Generative Social Science: Studies in Agent-Based Computational Modeling

Generative Social Science: Studies in Agent-Based Computational Modeling

Joshua M. Epstein
Copyright Date: 2006
Edition: STU - Student edition
Pages: 352
https://www.jstor.org/stable/j.ctt7rxj1
  • Cite this Item
  • Book Info
    Generative Social Science: Studies in Agent-Based Computational Modeling
    Book Description:

    Agent-based computational modeling is changing the face of social science. InGenerative Social Science, Joshua Epstein argues that this powerful, novel technique permits the social sciences to meet a fundamentally new standard of explanation, in which one "grows" the phenomenon of interest in an artificial society of interacting agents: heterogeneous, boundedly rational actors, represented as mathematical or software objects. After elaborating this notion of generative explanation in a pair of overarching foundational chapters, Epstein illustrates it with examples chosen from such far-flung fields as archaeology, civil conflict, the evolution of norms, epidemiology, retirement economics, spatial games, and organizational adaptation. In elegant chapter preludes, he explains how these widely diverse modeling studies support his sweeping case for generative explanation.

    This book represents a powerful consolidation of Epstein's interdisciplinary research activities in the decade since the publication of his and Robert Axtell's landmark volume,Growing Artificial Societies. Beautifully illustrated,Generative Social Scienceincludes a CD that contains animated movies of core model runs, and programs allowing users to easily change assumptions and explore models, making it an invaluable text for courses in modeling at all levels.

    eISBN: 978-1-4008-4287-2
    Subjects: Economics, Mathematics

Table of Contents

  1. Front Matter
    (pp. i-vi)
  2. Table of Contents
    (pp. vii-x)
  3. INTRODUCTION
    (pp. xi-xxii)

    The introduction toGrowing Artificial Societiesoffers the following thought on the future of explanation:

    What constitutes an explanation of an observed social phenomenon? Perhaps one day people will interpret the question, “Can you explain it?” as asking “Can yougrowit?” Artificial society modeling allows us to “grow” social structuresin silicodemonstrating that certain sets of microspecifications aresufficient to generatethe macrophenomena of interest . . . . We can, of course, use statistics to test the match between the true, observed, structures and the ones we grow. But the ability to grow them . . ....

  4. Prelude to Chapter 1: THE GENERATIVIST MANIFESTO
    (pp. 1-3)

    Every so often, there is a special conference of the Santa Fe Institute External Faculty. Initially called the Integrative Themes Workshops, their laudable purpose has been to explore crosscutting, unifying aspects of all the diverse efforts underway in the far-flung Institute, from physics, to computer science, to evolutionary ecology, to archaeology, to economics, to immunology.

    Over the years, these external faculty conferences have continued. Just as one would expect of a high-powered transdisciplinary community like SFI, the areas of disagreement probably dominate those of consensus. And so, with a healthy self-deprecating sense of its own heterogeneity (if not incoherence), the...

  5. Chapter 1 AGENT-BASED COMPUTATIONAL MODELS AND GENERATIVE SOCIAL SCIENCE
    (pp. 4-46)
    Joshua M. Epstein

    This article argues that the agent-based computational model—or artificial society—is a new scientific instrument.¹ It can powerfully advance a distinctive approach to social science, one for which the term “generative” seems appropriate. I will discuss this term more fully below, but in a strong form, the central idea is this: To the generativist, explaining the emergence² of macroscopic societal regularities, such as norms or price equilibria, requires that one answer the following question:

    ∗How could the decentralized local interactions of heterogeneous autonomous agents generate the given regularity?

    The agent-based computational model is well-suited to the study of this...

  6. Prelude to Chapter 2: CONFESSION OF A WANDERING BARK
    (pp. 47-49)

    My intellectual life began in music composition. I wrote three chamber pieces for small ensemble. Then, at Amherst, I became absorbed in mathematics and later, at MIT, in mathematical modeling. Now, I am immersed in agent-based modeling. It would certainly appear that I have wandered from music to mathematics to agent modeling, within which (judging by the current volume) I have wandered from civil violence to epidemiology to archaeology. But I do not feel that I have ever changedfields. What, then, is my field? In the next chapter, I quote Bertrand Russell on the topic of mathematical beauty. He...

  7. Chapter 2 REMARKS ON THE FOUNDATIONS OF AGENT-BASED GENERATIVE SOCIAL SCIENCE
    (pp. 50-71)
    Joshua M. Epstein

    The scientific enterprise is, first and foremost,explanatory. While agent-based modeling can change the social sciences in a variety of ways, in my view its central contribution is to facilitategenerativeexplanation (see Epstein 1999). To the generativist, explaining macroscopic social regularities, such as norms, spatial patterns, contagion dynamics, or institutions requires that one answer the following question:

    How could the autonomous local interactions of heterogeneous boundedly rational agents generate the given regularity?

    Accordingly, to explain macroscopic social patterns, we generate—or “grow”—them in agent models. This represents a departure from prevailing practice. It is fair to say that,...

  8. Prelude to Chapter 3: EQUILIBRIUM, EXPLANATION, AND GAUSS’S TOMBSTONE
    (pp. 72-74)

    Countless articles purport toexplainsocial phenomena by furnishing a Game in which the phenomenon is proved to be an equilibrium, typically a Nash equilibrium or some refinement thereof. To be sure, there are many cases in which the attempt to demonstrate that an important social regularity is Nash is deeply revealing. But there are at least three cases¹ where it won’t be:

    Case 1. The phenomenon of interest is anonequilibriumdynamic.

    Case 2. Equilibrium is attainable in principle, but not on acceptable time scales.

    Case 3. Equilibrium exists but is unattainable outright.

    Every model in this volume falls...

  9. Chapter 3 NON-EXPLANATORY EQUILIBRIA: AN EXTREMELY SIMPLE GAME WITH (MOSTLY) UNATTAINABLE FIXED POINTS
    (pp. 75-87)
    Joshua M. Epstein and Ross A. Hammond

    Much of game theory and mathematical economics is concerned with equilibria (see Kreps 1990, 405). Nash equilibrium is an important example. Indeed, in many quarters, “explaining an observed social pattern” is understood to mean “demonstrating that it is the Nash equilibrium of some game.” But, there is no explanatory significance to an equilibrium that is unattainable in principle. And there is debatable significance to equilibria that are attainable only on astronomical time scales. Yet, in a great many instances, the social pattern to be explained is simply shown to be an equilibrium. The questions, “Is the equilibrium attainable?” and “On...

  10. Prelude to Chapters 4–6: GENERATING CIVILIZATIONS: THE 1050 PROJECT AND THE ARTIFICIAL ANASAZI MODEL
    (pp. 88-89)

    The Artificial Anasazi model grew out of the 2050 Project. This was a multiyear collaboration of the Brookings Institution, the Santa Fe Institute, and the World Resources Institute, funded by the MacArthur Foundation. Although the project’s overall aim was to identify the conditions for sustainable development on a global scale to the year 2050, there were many Working Groups. Murray Gell-Mann asked if I would direct the Theoretical one. Having secured from Murray a guarantee that it would not involve any administration (“ordering tuna sandwiches for everybody”), I agreed, but in truth, the responsibilities of my directorship quickly reduced to...

  11. Chapter 4 UNDERSTANDING ANASAZI CULTURE CHANGE THROUGH AGENT-BASED MODELING
    (pp. 90-116)
    Jeffrey S. Dean, George J. Gumerman, Joshua M. Epstein, Robert L. Axtell, Alan C. Swedlund, Miles T. Parker and Stephen McCarroll

    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. Agent-based 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...

  12. Chapter 5 POPULATION GROWTH AND COLLAPSE IN A MULTIAGENT MODEL OF THE KAYENTA ANASAZI IN LONG HOUSE VALLEY
    (pp. 117-129)
    Robert L. Axtell, Joshua M. Epstein, Jeffrey S. Dean, George J. Gumerman, Alan C. Swedlund, Jason Harburger, Shubha Chakravarty, Ross Hammond, Jon Parker and Miles Parker

    As the only social science that has access to data of sufficient duration to reveal long-term changes in patterned human behavior, archaeology traditionally has been concerned with describing and explaining how societies adapt and evolve in response to changing conditions. A major impediment to rigorous investigation in archaeology—the inability to conduct reproducible experiments—is one shared with certain other sciences, such as astronomy, geophysics, and paleontology. Computational modeling is providing a way around these difficulties.¹

    Within anthropology and archaeology there has been a rapidly growing interest in so-called agent-based computational models (Gilbert and Doran 1994; Gilbert and Conte 1995;...

  13. Chapter 6 THE EVOLUTION OF SOCIAL BEHAVIOR IN THE PREHISTORIC AMERICAN SOUTHWEST
    (pp. 130-143)
    George J. Gumerman, Alan C. Swedlund, Jeffrey S. Dean and Joshua M. Epstein

    A central question that anthropologists have asked for generations concerns how cultures evolve or transform themselves from simple to more complex forms. Traditional study of human social change and cultural evolution has resulted in many useful generalizations concerning the trajectory of change through prehistory and classifications of types of organization. It is increasingly clear, however, that four fundamental problems have hindered the development of a powerful, unified theory for understanding change in human social norms and behaviors over long periods of time.

    The first of these problems is the use of whole societies as the unit of analysis. Group level...

  14. Prelude to Chapter 7: GENERATING PATTERNS IN THE TIMING OF RETIREMENT
    (pp. 144-145)

    What is the connection between individual rationality and aggregate efficiency? And what is the role of local interactions and social networks in determining that connection? Regarding the first question, the opening Generative chapter argues that individual rationality is neither necessary nor sufficient for the attainment of macroscopic efficiency. The two are logically independent; neither implies the other. The retirement model furnishes the necessity half of the independence proof: a society of autonomous agents arrives at the economically optimal retirement behavior even though the overwhelming majority do not optimize individually. More prosaically,the invisible hand does not require rational fingers. In...

  15. Chapter 7 COORDINATION IN TRANSIENT SOCIAL NETWORKS: AN AGENT-BASED COMPUTATIONAL MODEL OF THE TIMING OF RETIREMENT
    (pp. 146-174)
    Robert L. Axtell and Joshua M. Epstein

    Though motivated by a policy question, this work has theoretical dimensions. There are two related theoretical issues. One is the connection between individual rationality and aggregate efficiency—between optimization by individuals and optimality in the aggregate. The second is the role of social interactions and social networks in individual decision making and in determining macroscopic outcomes and dynamics. Regarding the first, much of mathematical social science assumes that aggregate efficiency requires individual optimization. Perhaps this is why bounded rationality is disturbing to some economists: they implicitly believe that if the individual is not sufficiently rational, it must follow that decentralized...

  16. Prelude to Chapter 8: GENERATING CLASSES WITHOUT CONQUEST
    (pp. 175-176)

    The preceding chapter concerned norms of retirement. We turn now to norms of a different, overtly discriminatory, type. In the Classes model, initially meaningless “tags” acquire socially organizing salience: tag-based discriminatory norms arise, and persist for long periods. The Classes model illustrates once more a core theme of the Generative chapter: that simple rules of individual behavior can generate—or map up to—macroscopic regularities, in this case, class structures. Of course, social classes can arise through outright conquest and subjugation. But, it is notable that they can “self-organize” as well. Indeed, a nice title for chapter 8 would have...

  17. Chapter 8 THE EMERGENCE OF CLASSES IN A MULTI-AGENT BARGAINING MODEL
    (pp. 177-195)
    Robert L. Axtell, Joshua M. Epstein and H. Peyton Young

    Norms are self-enforcing patterns of behavior: it is in everyone’s interest to conform given the expectation that others are going to conform. Many spheres of social interaction are governed by norms: dress codes, table manners, rules of deference, forms of communication, reciprocity in exchange, and so forth. In this chapter we are interested in norms that govern the distribution of property. In particular, we are concerned with the contrast betweendiscriminatory norms, which allocate different shares of the pie according to gender, race, ethnicity, age, and so forth, andequity norms, which do not so discriminate. An example of a...

  18. Prelude to Chapter 9: GENERATING ZONES OF COOPERATION IN THE PRISONER’S DILEMMA GAME
    (pp. 196-198)

    The famous two-person Prisoner’s Dilemma (PD) is widely assumed to raise the problem of cooperation in an arresting way (no pun intended). Two strategies are available to each player: cooperate or defect. In the one-shot game, the dominant strategy is to defect, even though higher payoffs would accrue to each individual were they to cooperate. As when Oedipus solves the riddle of the Sphinx, individual rationality leads to a suboptimal outcome. It is a very elegant puzzle and leads to a core question in social science: How can cooperation evolve in populations whose pairwise interactions have the PD payoff structure?...

  19. Chapter 9 ZONES OF COOPERATION IN DEMOGRAPHIC PRISONER’S DILEMMA
    (pp. 199-224)
    Joshua M. Epstein

    The prisoner’s dilemma (PD) game raises the problem of cooperation in a stark form. Two strategies are available to each player: cooperate or defect. The payoff to mutual cooperation (R) exceeds the payoff to mutual defection (P). But the highest payoff (T) goes to one who defects against a cooperator, while the lowest payoff (S) goes to one who cooperates against a defector. The lettersT,R,P, andSare used to denote the Temptation to defect, the Reward for mutual cooperation, the Punishment for mutual defection, and the Sucker’s payoff accruing to a sole cooperator. WithT>R>P>S, the...

  20. Prelude to Chapter 10: GENERATING THOUGHTLESS CONFORMITY TO NORMS
    (pp. 225-227)

    Let me try to get at something big with something small. Here are two small experiments: (1) Stand four inches too close to someone when conversing. (2) The next time someone at a conference table makes a clever academic quip, try laughing heartily for ten full seconds after the obligatory polite chuckles have subsided. In both cases, you’ll feel uncomfortable. Why? Because we have a very finely tuned sense of what’s inside a norm and what’s outside it. But we don’tthink aboutthat boundary in any conscious way; not, that is, until it is crossed.

    We don’t think about...

  21. Chapter 10 LEARNING TO BE THOUGHTLESS: SOCIAL NORMS AND INDIVIDUAL COMPUTATION
    (pp. 228-244)
    Joshua M. Epstein

    When I’d had my coffee this morning and went upstairs to get dressed for work, I never considered being a nudist for the day. When I got in my car to drive to work, it never crossed my mind to drive on the left. And when I joined my colleagues at lunch, I did not consider eating my salad barehanded; without a thought, I used a fork.

    The point here is that many social conventions havetwofeatures of interest. First, they are self-enforcing behavioral regularities (Lewis 1969; Axelrod 1986; Young 1993a, 1995). But second, once entrenched, we conformwithout...

  22. Prelude to Chapter 11: GENERATING PATTERNS OF SPONTANEOUS CIVIL VIOLENCE
    (pp. 245-246)

    This work grows out of a long-standing interest in security questions. My first three books¹ dealt exclusively with these issues, and developed mathematical models of combat, the Adaptive Dynamic Model (1985)² most notably. My fourth book,Nonlinear Dynamics, Mathematical Biology and Social Science(1997),³ included three chapters on conflict. In one of them, I explored the use of nonlinear ordinary differential equations and nonlinear reaction diffusion systems, all drawn from mathematical epidemiology, to model the spread of revolutions specifically. That chapter concluded with the thought that, beyond these methods, “it would also be interesting to attempt the formulation of such...

  23. Chapter 11 MODELING CIVIL VIOLENCE: AN AGENT-BASED COMPUTATIONAL APPROACH
    (pp. 247-270)
    Joshua M. Epstein

    This article presents an agent-based computational model of civil violence. For an introduction to the agent-based modeling technique, see Epstein and Axtell 1996. I present two variants of the civil violence model. In the first a central authority seeks to suppress decentralized rebellion. Where I use the term “revolution,” I do so advisedly, recognizing that no political or social order is represented in the model. Perforce, neither is the overthrow of an existing order, the latter being widely seen as definitive of revolutions properly speaking. The dynamics of decentralized upheaval, rather than its political substance, is the focus here.¹ In...

  24. Prelude to Chapter 12: GENERATING EPIDEMIC DYNAMICS
    (pp. 271-276)

    I love mathematics. I also believe that mathematical theories can offer fundamental insights. In areas where there is a well-developed and powerful mathematical theory, one should master the theory before building an agent model. Moreover, it is often of great value to “dock” the agent model—or some special case of it—to the classical mathematical one.

    In the field of epidemiology, there is a beautiful underlying mathematical theory that applies powerfully to an important class of cases. While we offer an agent-based epidemic model in chapter 12, the effort began with a “docking exercise” worth recounting as a prelude....

  25. Chapter 12 TOWARD A CONTAINMENT STRATEGY FOR SMALLPOX BIOTERROR: AN INDIVIDUAL-BASED COMPUTATIONAL APPROACH
    (pp. 277-306)
    Joshua M. Epstein, Derek A. T. Cummings, Shubha Chakravarty, Ramesh M. Singha and Donald S. Burke

    Since the September 11, 2001, terrorist attacks in New York and Washington and the subsequent anthrax outbreaks on the east coast of the United States, bioterror concerns have focused on smallpox. Routine smallpox vaccinations in the United States ended in 1972. The level of immunity remaining from these earlier vaccinations is uncertain but is assumed to be degraded substantially. For present modeling purposes, we assume it to be nil.

    As a weapon, smallpox would be very different from anthrax. Anthrax is not a communicable disease. Smallpox is highly communicable. With a case fatality rate of roughly 30 percent (meaning that...

  26. Prelude to Chapter 13: GENERATING OPTIMAL ORGANIZATIONS
    (pp. 307-308)

    Someone once asked the intriguing question, “What is Beethoven’s Ninth?” Surely, it is not merely the printed orchestral score of Beethoven’s Ninth, since the Ninth Symphony is a beautiful piece of music, while the score is a silent pile of paper with ink marks all over it. By the same token, there are as many audible realizations of that single printed score as there are conductors and orchestras (each with their idiosyncratic tempi, dynamics, phrasings, and other expressive particularities), open air amphitheaters, and intimate concert halls (each with their individual acoustics). It seems to me that “Beethoven’s Ninth” can only...

  27. Chapter 13 GROWING ADAPTIVE ORGANIZATIONS: AN AGENT-BASED COMPUTATIONAL APPROACH
    (pp. 309-344)
    Joshua M. Epstein

    What constitutes anadaptive organization? What would constituteoptimalstructural adaptation in a dynamic environment? Can one “grow” optimally adaptive organizations from the bottom up—that is, devise rules of individual behavior thatendogenously generateoptimal structural adaptations? There is, of course, a large literature on the origin of firms, on the size distribution of firms in an economy, and on a host of related topics.¹ However, I am unaware of any explicit model in whichindividual agents endogenously generate internal organizational structures that adapt optimally to dynamic environments. The present chapter develops such a model, using the agent-based technique...

  28. CODA
    (pp. 345-348)

    No one who is still growing intellectually ever feels that he has said all he can in a book. I suppose, therefore, that I should take consolation in the sense of incompleteness I feel in arbitrarily closing the discussion at this point; it is a sign of life.

    In a nutshell, I have tried to advance an argument for generative social science and to demonstrate its principal scientific instrument—the agentbased model—in a wide range of applications. If that argument is not now persuasive (and for some it will not be), its repetition at this juncture will not make...

  29. INDEX
    (pp. 349-356)