Anti-Chaos, Common Property, and the Emergence of Cooperation

Complex adaptive systems, as conceived by John Holland, are groups of agents engaged in a process of coadaptation, in which adaptive moves by individuals have consequences for the group. Holland and others have shown that under certain circumstances simple models of this process show surprising abilities to self-organize (Holland 1993; Kauffman 1993). Complex adaptive systems have interesting mathematical properties, and the process of "anti-chaos"-—the spontaneous crystallization of ordered patterns in initially disordered networks— has become a new area of interdisciplinary research. But the question of whether these models can illuminate real world processes is still largely open. Not long ago John Maynard Smith described the study of complex adaptive systems as "fact-free science" (1995). This chapter has two purposes. First, in response to Maynard Smith, I will show how the concept of ecological feedback in complex adaptive systems provides a simple and powerful explanation for the structure and persistence of cooperative networks among Balinese rice farmers. Second, I will generalize this explanation to shed light on the emergence of cooperation in a class of social systems where interactions with the natural world create both rewards and punishments. But before turning to these examples, in line with the purposes of this volume I will comment on the ideas and assumptions that underlie the use of models in this analysis. "Society is a human product. Society is an objective reality. Man [sic] is a social product." With this epigram Peter Berger and Thomas Luckmann neatly encapsulated a fundamental problem in social theory (1967:61). In American anthropology today this paradox is often posed as a conflict between "structure" and "agency," where the former refers to ideational, economic, institutional, or psychological systems that are represented as generating social reality; and the latter to the ability of individual social actors to modify their own social worlds. The same paradox recurs in classical social theory, such as Jürgen Habermas' insistence on the need to somehow reconcile actor-focused and system-level social theories (Habermas 1985, 1987).

Be There Then : A Modeling Approach to Settlement Determinants and Spatial Efficiency Among Late Ancestral Pueblo Populations of the Mesa Verde Region, U.S. Southwest

The archaeology of southwestern Colorado from A.D. 900 to 1300 presents a number of interesting problems, including population aggregation and abandonment. We report on an on-going project, implemented using the modeling libraries of Swarm, to model the settlement dynamics of this region, treating households as agents. Landscape detail includes an annual model of paleoproductivity, soils, vegetation, elevation, and water resource type and location. Individuals within households reproduce and die; households farm, relocate, and die; children within households marry and form new households. Household location is responsive to changing productivity (depleted in some scenarios) and, in some scenarios, water resources. Comparison of simulated settlement with the archaeological record highlights changes in the settlement and farming strategies between Pueblo II and Pueblo III times, including the increasing importance of water and sediment-control, and other alternatives to extensive dry farming. Our results suggest that degradation of the dry-farming niche may have contributed to these changes. This project began with a desire to understand why, during certain times in prehistory, most Pueblo peoples lived in relatively compact villages, while at other times, they lived in dispersed hamlets (Cordell et al. 1994). Our approach to this problem is based on a thread of accumulating research begun in the early 1980s when a dissertation from the University of Arizona by Barney Burns (1983) showed that it was possible to retrodict potential prehistoric maize yields in a portion of Southwest Colorado by combining prehistoric tree-ring records with historic crop-production records of local farmers. A few years later, Kohler et al. (1986; see also Orcutt et al. 1990) simulated agricultural catchment size and shape in a northern portion of the present study area, to arrive at the suggestion that avoiding violent confrontation over access to superior agricultural land was a major force in forming the villages that appeared in this area in the late A.D. 700s and again in the mid-800s.

The Evolution of Cooperation in an Ecological Context : An Agent-Based Model

The social and behavioral sciences have a long-standing interest in the factors that foster selfish (or individualistic) versus altruistic (or cooperative) behavior. Since the 1960s, evolutionary biologists have also devoted considerable attention to this issue. In the last 25 years, mathematical models (reviewed in Wilson and Sober 1994) have shown that, under particular demographic conditions, natural selection can favor traits that benefit group members as a whole, even when the bearers of those traits experience reduced reproductive success relative to other members of their group. This process, often referred to as "trait group selection" (D. S. Wilson 1975) can occur when the population consists of numerous, relatively small "trait groups," defined as collections of individuals who influence one another's fitness as a result of the trait in question. For example, consider a cooperative trait such as alarm calling, which benefits only individuals near the alarm caller. A trait group would include all individuals whose fitness depends on whether or not a given individual gives an alarm call. If the cooperative trait confers sufficiently large reproductive benefits on the average group member, it can spread. This is because trait groups that happen to include a large proportion of cooperators will send out many more offspring into the population as a whole than will groups containing few, or no cooperators. Thus, even though noncooperators out reproduce cooperators within trait groups (because they experience the benefits of the presence of cooperators without incurring the costs), this advantage can be offset by differences in rates of reproduction between trait groups. Numerous models of group selection (Wilson and Sober 1994) show that whether cooperative traits can spread depends on the relative magnitude of fitness effects at these two levels of selection (within and between trait groups). In addition, there is a growing body of empirical evidence for the operation of group selection in nature (e.g., Colwell 1981; Breden and Wade 1989; Bourke and Pranks 1995; Stevens et al. 1995; Seeley 1996; Miralles et al. 1997; Brookfield 1998) and under experimental conditions (reviewed in Goodnight and Stevens 1997).

Evolution of Inference

Rousseau began his discussion of the origin of language with a paradox that echoes through modern philosophy of language. How can we explain the genesis of speech without presupposing speech, reference without presupposing reference, meaning without presupposing meaning? A version of this paradox forms the basis of Quine's attack on the logical empiricist doctrine that logic derives its warrant from conventions of meaning—that logical truths are true and logical inferences are valid by virtue of such conventions. Quine raised the general skeptical question of how conventions of language could be established without preexisting language, as well as calling attention to more specific skeptical circularities. If conventions of logic are to be set up by explicit definition, or by axioms, must we not presuppose logic to unpack those conventions? David Lewis (1969) sought to answer these skeptical doubts within a game theoretical framework in his book, Convention. This account contains fundamental new insights, and I regard it as a major advance in the theory of meaning. Lewis sees a convention as being a special kind of strict Nash equilibrium in a game that models the relevant social interaction. To say that a convention is a Nash equilibrium is to say that if an individual deviates from a convention which others observe, he is no better off for that. To say that it is a strict Nash equilibrium is to say that he is actually worse off. To this, Lewis adds the additional requirement that an individual unilateral deviation makes everyone involved in the social interaction worse off, so that it is in the common interest to avoid such deviations. A theory of convention must answer two fundamental questions: how do we arrive at conventions?, and by virtue of what considerations do conventions remain in force? Within Lewis' game-theoretic setting, these questions become, respectively, the problems of equilibrium selection and equilibrium maintenance. On the face of it, the second problem may seem to have a trivial solution— the equilibrium is maintained because it is an equilibrium! No one has an incentive to deviate.

Agent-Based Modeling of Small-Scale Societies: State of the Art and Future Prospects

The thematic social sciences—economics, political science, psychology, and so on—often privilege that aspect of human action on which they focus. Can we fruitfully understand change in human affairs from the perspectives of these disciplines? Philosophers have (for millennia), and anthropologists and geographers (for little more than a century) have said "no," and have attempted to view human phenomena as a totality. Anthropology, a holistic discipline, at its best integrates human biology, cultural anthropology or ethnology, psychological anthropology, linguistics, and archaeology. But the task is daunting, and has led often to elegant, but very specific case studies. However, new theoretical approaches to nonlinear and adaptive systems and to modeling such approaches give hope that rigorous general formulations are possible. The Culture Group of the Santa Fe Institute focuses on long-term stability and transformation in cultural developments. In December 1997, with the support of the Wenner-Gren Foundation for Anthropological Research, a diversity of researchers gathered in Santa Fe to assess the progress of this working group and to chart future directions. We had many fruitful exchanges, ranging from general theoretical problems of cultural change and its explanation to the specifics of modeling actual cultural processes. The touchstones of the discussions were breakthroughs in the modeling of small-community networks in southwestern North America, but new developments in other theoretical and empirical areas also proved important in pointing toward future efforts. This volume presents the much discussed and revised papers from the Santa Fe meeting. The conference began, as does this volume, with overviews of the state of the art of modeling. George Gumerman, in his preface, touches on the roots of modeling whole social and cultural systems in North America, threads of inquiry which are picked up in many chapters of this volume. Tim Kohler, in his elegant introduction argues the advantages of agent-based modeling as the resolution of several outstanding problems in traditional social science. Nigel Gilbert then provides rich insight into recent work in Europe, little known to many North American social scientists outside the modeling community.

Modeling Sociality: The View from Europe

Social science research based on computer simulation, much of it using multiagent, multilevel models, has grown dramatically in Europe since the early 1990s. This growth has been inspired by the recent upsurge of work within computer science on distributed artificial intelthe metaphor between agents and people/social actors. This chapter reviews some recent and influential European examples of the multiagent simulation of social phenomena. One common thread running through what is otherwise a very heterogeneous collection of studies is the description and exploration of a small number of generalized "logics" or "abstract social processes." It has been possible to investigate these through the construction of "artificial societies," and it is this methodological discovery that partly accounts for the current energy and excitement in the field of computational social simulation. However, the assumption of a simple correspondence between agents and social actors needs to be applied with some care if it is to be useful in understanding human societies. The same epistemological puzzles and problems that sociologists have struggled over during the last hundred years can recur in trying to understand soci eties through computer simulations. Some of these problems will be described, again with reference to current European studies. While the use of simulation as a methodological tool is a commonplace in the natural sciences and engineering (e.g., Shannon 1975; Zeigler 1976), it still strikes many people as remarkable that one could use simulation in the social sciences. The very idea of modeling the obvious complexity, unpredictability, and autonomy of humans and their societies using computer simulation is considered by some social scientists as absurd. They suggest that if simulation of social phenomena could ever be possible, it would have to involve such simplification that nothing of value could be learned. Clearly, the whole enterprise is just an excuse for playing around with computers. While I do not agree with this view, there is a real question at the heart of many social scientists' skepticism.

Fractal House of Pharaoh: Ancient Egypt as a Complex Adaptive System, a Trial Formulation

In addition to understanding small-scale societies in their own right "from a complex systems perspective" (Boekhorst and Hemelrijk this volume), workshop participants expressed a goal of using insights about the dynamics of small-scale societies to better understand the "evolution of state-like structures" (Small this volume), or "the 'emergence' trajectories by which a smallscale society, in its environment, may move autonomously from relatively simple (distributed, no ranking or centralized decision making) to complex (ranking/hierarchy, with centralized decision making and a degree of specialization)" (Doran this volume). Small-scale societies are seen as "preceding conditions" to the development of "rank vs. egalitarian ideologies" (Wright this volume) such as are found in archaic states. Ancient Egypt is a salient example of such an archaic state. In the comparative study of civilizations, ancient Egypt has stood out as the quintessence of a centralized nation-state ruling a large territory. Egyptologists often operate through a vision of ancient Egyptian society, whether explicit or assumed, as highly absolutist. Pharaoh's control of society is complete, effected through an invasive and pervasive centralized bureaucracy. Anthropologists, taking their cue from Egyptologists, see Egypt as one of the earliest examples of a unified nation-state, with a redistributive economy centrally administered over the entirety of the Egyptian Nile Valley. I offer a prospectus for approaching Egyptian civilization as a complex adaptive system (CAS) based on loose analogies with concepts of emergent order and self-organization. This a narrative exploration of ways that ancient Egyptian society may be amenable to the kind of agent-based modeling applied to small-scale societies. Although I recognize that in discussions of "complexity theory" there is nothing close to unanimity or an agreed paradigm (Wilson 1998), some of the more general concepts may at least offer insightful new ways to view social complexity in Egypt. My prospectus is a workin- progress. My sources for complex systems studies are "the literature of metaphor (e.g., Cowan et al. 1994), and the popularizations of metaphysics"; that is to say, what follows is most certainly in Morowitz's (1998) category of meta-metaphor (and I will try to refrain from "word magic").

Understanding Anasazi Culture Change Through Agent-Based Modeling

Traditional narrative explanations of prehistory have become increasingly difficult to operationalize as models and to test against archaeological data. As such models become more sophisticated and complex, they also become less amenable to objective evaluation with anthropological data. Nor is it possible to experiment with living or prehistoric human beings or societies. Agentbased modeling offers intriguing possibilities for overcoming the experimental limitations of archaeology by representing the behavior of culturally relevant agents on landscapes. Manipulating the behavior of artificial agents on such landscapes allows us to, as it were, "rewind the tape" of sociocultural history and to experimentally examine the relative contributions of internal and external factors to sociocultural evolution (Gumerman and Kohler in press). Agent-based modeling allows the creation of variable resource (or other) landscapes that can be wholly imaginary or that can capture important aspects of real-world situations. These landscapes are populated with heterogeneous agents. Each agent is endowed with various attributes (e.g., life span, vision, movement capabilities, nutritional requirements, consumption and storage capacities) in order to replicate important features of individuals or relevant social units such as households, lineages, clans, and villages. A set of anthropologically plausible rules defines the ways in which agents interact with the environment and with one another. Altering the agents' attributes, their interaction rules, and features of the landscape allow experimental examination of behavioral responses to different initial conditions, relationships, and spatial and temporal parameters. The agents' repeated interactions with their social and physical landscapes reveal ways in which they respond to changing environmental and social conditions. As we will see, even relatively simple models may illuminate complex sociocultural realities. While potentially powerful, agent-based models in archaeology remain unverified until they are evaluated against actual cases. The degree of fit between a model and real-world situations allows the model's validity to be assessed. A close fit between all or part of a model and the test data indicates that the model, albeit highly simplified, has explanatory power. Lack of fit implies that the model is in some way inadequate.

Trajectories to Complexity in Artificial Societies: Rationality, Belief, and Emotions

This chapter illustrates and discusses the use of agent-based artificial societies to explore possible trajectories into social complexity through the integration of ideas from both anthropology and agent technology. Particular attention is paid to the role of rational cooperation, collective belief, and emotional dynamics in these trajectories. Some methodological problems associated with the use of artificial societies to build social theory are also discussed, especially how best to reduce the impact of our own cultural preconceptions. Computer simulation work in archaeology and anthropology is more than 25 years old (see Doran and Hodson 1975, chapter 11; Doran 1990; and compare Halpin to appear). After a period of enthusiasm in the early 1980s interest waned, but recently there have been a number of important computer-based studies of (human) social phenomena using so-called agent-based modeling (e.g., Kohler et al. this volume) and agent-based artificial societies (e.g., Epstein and Axtell 1996), and more are in progress. Both types of study involve (software) agents, that is, according to a standard textbook definition, entities which perceive and act in an environment (Russell and Norvig 1995:49). Reactive agents are typically built around a small number of relatively simple situation-to-action rules. Deliberative agents are more complex, typically posting goals and then forming and executing plans to achieve them. It is this rapidly developing "agent technology," largely based upon artificial intelligence studies, that is the driving force behind the new work. The methodology associated with both agent-based modeling and agentbased artificial societies emphasizes the ability to address explicitly processes of cognition, and hence phenomena that previous models could not tackle, and also the ability to explore what could happen rather than what has happened or is happening. However, unlike agent-based modeling, artificial societies are, in essence, models without a specific target system, and it has been argued that this type of modeling permits the study of societies and their processes in the abstract (Epstein and Axtell 1996; Doran 1997). An underlying assumption is that it is possible and useful for social scientists to explore wide-ranging and abstract social theories and that these theories can be expressed in terms of computational processes.

The Impact of Raiding on Settlement Patterns in the Northern Valley of Oaxaca: An Approach Using Decision Trees

A growing body of data indicates that armed conflict played a role in the creation of complex societies such as chiefdoms and states (Wright 1984; Spencer 1998). For example, according to Wright (1977:382), "most ethnographically reported chiefdoms seem to be involved in constant warfare," and large chiefdoms grew by absorbing their weaker neighbors. Marcus and Flannery suggest that warfare was often used to create a state out of rival chiefdoms: . . . We do not believe that a chiefdom simply turns into a state. We believe that states arise when one member of a group of chiefdoms begins to take over its neighbors, eventually turning them into subject provinces of a much larger polity. (Marcus and Flannery 1996:157) . . . As an example of this process, the authors cite Kamehameha's creation of a Hawaiian state out of five to seven rival chiefdoms between 1782 and 1810. They suggest that something similar happened in the Valley of Oaxaca, Mexico, when a chiefdom in the Etla region seized the defensible mountain top of Monte Albán and began systematically subduing rival chiefdoms in the southern and eastern parts of the valley. If this is the case, there should be a point in the sequence when considerations of defense began to influence settlement choice. In this chapter, our goal is to provide a preliminary description of our efforts in testing the suitability of this model to the Oaxacan case, and its potential use as the basis for a more general model of state formation. In order to test this hypothesis we need some way to operationalize it in terms of the archaeological record in the Valley of Oaxaca. The key phases of the model can be expressed as follows: 1. An early period in which raiding was minimal, and variables relevant to successful agriculture predominate in settlement choices. 2. A gradual rise in friction between social groups prior to state formation. This friction can be represented by archaeological evidence for raiding, the principle form of warfare in tribes and chiefdoms.