Major Application Areas of Complex Systems Research

The following should provide a sense of the integral nature of complex systems to advances in nanotechnology, biomedicine, information technology, cognitive science, and social and global systems. A level of complexity is found in their convergence.

Nanotechnology

Development of functional systems based on nanotechnological control is a major challenge beyond the creation of single elements. Indeed, the success of nanotechnology in controlling small elements can synergize well with the study of complex systems. To understand the significance of complex systems for nanotechnology, it is helpful to consider the smallest class of biological machines, also considered the smallest complex systems — proteins (Fersht 1999). Proteins are a marvel of engineering for design and manufacture. They also have many useful qualities that are not common in artificial systems, including robustness and adaptability through selection. The process of manufacturing a protein is divided into two parts, the creation of the molecular chain and the collapse of this chain to the functional form of the protein. The first step is ideal from a manufacturing point of view, since it enables direct manufacture from the template (RNA), which is derived from the information archive (DNA), which contains encoded descriptions of the protein chain. However, the chain that is formed in manufacture is not the functional form. The protein chain "self-organizes" (sometimes with assistance from other proteins) into its functional (folded) form. By manufacturing proteins in a form that is not the functional form, key aspects of the manufacturing process can be simplified, standardized, and made efficient while allowing a large variety of functional machines described in a simple language. The replication of DNA provides a mechanism of creating many equivalent information archives (by exponential growth) that can be transcribed to create templates to manufacture proteins in a massively parallel way when mass production is necessary. All of these processes rely upon rapid molecular dynamics. While proteins are functionally robust in any particular function, their functions can also be changed or adapted by changing the archive, which "describes" their function, but in an indirect and non-obvious way. The rapid parallel process of creation of proteins allows adaptation of new machines through large-scale variation and selection.

A good example of this process is found in the immune system response (Perelson and Wiegel 1999; Noest 2000; Segel and Cohen 2001; Pierre et al. 1997). The immune system maintains a large number of different proteins that serve as antibodies that can attach themselves to harmful antigens. When there is an infection, the antigens that attach most effectively are replicated in large numbers, and they are also subjected to a process of accelerated evolution through mutation and selection that generates even better-suited antibodies. Since this is not the evolutionary process of organisms, it is, in a sense, an artificial evolutionary process optimized (engineered) for the purpose of creating well-adapted proteins (machines). Antibodies are released into the blood as free molecules, but they are also used as tools by cells that hold them attached to their membranes so that the cells can attach to, "grab hold of," antigens. Finally, proteins also form complexes, are part of membranes and biochemical networks, showing how larger functional structures can be built out of simple machines. An artificial analog of the immune system's use of evolutionary dynamics is the development of ribozymes by in vitro selection, now being used for drug design (Herschlag and Cech 1990; Beaudry and Joyce 1992; Szostak 1999).

Proteins and ribozymes illustrate the crossover of biology and nanotechnology. They also illustrate how complex systems concepts of self-organization, description, and evolution are important to nanotechnology. Nanotechnological design and manufacturing may take advantage of the system of manufacture of proteins or other approaches may be used. Either way, the key insights of how proteins work shows the importance of understanding various forms of description (DNA); self-reproduction of the manufacturing equipment (DNA replication by polymerase chain reaction or cell replication); rapid template-based manufacture (RNA transcription to an amino-acid chain); self-organization into functional form (protein folding); and evolutionary adaptation through replication (mutation of DNA and selection of protein function) and modular construction (protein complexes). Understanding complex systems concepts thus will enable the development of practical approaches to nanotechnological design and manufacture and to adaptation to functional requirements of nanotechnological constructs.

Biomedical Systems

At the current time, the most direct large-scale application of complex systems methods is to the study of biochemical networks (gene regulatory networks, metabolic networks) that reveal the functioning of cells and the possibilities of medical intervention (Service 1999; Normile 1999; Weng, Bhalla and Iyengar 1999). The general studies of network structure described above are complementary to detailed studies of the mechanisms and function of specific biochemical systems (von Dassow et al. 2001). High-throughput data acquisition in genomics and proteomics is providing the impetus for constructing functional descriptions of biological systems (Strausberg and Austin 1999). This, however, is only the surface of the necessary applications of complex systems approaches that are intrinsic to the modern effort to understand biological organisms, their relationships to each other, and their relationship to evolutionary history. The key to a wider perspective is recognizing that the large quantities of data that currently are being collected are being organized into databases that reflect the data acquisition process rather than the potential use of this information. Opportunities for progress will grow dramatically when the information is organized in a form that provides a description of systems and system functions. Since cellular and multicellular organisms, including the human being, are not simply biochemical soups, this description must capture the spatiotemporal dynamics of the system as well as the biochemical network and its dynamics. In the context of describing human physiology from the molecular scale, researchers at the Oak Ridge National Laboratory working towards this goal call it the Virtual Human Project (Appleton 2000). This term has also been used to describe static images of a particular person at a particular time (NLM 2002).

The program of study of complex systems in biology requires not only the study of a particular organism (the human being) or a limited set of model organisms, as has been done in the context of genomics until now. The problem is to develop comparative studies of systems, understanding the variety that exists within a particular type of organism (e.g., among human beings) and the variety that exists across types of organisms. Ultimately, the purpose is to develop an understanding or description of the patterns of biological systems today as well as throughout the evolutionary process. The objective of understanding variety and evolution requires us to understand not just any particular biochemical system, but the space of possible biochemical systems filtered to the space of those that are found today, their general properties, their specific mechanisms, how these general properties carry across organisms, and how they are modified for different contexts. Moreover, new approaches that consider biological organisms through the relationship of structure and function, and through information flow are necessary to this understanding.

Increasing knowledge about biological systems is providing us with engineering opportunities and hazards. The great promise of our biotechnology is unrealizable without a better understanding of the systematic implications of interventions that we can do today. The frequent appearance of biotechnology in the popular press through objections to genetic engineering and cloning reveals the great specific knowledge and the limited systemic knowledge of these systems. The example of corn genetically modified for feed and its subsequent appearance in corn eaten by human beings (Quist and Chapela 2001) reveals the limited knowledge we have of indirect effects in biological systems. This is not a call to limit our efforts, simply to focus on approaches that emphasize the roles of indirect effects and explore their implications scientifically. Without such studies, not only are we shooting in the dark, but in addition we will be at the mercy of popular viewpoints.

Completion of the virtual human project would be a major advance toward creating models for medical intervention. Such models are necessary when it is impossible to test multidrug therapies or specialized therapies based upon individual genetic differences. Intervention in complex biological systems is an intricate problem. The narrow bridge that currently exists between medical double blind experiments and the large space of possible medical interventions can be greatly broadened through systemic models that reveal the functioning of cellular systems and their relationship to cellular function. While today individual medical drugs are tested statistically, the main fruit of models will be

• to reveal the relationship between the function of different chemicals and the possibility of multiple different types of interventions that can achieve similar outcomes

• the possibility of discovering small variations in treatment that can affect the system differently

• possibly most importantly, to reveal the role of variations between human beings in the difference of response to medical treatment

A key aspect of all of these is the development of complex systems representations of biological function that reveal the interdependence of biological system and function.

Indeed, the rapid development of medical technologies and the expectation of even more dramatic changes should provide an opportunity for, even require, a change in the culture of medical practice. Key to these changes should be understanding of the dynamic state of health. Conventional homeostatic perspectives on health are being modified to homeodynamic perspectives (Goldberger, Rigney, and West 1990; Lipsitz and Goldberger 1992). What is needed is a better understanding of the functional capabilities of a healthy individual to respond to changes in the external and internal environment for self-repair or -regulation. This is essential to enhance the individual's capability of maintaining his or her own health. For example, while physical decline is a problem associated with old age, it is known that repair and regulatory mechanisms begin to slow down earlier, e.g., in the upper 30s, when professional athletes typically end their careers. By studying the dynamic response of an individual and changes over his/her life cycle, it should be possible to understand these early aspects of aging and to develop interventions that maintain a higher standard of health. More generally, understanding of the network of regulatory and repair mechanisms should provide a better mechanism for dynamic monitoring — with biomedical sensors and imaging — health and disease and the impact of medical interventions. This would provide key information about the effectiveness of interventions for each individual, enabling feedback into the treatment process that can greatly enhance its reliability.

Information Systems

Various concepts have been advanced over the years for the importance of computers in performing large-scale computations or in replacing human beings through artificial intelligence. Today, the most apparent role of computers is as personal assistants and as communication devices and information archives for the socioeconomic network of human beings. The system of human beings and the Internet has become an integrated whole leading to a more intimately linked system. Less visibly, embedded computer systems are performing various specific functions in information processing for industrial age devices like cars. The functioning of the Internet and the possibility of future networking of embedded systems reflects the properties of the network as well as the properties of the complex demands upon it. While the Internet has some features that are designed, others are self-organizing, and the dynamic behaviors of the Internet reflect problems that may be better solved by using more concepts from complex systems that relate to interacting systems adapting in complex environments rather than conventional engineering design approaches.

Information systems that are being planned for business, government, military, medical, and other functions are currently in a schizophrenic state where it is not clear whether distributed intranets or integrated centralized databases will best suit function. While complex systems approaches generally suggest that creating centralized databases is often a poor choice in the context of complex function, the specific contexts and degree to which centralization is useful must be understood more carefully in terms of their functions and capabilities, both now and in the future (Bar-Yam 2001).

A major current priority is enabling computers to automatically configure themselves and carry out maintenance without human intervention (Horn 2001). Currently, computer networks are manually configured, and often the role of various choices in configuring them are not clear, especially for the performance of networks. Indeed, evidence indicates that network system performance can be changed dramatically using settings that are not recognized by the users or system administrators until chance brings them to their attention. The idea of developing more automatic processes is a small part of the more general perspective of developing adaptive information systems. This extends the concept of self-configuring and self-maintenance to endowing computer-based information systems with the ability to function effectively in diverse and variable environments. In order for this functioning to take place, information systems must, themselves, be able to recognize patterns of behavior in the demands upon them and in their own activity. This is a clear direction for development of both computer networks and embedded systems.

Development of adaptive information systems in networks involves the appearance of software agents. Such agents range from computer viruses to search engines and may have communication and functional capabilities that allow social interactions between them. In the virtual world, complex systems perspectives are imperative in considering such societies of agents. As only one example, the analogy of software agents to viruses and worms has also led to an immune system perspective in the design of adaptive responses (Forrest, Hofmeyr, and Somayaji 1997; Kephart et al. 1997).

While the information system as a system is an important application of complex systems concepts, complex systems concepts also are relevant to considering the problem of developing information systems as effective repositories of information for human use. This involves two aspects, the first of which is the development of repositories that contain descriptions of complex systems that human beings would like to understand. The example of biological databases in the previous section is only one example. Other examples are socio-economic systems, global systems, and astrophysical systems. In each case, the key issue is to gain an understanding of how such complex systems can be effectively represented. The second aspect of designing such information repositories is the recognition of human factors in the development of human-computer interfaces (Norman and Draper 1986; Nielsen 1993; Hutchins 1995). This is important in developing all aspects of computer-based information systems, which are used by human beings and designed explicitly or implicitly to serve human beings.

More broadly, the networked information system that is being developed serves as part of the human socio-economic-technological system. Various parts of this system, which includes human beings and information systems, as well as the system as a whole, are functional systems. The development and design of this self-organizing system and the role of science and technology is a clear area of application of complex systems understanding and methods. Since this is a functional system based upon a large amount of information, among the key questions is how should the system be organized when action and information are entangled.

Cognitive Systems

The decade of the 1990s was declared by President George Bush, senior (1990), the "decade of the brain," based, in part, on optimism that new experimental techniques such as Positron Emission Tomography (PET) imaging would provide a wealth of insights into the mechanisms of brain function. However, a comparison of the current experimental observations of cognitive processes with those of biochemical processes of gene expression patterns reveals the limitations that are still present in these observational techniques in studying the complex function of the brain. Indeed, it is reasonable to argue that the activity of neurons of a human being and their functional assignment is no less complex than the expression of genes of a single human cell.

Current experiments on gene expression patterns allow the possibility of knocking out individual genes to investigate the effect of each gene on the expression pattern of all other genes measured individually. The analogous capability in the context of cognitive function would be to incapacitate an individual neuron and investigate the effect on the firing patterns of all other neurons individually. Instead, neural studies are based upon sensory stimulation and measures of the average activity of large regions of cells. In gene expression studies, many cells are used with the same genome and a controlled history through replication, and averages are taken of the behavior of these cells. In contrast, in neural studies averages are often taken of the activity patterns of many individuals with distinct genetic and environmental backgrounds. The analogous biochemical experiment would be to average behavior of many cells of different types from a human body (muscle, bone, nerve, red blood cell, etc.) and different individuals, to obtain a single conclusion about the functional role of the genes.

The more precise and larger quantities of genome data have revealed the difficulties in understanding genomic function and the realization that gene function must be understood through models of genetic networks (Fuhrman et al. 1998). This is to be contrasted with the conclusions of cognitive studies that investigate the aggregate response of many individuals to large-scale sensory stimuli and infer functional assignments. Moreover, these functional assignments often have limited independently verifiable or falsifiable implications. More generally, a complex systems perspective suggests that it is necessary to recognize the limitations of the assignment of function to individual components ranging from molecules to subdivisions of the brain; the limitations of narrow perspectives on the role of environmental and contextual effects that consider functioning to be independent of effects other than the experimental stimulus; and the limitations of expectations that human differences are small and therefore that averaged observations have meaning in describing human function.

The problem of understanding brain and mind can be understood quite generally through the role of relationships between patterns in the world and patterns of neuronal activity and synaptic change. While the physical and biological structure of the system is the brain, the properties of the patterns identify the psychofunctioning of the mind. The relationship of external and internal patterns are further augmented by relationships between patterns within the brain. The functional role of patterns is achieved through the ability of internal patterns to represent both concrete and abstract entities and processes, ranging from the process of sensory-motor response to internal dialog. This complex nonlinear dynamic system has a great richness of valid statements that can be made about it, but identifying an integrated understanding of the brain/mind system cannot be captured by perspectives that limit their approach through the particular methodologies of the researchers involved. Indeed, the potential contributions of the diverse approaches to studies of brain and mind have been limited by the internal dynamics of the many-factioned scientific and engineering approaches.

The study of complex systems aspects of cognitive systems, including the description of patterns in the world and patterns in mind, the construction of descriptions of complex systems, and the limitations on information processing that are possible for complex systems, are relevant to the application of cognitive studies to the understanding of human factors in man-machine systems (Norman and Draper 1986; Nielsen 1993; Hutchins 1995) and more generally to the design of systems that include both human beings and computer-based information systems as functional systems. Such hybrid systems, mentioned previously in the section on information technology, reflect the importance of the converging technology approach.

The opportunity for progress in understanding the function of the networked, distributed neuro-physiological system also opens the possibility of greater understanding of development, learning, and aging (NIMH n.d.; Stern and Carstensen 2000; Mandell and Schlesinger 1990; Davidson, Teicher, and Bar-Yam 1997). While the current policy of education reform is using a uniform measure of accomplishment and development through standardized testing, it is clear that more effective measures must be based on a better understanding of cognitive development and individual differences. The importance of gaining such knowledge is high because evaluation of the effectiveness of new approaches to education typically requires a generation to see the impact of large-scale educational changes on society. The positive or negative effects of finer-scale changes appear to be largely inaccessible to current research. Thus, we see the direct connection between complex systems approaches to cognitive science and societal policy in addressing the key challenge of the education system. This in turn is linked to solution of many other complex societal problems, including poverty, drugs and crime, and also to effective functioning of our complex economic system requiring individuals with diverse and highly specialized capabilities.

Studies of the process of aging are also revealing the key role of environment on the retention of effective cognitive function (Stern and Carstensen 2000; Mandell and Schlesinger 1990; Davidson, Teicher, and Bar-Yam 1997). The notion of "use it or lose it," similar to the role of muscular exercise, suggests that unused capabilities are lost more rapidly than used ones. While this is clearly a simplification, since losses are not uniform across all types of capabilities and overuse can also cause deterioration, it is a helpful guideline that must be expanded upon in future research. This suggests that research should focus on the effects of the physical and social environments for the elderly and the challenges that they are presented with.

We can unify and summarize the complex systems discussion of the cognitive role of the environment for children, adults, and the elderly by noting that the complexity of the environment and the individual must be matched for effective functioning. If the environment is too complex, confusion and failure result; if the environment is too simple, deterioration of functional capability results. One approach to visualizing this process is to consider that the internal physical parts and patterns of activity are undergoing evolutionary selection dictated by the patterns of activity that result from environmental stimulation. This evolutionary approach also is relevant to the recognition that individual differences are analogous to different ecological niches. A more detailed research effort would not only consider the role of complexity but also the effect of specific patterns of environment and patterns of internal functioning, individual differences in child development, aging, adult functioning in teams, and hybrid human-computer systems.

Social Systems and Societal Challenges

While social systems are highly complex, there are still relatively simple collective behaviors that are not well understood. These include commercial fads, market cycles and panics, bubbles and busts. Understanding the fluctuating dynamics and predictability of markets continues to be a major challenge. It is important to emphasize that complex systems studies are not necessarily about predicting the market, but about understanding its predictability or lack thereof.

More generally, there are many complex social challenges associated with complex social systems ranging from military challenges to school and education system failures, healthcare errors, and problems with quality of service. Moreover, other major challenges remain in our inability to address fundamental social ills such as poverty (in both developed and undeveloped countries), drug use, and crime. To clarify some aspects of social systems from a complex systems perspective, it is helpful to focus on one of these, and the current military context is a convenient focal point.

Wars are major challenges to our national abilities. The current war on terrorism is no exception. In dealing with this challenge, our leadership, including the president and the military, has recognized that this conflict is highly complex. Instead of just sending in tens to hundreds of thousands of troops, as was done in the Gulf War, there is a strategy of using small teams of special forces to gain intelligence and lay the groundwork for carefully targeted, limited and necessary force.

A large-scale challenge can be met by many individuals doing the same thing at the same time, or repeating the same action, similar to a large military force. In contrast, a complex challenge must be met by many individuals doing many different things at different times. Each action has to directly match the local task that must be done. The jungles of Vietnam and the mountains of Afghanistan, reported to have high mountains and deep narrow valleys, are case studies in complex terrains. War is complex when targets are hidden, not only in the terrain but also among people — bystanders or friends. It is also complex when the enemy can itself do many different things, when the targets are diverse, the actions that must be taken are specific, and the difference between right and wrong action is subtle.

While we are still focused on the war on terrorism, it seems worthwhile to transfer the lessons learned from different kinds of military conflicts to other areas where we are trying to solve major problems. Over the past 20 years, the notion of war has been used to describe the War on Poverty, the War on Drugs, and other national challenges. These were called wars because they were believed to be challenges requiring the large force of old-style wars. They are not. They are complex challenges that require detailed intelligence and the application of the necessary forces in the right places. Allocating large budgets for the War on Poverty did not eliminate the problem; neither does neglect. The War on Drugs has taken a few turns, but even the recent social campaign "Just say no!" is a large-scale approach. Despite positive intentions, we have not won these wars because we are using the wrong strategy.

There are other complex challenges that we have dealt with using large forces. Third World development is the international version of the War on Poverty to which the World Bank and other organizations have applied large forces. Recently, more thoughtful approaches are being taken, but they have not gone far enough. There is a tendency to fall into the "central planning trap." When challenges become complex enough, even the very notion of central planning and control fails. Building functioning socioeconomic systems around the world is such a complex problem that it will require many people taking small and targeted steps — like the special forces in Afghanistan.

There are other challenges that we have not yet labeled wars, which are also suffering from the same large-force approach. Among these are cost containment in the medical system and improving the education system. In the medical system, the practice of cost controls through managed care is a large-force approach that started in the early 1980s. Today, the medical system quality of care is disintegrating under the stresses and turbulence generated by this strategy. Medical treatment is clearly one of the most complex tasks we are regularly engaged in. Across-the-board cost control should not be expected to work. We are just beginning to apply the same kind of large-scale strategy to the education system through standardized testing. Here again, a complex systems perspective suggests that the outcomes will not be as positive as the intentions.

The wide applicability of lessons learned from fighting complex wars, and the effective strategies that resulted, should be further understood through research projects that can better articulate the relevant lessons and how they pertain to solving the many and diverse complex social problems we face.

Global and Larger Systems

Global systems — physical, biological, and social — are potentially the most complex systems that are studied by science today. Complex systems methods can provide tools for analyzing their large-scale behavior. Geophysical and geobiological systems, including meteorology, plate tectonics and earthquakes, river and drainage networks, the biosphere and ecology, have been the motivation for and the application of complex systems methods and approaches (Dodds and Rothman 2000; Lorenz 1963; Bak and Tang 1989; Rundle, Turcotte, and Klein 1996; NOAA 2002). Such applications also extend to other planetary, solar, and astrophysical systems. Converging technologies to improve human performance may benefit from these previous case studies.

Among the key problems in studies of global systems is understanding the indirect effects of global human activity, which in many ways has reached the scale of the entire earth and biosphere. The possibility of human impact on global systems through overexploitation or other by-products of industrial activity has become a growing socio-political concern. Of particular concern are the impacts of human activity on the global climate (climate change and global warming), on the self-sustaining properties of the biosphere through exploitation and depletion of key resources (e.g., food resources like fish, energy resources like petroleum, deforestation, loss of biodiversity). Other global systems include global societal problems that can include the possibility of global economic fluctuations, societal collapse, and terrorism. Our effectiveness in addressing these questions will require greater levels of understanding and representations of indirect effects, as well as knowledge of effective mechanisms for intervention, if necessary. In this context, the objective is to determine which aspects of a system can be understood or predicted based upon available information, along with the level of uncertainty in such predictions. In some cases, the determination of risk or uncertainty is as important as the prediction of the expected outcome. Indeed, knowing "what is the worst that can happen" is often an important starting point for effective decision-making.

In general, the ability of humanity to address global problems depends on the collective behavior of people around the world. Global action is now typical in response to local natural disasters (earthquakes, floods, volcanoes, droughts); man-made problems from wars (Gulf War, Bosnia, Rwanda, the war on terrorism); and environmental concerns (international agreements on environment and development). In addition, there is a different sense in which addressing global concerns requires the participation of many individuals: The high complexity of these problems implies that many individuals must be involved in addressing these problems, and they must be highly diverse and yet coordinated. Thus, the development of complex systems using convergent technologies that facilitate human productivity and cooperative human functioning will be necessary to meet these challenges.

0 0

Post a comment