Perception and Misperception in International Politics

PRINCETON UNIVERSITY PRESS

Contents

Perception and the Level of Analysis Problem

Do Perceptions Matter?

Before discussing the causes and consequences of the ways in which decision-makers draw inferences from the world, we must ask a preliminary question: do the decision-makers' perceptions matter? This is an empirical question. Logic permits us to distinguish between the "psychological milieu" (the world as the actor sees it) and the "operational milieu" (the world in which the policy will be carried out) and to argue that policies and decisions must be mediated by statesmen's goals, calculations, and perceptions. But it does not follow from this that we must deal with these intervening variables in order to understand and predict behavior. This is not an uncommon claim:

One may describe particular events, conditions, and interactions between states without necessarily probing the nature and outcome of the processes through which state action evolves. However, and the qualification is crucial, if one wishes to probe the "why" questions underlying the events, conditions, and interaction patterns which rest upon state action, then decision-making analysis is certainly necessary. We would go so far as to say that the "why" questions cannot be answered without analysis of decision-making.

But theory and explanation need not fill in all the links between cause and effect. Indeed, this is impossible. One can always ask for the links between the links. High density theories have no privileged status; they are not automatically illuminating or fruitful. It is true that re-creating a decision-maker's goals, calculations, and perceptions is a satisfying method of explaining his behavior because the scholar, sharing as he does the decision-maker's characteristics of being a thinking, goal-seeking person, can easily say: "If that is the way the statesman saw the situation, it is no wonder that he acted as he did." But the comfort we feel with this form of explanation should not blind us to the fact that, unless there are significant variations in the ways people see the world that affect how they act, we need not explore decision-making in order to explain foreign policy. Most case studies assume that the details presented significantly affected the outcomes. This may not be true, however. "Pleikus are streetcars," McGeorge Bundy said in explaining that the Viet Cong attack on the American installation in February 1965 had affected only the timing of the American bombing of North Vietnam. If you are waiting for one, it will come along. The specifics of the triggering event cannot explain the outcome because so many probable events could have substituted for it. To understand the American policy of bombing the North we should not examine the attack on Pleiku. Had it not occurred, something else would have happened that would have permitted the same response. Logic alone cannot tell us that a similar claim about the decision-making process is invalid: the way people perceive others and make decisions only marginally influences outcomes. So we must seek empirical evidence on the question: do the important explanatory variables in international relations involve decision-making? In terms of perceptions this can be separated into two subsidiary questions: "are important differences in policy preferences traceable to differences in decision-makers' perceptions of their environments?" and "are there important differences between reality and shared or common perceptions?" Detailed affirmative answers to these questions will emerge in this book, but a brief general discussion is in order here.

These questions raise the familiar level of analysis problem. Although it has been much debated, agreement is lacking not only on the substantive dispute but even on the number of levels. Arnold Wolfers proposes two, Kenneth Waltz three, and James Rosenau five. To fill in the sequence, we will discuss four. One is the level of decision-making, the second is the level of the bureaucracy, the third is that of the nature of the state and the workings of domestic politics, and the fourth focuses on the international environment. Which level one focuses on is not arbitrary and is not a matter of taste — it is the product of beliefs (or often hunches) about the nature of the variables that influence the phenomena that concern one. To restate the first question in terms of the level of analysis problem, we need not adopt a decision-making approach if all states behave the same way in the same objective situation, if all states of the same kind (i.e. with the same internal characteristics and politics) behave the same way in the same objective situation, or if state behavior is determined by bureaucratic routines and interests.

Although the empirical questions are central here, we should also note that the level of analysis problem has important moral implications. When all people would respond the same way to a given situation, it is hard to praise or blame the decision-maker. Thus, those accused of war crimes will argue that their behavior did not differ from others who found themselves in the same circumstances. And the prosecution will charge, as it did against Tojo and his colleagues, that, "These defendants were not automatons; they were not replaceable cogs in a machine. ... It was theirs to choose whether their nation would lead an honored life ... or ... would become a symbol of evil throughout the world. They made their choice. For this choice they must bear the guilt." Similarly, if all nations follow similar courses of action, one cannot argue that some deserve to be branded as immorally aggressive. Thus in 1918 Bethmann-Hollweg rebutted those who blamed Germany for the war by pointing to the "general disposition towards war in the world ... how else explain the senseless and impassioned zeal which allowed countries like Italy, Rumania and even America, not originally involved in the war, no rest until they too had immersed themselves in the bloodbath?"

The three non-decision-making levels assert the importance of various aspects of the objective situation or the actor's role. They say that if we know enough about the setting — international, national, or bureaucratic — we can explain and predict the actor's behavior. An interesting sidelight is that if other actors believed that the setting is crucial they would not need to scrutinize the details of the state's recent behavior or try to understand the goals and beliefs held by the state's decision-makers. It would be fruitless and pointless to ask what the state's intentions are if its behavior is determined by the situation in which it finds itself. Instead, observers would try to predict how the context will change because this will tell them what the state's response will be. Decision-makers could then freely employ their powers of vicarious identification and simply ask themselves how they would act if they were in the other's shoes. They would not have to worry about the possibility that the other might have values and beliefs that differed from theirs. It is interesting, although not decisive, to note that decision-makers rarely feel confident about using this method. They usually believe both that others may not behave as they would and that the decision-makers within the other state differ among themselves. So they generally seek a great deal of information about the views of each significant person in the other country.

Of course it is unlikely that there is a single answer to the question of which level is most important. Rather than one level containing the variables that are most significant for all problems, the importance of each level may vary from one issue area to another. Furthermore, which level of analysis is the most important may be determined by how rich and detailed an answer we are seeking. The environment may influence the general outline of the state's policy but not its specific responses. Thus it can be argued that, while decision-making analysis is needed to explain why World War I broke out in August 1914, the external situation would have led the states to fight sooner or later. Or the importance of variables at each level may vary with the stages of a decision. For example, domestic politics may dictate that a given event be made the occasion for a change in policy; bargaining within the bureaucracy may explain what options are presented to the national leaders; the decision-maker's predisposition could account for the choice that was made; and the interests and routines of the bureaucracies could explain the way the decision was implemented. And the same variable may have different effects at different stages of the decision-making process — for example, conflicts among subordinates may increase the variety of information and the number of opportunities for decision that the top decision-maker gets, but may simultaneously decrease his ability to see that his decisions are faithfully implemented.

The importance of variables at one level can also vary with the state of variables at other levels. Rosenau suggests that the international environment is more important in determining the policy of small states than it is of large ones, and Stanley Hoffmann argues that nuclear weapons and bipolarity have reversed this relationship. More generally, the importance of the other levels decreases if the variables in one level are in extreme states. Thus, maneuvering within the bureaucracy may be more important when the top decision-makers are inexperienced or preoccupied with other matters. And Wolfers argues that states tend to behave the same way when they are faced with extreme danger or extreme opportunity, but that when environmental constraints are less severe there will be differences in behavior that must be explained at the decision-making level. More complex interactions among the levels are also possible. For example, the effect of internal instability on expansionism could vary with the opportunities for success in war. Unstable states may be more prone to aggression than stable ones when the chances of victory are high but might be more cautious than others when their leaders perceive a significant probability of defeat or even of temporary setback. Or the stability of the regime might influence its propensity for aggression, but the nature of the regime (e.g. whether it is democratic or dictatorial) might be more important in explaining how it makes peace.

To deal with all these questions would require another book. Here all I will try to do is to outline the kinds of evidence necessary to establish the validity of simple propositions about the importance of the various levels. In doing so, I will sketch the most extreme arguments for the importance of each level. It is obvious that the questions and arguments could be rephrased in more subtle terms but since I am concerned with the kinds of evidence that the propositions call for the gain in analytical clarity is worth the sacrifice involved in ignoring more complete explanations that combine a multitude of variables at many levels.

The International Environment

To argue that the international environment determines a state's behavior is to assert that all states react similarly to the same objective external situation. Changes in a state's domestic regime, its bureaucratic structure, and the personalities and opinions of its leaders do not lead to changes of policies. Changes in the external situation, however, do alter behavior, even when variables on the other levels remain constant. To test these claims, we would need good measures of all the variables, especially the nature of the objective situation and the state's policies. Even if we had such indicators, we would have to cope with the paucity of the most desired kinds of comparisons. This is easily understood by glancing at the similar issue in the study of individual behavior — the debate over the relative importance of situation and role versus idiosyncratic variables in determining individual behavior. Because so many people of widely differing backgrounds, personalities, and opinions fill the same role and because the same person fills many different roles, we can try to determine the relative impact of situational and idiosyncratic variables by examining how a person's behavior varies as his role changes and how people of widely differing characteristics perform in similar situations.

It is much harder to make the analogous comparisons in international relations. In only a few international systems do we find many cases in which states play, either simultaneously or consecutively, several roles and in which each role is filled by states that are otherwise quite different. This would occur in a long-lasting system where there were frequent changes in the relations among the actors. Thus each state might at one time be a neutral, a "holder of the balance," a state with aggressive designs, a defender faced by a state whose intentions are difficult to determine, and so on. To a limited degree this test is possible in a balance-of-power system. But it is not available for most other systems, for example the one prevailing since World War II. Most nations have not changed roles, and indeed cannot do so because of such permanent factors as size and geography. The United States can never play the role of a second-ranking state caught between two blocs of greater powers. France can never be the leader of one of two dominant blocs. And while the United States and France may have played roles similar to these in the past, the extensive differences in the situation mean that any differences in response that might be found would not show that roles are unimportant.

COMPULSION IN EXTREME CIRCUMSTANCES?

It is worthwhile to look at cases of the kind that are supposed to show most strongly the influence of external conditions. If there are differences of behavior here, the argument for not ignoring the other levels of analysis will apply a fortiori to a wider domain of cases. Arnold Wolfers argues that, the greater the external compulsion, the greater the homogeneity of behavior and therefore the less the need to study decision-making. In a well-known passage he says: "Imagine a number of individuals, varying widely in their predispositions, who find themselves inside a house on fire. It would be perfectly realistic to expect that these individuals, with rare exceptions, would feel compelled to run toward the exits. ... Surely, therefore, for an explanation of the rush for the exits, there is no need to analyze the individual decisions that produced it."

But the case is not as clear as this analogy suggests. If a situation were so compelling that all people would act alike, decision-makers would not hesitate nor feel torn among several alternative policies, nor would there be significant debates within the decision-making elite. In fact, key decisions that are easily reached, such as those involving the Truman Doctrine and Marshall Plan, stand out because they are so rare. For despite the implication of Wolfers' proposition that we know when we are faced by extreme danger, just as we can tell when the house is on fire, in fact this question is often bitterly contested. (To say that once decision-makers perceive the fire they will head for the exits leads us back to decision-making analysis.) For Churchill, the house was burning soon after Hitler took power in Germany; for Chamberlain, this was the case only after March 1939; and for others there never was a fire at all. To some decision-makers, the Soviet Union is a threat to which the United States is compelled to respond. To others the threat passed years ago. Again, to a growing number of scholars it never existed. Similarly, American statesmen see a much greater threat from communism in both Europe and Southeast Asia than do the leaders of our allies. Decision-makers may even agree that their state's existence is threatened but disagree about the source of the threat. This was true, for example, in the United States around the turn of the nineteenth century, when the Federalists believed France so much a menace that they favored war with her. At the same time, the Republicans believed England an equal menace. (It should be noted that this disagreement was rooted as much in differences in values and interests as in divergent empirical analyses of the situation.)