-
Physics of Complex Systems
Emergence, Innovation & Robustness in Evolutionary Systems
Information Processing & Computation in Complex Systems
Dynamics & Quantitative Studies of Human Behavior
Emergence, Organization & Dynamics of Living Systems
Dynamics & Quantitative Studies of Human Behavior & Institutions
A priority focus of research at the Institute continues to be the emergence, persistence and demise of social institutions and their co-evolution with distinctive human behaviors — such as altruistic cooperation, out-group hostility and adaptive learning — typically overlooked in standard economic and other behavioral science models. In line with the well-established scientific tradition of the Institute, this research is characterized by (a) its trans-disciplinary nature, (b) its use of nonlinear dynamical systems to study the explicit out-of-equilibrium dynamics of the relevant processes, (c) and sustained interaction between mathematical modeling and well-defined empirical case studies and problems of potentially great contemporary practical relevance. Summarized here are several research areas. For a comprehensive list of SFI researchers working on topics related to Dynamics of Human Behavior and Institutions, click here.
Financial Markets as an Empirical Laboratory
SFI researchers, including External Professor Brian Arthur and Professor Doyne Farmer , have a long track record of innovative research in financial markets, with a tradition of challenging mainstream paradigms. Whereas early work typically involved conceptual toy models, more recent work is closely grounded by studies of empirical data. A recent focus of SFI projects has been on understanding the price impact function, which measures the response of prices to trading. Price impact curves are important for their own sake, but are also important because they can be used to map out the ecology of financial investment strategies. Farmer, SFI External Professor Fabrizio Lillo and colleagues have shown that price impact functions have interesting scaling properties, and have recently introduced a method to separate mechanical and informational causes. Other work has shown that on short time scales prices are strongly influenced by fluctuations in liquidity, i.e., by fluctuations in the response of prices to trades. Farmer, SFI Professor Eric Smith and others have also developed the zero-intelligence method as a way to understand the role of financial institutions in shaping processes such as price formation. This approach begins by assuming that people make decisions more or less at random, and takes advantage of the resulting simplicity to develop a tractable model of their interactions through an institution such as the continuous double auction. This approach has had considerable success in explaining the relationship between trading order placement and price formation. Building on previous work, SFI financial markets research will continue to develop zero and low intelligence agent models as an approach to understanding price formation. A particularly exciting direction is in explaining the shape of price impact functions on long time scales.
Farmer and Lillo, in collaboration with SFI Professor Jon Wilkins , External Professor John Geanakoplos , Visiting Professor Jennifer Dunne , SFI Postdoctoral Fellow Michelle Girvan and others, also plan to use financial markets as a laboratory to study cultural evolution. Because financial markets provide a detailed record of human decision making under risk and uncertainty in a complex environment over long spans of time, they provide an ideal setting to study the context and history dependent aspects of decision making. Financial decision making strategies come under strong selection pressure — successful innovation is strongly rewarded and failure is severely punished. Significant changes take place over spans of time as short as a few years, making it possible to study evolutionary time scales.
The Cost of Money as a Public Good
The zero-intelligence focus on the dynamics and constraints of market clearing mechanisms is part of a larger focus at SFI on the central role of institutions as the foundations for economic behavior. It is well known that the dominant mathematical theory of microeconomics (called the neoclassical paradigm) does not involve money, nor does it involve any explicit representation of markets or other institutions. Yet much of the highest-quality and most mathematically regular economic data concerns money, and the aspect of social organization most amenable to quantitative study is the structure of markets and other institutions. Whereas in the neoclassical paradigm, institutions are superfluous except perhaps as expressions of the aggregation of individual preferences, research of SFI Professor Eric Smith and others has shown that many properties of price formation appear to be completely institutional in origin. Similarly, there are regularities in the progression through institutional forms of markets, observable across societies and historical epochs, particularly from the use of commodity moneys to government-certified coinage, to fiat money and most recently pure systems of account. Smith and SFI External Professor Martin Shubik have used the concepts of symmetry and universality to classify simple exchange markets, and have shown that systematic trade-offs between cost and allocative efficiency provide criteria for selection among these markets, independent of the history by which a country may arrive at one or another form. Future work concerns the endogenous creation of institutions such as markets, money systems, and banks, as part of the optimization problem faced by a society in making economic coordination possible.
Emergence and Robustness of Community Structures
Research on the emergence of community structures in social systems has so far been dominated by game-theoretical approaches, which frequently draw on evolutionary models. Although this approach has yielded important insights, the emphasis on equilibria (evolutionary stable states, etc) neglects transient, out-of-equilibrium dynamics that may be critical for understanding real communities, especially when they find themselves in changing environments.
SFI External Professors Erica Jen , Simon Levin , Sander van der Leeuwe , Lisa Curran , SFI Professor Steve Lansing and others are developing a multiple network dynamics perspective to identify the characteristic patterns of community structure over time. The basic hypothesis is that structure emerges through the interaction of factors including a community’s mode of production, language, social institutions, coupling with local natural resources, and relation to its parent communities. The simultaneous mapping of these networks, and the study of the mechanisms by which their interactions lead to robustness or fragility of community structure, will provide important insight into the social ecology and adaptation of communities under stress. Specific systems that serving as a starting point for developing methodological tools and conceptual frameworks include indigenous communities in Indonesia, the Brazilian Amazon, and Belize. See also research by Steve Lansing on genes, culture, and the evolution of language and social structure.
Geographic Structure, Demographic History, and Approximate Bayesian Computation
The size of social groups, the degree of movement between groups, population booms and crashes have had a decisive influence on the long-term evolution of human behavior. The goal of this research is to use the contemporary geographic distribution of genetic diversity to understand the demographic history of particular populations (patterns of migration and range expansion, and major historical changes, such as population bottlenecks). A portion of the Institute’s work in this area, led by SFI Professor Jon Wilkins, has focused on generating solutions to particular models of geographic structure, and applying these models to empirical data. Humans represent a radically non-equilibrium population. In addition to the relatively recent expansion of human populations out of Africa and the dramatic increase in the human population size, human patterns of migration and reproduction are sensitive to cultural practices, which can vary rapidly both across space and time. Wilkins and collaborators have begun incorporating these non-equilibrium cultural transformations into the analysis of one particular aspect of this problem: sex-biased dispersal in humans. The long-term goal is to develop approximate Bayesian computational methods that can account for the complexity of human demographic history and that can incorporate the wealth of information available from other fields (such as history, anthropology, linguistics, and archaeology) into a unified, statistically sound, inferential framework.
War Time Sexual Violence
Sexual violence varies in extent and form among civil wars as well as inter‑state wars, among ethnic wars as well as non‑ethnic, and among secessionist conflicts. Despite the challenges to gathering data on this sensitive topic, the variation does not appear to be a product of inadequately reported violence: there are well-documented cases at the low end of the spectrum of sexual violence as well as the high end. Through multi-sited field research and new analysis of existing data, research on sexual violence at SFI, led by SFI Professor Libby Wood seeks to explain the variation in sexual violence during war. Goals include (a) documenting variation in wartime sexual violence, with particular attention to cases with little sexual violence; (b) exploring in greater depth cases where the pattern of sexual violence appears to be of particular interest in accounting for that variation; (c) developing candidate explanations for the observed variation; and (d) to the extent feasible, assessing their empirical plausibility. This project analyzes the puzzle of variation of sexual violence in the context not just of war but also in light of the cultural norms and practices of particular societies, bearing in mind that war can radically change norms and practices and that small group dynamics can profoundly undermine both individual norms and armed group policy.
Agent-based Modeling in the Social Sciences
Agent-based modeling offers a useful theoretical portal from which to explore complex adaptive systems. This type of ``bottom-up'' modeling begins by capturing, in the form of simple algorithms, the essential behavior of the key agents in the system under investigation. These computations are then allowed to interact with one another, resulting in a new model system that can be explored. Agent-based models are well suited for the analysis of dynamical systems of heterogeneous, adaptive agents. These types of systems are often difficult to capture using more traditional modeling tools like axiomatic mathematics. Over the last two decades agent-based models have become an important tool in a variety of fields, and have been used to explore problems ranging from urban segregation to the flocking of birds to trading in markets. SFI External Professors George Gumerman and Tim Kohler and others are simulating ancient societies of the southwest, SFI Professor Steve Lansing is modeling the relation between rule based marriage and genetic relatedness in Bali, and SFI External Professors John Miller and Scott Page are modeling complex adaptive social systems using thoughtful agents.
The Dynamics of Civilizations
8000 years ago, the ancestors of humanity--whether they lived in sedentary villages or mobile bands, whether they collected wild foods or herded flocks or cultivated gardens--lived in a world of kin and ritual in which the exploitation of many by a few was held in check. In a millennium or less, first in Mesopotamia, then in Egypt, then in India, China, Peru, Mexico, and elsewhere, villagers were incorporated into larger polities focused on sprawling urban centers and ruled by emergent elites. Did people willingly give up their independence to join such new polities or were they coerced by despots? The Dynamics of Civilizations Group at SFI, led by SFI External Professors Henry Wright and Doug White , and Science Board Member Bob Adams , aims to explain this fundamental transformation in the human career accelerated during the 20th century.
The Evolution of Human Behaviors & Institutions
This project, run by SFI Professor Sam Bowles , seeks to understand how the institutions that regulate social interactions – such as economic exchange, marital matching, and cooperation and conflict within and between groups – shape the evolution of individual preferences, norms, and other motivations, and in turn how the resulting individual behaviors shape the evolution of social institutions. Methods include stochastic evolutionary game theory, gene-culture co-evolutionary models, agent-based simulations, and behavioral experiments. To sharpen and discipline the theory-building process, Bowles and colleagues address such empirical puzzles as the innovation, persistence and demise of institutions regulating economic activity and the distribution of wealth. Another important theme is the nature and diversity of other-regarding preferences such as altruism and in-group bias, and their evolutionary origins and contemporary dynamics.
Research topics include: long-term dynamics of hierarchy and economic disparity (Bowles , SFI External Professor Hillard Kaplan , and others), institutional niche construction and the evolution of a cooperative species (Bowles), social preferences in nonhuman primates (SFI Postdoctoral Fellow Charles Efferson ), parochial altruism, outsider hostility and war (Bowles, SFI Professor Libby Wood , SFI Postdoctoral Fellow Dan Hruschka , and others), social learning, culture, and institutional equilibrium selection (SFI External Professor Scott Page , Efferson), network emergence of cooperation, outsider hostility, and cliques (Hruschka, Bowles, others), how to govern the ‘cooperative species’ (Bowles and others), and the dynamics of cooperative institutions (Page, Bowles).
The Evolution of Human Languages
As time passes, languages change. Vocabulary items (and grammatical features) are passed along from care-givers to children, often with alterations of meaning and/or pronunciation. This process of “vertical” transmission may be contrasted with the “horizontal” process of borrowing features from neighboring languages. A language evolves into “daughter” languages descended by vertical transmission from a common “proto-language,” with sound changes tending to follow definite rules. Meanwhile, borrowing occurs as well, complicating the picture.
Vertical transmission thus creates groupings of languages, with each grouping descended from a common proto-language, and a whole phylogeny can be built up in this way, with “branches” grouped into families, which form super-families, and so forth. Thus English belongs, together with German, Dutch, Danish, etc., to the Germanic branch of the Indo-European family, which also includes branches such as Slavic, Hellenic, Celtic, Iranian, and so forth. The Italic branch includes Latin and the romance languages descended from it.
Historical linguists reconstruct ancient proto-languages using their surviving descendants and figuring out the laws of sound changes involved, while also paying attention to grammatical and semantic shifts. Sometimes a proto-language is attested, as in the case of Latin, which could have been reconstructed from its Romance daughter languages if we hadn’t already known it.
The various proto-languages have a variety of ages. “Glottochronology,” based on percentage of overlap in basic vocabulary, supplies a rough method of dating the time when a proto-language splits up. It gives results compatible with other methods of estimating age, based on history and pre-history. Thus Indo-European is assigned an age something like seven thousand years. Other well-established families of comparable age are Uralic (including Finnish, Hungarian, and the Samoyed languages), Altaic (including Turkish, Mongolian, Japanese, and Korean), Semitic (including Hebrew, Arabic, and ancient Akkadian), etc. The program includes research on the mathematical formulae used in glottochronology.
The work of comparative linguists on a given language grouping culminates in the preparation of an etymological dictionary of the grouping, in which the sound correspondences are worked out, the proto-language is reconstructed, and numerous vocabulary items are traced back to that protolanguage. The founding director of the Evolution of Human languages program, the late Sergei Starostin, created, along with two colleagues, a three-volume etymological dictionary of the Altaic family. Such an undertaking requires years of effort. Other linguists of the program have compiled etymological dictionaries at the family level: a new one for Indo-European by Nikolaev, a new one for Semitic by Militarev and Kogan, a Chadic one by Olga Stolbova, and one for Mon-Khmer by Ilia Peiros.
The program searches for still larger groupings (super-families and super-super-families) and attempts to cover, in the end, all human languages, conceivably deriving them all from a single universal proto-language. If that radical hypothesis turns out to be correct, we can already point to a few very widespread words or roots that may be traceable all the way back. One of those is “man” for person, as in the English word. Another is “bak,” as in English “back.” The kinship term “kaka” for a mother’s brother or father is yet another. The root “men” for thinking may also belong on this list.
Meanwhile, the emphasis is on very large groupings short of the largest one, The Eurasiatic superfamily (about the same as ”Nostratic”) includes Indo-European, Uralic, Altaic, and several other families. Work has started on an etymological dictionary of Eurasiatic.
Virtually all the languages of Eurasia belong to four such super-families: Eurasiatic; Afroasiatic (which includes families ranging from Semitic to Chadic and to Cushitic in the horn of Africa; Dene-Caucasian, which ranges from Na-Dene (including Navajo) to Sino-Tibetan, the languages of the North Caucasus, and Basque; and Austric, the super-family.that includes the Polynesian and Indonesian tongues, Khmer, and many other languages of Southeast Asia and the islands. EHL linguists have found evidence that these four super-families belong to a single super-super-family (provisionally named Borean), which must go back to the middle of the last ice age, perhaps eighteen to twenty thousand years ago. Most or all native languages of the Americas may belong here too. That hypothesis is under investigation.
Perhaps some population bottleneck at that time produced this remarkable result that a single language of that era is the ancestor of virtually all the modern languages of Eurasia and perhaps the Americas as well.
It is hoped that with more data from places such as New Guinea, which has nearly a thousand languages, the world picture can be clarified further. It seems now that the “Indo-Pacific” region contains at least two more super-families. One of those appears to include, in New Guinea, the proto-language of the Tasmanian languages and also the proto-language of all the other Australian aboriginal languages.
Finally, EHL linguists are looking into the three additional super-families of sub-Saharan Africa identified by Joseph Greenberg around fifty years ago.
The program, with its associated archaeologists, cultural and physical anthropologists, and geneticists, features efforts to understand the early migrations of modern human beings that contribute to today’s linguistic situation. These efforts include attempts to reconstruct the cultures of the speakers of proto-languages and to locate the corresponding homelands.
All results of the program are available on the Internet through the EHL network of databases. This kind of sharing is very unusual in linguistics.
Please direct comments on the content of this page to Jessica Flack ().