kinds of databases; for major resources, editorial boards may be a good mechanism to aid in this process. As in all sciences, replication of experimental results will continue to be important. The task forces also supported the concept that university tenure committees should consider certain types of data sharing, particularly of peer-reviewed data, as evidence of professional competence, comparable to journal publication and teaching evaluations. The committee was further encouraged to examine the policies developed recently by some journals devoted to gene mapping and gene and protein sequencing for deposit of data into established databases. Other areas of sociological concern were how to ensure that electronic resources would be accessible to more than a few well-funded laboratories and how to overcome resistance to the integration of technology into the way people work. An additional concern of many participants was the changing work force that would result from greater use of technology and the need for computer specialists with expertise in neurobiology as well as neurobiologists with expertise in computer science.
To overcome these technological and sociological impediments, the
task forces in particular strongly recommended that the committee call for the establishment of pilot projects. The primary goal of these projects would be to provide a desperately needed base of experience from which to establish a family of usable computerized resources. The task forces suggested a two-phase effort, with the pilot projects as a necessary first step, followed by a more global incorporation of computerized resources into the neuroscience research enterprise.
A key area of consensus was that pilot projects would require coordination and that oversight and evaluation mechanisms were crucial to the eventual implementation of the resource complex. Structures suggested for coordination of the pilot projects included advisory boards, host institutions, and formal meetings.
A final area of discussion was funding. Despite general support for the proposed effort, many participants expressed concern about how and where funds might be secured. There was considerable dialogue about current constraints on biomedical research support, and virtually all participants expressed the opinion that funds for this project should be obtained only through additional appropriations.
The Brain Mapping Initiative: Committee Recommendations
After considering the opinions expressed and the input received through its activities and during its own deliberations, the committee concluded that an environment of opportunity now exists to enhance neuroscience research by a more global incorporation of computer and information technologies. Considering past experiences in other scientific disciplines, however, it is also apparent that to ensure the greatest benefit from these technologies, they should be incorporated with care and with a clear vision of the intended goal. Neuroscience is diverse in its methodology and levels of inquiry. To achieve true understanding, all of the available information must be coordinated into an integrated, meaningful picture. At present, the detailed information being generated at every level of neural organization is difficult to grasp and integrate. Even searches for information regarding relatively specific neural levels or processes are hindered because the information is widely scattered through scores of different journals, review papers, symposia summaries, and books. Added to these difficulties are the unique visual requirements of neuroscience. The mass of information is steadily expanding because researchers can now generate two- and three-dimensional graphic images relatively quickly and easily. The increasing use of computers to collect the various kinds of neuroscience data needed by researchers and the development currently under way of the technology to link these
computerized research environments makes this an opportune time to begin a Brain Mapping Initiative. Therefore,
the committee recommends that the Brain Mapping Initiative be established with the long-term objective of developing three-dimensional computerized maps and models of the structure, functions, connectivity, pharmacology, and molecular biology of human, rat, and monkey1brains across developmental stages and reflecting both normal and disease states.
The Brain Mapping Initiative is meant to include features that were originally proposed for a National Neural Circuitry Database, although the new initiative, unlike the earlier project, would not be a single-entity database. Rather, the Brain Mapping Initiative would lead to the establishment of a complex of interrelated, integrated databases accessible from individual laboratories. The committee is aware of the broad scale of such an undertaking and that the successful implementation of this program will require transformation of the way information is acquired, communicated, and analyzed by neuroscientists. Therefore, the committee envisions the initiative as a longterm endeavor to be accomplished in two phases. Phase 1 would comprise an organized initiation of seed or pilot projects with the overall goals of gaining experience in the incorporation of the required technologies and applying that experience to long-range planning for phase 2. Phase 2 would involve construction of the maps and models necessary to provide a complex of digital and electronic resources to enhance neuroscience research. To begin phase 1,
the committee recommends the establishment of pilot projects or consortia. These projects should be peer-reviewed by neuroscientists and computer scientists; they should also be investigator initiated, involve geographically dispersed laboratories, and include neuroscientists with varied levels of computer experience. The projects should develop common formats for the exchange of data and focus on different types of computer data representations (geometric, structural, image, and free text). Selection of projects should be on the basis of research quality and value to the evolution of a complex of electronic resources for mapping the brain.
The pilot projects the committee recommends should be a coordinated program of separate efforts but with certain common goals. In the committee's vision, the pilot program would consist of several
groups of investigators, each working on specific neuroscience topics, with the primary goals of mapping the brain's anatomy, chemistry, and functions, and forging pathways for the integration of computer and information technology into the overall neuroscience research effort. Consortia could be organized among geographically distinct institutions or as centers housed within a single institution. (If the centers approach is chosen, special attention should be given to involving investigators from geographically distant institutions as users of the technologies developed.) Research topics for the phase 1 projects should reflect the vertical hierarchy of the brain from behavior and emotion to molecular biological and genetic mechanisms, as well as the horizontal range of inquiry including the anatomy, physiology, neurochemistry, and molecular and developmental biology of specific brain systems.
In terms of technical developments, the overall goals of the pilot project program would be the following:
Develop electronic data collection and storage methods for data types at each level of the neural hierarchy.
Identify the kinds of data, level of resolution, and experimental information necessary to facilitate new insights and stimulate research.
Examine and evaluate the wide range of available capabilities that can increase use of the resources and enhance access to meaningful information.
Develop a variety of databases from formal, consensus databases to informal databases for research collaboration.
Develop and experiment with different software for translation across different computing environments, for user interfaces, for network transmission of images, for data searching, and for image generation and comparison.
Begin to develop standard data formats, nomenclature, and data collection schemes, and to evaluate the evolution of these standards.
Gain experience in data sharing and communication through electronic means, including networks and transportable media.
Communicate with others in the program to share and evaluate experiences and technological developments.
To ensure the greatest benefits possible from the phase 1 projects, the committee agreed to emphasize certain areas by formulating the recommendations below.
The committee recognizes that neuroscience efforts proceed internationally and recommends that an international registry of neuroscience databases and contacts be established so that appropriate linkages can be created in the future.
Such a registry, which should be available to users through computer networks, would help to identify currently available resources and provide a mechanism to coordinate the efforts of phase 1 investigators and investigators in other countries.
The committee recommends the establishment of an archive of public domain software, accessible through computer networks.
The committee expects that phase 1 projects will have special needs for novel software. Public domain software is available at little or no cost to anyone who wants to use it, and these programs should be explored. In addition, such an archive would encourage the formation of neuroscience “news groups,” or groups of users with similar interests, who could communicate by computer bulletin boards or electronic mail.
The committee recommends that an administrative structure be established to coordinate phase 1 activities. This Brain Map Advisory Panel (BMAP) should be composed of neuroscientists and computer and information scientists, with additional input from funding agency administrators. The panel would be responsible for the overall direction, evaluation, and coordination of consortia and for the development of necessary policies relating to establishment of a brain mapping effort. The committee also recommends that the Advisory Panel be responsible for consideration and development of editorial functions and policies relating to the ethical and sociological issues that will arise, including, but not limited to, correctness of information and quality control, intellectual property rights, rights to privacy, and freedom of information.
To develop the proper basis for a coordinated brain mapping effort, there must be communication among the consortia and some type of central oversight. The committee considered several mechanisms for providing such oversight and chose an advisory panel structure. The BMAP could also act as a clearinghouse for information and perform certain other functions:
examine the needs of the entire neuroscience community;
evaluate various aspects of database development including resource use, standards development, and effectiveness of incentives for data sharing;
gather information from the consortia on emerging trends in the computer industry; and
facilitate acceptance of the database among neuroscientists.
The BMAP should exercise oversight not in a top-down manner but as the result of communication between members of the consortia and the panel. This communication could occur through a variety of mechanisms including consortia representation on the panel and the sponsorship of regular meetings for consortia investigators. The panel should also coordinate establishment of the international registry of neuroscience databases and contacts and the archive of public domain software recommended by the committee. Finally, the panel should coordinate the Brain Mapping Initiative's interaction with funding agencies and with other, related scientific initiatives, and undertake the long-range planning of phase 2.
If electronic resources are to be accepted and utilized, scientists must be able to trust the accuracy of the information contained in the resource. One of the tasks of phase 1 investigators should be to begin to define mechanisms that ensure the appropriate use and labeling of different levels of data (from preliminary to peer-reviewed) and that allow for the deletion of information that becomes obsolete. In the committee's opinion, one of the best ways to achieve such goals is to develop edited archives and databases.
The committee recommends that the phase 1 and 2 projects of the Brain Mapping Initiative maintain a close relationship with the gene mapping and sequencing community and the Human Genome Project, and with other scientific computing efforts, including network initiatives such as NSFNET and the proposed National Research and Education Network. As part of these efforts, the committee further recommends that linkages be established with protein sequence and genome databases to enhance access to information about brain-specific genes.
The committee believes that the importance of gene mapping and sequencing to neuroscience should not be underestimated. In addition, much can be learned from existing database initiatives in this area, including the array of public and private databases that support the Human Genome Project and other established scientific networking efforts. Interaction with the ongoing scientific computing efforts in the global change research, astrophysics, and earth mapping communities would also be desirable.
The committee recommends that federal funding agencies develop requests for applications and/or cooperative agreements to support the formation of consortia and the activities of the Brain Map Advisory Panel. Limited use of contract mechanisms should also be considered when appropriate to the overall goals of the initiative.
The federal government uses several mechanisms to provide federal funds for research. One is contracts in which the government funds projects that its agencies propose, to be completed by an outside investigator according to a contract written by the government. Another mechanism, typical of NIH, ADAMHA, and NSF is grants to investigators or groups of investigators for research projects that are proposed, accomplished, and supervised by the investigators themselves. A third method, the cooperative agreement used by NSF, combines aspects of both grants and contracts. In the committee's opinion, the proposed Brain Mapping Initiative favors the use of grants because the development of usable resources should be intricately combined with the research itself. This kind of development is best carried out by scientists actively involved in neuroscience research.
The committee recommends that phases 1 and 2 of the Brain Mapping Initiative be international in scope and that they be funded by multiple sources in a coordinated fashion. The structure for administering the funding should ensure program stability and effectiveness. Possible funding structures include the identification of a lead agency or institute, or the establishment of formal administrative structures among two or more agencies.
The sources of neuroscience funding include the three institutes of ADAMHA and almost every institute of the NIH, as well as many other governmental agencies. The proposed organization of the Brain Mapping Initiative, especially the inclusion of an advisory panel and computer scientists, will not fit well into the usual funding structures administered by the NIH Division of Research Grants. To be successful, the proposed phase 1 projects require the involvement of multiple components of the federal biomedical research complex, as well as communication and cooperation among appropriate agencies of the Public Health Service, the Departments of Defense and Energy, and private foundations that fund biomedical science.
The committee concludes that the expected benefits of the proposed Brain Mapping Initiative justify the investment of necessary resources and recommends the appropriation
of additional funding to support the establishment of phase 1 projects.
The committee is sensitive to the current fiscal constraints now being felt by the entire U.S. biomedical research effort and recognizes the view held by many scientists that large-scale projects pose a threat to the basic research enterprise. The committee believes, however, that the Brain Mapping Initiative is an important project that should be undertaken and that it could begin with an additional allocation of approximately $10 million annually, with an overall evaluation to occur at the end of five years. This investment represents only a small part (less than 1 percent) of the entire U.S. neuroscience research effort, which is estimated to be more than $1.5 billion annually. It is the committee's considered opinion that the success and probable benefits of the initiative proposed here depend on and justify the appropriation of this additional support.
It is clear that neuroscience stands at the threshold of a tremendous opportunity to unlock the mysteries of the brain and its functions. Securing the benefits inherent in this opportunity requires a concerted, interdisciplinary effort on the part of the many basic and clinical neuroscientists engaged in research worldwide. It is equally clear that this scientific enterprise is increasingly reliant on the use of sophisticated methodologies and computer technologies. The Brain Mapping Initiative proposed in this report will allow investigators studying the brain to view data in new ways, to communicate data to each other more efficiently, and to access data from any of the neuroscience subspecialties. So enabled, neuroscientists will be able to map the brain and its functions, thus realizing the full potential of the electronic resources now available to the scientific enterprise and ensuring that society will receive the greatest possible benefits from neuroscience research.
1. These species are intended as starting points. The committee also recognizes the need to include data from other, vertebrate and invertebrate, species.