The Institute Different

Courtesy Santa Fe InstituteThe Santa Fe Institute, situated in the hills above Santa Fe, N.M. Even its interior design serves the unusual purpose of the Santa Fe Institute (SFI). At the top of a winding drive on the outskirts of the New Mexico capital that calls itself "the city different," SFI occupies a 1950s hacienda defined by three descending "pods." First is reception and administration. Second is a community area, full of comfortable furniture, with big views of the city and mountains. Centerpiece of this second pod is an inner courtyard, where the approximately 200 scientists from throughout the world who visit each year can meet for sumptuous, British-style teas. Around the courtyard are conference rooms. The third pod, in the back, is a warren of small offices. Progress from reception to the offices therefore requires passage through the community pod, where the chances of overhearing a conversation or being waylaid to chat are maximized. This layout encourages the institute to do what it does best: facilitate communication between scientists of different fields, in the belief that ideas worth interdisciplinary pursuit will emerge. The people that SFI draws together, and the ways in which they collaborate, are at once quite unlike typical scientific investigation and at the forefront of its evolution. The nonprofit organization has only 25 staff, seven residential faculty, a dozen postdoctoral fellows, and five graduate students, but about 60 external faculty visit for up to one month annually. Each is selected after a screening process and the institute pays for transportation and off-campus housing. They aren't salaried, but neither are they required by SFI to publish, teach, or otherwise produce in conventional ways. The institute's $6 million annual budget is derived in approximately equal amounts from research grants, donations from individuals and private foundations, and corporate contributions. What the scientists study is a broad range of complex adaptive systems (CAS). Basically, this is any system that exhibits random variation and selection, resulting in learning or evolution. A CAS can be microscopic, like the immune system or the central nervous system, it can be an organism that includes those systems, and it can even be a composite of such organisms, like a termite mound, an ecology, or an economy. Exact definitions differ--for example, some SFI researchers contend that a computer-generated program's artificial intelligence makes it a complex adaptive system, while others disagree--but no one disputes that a CAS can self-organize, and that phenomena emerge from it that cannot be fully understood by analyzing parts of the system and then combining them. One example of such phenomena is the emergence of mind from the human brain. Studying a CAS therefore requires a nonlinear approach to supplement conventional reductive methods. SFI as a Market The institute itself is a complex adaptive system. "In some ways, SFI is like a market, with a constant flow of participants entering and leaving," comments longtime external faculty member John H. Holland, a leader in the field of genetic algorithms, who holds two professorships at the University of Michigan, in electrical engineering and computer science, and in psychology. "It is also like a market in that the aggregate produces phenomena not directly attributable to the participants. The studies and research that emerge result from interactions that would be unlikely under the 'departmental' system that holds in most university, government, and industrial organizations." One of SFI's founders, George A. Cowan, cautions that the addition of human behavior to such a system brings with it many more constraints than for a nonhuman CAS. "The commonalities are there," he concurs, "but the phenomenology and the dynamics are very, very different." That isn't to say Cowan, now an SFI distinguished fellow, would eschew study of such systems. When he catalyzed the initial meetings that led to the institute's formation in 1984, he had a background in multidisciplinary research as a chemist at Los Alamos National Laboratory (LANL), about 23 miles northwest of Santa Fe. "I was familiar with the good things that happen when you get people from various disciplines that overlap to talk to one another." The founding group discussed interfaces between disciplines, emergent phenomena, complexity, and nonlinear dynamics. Murray Gell-Mann Theoretical physicist Murray Gell-Mann, another early instigator of the institute, recalls that he was hoping it would be highly transdisciplinary and could find a grand synthesis on which to work, such as the unified theory of all particles and forces, or biological evolution and genetics. Founding workshops were organized, featuring numerous speakers. "It turned out that just about every speech was somehow related to simplicity, complexity, adaptation, and evolution," he remembers. "And that became our grand synthesis." The specialties represented today at SFI include mathematics, computer science, physics, chemistry, immunology, population biology, ecology, evolutionary biology, neurobiology, psychology, cultural anthropology, linguistics, archaeology, history, political science, and economics. Cowan believes that SFI's inclusion of physical scientists in questions involving the social sciences has stimulated a new approach to research. He advocates more such collaborations, perhaps combining neuroscience, learning theory, and the behavioral sciences. Courtesy Santa Fe Institute Cosma Shalizi, graduate fellow, working on a physics problem at the institute. Atypical Science Examples of such transdisciplinary projects widely vary. A political scientist applies the algorithms of a theoretical chemist to describe developments in the economy of medieval Florence. A physicist and two biologists formulate a scaling theory that can make accurate predictions about energy transport networks in numerous biological systems.1 A linguist employs a geneticist's DNA recombination model to study the evolution of language. The titles from a current list of 180 SFI working papers are revealing: "Physicists Attempt to Scale the Ivory Towers of Finance," "Life and Evolution in Computers," "Cooperative Transport by Ants and Robots," "Power Spectra of Extinction in the Fossil Record," and "The Structure of Scientific Collaboration Networks." "Science here is a lot of fun, it's much more fun than I've seen it elsewhere," says Thomas B. Kepler, who last year resigned his directorship of the biomathematics program at North Carolina State University to become the institute's vice president for academic affairs. "A lot of what goes on here is like improvisation, and it's critical to maintain that attitude. When jazz is fun, the improvisation really takes off and envelops the listener." Much of that fun is the intellectual challenge. As Holland notes, "Because most of the participants are not employees of the institute, showing up on a voluntary basis, there is a high premium on offering an environment and ideas that are exceptional. Otherwise, the faculty will simply vote with its feet, ceasing to participate. This is in stark contrast to most think tanks." Kelper adds, "Part of the beauty of SFI, and of interdisciplinary research in general, is that you have to see what you work with every day in a new way, from someone else's perspective." He likens the process to regarding a four-dimensional object from numerous three-dimensional angles. Eventually, a sense of the object emerges. "Once one is able to see it, it's a transformative experience." Although the work at SFI is largely theoretical--the campus has no labs, and computer modeling is the only onsite research that could be considered experimental--Kepler is eager to develop additional collaborations with other institutions; particularly in biomedicine, where he sees potential for useful contributions of complex adaptive theory to experimentation. One such SFI triumph was the mathematical analysis in the mid-1990s by theoretical immunologist Alan S. Perelson of data from David D. Ho, scientific director of the Aaron Diamond Research Center at Rockefeller University. Perelson, theoretical biology and biophysics group leader at LANL, showed with the assistance of former SFI and LANL postdoc Avidan U. Neumann that HIV-1 has a rapid turnover, rather than the previously theorized dormant period before AIDS appears.2 This discovery helped to pave the way for the successful "hit early, hit hard" strategy of combination therapy. CAS Characteristics Gell-Mann, a 1969 Nobel laureate in physics for his work on the classification of elementary particles and their interactions, also studies aspects of simplicity and complexity, a transdisciplinary endeavor he calls plectics. The general characteristics of a complex adaptive system, he asserts, include its ability to identify "regularities," as opposed to random data, in the stream of information it receives about itself and its surroundings. These regularities are then compressed into a schema, or internal model, that can supply descriptions of the real world, make predictions about what will happen there, and thus prescribe behavior for the CAS. The results obtained by this schema in the real world feed back into the system, to affect the schema's standing in competition with rivals that have arisen from mutations of various sorts.3 In biology, the genotype is a schema. In human society, the schemata can be laws, customs, myths, or institutions. The scientific enterprise also is a CAS, in which regularities are identified from a vast quantity of data and compressed into a theory, which is the schema. In the case of SFI, the schemata are its bylaws, regulations, and traditions that comprise a "cultural DNA." For example, the institute has a rule that limits on-site scientists to 50 at any time. Such schemata sometimes give way to competitors as SFI evolves. For instance, Gell-Mann notes that an original intention was to grant degrees, but this was abandoned with the realization that accreditation probably would require a full set of courses and a huge faculty to teach them. Ellen Goldberg SFI president Ellen Goldberg observes that promotion and tenure in universities occur within departments, and interdisciplinary work accordingly risks failure to find funding or other support. Five years ago, she left the University of New Mexico as associate provost and director of graduate research to lead SFI. She comments that a complex adaptive system is built on relatively simple rules, and she offers an early lesson from Cowan that underscores this point. "He told me, 'Ellen, all you need to do is look for smart people and allow them to do their thing. You don't need an agenda.'" She continues, "In many ways, I see us as self-organizing," but she adds that the impression that a CAS operates without a central authority is mistaken. Gell-Mann explains that although a model of a CAS might omit a central authority for the sake of mathematical simplicity, most such real systems typically do not function that way. In the case of human society, for example, "One of the most important things such an aggregation can evolve is a government." In addition to Goldberg and Kepler, SFI has a distinguished, widely representative science board that meets annually, a science steering committee that addresses all appointments and projects, and a board of trustees. A complex adaptive system has a tendency to spawn other such systems, in the way that, perhaps, biological evolution gave rise to human thinking. Although there have been numerous attempts to emulate SFI, its principals do not yet know of another such free-standing campus in the world. "It's not an easy endeavor, and we see our job as trying to help these groups put together SFI-like interactions," Goldberg says. Funded by a private donation, the institute last year began assisting interdisciplinary efforts in China, India, and Russia. In 2001, the project is being extended to Eastern Europe, Africa, and Latin America. In 2002, SFI plans to bring together the leaders of these fledgling collaborations for a workshop in Santa Fe. As Gell-Mann is fond of noting, "We have a worldwide family." And the family tree continues to branch. Steve Bunk (sbunk@uswest.net) is a contributing editor for The Scientist. References 1. S. Bunk, "Do energy transport systems shape organisms?" The Scientist, 12[24]:14, Dec. 7, 1998. 2. D.D. Ho et al., "Rapid turnover of plasma virions and CD4 lymphocytes in HIV-1 infection," Nature, 373:123-6, 1995. 3. M. Gell-Mann, The Quark and the Jaguar: Adventures in the Simple and the Complex, W.H. Freeman: New York, 1994. Complex Social Behavior: What Does 'Hierarchy' Mean? Among the first things done by an 18-member Social Insects Working Group gathered around a big table at the Santa Fe Institute (SFI) was to subdivide the group. For three days in October 2000, researchers considered issues in the evolution of social organization, and each six-person "breakout group" was required to contain a similar balance of disciplines. Their specialties included genetics, molecular biology, neurobiology, modeling, ecology, and even the philosophy of science. Some scientists worked with honeybees, others with wasps, or ants. Their interests and skills were close enough to allow good communication but different enough to spark new synergies. The scientists traveled from Germany, Belgium, Norway, Costa Rica, and from eight states. Among their goals was the development of research collaborations that would help to explain how complex social behavior, at many levels of life, comes to be. The three breakout groups repaired to smaller rooms, where they spent much time on terminology. Each subgroup considered a different issue. One was working on the origins of social organization, and it set about identifying features that mark transitions in the evolution of social insects. But the participants wanted to describe these features in a way that would allow comparisons with non-insect complex systems. The words "solitary," "group living," and "hierarchical groups" were entered on the blackboard, and the trouble started. What was meant by "hierarchy?" Is it characterized by reproductive dominance, is it social, is it behavioral? And does the structural shift from solitary to group infer behavior, or can groups be mere aggregates? This was just the beginning of such questions. For example, shouldn't groups be subdivided into the noncohesive and the cohesive? And isn't "brood tending" a more encompassing term than "brood care?" And is "differentiation" merely morphological, or could it be either flexible or irreversible at different stages of group structure? After a break, the subgroup decided to identify other entities that could be compared to insects, to consider if the transition from solitary to group living involves similar changes that affect cohesion and differentiation. They produced possible comparisons in multidomain proteins, in ecosystems, and in human populations. Finally, they decide that a tension exists in complex systems between cohesiveness and differentiation (a conclusion well-demonstrated by their own interactions). The Working Group declared the meeting wonderfully successful. Roughly twice a month, such gatherings are held at SFI, a few leading to major research projects. This group settled on numerous paths for future collaboration, including more work at the institute. It even set up a Web page, sfi.cyberbee.net. --Steve Bunk

The Institute Different

Interested in reading more?

Become a Member of