International Norm Complexity: Modeling the Evolution of Norms, Communities, and Action in the European Union Presentation to the DIMETIC Maastricht Session 16 October 2008 Mark Rouleau PhD Candidate Dr. Claudio Cioffi-Revilla Dissertation Advisor George Mason University Center for Social Complexity 379 Research 1 Building, MS 6B2 Fairfax, VA 22030 USA Correspondence: mrouleau@gmu.edu
Table of Contents Motivation... 2 Background: Review of Literature... 2 Theoretical Framework... 5 Principal Research Hypotheses... 9 Methodology... 12 Case Studies... 16 Concluding Remarks... 20 Bibliography... 22 Notes... 29 Proposed Dissertation Committee: Claudio Cioffi-Revilla (Chair) George Mason University, Computational Social Science Maxsim Tsvetovat (ABM) George Mason University, Computational Social Science Desmond Dinan (EU Integration) George Mason University, School of Public Policy Matthew Hoffmann (Social Constructivism) University of Toronto, Political Science 1
Motivation Constructivism has successfully challenged the traditional political science understanding of international relations (IR) with the introduction of social norms. However, constructivism has remained largely a counterargument to traditional IR without advancing a viable (i.e. formal and testable) theory of its own. This has had two important negative consequences for constructivist research. First, it has made the validation of constructivism imprecise and woolly. The lack of a formal framework has often led to the production of vague and unfalsifiable assumptions. Second, constructivism s quest to counter traditional theory has encouraged a copious research concentration on international cooperation at the expense of understanding the mechanisms behind social deviance. This focus on conformance has prevented constructivism from fully grasping the importance of social complexity in the evolution of norms. My proposed dissertation will overcome both of these problems through the synthesis of constructivism, complexity theory, and Agent-Based Modeling (ABM). The proposed research will formalize a new ontological framework that will fortify and expand upon the core theoretical assumptions of constructivism. I will use ABM to test the theoretical limitations of the proposed framework. The results of the ABM simulation will then guide the application of the proposed framework to the study of two empirical examples of European integration. The first case will explore the constitutional development of the Common Foreign and Security Policy (CFSP) pillar of the European Union (EU). The second case will investigate the constitutionalization and promotion of the respect for human rights within the EU. The resulting dissertation will illustrate the value of complexity theory and ABM in strengthening the constructivist paradigm. Background: Review of Literature I begin with a brief review of the theory, methodology, and empirical background motivating the proposed work. On the theoretical side, I propose to reinforce and extend the IR paradigm of social constructivism. Constructivism has proven to be an important alternative to traditional IR theory. i It has successfully contested the conventional, static conception of international relations. ii It has also highlighted the influence of norms shared and evolving conceptions of appropriateness amongst actors of a given identity in the shaping of state interests. iii Constructivists have uprooted the standard IR logic of consequences and replaced it with a dynamic logic of appropriateness, which accounts for systemic change. iv In this way, constructivists have recast interest formation into an endogenous process where social interaction guides state behavior. v Yet, despite the achievements of constructivism, certain limitations of the paradigm have been difficult to overcome. Constructivism has a number of weaknesses that stem from the eclectic development of the paradigm over the past two decades. Constructivism surfaced in pockets, within IR, 2
primarily through independent challenges to neo-realist and neo-liberal theory. vi This lack of synchronization left constructivism without a single unified framework. As a result, variation exists in the interpretation of the paradigm and its terminology. To critics, this can have the effect of complicating the communication of constructivist research. vii Communication is further obscured by the lack of formalism within constructivism. Lacking a formal framework for causal mechanisms, constructivists have struggled to devise an uncontroversial depiction of the role of norms in international relations. viii Martha Finnemore and Kathryn Sikkink have created a concise constructivist framework known as the Norm Life Cycle (NLC) to tackle this problem. ix The NLC was an important contribution to constructivist literature because it provided a formal definition of the macro influence of norms upon actors striving to maintain social acceptance. It has done so by framing the emergence and evolution of norms within a three-stage path of norm development. x This path begins with norm entrepreneurship, leads to the formation of a critical mass of followers, and ends with a norm cascade yielding norm internalization. The NLC framework added a necessary level of precision to the constructivist paradigm. xi However, the NLC did little to improve constructivism s ability to substantiate abstract claims. Theoretical validation remained woolly due to the reliance on historical reinterpretation and the ambiguity of vital constructivist concepts. xii Thus, despite Finnemore and Sikkink s laudable efforts, critics continued to focus on the weakness of constructivism s methodological approach rather than assessing its theoretical capability. xiii Presently, constructivism still lacks a reliable procedure to verify and expand upon its theoretical claims. Verbal argumentation can take the paradigm only so far. However, recent research by Matthew Hoffmann suggests how constructivists may turn to alternative methodologies, such as ABM, to overcome this predicament. xiv Hoffmann saw great potential in the NLC framework but he also saw two areas for improvement. First, he argued that the NLC framework defined the macro structure of the evolution of norms but it did not explain the micro mechanisms driving individual actors. His work drew upon complexity theory to produce an adaptive decision-making scheme that could solve this problem. xv Second, Hoffmann understood that the NLC lacked a sufficient means to explore the validity of its theoretical assumptions. Thus, Hoffmann used agentbased experimentation to test and then support the validation of the verbal NLC model. xvi Hoffmann s work proved it was possible to confirm constructivist assumptions in a positivist fashion within a lab-like environment. xvii The ABM simulation of Hoffmann provides a critical advancement upon the original NLC framework of Finnemore and Sikkink. However, the potential for extending both the framework and the model remains high. For example, the NLC does not problematize the essential impetus for norm change, the norm entrepreneur. xviii It does not explain why norm entrepreneurs appear within the system. Nor does it suggest why norm entrepreneurs are immune to the pressures of the logic of appropriateness. Furthermore, the exogenous nature of norm entrepreneurship in the NLC discounts both unintentional (i.e. not managed or controlled) norm shifts and norm changes within norm 3
entrepreneurs themselves. Finally, recent IR work highlights the flawed logic of treating states as unitary actors. xix Yet, the NLC framework does not provide a formal means to model actors that are merely aggregations of individuals playing multiple roles and/or driven by a diverse set of norms and identities. xx The NLC captures the norm-based actions of individual actors but it does not formalize the aggregation of these actors into complex organizational units such as states. I believe that extending the NLC to address these issues would greatly benefit the study of complex IR topics such as European integration. Ian Manners and a number of EU constructivists have produced numerous works outlining the impact of norms and identities for European integration. xxi In this vein, Jeffrey Checkel and others have confirmed the significance of the relationship between domestic society and political elites in the diffusion of, and compliance with, international norms. xxii Scholars such as Cristoph Meyer have also documented the emergence of a common European identity and the harmonization of norms amongst the individual states of Europe. xxiii However, as with most constructivists, the key weakness of the EU group has been the informal application of constructivist logic to complex social topics. Critics have labeled this work esoteric and unfalsifiable. xxiv Thus, regardless of the depth or ingenuity of this research, its primary failure has been the inability to rigorously test theoretical claims in a scientific manner. I believe that the formal framing and testing of constructivism, following the example of Finnemore, Sikkink, and Hoffmann, provides a solution to this problem. I also argue that the proposed extension to the NLC will be necessary for the use of this framework in the study of European integration. Scholars of the multi-level governance approach have shown that norm evolution in the EU involves much more complexity than that which is represented in the simplified NLC framework of Finnemore and Sikkink. xxv These scholars have argued that understanding the internal dynamics of states can be just as important as grasping the role of systemic pressure in the life cycle of norms. xxvi Therefore, I argue that the NLC must take a cue from this work and open up the black box of the state. It must attempt to define the social overlap amongst internal and external relations. In this way, the NLC will be able to investigate the flow of norms both horizontally (i.e. amongst actors) and vertically (i.e. within complex actors) to establish a comprehensive picture of norm diffusion and internalization. This extension will provide a better indication of the limits and potential range of both the NLC and constructivist logic. It will also permit the NLC to relax its dependence on unified state actors and to explore problems where state sovereignty is in question, such as the case of European integration. Given the complexity of this extension, an ABM formalization of concepts and causal mechanisms will be crucial for the proposed framework to be both useful and convincing. I will also endeavor to keep the work germane to EU integration by applying the proposed framework to two empirical case studies. European integration is, of course, a massive subject. EU scholars have often made it more digestible by focusing on integration through the lens of the constitutional process. xxvii Constitutional development provides a formal documentation of intentions 4
and observable institutional change. Actors openly reveal preferences and affiliations as they debate constitutional amendments. I propose to leverage this data to trace the evolution of foreign policy norms within the framework of the emerging EU constitution. My first case study will explore the norm of collective security in the development of the EU s CFSP pillar. The study will use the NLC extension to explain how and why individual state perceptions of collective security have shaped and been shaped by the constitutional process. This will help us to better understand the evolution of the CFSP pillar as a product of normative interaction. My second case study will investigate the norm of human rights, the push to constitutionalize this norm as a founding principal of the EU, and the emergence of the EU as an international human rights norm entrepreneur. Previous work in this area has detailed the significance of a human rights norm within Europe. xxviii However, this research has done much less to explain the diffusion, evolution, and internalization of this norm. The proposed study will use the NLC extension to demonstrate the role of norm entrepreneurs in the development of the EU s human rights norm and in the efforts to codify this norm within succeeding treaties on European union. The study will also highlight the movement to create the Charter of Fundamental Rights and its effects upon the emergence of the EU as an entrepreneurial actor in its own right. Yet, before delving into the details of these case studies, it is necessary to first layout the proposed theoretical framework from which to understand these events. Theoretical Framework As noted above, Finnemore and Sikkink s NLC framework requires a better representation of organizational and social complexity to explore European integration. I propose to follow the example of Herbert Simon xxix in incorporating complex social entities the state or other international organizations into the NLC. Simon helps us to understand the role played by simple units in aggregate structures. The key to his work is a feature he termed near decomposability. Simon describes near decomposability as an advanced organizational trait with two important characteristics: 1) a hierarchical ordering of units and 2) a loose coupling of interaction amongst units. xxx I see great value in introducing near decomposability into the ontological framework of constructivism. This will strengthen our understanding of cognitive dissonance and entrepreneurship in the evolution of norms. It will also outline the relevant norm actors (i.e. the norm entrepreneur and the target audience), the likelihood of entrepreneur success (i.e. possible norm conflicts to leverage), and the potential for the percolation of norms within a particular social environment. Given that Simon s work did not apply directly to IR, the first step of the NLC extension is to redefine the relevant norm actors of an international system, like the EU, in terms of near decomposability. To do this, we must first define the basic unit of social interaction. I argue that this fundamental component is the individual and not the state, as claimed by traditional IR theory. xxxi I believe that individuals are at the core of norm evolution because it is ultimately through individuals that norms flow. Individuals are also the true 5
action takers of the international system. It is individuals, not states, that are responsible for norm entrepreneurship and internalization. xxxii Finally, the executive actions of individuals define the behavior and normative positions of aggregate actors. xxxiii Therefore, it is crucial that the proposed framework formalizes the individual within the internal dynamics of aggregate decision-making. Yet, individuals are clearly not the only social actors within the international system. Lone individuals are only capable of managing political issues of relative simplicity and narrow social scope. xxxiv As issue complexity and social scope increases, individuals form larger political units, with more far-reaching institutional arrangements, to address wider social problems. xxxv The proposed framework will represents each of these political units as a complex aggregation of individuals, or lower level aggregate units, couched within a hierarchy of organizational ties. Each hierarchical order will follow the loose coupling of Simon s near decomposability. xxxvi Executive agents following the demands of domestic and international pressure will control the actions of these aggregate actors. xxxvii This form of organizational complexity is empirically present in Europe where individuals make up the population of states and states, represented by individuals, participate in international organizations of varying social scope (i.e. the Western European Union, the EU, NATO, the United Nations, etc.). As will be shown below, Europe also presents a visible example of structural transition with interesting consequences for norm evolution. For a static view of this world, see the graphic interpretation of the proposed framework extension in figure 1. xxxviii 6
Figure 1. This diagram is a rough interpretation of the nearly decomposable international system. Within the diagram, we see that individuals serve as the fundamental units of the system. They also serve as the representative voice of aggregate collectivities such as states, Non-Governmental Organizations (NGO), and International Organizations (IO). Individuals are the only units within the system that possess spatial relationships to other individuals. The geographic place of citizenship defines the individual s allegiance to a single state and to the other individuals within this state. Units above the individual have only social relationships with other units in the system. Leader or representative individuals often serve as the filter through which relations flow inside of an aggregation unit such as the state. Finally, the two axes of this diagram indicate the application of this hierarchical structure in regards to a single international issue such as the respect of prisoners of war. The x-axis denotes the spatial scope of this problem and the y-axis denotes the social scope or aggregate complexity of the organizational flow used to address this problem. However, it is important to note that this diagram represents the relevant social actors for only one international issue. Each issue may have its own set of relevant actors but each issue will also follow an organizational structure similar in nature to this nearly decomposable framework. One important point to note is that this near decomposable ontology is not intended to replace the logic of appropriateness of social constructivism. In fact, I claim that it should be possible to consistently apply the logic of appropriateness to all units of the nearly decomposable international system, as it is now applied to states. Importantly, this detailed ontological framework permits us to investigate the multiple and overlapping spheres of appropriateness actors face within the international system. This gives us a better understanding of the locus of norm conflict and, hence, the impetus for entrepreneurship activity leading to the evolution of norms. To illustrate how the proposed NLC extension fills the gaps of the initial NLC, I offer a hypothesized narrative of the emergence of entrepreneurship activity. This narrative uses near decomposability to provide a logically consistent constructivist explanation of norm entrepreneurship and norm change. The narrative begins by explaining the drive behind entrepreneurship activity. I argue that norm entrepreneurship arises in response to new and potentially conflicting interpretations of appropriate behavior. These new normative interpretations appear within the social environment in two ways. First, as understood by most constructivists, actors generate new interpretations in response to stimuli or encounters that have no experiential precedence. xxxix Second, often left unnoticed by constructivism, actors also face interpretations new to their quasi-isolated social environment when nearly decomposable units interact. In this way, the impetus for entrepreneurship lies in the intermingling of opposing norm interpretations. Yet, this does not explain why actors would internalize new interpretations of appropriateness in opposition to the currently entrenched norms of their social environment. Where do social deviants, such as norm entrepreneurs, fit into the logically appropriate world defined by constructivism? We must look to the near decomposable makeup of aggregate actors to answer this question. I believe that the inconsistent overlap between organizational units and social entities serves as the locus of social deviance and, hence, the motivation for norm entrepreneurship. A brief look at the social life of norms will show how this claim avoids deviating from the logic of appropriateness to define the emergence of norm entrepreneurs. As sociologists inform us, groups of socially interactive individuals provide the medium in which norms survive. xl Therefore, we can further assert that the set of individuals who accept a given norm define a norm group (see figure 2). The 7
members of these norm groups often share overlapping membership with other norm groups. This norm group overlap serves to delineate the bounds of norm communities a group of overlapping norm groups whose members become constrained by a package of norms, tied to multiple issue sectors. xli However, norm groups and communities frequently span and overlap aggregations of organized units such as the state. I believe that the difference between this social and organizational overlap provides the impetus for entrepreneurial development and hence norm evolution. In other words, the complex social overlap within aggregate norm actors provides the positive feedback from one perspective needed for social deviance from another perspective to develop and expand. Figure 2. This diagram is a bird s-eye-view of the political organizational hierarchy from figure 1 with hypothetical norm groups included. The hypothetical norm groups refer to a single international issue similar to the situation depicted in figure 1. I have omitted the social relations from this diagram for greater clarity. However, one thing to remember is that norm groups require members to possess at least one social connection to another member of the same norm group. The aim of this diagram is to illustrate the splitidentity makeup of states in terms of a single international issue. We can see from this graphic that endowing states which merely represent aggregates of individuals with normative agency is a great mischaracterization of international relations. States are complex entities. Some states may be highly diverse in terms of interpreting norms while other may be rather homogenous. This split-personality nature of states drives norm change in the international system with pressures coming from both the domestic and international spheres. Norm groups and norm communities shape the complex motivations for action within any social system. xlii Individual actors respond to the pressures of a norm group or 8
community as a way to remain socially accepted. xliii Yet, maintaining social acceptance is a complicated process not fully understood by constructivist frameworks such as the NLC. The NLC ineffectually deals with the fact that individuals identify with multiple entangling norm pressures. The overlap in norm group membership amongst aggregate organizational units further complicates this process by linking diverse actors to a chain of normative commitments such as a norm-package. xliv This specific understanding of entrepreneurship allows us to formalize the means by which noise or cognitive dissonance enters and remains semi-stable within the social system. Individuals attempt to align their experiences with the many cues put forth by their multi-part identities but often find themselves unable to please everyone at the same time. Social overlap forces actors into an uneasy middle position that provides the impetus and/or support of entrepreneurial activity. Finally, I claim that the meta-stability (i.e. high entropy) of norm evolution is the result of the loose coupling of units within the international system. Loose coupling amongst units slowly exposes individual actors to the normative interpretations of other actors from other norm groups or communities. Once introduced into a unit, these new interpretations either gain further acceptance or gradually fade. Norm internalization is dependent upon the potential for norm acceptance in the heterogeneous and overlapping affiliations of the actors of a given unit. Furthermore, before total acceptance or fade comes about, it is often the case that new behaviors again enter the unit. The result is a meta-stable pattern of norm evolution. Principal Research Hypotheses The theoretical arguments put forth in the preceding section lead to a suite of research hypotheses. I propose to validate these hypotheses using ABM and empirical analysis in the following two sections. From the complex and nearly decomposable view of the international system, I generate three sets of incremental and testable research hypotheses. The first set involves three hypotheses that address the minimal properties of the logic of appropriateness. These hypotheses are as follows: Hypothesis 1-A (Expected Homogeneity): given a highly simplified environment, if all actors of a fully linked social system follow the logic of appropriateness for a single norm, then total normative homogeneity will result in all instances. Hypothesis 1-B (Expected Heterogeneity): if no social link exists between units, as defined in hypothesis 1-A, then within-unit homogeneity and unit-to-unit heterogeneity will result. Hypothesis 1-C (Expected Dynamism): given a set of heterogeneous units, as described in hypothesis 1-B, an increase in social ties amongst clusters will produce norm change and will eventually result in system-wide homogeneity. Hypothesis 1-D (Expected Conflict): given a social system as described in hypothesis 1- C, the greater the number of norm interpretations, the longer the period of normative conflict. Furthermore, assuming that actors fuse multiple norms into a package of expected behaviors, the greater the number of norms within a norm-package, the longer the period of normative conflict. 9
These four hypotheses describe the default expectations from social behavior defined by the logic of appropriateness. They state that, with all else constant, social interaction simply leads to homogeneity. Although it is clear that the social world is not quite this simple, these four hypotheses do provide the roots for the generation of cognitive dissonance. Hypothesis 1-B indicates that isolation is the key to heterogeneity in a logically appropriate world. Hypothesis 1-C implies that the introduction of social linkages amongst heterogeneous and isolated units brings about norm change. Finally, hypothesis 1-D provides a clue as to the role of complexity in the evolution of norms. However, a more realistic picture of norms would require some pattern of meta-stable cycling as opposed to a trajectory that consistently leads to system-wide homogeneity. This is why the proposed near decomposable ontology is needed. I argue that the near decomposability of political units provides the proper mixture of isolation and social linkage needed to prevent homogeneity and to ensure meta-stability. The proposed near decomposable ontology draws together the insights from hypotheses 1-B, 1-C, and 1-D to outline the social circumstances needed to generate cognitive dissonance and meta-stability. A near decomposable view of the world leads to the following three hypotheses: Hypothesis 2-A (Unit and System Homogeneity): Let each unit be arranged in an organizational hierarchy with a set of social linkages amongst units. The greater the social linkages within a unit, the higher the probability of unit homogeneity. And, the greater the linkages amongst units, the higher the probability of system-wide homogeneity. Hypothesis 2-B (Within Unit Conflict): An increase in organizational links without an accompanying increase in social links is likely to lead to greater norm conflict. On the other hand, a simultaneous increase in organizational and social links will lead to an initial spike in norm conflict followed by a growing trend towards normative homogenization. Hypothesis 2-C (Nearly Decomposable Meta-Stability): If too few links amongst otherwise isolated units equate to normative stalemate and unit-to-unit heterogeneity, and too many links amongst otherwise isolated units equate to normative flood and systemwide homogeneity, then there must exists a critical mass of links amongst otherwise isolated units that will prevent the system from settling upon a well-defined normative equilibrium over a sizeable interval of time. Hypothesis 2-D (Persistent Meta-Stability): Given a nearly decomposable system as described in hypothesis 2-A, let each unit consist of a large set of actors with dynamic social ties. Let the dynamics of social ties be governed by the interaction potential of the organizational ties of a given individual s unit. A nearly decomposable system of this type will remain meta-stable without permanent settlement upon an absolute normative equilibrium over a sizeable interval of time. The previous four hypotheses describe the conditions needed to disrupt the normative equilibrium expected from the logic of appropriateness. I argue that this normative interference defines the locus of cognitive dissonance. These hypotheses also highlight the potential for the interaction of conflicting norm interpretations (hypothesis 2-B) and the development of meta-stable norm patterns from the unmanaged actions of minimal 10
actors (hypothesis 2-C and 2-D). However, hypotheses 2-C and 2-D require the caveat over a sizeable interval of time to remain consistent with Simon s conception of near decomposability. As Simon would argue, given all else constant and given enough time for norm diffusion to take place, it is expected that system-wide homogeneity will eventually result. Yet, this picture of cognitive dissonance provides the time and social space needed to produce norm entrepreneurs. The next step is to define the role of norm entrepreneurs in the following set of hypotheses: Assume that the probability of norm entrepreneurship is defined by the following equation: P(e) = (s/x) * (1/t) Let P(e) equal the probability of norm entrepreneurship. Let the variables s, x, and t represent support, exposure, and tolerance respectively. Support is a perceived level of acceptance of a given norm interpretation within one s social network. Exposure is a measure of the number of encounters with an alternative norm interpretation. Tolerance is a threshold value drawn from a Power Law or normal distribution that represents the likelihood of an individual actively engaging in protest to alternative norm interpretations. Assume a logic of appropriateness that enforces greater social sanction penalties upon individuals if norm opposition occurs between members possessing a direct social link. Also assume that, once individuals become norm entrepreneurs, the social sanction penalty decreases during entrepreneurship. Hypothesis 3-A (Probability of Norm Entrepreneurship): Given individual actors with the social characteristics described above, the probability of norm entrepreneurship decreases as the linkages amongst units increase to full linkage or decrease towards no linkage. Hypothesis 3-B (Maximum Likelihood of Entrepreneurship): There exists a maximum likelihood of norm entrepreneurship within a social system somewhere between a fully linked and fully isolated state. Therefore, the closer the social system is to a nearly decomposable system, the greater the likelihood of norm entrepreneurship. Hypothesis 3-C (Probability Density of Entrepreneurship): The relationship between the probability of norm entrepreneurship and the connectivity of social units is defined by a log normal distribution. Hypothesis 3-D (Volatility of Entrepreneurship): An increase in norm entrepreneurship activity leads to an increase in norm volatility. Therefore, the lower the average individual tolerance the greater the likelihood of a single entrepreneurial act creating conflict cascades throughout the system. The preceding set of hypotheses describes the role played by norm entrepreneurs in keeping a social system meta-stable. Entrepreneurs provide the necessary agitation to prevent a system from settling upon its eventually expected equilibrium. However, these hypotheses also assert that the success of entrepreneurship in disrupting this equilibrium is dependent upon a suitable level of social connectivity. I argue that a nearly decomposable system provides the ideal level of social interaction needed to maintain cognitive dissonance and norm entrepreneurship. Therefore, as a system moves closer to a fully linked or fully isolated system, the space of cognitive dissonance shrinks, the likelihood of entrepreneurship decreases, and norm evolution diminishes. I believe it is 11
this ability to capture the mechanisms underlying the dynamics of norm change that advances the proposed NLC extension beyond its predecessor. I argue that the near decomposable extension to the NLC advances our formalization of norm development. It provides a tangible boundary to the sometimes-loosely applied assumptions of current constructivist work. It also places critical actors such as norm entrepreneurs into an adaptable but well-structured hierarchy of interaction. Finally, it explains the source of norm deviance and the development of norm entrepreneurship in a way that elucidates constructivist concepts such as cognitive dissonance or noise. With this new framework, we can begin to investigate the stress and consequences of having to manage multi-level, multi-part, and overlapping identities. Consequently, I believe that this framework can explain more of the emergent behavior that results from the assumptions of constructivist logic, especially in regards to norm entrepreneurs. However, to confirm the validity of this complex perspective on norms, I argue that constructivists must develop their research using alternative methodologies designed with complexity in mind. This is why I now turn to propose the use of ABM to test the theoretical conjectures detailed above. Methodology Constructivists and other social scientists have slowly begun to adopt ABM to perform theoretical validation for a number of reasons. xlv For example, unlike EBM, ABM permits the formalization of theory grounded upon stochastic actor interaction. xlvi The object-oriented design (OOD) of ABM provides a clear and well-structured approach to defining complex relationships amongst heterogeneous actors at varying scales of aggregation. xlvii OOD also enables heterogeneity amongst agents. This has two important implications. First, agent activation is able to execute in a quasi-parallel fashion that mimics the simultaneity and disproportionate relations found empirically. xlviii Second, the OOD of ABM does not require ideal-type agents such as those that serve as the foundation to equation-based models. xlix Rather, agent-based models simulate the interactions of realistic imperfect actors that make mistakes and that are not bound to a predefined notion of utility maximization. This feature is critical to the modeling of constructivism because constructivists reject the notion of individualistic actors endowed with an exogenously defined decision-making scheme. In fact, ABM may be the only formal methodology that allows for the materialization of rule-sets at run-time through the adaptation and interaction of individual agents. ABM also permits the formalization of insights from complexity theory. l Consequently, ABM encourages the investigation of non-equilibrium dynamics. This ability to move beyond the strictures of equilibrium permits ABM to explore complex relationships considered mathematically intractable. li This is why I propose to use ABM to investigate the forces of social change. ABM s emphasis on dynamism and the discovery of metastable patterns is quite amenable to the proposed research. lii The proposed study will formalize the role of feedback (i.e. the development of cognitive dissonance) and emergence (i.e. the emergence of norm entrepreneurs and the evolution of norms) within the constructivist paradigm. Such formalization would not be possible within EBM. 12
Equation-based models use projections of the aggregate to predict the future state of the world. This results in washing out the consequences of feedback leading to emergent behavior. In this regard, ABM is a superior formal methodology that supports the study of complex social interactions. ABM is also appealing because of its ability to convincingly simulate the flow of norms. For example, sociologists understand the important distinction between spatial and social ties in the life of a norm. liii Space and social networks both define the potential for interaction and the spread of norms. However, space and social networks do not have the same affect upon actors. One is more likely to place greater weight upon the normative influence of an actor defined by its social network rather than those of its local neighborhood. liv At the same time, space does provide a clue to the likelihood of contact and the potential for exposure to new norms. lv Thus, constructivist models must be able to represent the distinction between spatial and social ties. However, such a separation is not possible within an equation-based model. ABM, on the other hand, is capable of representing this distinction. lvi Finally, researchers have used ABM to improve the scientific rigor of the social sciences in general. This is because ABM presents researchers the means to formally validate theoretical assumptions. lvii It does so by providing the social laboratory desired to test hypotheses. lviii The ability to rerun a stochastic process allows researchers to gain a better understanding of the general principals guiding empirical events. lix Researchers can undertake multiple simulation runs, using a suite of parameter values, to iteratively explore the consequences of changing variables of interest. Lastly, a formal rendering of social science theory will require researchers to tighten assumptions and to ensure a strict definition of concepts. lx This will certainly improve both the communication and validation techniques of social science paradigms such as constructivism. Formalization allows social scientists to engage in experimentation in a way that is not open to qualitative research methods. For the reasons just outlined, I believe that the complex and informal paradigm of constructivism has much to gain from ABM. Therefore, I now turn to propose an agentbased model that will formalize the expanded constructivist ontology and theoretical assumptions described in the three sections above. The target system for this model will be the policy-making domain of the European Union. To reach this target, model development will unfold in three stages. Each stage will incrementally assess the validity of the hypotheses proposed in the previous section. First, the model will verify the outcomes expected from the logic of appropriateness through a set of simple and highly abstract experiments. Next, the model will explore the generation of cognitive dissonance and meta-stability using a nearly decomposable framework to define agent relations. Then, the model will use the agent decision-making scheme proposed in the hypothesis set 3 to explore the emergence of norm entrepreneurs within this nearly decomposable social environment. Finally, the model will focus on organizational change (i.e. integration) and its correlation with policy development. 13
The proposed model will gradually develop into a high-level representation of an international system such as the European Union. The first stage of development will produce a highly abstract model similar to Hoffmann s original NLC simulation. Agents will play the role of individuals and groups of agents will define aggregate units such as the state. Agent decision-making will be minimal and will closely adhere to the logic of appropriateness modeled by Hoffmann. This will require the adoption of Hoffmann s pick a number schema, which converts norm development into a game of matching the behavior of the majority. In this context, the model will test hypotheses 1-A through 1-C by modifying the social connections amongst individual agents as detailed within each hypothesis. The second stage of model development will involve the introduction of near decomposability. At this point, the model will examine various social and organizational structures to determine the conditions needed to bring about cognitive dissonance and meta-stability without the influence of exogenous noise. This stage will begin with the simple aggregate groups of the previous phase of model development. The model will slowly increase the number of links amongst isolated units to determine the window of near decomposability that produces meta-stable norm conflict and evolution. The same experiment will then be performed with increased social and organizational complexity. This will require the division of individual agents into multiple and potentially overlapping aggregate units. The goal will be to recreate cognitive dissonance at the level of the state. The model will use various organizational structures to represent the domestic control of aggregate actors. However, all organizational structures will require select individuals to play the role of executive decision-makers for each aggregate unit. Attempts will also be made to investigate the importance of dynamic social ties, migration amongst units, and the impact of alternative governance regimes upon the evolution of norms and policy. The final stage of model development will draw upon the lessons learned in generating cognitive dissonance within a near decomposable setting. In this stage, the agent decision-making schema will incorporate the norm entrepreneurship functionality proposed in hypothesis set 3. The model will use norm entrepreneurship to address two research questions. First, the model will compare results in this final stage to the nonentrepreneurship stage to formalize the role of norm entrepreneurs in the meta-stable development of norms. Experiments will test the validity of hypotheses 3-A and 3-B, which claim that entrepreneurs prevent the system from settling upon a final normative equilibrium. Second, the model will investigate the consequences of an event that alters the structure of near decomposability such as the integration of Europe. The model will simulate integration by increasing the organizational and social linkages amongst an initially nearly decomposable set of units. This will permit the testing of hypotheses 2-B, 3-C, and 3-D which would posit a preliminary spike in entrepreneurship followed by a growing trend towards normative homogenization. I argue that the results of this final experiment will support and guide the validation of the theoretical claims within the empirical case studies. 14
The final model produced from the proposed design will be a complex formalization of norm evolution. A brief description of the main simulation loop of this model may be helpful. A typical simulation will begin by distributing agents throughout a grid to define spatial relations (i.e. integration experiments will explore the full significance of this parameter). The model will then provide agents with an initial Small World set of social ties. lxi Next, the model will choose the initial executive agents to represent aggregate actors and draw relevant organizational ties to these agents. During each round, the simulation will permit agents to perform a number of actions. First, the simulation will ask activated agents to reevaluate their policy positions (the current rule in Hoffmann s model). Similar to Hoffmann s NLC model, evaluation will require agents to increment the scores of successful rules and decrement the scores of unsuccessful rules. lxii I will define success by the policy positions that find social agreement within the system. However, only local (not global) evaluation will take place. Furthermore, agents will also be required either to execute policy for an aggregate unit (if the agent is an executive agent) or to exercise their right to change the current leadership (voting executives out of office). Finally, the simulation will also ask agents to reevaluate their social ties and to consider norm entrepreneurship. This step will permit agents to add or drop links to other individuals, to form new organizational groups (international organizations), or to promote new normative interpretations in response to the current policy regime. As we can begin to see, social ties will serve as a complex percolation medium for norms within this model. Agents will face pressure from various directions to change their policy perspectives. Bottom-up pressure will come from the organizational ties simulating domestic demands upon executive agents. Lateral pressure will come upon all agents from norm groups or communities. Finally, agents will also face top-down pressure from the international organizations to which they belong. The result will be a complex interaction of norms and policy development (see figure 3). In a setting such as this, we can begin to explore the conditions needed to produce entrepreneurial activity. I believe that the proposed model will be able to do so by providing the complex and multi-faceted social environment needed to sustain the form of social deviance required for norm entrepreneurs. It will do so in a way that explains the existence of norm entrepreneurs, not as immune to constructivist logic, but simply as actors caught within the complex demands of their own social system. However, this exercise in theory testing would also require empirical validation. This is why I propose to apply the model findings to the case of European integration. 15
Figure 3. UML Class diagram of the proposed simulation. The diagram depicts four key organizational actors (individuals, states, Non-Governmental Organizations, and International Organizations) and the social relationships amongst individual norm followers. This is simply a rough outline of the overall simulation which will likely develop in three stages. Case Studies Nearly 60 years of integration in Europe presents a clear empirical demonstration of the mechanisms of norm evolution within the proposed NLC extension. European integration is momentous because it provides discernible change over time regarding the key independent variables of the proposed research: the social and organizational ties amongst states. European integration also offers variation in the dependent variable of interest: the current norm interpretation of individual states. Consequently, the study of European integration encourages researchers to explain both the increasing trend towards normative homogeneity as well as the source of enduring heterogeneity. I argue that, unlike the original NLC framework, the proposed theoretical extension is capable of accomplishing both of these goals. I will demonstrate the validity of the proposed research through a comparative analysis of norm evolution amongst the individual states of Europe. I will focus on two specific 16
examples of norm change and highlight areas of growing similarity and residual conflict. I plan to trace the evolution of EU norms within the constitutional process, focusing on constitutional speeches/debates, treaty documents, referendum campaigns, and visible institutional change. The first case study will investigate changes in individual interpretations towards the appropriate means of defending the European community. The second case study will explore the respect of human rights and the means considered appropriate to accomplish this objective. Both cases show, as can be confirmed in the constitutional process, that a concurrent increase in social and organizational ties leads to an increasing homogenization of norm interpretations. lxiii However, in both cases, it is also clear that aspects of lasting, stable heterogeneity remain and require further explanation. lxiv I argue that insights drawn from the proposed theoretical extension and agent-based simulation will provide a formal understanding of the mechanisms driving the dynamics of norm evolution within each of these case studies. The first case study will investigate the evolution of a collective defense norm within the EU and its role in the development of the CFSP pillar. The study will show how the collective defense norm began during a period of elevated cognitive dissonance following the Second World War. The study will argue that, lacking a critical mass of organizational ties amongst the states of Europe, a number of competing interpretations of collective defense were able to develop during the early stages of the Cold War (hypothesis 1-B). lxv The study will break these early strands of collective defense into three distinct norm groups: 1) neutrals states wishing to remain militarily non-aligned, 2) US-allies state favoring a defense partnership with the US through NATO, and 3) independents states seeking to relax military dependence on the US. The study will then show how progressive integration, coupled with efforts to constitutionalize collective defense within the four successive treaties on European Union, has produced norm compromise in each of these three dimensions. The empirical analysis of the first case study will support three critical assumptions proposed within the NLC extension. First, the study will highlight the importance of cognitive dissonance in the emergence of norm entrepreneurs (hypothesis set 3). The study will show how normative tension between France, the U.S., and Britain led to the entrepreneurship activity of Charles de Gaulle and the divorce of the independent strand from the US-allies norm group. lxvi The study will also outline the maintenance of de Gaulle s normative opposition through Franco-German rapprochement. lxvii Finally, the study will illustrate how the burgeoning of the EEC into the European Union served to later reinforce de Gaulle s independent strand through the efforts of leaders like Jacques Chirac. lxviii The study will support this claim by highlighting evidence of the independent strand within the current EU constitutional debates. lxix Second, the proposed study will show how a concurrent increase in social and organizational ties leads to the homogenization of norm interpretations (hypotheses 2-A and 2-B). The study will focus specifically on the increasing homogenization of collective defense interpretations amongst the three normative strands reflected in the Maastricht Treaty. lxx Maastricht is considered a turning point in collective defense relations for two important reasons. First, it represents the earliest formal documentation 17