The Verbal Model

1. Finnemore and Sikkink 1998.

2. See Ruggie 1998; Wendt 1999; Adler 1997; Checkel 1998.

3. I do not make an attempt to delineate multiple strains of social constructivism—see Adler 1997; Guzzini 2000—nor do I attempt to fully capture all intraconstructivist debate—see Checkel 1998; 2001; Sending 2002.

4. Crawford and Ostrom 1995.

5. Finnemore and Sikkink 1998, 891.

6. These characteristics are drawn from a wide reading of norms literature from rationalist and constructivist viewpoints (see e.g., Ensminger and Knight 1997; Ostrom 2000; Bicchieri et al. 1997; Finnemore and Sikkink 1998; Katzenstein 1996; Kratochwil 1989; Nadelman 1990).

7. This is well known in nonconstructivist approaches as well. See especially Epstein 2000.

8. See Kratochwil 1989; Katzenstein 1996; Finnemore 1996a.

10. Popper denoted this World Three. See Popper 1982.

11. Yee 1996, as quoted in Adler 1997, 327.

12. See Kratochwil 1989, 25-28.

13. The term logic of appropriateness originated with March and Olsen


14. For elaboration on this see Checkel 2001; Guzzini 2000; Sending


15. March and Olsen 1989, 22.

16. Checkel 1998; 2001; Risse, Ropp, and Sikkink 1999; Guzzini 2000.

17. Finnemore and Sikkink 1998, 901. This is not to say that norms cannot survive deviation.

18. See e.g., Cederman 1997; 2003; Bak and Chen 1991; Brunk 2001; 2002; Hoffmann 2003a.

19. Finnemore and Sikkink 1998, 902.

22. Manicas 1997, 11.

23. Checkel 1998, 342.

24. Johnston 1999, 11. See also Johnston 2001.

25. Rosenau 1990.

26. See e.g., Nadelman 1990; Young 1991; 1999; Moravcsik 1999a; 1999b; Lustick 1993; Schneider and Teske 1992; Bianco and Bates 1990.

27. They discuss possible motivations (i.e., empathy, altruism) and external forces (i.e., socialization) but they do not address processes within the agents being persuaded or socialized.

28. Sending 2002, 445.

29. See Johnston 2001; Risse, Ropp, and Sikkink 1999; Checkel 2001; Bernstein 2001.

30. For interesting arguments on the need for political science to move beyond Newtonian models see Gaddis 1996; Bernstein et al. 2000; and Hoffmann and Riley 2002.

31. This section is designed to briefly introduce complexity theory in order to use aspects of it in building an analytical framework—it is not a comprehensive exposition on this vast subject.

32. Holland 1992, 17.

33. Simple here is used in comparison to the more complex aggregations. For instance, the behavior of a firm (an agent in the economy) is less complex than the behavior of the economy as a whole, even though the firm is itself a complex agent.

34. Holland 1992; 1995 discusses this concept.

35. Holland 1995; Mitchell 1996; Nadel and Stein 1991; 1992; 1993; 1995; Cowan, Pines, and Meltzer 1994; and Epstein and Axtell 1996. For political science applications see Axelrod1997; Jervis 1997; and Cederman 1997; 2003.

36. The aggregated results of agent actions in a complex system are "emergent" properties—characteristics that are irreducible to the sum of independent agent actions.

37. See Holland 1995 and Gell-Mann 1994a. Schema is a familiar term in political science—see Khong 1992—but it is used slightly differently here. Holland's and Gell-Mann's formulations are reminiscent of and compatible to similar ideas drawn from psychology: knowledge structures, etc. See Piaget 1971; 1977; Minsky 1985; and Schank and Abelson 1977.

38. Emergence is an oft-debated and imprecise concept. For an introduction to emergence and emergent processes, see Holland 1998.

39. Applications differ on how much of an agent's internal rule model is hardwired.

40. Agents are not always or even usually treated as unitary. When dealing with meta-agents (agents composed of other agents) Subagents within the agent do evaluation.

41. The evaluation stage of complex adaptation is crucial. At this stage, agents alter their rule models, which is key for understanding how ideas become an ingrained part of internal rule models. In addition, however, this stage adds variation in a population of agents, because different agents may have different evaluation processes, and different criteria for fitness.

42. Arthur 1994b, 406. Arthur cites: Schank and Abelson 1977; Rumelhart 1980; Bower and Hilgard 1981; and Holland, Holyoak, Nesbitt, and Thagard 1986 as works from psychology supporting the inductive view of human reasoning.

43. Gell-Mann 1994a, 18-19.

44. March and Olsen 1984, 745.

45. Gell-Mann 1994a, 18-19.

46. In computer modeling, the initial composition of the internal rule models is obviously initialized with specific rules, but they can and, in most modeling examples, do change over time as the agent learns about her environment.

47. New rules can come from outside the agent via mimicry, persuasion, coercion, or suggestion. Alternatively, they can come from inside the agent via trial and error or through mutation, a genetics metaphor complexity theorists sometimes use.

48. Kauffman 1995, 27.

49. Rosenau 1981, 7.

50. This can take place explicitly or implicitly. With explicit evaluation an agent takes stock of outcomes and alters/reifies her rules. Implicit evaluation often takes place at the population level with a natural selection mechanism/metaphor. Agents' reproduction is based on how well they did in previous rounds and the rule model distribution in the population evolves over time.

51. Bernstein 2001 discusses the manner in which new norms fit with existing social structures as a possible mechanism for "selection." This is a complementary viewpoint to complexity theory, but the problem still remains as to how the evaluation of normative fit takes place.

53. Dieter-Opp 1982, 144.

54. Holland 1995, 53-57.

55. Thinking of politics in adaptive terms is not a new endeavor. See Rosenau 1981.

56. See Klotz 1995a; Finnemore 1996a; 2003; Bernstein 2001; Price 1997.

57. Stein 1989, xiii-xiv.

58. Arthur 1994a; and Pierson 2000.

59. There is also a strong relationship between coevolution and path dependence. Coevolution and initial conditions lead to path dependency. The classic example is the QWERTY keyboard. Initially, for mechanical reasons the QWERTY keyboard was developed. Typists and typewriter manufacturers coe-volved with this new technology and technique until all other types of keyboards became "extinct." Now the QWERTY keyboard is the only path left—even though it is not necessarily the most efficient way to set up a keyboard. There is no way, by just looking at the keyboard, to determine why it is the way it is. The initial conditions, coevolution, and path dependency are too important.

60. Giddens 1984, 27.

62. Arthur 1994a, 27.

63. Giddens 1984, chapter 5.

64. For cogent analyses of the construction of "natural" taboos on chemical and nuclear weapons see Price 1997; and Price and Tannenwald 1996.

65. Holland 1992.

67. Rosenau 1995; 1997b; Hoffmann 2003b; 2003c.

69. This is not to imply that all agents are equally able to construct world politics. Power is obviously an enormous factor in the construction of world politics. U.S. actions often have a greater impact on structures than do actions of other states. See Hoffmann 2003c.

70. Holland 1992, 18.

71. Weldes 1996, 281.

73. Manicas 1997, 11.

74. Giddens 1984, 5.

76. Sandholtz 1998, 8.

77. This dual reality also fits well with the sociology of knowledge strain of thought that preceded constructivism. In it, Berger and Luckmann make clear that social facts are real and external to agents. They posit that "the reality of everyday life appears already objectified, that is constituted by an order of objects that have been designated as objects before my appearance on the scene." However, this world is dependent upon human action that results from and is directed by the internalization of the objective social order that the action itself creates and sustains. See Berger and Luckmann 1966, 21-22, 60-61.

78. On internal rules models see Holland 1995; Gell-Mann 1994a. On schema see Khong 1992; Piaget 1971; Schank and Abelson 1977.

79. Holland 1995, 34-37.

81. Giddens 1984, 22.

82. See Bernstein 2001; Florini 1996.

83. Ullmann-Margalit 1977, 8.

84. I am not claiming that ontologically primitive agents stand outside themselves picking interests—the agents are always in the process of figuring out what is appropriate and this process is constrained and constituted by the agent's past experiences. The constitution of agent interests is never completed but this does not have to imply that agents lack the ability to reflect on themselves and their social environment.

85. Risse 2000; Risse, Ropp, and Sikkink 1999; Checkel 2001; Johnston


86. Bernstein 2002, 208.

87. Because this framework is to be used to investigate a specific question, I do not discuss all the various ways that a CAS perspective could inform con-structivist analysis. Instead, I concentrate on norm emergence and the initial influence that the spread of the universal participation requirement had on the climate change negotiations.

88. Kratochwil 1989, 61.

89. Holland 1995, 53-56.

91. Of course, external evaluation and persuasion, socialization, and coercion take place as well.

92. In the cases at hand, persuasion is the most prevalent mechanism. For more on persuasion see Johnston 2001; and Risse 2000.

93. It is possible to encounter infinite regress with this analysis. The United States is an agent with rule models, but it is also a structure for other agents acting within it. Those other agents have rule models that cause their behavior (evaluating U.S. actions) and alter their structure (U.S. rule models). In the analysis of the ozone depletion and climate change cases, I limit the regress and merely focus on the evaluations that come from various groups in the United States.

94. CAS perspectives consider that agents have both active and latent rules in their rule models.

0 0

Post a comment