A Tale of Two Systems: Learning to Cooperate and Compete in the International System

I argue that states and leaders in competitive frameworks learn to behave with realist policies from their interactions with other states and leaders and, in the absence of these interactions, rely on other mechanisms. Contrary to what scholars of the realist tradition maintain, I do not argue that the tenants consistent with realism are effectively human nature or due to the self-help, anarchic structure of the international system. Instead, I maintain that leaders in conflictual relationships learn these methods are an effective way in which to respond to the world around them as they learn the constraints that are placed on them by other states’ leaders.

The Problem Classical realism sees humanity as being fundamentally flawed and desirous of maximizing power to feel safe.i Morgenthau even goes so far to claim that human nature is static and the principles of realism are ingrained in human political behavior and interests are defined in terms of power.ii However, structural realism bypasses the first image of analysis (the individual) by claiming that human behavior is muted by structure, or lack thereof, at the international level, thus focusing more on the anarchy of the international system.iii Non-realist scholars such as Hedley Bull even go so far as to admit that the current organizational structure of the international system is anarchic, while also claiming that anarchy can be minimized with common rules and institutions to establish norms.iv I argue that this behavior is learned from interactions in a competitive environment in which states and their leaders learn to accept realist behaviors as both useful tools and, in some instances, as an acceptable method of state behavior.And in times of turmoil and increasing normalization of violence, we should see both conflictual and peaceful interactions between states not as opposing ways of dealing with other actors, but as learned responses that emerge from a history of interactions.If we understand peace and conflict as learned processes, we can better understand how both peace and conflict diffuse to other actors in the system and how conflict can be mitigated against as peaceful interactions are increased among actors.In fact, conflict begets conflict and peace begets peace.
But what does it mean to learn to behave like a realist and learn these norms of conflict?In learning realism, leaders look at the international system and their interactions in "bad neighborhoods" (few democracies, territorial disputes, poor historical relations) and perceive that states behave consistent with realist assumptions to maximize power and pursue their own interests.Additionally, they learn the limits and how to use power in order to achieve outcomes as they are socialized by the leaders of these other states.v Neorealism can be explained as a socialization process in which states can make sense of the world and then interact with it accordingly.vi However, with a constructivist perspective, we can understand interactions in terms of learning based off prior interactions with others in the international system.vii With either theoretical argument, however, we see that there is room to understand international interactions, both conflictual and cooperative, in terms of a set of learning experiences.And states that have negative interactions with one another will be likely to continue reciprocating that ill-will.
Another way in which to understand the concept of learning is to see it as the process of establishing the norms of violence and the norms of cooperation.Finnemore and Sikkink argue that norms emerge via a life cycle in which a behavior emerges, there is a cascade of the behavior, and internalization of the norm.viii Bull ix and other scholars such as Manning x in the English School generally argue that norms emerge out of anarchy and, in so doing, they add a form of non-hierarchical order to the international system.Learning to behave cooperatively or conflictually can thus be understood as the means by which the norms of cooperation and conflict emerge between actors and add a layer of order, or norm of practice, to international politics.
Though it is impossible to test this theory due to the scarcity of data in a temporal sense, I maintain that we can see instances of learning realist behaviors from their interactions with other leaders in the international system, particularly among new leaders and new states, even in today's world.xi Anthropologically, one can make the argument that the first political communities were bound by realist logic of organized violence over things such as resources, hunting grounds or territory.However, it is impossible to know if these individuals had learned this type of behavior, if it was biologically innate, or due to the anarchic nature.This leaves us, then, only with the world we currently have to look for evidence of updating interactions through a learning process.While we can look historically to understand the emergence of cooperative or conflictual behavior, it is difficult to separate out causality in a biological, historical, or anthropological sense.Thus, methodologically, understanding conflict and cooperation as learned behaviors is difficult to test, both quantitatively and qualitatively.

Moving Forward
If Kuhn xii is correct and the only way forward from one paradigm to the next is that the old paradigm must be rejected by new, better explanations, then realism must be seen as a constructed behavior learned through repeated interactions.And, as Vasquez claims, "when more than one explanation exists for a finding, it is possible to appraise their relative merits by: (a) examining their logical coherence and plausibility, (b) comparing their consistency with all the relevant evidenced, including studies indirectly related to the question, and (c) specifying tests in which the different explanations offer different predictions".xiii In answer to point a, the literature has increasingly shown that the logical coherence and plausibility of realist claims in the arms race, conflict, and alliance literature does not hold up to scrutiny.xiv To point b, the consistency of findings has varied across some of these areas, indicating that the logic of realist practice may not hold to be as true as once supposed.For example, the arms race literature (as methodologically flawed as it may be) now largely agrees that arms races increase the probability of conflict (which also may apply to Militarized Interstate Disputes (threats, displays, and uses of force that fall short of war).As for alliances, as Gibler xv and others have demonstrated, alliance formation acts more as a signaling game and show of credible/non-credible commitments than as a hard-set game of balancing between states.And to point c, I leave this to later work to specify the tests whereby we may understand the causal mechanisms by which conflict and cooperation are learned behaviors.
Instead of supporting realist claims, the literature has increasingly shown that the realist/neorealist understanding of how international relations works is fundamentally flawed for several reasons.First, there has been a lack of focus on the first image, or the role that individual actors play, in understanding how international relations takes shape, even when they act under the umbrella of the state.Second, and most importantly, realist scholars have ignored the updating and learning process that both leaders and states (collectively) undergo and the role this plays in informing how actors behave when interacting with one another.In this sense, then, realism is ahistorical in that it does not account for the lessons of history and how states can change their interactions with one another based on a new set of interactions.This updating from interactions is similar to what Reiter xvi argues for alliances and Nevin xvii argues for war-making.In this understanding, the broader context of how states and leaders interact with other states and leaders shapes the way in which they will behave in the practice of international politics.In other words, leaders learn lessons from each interaction they have with other states.Often, they adapt these lessons to their own agenda to maximize what they would like to extract from other states.Instead of realism being understood as an inherent behavior of humans or structure, I argue that some leaders and states learn to behave consistent with realist logic because they have learned this type of behavior from other states in negative interactions.This process is similar to how Mitchell, Kadera, and Crescenzi xviii find evidence to support the penetration of democratic norms into non-democratic states.But, instead of learning democratic norms or norms of cooperation, they learn a different set of behaviors because of their interaction or entrance into the international system coincides with the conflictual interactions they have with other states in the international system.Unlike Linebarger xix and others, xx I agree with Reiter xxi that the learning process is not confined to mere observations of other states and their behavior.Rather than merely observing other states, states and their leaders learn not just from the interactions of other states, but also from their own interactions with other states.This is an important distinction that cannot go unnoticed.Though a leader can examine the actions of another leader or state and determine whether or not the action was successful or not, the same leader may also look to their own personal or state experiences as direction for policy.And these lessons can be broadly categorized as lessons of violence and cooperation.

Diffusion of Violence
Pitcher, Hamblin, and Miller argue that previous sociological literature focused on social conditions and psychological reasons for violence.xxii They argue, instead, that violence should be understood as a diffusive process that "is both instigated and inhabited via direct and vicarious learning."xxiii Jones, Bremer, and Singer, xxiv as mentioned above, note that international disputes tend to cluster geographically and that violence generally occurs within specific regions and states, seeming to indicate a diffusive process.
Vasquez notes that a unified theory of the spread of violence is lacking among scholars of peace and conflict studies.xxv Instead, scholars tend to focus on the individual aspects of when peace or conflict happens.A diffusive process, on the other hand, allows us to see a unified concept of how violence and peace spread temporally and spatially.Vasquez maintains that alliances, rivalries, territorial contiguity, pure contagions (such as ongoing civil wars) are diffusive processes by which conflict spreads from one actor to another.xxvi Within the civil war diffusion literature, several recent advancements in the learning of politics are noteworthy.Linebarger xxvii argues that civil strife metastasizes via a learning process in which the conflict is learned as a successful course of action from interactions with other rebel groups that were successful.Though it is likely that this diffusion of civil violence occurs best when states are contiguous, Linebarger finds evidence of rebel groups learning from noncontiguous, even ideologically dissimilar groups.xxviii However, he argues this diffusion process likely occurs from two pathways: 1.) from ongoing civil conflicts and; 2.) from successful rebels in government.xxix Buhaug and Gleditsch note that civil wars tend to cluster geographically and that a common link in this clustering is ethnic linkages in spatially proximate conflicts.xxx Likewise, Schutte and Weidmann find that civil wars spatially relocate as violence and as conflict escalate in one place, it escalates in another, implying a process of learning in civil conflict.xxxi And using a space-time analysis of conflict in the North Caucuses in Russia, O'Loughlin and Witmer find strong evidence that the direction and tone of violence in the region stemmed from the nucleus of violence in Chechnya.xxxii   Thus, both within the inter and intrastate divisions of conflict study, we see a similar trend: the spread of violent activities (or the preparation of violence in the case of the alliance literature) behaves consistent with a diffusive process in which states learn from others' (particularly similar others) and their own interactions with other states.

Diffusion of Peace
O'Loughlin et al., find convincing evidence of temporal and spatial clustering of democracies and autocracies.xxxiii Mitchell, Kadera, and Crescenzi, xxxiv in addition to Mitchell, xxxv find that autocracies will begin to behave like democracies as the number of democratic regimes increases globally or within a region.xxxvi Similarly, Harrison examines the democratic peace from the systemic level and makes a neofunctionalist argument in support of spillover from democratic regimes to autocratic regime behavior.xxxvii Likewise, Hayes labels this democratic wave and establishment of new norms as a 'third-wave' of constructivist thought.xxxviii   Harrison takes a similar constructivist argument a step further towards liberalism and attempts to reform the liberal argument to claim that the so-called "liberal peace" acts as a socializing mechanism to create a critical mass of democracies that "push" their norms onto non-compliant states.xxxix   If the literature is correct that liberal international norms are slowly being adopted by non-liberal entities, this also implies the counterfactual for conflictual interactions.In other words, this provides evidence for a learning by a diffusion process in which leaders and states see how other states and leaders behave in the international system, implying some leaders will learn to behave conflictually with their neighbors.
Constructivism and the diffusion literature can move our understandings of the violence and peace literatures forward because these causal explanations understand the role of ideas and experiences in how states, their leaders, and their publics perceive and respond to interactions with other states in foreign affairs.A constructivist understanding of learning can show instances where negative interactions between states can lead to exhibitions of conflictual realist behaviors.Mitchell, Kadera, and Crescenzi, xl as well as Mitchell xli have successfully demonstrated that liberal norms become more apparent as the density of democracies increases in a given neighborhood.Similarly, it stands to reason that examples of states learning to behave like realists is apparent when states have negative interactions with other states.Both approaches, taken together, provide an illustration of constructivist claims (see Figure 1): it is possible to build relations between states that are liberal or realist in nature, depending on their previous interactions with one another and the current context of those interactions.States, though, must learn to behave consistent with these frameworks.

Theory
If we assume that states are rational actors that are "pulled" in different directions by domestic and international events, as well as by human individuals, we can assume that states and their leaders will learn from previous interactions.As humans make up those who are both leaders and members of states and institutions, one must assume that if a diffusive updating or learning process is not occurring, then individuals and state institutional behavior, and subsequently international politics, are static, unchanging spheres.
If, however, we assume that states are not static black boxes and that the practice of relations between states is in constant flux due to internal and external dynamics, then we assume that states have learned to behave towards one another consistent with their prior interactions.If we assume that the world of relations between states is not a static, unbending environment, then we must assume that the environment in which states and leaders exist is changing and that both entities adjust their relations with one another.Broadly speaking, I term this the learning process.I define learning as events that a state or a leader has previously been involved in that directs the course of similar events in policymaking.xlii My definition of learning is important because it causally links the past with present and potential future actions.As stated above, the learning process is not merely observations of other states or third-party actors.Instead, the learning process also must be viewed for what leaders and states learn from their own interactions.They do not merely learn by observing others' actions (passive learning), but also by contextualizing their own interactions (active learning).
Furthermore, the last part of this definition is important.To learn a lesson without implementing it is problematic for several reasons: 1.) cognitively, it is impossible to know if the state or leader actually ever "learned" anything or is just making a post-hoc judgment of having learned something; 2.) it makes a normative judgement call that it was the "right" decision based off the events; 3.) the causal link between actions and lessons is broken; 4.) if the lesson was not implemented, then the lesson was either not vivid enough or not recognized at all.Importantly, this definition includes the concept of a change in behavior because of experience.Experience is the key component here because, for the purposes of this argument with respect to realist behaviors, we can measure their (both the states' and leaders') previous exposure to the types of behavior that underlie those interactions.We can measure experience in two ways: 1.) in temporal exposure to the system or in a position of leadership; 2.) in terms of the definition of cognitivism, such as prior experience in making alliances or disputes.This conceptualization of experience mirrors what we see in the real world: time matters for promotions, even in an academic setting.Likewise, showing that one has prior experience (such as academic publishing), shows both socialization and productivity.
We can assume that, as states are made up of individuals who are constantly interacting and learning from the world, these learning experiences should be externalized and exhibited by states, and that we can in turn find evidence of learning.We can also assume that the personal experiences of leaders xliii and the collective experiences of states permit those leaders and states to acquire certain understandings of how they perceive the world works and how they fit into that picture.This, in turn, affects their behavior with respect to other states and the world views that they hold with respect to human and social interaction.It is also important to note, though, that the causal arrow can point both directions.Experiences of states and leaders can affect the behaviors they chose to take and these cumulative behaviors in turn affect the experiences that states undergo.
Along these same lines, there are multiple arguments for how interactions between states occur, which is commonly known as the level of analysis problem.The three primary models are systemic explanations of relations between states, xliv domestic (institutional) models of the constraints placed on leaders in autocracies and democracies alike xlv and individual level models that emphasize background experiences and the free agency of world leaders.xlvi I take a middle-of-the-road approach in the levels-of-analysis debate, arguing that leaders primarily take knowledge from their interactions with other leaders in the international system and in the absence of interactions, they rely on other states' interactions as indicators of acceptable behavior and personal experiences and the nature of their domestic rise to power.The effect is a "state/leader dyad" in which both states as functional units and leaders absorb information from their interactions with one another and, in turn, they reciprocate a similar style of interactions with one another.This behavior occurs by one of two diffusive mechanisms: 1.) they see it as a way to counter the other state or; 2.) it is viewed as a legitimate way for states to behave.State leaders then use this background knowledge as a basis for their policies with other states.This diffusive process of learning is greatest for both new leaders and new states in the international system as they are socialized into acceptable behaviors with the states they engage politically.
Mechanically, leaders who predominately interact with other democratic leaders are more likely to absorb the lessons associated with democracies.xlvii For instance, there exists much support for the conclusion that democracies tend to geographically or regionally cluster, indicating a diffusive process.xlviii One shortcoming of this approach is that these analyses are exclusively confined to institutional or systemic level explanations.There are multiple explanations as to why democratic regimes exhibit a geographic clustering effect, which appear to largely mirror the institutional and systemic approaches.But it also is likely that the interactions between leaders with one another on a personal level also have something to add here.Though institutional and systemic explanations of democracy like this are noteworthy, scholars cannot discount the role of human agency of individual leaders in being the genesis of such causal outcomes.For instance, José Figueres Ferrer, the former rebel leader of Costa Rica turned president, disbanded the Costa Rican military and established a democracy in a region surrounded by autocratic regime.xlix   In contrast, for leaders who predominately interact with autocratic regimes or have negative interactions with other leaders (particularly over salient stakes such as territory), the lessons that they are more to learn are that force is an acceptable foreign policy strategy.This will thus be exhibited in the policy choices they pursue, particularly in inherently competitive realist frameworks such as alliances, arms races, and disputes falling short of war.These competitive frameworks should offer the most evidence that leaders learn from previous interactions with leaders and from their own personal and domestic backgrounds.And when leaders learn to use alliances or arms races or continue disputes with another aggressor state, they are exhibiting the learned behavior from their prior interactions.

Conclusion
An understanding of the interactions between states and their leaders as a relic of learning is useful for several reasons.First, it allows for updating to occur with states in the international system.Second, it provides an explanation of why conflict or cooperation occurs between states based off a shared positive or negative history with one another.Finally, it is a theoretical shift away from the predominate ahistorical paradigms of international relations which, while parsimonious, do not account for the complexities of interactions in the world.Instead of focusing on states and their leaders as greedy power-maximizers (classical realism), or that states can always cooperate with one another to solve their informational and political problems (as in neoliberalism or liberalism), an understanding of states learning in the prior context of interactions is useful precisely because it provides context.
Using this framework as an understanding of international political interactions, we can better conceptualize both friendly and conflictual interactions between actors in the international system.This is appealing because it does not assume that states are either naturally cooperative or naturally conflictual.Rather, it assumes that states will reciprocate either the ill-will or good feelings that stem from their interactions with other states in the international system, providing a path to peace or a path to war.x Manning, Charles.The Nature of International Society.Macmillan, 1962.xi Due to the large n, it will be easiest to see examples of new leaders and states learning from their interactions in MIDs.With the small number of alliances (even dyadically), it is more difficult to observe learning with respect to my argument.Similarly, as stated above, the arms race literature has numerous biases built into which cases are observed and which are not, thus it is problematic to examine arms races with respect to learning from the interactions between one another.xii   xxxvi See also Dixon, 1993; Dixon, 1994; Dixon, 1998; McLaughlin Mitchell and Diehl, 2007  xxxvii