NUCLEAR PROLIFERATION AND THE USE OF FORCE: NUCLEAR COERCION AND NUCLEAR LEARNING by Michael David Cohen A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in The Faculty of Graduate Studies (Political Science) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2012 © Michael David Cohen, 2012 ii Abstract What is the effect of developing nuclear weapons on a state’s conflict propensity? Extant answers to this question do not model the effects of time and thus leave policy-makers addressing cases such as Iran and North Korea with two contradictory answers – that nuclear proliferation is stabilising and destabilising – that both have extensive empirical support. I argue that nuclear proliferation is dangerous when decision-makers learn that it is safe and safe when they learn it to be dangerous. I develop and test a psychological nuclear learning model that, consistent with recent quantitative research, explains why weak, revisionist/dissatisfied nuclear powers are highly conflict prone but the same experienced nuclear powers are not. In the model, three biases associated with the availability heuristic make weak revisionist new nuclear powers war prone. (1) Illusory correlations cause decision-makers to believe that the immense destructive potential of nuclear weapons causes them to offer similarly large coercive power; (2) self-serving attribution biases cause decision-makers to infer a causal relationship between nuclear compellence threats and subsequent compliance and (3) decision-maker’s nuclear threats, because they are more cognitively accessible than other contextual variables and the operation of these variables in historical cases, cause them to overestimate the probability of successful nuclear coercion. Such new nuclear powers inevitably practice nuclear coercion, which causes a nuclear crisis and fear of imminent nuclear war. I show that this fear moderates the high war propensity of new nuclear powers. It causes (1) pessimistic estimates of the probability of inadvertent nuclear escalation resulting from nuclear coercion, (2) moderation in future nuclear diplomacy and (3) pessimistic risk choices in logically unrelated foreign policies. Recently released Soviet archival data and elite interviews in South Asia, including with former Pakistani President Pervez Musharraf, allow me to test the psychological nuclear learning model against realist, domestic politics and rational learning models. iii Preface Approval for some research for this dissertation was granted under UBC Behavioural Research Ethics Board Certificate No. H10-00680 iv Table of Contents Abstract ......................................................................................................................................................... ii Preface ........................................................................................................................................................... iii Table of Contents ......................................................................................................................................... iv List of Tables ................................................................................................................................................ vii List of Figures ............................................................................................................................................... viii Acknowledgements ...................................................................................................................................... ix Dedication ..................................................................................................................................................... xi 1 Introduction ............................................................................................................................................... 1 1.1 Introduction .......................................................................................................................... 1 1.2 Nuclear Proliferation, Conflict and Time ............................................................................. 7 1.3 Unpacking the Effect ............................................................................................................ 13 1.4 Explanations: The Conventional Wisdom Fails ................................................................... 15 2 Two Nuclear Learning Models ................................................................................................................ 28 2.1 Two Nuclear Learning Models ............................................................................................. 28 2.2 Nuclear Proliferation and Nuclear Coercion in a Rational Learning Model ........................ 37 2.3 Nuclear Proliferation and Nuclear Coercion in a Psychological Learning Model ............... 44 2.4 Dissertation Summary .......................................................................................................... 57 3 Soviet Nuclear Missiles, 1958-1962 .......................................................................................................... 65 3.1 Soviet Nuclear Missiles ........................................................................................................ 65 3.2 Was the Soviet Union a Weak Revisionist State? ................................................................ 66 3.3 The Availability Heuristic .................................................................................................... 82 3.3.1 The Suez Crisis, 1956 ................................................................................................. 82 3.3.2 The First Berlin Crisis, 1958-59 ................................................................................. 87 3.3.3 The Second Berlin Crisis, 1961 .................................................................................. 89 v 3.4 Fear ....................................................................................................................................... 91 3.4.1. The Cuban Missile Crisis, 1962 ................................................................................ 91 3.5. Kennedy’s Lessons from the Cuban Missile Crisis ............................................................. 101 4 Pakistani Nuclear Missiles, 1998-2002 .................................................................................................... 107 4.1 Pakistani Nuclear Missiles ................................................................................................... 107 4.2 The Kargil War, 1999 ........................................................................................................... 113 4.3 Pakistani Lessons from Kargil ............................................................................................. 123 4.3.1. Psychological Causes: Availability ........................................................................... 125 4.3.2. Lessons ..................................................................................................................... 127 4.3.3. Fear ........................................................................................................................... 128 4.4 A Further Test: Indian Lessons from Kargil ........................................................................ 131 4.4.1. Expectations of Kargil ............................................................................................... 131 4.4.2. Lessons from Kargil .................................................................................................. 135 4.5 American Crisis Management in South Asia ....................................................................... 138 5 Lyndon Johnson’s Lessons from Cuba ................................................................................................... 142 5.1 Lyndon Johnson’s lessons from Cuba .................................................................................. 142 5.2 Lyndon Johnson and the Cold War ...................................................................................... 147 5.3 Johnson’s Beliefs: Congressman and Senator, 1945-60 ...................................................... 153 5.4 Johnson’s Beliefs: Vice President, 1961-November 1963 ................................................... 163 5.4.1. Johnson’s Beliefs on Nuclear Weapons .................................................................... 165 5.5 Johnson’s Beliefs: President, November 1963-1965 ........................................................... 167 5.6 Conclusion ............................................................................................................................ 180 6 Conclusion ................................................................................................................................................. 182 6.1 Summary .............................................................................................................................. 182 6.2 Problems and Rejoinders ...................................................................................................... 189 6.3 Iran and North Korea ............................................................................................................ 193 vi 6.4 Managing Nuclear Proliferation ........................................................................................... 200 6.5 Other Implications ................................................................................................................ 205 Bibliography ................................................................................................................................................. 209 vii List of Tables Table 1 Psychological and Rational Choice Model Expectations of Weak Revisionist New Nuclear Powers......................................................................................................56 Table 2 Universe of Cases of New Nuclear Powers...........................................................59 Table 3 Soviet Deployed ICBMs, 1955-64.........................................................................82 Table 4 Kennedy’s and Johnson’s Beliefs..........................................................................147 viii List of Figures Figure 1 The Psychological Nuclear Learning Model.........................................................57 ix Acknowledgements Ten years ago, in the midst of an Economics degree at the University of Sydney, I became very curious about the impact that nuclear weapons have had on international politics. This thesis is the result of that curiosity. I have accumulated a number of personal debts along the way, and while this brief acknowledgement cannot do these numerous individuals’ services justice, I would like to express my sincere gratitude and appreciation for their commitment to my intellectual development. I could not have asked for a more encouraging group of scholars, friends and family, and want to absolve them all for any deficiencies in this final manuscript. At the University of Sydney, John M. Hobson, the descendent of the famous imperialism theorist John Hobson, first sparked my interest in International Politics. I had been interested in international affairs before taking his course, but his exposition of key theories of international politics seems critical to my pursual of doctoral studies in Political Science and International Relations. My greatest academic debts, however, are to two scholars in the Department of Political Science at UBC. Dick Price was an obvious choice of supervisor for my Master’s Thesis on the nuclear taboo. He perfectly balanced the tricky task of pushing me in the right direction while allowing me to follow my own intuitions. I am also extremely grateful to him for taking time out of his sabbatical year to help me through my comprehensive exams. Although my initial explorations for dissertation work led me further and further away from areas in which he had expertise, he, ever committed, offered to supervise me when it became clear that my dissertation would involve methodologies and literatures that he was unfamiliar with. I am very thankful for his faith in my abilities, prompt attention to my numerous emails, insights on teaching, and good humour. I walked out of Alan Jacob’s first qualitative methods class awestruck at his intelligence, attention to detail and strong commitment to rigorous social science. Six years later, I am still amazed at these qualities. Alan’s input has been tremendously influential in the development of this dissertation, and he is responsible for a substantial improvement between early and later drafts. I am convinced that very few scholars read their students draft dissertation manuscripts as carefully as he did. His comments always stimulated helpful subsequent developments and refinements, and I only hope that I will have the benefit of his input in future research projects. I am also thankful to Wade Huntley for his commitment to a term of nuclear deterrence and proliferation directed readings, generous time talking about nuclear weapons and early dissertation prospectus related issues over several lunches. After his unexpected departure from UBC, Professor Patrick Morgan of the University of California Irvine, who had been friendly enough to share his time and insight with me at several conferences, generously offered to serve on my committee. His profound knowledge of both nuclear deterrence and International Security, kind manner, patience with numerous emails and generous provision of lavish lunches at numerous conferences to talk through dissertation related issues have all been tremendously appreciated. At UBC, I have also benefitted from helpful discussions with Arjun Chowdhury, Brian Job, Chris Kam, Ben Nyblade, Fred Cutler, David Santoro, Adam Bower, Clare McGovern, John McAndrews and Nathan Allen (who suggested the first half of the title of this dissertation). The encouraging friendship and numerous summer discussions with Professor John M. Owen have also been very well received. Seva Gunitsky, Walter Ladwig, Dane Swango, Gaurav Kampani, Kyle Lascurettes and Rich Maas, also completing their dissertations x on International Security related issues at other universities, have been a source of encouragement and stimulation. Finally, this list of scholarly debts would not be complete without mentioning the indirect influence of Professor Robert Jervis. I had the pleasure of briefly meeting him several years ago, and despite several helpful email exchanges, he has not directly influenced this dissertation. I have nonetheless found his attention to the impact of psychological biases on International Security, commitment to think systemically about complex strategic issues and commitment to document these historically a strong scholarly model. Jervis’s influence on this thesis is strong, and I hope that if he ever reads it he will not be disappointed. My family and friends have, however, had to put up with far more of me than my academic colleagues, and I owe them a far greater debt. My mother, Jenny, has encouraged me in my academic pursuits for as long as I can remember. My father, Phil, has also encouraged me to pursue my academic goals in his unique way. I am very thankful to my parents for encouraging me to work hard throughout twenty-three years in educational institutions. My parents-in-law, Tim and Georgina, have also been a source of encouragement, memorable trips to the Sydney Cricket Ground and good humour. In Vancouver, Don Lewis, the late Ken Stevenson, Nathaniel Funk, George Sterling, Tom Wilson, Tom Spreter and Katie Calloway have been strong sources of support, and often encouraging distractions, through my doctoral work. Along the way, my wife Anna and I have been blessed with the birth of our two sons Samuel and Robert. They have been tremendous sources of joy, and have helped me complete this dissertation more than they know. Not only have they provided a strong incentive for me to finish graduate school, but their good spirits, simple needs and easy sources of contentment have been a real model as I have had to deal with the realities of graduate school. I hope this final product justifies the countless times that I had to say goodbye as I left them to work on “Daddy’s book.” My greatest debt, of course, goes to my wife Anna. This thesis would simply have been impossible without her, and it is hard to know how to start to thank her. Two months after our wedding day we left Sydney for Vancouver where Anna knew nobody and had no guaranteed employment. Her own work provided welcome financial support, and she often single-handedly raised our sons while I worked on this manuscript. But more importantly than that, she has been my best friend and fiercest critic. She has stuck with me through good times and bad, put up with far too many discussions about nuclear learning, and reinforced what commitment means. I dedicate this thesis to her, and hope that I can repay the many hours that I have spent away from her to complete it. xi To Anna 1 Chapter 1: Introduction 1.1. Introduction The most likely current system-changing force is nuclear proliferation, although how proliferators would use this security is far from clear. Robert Jervis, 20091 Interest in the impact of nuclear weapons on international politics is undergoing a renaissance. Ever since the bombing of Hiroshima sixty-five years ago, the role of nuclear weapons in international politics has been a central part of the study of international politics. Early studies that addressed this by Bernard Brodie, Albert Wohlstetter and Thomas Schelling had a profound influence on the International Relations discipline, and assessments of nuclear strategy, deterrence and arms control were still common in the Cold War's final decades. The end of the Cold War however witnessed a redirection of intellectual initiative. In the last two decades many scholars have directed their attention to other security issues such as civil war, ethnic conflict, insurgency and terrorism. The demise of the Soviet Union and end of the superpower strategic rivalry seemed to herald the end of rigorous scholarship on the consequences of the nuclear revolution. Yet the 1990s witnessed a resurgence of nuclear security concerns. While the newly independent Ukraine, Kazakhstan and Belarus handed over their inherited arsenals to 1 Robert Jervis, “Unipolarity: A Structural Perspective,” in World Politics 2009 61 (1) pp. 188- 213, p. 212 2 Russia and the Apartheid regime in South Africa underwent liberal reforms and eliminated its small arsenal, India, Pakistan and North Korea (in 2006 and 2009) all tested nuclear weapons and delivery vehicles. The former states fought a war and engaged in a ten month mobilised military standoff in the aftermath of the tests. North Korea became the center of an international crisis when President Bill Clinton almost authorised airstrikes to destroy Pyongyang's uranium and plutonium enrichment facilities in 1994.2 Israel has continued to deny the existence of its own nuclear deterrent and Iran, a long-time sponsor of terrorist groups and spoiler of multiple Middle East Peace processes, has persisted with uranium enrichment and missile testing in the face of economic sanctions from the United States and, more recently, other Security Council members. Tehran now apparently possesses enough nuclear fuel that, with further enrichment, would be sufficient to make two nuclear weapons, and has successfully tested missiles capable of delivering nuclear payloads to Israel and elsewhere in the Middle East.3 This spread of nuclear weapons has challenged policy-makers to address fundamental strategic changes in the important Middle East, South Asian and East Asian regions. The spread of such destructive weapons to relatively weak actors is a historically unique development. Unfortunately, International Relations scholars have responded with neither a satisfactory explanation of the consequences of these transformations nor 2 Bill Clinton, My Life: The Presidential Years. New York: Vintage, 2004 p.164 3 David E. Sanger and William J. Broad, U.N. Says Iran Has Fuel for 2 Nuclear Weapons, The New York Times, May 31 2010. See also William J. Broad, John Markoff and David E. Sanger, “Israeli Test on Worm Called Crucial in Iran Nuclear Delay,” New York Times, January 15, 2011 3 tactical policies for navigating these turbulent waters.4 We still know relatively little about the relationship between nuclear proliferation and conventional violence. Some recent studies on nuclear proliferation have made important contributions. Most of the literature on the causes of nuclear proliferation was, until recently, descriptive case histories that lacked any comparative basis.5 Two studies that addressed variation in ruling coalitions and national identities reached similar conclusions on the causes of nuclear proliferation. Jacques Hymans showed that leaders who both define their nation as being naturally in conflict with an adversary and have a strong emotional tendency to appeal to nationalist ideals are likely to pursue an operational nuclear arsenal. Etel Solingen showed that “inward looking models [of economic development] approximate necessary if not sufficient conditions for nuclear weapons programs;” states that are isolated from the global economy, nationalistic and radical-confessionalist are more likely to acquire nuclear weapons than those states that are integrated into the global economy and liberal and/or democratic.6 4 For a comprehensive discussion of current nuclear weapons issues in twenty-first century Asia, see T.V. Paul, “Nuclear Weapons and Asian Security in the Twenty-first Century,” in N. S. Sisodia, V. Krishnappa and Priyanka Singh, Proliferation and Emerging Nuclear Order in the Twenty-first Century, 2009. New Delhi: Institute for Defence Studies and Analyses, pp. 29-43 5 By proliferation I, like the literature, refer to the horizontal spread of nuclear weapons where previously non-nuclear states acquire a nuclear arsenal. The impact of vertical proliferation – where nuclear powers increase the size, sophistication or diversification of their arsenal has also received attention. See Robert Jervis, “Why Nuclear Superiority Doesn't Matter,” Political Science Quarterly 94 (4) Winter 1979-80; Matthew Kroenig, Nuclear Superiority or the Balance of Resolve? Explaining Nuclear Crisis Outcomes, http://www.matthewkroenig.com/Kroenig_Nuclear%20Superiority_July2010.pdf 6 Jacques E. C. Hymans, The Psychology of Nuclear Proliferation: Identity, Emotions, and Foreign Policy, 2006. Cambridge: Cambridge University Press; Etel Solingen, Nuclear Logics: Contrasting Paths in East Asia and the Middle East, 2006. Princeton: Princeton University Press, p.46. See also T.V. Paul, Power Versus Prudence: Why Nations Forgo Nuclear Weapons, 2000. Montreal and Kingston: McGill-Queens University Press. 4 While we know much more about the causes of nuclear proliferation, much less has been established about its consequences. Most of the earlier literature on the consequences of nuclear weapons for international politics addressed nuclear strategy, deterrence and arms control in the Cold War superpower context. This addressed the salient concern of the era, but did not establish whether the strategic dynamics between other nuclear powers after 1990 are different. Unlike the Cold War Superpowers, other nuclear adversaries face challenges and opportunities of external interference. In some cases, operational nuclear weapons may be designed to usher favourable assistance from a superpower patron. These states also tend to have smaller and less sophisticated conventional forces and less secure command and control networks. Many of them also have historical enmities with their principal adversary – that may also have either nuclear weapons or ambitions to acquire them – that predate nuclear proliferation. Whether the Cold War experience is a reliable model for the consequences of nuclear proliferation in South Asia, East Asia and the Middle East remains unclear. Paucity of data in new nuclear powers forced scholars to examine the United States case and tentatively extrapolate their findings elsewhere.7 Scholars have more recently begun to address nuclear deterrence in South Asia. As I address in greater detail below, Paul Kapur showed that nuclear missile acquisition emboldened the Pakistani military to pursue a coercive strategy over Kashmir in 1999 that was designed to encourage favourable American intervention. Vipin Narang recently differentiated three different nuclear postures. He argued that Pakistan's conventional 7 Two of the influential studies from this period are Scott D. Sagan, The Limits of Safety: Organisations, Accidents and Nuclear Weapons, 1993. Princeton: Princeton University Press, and Peter D. Feaver, Guarding the Guardians: Civilian Control of Nuclear Weapons in the United States, 1992. Ithaca: Cornell University Press. 5 inferiority to India and weak civilian control over the military has caused what he calls an ‘asymmetric escalation posture’ that involves nuclear weapons deterring not only nuclear threats but conventional attacks through deploying nuclear assets quickly to enable their battlefield release through pre-delegative procedures to military users in crises.8 Other research has highlighted the influence of ideational variables on the use of nuclear weapons. Nina Tannenwald and Richard Price have shown that the long-time staple of deterrence theory – a credible commitment to defend one's interests and an at least partly survivable retaliatory capability – may be neither necessary nor sufficient to prevent the use of nuclear and chemical weapons. The nuclear taboo – a norm that stigmatises the use of nuclear weapons as morally reprehensible – may have been necessary to cause nuclear restraint in some cases.9 Very little attention has been directed to the relationship between nuclear proliferation and conventional war and conflict. Almost all of the literature on the consequences of nuclear proliferation assumed that the most important consequence of the spread of nuclear weapons is nuclear war. Kenneth Waltz and Scott Sagan's famous debate, The Spread of Nuclear Weapons, contested the probability of multiple mechanisms that could cause nuclear war and became the benchmark study to address this issue. While Waltz 8 S. Paul Kapur, “Ten Years of Instability in Nuclear South Asia,” International Security 33 (2) 2008 pp. 71-94; Vipin Narang, “Posturing for Peace? Pakistan's Nuclear Postures and Nuclear Stability,” in International Security 34 (3) 2009 pp. 38-78 9 Richard Price and Nina Tannenwald, “Norms and Deterrence: The Nuclear and Chemical Weapons Taboos,” in Peter Katzenstein, ed. The Culture of National Security, 1996. New York, Columbia University Press; Nina Tannenwald, The Nuclear Taboo: The United States and the Non-Use of Nuclear Weapons since 1945, 2008. New York: Cambridge University Press. See also T. V. Paul, The Tradition of Non-Use of Nuclear Weapons, 2009. Stanford: Stanford University Press 6 and Sagan disagreed on the dangers posed by nuclear proliferation, both agreed that nuclear war is the most important consequence to address.10 Subsequent contributions to the debate did not challenge this, and much scholarly and non-scholarly attention has assumed that nuclear use or nuclear war in the Korean peninsula or Middle East is the most important outcome to address. This scholarly bet has paid off in addressing how the most undesirable consequence of nuclear proliferation might be avoided. But it has failed to address the actual consequences of nuclear proliferation. While nuclear weapons have spread to nine states in over six decades since 1945, they have not been used in war. A preoccupation with nuclear war has prevented scholars from identifying what has probably been the most destructive consequence of the spread of nuclear weapons. Scholars have assumed that nuclear proliferation does not cause conventional conflict because statistical tests showed no relationship between these variables.11 But these models have ignored the role of time. Temporally disaggregating the effect of nuclear proliferation on conflict propensity uncovers a robust correlation. Almost two thirds of states that have developed nuclear weapons have fought a war or conflict within a few years of acquiring them. While new nuclear powers are highly conflict prone, the same experienced nuclear powers are not. This dissertation addresses the relationship between nuclear proliferation, conflict and time. More specifically, I shall argue that inattention to 10 Kenneth N. Waltz and Scott D. Sagan, The Spread of Nuclear Weapons: A Debate Renewed, 2003. New York: Norton. 11 I hereafter refer to conventional war as war. See Paul Huth, D. Scott Bennett and Christopher Gelpi, “System Uncertainty, Risk Propensity, and International Conflict among the Great Powers,” Journal of Conflict Resolution, September 1992, Vol 36 No. 3 pp. 478-517; D. Scott Bennett and Allan C. Stam, The Behavioural Origins of War, 2003. Michigan: University of Michigan Press, pp. 135-6. 7 the role of time has led scholars to miss a central dynamic of the spread of nuclear weapons.12 This dissertation will address the robust effect of experience with nuclear weapons and show how it influences the conflict propensity of weak and revisionist states. 1.2. Nuclear Proliferation, Conflict and Time Recent studies applied multivariate regression analysis to identify a robust relationship between nuclear proliferation, conflict and time. Michael Horowitz utilised widely agreed upon nuclear weapon proliferation dates and readily available militarised international dispute databases and found that experience with nuclear weapons has a strong effect on the probability that a state reciprocates an international dispute. While the acquisition of nuclear weapons makes states significantly more likely to reciprocate militarised challenges in the short-term, they are less likely to reciprocate in the long term. Moreover, an experienced nuclear power is less likely to reciprocate a military challenge than it was before nuclear weapon development. 12 For an exception, see Phillip Potter, “Does Experience Matter? American Presidential Experience, Age and International Conflict,” Journal of Conflict Resolution 2007 51 (3) pp.351- 379. Potter demonstrated the effect of experience in crisis involvement. American Presidents between 1918 and 2001 were likely to be involved in over twice as many crises in their first two years of office than in subsequent years. Other scholars have begun to address the role of time horizons. See Philip Streich and Jack S. Levy, “Time Horizons, Discounting, and Intertemporal Choice,” Journal of Conflict Resolution, 2007 51 (2) pp. 199-226; Aaron Rapport and Ronald R. Krebs, “International Relations and the Psychology of Time Horizons,” American Political Science Association 2010 Annual Meeting paper; David Edelstein, “Time Horizons and the Causes of Conflict and Cooperation,” American Political Science Association 2010 Annual Meeting paper. 8 Horowitz conducted a statistical analysis that controlled for the balance of power, alliances, superpower involvement, regime type, individual nuclear powers and the issues at stake in the dispute. He found that the probability of nuclear powers reciprocating disputes quickly increases then decreases over time. The probability that a nuclear state will reciprocate a dispute with a non-nuclear state drops from .53 one year after developing nuclear weapons to .23 in year 56.13 Horowitz found that two new nuclear powers are 67% more likely to reciprocate a dispute than two average non-nuclear states. Two experienced nuclear states are 65% less likely to reciprocate than two average non- nuclear states. The probability of dispute reciprocation between an experienced and new nuclear power is 26% greater than two non-nuclear states, and the probability of a very experienced state and a somewhat experienced state reciprocating is 42% less than two non-nuclear states. Horowitz also found that the effect of experience with nuclear weapons is strongest for nuclear challengers.14 Other quantitative research found that nuclear acquisition neither increases nor decreases overall conflict propensity but has important effects on the intensity, duration 13 Michael Horowitz, “The Spread of Nuclear Weapons and International Conflict: Does Experience Matter?” Journal of Conflict Resolution 2009 53 No. 2 (April) pp. 234-257, p.244. 14 Michael Horowitz, “The Spread of Nuclear Weapons and International Conflict: Does Experience Matter?” Journal of Conflict Resolution 53 (2) 2009 pp. 234-257, pp.247-255. One might argue that the small number of cases of nuclear powers undermines the validity of the statistical results. Such challenges may be serious for studies that address the cause of nuclear proliferation and specify it as the dependent variable since there is little variation in it. But studies such as Horowitz’s that address the impact of nuclear proliferation on state conflict propensity specify the latter as the dependent variable. The universe of cases is thus all disputes involving nuclear (and non-nuclear) states, offering significant variation on the dependent variable. As I argue in Chapter 2, within case analysis adopted in this dissertation will aim to establish the causal mechanisms within the correlations that the statistical results identify. 9 and timing of conflict.15 Erik Gartzke conducted a similar statistical test when the dependent variable was dispute initiation rather than reciprocation and found similarly robust results.16 Gartzke found that while the overall effect of nuclear proliferation on conflict propensity is neutral, there is crucial variation in the effect of proliferation over time. The variables that captured variation in nuclear acquisition also captured variation in dispute propensity; nuclear weapons may not cause conflicts because states that acquire nuclear weapons are also more likely to fight. But Gartzke controlled for the distribution of nuclear weapons, conventional military and economic power, regime type, distance, territorial contiguity and alliance membership and found that nuclear acquisition influences the quality and timing of disputes. New nuclear powers are very prone to initiate militarised conflict, and experience makes them less likely to do so. Over the long run they become as likely to initiate disputes as they were before nuclear acquisition.17 15 The most important studies are Erik Gartzke, “Nuclear Proliferation Dynamics and Conventional Conflict,” http://dss.ucsd.edu/~egartzke/papers/nuketime_05032010.pdf, and Erik Gartzke and Dong Joon Jo, “Bargaining, Nuclear Proliferation and Interstate Disputes” Journal of Conflict Resolution 53 (2) 2009 pp. 209-233. Kyle Beardsley and Victor Asal showed that nuclear states are 60% more likely that non-nuclear states to use a military response against a non-nuclear state, and 35% more likely to do so against nuclear opponents. A non-nuclear state is 35% less likely to use a military response against a nuclear adversary than a non-nuclear one. The use of force (what the authors call “violence pre-eminent”) by a non-nuclear state is 60% less likely against a nuclear adversary than a non-nuclear one, and about 75% less likely between two nuclear opponents than two non-nuclear ones. See Kyle Beardsley and Victor Asal, “Nuclear Weapons as Shields,” Conflict Management and Peace Science 2009 (26) pp. 235-255; Beardsley and Asal, “Winning with the Bomb,” Journal of Conflict Resolution 53 (2) 2009 pp. 278-301; Robert Rauchhaus, “Evaluating the Nuclear Peace Hypothesis: A Quantitative Approach,” Journal of Conflict Resolution 53 (2) 2009 pp. 258-277 and Daniel Geller, “Nuclear Weapons, Deterrence and Crisis Escalation,” Journal of Conflict Resolution 34 (1990), pp. 291-310. 16 Gartzke, Proliferation Dynamics 17 Gartzke, Proliferation Dynamics 10 Although Horowitz and Gartzke confirmed that a relationship exists between nuclear weapons proliferation, conventional conflict and time, they have not established how experience matters. This is deeply problematic because only knowing how experience matters allows policies to be formulated that can address strategic objectives. There are several plausible causal mechanisms, and without establishing the most plausible one in future cases we cannot develop policies to engage Iran and North Korea. Policies that address one mechanism may be ineffective or dangerous in others. Two potential explanations involve decision-maker competition and decision- maker learning. In the former, decision-makers in new nuclear powers could disagree over the desirability of developing nuclear weapons. Starting crises or wars could attract resources to a command and control infrastructure and a survivable arsenal and perhaps rally domestic constituencies around the flag. Competition over what resources to commit to a nuclear program could provide incentives for some decision-makers to start crises to commit resources to the necessary infrastructure. This would explain the conflict propensity of new nuclear powers. However once such resources have been committed to the nuclear program, there would be little need for such experienced nuclear power decision-makers to start crises. Decision-maker competition might thus explain the effect of experience with nuclear weapons. In the latter, the high conflict propensity of new nuclear powers reflects a learning process where decision-makers adjust their beliefs about the effectiveness of nuclear coercion.18 Without knowing the causal mechanism 18 Nuclear coercion is defined as nuclear compellence: threats to change some aspect of the status quo backed up by threats, often implicit, of nuclear escalation. Nuclear coercion by weak revisionist new nuclear powers may involve a deployment or use of force and demand for some change to the status quo; inaction will be threatened to lead to further changes to the status quo perhaps made possible by the initial fait accompli. 11 between experience with nuclear weapons and decreased conflict propensity, we cannot formulate policies to ensure that Iran and North Korea behave like other experienced nuclear powers. If the decision-maker competition hypothesis is correct, deterrence policies that threaten punishment for aggression might cause the very aggression that they are designed to prevent. But if the decision-maker learning hypothesis is correct, such deterrence policies might dampen a new nuclear power’s ambitions. Without establishing the causal mechanism it is impossible to know the most appropriate policy. Moreover, it is unclear whether multivariate statistical techniques can establish how experience with nuclear weapons influences conflict propensity. Erik Gartzke found that the shock and decay in dispute behaviour associated with nuclear proliferation resulted from his information and uncertainty variables, and concluded that “uncertainty about nuclear status and a gradual learning process may allow nations to better bargain in the face of differences, tensions, and nuclear weapons.”19 But Gartzke does not establish what lessons pacify new nuclear powers. His proxy for uncertainty about nuclear status may be capturing some of the variation that the learning variable explains; it is thus unclear whether the relevant learning is about an adversary’s nuclear status or something about one’s own nuclear weapons. Moreover, the speed of learning is also unclear. Gartzke modelled learning to occur at a constant rate; if new nuclear power decision- makers tend to learn more than experienced nuclear powers, or perhaps after one episode and not thereafter, his statistical model would not capture this effect but rather model a 19Gartzke, Proliferation Dynamics, p.21 12 constant rate of learning. 20 Only through a more rigorous operationalisation of learning that specifies when leaders learn what lessons can we know how experience with nuclear weapons matters. Process tracing how causal mechanisms operate through historical cases is a method ideally suited to this challenge.21 The nuclear proliferation literature requires such heavy doses of historical analysis to document these mechanisms. Peter Feaver’s 1995 assessment remains accurate: empirical enquiries into how established nuclear states have in fact interacted with emerging proliferators is still a high research priority. David Karl urged scholars to go beyond theoretical arguments over whether proliferation is good or bad and undertake empirical investigations into the actual behaviour of new nuclear powers. More recently, Alex Montgomery and Scott Sagan recently claimed that additional research is required that directly examines the dynamics of escalation from the threat to the use of force and from the use of force to fatal militarised interstate disputes to determine whether the fear of escalation by one side or both determined the governments’ behaviour or whether other variables were more important. Robert Jervis recently claimed that while the most likely current system-changing force is nuclear proliferation, how proliferators would use this security is far from clear.22 Clearly, establishing how experience with nuclear weapons 20 Gartzke, Proliferation Dynamics, p.12, and Gartzke, personal correspondence with author. 21 Alexander L. George and Andrew Bennett, Case Studies and Theory Development in the Social Sciences, 2005. Cambridge: MIT Press 22 Peter D. Feaver “Optimists, Pessimists, and Theories of Nuclear Proliferation Management,” Security Studies 4 (4) 1995 pp. 754-772, p. 769; David J. Karl, “Proliferation Pessimism and Emerging Nuclear Powers,” International Security 21 1997 pp. 87-119, pp. 118-19; Alexander H. Montgomery and Scott D. Sagan “The Perils of Predicting Proliferation,” Journal of Conflict Resolution, 2009 Vol. 53 (2) April pp. 320-328, p. 321; Robert Jervis, “Unipolarity: A Structural Perspective,” in World Politics 2009 61 (1) pp. 188-213, p. 212 13 causes restraint is a high research priority. This dissertation aims to establish this causal mechanism. 1.3.Unpacking the Effect The statistical effect of experience with nuclear weapons is explained by two types of states fighting two types of adversaries. The first group is likely only correlation with no causal relationship between nuclear weapons and state conflict propensity. The second may involve a causal relationship involving nuclear weapons and conflict propensity. In this section I show that the conventional wisdom cannot explain the second group of cases. Some cases involve major powers fighting against weaker non-nuclear adversaries before 1960. The end of World War Two, creation of early Cold War spheres of influence, demise of European colonial empires and rise of nationalist independence movements throughout the third world offers an explanation of the effect of experience with nuclear weapons. These structural changes found the United States and Soviet Union facing a new post-war international system with much of Asia and Europe up for grabs, and the British and French facing extensive unrest in their former colonial empires. 14 The United States and Soviet Union may have intervened in South Korea and Hungary not because they were emboldened by their recently developed nuclear weapons but to establish Cold War spheres of influence. Similarly, Britain may have fought in Egypt not because of nuclear weapons but to maintain great power status and regional influence in the face of impending decline. These structural changes are exogenous to nuclear proliferation and explain the effect of experience with nuclear weapons. It is possible that new nuclear power status influenced conventional war involvement in these early Cold War cases, but also plausible that the Korean war, Soviet response to the Hungarian uprising and Soviet response to the Suez war would have played out similarly without American, Soviet and British nuclear weapon development. By 1960 however, Cold War spheres of influence had been more or less established and the nature of the post-war order was mostly clear. The structural change hypothesis cannot explain another set of cases. These involve revisionist new nuclear powers that lack military superiority over their principal nuclear weapon equipped adversary when fighting them after 1960. The structural change hypothesis cannot explain the most dangerous behaviour of the Cold War – Soviet coercion over West Berlin and the Cuban Missile crisis – without reference to Soviet nuclear tipped ballistic missile development in the late 1950s. Structural variables can explain why the status of West Berlin was such a threat to the Soviet Union but the structural hypothesis cannot explain why Soviet coercion began as late as 1958 without reference to Soviet nuclear missiles. The structural change hypothesis does not explain why India and Pakistan fought a war at the Kargil sector of Kashmir in 1999, Mao Zedong found China in a military crisis with Soviet forces at Zhenbao island in 1969, and Israel fought two wars 15 with its Arab adversaries shortly after developing a small nuclear arsenal. Moreover, since the massive systemic changes and widespread third world liberation movements that underlie the structural change mechanism are unlikely to explain future conflicts in the Middle East and East Asia, we must look elsewhere to anticipate and understand the effect of experience with nuclear weapons in these cases.23 1.4. Explanations: The Conventional Wisdom Fails In this section I consider orthodox explanations and find that they do not explain the effect of experience with nuclear weapons. Glenn Snyder hypothesised that a relationship exists between nuclear deterrence and conflict.24 If the leader of a nuclear power believes that nuclear war with its nuclear weapons equipped adversary is unlikely due to the fear of escalation, he may engage in conventional aggression or other undesirable behaviour not likely to warrant a nuclear response. The key insight was that nuclear weapon development and the belief that it was safe could be necessary for such provocations. Such aggression could be defended against in a pre-nuclear world; without nuclear weapon acquisition such aggression would be, all else equal, less likely to occur. The nuclear revolution undermined defences and transformed nuclear adversaries' incentive 23 The remainder of this dissertation addresses weak revisionist new nuclear powers and refers to them as nuclear powers. 24 Glenn H. Snyder “The Balance of Power and the Balance of Terror,” in Paul Seabury (ed.) Balance of Power, 1965 Chandler: San Francisco, pp. 184-201, p. 199 16 structures. Conflict of interest could now never be total because destruction could be. In the nuclear age, leaders of nuclear states could make challenges to the status quo that were insufficient to warrant a nuclear response but which, because of the danger of nuclear escalation, could not be safely defended against. Such revisionism could be more swiftly contained in the pre-nuclear age. Snyder however suggested that the temptation to be emboldened after nuclear acquisition might be heavily qualified by fear of nuclear escalation. He elaborated – with an important qualification that has rarely been noted in the mostly speculative literature that has assessed his hypothesis – that this greater likelihood of gradual conventional escalation due to a stable strategic equilibrium may also tend to minimise conventional provocation.25 Decision-makers face uncertainty about whether conventional provocation will escalate into nuclear war. This creates uncertainty about our predictions; decision- makers may be emboldened to pursue their revisionist goals but may also fear the possibility of escalation to nuclear war from a conventional crisis. Snyder concluded that the first hypothesis is “probably dominant” but must be “heavily qualified” by the second. How might this explain the relationship between nuclear proliferation, war and time? Kenneth Waltz and others have argued that neither new nor experienced nuclear powers should be war prone. While nuclear weapons provide insurance against external coercion, they do not offer much coercive leverage.26 In this argument, the threat to use nuclear weapons is not credible and nuclear retaliation is highly undesirable, so nuclear 25 Snyder, The Balance of Power and the Balance of Terror, p.199 26 See, for example, Kenneth N. Waltz, “Nuclear Myths and Political Realities,” American Political Science Review 84 (3) 1990 17 coercion will not be effective. If nuclear coercion is ineffective it should not be used. Barry Posen thus argued that if Iran provoked a crisis or war, “Tehran could not rule out the possibility that others with more and better nuclear weapons would strike Iran first.”27 This argument explains why experienced nuclear powers are unlikely to fight but does not address the high war propensity of new nuclear powers. A mechanism is required that explains the effect of experience with nuclear weapons. Why are new nuclear powers – but not the same experienced nuclear powers – war prone? Paul Kapur's contribution to this question stipulated when new nuclear powers would be emboldened as Snyder suggested and became the orthodox wisdom.28 Kapur showed that such nuclear powers have to be both conventionally inferior to their principal adversary and possess revisionist territorial ambitions against them. In Kapur’s argument, nuclear weapon development by a conventionally inferior state without revisionist ambitions should not lead to revisionism because there is no underlying strategic or territorial prize for the new nuclear state. Likewise, nuclear weapon development by a revisionist state with conventional superiority should not cause emboldenment because conventional superiority should have allowed revisionist goals to be satisfied before nuclear proliferation.29 Conventional superiority, revisionist preferences and nuclear 27 Barry R. Posen, “We Can Live With a Nuclear Iran,” New York Times, February 27, 2006. See also Sumit Ganguly, “Nuclear Stability in South Asia,” International Security 33 (2) Fall 2008 pp. 45-70 28 S. Paul Kapur, Dangerous Deterrent: Nuclear Weapons Proliferation and Confllict in South Asia. Stanford: Stanford University Press, 2007; Kapur, “Ten Years of Instability in Nuclear South Asia,” International Security 33 (2) pp. 71-94; Kapur, “Revisionist Ambitions, Capabilities, and Nuclear Instability,” in Scott D. Sagan (ed.), Inside Nuclear South Asia, 2009. Stanford: Stanford University Press, pp.184-218 29 Indeed, this suggests that there should not be any revisionist states that are conventionally superior to their principal adversary because they will have realised their goals through 18 weapon development might cause revisionism if a conventionally inferior principal adversary already possessed nuclear weapons, but such scenarios have not yet occurred.30 On the other hand, Kapur pointed out that nuclear weapon development by conventionally inferior revisionist states would offer them a potential means to challenge the undesirable status quo. Before nuclear proliferation the weaker states' aggression could be either defended against or deterred by its more militarily powerful adversary through threats of conventional or nuclear escalation. Such retaliation often incurs high costs on the weaker power as Pakistan experienced in 1971 when West Bengal became independent Bangladesh. The weaker revisionist state's development of nuclear weapons however constitutes a shield against such escalatory threats. Whereas the conventionally superior state was prepared to respond to aggression before its adversaries nuclear acquisition with escalation because it was not risky, after its revisionist adversaries nuclear acquisition there is no guarantee that such escalation won't cause larger destruction. Moreover, such weaker nuclear powers will often develop nuclear postures that deter nuclear and conventional retaliation.31 Their revisionist preferences and insecurity against a more powerful adversary should make it easier for them to credibly commit to escalate a crisis than their more powerful adversary. The stronger state’s status quo preferences and relatively benign security environment will make it harder to credibly conventional superiority and selected themselves out of this category. One might however argue that China had revisionist goals against conventionally inferior India after nuclear proliferation. 30 Future South Korean proliferation would not count because Seoul has status quo preferences. Iranian proliferation would not count because Iran is not conventionally superior to Israel. 31 See Narang, Posturing for Peace 19 commit to forcefully punish aggression from its nuclear weapon equipped weaker adversary. Conventionally inferior revisionist proliferators thus might believe that the bomb would offer them a means to achieve their revisionist designs.32 This specification of the necessary requirements for when nuclear proliferation would cause conventional war was a critical contribution. Kapur offered an impressive array of interview data, including an interview with former Pakistani Prime Minister Benazir Bhutto, to show that Pakistani nuclear weapons loomed large in the strategic thinking behind the 1999 Kargil War (see Chapter 4). Kapur’s model explains the front end of the relationship between nuclear proliferation and war. But he cannot explain why experienced nuclear powers are not war prone because he does not model time. Kapur has an explanation for Pakistan’s changing behaviour over time, but it is ad hoc and cannot explain the broader cross-national variation. He argued that the September 11 attacks, Operation Enduring Freedom in Afghanistan and President Bush's desire for Pakistani cooperation caused Musharraf to contain his revisionist ambitions in Kashmir.33 But my research suggests that Musharraf had decided to stop sponsoring the insurgency in Kashmir almost six months before 32 Negeen Pegahi's understanding of this mechanism – whereby before nuclear proliferation the conventionally superior state could engage in a devastatingly destructive attack and after proliferation it cannot – is incorrect. Nuclear acquisition by the revisionist adversary changes the conventionally superior status quo state's risk tolerance, not military capability. See Kapur, “Revisionist Ambitions, Capabilities, and Nuclear Instability,” in Scott D. Sagan (ed.), Inside Nuclear South Asia, 2009. Stanford: Stanford University Press, pp.184-218, p.193, p.195 33 Kapur, Ten Years of Instability, pp.85-86 20 September 2001.34 Moreover, during a major war scare in the ten month 2001-02 crisis he experienced fear of nuclear escalation that may have been sufficient to moderate Pakistani revisionism. Kapur’s prediction that Pakistani revisionism in Kashmir will increase after the American commitment to fighting the Taliban ends neglects these alternative explanations and the systematic effect of experience with nuclear weapons. The September 11 attacks may not have been necessary to cause Pakistani restraint in Kashmir. Having shown that orthodox explanations do not explain the effect, I here consider other explanations and find that they do no better. Perhaps experienced nuclear powers are less war prone because their nuclear adversary’s created robust defences after the first crisis or war? If this were true, they should have been deterred from realising their revisionist ambitions by their adversary’s defences. But nothing could have prevented Nikita Khrushchev from reinstalling Soviet nuclear missiles in Cuba or elsewhere or Pervez Musharraf from attempting another fait accompli in Kashmir. Robust territorial defences were impossible in both cases. The rugged and mountainous terrain in the Caribbean, Kashmir and Zhenbao do not allow the construction of strong defences. Israel's successes in its wars after nuclear weapon development may have convinced Egypt and Syria not to similarly challenge it again. But Palestinian successes in Lebanon, Gaza and the West Bank suggest that Israel cannot achieve a perfect defence against Palestinian revisionism. Consolidated defences do not explain the effect of experience with nuclear weapons. 34 Pervez Musharraf, author interview, Seattle, March 2010. He learned from Indian Prime Minister Vajpayee’s willingness to meet him in Agra in April 2001 that his sponsorship of the insurgency in Kashmir had forced Vajpayee to negotiate. See Chapter 4. 21 Secondly, perhaps new nuclear powers are war prone because of other states’ attempts to destroy their nuclear programs or weapons before they become operational. But if this explanation were correct, wars should occur before states become new nuclear powers. Why not attack before the nuclear weapons become operational rather than wait until nuclear escalation is possible? Preventive motivations to destroy nascent nuclear arsenals have often caused crises but rarely cause wars. The historical record suggests that the preventive war hypothesis could only be correct in the 1967 Six-Day war case. But the evidence is inconclusive. That Leonid Brezhnev planned a “decisive blow” to Israel and ordered his air force to target Israel’s secret nuclear enrichment facility at Dimona does not confirm that Soviet concerns about Israeli nuclear weapon development caused the Six-Day war.35 Brezhnev may have intended to cause a crisis that would weaken American influence in the region. Future preventive motivations for war might cause new nuclear powers such as Iran and North Korea to fight, but this case remains inconclusive. Another potential explanation for the effect of experience with nuclear weapons models decision-maker competition over the decision to establish or commit further resources to a nuclear weapons program. Decision’s to divert civilian nuclear programs to military purposes or commit further resources to the nuclear weapons program are presumably contested. Decision-makers who think that nuclear weapons programs are a waste of money might have plausible cases that nuclear weapons are not worth their significant long term financial costs, international isolation, and regional uncertainty and insecurity. Nuclear proliferation might create motives for pro-proliferation decision- 35 Isabella Ginor and Gideon Remez, Foxbats Over Dimona: The Soviets’ Nuclear Gamble in the Six-Day War. New Haven: Yale University Press, 2007 22 makers to fight wars to justify the significant financial commitment and quell opposition from competing decision-makers and the population who do not favour nuclear acquisition. Such a war might also hold the promise of rallying domestic support. People might feel less committed to the state in times of peace but be prepared to make sacrifices in times of war. Such wars might increase the popular support of pro-nuclear decision- makers and undermine that of their anti-nuclear competitors. Moreover, organisational biases within the military might cause incentives to fight after nuclear acquisition to commit further resources to a command and control infrastructure, weaponisation or delivery vehicle development. These institutions divert usually scarce resources from elsewhere, and civilian leaders may not want the financial commitment to this infrastructure that military elites’ desire. A war would strain the plausibility of the former’s case. If this hypothesis is correct, nuclear proliferation should give rise to a conventional war in the short term as the pro-nuclear group quell resistance from anti-nuclear members of the civilian, military or non-state arenas. The ascendency of the former over the longer term should provide fewer incentives to fight. Nuclear programs are rarely dismantled, and sunk costs and institutional inertias would reduce incentives to fight to commit further resources to a nuclear weapons program. The decision-maker competition hypothesis would thus lead us to expect that experienced nuclear powers would not fight. If it is correct, intra-decision-maker rivalry should be greatest immediately after acquisition of nuclear weapons and necessary delivery vehicles and decrease over the longer-term. The decision-maker competition explanation does not however find any empirical support. Israeli decision-makers were divided over whether nuclear weapons should have 23 been developed in 1967 but there is no evidence that any of them would have been motivated to start a war to safeguard the program. The decision-maker competition model does not explain the Chinese case because Mao dominated Chinese politics.36 Moreover, the Soviet Union and Pakistan exhibited the reverse dynamic. The decision-maker competition hypothesis would predict that contestation amongst the Soviet political elite would be greatest shortly before the first Berlin crisis in 1959. But by this time Khrushchev had consolidated his control over Soviet policy. Moreover, the 1957 coup attempt that resulted in the political isolation of the remaining Stalinists does not seem to have been about the Soviet nuclear program but whether the underlying Stalinist principles of Soviet foreign policy should have been discarded.37By 1955, Khrushchev had assumed leadership in the Soviet Presidium (former Politburo) after organising the death or marginalisation of his three colleagues that found themselves in charge of Soviet decision-making after Stalin's death in 1953. Decisions in the Presidium came to be based on a majority-win voting system amongst the dozen senior Presidium members, but over time Khrushchev replaced the aging Stalin era politicians with younger loyalists and consolidated his control over Soviet policy. His more outlandish hawkish policies were often challenged, but by 1957 and certainly 1960 Khrushchev usually got his way with at most minimal qualifications.38 In 1958 Khrushchev had initially proposed the sudden signing of a peace treaty with East Germany but was forced to concede to a six month 36 See Roderick MacFarquhar and Michael Schoenhals, Mao’s Last Revolution, 2008, Harvard University Press; Philip Short, Mao: A Life, 2004, John Murray. 37 On the 1957 coup see William J. Thompson, Khrushchev: A Political Life, 1997. New York: St Martin's Griffin, pp.179-184; William Taubman, Khrushchev, 2003. New York: Norton, pp. 310- 324; Fursenko and Naftali, Khrushchev's Cold War, pp. 145-149. One might argue that the debate over Stalinist foreign policy would have at least partly been about nuclear weapons. But Khrushchev began his coercion in 1958 after he had deposed the coup plotters from the Presidium. 38 Fursenko and Naftali, Khrushchev’s Cold War, pp. 15-32 24 ultimatum. In the 1961 Berlin crisis Khrushchev fathered all the offensive policies directed at the United States; there was no pressure to take more risk. By early 1962 Khrushchev's two unsuccessful Berlin bluffs caused no outward revolt amongst Presidium members. After the May 1962 Presidium meeting, when Khrushchev first announced his Cuban Missile strategy for Berlin, descended into discord, Khrushchev halted the meeting and asked for a three day recess. The subsequent vote was unanimous.39 Weeks later he got all the intermediate and tactical missiles that he asked his colleagues for and radically transformed Soviet Berlin policy without any debate. Rather than dissent causing the crises, the crises caused dissent. The risk that Khrushchev engaged in by putting missiles in Cuba and the little gains that this offered, in addition to his generally authoritarian management of Kremlin decision-making, caused several Presidium members to stage a coup against him in 1964. Soviet restraint in Western Europe and the Caribbean after October 1964 may have been caused by decision-maker competition during the Brezhnev era, but such competition did not cause the Berlin and Cuban Missile crises. The Pakistani case also falsifies the decision-maker competition explanation. The 1999 Kargil war was not caused by a desire to overcome political or military resistance to the nuclear program; India’s nuclear test would have eliminated any such resistance. Nor is it likely that competition within the military caused preferences for the war to attract 39 Kremlin Decision-Making Project, Miller Centre for Public Affairs (hereafter referred to as KDMP. These online notes seem to mostly be stenographic accounts of State Meetings, referred to in Fursenko and Naftali's Khrushcev's Cold War as Archives of the Kremlin or AOK). Available at http://millercenter.org/scripps/archive/kremlin, accessed September 14, 2010), May 21 1962 25 resources to the nuclear program because the threat from India already ensured the nuclear program’s importance. Pakistan has been characterised by intense competition between the civilian leadership and army for decades. While much remains unknown about these dynamics, it appears that Pakistani Prime Minister Nawaz Sharif signed off on the Army- proposed Kargil plan while he remained unaware of its full operational details or likely consequences. This is hardly a story of competition between the civilian and military leadership or hawkish and dovish military authorities causing the use of force to justify a nuclear program. Rather, similarly to Cuba, the war was not caused by but exacerbated civil-military relations. Disagreements over how to navigate Pakistan out of an unsuccessful military stalemate in Kashmir caused General of the Army Staff Pervez Musharraf to take control of the government within months of the wars’ end. Pakistani restraint in Kashmir between 2001 and 2007 is unlikely to have resulted from decision-maker competition because Musharraf's hold on civilian and military power was quite high. American diplomatic pressures amidst Operation Enduring Freedom in Afghanistan have increased the incentives for restraint since 2008, but it is unlikely that reduced decision-maker competition in the post-Musharraf years caused restraint in Kashmir. At best, Nawaz Sharif and Asif Zardari's preoccupation with power sharing, the public furore over Musharraf's alleged corruption and Pakistan's economic malaise prohibited further aggression in Kashmir. The decision-maker competition model does not speak to these developments. A final and perhaps obvious explanation is changes in the balance of military and/or economic power. Structural realism assumes that these variables explain variation 26 in war and alliances, although there is sharp disagreement over what type of balance of power causes what level of international violence. Do changes in the balance of power explain the effect of experience with nuclear weapons? Unfortunately, such structural variables do not explain the effect. At most, they explain why new nuclear powers are war prone insofar as nuclear proliferation changes the balance of power. But this seemingly obvious hypothesis does not explain why the same experienced nuclear powers are not war prone. It is also unclear that weapons that most people believe are militarily unusable should be included in assessments of the military and economic balance. More importantly, the balance of non-nuclear economic and military power in the Cold War and South Asian rivalries did not co-vary with their war propensity. The Soviet Union and Pakistan did not develop more military power or economic growth before or after the Cuban Missile Crisis and Kargil war. Variation in Chinese clashes with Soviet forces during the 1960s is not explained by changes in the balance of power. Explanations based on the balance of military and economic power do no better in explaining the effect of experience with nuclear weapons than domestic politics explanations. Six decades into the nuclear age, perhaps the greatest effect of nuclear weapons on human warfare has only just been established and remains inadequately understood. There is a systematic effect of experience with nuclear weapons on the conflict propensity of weak revisionist powers. Yet neither the conventional wisdom nor other explanations can explain why such new nuclear powers are highly conflict prone but the same experienced states are as conflict prone as they were before developing nuclear weapons. Realism and liberalism, the two long dominant approaches to the study of international conflict, offer little to explain the relationship between nuclear proliferation, 27 conflict and time. Uncertainty characterises the immediate aftermath of nuclear proliferation, and usually generates opportunities for learning. This raises several questions. Can the individual level of analysis, and more recently developed psychological and constructivist theories, explain the effect? Perhaps nuclear proliferation become partly safe as people believe it is unsafe. Does learning about the political utility of nuclear weapons explain why experienced nuclear powers do not fight? What causes such beliefs to form and change? We need to explain why weak revisionist experienced nuclear powers do not fight with a causal mechanism that also explains their earlier belligerence. The next chapter develops psychological and rational choice learning models to do this. 28 Chapter 2: Two Nuclear Learning Models 2.1. Two Nuclear Learning Models The previous chapter identified a robust relationship between nuclear proliferation, conflict and time. Weak revisionist new nuclear powers are conflict prone but the same experienced nuclear powers are not. Modifications of the two orthodox approaches to the study of the consequences of nuclear proliferation fail to explain this variation. Horowitz and Gartzke’s multivariate regressions controlled for the balance of military and economic power, regime type and distance; Gartzke found that his information variables were statistically significant. Decision-maker learning over time about the benefits of nuclear coercion seems a plausible explanation of the effect. In this chapter, I develop a psychological nuclear learning model that I contrast against a rational choice baseline model. The study of learning and foreign policy change became quite popular amongst International Relations scholars in the early 1990s as a mechanism to explain Soviet foreign policy change and the end of the Cold War.40 Disagreements over definitions and several methodological problems led most scholars to quickly lose interest in the concept. By the mid 1990s, with the rising popularity of constructivism as a new theoretical 40 Much of the learning literature was motivated by the search for an explanation of Soviet policy changes in 1989-1991. Some of the key contributions were Philip Tetlock, “Learning in U.S. And Soviet Foreign Policy: In Search of an Elusive Concept,” in George Breslauer and Philip Tetlock (eds.) Learning in U.S. And Soviet Foreign Policy, 1991. Boulder: Westview Press; Jack S. Levy, “Learning and Foreign Policy: Sweeping a Conceptual Minefield,” International Organization 1994 48 (2), Yuen Foong Khong, Analogies at War, 1992. New Jersey: Princeton University Press, Janice Gross Stein, “Political Learning by Doing: Gorbachev as Uncommitted Thinker and Motivated Learner,” International Organization 1994 48 (2); Stein, “Deterrence and learning in an enduring rivalry: Egypt and Israel, 1948-1973,” in Security Studies 6 (1) 1996, and Andrew O. Bennett, Condemned to Repetition? The Rise, Fall and Reprise of Soviet-Russian Military Intervention, 1973-1996. MIT Press: Cambridge, Massachusetts, 1999. 29 approach and the publication of Peter Katzenstein's important edited volume The Culture of National Security, interest in learning among International Relations scholars plateaued.41 Thus Janice Stein noted in 2002 that while the “analysis of political learning should be a central focus of analysis,” “surprisingly it has not been.”42 Yet the key variables in constructivist explanations – actor identities, shared norms and constitutive structures of normative values – presuppose assumptions about beliefs, perception and information. People must learn their own and others’ identities, and what is learned and relearned depends on the distribution of information and variation in beliefs, perceptions and interests. Nuclear proliferation might change actors’ identities, but such ideational dynamics are caused by perceptions of uncertainty about how this might change their behaviour and beliefs about the most probable consequence. Constructivist theory has little to offer an explanation of the relationship between nuclear proliferation, conflict and time without reference to these psychological variables. Below I define what does and does not count as learning and offer two empirically falsifiable nuclear learning models; one psychological, one rational. I adopt Jack Levy’s definition of learning as a change of beliefs, degree of confidence in one's beliefs, or the development of new beliefs, skills or procedures as a result of the 41 The literature is now vast, but key contributions include Peter Katzenstein (ed.), The Culture of National Security, 1996. New York: Columbia University Press; Alexander Wendt, Social Theory of International Politics, 1999. New York: Cambridge University Press; Ted Hopf, “The Promise of Constructivism in International Relations Theory,” International Security 23 (1) 1998 pp. 171- 200; James Fearon and Alexander Wendt, “Rationalism v. Constructivism: A Skeptical View,” in Walter Carlsnaes et al, Handbook of International Relations, 2005. Sage: California, pp. 52-72; Martha Finnemore and Kathryn Sikkink, “International Norm Dynamics and Political Change,” International Organisation 52 1998 pp. 887-917 and Richard M. Price, “Reversing the Gun Sights: Transnational Civil Society Targets Landmines,” International Organisation 1998 52 pp. 613-644 42 Janice Gross Stein, “Political Learning and Political Psychology: A Question of Norms,” in Kristen Renwick Monroe (ed.) Political Psychology, 2002. New Jersey: Lawrence Erlbaum Associates, pp. 107-115, p. 107. 30 observation and interpretation of experience.43 Causal learning refers to changing beliefs about the laws or hypotheses of cause and effect, the consequences of actions and the optimal strategies under various conditions. Such learning might cause either a change of preference, or, more commonly, different strategies to achieve the same preference or different tactics to pursue the same strategy.44 Diagnostic learning refers to changes in beliefs about the definition of the situation or the preferences, intentions, or relative capabilities of others.45 For the purposes of this dissertation, nuclear learning is causal learning and is defined as learning about the efficacy and dangers of nuclear coercion.46 Such causal learning might give rise to diagnostic learning. Learning is not a passive activity in which historical events generate their own lessons that all actors absorb. People actively interpret historical experience through the lens of their own analytical assumptions and beliefs. If all political leaders drew the same inferences from experience, and if similar experiences led to similar inferences, learning would be epiphenomenal and explain little of the variation in policy outcomes. The claim that organizations or governments can be treated as organisms that have goals, beliefs and memories is not analytically viable. Organisations only learn through the individuals who serve in them and who encode individually learned inferences from experience into organizational routines and governmental procedures.47 Learning is also distinct from 43 See Levy, Learning and Foreign Policy, p.283. Much of the next few paragraphs is influenced by Levy’s discussion. 44 Tetlock, Learning in U.S. And Soviet Foreign Policy, pp. 27-31. 45 Levy, Learning and Foreign Policy, p.285 46 Nuclear coercion is defined as nuclear compellence: threats to change some aspect of the status quo backed up by threats, often implicit, of nuclear escalation. Nuclear coercion by weak revisionist new nuclear powers may involve a deployment or use of force and demand for some change to the status quo; inaction will be threatened to lead to further changes to the status quo perhaps made possible by the initial fait accompli. 47 Organisational learning thus involves a multistage process where environmental feedback 31 turnover; turnover might result from generational change, political competition or realignment. Leadership succession introduces new members with different goals, beliefs and abilities and need not involve learning. This definition of learning differs from alternative conceptions in the literature in five important respects. Firstly, learning is neither necessary nor sufficient for policy change. States alter their foreign policies because of changes in the external environment, leadership change, bureaucratic or societal coalition realignment and individual belief change about the optimal means of achieving different goals. Only the latter involves learning. Actors may learn from experience but be prevented by domestic, economic or bureaucratic constraints from implementing their preferred policies. A great deal of research suggests that learning may reinforce decision-makers current beliefs and inhibit policy change.48 If we follow William Jarosz and Joseph Nye's advice and conflate learning with policy change, we cannot understand when learning does and does not cause such change.49 causes individual learning, learning causes individual action to change organizational procedures and procedure change causes a change in organisational behaviour, which may lead to further feedback. Government learning is even more problematic than that of organizations because it involves the aggregation of learning by multiple organisations and multiple individuals who may act either through organizations or independently of them. Whether organizational or government learning is more important than individual learning by top government decision-makers as a source of policy change depends in part on the degree of centralisation and bureaucratisation in the political system and the nature of the issue area. Organisational learning may be more important than individual learning in international economic or environmental policy than in security policy and more important in arms control and issues relating to force structure and strategic doctrine than in decisions to use military force. See Levy, Learning and Foreign Policy, pp. 288-89. 48 Robert Jervis, Perception and Misperception in International Politics, 1976. New Jersey: Princeton University Press, pp 217-287, Richard Nisbett and Lee Ross, Human Inference: Strategies and Shortcomings in Social Judgement, 1980. Prentice Hall; Philip E. Tetlock, Expert Political Judgement: How Good Is It? How Can We Know?, 2005. New Jersey, Princeton University Press 49 William W. Jarosz and Joseph N. Nye, Jr, “The Shadow of the Past: Learning from History in National Security Decision Making,” in Philip E. Tetlock et al, (eds.) Behaviour, Society and International Conflict, 1993, p. 130, 180 New York: Oxford University Press. Peter Hall adopts a similar definition. See “Policy Paradigms, Social Learning and the State: The Case of Economic 32 Moreover, we may neglect much learning that occurs. Secondly, learning necessarily involves neither developing more accurate beliefs nor beliefs that better enable the achievement of strategic objectives. Belief change is not always accurate, and knowledge does not always translate into skill. The accuracy criterion does allow the differentiation of learning from less interesting forms of belief change such as paranoid thinking; if learning is belief change, why call it learning? Yet this stipulation requires a standard by which we can identify and measure accuracy. In the absence of a consensus on empirical laws by which the accuracy of learning can be assessed, observers will evaluate learning on the basis of their own causal assumptions and normative biases. The correct lessons of the Vietnam war or U.S. President Obama's recent economic stimulus package will not be resolved soon. If learning requires accuracy, we cannot empirically investigate the accuracy of learning or the conditions, issue-areas, and organizational structures that facilitate it. Moreover, the accuracy criterion requires that we know what information an actor had available at an earlier time. If somebody changed their beliefs based on evidence which at the time pointed toward firm conclusions but which was later challenged by new evidence, are we prepared to not call this learning? These problems also apply to efficiency. No consensus exists on metrics for measuring efficiency – some might include the economic use of time, money or even emotional investment – and what reasonably appeared efficient at one time may not appear so later when new information is available. The question of accuracy and efficiency in the observation and interpretation of experience is not unimportant but should be resolved by empirical analysis and not definition. Janice Stein has argued that an “evaluation of the structure and content of cognitive Policy Making in Britain,” Comparative Politics 1993 25 (3) pp.275-296 33 change is inescapably built into the concept of political learning.”50 “Defining learning as any belief change,” Stein argues “makes the concept devoid of content and a theoretically irrelevant term.”51 Most scholars would indeed not only be interested when leaders learn what lessons but when they learn a certain type of lesson such as those involving the initiation or resolution of wars. But we do not know enough about the different contexts when leaders learn different lessons to confine our definition of learning to certain types of belief change. A research program that defines learning as belief change can address when leaders learn what lessons, and this can establish the types of belief change that cause policy change or more specific lessons. Thirdly, learning need not involve change in the structure (rather than the content) of beliefs. Philip Tetlock suggested that learning involves greater complexity and capacity for self-criticism in the cognitive structure of one's image of the international environment. Such cognitive structural learning involves any increase in the number of distinct arguments underlying a set of beliefs, the degree of inconsistency among them, and the extent to which distinct elements are integrated into a larger framework that resolves inconsistencies and value trade-offs.52 The increased complexity criterion however smuggles the accuracy criterion and its attendant problems through the back- door. Complex cognitive structures do not always cause a better understanding of the world or a better ability to advance one's own interests. Under some conditions cognitive complexity might impede accurate learning rather than enhance it. More cognitively complex belief systems are less falsifiable and more resistant to change in the face of discrepant information because they make it easier for actors to assimilate inconsistent 50 Stein, Learning By Doing, p.171 51 Stein, Political Learning and Political Psychology, p. 112 52 Tetlock, Learning in U.S. And Soviet Foreign Policy, pp.32-35. 34 information into their existing beliefs as exceptions. Increases in cognitive complexity may be a way of avoiding belief change rather than facilitating it. Moreover, conflating learning and cognitive complexity prohibits assessing when cognitive complexity causes learning. Whether cognitive complexity causes learning should be resolved by empirical analysis and not definition. Fourth, learning models of foreign policy change are distinct from models of structural adjustment. The latter assume that individuals recognise changing environmental conditions but that there are well defined and predictable relationships between structural antecedents and behavioural responses. States that adapt to the environment will behave similarly under similar conditions at different times. Learning models assume that these intervening linkages are much less deterministic. Leaders that learn may behave differently under similar conditions in different cases due to variation in prior beliefs, information processing capabilities and the interaction of these two variables with incoming stimuli from the external environment. This will usually lead to different perceptions about the consequences of behaviour and different preferences under similar structures. Both structural adjustment and learning models involve learning, but in the structural adjustment model changing external forces, not learning, influence behaviour. This permits structural analysts to ignore the intervening perceptual process and construct a more parsimonious explanation. Neorealism, for example, posits that a given international structure has a fairly well defined set of consequences such that structures determine or at least severely constrain outcomes. Intervening perceptual variables have no explanatory power.53 In the learning model, intervening perceptual variables are not epiphenomenal but vary across actors in terms of both the accuracy and rate of learning 53 Kenneth N. Waltz, Theory of International Politics, 1979. New York: Mcgraw-Hill. 35 that they generate and thus have significant explanatory power. Neorealist models involve the rational or at least efficient adjustment to changing structural incentives, whereas learning models usually possess significant variation in individual responses to structural changes deriving from variation in cognitive structures, beliefs and perceptual processes. Finally, rational choice learning models operationalise learning as preference adjustment to a changed distribution of information. Unlike structural adaption models, game theoretic rational choice models place the transitive preference maximising individual at the centre of analysis. Yet Bayesian probability updating follows directly from environmental stimuli such that, like structural adaption models, mediating cognitive variables have no independent impact. Rational choice learning models assume that individuals with equal preferences exposed to the same information will exhibit equal behaviour; variation in the individual perception and processing of external stimuli is either not modelled or assumed away. Moreover, rational choice models assume that individuals will always scan the environment for a sufficient variety of information to make the optimal or near optimal decision. Thus in his account of the Cuban Missile Crisis, Harrison Wagner argued that to understand how President Kennedy's behaviour might have caused Nikita Khrushchev to revise his estimate of the probability of an American strike on Cuba, we do not need to address the Soviet leader's beliefs or information processing abilities. Anybody with similar preferences and information to Khrushchev would have, according to Wagner's model, been attentive to the same important parts of President Kennedy's behaviour, reached the same conclusions and acted similarly. The key assumption here is that all individuals would have scanned the environment sufficiently to receive enough 36 information to make the best or near-best decision and interpreted the important information similarly. New information about adversary resolve is therefore, in rational choice models, necessary for learning. Because a bluffer can always mimic the behaviour of a non- bluffer, there has to be a cost to bluffing for learning to occur.54 Bluffing in a costly- bluffing world signals resolve to incur costs to achieve goals that generates learning. If bluffing is not costly it does not signal resolve and bluffing generates no possibilities for learning. Optimal attention and information processing assumptions ensure that behaviour that offers no new information but that might otherwise be salient will not cause learning. Thus Wagner stipulated that if one player is believed to be certain to issue a threat and does, his opponent will not change his probability estimate of further undesirable behaviour because no new information has been received.55 Psychological models on the other hand would predict that threats when bluffing is not costly might cause learning because of variation in different prior beliefs, attention spans and information processing capabilities. In the next section I develop two empirically falsifiable rational choice and psychological nuclear learning models to explain the effect of experience with nuclear weapons. A rational choice framework is an obvious choice because these models are the predominant analytical tool within North American political 54 R. Harrison Wagner, “Uncertainty, Rational Learning and Bargaining in the Cuban Missile Crisis,” in Peter C. Ordeshook (ed.) Models of Strategic Choice in Politics, 1989. Ann Arbor: University of Michigan Press, pp.177-205, p.201 55 Op. cit., p.186. For similar assumptions in similar models, see James D. Fearon, “Domestic Political Audiences and the Escalation of International Disputes,” American Political Science Review, Volume 88 (3) 1994 pp. 577-592 and Alastair Smith, “International Crises and Domestic Politics,” American Political Science Review 92 (3) 1998 pp. 623-638. Recent game theoretic approaches have begun to explore the conditions when costless signalling can cause belief change. See Robert F. Trager, “Diplomatic Calculus in Anarchy: How Communication Matters,” American Political Science Review 104 (2) 2010 pp. 347-368 37 science and international relations, and recent scholarship, as I will show, has begun to model learning processes. A psychological framework is also plausible to model nuclear learning and explain the effect of experience with nuclear weapons because there are grounds to believe that psychological variables explain processes of learning better than abstract deductive rational choice assumptions. 2.2. Nuclear Proliferation and Nuclear Coercion in a Rational Learning Model Most rational choice models used in International Relations employ expected utility theory in game theoretic contexts. Behaviour here can be conceptualised as choices from a group of alternatives, each of which has a value and probability of occurrence. According to the theory, people assess the probability and calculate the utility of each outcome and choose the option that maximises probability-weighted utility. Despite their popularity, rational choice game theoretic models have not been applied to explain the effect of experience with nuclear weapons or the effect of nuclear weapons on state conflict propensity. Such explanations would model a two player game with complete information about the changed distribution of capabilities but incomplete information about the distribution of resolve. Thus Peter Feaver and Emerson Niou, in the only published theoretical model of the consequences of nuclear proliferation, concluded that the solution to their game hinges upon determining whether the proliferator is risk acceptant or risk averse under conditions of complete information wherein each side is uncertain of the other’s resolve.56 In such game theoretic models, uncertainty about 56 . See Feaver and Niou, “Managing Nuclear Proliferation: Condemn, Strike or Assist?,” International Studies Quarterly, 1996 40 pp.209-234, p.219. Feaver and Niou do not attempt to 38 resolve causes a crisis or war, fighting war alleviates these problems because battlefield behaviour reveals true resolve, and the crisis or war ends when these problems have been reduced sufficiently or eliminated.57 The most likely mechanism to explain the effect of experience with nuclear weapons would involve the revisionist new nuclear power decision-maker challenging the status quo and learning about the ability of their nuclear weapons to revise the status quo given their adversary’s resolve to contain this.58 Throughout this dissertation, I use the term rational choice models to refer to prominent rational learning models that share a common set of assumptions about how learning occurs. I do not intend to address or critique the diverse array of rational choice models used in Political Science and International Relations that do not model learning. Most rational choice learning models assumed that bargaining and learning stop once bargaining fails and fighting begins.59 These have little to offer a rational learning explain the relationship between nuclear proliferation and war; they are concerned with establishing the optimal U.S. policy for engaging new nuclear powers. 57 Dan Reiter, How Wars End, 2009. New Jersey: Princeton University Press, pp. 2-3. Unless pre- proliferation reputations are so strong that new data won’t require probability updating, priors should be updated based on the probability that the crisis behaviour reflects a resolved type. It is actually unclear that worrying about such reputations is rational since research suggests that reputations do not influence past attributions and future predictions of crisis behaviour. See Jonathan Mercer, Reputation and International Politics, 1996. Ithaca: Cornell University Press; Daryl Press, Calculating Credibility: How Leaders Assess Military Threats, 2005. Ithaca: Cornell University Press. 58 Alternatively, the experienced nuclear power elite may challenge the status quo out of the desire to destroy their weaker adversary’s new nuclear capabilities. They could never be sure that belligerence by their weaker adversary would not be evidence of imminent aggression. This mechanism could also cause the weaker new nuclear power to fight; elites would plausibly worry about their more powerful adversaries preventive motivations and be unsure of their resolve to destroy their offensive military capabilities. 59 The literature is vast. Some important contributions are James Fearon, “Rationalist Explanations for War,” International Organization 1995 (Summer) pp. 379-414; Fearon, “Signalling Versus the Balance of Power and Interests: An Empirical Test of a Crisis Bargaining Model,” Journal of Conflict Resolution 1994 (February) pp. 236-69; Robert Powell, In the Shadow of Power, 1999; Princeton: Princeton University Press; Kenneth Schultz, “Do 39 model of the effect of experience with nuclear weapons because the learning would occur after fighting began.60 Some influential recent models have however relaxed this assumption and modelled bargaining and learning during fighting. In Robert Powell’s model, one learns about another’s type “by watching how that actor behaves and then updating its beliefs about that actor’s type in light of the observed information.”61 In Branislav Slantchev’s model, “players use all the information available;” “the uninformed state learns about its opponent by observing both its strategic behaviour at the bargaining table and the nonmanipulable battlefield performance.”62 In Mark Crescenzi’s model, states assess the information available from all the possible other states in the system to determine what their dyadic partner would do in crisis situations; “learning is assumed to be experiential in that states learn from the experiences and behaviour of other states; diagnostic in that states use the experiences of others to update their beliefs about the intentions of others; and vicarious, or diffuse, in that states learn from experiences in which they are not directly involved.”63 Michael Tomz assumed that investors update Democratic Institutions Constrain or Inform?” International Organization 1999 53 (Spring) pp. 23-66; Smith, International Crises and Domestic Politics, Harrison R. Wagner, “Peace, War and the Balance of Power,” American Political Science Review 1994 88 (September) pp. 593-607. 60 Harrison R. Wagner, “Bargaining and War,” American Journal of Political Science 2000, 44 (July) pp. 469-484. Other research modelled learning but assumed that it ended after one or a few interactions. These models would have little to offer a rational choice explanation of the effect of experience with nuclear weapons because learning may occur after a few interactions of negotiating or fighting. See Darren Filson and Suzanne Werner, “A Bargaining Model of War and Peace,” American Journal of Political Science 2002 46 pp. 819-38; Branislav Slantchev, “The Power to Hurt: Costly Conflict with Completely Informed States,” American Political Science Review 2002 (47) pp. 1-11; Alastair Smith, “Fighting Battles, Winning Wars,” Journal of Conflict Resolution, 1998 42 (June) pp. 301-320; Alastair Smith and Allan C. Stam, “Bargaining and the Nature of War,” Journal of Conflict Resolution 2004 48 (6) pp. 783-813 61 Robert Powell, “Bargaining and Learning While Fighting,” American Journal of Political Science, 48 (2), April 2004, pp.344-361, pp. 358-59. 62 Branislav L. Slantchev, “The Principle of Convergence in Wartime Negotiations,” American Political Science Review 97 (4) November 2003, pp. 621-632, p.622 63 Mark J. C. Crescenzi, “Reputation and Interstate Conflict,” American Journal of Political Science, 51 (2) April 2007 pp. 382-396, p.384 40 their beliefs about a government in response to new facts. At each stage in the learning process, investors strike a compromise between their prior views and fresh data. Although Tomz claims that his argument “does not require that all people possess identical cognitive abilities,” it does assume that all people will use their different cognitive abilities to reach similarly optimal updating of the same information.64 These models make common assumptions about attention to the information relevant to realise preferences and belief updating based on this information. Given these assumptions in the most influential rational choice research on learning, a critical expected causal process observation of rational choice learning explanations of the effects of experience with nuclear weapons would be that new nuclear power decision-makers’ attention spans and information processing capabilities render them attentive to enough history or logic to accurately estimate the probability of successful nuclear coercion. If Bayesian updating assumptions model sufficient cognitive capacity to infer key information about adversaries, the obvious interests and potential costs at stake imply that decision-makers should also assess the probability of nuclear coercion realising their objectives. Rational choice models would assume that the subjective probability of effective nuclear coercion should be driven by a systematic analysis of all available diagnostic information about all cases across space and time that are deemed relevant, generating a subjective probability close to the objective probability.65 Whether the relevant information is assumed to be the adversary’s 64 Michael Tomz, Reputation and International Cooperation: Sovereign Debt Across Three Centuries, 2007. New Jersey: Princeton University Press. 65 A possible further expected observation is an assessment of the resolve of the experienced nuclear adversary. The new nuclear power elite should be unsure of the experienced nuclear power elite’s resolve to forcefully respond to nuclear coercion and how much a resulting crisis 41 battlefield performance (Powell), bargaining behaviour and battlefield performance (Slantchev) or the universe of possible conflict dyads (Crescenzi), rational choice models assume that decision-makers are attentive to the relevant data and weigh it properly in relation to its diagnostic value. Three problems challenge a specification of how much nuclear coercion rational choice learning models predict new nuclear powers will pursue. Firstly, there are grounds for arguing that that it is not clear what a rational assessment of the historical record on nuclear coercion looks like given different interpretations of historical cases and their applicability to contemporary cases. A decision-maker who believed that some historical cases of nuclear coercion are successful or irrelevant to his or her own would, on this argument, rationally pursue more nuclear coercion that one who believed otherwise. Secondly, one might argue that rational choice game theoretic analyses tend to model the distribution of information and preferences and generally do not model belief change or learning from history. Rational learning models might therefore not be a competitive theory to explain the effect of experience with nuclear weapons. Thirdly, random errors could distort rational judgements. I address each of these points in turn. Irrespective of what one defines as the relevant universe of cases in assessing the efficacy of nuclear coercion, a rational analysis that is attentive to the key variables in the relevant cases must conclude that its success rate is very low. Matthew Fuhrmann and Todd Sechser, for example, conducted a statistical analysis of all cases of nuclear may escalate. The new nuclear power elite might believe that his experienced nuclear power adversary will be restrained because his own nuclear weapons could cause escalation, but may also believe that his adversary understands this dynamic and will respond to his coercion with prompt escalation. Similarly, he may believe that nuclear coercion will be effective or might believe that it would quickly escalate, perhaps to nuclear war. 42 compellence and found that having nuclear weapons does not make compliance more likely; the average success rate of nuclear coercion is zero.66 Soviet and American threats in the Suez and Cuban Missile crises come closest to successful cases of nuclear coercion, but other variables in both cases – as I will show in chapter 3, the depreciating value of the British Pound and Khrushchev’s experience of fear of nuclear war – caused compliance. One can debate whether the effectiveness of nuclear coercive threats in these cases is minimal or zero, but a decision-maker who behaves consistently with Bayesian assumptions of rational choice models would not ignore these other variables or conclude that other cases of nuclear coercion were successful.67 Thus in his account of (subjectivist) Bayesian assumptions about rational beliefs, Dale Poirier claimed that “the only restriction to be placed on such personal degrees of belief is that the individual never assigns probabilities to events in such a way that...the individual loses no matter what the outcome of the lottery.”68 If nuclear coercion is not effective, beliefs to the contrary are irrational. Rational analysis must therefore conclude that nuclear coercion is unlikely to realise foreign policy objectives. Rational choice learning models would predict that however attentive to history the decision-maker is, she would not pursue a nuclear coercion strategy that is almost certain to fail but rather, if she presides over a weak revisionist state, fit her ambitions with her capabilities and give up on what she cannot 66 Todd Sechser and Matthew Fuhrmann, “The Coercive Limits of Nuclear Weapons,” presented at the International Studies Association Annual Meeting, Montreal, 2011. See also Lebow and Stein, We All Lost the Cold War, p. 364; McGeorge Bundy, Danger and Survival: Choices About the Bomb in the First Fifty Years, 1988. New York: Random House; Richard K. Betts, Nuclear Blackmail and Nuclear Balance, 1987. Washington D.C.: Brookings Institute Press. 67 An obvious implication of this is that the many cases of nuclear compellence should not have occurred if eites behaved rationally. The psychological nuclear learning model outlined below explains the causes of these beliefs. 68 Dale Poirier, “Frequentist and Subjectivist Perspectives on the Problems of Model Building in Economics,” Journal of Economic Perspectives, Vol. 2, No. 1 (Winter 1988) pp. 121-144, p.122 43 achieve. Even if the status quo is intolerable, rational choice learning models would predict that she would learn enough from history to accurately consider the costs and benefits of nuclear coercion and not pursue the strategy. Rational choice nuclear learning models would thus lead us to expect that states would not practice nuclear coercion. The second criticism is correct to note that most rational choice game theoretic accounts model the distribution of information and preferences. But recent research by influential scholars such as Robert Powell and Branislav Slantchev has modelled learning during wars. The criticism that influential game theorists do not model learning can no longer be sustained. The third criticism suggests that even if it is true that rational decision-makers should not practice nuclear coercion, random errors might cause it. But if errors are random, new nuclear powers should pursue nuclear coercion randomly and not systematically. A rational choice explanation of the effect of experience with nuclear weapons that allows for error would lead us to expect that decision-makers in weak revisionist new nuclear powers would believe that nuclear coercion is their optimal strategy as often as they conclude that it is not. A higher proportion of beliefs that nuclear coercion is an optimal strategy would be error that is systematic rather than random, and, as argued above, inexplicable through rational choice frameworks. Rational choice models that allow for error would thus lead us to expect new nuclear power decision- makers to adopt nuclear coercion at most half the time, although models that would not allow for error would lead us to expect no nuclear coercion. Moreover, these models would lead us to expect decision-makers to be attentive to sufficient logic or history to estimate the true power of the strategy. 44 2.3. Nuclear Proliferation and Nuclear Coercion in a Psychological Learning Model I have outlined a rational choice nuclear learning model that predicts new nuclear powers to either not practice nuclear coercion or practice it randomly. In this section I develop a psychological nuclear learning model that predicts that (weak revisionist) new nuclear powers will systematically pursue nuclear coercion until they experience sufficient fear of nuclear escalation. Psychological theories start from the robust empirical finding that human rationality is inherently bounded by innate, insuperable limitations on information processing and the capacity of working memory. Attention is finite, systematically searching for information is costly and important information to relevant issues may be ambiguous or inaccessible and misinterpreted if not missed. Decades of cognitive and social psychological research have demonstrated that people commonly resort to inferential shortcuts to proceed economically in the face of these inherent limits on information processing. Only these cognitive shortcuts with their efficiency gain allow people to make important decisions in conditions often characterised by high uncertainty. They are natural ways in which people interpret information quickly and efficiently to categorize and respond to their surroundings and simplify otherwise complex and difficult more normatively appropriate inferential procedures. These heuristics aid the 45 speed and utility, but not always the accuracy, of human inference, and cause significant, systematic and lasting biases in human inference.69 Rational choice models can incorporate issues of imperfect information, but they cannot accommodate selective attention and the systematically unbalanced, distorted processing of information that psychologists have identified. Inattention to crucial data points and the incorrect diagnostic value of other observations leads to preferences that are substantially different from normatively appropriate expected utility models. The inferences that people commonly draw from the uncertain information they have leads decision making to diverge significantly from rational choice stipulations. These biases are widespread and do not cancel each other out in aggregate. Much of the early research found these biases in college undergraduates but more recent studies have shown that foreign policy decision-makers in high stake political contests suffer from similar biases when they perceive information, assess risks and make decisions. Since Robert Jervis’s seminal contribution on the importance of psychological variables to international security, scholars have shown how psychological mechanisms bias decision-maker attributions of adversary resolve, learning from history, and risk propensities when they have strong incentives to behave rationally.70 Yuen Foong Khong, for example, showed 69 See, for example, Susan T. Fiske and Shelley E. Taylor, Social Cognition: From Brains to Culture, 2008. New York: McGraw-Hill; Daniel Kahneman, Paul Slovic and Amos Tversky, Judgment under Uncertainty: Heuristics and Biases, 1982. Cambridge, U.K.: Cambridge University Press; R. E. Nisbett and L. D. Ross, Human inference: Strategies and shortcomings of social judgment, 1980. New Jersey: Prentice Hall. 70 See Robert Jervis, Perception and Misperception in International Politics, 1976. New Jersey: Princeton University Press; Dan Reiter, Crucible of Beliefs: Learning, Alliances and World Wars, 1996. Ithaca: Cornell University Press; Yuen Foong Khong, Analogies at War: Korea, Munich, Dien Bien Phu and the Vietnam Decision of 1965, 1992. New Jersey: Princeton University Press; Jonathan Mercer, Reputation and International Politics, 1996. Ithaca: Cornell University Press; Rose McDermott, Risk Taking in International Politics: Prospect Theory in American Foreign Policy, 1998. Ann Arbor: University of Michigan Press, and Philip E. Tetlock, Expert Political 46 that the availability heuristic systematically biased Lyndon Johnson and his associates’ assessment of the options for continuing the Vietnam war in 1965. Ignoring such psychological biases or assuming them away misses a critical variable that explains variation in foreign policy, crisis behaviour or uses of force. Nobody has yet assessed the impact of psychological biases on the conflict propensity of new nuclear powers. Uncertainty about possible coercive leverage and reputation for resolve characterise the immediate aftermath of nuclear proliferation and should generate a propensity for biased information processing and learning. In this section I outline how the availability heuristic makes weak revisionist new nuclear power decision-makers conflict prone; the experience of fear of imminent nuclear war makes them adopt more pessimistic risk estimates about the effectiveness of nuclear coercion and the probability of it causing nuclear escalation and less conflict prone in their nuclear and foreign policies. The availability heuristic inclines people to assess the likelihood of an event or outcome based on its ease of accessibility or mental construction. Rather than directing sufficient attention to current, past or plausible data points to estimate probabilities that generate causal inferences that are close to optimal outcomes as rational choice models predict, psychological research has shown that cognitively accessible data absorbs too much attention and causes misinterpretation or neglect of other important but less mentally accessible information. Recently and frequently activated data is more cognitively accessible; it comes to mind more easily than inactivated ideas. Such data profoundly influences the encoding of external stimuli into internal cognitive representations, and guides interpretation and mental representation of relevant, moderate Judgement: How Good Is It? How Can We Know?, 2006. New Jersey: Princeton University Press. 47 and ambiguous stimuli.71 Often something comes to mind quickly because its frequency is high. But people’s wants, needs and expectations can increase or decrease the availability of some classes of phenomena or events without altering their overall number. This biases probability judgements. As Tversky and Kahneman explain: A person is said to employ the availability heuristic whenever he estimates frequency of probability by the ease with which instances or associations could be brought to mind. The availability heuristic...uses strength of association as a basis for the judgement of frequency. Availability is an ecologically valid cue for the judgement of frequency because, in general, frequent events are easier to recall or imagine than infrequent ones. However, availability is also affected by various factors which are unrelated to actual frequency. If the availability heuristic is applied, then such facts will affect the perceived likelihood of classes and the subjective probability of events. Consequently, the availability heuristic leads to systematic biases.72 The original availability experiment was one designed to test for the frequency of occurrences. Subjects were asked to judge the probability that a particular consonant, specifically the letter R or K, is more likely to be found in the first or third position of a word. The majority of subjects guessed the letters to be found more frequently in the first position, although the opposite was the case for all of the letters used in the experiment. Individuals thus made their assessments of frequency based on how easily they could construct examples of each case. Words that begin with these consonants are more cognitively available than words in which the consonant appears in the third position. Tversky and Kahneman concluded that “classes whose instances are easy to construct or 71 Fiske and Taylor, Social Cognition, p. 74 72 Kahneman and Tversky, Judgement under Uncertainty, p. 18 48 imagine will be perceived as more frequent than classes of the same size whose instances are less available.”73 Two considerations are particularly important in determining the content of what might be available to an individual decision maker.74 Firstly, a key aspect of availability is the relative potential for retrieval of examples from memory. Individuals make frequency estimates based on the ease with which they can recall examples of the subject under consideration. Recent events are more easily recalled than more distant ones. Events experienced firsthand are more cognitively available than those not so experienced. Similarly, such events are more accessible than the universe of logically plausible non-occurrences. Overlearned and familiar information is more easily recalled than novel information. In all of these cases, inferential problems arise whenever these determinants of availability are imperfectly correlated with the actual frequency of events. Secondly, availability is influenced by illusory correlations. When using the heuristic, people tend to believe that causes have resemblances with their effects.75 Large effects, for instance, tend to be attributed to large causes. In the early tests for this bias, skilled clinicians were asked whether they could see any association between patients’ responses to a Rorschach test and certain then clinical diagnoses.76 The subjects saw 73 Op. Cit., p.166 74 For a similar discussion, see Rose McDermott, “Arms Control and the First Reagan Administration: Belief-Systems and Policy Choices,” Journal of Cold War Studies, Volume 4, Number 4, Fall 2002, pp. 29-59 75 Fiske and Taylor, Social Cognition, p.135; P. Slovic, B. Fischhoff and S. Lichtenstein, “Cognitive processes and societal risk taking,” in J. S. Carroll and J. W. Payne (eds.), Cognition and social behaviour, 1976, pp. 165-184. New Jersey: Erlbaum; L. J. Chapman, “Illusory correlation in observational report,” Journal of Verbal learning and Verbal Behaviour, 6 1967 pp.151-55. 76 L. J. Chapman, “Illusory correlation in observational report,” Journal of Verbal learning and Verbal Behaviour, 6 1967 pp.151-55; L. J. Chapman and J. P. Chapman, “Illusory Correlation as 49 multiple associations that did not in fact occur: between eyes and paranoia; and genitals and homosexuality. Moreover, they failed to see associations that did in fact occur, such as between half-human, half-animal figures and homosexuality. These clinicians saw associations that they believed to be true and failed to see those that existed but were not as available to them as established stereotypes. Illusory correlations thus exert an impact on availability even when they might be false. Attribution heuristics cause similar effects to illusory correlation biases. Rational choice learning models assume that decision-makers consider enough data, counterfactuals and alternative explanations to make near optimal causal and descriptive attributions by systematically assessing the most likely cause of an outcome. But a major finding from social psychology is the self-serving attribution bias. People systematically exhibit a tendency to take credit for success and underestimate responsibility for failure. Success is attributed to one’s own actions and the role of chance or context is neglected. Failure is attributed to other’s actions, external variables or bad luck.77 These two mechanisms of availability suggest that decision-makers in weak revisionist new nuclear powers will be much more dangerous than rationalist assumptions suggest. Rational choice models assume that subjective probabilities of effective nuclear an Obstacle to the Use of Valid Psychodiagnostic Signs,” Journal of Abnormal Psychology, Vol. 74, No. 3 June 1969 pp. 271-80 77 The experimental evidence that people take credit for success is stronger than evidence that they deny responsibility for failure. See D. Miller and M. Ross, “Self-serving biases in the attribution of causality: Fact or fiction?,” Psychological Bulletin 82 1975 pp. 213-225; R. Arkin, H. Cooper and T. Kolditz, “A statistical review of the literature concerning the self-serving attribution bias in interpersonal influence situations,” Journal of Personality, 48 1980 pp. 435- 448; B. Mullen and C. A. Riordan, “Self-serving attributions for performance in naturalistic settings; A meta-analytic review,” Journal of Applied Social Psychology 18 1988 pp. 3-22. For an application to international politics, see Mercer, Reputation and International Politics. 50 coercion should be driven by a systematic analysis of all diagnostic information that should generate a subjective probability close to the objective probability. The psychological nuclear learning model predicts that decision-makers in weak revisionist new nuclear powers will make two dangerous inferential leaps. Firstly, they will not learn from history. Their own recently developed nuclear weapons will be more cognitively available than other similar historical cases of unsuccessful or unattempted nuclear coercion. One might argue that the complexity associated with determining what successful nuclear coercion is and historical contingency makes such inattention to history rational. But there have been only a few weak dissatisfied new nuclear powers throughout history, so rational choice models should lead us to expect foreign policy decision-makers to be especially attentive to these cases to form an estimate of the probability of successful nuclear coercion. Psychological research however suggests that decision-makers will direct little to no attention to them because they do not learn from foreign wars.78 Secondly, they will believe that nuclear weapons’ enormous destructive power must translate into similar coercive power. Like the skilled clinicians, they will see a correlation that does not exist and miss real correlations.79 Desirable outcomes such as adversary irresolution or failure will be attributed to the new nuclear power; specifically, nuclear coercion or the latent coercive power of nuclear weapons. The role of other variables and chance in causing the desired outcome will be neglected. 78 Jervis, Perception and Misperception, pp. 241-2. 79 It is unclear that the relevant comparison for coercive leverage will be the destructive capacity of the weapon. Another possibility is that elites believe that their coercive leverage is as weak as the small number of times nuclear weapons have been used in wars. But most people tend to use the former for comparison, and the availability heuristic may also explain the selection of this for comparison. 51 Rational learning models such as those of Powell, Slantchev and Crescenzi would predict decision-makers to be attentive to enough history or logic to accurately estimate the low efficacy of nuclear coercion. The availability heuristic however suggests that decision-makers in weak revisionist new nuclear powers will learn that nuclear coercion is a powerful strategy. Moreover, they will learn that the probability of nuclear coercion causing inadvertent nuclear escalation is low. While status quo nuclear power decision- makers may also learn these lessons but not be emboldened to challenge the status quo, decision-makers in weak revisionist new nuclear powers that cannot achieve their goals through internal or external balancing will be highly likely to learn that nuclear coercion might achieve their regional objectives.80 But nuclear compellence is harder than deterrence. All else equal, a threat to respond to a challenge with nuclear escalation is more credible than a threat to start a nuclear war if one’s demands are not met. Moreover, if the challenger slightly changes 80 Elites in status quo new nuclear powers are also likely to be subject to these biases but would be less likely to pursue nuclear coercion because of their lower motivation to do so. Wishful thinking further biases causal inference and cause underestimation of a policy’s costs, overestimation of the probability of its success and a neglect of trade-offs between it and other policies. Structural adaption and rational choice models might expect elites in weak revisionist states that cannot achieve their objectives through internal or external balancing to estimate the average efficacy of nuclear coercion, give up their objectives and settle for what they can realistically achieve. The development of nuclear weapons – and a nuclear coercion strategy that is unlikely to realise revisionist objectives – should not, in these models, interfere with this adjustment. But psychological research suggests that political and military elites avoid painful value trade-offs like this between accepting an intolerable status quo and pursuing a dangerous coercive policy that probably would not succeed. Forfeiting long-held desires is psychologically painful; it would be much more bearable to persist in believing that revisionist objectives could somehow be achieved through recently developed nuclear weapons. According to findings on wishful thinking biases, people facing such undesirable options often ignore available information about the defects of the least undesirable policy, and may develop rationalisations for them that cause neglect of potentially unsafe features. See Irving L. Janis and Leon Mann, Decision Making: A Psychological Analysis of Conflict, Choice and Commitment, 1977. New York: Free Press, pp. 91-95; Robert Jervis, “Kargil, deterrence and international relations theory,” in Peter Lavoy (ed.) Asymmetric Warfare in South Asia: The Causes and Consequences of the Kargil Conflict,” 2009, pp. 377-397. New York: Cambridge University Press, pp. 385-6 52 the status quo and makes demands for more changes, there will usually be a way to restore it without causing nuclear escalation. But such coercion may cause a war or crisis that may give rise to fear of nuclear escalation. While the availability heuristic makes revisionist new nuclear powers conflict prone, I argue that the experience of fear causes “fundamental learning” about nuclear coercion that makes them adopt pessimistic risk estimates in their nuclear and foreign policies.81 A classic study from the 1980s found that people tend to make judgements that are compatible with their current mood, even when the subject matter is unrelated to the content of that mood. Positive emotions trigger optimist risk assessments and negative emotions trigger pessimistic risk assessments.82 The simple act of feeling an emotion influences behaviour in other areas logically unrelated to that which generated the emotion. For example, an account of a fatal stabbing did not increase the subject’s frequency estimate of a closely related risk, such as homicide, more than the estimates of unrelated risks such as natural hazards. Subsequent studies replicated the same phenomenon.83 In a series of influential articles, psychologist Jennifer Lerner and her 81 Fundamental learning is differentiated from peripheral learning because it changes central beliefs. See Bennettt, Condemned to Repetition, p. 84. Fear is defined as dread of impending disaster that tends to cause intense urges to defend oneself by escaping a situation. It is differentiated from anxiety, which is an ineffable and unpleasant feeling of foreboding. See Arne Ohman, “Fear and Anxiety: Overlaps and Dissociations,” in Michael Lewis, Jeannette M. Haviland-Jones and Lisa Feldman Barrett (eds.), Handbook of Emotions, Third Edition, 2008. New York, Guilford Press, pp. 709-729, p. 710. 82 Eric J. Johnson and Amos Tversky, “Affect, Generalization, and the Perception of Risk,” Journal of Personality and Social Psychology, 1983, Vol. 45, No. 1, 20-31. See also Craig A. Smith and Phoebe C. Ellsworth, “Patterns of cognitive appraisal in emotion,” Journal of Personality and Social Psychology, 48 (1985) 813-838 83 See, for example, David DeSteno, Richard E. Petty, Duane T. Wegener and Derek D. Rucker, “Beyond valence in the perception of likelihood: The role of emotion specificity,” Journal of Personality and Social Psychology, Vol. 78 (3), March 2000, 397-416; Dacher Keltner, Phoebe C. Ellsworth and Kari Edwards, “Beyond Simple Pessimism: Effects of Sadness and Anger on Social Perception,” Journal of Personality and Social Psychology (1993) Vol. 64 No. 5 740-752; 53 colleagues showed that experiencing fear systematically causes pessimistic risk estimates in the domain that caused the fear and logically unrelated domains.84 Earlier research tended to assume that emotions arise from, but do not cause, specific cognitive appraisals.85 But Lerner and Keltner found that emotions also elicit specific cognitive appraisals that, although tailored to help the individual respond to the event that evoked the emotion, persist beyond the eliciting situation and become an implicit perceptual lens for interpreting subsequent situations. Different emotions have different effects. Anger produces optimism about future outcomes and risk acceptance that stems from low levels of uncertainty. But fear and distress generate anxiety and stem from high appraisals of uncertainty and the need for situational control. This leads to pessimistic judgements of the future as individuals discern the source of threat and how it can be contained. Experimental studies found that fear triggered in one situation evokes more pessimistic risk estimates in unrelated situations. In a series of laboratory experiments, Lerner and Keltner found that fearful individuals consistently made relatively pessimistic judgements and choices regardless of whether fear was naturally occurring or experimentally induced, judgement targets were relevant to the subject, probabilities of different outcomes were known or not, and estimates were expressed publically or Larissa Z. Tiedens and Susan Linton, (2001) “Judgement under emotional certainty and uncertainty: The effects of specific emotions on information processing,” Journal of Personality and Social Psychology, 81, 973-988; Fiske and Taylor, Social Cognition, p. 135 84 Jennifer Lerner and Dacher Keltner, “Beyond valence: Toward a model of emotion-specific influences on judgment and choice,” Cognition and Emotion, (2000) 14, 473-493; Lerner and Keltner, “Fear, Anger and Risk,” Journal of Personality and Social Psychology, Vol. 81 No. 1 2001 pp. 146-159; Jennifer Lerner, Roxana M. Gonzalez, Deborah A. Small and Baruch Fischhoff, “Effects of fear and anger on perceived risks of terrorism: A national field experiment,” Psychological Science, (2003) 14, 144-150; Baruch Fischhoff, Roxana M. Gonzalez, Jennifer S. Lerner, and Deborah A. Small, (2005) “Evolving judgments of terror risks: Foresight, hindsight, and emotion,” Journal of Experimental Psychology: Applied, 11, 124-139. 85 Smith and Ellsworth, Patterns of cognitive appraisal in emotion. 54 privately. Moreover, fear elicits specific cognitive appraisals that persist beyond the eliciting situation and become an implicit perceptual lens for interpreting subsequent situations.86 In one experiment, participants predicted the likelihood that specific positive and negative events would occur in their own life compared with the lives of relevant peers. The differences observed for fear (and anger) influenced not only choices with known probabilities and little personal consequence but also judgements with unknown probabilities and real personal consequence.87 In a subsequent study, respondents exposed to a fear inducing manipulation assigned, on average, a 7.8% higher probability to five negative consequences of terror than did respondents exposed to an anger- inducing manipulation. These emotions carried over to probability judgements for routine risks having no obvious connection to the terrorism related manipulations.88 The experimental research on fear suggests that foreign policy decision-makers that experience fear of nuclear escalation or war will learn two lessons that explain the effect of experience with nuclear weapons. Firstly, they will learn, from their own nuclear crisis rather than history, that the probability of nuclear coercion causing inadvertent escalation is high. Secondly, they will learn that nuclear coercion is not an effective strategy for realising their revisionist objectives. They will adopt such pessimistic risk estimates not only in nuclear diplomacy but in unrelated foreign policies; this should cause restraint in a nuclear crisis (if restraint is not 100% certain to occur), safer nuclear diplomacy and 86 Literature on fear conditioning suggests that sufficient levels of fear can cause emotional recalibrations that can last for the remainder of one’s life. See Arne Ohman and Susan Mineka, “Fears, Phobias and Preparedness: Toward an Evolved Module of Fear and Fear Learning,” Psychological Review, Vol. 108 No. 3 2001 pp. 483-522 87 Lerner and Keltner, Fear, Anger and Risk, p.150 88 Lerner et al, Effects of Fear and Anger, p.147 55 less dangerous foreign policies (in any setting). Decision-makers are still likely to remain inattentive to historical cases of new nuclear power coercion, but they will learn through fear experienced in their own crisis or war that nuclear coercion is dangerous. The contrasts with the predictions of rational choice models are strong. Rational choice models would lead us to expect a reassessment of the efficacy of nuclear coercion after the first failed attempt, but psychological research suggests that much more is required to cause such a reassessment.89 When fear of nuclear escalation is not experienced, actors should not exhibit these pacifying effects and continue to adopt dangerous nuclear crisis behaviour, nuclear diplomacy and foreign policies. Fear of nuclear escalation may cause not only nuclear restraint but, at minimum, caution in other foreign policies such as conventional postures, alliances and arms control.90 Importantly, fear of imminent nuclear war might not be caused by the first coercive attempt; several attempts may occur before the challenger experiences fear, adopts pessimistic risk estimates and tacitly accepts the status quo. Moreover, while new nuclear powers are more likely to cause a crisis than their experienced nuclear rival, there is nothing in the experimental research that suggests that new nuclear power decision- makers are more likely to experience fear in a nuclear crisis than their experienced nuclear adversaries. But it is the new nuclear power’s experience of fear that explains the 89 One can argue that whether one failed attempt should rationally cause a reassessment of the strategy depends on the strength of prior beliefs. But these beliefs will also be biased through the availability heuristic. 90 When policy-makers change their minds, they tend to abandon beliefs about appropriate tactics before giving up on an entire strategy and abandon strategies before questioning fundamental assumptions about other states and the international system. Jervis, Perception and Misperception, pp. 275-6; Philip E. Tetlock, “Social Psychology and World Politics,” in Daniel T. Gilbert et al, The Handbook of Social Psychology, Fourth Edition, 1998, pp.868-914. New York: McGraw-Hill,p. 880 56 effect of experience with nuclear weapons. It is not clear that the experienced nuclear power must also experience fear to explain the effect. While the status quo experienced nuclear power may be responsible for the weak new nuclear power’s revisionism, the latter’s challenges make enduring rivalries so conflict prone after they develop nuclear weapons.91 TABLE 1: PSYCHOLOGICAL AND RATIONAL CHOICE MODEL EXPECTATIONS OF WEAK REVISIONIST NEW NUCLEAR POWERS Rational Choice Expectation Psychological Bias Expected Psychological Process and Outcome STAGE 1 learn from history; estimate low efficacy of nuclear coercion; do not practice it Availability underlearn from history, overlearn from personal experience; overestimate efficacy of nuclear coercion and underestimate probability that it will cause inadvertent nuclear escalation; practice it most of the time STAGE 2 Fear underlearn from history, overlearn from personal experience; estimate low average efficacy of nuclear coercion and high probability that nuclear coercion will escalate to nuclear war; do not practice it. Adopt pessimistic risk estimates in nuclear diplomacy and unrelated foreign policies 91 For an argument that both states in an enduring rivalry need to learn to explain such change, see Jeffrey W. Knopf, “The Importance of International Learning,” Review of International Studies (2003) 29, pp. 185-207. 57 FIGURE 1: THE PSYCHOLOGICAL NUCLEAR LEARNING MODEL The effects of the availability heuristic and fear imply that nuclear proliferation is dangerous when decision-makers learn that it is safe and safe when they learn that it is dangerous. Initially, their learning of biased lessons from a limited reading of history is destabilising. But the fear subsequently experienced through their own war or crisis teaches them of the dangers of nuclear coercion and explains the effect of experience with nuclear weapons 2.4. Dissertation Summary Chapter 1 showed that the most important consequence of nuclear proliferation is conventional conflict. Yet neither key contributions to the literature nor other plausible Weak, Revisionist new nuclear power decision-maker learns that nuclear coercion is effective and unlikely to cause nuclear escalation, practices nuclear coercion Experience fear of imminent nuclear war in crisis or war? NO (arrow to above box) Experience fear of imminent nuclear war in crisis or war? YES Learn that nuclear coercion is likely to cause inadvertent escalation and that it is a generally weak coercive strategy. Adopt pessimistic risk estimates in nuclear diplomacy and foreign policy; pursue revisionist objectives through less dangerous strategies 58 hypotheses explain the effect of experience with nuclear weapons on weak revisionist states. This chapter has introduced learning – defined as the change of beliefs that result from the observation and interpretation of experience – as a plausible causal mechanism. It has outlined two empirically falsifiable rational choice and psychological models that explain the effect of experience with nuclear weapons. The subsequent two chapters test these models. I now proceed to justify my case selection and methodological strategies. I have argued in the previous chapter that the effect of experience with nuclear weapons is not puzzling for new nuclear powers that fought weaker non-nuclear adversaries before 1960 such as the United States and Britain. The universe of cases relevant to the puzzle of the effect of experience with nuclear weapons is the eight other nuclear powers. Yet it is unclear that the learning processes in the two models would occur in new nuclear powers that are not conventionally weaker than their adversaries and territorially revisionist. The point of nuclear coercion is, after all, to change the status quo; weak revisionist new nuclear powers would be most likely to practice it. I thus make new nuclear powers that are weak and revisionist a boundary condition of the theories developed and tested here. Such a criteria only applies to the Soviet Union, Pakistan and possibly China. But these states most closely resemble Iran and North Korea. Moreover, further research should address whether the nuclear learning models explain the conflict propensity of status quo new nuclear powers such as France, Israel and South Africa or great powers before 1960 such as the United States and Britain. While this dissertation focuses on whether the learning models apply to revisionist states that are prepared to challenge their conventionally stronger principal adversary over important parts of the status quo, future research should address whether they also explain states that were 59 content with the status quo before proliferation. It is plausible, for example, that status quo new nuclear powers are emboldened to fight more aggressively or bargain harder in conflicts that would have occurred absent nuclear proliferation. The universe of cases relevant to the models developed and tested in this dissertation is thus the Soviet Union, China, Pakistan and perhaps North Korea (if the latter can be classified as a nuclear power). TABLE 2: UNIVERSE OF CASES OF NEW NUCLEAR POWERS CONVENTIONALLY STRONGER THAN PRINCIPAL ADVERSARY CONVENTIONALLY WEAKER THAN PRINCIPAL ADVERSARY STATUS QUO OBJECTIVES RE PRINCIPAL ADVERSARY UNITED STATES 1945? INDIA 1990 ISRAEL 1967 SOUTH AFRICA 1982 BRITAIN 1952 FRANCE 1960 CHINA 1964? UKRAINE, BELARUS, KAZAKHSTAN 1991 REVISIONIST OBJECTIVES RE PRINCIPAL ADVERSARY UNITED STATES 1945?92 SOVIET UNION 1958 CHINA 1964? PAKISTAN 1989 IRAN? NORTH KOREA? Limited data on the perceptions, beliefs and attention spans of political and military decision-makers in these cases however offers a poor environment for testing psychological explanations against rational choice alternatives. Without extensive archival or interview data that addresses these variables definitive conclusions will be elusive. To address this problem, I draw heavily on the Soviet case because it offers 92 The British conflicts in Malaysia and Kenya might warrant this country to be placed in this category too. 60 several independent sources on the beliefs of Nikita Khrushchev that offer a plausibility probe of the psychological learning hypothesis. To show that conclusions from this case travel beyond the Cold War, I also examine the Pakistani case through extensive data gathered by Paul Kapur and others and an original interview with former President Pervez Musharraf. Much of the contribution of this dissertation is the theoretical predictions of the behaviour of weak revisionist new nuclear powers as well as the empirical findings, and these will be signposted throughout.93 I utilise the qualitative research method of process tracing to derive causal process observations and establish causal mechanisms in correlations that were established by earlier quantitative research. Causal process observations and process tracing are natural methodological choices to examine processes of decision-maker learning that, based on the definition specified above, would not be amenable to quantitative statistical analysis due to data availability and measurement validity problems. It is very hard to identify learning without detailed process observations. Analysing the effect of learning with statistical techniques requires establishing an empirical indicator for learning that is equally valid in all observations. If this cannot be achieved, the statistical model will not capture the true effect of learning, produce biased estimates of the effect and perhaps not test the learning hypothesis. For example, it is unclear what learning the uncertainty 93 Limited data availability prohibit me from testing the nuclear learning hypotheses on the case of China in this dissertation. Some recent archival evidence however suggests that Mao Zedong did not harbour revisionist territorial ambitions because he was too preoccupied with the Cultural Revolution to concomitantly fight an external war. Mao’s ambush of Soviet patrols on the Sino- Soviet border at Zhenbao river in 1969 seems to have been caused by his desire to undermine his domestic critics and consolidate his sweeping reforms rather than coerce the Soviet Union. See Lyle J. Goldstein, “Return to Zhenbao Island: Who Started Shooting and Why it Matters,” The China Quarterly 2001 No. 168 pp.985-997 and Yang Kuisong, “The Sino-Soviet Border Clash of 1969: from Zhenbao Island to Sino-American rappochement,” Cold War History, No. 1 August 2000, pp. 21-51 61 variables in Gartzke’s model refers to. Given the inherently complex nature of belief change and the challenges of developing a metric that codes all beliefs, testing genuinely learning based hypotheses through statistical models is an elusive goal. One response to this challenge would be to define learning as policy change so that all policy changes are similar units of learning. But this choice would prohibit assessing when – and how - learning actually causes policy change because learning is conflated with policy change. Thus Russell Leng conflated policy change with learning and offered the modest conclusion on the Indo-Pakistani rivalry that “whatever learning occurred remained within the bounds of realpolitik assumptions about inter-rivalry relations” and “most often has reinforced behaviour that has encouraged the recurrence of crises and wars.”94 To understand what lessons cause what policy changes we need to analyse the perception, reasoning and information processing of political decision-makers. Causal process observations derived from archival data and secondary historical accounts are ideally suited to this task. Process tracing is more likely to establish how learning moderates the conflict propensity of new nuclear powers. The obvious limitation of this strategy is not only that the findings may not apply to other weak revisionist new nuclear powers or other stronger nuclear powers, but that the data available to test learning models will not allow definitive conclusions in the selected cases. But using a poor indicator for learning in statistical models would offer no firmer conclusions. In an environment where so little has been established about the causes of variation in the conflict propensity of new nuclear powers, it is better to pursue 94 Russell J. Leng, “Realpolitik and learning in the India-Pakistan rivalry,” in T.V. Paul, (ed.) The India-Pakistan Conflict: An Enduring Rivalry, 2005. New York: Cambridge University Press, pp. 103-127, p.121, 103. See also Leng, Bargaining and Learning in Recurring Crises: The Soviet- American, Egyptian-Israeli and Indo-Pakistani Rivalries, 2000. Ann Arbor: University of Michigan Press. 62 a “narrow and deep” case selection strategy that addresses a few cases through a detailed examination of the historical record than a “wide and shallow” analysis using indicators that are unlikely to capture the cognitive dynamics of interest. Such a strategy might suggest an empirical indicator that could be incorporated into statistical models in future research. In the next section I summarise the remainder of this dissertation. In Chapter 3 I utilise recently declassified archival and autobiographical data of Nikita Khrushchev to show how Soviet coercive behaviour over Berlin in the Cold War is better explained by the psychological nuclear learning model. That Soviet nuclear coercion persisted after its first failure until Khrushchev experienced fear of nuclear escalation is better explained by the psychological explanation than the rational choice alternative. To show that the validity of the psychological model is not confined to the Cold War, Chapter 4 uses the secondary literature and recently obtained elite interviews, including with former President Pervez Musharraf, to test the two learning models in the Pakistani case. I show that territorial contiguity and terrorist groups in South Asia differentiated the Pakistani strategy of nuclear coercion but not the psychological logic of the consequences of nuclear weapons. While India’s Cold Start doctrine and continued terrorist attacks continue to pose challenges to nuclear stability, the fear of nuclear escalation that Pervez Musharraf experienced at the second peak of the ten month crisis suggests that the psychological explanation may also explain South Asia. Chapters 3 and 4 show that the psychological learning model explains the Soviet and Pakistani cases better than rational choice learning models. But decision-makers do not stay in power forever. Decision-makers could enact institutional changes that commit their successors to their own policies, but different leaders who have been influenced by 63 different lessons should pursue different policies. A satisfactory test of the hypotheses requires data on the beliefs of Khrushchev’s and Musharraf’s successors. If the psychological learning model explains the effect of experience with nuclear weapons, Leonid Brezhnev should have learned the same lessons as Khrushchev, and Musharraf’s civilian and military successors should have learned similar lessons to him. Although Brezhnev was intimately involved in Soviet policy at the time of the Missile crisis, data on his beliefs is not available. In the Pakistani case, Prime Minister Asif Zardari and Army General Ashfaq Kayani declined to be interviewed. To address this problem, in Chapter 5 I select a case that allows me to address the origins of foreign policy beliefs. Do leaders learn from their predecessors or from their own cognitively accessible experiences? Data from the relevant cases s lacking so we need to turn to other cases from which we could at least begin to make some inferences about the more limited question of the extent to which decision makers of nuclear states might learn about crises from their predecessors. There is much data on Johnson’s beliefs over time that allows an assessment of their origins. I show that Johnson’s behaviour was influenced much more by his cognitively accessible Munich experiences than any less cognitively accessible lessons from Kennedy. His earlier lessons from Munich influenced him to escalate the Vietnam war; he was relatively uninvolved in the Missile crisis and seems to not have learned anything from it. This provides further support for the claim that formative experiences strongly influence policy choices. Moreover, it suggests that if Brezhnev and Zardari experienced fear of nuclear escalation in their respective crises, the psychological explanation may explain the pacifying effect of experience with nuclear weapons in the Soviet and Pakistani cases. 64 Chapter 6 concludes the dissertation. Firstly, I argue that although the term ‘revisionist’ has important ambiguities, these do not undermine the psychological learning model’s explanation of the Soviet and Pakistani cases. Further research should examine whether nuclear weapons emboldened relatively status quo states such as Israel, South Africa and India to challenge the status quo in a manner that they would not have done before. Secondly, I apply the psychological learning model to briefly predict the consequences of North Korea and Iran developing nuclear weapons capable of targeting their chief adversaries. It is unclear whether the relevant adversary is the United States or South Korea, Japan, Israel or Saudi Arabia. The psychological model suggests that if these states are dissatisfied with major aspects of the status quo, their development of nuclear weapons capable of targeting their chief adversary will cause a crisis or war in the short term. But the relevant decision-makers will inevitably learn through fear about the limits of nuclear coercion in a crisis or war and pursue less dangerous nuclear and foreign policies to realise their objectives. I conclude with implications of the psychological learning model for other political, military and commercial strategic environments. 65 Chapter 3: Soviet Nuclear Missiles, 1958-1962 3.1. Soviet Nuclear Missiles: 1958-1962 [It would] take some time until the present psychological state of distrust and fear were overcome. What was necessary were some events or series of events which might change the psychological climate. President Dwight D. Eisenhower, July 195595 The Cuban Missile crisis was the best thing to happen to us since the second World War. It helped us avoid further confrontation with the Soviets; it resolved the Berlin issue; and it established new basic understandings about U.S.-Soviet interaction. Thomas Schelling, 198996 Nikita Khrushchev’s three unsuccessful coercive attempts to force the United States out of West Berlin after developing strategic nuclear missiles constituted the most dangerous years of the Cold War. Rational choice learning models would not predict him to have maintained his policy of nuclear coercion throughout the five year period after he developed the capability to target the United States with nuclear missiles, and would not predict Soviet nuclear diplomacy and foreign policy to have fundamentally changed after the Cuban Missile crisis. The psychological nuclear learning model leads us to expect Khrushchev to have exaggerated the efficacy of nuclear coercion until he learned through fear of imminent nuclear war that the probability of nuclear escalation resulting from 95 Foreign Relations of the United States (hereafter referred to as FRUS), 1955-57, Vol. 5, pp.412- 413 96 James G. Blight and David A. Welch, On the Brink: Americans and Soviets Reexamine the Cuban Missile Crisis. New York: Hill and Wang, 1989, p. 104 66 nuclear coercion is high. It also predicts that the fear that Khrushchev experienced during the crisis would have caused pessimistic risk estimates in nuclear diplomacy and other foreign policies. I firstly provide some historical background to establish that Khrushchev’s Soviet Union was economically weaker than the United States and lacked conventional military superiority to realise revisionist territorial ambitions. Much of the evidence to test the central claims of this dissertation are drawn from this case, so it is important to establish that the Soviet case is similar in important ways to states like Iran and North Korea. I also show that while the Soviet Union tested atomic and hydrogen bombs in 1949 and 1953, Khrushchev only developed strategic nuclear missiles capable of targeting the United States in 1959. Readers familiar with these details may wish to skip to section 3.3 where I show through case studies of the two Berlin crises and Cuban Missile crisis that the psychological nuclear learning model explains the Soviet case better than the rational choice alternative. In the conclusion, I subject the psychological learning model to a further test. I argue that John Kennedy was unlikely to have learned about the dangers of nuclear coercion from the Cuban Missile crisis because he did not experience fear during the Missile Crisis. 3.2. Was the Soviet Union a Weak Revisionist State? One might argue that the consequences of nuclear weapon development by a state as powerful as the Soviet Union would provide little insight about the consequences of nuclear proliferation for weak revisionist states in the twenty-first century. Did 67 Khrushchev’s Soviet Union have characteristics that many today attribute to Iran and North Korea? Specifically, was the United State’s Cold War strategic competitor, fifteen years after the end of the second World War, economically inferior to it and unable to realise territorially revisionist goals through conventional military superiority? These are critical questions, because if Iranian decision-makers are presently dissatisfied with American influence in the Persian Gulf, understanding how nuclear weapons will influence their conflict propensity requires selecting cases of revisionist states. Findings from the Soviet case will only travel to Iran and North Korea (if these states are revisionist) if Khrushchev was also prepared to accept significant risk to reduce American regional influence. In this section, I show that Khrushchev’s Soviet Union fits the model of a weak revisionist state. Moreover, I show that while the Soviet Union tested atomic and hydrogen bombs in 1949 and 1953, Khrushchev only acquired the capability to target the United States with nuclear bombs in 1959. I firstly address the Soviet economy, secondly address Soviet grand strategy and conclude with Soviet nuclear missile development. Khrushchev’s Soviet Union was economically weaker than the United States on almost all significant indicators, and faced daunting external and internal economic challenges in the 1950s. Most fundamentally, the Soviet command economy was failing to provide basic subsistence requirements for many Soviet workers. These economic shortcomings undermined the capability of the Soviet government to militarily and economically compete with the United States, and undermined the promise and legitimacy of Soviet communism. These shortcomings were not only internal problems; the Soviet controlled command economies in Eastern Europe, a bulwark against 68 American influence there, faced similar challenges. Eastern Europe, at least since Khrushchev’s 1953 anti-Stalin turn, was showing signs of widespread discontent with Soviet tutelage; American influence in Western Europe however was much desired in London, Paris and Bonn as a bulwark against Soviet expansion. The cheap aid of the Marshall Plan and the NATO alliance, largely a response to Stalin's failed West Berlin grab in 1948, brought close economic integration and security cooperation between major Western European powers and the United States. If the Soviet Union was economically weaker than the United States, what about the military balance? Did Khrushchev’s Soviet Union possess superior military capabilities to realise his Berlin ambitions? Many analysts at the time believed that the Soviet Union was conventionally superior to NATO forces. One could argue that in total conventional war the Soviet Union would have prevailed over NATO forces, but such a war may have escalated to nuclear war where winners and losers would be hard to determine. A more relevant indicator of superior conventional military power may be the ability to make and sustain revisions to the status quo. The most likely form of such Soviet revisionism was usually considered to be a surprise attack with standing forces because a major mobilisation would eliminate the benefits of surprise. Several influential articles have argued that NATO forces would have been able to deter or withstand such a Soviet/Warsaw Pact surprise attack. Matthew Evangelista focussed on the 1947-48 period and found that “Soviet troops were not capable of executing the kind of invasion feared in the West during the 69 late 1940s.”97 Alain Enthoven and Wayne Smith addressed the 1960s and concluded that “NATO and the Warsaw Pact were roughly equal in terms of soldiers, guns, vehicles, infantrymen, and the like. In many respects, we were ‘superior:’ in some respects, they were.”98 Barry Posen addressed the 1980s and concluded that “NATO forces are fully competitive with the Warsaw Pact in Central Europe,” and “would probably thwart a conventional attack.”99 John Mearsheimer also addressed the 1980s and argued that “there is ample reason for thinking the Soviets cannot overrun Germany with conventional forces....NATO has a good chance of defeating a conventional Warsaw Pact attack in Central Europe.” Moreover, Mearsheimer argued that “the common image of overwhelming Pact materiel superiority, created by misleading “bean counts” of unrepresentative classes of equipment, is simply incorrect.”100 While Khrushchev could not look to internal balancing to realise his revisionist ambitions, external balancing possibilities were similarly bleak. Khrushchev had few allies to turn to in the communist bloc. Mao Zedong had long received more benefits from the Soviet Union than he had offered in return, but by 1960 Mao’s growing radicalism had undermined the Sino-Soviet alliance. Khrushchev lacked the capability to achieve any strategic goals through internal or external balancing; he could not afford to militarily compete with the United States 97 Matthew Evangelista, “Stalin’s Postwar Army Reappraised,” International Security, Vol. 7, No. 3, (Winter 1982-83), pp. 110-138, p. 111. He elaborates that (p.119) when one considers that “Soviet divisional manpower has historically numbered 50 to 60 percent of Western divisional manpower, and that Soviet divisions have far fewer support troops, the picture looks different...an image of rough parity emerges.” 98 Alain C. Enthoven and K. Wayne Smith, How Much is Enough? Shaping the Defense Program, 1961-1969. Harper and Row, 1971, p. 142 99 Barry R. Posen, “Is NATO Decisively Outnumbered?,” in International Security Vol. 12 No. 4 (Spring 1988), pp. 186-202, p. 189, 200. See also Barry R. Posen, “Measuring the European Conventional Balance: Coping with Complexity in Threat Assessment,” in International Security, Vol. 9, No. 3 (Winter 1984-1985), pp. 47-88. 100 John J. Mearsheimer, “Numbers, Strategy and the European Balance,” in International Security, Vol. 12, No. 4, (Spring 1988) pp. 174-185, pp. 184, 180-181. 70 and was out of allies. The subsequent paragraph provides some examples of Soviet economic weaknesses. The orthodox wisdom in the 1950s was that Soviet progress in economic growth, space exploration and missile development was high. But the economy was weak and undermined Soviet military power. At a Presidium meeting in June 1955, Khrushchev noted that the “economy is falling into decay.”101 Khrushchev confided to an American in 1959 that the Soviet Union was like “a hungry person who had just awakened and wanted to eat. The Soviet Union was not trying now to develop the production of any sophisticated consumer goods; it was simply trying to satisfy the basic needs.”102 Similar problems threatened Eastern Europe. Khrushchev and the majority of Presidium members concluded that the Polish and Hungarian governments had brought workers and students into the streets in 1956 by failing to provide them with adequate standards of living. They believed that they faced a similar threat at home; Khrushchev suggested an emergency investment in Soviet residential housing construction and more funds targeted at raising the material well-being of Soviet workers.103 Any fiscal qualms amongst Presidium members were outweighed by the conviction that Soviet standards of living, especially for the industrial working class, had to be improved to forestall political trouble. The Soviet leader later recalled that ''we were scared, really scared. We were afraid the thaw might unleash a flood, which we wouldn't be able to control and which could drown 101KDMP, June 20 1955. KDMP refers to the Kremlin Decision Making Project archival notes from Khrushcev’s Presidium made available at the Miller Center; http://millercenter.org/scripps/archive/kremlin 102 Memorandum of Conversation, Lodge, “Train Trip from Los Angeles to San Francisco,” September 21, 1959, RG 59, Office of Soviet Union Affairs, Subject Files, 1957-1963, Box 4, National Archives and Records Administration, College Park, Maryland. 103 KDMP, November 20, 1956 71 us.''104 By 1961 Khrushchev's seven-year plan to improve domestic conditions in the Soviet Union was failing. On his inspection tours around the USSR, Khrushchev saw mounting evidence that the Soviet state was not even meeting the most basic of its citizen’s requirements. Soviet workers were experiencing chronic shortages in milk, meat and eggs. At one point, a senior party official told Khrushchev that “if one were to remove Moscow and Leningrad from our responsibility, then we could feed ourselves.”105 Housing starts had fallen so far short that Khrushchev decreed that all new apartment buildings would be built higher than the standard five stories to try to catch up with demand. But he had to rescind the order when he received news that the USSR lacked the raw materials to build elevators for taller apartment buildings. Party bosses were soon forced to decide how they would fudge the numbers for all aspects of the seven-year plan so that the gap between promise and reality would not appear as huge as it was.106 In the spring of 1962 Khrushchev was forced to admit the failure of his programs to increase croplands and farm output. His forced announcement of 20-30 percent increases in state controlled prices of basic foods triggered protests and military reprisals that killed twenty three protesters and led to hundreds being arrested and over a dozen executed.107 In February 1963 Khrushchev reminded his audience near Moscow that their standard of living would not rise as rapidly as promised. Meat production in 1962 was only 40% of 104 Anne Appelbaum, Gulag: A History, 2003. New York: Doubleday, p.514 105 Alexsandr Fursenko and Timothy Naftali, Khrushchev's Cold War, 2007. New York: Norton. p.343 106 Harrison E. Salisbury, A New Russia?, 1962. New York: Harper and Row, p.120; KDMP, March 23, 1961 107 Vladislav M. Zubok and Constantine Pleshakov, Inside the Kremlin's Cold War: From Stalin to Khrushchev, 1996. Cambridge, MA: Harvard University Press, pp. 263-265 72 what he had called for then in 1957.108 The British Ambassador to Moscow wrote in 1959 that “Khrushchev's chief ambition was not the extension of an empire, but to be able in his lifetime to declare communism in the Soviet Union with its corollary of overtaking the United States in gross production.”109 In February 1964 the CIA advised Lyndon Johnson that the Soviet Union had “little choice but to find ways to contain the arms race and reduce its burden on the Soviet economy.”110 In the next section I show how these weaknesses underlay Soviet grand strategy. Khrushchev’s desperation for Soviet economic consolidation fuelled his desire for disarmament. He wanted to reduce the military costs of the Cold War so that more money could flow into the civilian economy. Accommodation with the United States and avoidance of war were means to these ends. Soviet economic problems did not hold much promise for Soviet victory in the East-West struggle that Khrushchev was convinced history would vindicate him on. His strategy to address them seems to have involved a push for a strategic military detente with the United States so that competition could be confined to economic and ideological grounds. He wanted to avoid war and military confrontation at a time when he knew he would lose and consolidate the Soviet economy and its international appeal such that military competition, if it was to occur, would be kinder to him.111 Khrushchev wanted to lift the enormous economic burdens of defence budgets. 108 Harry Schwartz, “Soviet Wipes out Promise to Raise Meet Allowances,” New York Times, March 4, 1963; Lincoln Broomfield, Walter C. Clemens, Jr., and Franklyn Griffiths, Khrushchev and the Arms Race: Soviet Interests in Arms Control and Disarmament 1954-1962, 1966. Cambridge: Massachusetts, M.I.T. Press, pp. 230-231 109 Sir Patrick Reilly to Macmillan, August 25, 1958, Prem 11/5115, British National Archives, Kew, United Kingdom, cited in Fursenko and Naftali, Khrushchev’s Cold War 110 F.R.U.S., 1964-1968, Volume XIV, The Soviet Union, p.21 111 James G. Richter, Khrushchev's Double Bind: International Pressures and Domestic Coalitional Politics, 1994. Baltimore: Johns Hopkins University Press, p.115 73 He had long feared that the race to keep up with Washington in military firepower would bankrupt his country.112 He enacted reforms that dramatically cut the Soviet armed forces and his power projection ability to devote scarce resources to the fragile Soviet economy.113 In 1960 he turned down a proposal for the creation of a vast system of intercontinental ballistic missiles (ICBMs) that would offer parity with the United States on the grounds that “missiles are not cucumbers.”114 The Soviet leader wanted a nuclear test ban treaty and nuclear disarmament but these had to follow a settlement to the German problem. Neither Eisenhower nor Kennedy would sign a nuclear arms control treaty if they believed that a crisis in Berlin, where Khrushchev had vastly superior conventional forces, was imminent. I next address the place of West Berlin in Soviet grand strategy. Since Hitler's defeat, American, British and French troops in West Germany and isolated West Berlin faced Soviet troops in East Germany. Stalin, Truman, Khrushchev and Eisenhower all believed that this status quo was untenable. It was not only expensive. The German people would inevitably demand unification or at least more autonomy; occupying Germany might provoke the future German aggression that it was designed to prevent. But Stalin and Khrushchev worried that a unified Germany or even rearmed West Germany would threaten the Soviet Union. Memories of Hitler's 1941 betrayal were vivid, and both Soviet leaders worried about Bonn acquiring partial control over 112 Sergei Khrushchev, Nikita Khrushchev and the Creation of a Superpower, 2000, trans. by Shirley Benson. University Park: Pennsylvania State University Press, p.281 113 S. Khrushchev, Nikita Khrushchev and the Creation of a Superpower, p. Viii, 89. This was of course based on the belief that “the basis of our defence is strategic missiles” and that “planes, tanks, and artillery would gradually become obsolete and disappear from the scene.” op. Cit., pp. 216-217, 674-675 114 Matthew Evangelista, “Why Keep Such an Army?: Khrushchev's Troop Reductions,” Working Paper 19, Cold War International History Project (hereafter referred to as CWIHP) Bulletin, December 1997, p. 39 74 American nuclear weapons or their own nuclear arsenal. The major goal of Stalin's post- war foreign policy had been to prevent the integration of West Germany into the Western bloc. Stalin had pushed for a reunification of Germany on the condition that it was neutral and demilitarised. He would have preferred the untenable status quo to a West Germany integrated into the Western bloc, and perhaps pushed for a neutral and demilitarised Germany because he knew that neither Truman nor Eisenhower could accept this. Although he had created a Communist regime in his occupation zone, the protection of East Germany was not the organizing principle of Stalin's German strategy. Stalin’s attempt to evict the Western powers from West Berlin pushed West Germany further into Washington's arms. In the fall of 1954 the Western powers drafted a series of agreements that would bring West Germany into the NATO alliance. Once this occurred in 1955, Khrushchev's concern was however not the Stalinist worry that German Chancellor Adenauer would use the NATO alliance to prepare an attack on the Soviet Union. Khrushchev only began to seriously worry about German aggression in 1958 when Bonn openly sought nuclear weapons. Khrushchev was forced to accept an independent West Germany allied with NATO. But this further committed him to ensuring the territorial integrity, economic growth and source of influence in Eastern Europe, East Germany, remained secure. But American access rights to West Berlin, guaranteed since Stalin, Truman and Churchill met at Potsdam in 1945, allowed American air and rail transports to violate East German sovereignty on a daily basis. West Berlin and its threat to East Germany thus lay at the root of Nikita Khrushchev's revisionist motivations.115 115 Hope M. Harrison, Driving the Soviets up the Wall: Soviet-East German Relations, 1953-1961, 2003. New Jersey: Princeton University Press, p.5; Vladislav Zubok, A Failed Empire: The Soviet 75 The post-war settlement permitted the Western powers to freely move troops in and out of West Berlin, and as of November 1958 there were already eleven thousand NATO soldiers in the city. Moreover, West Berlin constituted a black hole that sucked skilled workers such as doctors, engineers and scientists out of East Germany to the West. The perpetuation of the status quo would thus further weaken East Germany. Every month since the June 1953 East German uprising an average of fifteen thousand people fled East Germany through West Berlin. Almost three hundred thousand of the states’ most skilled citizens had escaped by 1955. West Berlin posed a threat to East Germany – and therefore the Soviet Union – in several ways. In addition to the ‘brain-drain’ problem, the approximately fifty thousand East Berliners working in West Berlin took advantage of the East's low living costs but took their labour to the West Berlin market, further drying up sources of East German economic dynamism. Furthermore, many West Berliners came to East Berlin to take advantage of cheaper state subsidised goods and services and consume them in West Berlin. This depleted the availability of goods and services for East Berliners and further increased their incentives to move west. All of this severely challenged the autonomy of East Germany; on the eve of the Vienna summit the East Germans had communicated to Moscow that their immediate objective was to close the sectoral border through which so many East Germans were fleeing.116 Khrushchev told visiting U.S. Senator Hubert Humphrey in 1958 that Berlin was a “bone in my throat,” and that he intended to “cut this knot.”117 Marc Trachtenberg has influentially argued that Khrushchev's risky foreign policy Union in the Cold War from Stalin to Gorbachev, 2007. Chapel Hill: University of North Carolina Press, p.109 116Harrison, Driving the Soviets up the Wall, pp. 170-171 117 Embassy in Moscow to U.S. Department of State, December 3, 1958, F.R.U.S. 1958-1960, Vol. 8, pp. 148-152 76 and the resulting Cold War crisis years were caused by Soviet fears of a West German nuclear bomb. In this account, the pressure on West Berlin was designed to force the United States to pressure its West German ally to abstain from acquiring the bomb.118 Eisenhower was indeed sending nuclear weapons to Europe not only for American use but to be eventually turned over to the allies, for their own use, in the event of war. The United States and other NATO members did not want an indigenous German nuclear program, and had to balance Germany’s security needs with most NATO members’ fears of a German bomb. However there are four reasons why Khrushchev's fears of a German bomb were important but not central to Soviet policy. Firstly, as Trachtenberg notes, if West German nuclear proliferation was as important to Khrushchev as he claimed, Khrushchev should have accepted Kennedy’s proposed settlement that committed to a non-nuclear West Germany in late 1961 before he deployed missiles to Cuba.119 Khrushchev still refused the agreement because it allowed Kennedy to maintain West Berlin securely tied to the West with Western military forces as guarantors of the cities odd status. Trachtenberg admitted that this is a 'major problem' for his argument, and conceded that “by the early 1960s additional factors must have come into play.”120 Yet this small note in the preface takes the wind out of the sails of much of his argument. Something other than West German nuclear ambitions – which must have been the economic and territorial challenges facing East Germany – influenced Khrushchev’s decision to reject Kennedy’s deal and, I shall argue, deploy Soviet nuclear missiles to Cuba. Moreover, unlike Eisenhower, Kennedy had never intended to let West 118 Marc Trachtenberg, A Constructed Peace: The Making of the European Settlement, 1945-1963, 1999. New Jersey, Princeton University Press, pp. 251-255, 282 119 op. cit., pp. 345-350, 380-388 120 op. cit., pp. ix-x 77 Germany acquire nuclear weapons.121 If Khrushchev was aware of this, Trachtenberg's argument does not fare well for the second Berlin crisis because Khrushchev would have known that Kennedy would not have permitted West Germany to get the bomb. German nuclear ambitions were relevant in the first crisis in 1958 but concerns for the consolidation and sovereignty of East Germany were great here too. Secondly, as Trachtenberg realises, Soviet concerns about a German bomb were closely associated with concerns for East German sovereignty. A West German bomb may have emboldened Bonn to challenge Soviet influence in East Germany. Many West Germans favoured unification; a nuclear West Germany or perhaps even a West Germany with access to the proposed NATO nuclear force would plausibly have been tempted to intervene in any Soviet attempts to crush future East German uprisings, and might take a generally more hostile line toward Soviet interests.122 The perpetuation of the fragile German Democratic Republic was always challenged by the superior economic development of West Germany, and Khrushchev may have believed that Bonn's acquisition of nuclear weapons would have accelerated these trends. A concern for the sovereignty of communist East Germany thus implied concern for West German nuclear aspirations. If Khrushchev was only concerned with a West German bomb, he should not have targeted his coercive attempts exclusively at West Berlin rather than West Germany itself. If his concern was East German sovereignty however, he should have targeted West Berlin. Khrushchev's concern for consolidating East Germany was much more important to him than containing West German nuclear ambitions. Bonn's nuclear ambitions were 121 op. cit., pp. 283-284 122 One could argue that these concerns are irrational because nuclear weapons do not usually allow coercive advantages, but Khrushchev believed that they offered him significant leverage in Berlin. It is therefore plausible that he worried that they would offer Adenauer similar power. 78 mentioned much less frequently in his interactions with the East Germans and Western powers than his concern for the threat that West Berlin posed to East Germany.123 In the last Presidium meeting on May 26, 1961, before Khrushchev left for his June 3 meeting with Kennedy in Vienna, he informed his colleagues that “we are not afraid of German aggression...the most dangerous is America. We cannot vouch for America. [It is] governed by different groups and sudden coincidental events...That's why certain forces could emerge and find a pretext to go to war against us.”124 Thirdly, Trachtenberg did not have access to the Soviet archival data that scholars such as Alexander Fursenko, Timothy Naftali and Hope Harrison did. Khrushchev made reference to the status of West Berlin and its threats to East Germany before all three coercion attempts, and only mentioned West German nuclear aspirations before the 1958- 59 crisis. Even here, West Berlin is emphasised before that of a potential West German bomb. Fourth, the Cold War crisis years should not have ended in 1962 if fears of a German bomb were central to Soviet behaviour. West Germany signed the limited test ban treaty but expressed similar desires for a bomb in the future, so something else must have caused Soviet restraint. Soviet strategy in the Khrushchev era was therefore motivated by generating a reduced Western presence in West Berlin that would consolidate the territorial integrity and economic development of East Germany. The latter was not recognised by the major 123 Harrison, Driving the Soviets up the Wall, pp. 114-115. Harrison elaborated that Ulbricht tried to push Khrushchev to raise the West German nuclear issue more frequently in public because he believed the German public would be much more supportive of a campaign against West German nuclear weapons than a campaign for a peace treaty with the Soviet Union. 124 op. Cit., p.356 fn 72. Robert Kennedy learned through his backchannel with Georgi Bolshakov that “Khrushchev still honestly believes the country is run by the Rockefellers, the Morgans and the big financiers on Wall Street.” Arthur Schlesinger Jr., Robert Kennedy and His Times, 1978. Boston: Houghton Mifflin, p.499 79 western powers, including West Germany, and the allied access routes connecting West Germany with West Berlin, guaranteed since the Potsdam agreements of 1945, allowed daily road, rail and air violations of East German sovereignty. East Germany's demise would perhaps throw doubt on Soviet commitment to other communist satellite states and undermine Soviet influence in its East European buffer with the West and, perhaps, the Soviet Union itself. Khrushchev desperately wanted to shore up his East German client through cutting off the flow of refugees to the West, curtailing Western violations of its sovereignty and gaining Western recognition of Soviet influence in East Germany. Khrushchev told his colleagues in January 1962, after the failed Berlin ultimatums and before the missile deployment to Cuba, that “it is better to have [problems with] Berlin and aggravation with the West rather than making concessions at the expense of the GDR's sovereignty.”125 He assured East German President Walter Ulbricht that “East German needs are our needs.”126 But American resolve to defend West Berlin from Soviet encroachments was viewed by the British, French and West Germans as a litmus test for the American resolve to defend Western Europe. Khrushchev's multiple proposals for reduced Western access rights, a neutral West Berlin defended by a temporary United Nations force or access to West Berlin but only through a terminus in East Germany all fell on deaf ears in Washington. If Khrushchev signed a peace treaty with East Germany, the Americans would have to get permission from the truculent East Germans to cross East German territory by air or land to visit their political island of West Berlin. Any concessions here by an American President during Khrushchev's era would have been electoral suicide. 125 KDMP, January 8, 1962 126 Douglas Selvage, “The End of the Berlin Crisis: New Evidence from the Polish and East German Archives,” CWIHP Bulletin 11, 1998, pp. 218-229, p.218 80 Khrushchev, deeply dissatisfied with the status quo in divided Germany, thus found himself out of options.127 He and his colleagues agreed with Defence Minister Rodion Malinovsky that “our inferior position was impossible to us.”128 Yet with “things still going poorly” all he could do was “grasp at straws.”129 I have argued that Khrushchev’s Soviet Union was militarily and economically weaker than the United States and territorially revisionist. Yet why was the Soviet Union not emboldened to further challenge the status quo after breaking the U.S. atomic monopoly in 1949? I now show that Khrushchev only acquired the means to target the United States with nuclear weapons ten years later in 1959. Since the missile age, new nuclear powers have acquired missiles capable of targeting their principal adversaries before they have developed nuclear weapons. But the Soviet Union first tested an atom bomb a decade before developing the capability to target the United States with it. The small Soviet bomber force was too vulnerable to NATO air defences. The United States had weapons stationed close to Russia in Europe from the early Cold War, but the Soviet Union lacked a similar nearby facility and developed the capability to target the United States with nuclear weapons a decade after developing them. Despite Soviet successes in testing an atomic bomb in 1949 and a hydrogen bomb in 1953, Moscow's claims to being a nuclear superpower on par with the United States were hollow. President Truman had bomber bases in Western Europe and could authorise over one hundred nuclear weapons to be dropped on Soviet targets by 1949. Since the 127 Constantine Pleshakov, “Studying Soviet Strategies and Decisionmaking in the Cold War Years,” in Odd Arne Westad (ed.) Reviewing the Cold War: Approaches, Interpretations, Theory, 2000. Frank Cass: London, pp. 232-241, p.233 128 Ernest R. May and Philip D. Zelikow, eds., The Kennedy Tapes: Inside the White House During the Cuban Missile Crisis, 2002. New York: W. W. Norton, p.422 129 Sergei Khrushchev, Khrushchev on Khrushchev: An Inside Account of the Man and his Era, 1990. Little, Brown and Company, p.21 81 Soviet Union, unlike the United States, could not deploy its bombers close to its adversary's borders, delivering nuclear weapons to United States territory required the development of intercontinental missile platforms. By 1955 Khrushchev had no means of using a nuclear device against an American city. His latest bomber, the M-4, had a combat radius of five thousand miles and could not reach either U.S. Coast from the nearest point of Soviet territory. The Soviet Navy in 1955 also lacked the capacity to deliver a nuclear strike against Washington. Stalin had promoted the development of submarines, but missile-launching craft and aircraft carriers were still years from production. Later aircraft, the 3M and Tu-95, entered service in 1957 and remained until 1959 the only Soviet means of reaching United States territory with nuclear weapons. However vulnerability to American air defences and jet interceptors limited the deployment of these aircraft.130 Khrushchev could only harm the United States through hitting one of its NATO allies. The Tu-16 Badger bomber finally made the Soviet nuclear threat credible in Europe in the late 1950s but these planes were also vulnerable to NATO's antiaircraft defences.131 Khrushchev's decision to limit long-range bomber production, coupled with the lack of long-range missiles, meant that the Soviet Union would be overwhelmed by United States and NATO nuclear forces with no comparable response in any war as late as 1958.132 The Kremlin had approved a plan in March 1955 to deploy R-5M medium range missiles – the first capable of targeting London and Paris with nuclear weapons – to East Germany so that NATO members could be targeted. Orthodox histories date the 130 See Steven J. Zaloga, The Kremlin's Nuclear Sword: The Rise and Fall of Russia's Strategic Nuclear Forces, 1945-2000. Washington: Smithsonian Institute Press, 2002, pp. 26-31; Pavel Podvig (ed.) Russian Strategic Nuclear Forces. Massachusetts: The MIT Press, 2001, pp. 1-6. 131 Fursenko and Naftali, Khrushchev's Cold War, pp. 39-40 132 op. cit, p.166, p.177 82 deployment to have occurred in 1956.133 New archival evidence shows that the deployment did not take place until 1958, and the nuclear warheads were not transferred to East Germany until April 1959.134 The first Soviet ICBM capable of targeting the United States, the R-7, had been used to put Sputnik into space in October 1957 but needed more engineering before it could be outfitted with a nuclear warhead and deployed. The first successful launch of the R-7 occurred in July 1959. The Soviet military stationed two R-7 missiles, each of which carried a three-megaton warhead, at two launch facilities; Khrushchev finally had the ability to launch nuclear missiles at the United States in the summer of 1959.135 The next sections assess the psychological and rational choice hypotheses in Khrushchev’s lessons from the Suez crisis, the two Berlin crises and the Cuban Missile crisis. TABLE 3: SOVIET DEPLOYED ICBM'S, 1955-1964 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 0 0 0 0 4 4 c14 c47 c109 c191 3.3. The Availability Heuristic 3.3.1 The Suez Crisis, 1956 Khrushchev placed great importance in spreading Soviet influence in the third world; he had made the Kremlin's relationship with Egyptian President Abdel Gamal Nasser the centrepiece of his strategy of building alliances in the third world and of staking a claim 133 Podvig, Russian Strategic Nuclear Forces, p.3; Zaloga, The Kremlin's Nuclear Sword, p.238 134 Matthias Uhl and Vladimir I. Ivkin, “'Operation Atom': The Soviet Union's Stationing of Nuclear Missiles in the German Democratic Republic, 1959,” CWIHP Bulletin, issue 12/13 (Winter-Spring 2001), pp. 299-307. 135 Podvig, Russian Strategic Nuclear Forces, p.181; Zaloga, The Kremlin's Nuclear Sword, p.232; S. Khrushchev, Nikita Khrushchev and the Creation of a Superpower, pp. 282-283 83 to influence in the resource rich Middle East. However Nasser's nationalisation of the Suez Canal, where two thirds of all the oil imported by Europe had to pass through, came at a bad time for the Soviet leader. Khrushchev lacked the power projection capabilities to defend Nasser from British and French aggression.136 After Nasser nationalised the Canal and faced intervention from Israel and then Britain and France, it took only four days for his military position to deteriorate. Coercing London and Paris to leave Suez with threats of nuclear retaliation when he did not have the ability fulfil his threat was his only option to compel a ceasefire. He would have known from Western newspapers that the West was very closely watching the development of his R5-M medium-range ballistic nuclear missile that was capable, if deployed in Eastern Europe, of reaching London and Paris. There were no R-5Ms on combat duty in November 1956 but Khrushchev must have believed that he had no choice but to exploit Western fears of Soviet missile capabilities. Khrushchev suggested sending a threatening letter to the British, French and Israelis that was broadcast on Moscow radio on November 5. There was no explicit threat to use nuclear weapons but the British, French and Americans would have feared this the most: In what position would Britain have found herself had it been attacked by more powerful states possessing all types of modern weapons of destruction? We are full of determination to crush the aggressor and re-establish peace in the East by using force. 137 136 Zachary Karabell, Parting the Desert: The Creation of the Suez Canal, 2003. New York: Knopf, p.269 137KDMP, November 5, 1956; Fursenko and Naftali, Khrushchev's Cold War, p.134 fn 50. 84 Khrushchev's bluff was neither credible nor effective. The British sued for a ceasefire for primarily economic reasons. The British intelligence chiefs had only one question to ask the CIA station chief in London at their tense meeting on November 6. When the American reported that Khrushchev did not have missiles that could have reached London, the British contingent apparently became much more relaxed.138 Khrushchev's threatening message was of less interest to Prime Minister Anthony Eden than the pending possibility of a financial crisis triggered by his adventurism in Egypt. In the eight days since the initial Israeli attack that started hostilities, foreign currency traders had become concerned about Britain's future oil supply and began dumping their holdings of British Pounds. Heavy international sales were now placing strong downward pressure on the value of the Pound Sterling, and Eden was hard pressed to buy enough of it to sustain its value. Eden had assumed that Eisenhower would begrudgingly support any military action against Nasser, and was confident that the U.S. Treasury would step in and buy enough of the Pound in currency markets. But Eisenhower was in no mood to support such adventurism when he was trying to bolster the appeal of capitalism in the Middle East, and allowed the Pound's value to further decline. The British were losing gold and dollars desperately needed to buy oil from Venezuela and elsewhere as the military operation continued. When Eden's intelligence staff learned of the emptiness of the Soviet nuclear threat, his finance minister was telling him that without American assistance he could not afford the war.139 The British Cabinet decided to seek a cease-fire in Egypt that morning despite having neither removed Nasser nor re-established control 138Fursenko and Naftali, Khrushchev's Cold War, p. 136 fn 55 139 Diane Kunz, The Economic Diplomacy of the Suez Crisis, 1991. Chapel Hill: University of North Carolina Press, pp. 116-152 85 of the Suez Canal. French support for withdrawal was also quickly forthcoming.140 But Khrushchev's coercive threat did not compel Eden; the British Prime Minister would likely have acted similarly if Khrushchev had not issued it. Rational choice learning models would lead us to expect Khrushchev to have systematically assessed the potential causes of the Anglo-Israeli retreat before concluding that his nuclear coercion was successful. He could have considered, for example, the pressure that the crisis was causing on the British Pound, or whether similar coercive threats by Stalin (in Berlin) or Eisenhower (in Korea) were effective. But there is no evidence in the extensive primary and secondary accounts that he did so. The psychological nuclear learning model would lead us to expect him to have overestimated the effect of his cognitively accessible coercive threat and neglected other logically possible but less accessible variables. It would predict him to have attributed the desirable outcome to his own behaviour, believed that highly destructive nuclear weapons had similarly large coercive power and concluded that nuclear weapons have similarly powerful coercive effects. The psychological model better explains Khrushchev’s inference. Khrushchev's son Sergei later noted that The cease-fire resulting from our message made an enormous impression on Father. He was extraordinarily proud of his victory. He had a way to influence the course of international affairs and would return to it more than once. He resorted to similar arguments when tensions arose around Syria, Iran, Jordan and Iraq. Father became convinced that the mere mention of nuclear-armed missiles had a powerful effect....The consequences of Suez were felt throughout the years to come. Its echo could even be heard in the...Cuban Missile Crisis. Over and over again he recalled the previous year's events in Suez, the phenomenal effect produced by 140 cable, Paris to DOS, November 6, 1956, 2:00AM, F.R.U.S., 1955-1957, Vol. 16 p.1012; p.1029, note 1 86 the mere mention of the R-5.141 He told Mao Zedong in July 1958 that “when we wrote letters to Eden and Mollet during the Suez events they immediately stopped their aggression.”142 He also explained to the Chinese Chairman that this was the key to saving the new progressive regime in Iraq.143 After Eisenhower deployed troops to the Middle East after the Iraqi monarchy had been deposed, Khrushchev indirectly communicated to him that “any United States or British move towards Iraq will mean war: Russia will ignore European bases and attack the United States directly.”144 Khrushchev informed his Presidium colleagues that for the second time in eighteen months fears of Soviet power had prevented the militarily and economically stronger United States from destroying one of his third world allies.145 His son Sergei wrote that he “celebrated a victory.”146 Rational choice learning models cannot explain why he systematically overestimated the power of his nuclear coercion. If Khrushchev's lessons from the Suez crisis caused his subsequent nuclear coercion in the Middle East, Berlin and Cuba, other Presidium members at the time who learned different lessons from that war should have had different policy preferences. Yet there is no evidence of this. Openly disagreeing with Khrushchev on this issue would have had 141 S. Khrushchev, Nikita Khrushchev and the Creation of a Superpower, pp. 211-212, 264 142 Vladislav M. Zubok, “The Mao-Khrushchev Conversations, 31 July – 3 August 1958 and 2 October 1959,” CWIHP Bulletin, 2001, Issue 12-13, pp.244-272, p. 14. 143 Mao Zedong also believed this. The East German ambassador in Beijing remarked at the end of 1961 that the Chinese had believed that “in the case of the Suez aggression, the Soviet ultimatum, which was taken seriously, scared the imperialists and forced them to stop their aggression.” See Harrison, Driving the Soviets Up the Wall, p.240 144 “Holeman,” Box 347, General Correspondence, Richard M. Nixon Pre-Presidential Papers, Vice-President, National Archives and Records Administration, Laguna Niguel, California 145 KDMP, August 4, 1958 146 S. Khrushchev, Nikita Khrushchev and the Creation of a Superpower, p.292 87 undesirable consequences for Presidium members, and few of them left published memoirs. It is possible that some Presidium members learned different lessons but did not voice them and rather supported Khrushchev as the Chairman consolidated his political power. Anastas Mikoyan, one of the few Presidium members who was prepared to openly disagree with Khrushchev, told the delegates to the 1957 Party Plenum that “everybody acknowledged that with this [nuclear coercion] we decided the fate of Egypt.”147 If Khrushchev’s colleagues influenced his own preferences and believed in the power of nuclear coercion when attention to other variables in this case would have suggested otherwise, the psychological nuclear learning model may explain the effect of experience with nuclear weapons better than rational choice learning models. 3.3.2. The First Berlin Crisis, 1958-59 Khrushchev learned of the efficacy of nuclear coercion from the Suez crisis. But these threats were bluffs, and as he became aware of his pending ability to target the United States, London and Paris with nuclear missiles, he was further emboldened to challenge the status quo in occupied Germany. He announced a radical change to his German policy in November 10 1958.148 In November 1958 there were already eleven thousand NATO soldiers in the city, and the air and rail transport corridors were used multiple times daily. He would give East German president Ulbricht what he wanted and sign a peace treaty with East Germany that would end effectively Western access rights to West Berlin and 147 cited in Zubok, A Failed Empire, p.130 148 Khrushchev address, November 10, 1958, U.S. Department of State, Documents on Germany, 1944-1985. Office of the Historian, Bureau of Public Affairs (Washington, D.C.: Government Printing Office, 1985), pp. 542-546 88 uproot the thirteen year status quo.149 This would guarantee the juridical equality of the two Germanys. However two days later Soviet guards stopped three U.S. Army trucks that were leaving West Berlin by the Babelsburg checkpoint. The commanding U.S. officer refused to permit inspection and the standoff was broken off eight hours later when a platoon of American tanks arrived at the scene.150 At the next Presidium meeting on November 20 Anastas Mikoyan rallied the group to derail Khrushchev's Berlin policy. The role of the Babelsburg incident is unclear but the Soviet leader's rash plan was voted down.151 The subsequent Soviet policy reflected a middle ground between Khrushchev and Mikoyan. Moscow promised to wait six months before unilaterally signing its peace treaty with East Germany. But given that the East German nuclear missile deployment would be operational within six months, Khrushchev was effectively forced to abandon an empty threat in favour of one that he could carry out. The new proposal also called for negotiations regarding the occupation regime in West Berlin to become a free city linked neither to NATO nor the Warsaw Pact. But Eisenhower knew that compellence is harder than deterrence; if he did nothing Khrushchev had to either take military action or back down.152 When Khrushchev visited him in Washington in September 1959, four months after the ultimatum expired, the Soviet leader announced that he would not insist on any deadline in the negotiations over Berlin. Rational choice learning models would not lead us to expect Khrushchev to 149 Hope Harrison, “Ulbricht and the Concrete Rose: New Archival Evidence on the Dynamics of Soviet-East German Relations and the Berlin Crisis,” CWIHP Working Paper 5, pp. 8-21 150 F.R.U.S., 1958-1960, Vol 8., p.69, n. 1. 151 Fursenko and Naftali, Khrushchev's Cold War, pp. 203-204. 152 Telephone Conversation, John Foster Dulles and Dwight D. Eisenhower, November 30 1958, F.R.U.S., 1958-1960, Vol. 8, pp. 134, 142-43 89 practice nuclear coercion. They might also lead us to expect him to have learned from the first Berlin crisis of the limits of nuclear coercion. The psychological learning model would lead us to expect him to continue to believe in the efficacy of nuclear coercion and underestimate the risks of nuclear escalation until he learned of its dangers through fear. There is some evidence that he considered the dangers of his strategy. He told Polish leader Wladislaw Gomulka that “there will be tensions of course...there will be a blockade. They will test to see our reaction...we will have to show a great deal of cold blood in this matter.”153 But, consistently with the psychological model, he continued to believe in the efficacy of nuclear coercion because he had not experienced fear. He informed his colleagues in February 1959 that if the Western powers “wish to force your way through, using all appropriate means...we will answer with all appropriate means.”154 3.3.3. The Second Berlin Crisis, 1961 Khrushchev worried that a second coercive threat to Eisenhower would hand Richard Nixon the Presidency. He refrained from the strategy throughout the election year of 1960 and issued his second coercive threat to John Kennedy early in his first term in 1961 at the Vienna summit. He offered an interim agreement on West Berlin with a fixed time limit, after which West Berlin would become a free city without any occupation forces or special NATO access routes. When Kennedy proved unwilling to accept this, he resorted to an ultimatum; if Kennedy did not concede after the ultimatum expired, he would sign a peace treaty with Ulbricht and turn over all air, rail and road access routes to the East 153 Douglas Selvage, “Khrushchev's November 1958 Berlin Ultimatum: New Evidence from the Polish Archives,” CWIHP Bulletin 11, 1998, pp.200-203, p.202 154 KDMP, February 21, 1959 90 German government. But Khrushchev’s ultimatum speeches had caused more people to flee East Germany and John Kennedy to increase his forces in West Berlin. By late June and early July, the flow of East German refugees through West Berlin was one hundred thousand; the largest it had been since October 1955.155 Khrushchev's desperation and Ulbricht's persistence caused the construction of the Berlin wall. The Berlin wall had temporarily stopped the refugee flow out of East Germany, but its sovereignty was still violated several times daily and Kennedy had substantially increased his troop presence in Berlin. Several weeks after the construction of the wall on September 24 Khrushchev informed Kennedy through his press secretary that “the storm in Berlin is over.”156 There is some evidence that Khrushchev considered the dangers of his nuclear strategy in Berlin. He instructed his Ambassador to East Germany to assure Ulbricht that “we are approaching this question seriously and if this drags us into war, there will be war.”157 When it became clear that there would be no Western efforts to remove the Berlin wall or otherwise penalise the Eastern bloc, Khrushchev remarked that “war might have broken out.”158 Khrushchev’s consideration of the dangers of nuclear coercion is consistent with rational learning models although his practice of the strategy is not. But his continued belief in the efficacy of nuclear coercion after two failed attempts, before he experienced fear, is better explained by the psychological learning model. While he criticised Stalin in his memoirs for “not gauging our possibilities realistically” and “not thinking it through properly” in the 1948 Berlin blockade, he made the same inferential 155 Strobe Talbott (ed.) Khrushchev Remembers. Boston: Little Brown, 1970, p.456; Honore M. Catudal, Kennedy and the Berlin Wall Crisis: A Case Study in U.S. Decisionmaking, 1980 Berlin: Berlin-Verlag, p.164 156 Pierre Salinger, With Kennedy, 1966. Garden City, N.Y.: Doubleday, p.191 157 Fursenko and Naftali, Khrushchev`s Cold War, p.377 fn 51 158 Jerold L. Schecter, (ed.) Khrushchev Remembers: The Glasnost Tapes, 1990. Boston: Little Brown, p.170 91 errors three times between 1958 and 1962 of not accurately estimating the power of nuclear coercion.159 In October 1961, he told Polish President Gomulka that “if we do not apply pressure, then we will have to give up on signing a peace treaty with the GDR.”160 Khrushchev believed that his second Berlin strategy left a “more than 95% probability that there will be no war.”161 He had no intention of altering his bottom line in the Berlin negotiations, and still insisted on no international corridors, no Western troops in West Berlin, East Berlin to remain part of East Germany and West Berlin to be a neutral, international city. Moreover, he proposed creating permanent international tension to achieve his Berlin objectives despite military and economic inferiority. At a January 8 1962 rump Presidium meeting Khrushchev concluded that his ultimatums had failed. But he elaborated that It is now too early to say that we will not win. We should still press on. I take the worst case: They won't agree. But it means agreeing right now that it will bring nothing. It's too early. So it's worthwhile playing this game.162 3.4. Fear 3.4.1. The Cuban Missile Crisis, 1962 After the two failed Berlin ultimatums and slow pace of Soviet ICBM development, Khrushchev asked his Defence Minister in April 1962 about the possibility of establishing his own off-shore missile base closer to the United States to reduce the 159 Cited in William Taubman, Khrushchev, 2003. New York: Norton, p.332 160 Selvage, The End of the Berlin Crisis, p.224 161 Fursenko and Naftali, Khrushchev's Cold War, p.340 fn 10; p.356 fn 72. 162 Fursenko and Naftali, Khrushchev's Cold War, p. 412 fn 10 92 American strategic missile advantage.163 He would have been aware of the American deployment, begun in 1961, of intermediate range Jupiter missiles in Italy and Turkey.164 Deploying nuclear missiles to Cuba was riskier than the earlier Berlin ultimatums because Khrushchev knew that Kennedy would not allow him to get away with such a dramatic revision to the status quo so close to American shores. The cautious Mikoyan warned that “with this approach we risk provoking an attack on them and losing everything,” but only gained a three day recess at the Presidium meeting.165 Khrushchev subsequently rallied support and the subsequent vote was unanimous with even Mikoyan recorded as speaking in favour of the Cuban missile plan.166 Operation Anadyr involved the dispatch of forty nuclear missiles divided into five nuclear missile regiments, three with medium range R-12s and two with intermediate range R-14s. Atlanta, Georgia, was in range of a medium-range missile launched from Cuba, whereas an intermediate range missile could hit the U.S. Strategic missile bases in the Midwest and Washington D.C. These missiles represented a major augmentation of Soviet strategic power.167 They were to remain under Soviet command, although Havana 163 Dmitri Volkogonov, Autopsy for an Empire: The Seven Leaders Who Built the Soviet Regime, 1998. New York: Free Press, p.236. Volkogonov was adviser to President Boris Yeltsin and had exclusive and extensive access to Soviet-era political and military records. The only record of this conversation comes from this source. 164 See Philip Nash, The Other Missiles of October: Eisenhower, Kennedy and the Jupiters, 1957- 1963, 1997. Chapel Hill: University of North Carolina Press. On Soviet ICBM development at this time, see S. Khrushchev, Khrushchev and the Making of a Superpower, pp. 468-482; Podvig (ed.), Russian Strategic Nuclear Forces, pp. 189-195. 165 Anastas Mikoyan, Tak Bylo [As It Was], 1994. Moscow: Vagrius, p.606 166 Fursenko and Naftali, Khrushchev's Cold War, p.436 fn 83 167 The strategic missiles were the centrepiece of an extensive Soviet military presence in Cuba. Four motorised regiments, two tank battalions each fitted with the newest Soviet tank, a MiG-21 fighter wing, some anti-aircraft gun batteries and twelve SA-2 surface-to-air missile detachments with 144 launchers would protect the nuclear missiles. Total deployment of Soviet forces would be 50,874. Additional nuclear striking power would come from forty-two Il-28 light bombers, each equipped with six nuclear bombs and which could reach Florida, and two cruise missile 93 would be given a joint defence agreement to assure them that Soviet military means would be used to defend their country. Khrushchev was aware of the powerful reach of American intelligence, but would only reveal the presence of the missiles after their deployment. One Soviet diplomat later recalled that “not a single specialist believed that this could be done secretly.”168 The Soviet delegation sent to Cuba to assess the feasibility of an offshore Soviet missile base however reported that the terrain and camouflage efforts would shield the missile sites from American surveillance.169 Most observers originally believed that the Cuban Missile Crisis was motivated by offensive Berlin concerns rather than defensive Cuban motivations.170 A second generation of scholarship argued that defensive motivations in Cuba were more important.171 A consideration of the different requirements of defensive (Cuba) and offensive (Berlin) strategies suggests that the strategy was aimed primarily at Berlin. regiments comprising eighty nuclear-tipped missiles positioned opposite likely U.S. Landing beaches. In addition to this impressive land and air component, the Soviet armed forces intended to establish a submarine base in Cuba that would simplify the logistics of maintaining patrols of the North American coastline. A massive flotilla that would establish a naval presence for the Soviet Union around the island would accompany these submarines. The motorised brigades would later receive twelve tactical nuclear missiles, and four foxtrot class diesel submarines were armed with nuclear missiles and a special target list so that “upon the signal from Moscow [they could] launch an attack on the most important American coastal targets.” Fursenko and Naftali, Khrushchev's Cold War, p. 440 fn 4; p.455 fn 50; p.461 fn 66 168 Taubman, Khrushchev, pp. 551-552 169 Anatoli I. Gribkov, “The View from Moscow and Havana,” in Operation ANADYR: U.S. And Soviet Generals Recount the Cuban Missile Crisis, ed. Alfred Friendly, Jr, 1994. Chicago: Edition Q, pp. 3-76, 15-16. 170 Sovietologist Llewellyn Thompson often said that the missile deployment was more about the Berlin problem than the Cuban one. May and Zelikow, Kennedy Tapes, pp. 37-41, 62, 67-68, 87 115, 159-160. Kennedy accurately predicted on October 9 that “as a result of the Soviet actions on Cuba, there was much less prospect of reaching agreement on Berlin” because “Khrushchev might try to force something.” Memorandum of Conversation, John F. Kennedy and Maurice Couve de Murville, October 9, 1962, F.R.U.S., Vol. 15, pp. 351-355 171 For the argument that Soviet behaviour was motivated by the defence of Cuba, see James G. Blight, Bruce J. Allyn and David A. Welch, Cuba on the Brink: Castro, The Missile Crisis, and the Soviet Collapse. New York, Pantheon, 1993; James G. Blight and David A. Welch, On the Brink: Americans and Soviets Reexamine the Cuban Missile Crisis. New York: Hill and Wang, 1989, pp. 238-247. 94 Khrushchev had always wanted to consolidate communist gains in Cuba from American intervention but would never have installed such a large missile base off Soviet soil to achieve this goal. The extensive military deployment that Khrushchev ordered was much more than what Castro needed to defend his regime. If consolidating Castro's gains in Cuba was all that mattered to Khrushchev, he could have deployed only conventional forces or deployed conventional forces with tactical nuclear missiles that would have provided more insurance against an invasion of the island without threatening the American mainland. He might also have signed a defence treaty with Cuba without deploying forces. Thus Fidel Castro remarked in 1992 that “if it was a matter of defending Cuba without creating an international problem, the presence of tactical weapons would not have created the same problem that the strategic weapons did.”172 Recent archival evidence from the Kremlin by Alexsandr Fursenko and Timothy Naftali strongly suggests that the Cuban missile deployment was more the product of offensive Berlin considerations than a perceived mortal threat to Fidel Castro's regime.173 When Khrushchev presented his Cuban strategy to the Presidium for approval, he emphasised to second tier officials the altruism of the scheme and claimed to be purely motivated by the defensive needs of Cuba. In his memoirs, he justified the deployment as consolidating communist gains in Latin America.174 But the Soviet leader admitted to his senior Presidium colleagues that “this will be an aggressive policy,” and that although appearing to be designed to deter American aggression, the missiles were designed to 172 James G. Blight, Bruce J. Allyn and David A. Welch, Cuba on the Brink: Castro, the Missile Crisis, and the Soviet Collapse, 1993. New York, Pantheon, p.250 173 Timothy Naftali, “Author's Response,” in Khrushchev's Cold War: The Inside Story of an American Adversary, H-Diplo Review, p.37. http://www.h- net.org/~diplo/roundtables/PDF/KhrushchevsColdWar-Roundtable.pdf accessed August 19, 2010 174 Khrushchev Remembers, p.493 95 give the Soviet Union coercive advantages in Berlin.175 Khrushchev used a July 1 Presidium meeting to unveil his Berlin agenda to his colleagues. He announced that he wanted to renew the push for a settlement on West Berlin. He planned to travel to the United States after the mid-term elections and the missiles were deployed to talk to Kennedy. He ordered a United Nations speech emphasising the German situation to be written for a presentation at the United Nations after he had publically announced his Cuban Missile base.176 The Western powers and the two Germanys would be forced to sign a peace treaty. Khrushchev informed his colleagues that Kennedy had to be forced to accept it even if it meant taking the “path of aggravating things.”177 This is unlikely to have occurred if the Cuban deployment was exclusively motivated by consolidating Castro’s regime. Khrushchev’s continued reliance on nuclear coercion is not well explained by rational choice models that predict him to pursue the strategy at most half of the time after developing nuclear missiles. The psychological learning model better explains his reliance on the strategy until he experienced fear of nuclear escalation in the final days of the crisis. The psychological model also explains his pessimistic risk estimates in other unrelated foreign policies after this date. As soon as John Kennedy announced that he would deliver a presidential address, Khrushchev worried that he would authorise an invasion of Cuba and began to learn about the dangers of nuclear escalation resulting from nuclear coercion through experiencing fear of nuclear escalation. He became convinced that he should take the 175 KDMP, May 21, 1962; Fursenko and Naftali, Khrushchev's Cold War, p.435 fn 80; Richard Ned Lebow and Janice Gross Stein, We All Lost the Cold War, 1994. New Jersey: Princeton University Press, p.48 176 KDMP, June 10, 1962 177 op. Cit. 96 initiative before events spiralled out of control, and admitted the fear that he had felt since hearing of Kennedy's speech; “we started out and then got afraid.” He admitted that “we didn't want to unleash a war,” and reasoned that “the tragedy is that they can attack, and we shall respond. This may end in a big war.”178 After learning of Kennedy’s blockade of Cuba, Khrushchev had sent Kennedy a letter on October 24 that informed him that “we will not simply be bystanders with regard to piratical acts by American ships on the high seas.”179 However by the end of the next day, before he received any response from Kennedy about the American response to this, fear of inadvertent escalation made him adopt pessimistic risk estimates with regards to the probability that his Cuba coercive strategy would cause inadvertent nuclear escalation. Khrushchev's son Sergei wrote that during the night [of the 24th] father had already begun to doubt the advisability of the evening's decision to continue straight ahead [and challenge the blockade]. That Thursday was when a change took place in his thinking.180 He decided that Soviet ships in the Atlantic bound for Cuba should temporarily go no further than the approaches to Cuba, and ordered the ships still in the Mediterranean to immediately return to their Black Sea ports.181 He wrote to Kennedy that if he would offer a pledge not to invade Cuba, Khrushchev would order the removal of the ballistic missiles.182 He subsequently sent another letter to Kennedy demanding both the 178 KDMP, October 22, 1962 179 Nikita S. Khrushchev to John F. Kennedy, October 24, 1962, F.R.U.S., 1961-1963, Vol. 6, p.170 180 S. Khrushchev, Nikita Khrushchev and the Creation of a Superpower, pp.572, 576 181 KDMP, October 25, 1962 182 KDMP, October 25, 1962. If Kennedy accepted this, Khrushchev could later maintain the lie that his missile base in Cuba was only designed for Castro's defence and could claim that he had 97 conditions in the earlier letter and the withdrawal of the Jupiter missiles in Turkey. But Khrushchev had decided to agree to the first proposed deal well before he received word that Kennedy had offered to secretly dismantle the Jupiters.183 Khrushchev admitted in his memoirs that the Cuban Missile crisis was “highly instructive.”184 One of the 1964 coup plotters alleged that the crisis had “frightened terribly the organizer of this very danger.”185 In early December 1962, Khrushchev informed peace activist Norman Cousins of the source of his desire for a deal a quick end to the crisis: The Chinese say I was scared. Of course I was scared. It would have been insane not to have been scared. I was frightened about what could happen to my country – or your country or all the other countries that would be devastated by a nuclear war. If being frightened meant that I helped avert such insanity then I'm glad I was frightened. One of the problems in the world today is that not enough people are sufficiently frightened by the danger of nuclear war.186 Rational choice models cannot accommodate Khrushchev’s systematic reliance on deterred an American attack on Cuba. He also vaguely referred to Berlin. “We have succeeded in some things and not in others.” Fursenko and Naftali, Khrushchev's Cold War, p.483 fn 67 183 KDMP, October 28, 1962. Many of Kennedy's advisers warned against giving Khrushchev the Jupiter concession, and the Turks were signalling that they did not want the Jupiters to be part of any trade. Kennedy believed that removing the Jupiter missiles were a good price to pay for reducing the risk of nuclear war but insisted that this be kept secret. 184 Nikita Khrushchev, Memoirs of Nikita Khrushchev, Volume 3: Statesman, 1953-1964, 2004. Edited by Sergei Khrushchev, Translated by George Shriver. University Park: Pennsylvania State University Press, p. 356 185 cited in Fursenko and Naftali, One Hell of a Gamble: Khrushchev, Castro and Kennedy, 1958- 1964, 1998. New York: Norton p.352 186 Norman Cousins, The Improbable Triumvirate: John F. Kennedy, Pope John, Nikita Khrushchev, 1972. New York, Norton, p.46, italics added. The evidence on Khrushchev’s fear of nuclear war supports Lebow and Stein’s claim that Kennedy’s compellence threat to Khrushchev to remove his missiles from Cuba was less effective than is commonly claimed. The Soviet withdrawal of nuclear missiles from Cuba shortly after Kennedy’s threat to do so encouraged the belief that nuclear crises could be won by using military threats to convey resolve. But Khrushchev’s fear of nuclear war may have caused him to remove the missiles and de-escalate the crisis absent Kennedy’s nuclear threat. See Lebow and Stein, We All Lost the Cold War, pp. 291-93 98 nuclear coercion before experiencing fear of nuclear war and his rejection of the strategy thereafter. The psychological learning model not only explains this but also accounts for his pessimistic risk estimates in other logically unrelated foreign policies. He told the American ambassador in November that “we may not love each other, but we have to live together and may even have to embrace each other if the world is to survive.”187 In December he sent letters to Kennedy explicitly dispelling the threat of a renewed Berlin ultimatum. At a December 3 Presidium meeting Khrushchev hinted that “for the foreseeable future the Soviet use of pressure tactics was dead.”188 His son Sergei later recalled that after the crisis he was “very interested in cooperating with Kennedy.”189 Fear of imminent nuclear war experienced during the crisis had made him adopt pessimistic risk estimates in other unrelated areas such as arms control, Berlin, Southeast Asia and the spread of communism to the third world. While his resistance to on-site inspections in the Soviet Union had prevented any progress on the nuclear test ban negotiations, he told the outgoing British ambassador that he would permit international inspectors to supervise the operation of automatic seismic stations on Soviet territory.190 On July 2 1963 Khrushchev announced that he was prepared to accept a partial test ban treaty, and within three weeks Washington, London and Moscow had drafted and signed the Limited Test Ban treaty. Earlier reversals in the third world in Syria, Egypt, Iraq, Cuba and the Congo had caused Khrushchev to threaten the United States with intervention if they supported anti-communist forces. But after experiencing fear during the Cuban Missile crisis he passively accepted the stalling of the Iraqi revolution in 187 Fursenko and Naftali, Khrushchev's Cold War, p.498 fn 27 188 Fursenko and Naftali, Khrushchev's Cold War, pp. 506-507. 189 Lebow and Stein, We All Lost the Cold War, p.145 fn 215 190 Glenn T. Seaborg, Kennedy, Khrushchev and the Test Ban, 1981. Berkeley, University of California Press, p. 177 99 February 1963 when General Qasim was murdered and an anti-communist Ba-athist regime came to power.191 Khrushchev also vowed to stop allowing Soviet aircraft in Laos to violate the spirit of the July 1962 neutrality agreement. Finally, as the prospects of the American sponsored South Vietnamese government looked bleak, Khrushchev refused to interfere on the grounds that Mao Zedong would try to force a confrontation between him and Kennedy.192 International politics had changed after the Cuban Missile crisis.193 Soviet Ambassador to Washington Anatoly Dobrynin concluded that the crisis “revealed the mortal danger of a direct armed confrontation of the two great powers...and agonizing realization of the disastrous consequences. It was this insight that made the settlement possible.”194 The evidence shows that the personal experience of fear of nuclear war influenced Khrushchev’s beliefs and subsequent Soviet policy in a way that the historical record of nuclear coercion could not. Prior to reaching the nuclear brink, he underlearned from history and overlearned from personal experience that the probability of nuclear war was low. After the crisis, he overlearned from his Cuban experiences of the dangers of nuclear 191 Fursenko and Naftali, Khrushchev’s Cold War, p.513 192 S. Khrushchev, Nikita Khrushchev and the Creation of a Superpower, p.695 193 It is impossible to know what Soviet foreign policy in the 1958-1963 period would have been if Malenkov or Molotov had sidelined Khrushchev and dominated the Kremlin. Yet both men would have faced severe economic shortages and deficits, a resurgent West Germany and temptations to pursue nuclear coercion after developing ICBMs. The years after the Soviet Union acquired such a nuclear missile capability were bound to exhibit higher Soviet risk acceptance and a higher danger of conventional and nuclear war. The irony of the Cuban missile crisis is that a less risky strategy may have caused a more moderate crisis, and this may have produced experiences less likely to cause fear, subsequent nuclear learning and risk aversion. If the Cuban Missile Crisis had rather been a third Berlin ultimatum, it may have led to further subsequent coercive strategies. Had Khrushchev survived the 1964 coup he may have attempted to coerce President Johnson out of West Berlin or further escalated the Vietnam war in response to Johnson's Americanisation policies. But the absence of risk acceptance or aggression in most other aspects of Soviet foreign policy for the two years after October 1962 suggest that he would have displayed similar risk aversion in Berlin and Vietnam to his immediate successors. 194Anatoly Dobrynin, In Confidence: Moscow's Ambassador to America's Six Cold War Presidents, 1962-1986, 1995. New York: Times Books, p.93 100 escalation resulting from nuclear coercion and the true power of nuclear coercion. Belief change caused pessimistic risk estimates that translated into policy preferences favouring confidence building measures and diplomacy rather than nuclear ultimatums. The influence of fear in other foreign policies is not easily accommodated by rational learning models. Contrary to the predictions of the rational choice learning model, Khrushchev systematically pursued nuclear coercion from the time he developed the nuclear capability to target the United States in 1959 to when he experienced fear. Of the four years between Khrushchev’s first Berlin ultimatum in November 1958 to his experience of fear of nuclear war in October 1962, Eisenhower or Kennedy were facing a soviet nuclear coercive threat for 46% of the time (22 months).195 Almost half of the other months were in the election year of 1960, and Khrushchev likely refrained from the strategy during this time not because he doubted its power but because he wanted to ensure that Kennedy defeated Richard Nixon. Khrushchev’s nuclear coercive threats were not random and were too frequent to be attributed to error. Rather, consistently with the psychological nuclear learning model, they occurred systematically until he experienced fear of nuclear escalation. Moreover, there is no evidence in the extensive primary and secondary literature and declassified Soviet archival sources that Khrushchev was attentive to historical cases of nuclear coercion throughout the crisis years. Nuclear proliferation was dangerous when Khrushchev believed it to be safe but became safe when he learned that it was dangerous. His consistent reliance on nuclear coercion to eject the United States from West Berlin until he experienced fear is better explained by the psychological nuclear learning model. The pervasive impact of Khrushchev’s fear on 195 This figure is calculated based on the dates noted throughout this chapter. Khrushchev’s three coercive attempts lasted from November 1958 through September 1959, June 1961 through September 1961 and April 1962 through October 1962. 101 subsequent Soviet nuclear and foreign policy is also better explained by his experience of fear than expected utility calculations. 3.5. Kennedy’s lessons from the Cuban Missile Crisis The evidence on John Kennedy’s lessons from the Cuban Missile Crisis, and whether he experienced fear of nuclear escalation, is less clear. Kennedy experienced what his brother referred to as “worry” at the same time in the crisis that Khrushchev experienced fear of nuclear escalation but he may not have experienced fear. His experiences at the Bay of Pigs taught him to not trust the preferences of his military advisers, and may have influenced his caution during the crisis. In this section I argue that because he may not have experienced fear, he was unlikely to have adopted pessimistic risk estimates in nuclear diplomacy or unrelated foreign policies. Some evidence suggests, however, that almost a year after the crisis he authorised investigations into establishing a new relationship with Fidel Castro. It is unclear if the crisis caused pessimistic risk estimates in unrelated foreign policies such as his relationship with Castro. Kennedy’s lessons from the crisis thus provide little evidence that decision-makers in both states in an enduring rivalry need to experience fear of nuclear escalation to explain the effect of experience with nuclear weapons. Robert Kennedy, the President’s brother and intimate adviser, referred to a moment when the President experienced worry but this may not have been fear. On the same day that Sergei Khrushchev claimed that a change took place in his father’s 102 thinking regarding his response to the American blockade of Cuba, Kennedy may have experienced fear of nuclear escalation. Robert McNamara informed him on October 24 that due to the inability of Soviet submarines to receive orders from Moscow, he had unilaterally changed the rules of engagement for U.S. destroyers trailing them. He had deviated from standard international practice in authorising practice depth charges that could hit but not destroy submarines to be used as warning notices and instructions for unidentified submarines to surface.196 But Soviet submarine captains could not know of this new procedure and that practice depth charges hitting their submarine would not destroy them, and Kennedy likely worried about the submarine captains’ immediate response to the charge. He responded to McNamara that “we don’t want to have the first thing we attack as a Soviet submarine.”197 Robert Kennedy referred to this as “the time of greatest worry by the President.” The President’s hand apparently “went up to his face and covered his mouth and he closed his fist. His eyes were tense, almost grey” and, Kennedy noted, “we just stared at each other across the table.”198 It is unclear if Kennedy’s worry was fear of nuclear escalation. Moreover, it does not seem to have caused pessimistic risk estimates in nuclear diplomacy or unrelated foreign policies, although it may have caused pessimistic risk estimates with regards to challenging Castro in the months after the crisis. His ‘moment of greatest worry’ did not influence his strategy to remove the Soviet missiles in Cuba. Although Kennedy had immediately committed to remove the 196 Ernest R. May and Philip D. Zelikow, (eds.) The Kennedy Tapes: Inside the White House During the Cuban Missile Crisis, 2001. New York: Norton, p. 229 197 Op. Cit., p. 230 198 Arthur Schlesinger, Jr., Robert Kennedy and His Times, New York: Random House, 1978, p.514 103 missiles from Cuba when he first learned about their presence, he was averse to ordering a military strike or invasion from the beginning of the crisis. His preferences for removing the missiles did not change after October 24. Similarly, his worry does not seem to have influenced his decision to remove the Jupiter missiles in Turkey. This decision occurred shortly after October 24, but, as May and Zelikow have shown, Kennedy had thought about doing so well before October 24 and likely only did so because Khrushchev stipulated it.199 Kennedy’s proclivity to offer concessions on Berlin was also not influenced by his worry about Soviet submarines. He seems to have never seriously considered giving in to Khrushchev’s Berlin demands before or after October 24. He was certainly worried about the hawkish preferences of his military advisers but learned of the dangers of his military advisers hawkish views from the failed Bay of Pigs invasion, long before the Cuban Missile Crisis. Perhaps Kennedy learned from the crisis of the importance of cooperating with the Soviet Union to avoid nuclear war. He proclaimed at his famous American University speech in June 1963 that “we have no more urgent task” than “peace because of the new face of war.” But he may have already adopted pessimistic risk estimates in his approach to Khrushchev before October 24. His statement to the Joint Chiefs on October 19, after he had learned of the Soviet missiles in Cuba but before he informed Khrushchev of the blockade and, exhibited the same caution of his American University speech eight months later: 199 May and Zelikow, Kennedy Tapes; Roger Hilsman, To Move a Nation: The Politics of Foreign Policy in the Administration of John F. Kennedy. New York: Doubleday, 1967: pp. 202- 203; Barton J. Bernstein, “Reconsidering the Missile Crisis: Dealing with the Problems of the American Jupiters in Turkey,” in James A. Nathan (ed.) The Cuban Missile Crisis Revisited. New York, St Martin’s, 1992, pp. 55-128. 104 The argument for the blockade was that what we want to do is avoid, if we can, nuclear war by escalation or imbalance. The Soviets increase; we use (force); they blockade Berlin. They blockade for military purposes. Then we take an initial action so that...We've got to have some degree of control. Those people (the Soviets) last night were so away from reality that there's no telling what the response would be.200 Robert Kennedy said that his brother believed “from the start that...there was always the chance of error, of mistake, miscalculation or misunderstanding.”201 Kennedy's National Security Adviser, McGeorge Bundy, claimed that Kennedy “may well have believed the key elements of his American University speech before 1963, but he did not feel that he could say them publicly.”202 One historian noted that there was not a sentence in this [American University] speech with which John Kennedy would have privately disagreed in 1960.”203 Some evidence however suggests that a year after the crisis Kennedy learned about the limits of his hard line policy toward Cuba. After the failed Bay of Pigs attempt in April 1961, Kennedy and his inner circle developed a hatred of Castro. Kennedy authorised a new set of covert operations, code named Mongoose, to undermine Castro’s government. A November 1961 government directive stated that the United States would “help the people of Cuba overthrow the Communist regime from within Cuba and institute a new government with which the United States can live at peace.” An update of 200 May and Zelikow, Kennedy Tapes, p.120 201Robert F. Kennedy, Thirteen Days: A Memoir of the Cuban Missile Crisis, 1969. New York, Norton pp. 126-127 202 Fursenko and Naftali, Khrushchev’s Cold War, p. 523 fn 114 203 Michael C. Beschloss, The Crisis Years: Kennedy and Khrushchev, 1960-1963. Harper Collins, 1991, p.600 105 March 1962 added “The U.S. will make use of indigenous resources, internal and external, but recognizes that final success will require decisive U.S. military intervention.” The President and his brother authorised military services prepared for possible intervention. 204 However this hard line changed after the Cuban Missile crisis. In early November 1963 Kennedy allowed an official to go to Havana for secret talks to improve relations with Washington if official involvement could be plausibly denied. Kennedy was apparently “very interested” in “thinking along more flexible lines” with Castro.205 This pessimistic risk estimates with regards to more risky policies toward Castro may have been caused by fear experienced at Cuba, but by early 1962 U.S. policy was already limited to isolating Cuba and limiting Cuban influence.206 By the end of April, Kennedy refused to set up training camps for a new landing attempt or to say a word encouraging hope that he would eventually approve an armed invasion.207 Kennedy’s motivations for this policy change are unclear though, and it is far from clear that they resulted from his Cuban Missile Crisis experiences the previous year. I have argued that the psychological nuclear learning model explains the effect of experience with nuclear weapons better than rational choice learning models in the Soviet case. Khrushchev systematically overestimated the power of nuclear coercion until he learned its true coercive power through fear in the Cuban Missile crisis. In the American case, Kennedy seems to have been strongly influenced by the Bay of Pigs. The influence 204 Lawrence Chang and Peter Kornbluh (eds.) The Cuban Missile Crisis, 1962: A National Security Archive Documents Reader (New York: New Press, 1992), Nos. 5 and 6 205 National Security Archive, Cuba Files, Kennedy and Castro: The Secret History. http://www.gwu.edu/~nsarchiv/NSAEBB/NSAEBB103/630304.pdf, Accessed September 8 2011 206 Chang and Kornbluh, Cuban Missile Crisis, p.351 207 Passavoy to Record, “Topics discussed during Meeting of Dr. Miro Cardona with the President,” 25 April 1962, “Cuba: Subjects, Miro Cardona, Material Sent to Palm Beach,” folder, National Security Files, Box 45, John F. Kennedy Library. 106 of this one event is better explained by the psychological nuclear learning model than rational choice learning models, but Kennedy’s not experiencing fear at Cuba and not adopting pessimistic risk estimates could be explained by psychological or rational models. It is unclear, however, that the nuclear learning models developed in Chapter 2 apply to cases other than the Cold War. In the next Chapter I apply them to nuclear South Asia. 107 Chapter 4: Pakistani Nuclear Missiles, 1998-2002 4.1. Pakistani Nuclear Missiles, 1998-2002 I have shown that the psychological nuclear learning model explains the Soviet case better than the rational choice alternative. But does this conclusion travel to South Asia? Pakistani nuclear diplomacy is a hard case to test psychological explanations of foreign policies against rational choice alternatives. Very little data on the beliefs of senior civilian and military decision-makers exists. In this chapter, I utilise interview data and the secondary literature to show that Pakistani behaviour after developing nuclear weapons capable of targeting its principal adversary is similar to Soviet behaviour in the 1960s despite territorial contiguity to its principal adversary, a historical animosity predating nuclear weapons and formerly Pakistani sponsored non-state terrorist groups. I provide some evidence about the Pakistani (and Indian) causes of and lessons from the Kargil war and ten month crisis that are consistent with the psychological nuclear learning model but insufficient to dismiss the rational choice alternative. Moreover, I cannot tell whether Musharraf’s successors, current Prime Minister Zardari and Army General Kayani, experienced fear of nuclear escalation during the crisis years. This leaves me unable to address the common argument that South Asia is more dangerous than the Cold War. But I show that the psychological learning model is a plausible explanation of Pakistani behaviour and the impact of nuclear proliferation on war propensity in South Asia. How have Pakistani nuclear weapons influenced the state’s conflict propensity? 108 Does the psychological nuclear learning model explain Pakistani behaviour? India and Pakistan were able to produce nuclear weapons at very short notice by 1990. After almost three decades of peace since the second Indo-Pakistani war in 1971, the 1998 Indian and Pakistani nuclear tests were followed by the two states fighting a war at the Kargil sector of Kashmir in 1999 and subsequently engaging in a ten month mobilised military crisis in 2001-02. Indian and Pakistani dialogue over Kashmir has progressed since then but has been periodically challenged by a number of terrorist attacks in India by once Pakistani sponsored Kashmiri insurgent groups. The question of the impact of nuclear weapons on Pakistani conflict propensity has major policy relevance at a time when the Kashmir dispute remains unresolved and the United States desires Pakistani support in its war in Afghanistan. If the Pakistani army is preparing to fight its long time Indian enemy, Pakistan will be unwilling to commit military resources and political capital to the American war in Afghanistan. The Pakistani case also raises theoretical questions. Pakistan is, like the Soviet Union, militarily and economically inferior to its chief adversary India and territorially revisionist. However, two aspects of the Pakistani case differentiate it from the Soviet Union. Firstly, much Pakistani coercive behaviour has been motivated by the desire for favourable American intervention in Kashmir. Pakistan could never achieve its territorial goals there without support from a third party. Khrushchev, on the other hand, never desired third party intervention in Berlin. Secondly, formerly Pakistani sponsored Kashmiri insurgent groups that desire to destabilise Indo-Pakistani cooperation have struck targets inside India. Khrushchev never struck targets inside the United States. The psychological nuclear learning model may explain Khrushchev’s Soviet Union, but does 109 it explain a case such as Pakistan with these two additional independent variables? The answer to this question is critical, because future nuclear powers such as Iran and North Korea will resemble Pakistan in these respects more than the Soviet Union. The scholarly and popular consensus is that nuclear weapons have had a strong and dangerous impact on Pakistani conflict propensity. In the aftermath of the 1998 Indian and Pakistani nuclear tests, U.S. President Bill Clinton famously referred to Kashmir as the “most dangerous place in the world.”208 President George W. Bush’s senior advisers, if not the President himself, believed that inadvertent escalation would cause nuclear war in South Asia.209 More recently, President Barack Obama’s coordinator for arms control and WMD terrorism claimed that “the risk of a conflict escalating to a nuclear war is probably higher in South Asia than in anywhere else in the world.”210 Some scholars have echoed Waltz’s argument that the consequences of nuclear war have deterred major aggression and will continue to do so.211 But many more have argued that Pakistani nuclear weapons emboldened Pakistani military decision-makers to pursue their revisionist Kashmir objectives without fear of retaliation. Indian retaliation and inadvertent escalation, this argument goes, could cause future nuclear war.212 But if the quantitative cross-national findings on the effects of experience with nuclear weapons 208 Peter Popham, ‘The World’s most dangerous place’ is already at war,’ The Independent: Asia, March 18 2000. 209 Polly Nayak and Michael Krepon, U.S. Crisis Management in South Asia's Twin Peaks Crisis, Henry L. Stimson Centre, Report No. 57, September 2006 210 Peter Crail, Daniel Horner and Daryl G. Kimball, “Pursuing the Prague Agenda: An Interview with White House Coordinator Gary Samore,” Arms Control Today, May 2011. http://www.armscontrol.org/act/2011_05/Samore, accessed May 19 2011 211 See, for example, Sumit Ganguly, “Nuclear Stability in South Asia,” International Security 33 (2) Fall 2008 pp. 45-70. 212 The most influential statements are S. Paul Kapur, “Ten Years of Instability in Nuclear South Asia,” International Security 33 (2) 2008 pp. 71-94; and Kapur, “Revisionist Ambitions, Capabilities, and Nuclear Instability,” in Scott D. Sagan (ed.), Inside Nuclear South Asia, 2009. Stanford: Stanford University Press, pp.184-218. 110 apply to South Asia, the impact of Pakistani nuclear weapons on the state’s conflict propensity may have been more complicated. Nuclear proliferation may have initially made Pakistan more conflict prone, but Pakistani decision-makers may have learned about the limits of nuclear coercion through fear and become less conflict prone over time. In this chapter I apply the psychological and rational choice nuclear learning models to the Pakistani case. Interview data, including with former Pakistani President Pervez Musharraf, allows a plausibility probe of the two learning models. I make two contributions. Firstly, I offer a theoretical account of the initial instability caused by nuclear proliferation to South Asia and the subsequent less dangerous period with the same theoretical mechanism. This is not insignificant. In the current orthodox wisdom, Paul Kapur explained Pakistani restraint with ad hoc variables. In his argument, the American response to the September 11 attacks was necessary to contain Pakistani revisionism. But my interview data suggests that Musharraf learned through fear about the dangers of nuclear coercion during the second peak of the ten month crisis, shortly after the September 11 attacks. The psychological learning model may explain Pakistani behaviour. The September 11 attacks may therefore have been unnecessary for Pakistani restraint because Musharraf’s learning about the danger of nuclear coercion through fear may have caused this. Secondly, I show that the Pakistani desire for favourable American intervention and ability to sponsor terrorists that can strike inside India alters the logic of nuclear coercion but not the mechanisms of nuclear learning. These two actors influenced when Pervez Musharraf coerced India and learned through fear about the dangers of nuclear 111 coercion. New nuclear states that rely on favourable third party intervention before coercing their adversary will wait for the opportune time when they believe this will occur. Rather than coercing India immediately after developing nuclear weapons, Pakistani decision-makers waited until they believed that they would receive favourable American intervention after the 1998 Indian nuclear tests. American crisis management may have caused Indian and Pakistani decision- makers to learn that nuclear coercion is not dangerous because Washington will contain escalation; it might encourage the Pakistani revisionist aggression (and aggressive Indian retaliation) that it was designed to prevent. Although this makes South Asia more dangerous than the Cold War, it has been balanced by terrorist attacks inside India that have encouraged Indian retaliation, inadvertent escalation and Pakistani learning about the dangers of nuclear coercion. While Pakistani decision-makers initially sponsored the Kashmiri insurgent groups to cause enough destruction to force Indian Prime Ministers to negotiate the status of Kashmir, these groups have developed substantial autonomy from Pakistani state institutions and have come to challenge their former sponsors. Fearing that their interests would be neglected in Indo-Pakistani negotiations, they have attacked targets inside India to derail Indo-Pakistani cooperation. This has caused undesired nuclear crises from which some Pakistani decision-makers have learned; I show below that former President Pervez Musharraf learned of the dangers of nuclear coercion through fear of nuclear escalation in the 2001-02 ten month crisis. United States intervention and Kashmiri terrorist groups have thus differentiated mechanisms of Pakistani nuclear learning, but Pervez Musharraf’s lessons about nuclear coercion were 112 similar to Khrushchev’s.213 There is insufficient data on the sources of Musharraf`s lessons so I cannot systematically test the psychological and rational choice nuclear learning models. Moreover, whether Musharraf’s civilian and military successors have learned these lessons is unclear, so I cannot tell whether recent terrorist attacks were sponsored by elements of the Pakistani state and thus whether the psychological theory would lead us to expect Pakistani nuclear and foreign policy restraint to have occurred. I show that the psychological nuclear learning model explains important dynamics of Pakistani nuclear strategy that rational choice models cannot easily account for. Some evidence suggests that rather than considering the historical record of nuclear coercion and American responses to such coercion, the Pakistani planners relied too much on their own recent and salient experiences of Indian isolation after the 1998 nuclear tests. Moreover I suggest that Musharraf`s experience of fear may have moderated Pakistani revisionism after 2002. I also subject the psychological nuclear learning model to a further test by assessing how well it explains Indian Prime Minister Vajpayee’s and his associates’ lessons from the Kargil war. I firstly provide a brief historical introduction to establish that Pakistan is a weak revisionist state such that it is a most likely case for the nuclear learning models. I then turn to the Kargil war and address whether Pakistani nuclear weapons were sufficient to cause it. Having established this, I proceed to apply the psychological and rational choice learning models to the causes and consequences of the Kargil war. I conclude with some speculative remarks about American nuclear crisis management in South Asia. 213 One might argue that this stalling of learning the limits of nuclear coercion makes South Asia more dangerous than the Cold War. But terrorist group intervention could speed up the learning process through attacking sooner. 113 4.2. The 1999 Kargil War Indo-Pakistani relations have been dominated by the dispute over Kashmir for much of the time since independence.214 India currently controls two thirds of the area; a Pakistani sponsored insurgency against Indian-held Kashmir began in the late 1980s. Support of the insurgency and the ultimate goal of liberating Indian held Kashmir is, according to Pervez Musharraf, the “core issue” in Indo-Pakistani relations.215 The Maharajah of Kashmir signed an instrument of accession and joined India in 1947, but the Pakistani government has long believed that Kashmir's accession to India was illegal and undemocratic, and continues to maintain that the Indian refusal to hold a plebiscite on the question of accession denies the Kashmiris their self-determination rights. 216 The 1947- 48, 1965 and 1999 wars have all been about the ownership of the territory, and the Kashmir issue is at the heart of Indo-Pakistani tensions for nationalist and strategic reasons. The disposition of Muslim majority Kashmir has significant implications for Indian and Pakistani statehood. The justification for Pakistan's creation in 1948 was the need for a religious home for South Asian Muslims. If Pakistan were to cede Kashmir – the state of Jammu and Kashmir is about two thirds Muslim and Kashmir valley is about ninety- five percent Muslim – to India, it would suggest that South Asian Muslims do not need a 214 Ian Talbot, Pakistan: A Modern History, 1999. New Delhi: Oxford University Press, p.113; Robert G. Wirsing, India, Pakistan and the Kashmir Dispute: On Regional Conflict and Its Resolution, 1998. New York: St Martin's; Stephen P. Cohen, India: Emerging Power, 2001. Washington, DC: Brookings Institution Press, pp.217, 220 215 K.J.M. Varma, “Pak Not Ready to Sideline Kashmir: Musharraf,” Press Trust of India, June 17, 2003 216 Sumit Ganguly, The Crisis in Kashmir: Portents of War, Hopes of Peace, 1997. New York: Cambridge University Press, p.10 114 religious homeland and could live in a secular multi-ethnic Indian state. This would call Pakistan's sovereignty into question. India was conceived as a heterogeneous democracy capable of being a homeland to all South Asia's diverse religious and ethnic groups. If India were to cede Kashmir to Pakistan, it would suggest that a religious or ethnic minority cannot live within a heterogeneous South Asian democracy. This would challenge fundamental Indian conceptions of multi-ethnicity and would be a dangerous public perception for a state as diverse as India.217 Strategically, Kashmir contains the headwaters of the important Indus and Punjab rivers. Many Indians believe that foreign control of Kashmir's mountains may leave parts of Punjab and perhaps Delhi vulnerable to attack. Pakistanis believe that Islamabad's proximity to Kashmir leave the capital open to an Indian assault. Finally, Kashmir borders strategically important China and Afghanistan.218 India and Pakistan therefore view possession of Kashmir as a primary security goal. Within a year of the 1998 nuclear tests, members of the Pakistani Northern Light Infantry Unit, disguised as tribal mujahedeen, penetrated twelve kilometres into Indian held Kashmir at the Kargil-Dras sector of the Line of Control (LoC). While Army chief Pervez Musharraf and Prime Minister Nawaz Sharif later blamed each other on the aggression after Pakistani complicity was revealed, it is likely that Musharraf vaguely informed Sharif of the nature of the operation. Sharif approved it but did not realise its likely operational procedures or low probability of succeeding.219 Incursions into enemy territory in Kashmir had been common since the 1980s but the depth of intrusion and 217 Wirsing, India, Pakistan and the Kashmir dispute, p.125; Ganguly, The Crisis in Kashmir; Cohen, India, p.215, Talbot, Pakistan, p.114 218 Cohen, India, pp.212-13;Talbot, Pakistan, p.114. 219 Shuja Nawaz, Crossed Swords: Pakistan, Its Army and the Wars Within, 2008. Oxford: Oxford University Press, pp. 500-518 115 number of intruders at Kargil in 1999 was different to earlier challenges. Had Pakistan been able to consolidate its position, it could have freely shelled Indian Highway 1A, the road leading to the Indian positions on the strategically important Siachen glacier, and perhaps compelled a partial Indian withdrawal. This would not have changed the fundamental status quo in Kashmir, but it would have been an important symbolic victory that might have undermined Indian confidence and perhaps led Delhi to contemplate concessions. The aggression occurred within months of the 1999 Lahore peace agreement that committed Indian and Pakistani leaders to establish confidence building measures and notifications of future weapons tests. Indian Prime Minister Atal Vajpayee and his advisers were surprised at the aggression and took several weeks to redeploy troops and dislodge the Pakistani intruders through intense peak to peak combat. After two months Indian forces dislodged most of the Pakistani isolated outposts. After a meeting with President Clinton, Sharif withdrew his remaining troops to Pakistani held Kashmir. Did Pakistani nuclear weapons cause or influence the Pakistani intrusions into Kargil? There is no conclusive evidence that links Pakistani nuclear weapons to the initiation of the Kargil war and, moreover, establishes it as a case of nuclear coercion. Pervez Musharraf and senior Pakistani officials still deny that the mujahedeen were disguised Pakistani infantrymen. India and Pakistan have fought earlier wars – although not since 1971 – and it is plausible that the Kargil war would have occurred without Pakistani nuclear weapons. In his definitive volume on the war, Peter Lavoy and his co- authors argued that the Kargil planners were not emboldened to undertake the operation because Pakistan's nuclear weapons capability was demonstrated in the previous summer: 116 None of the planners of the Kargil operation had any experience in their career with nuclear theory and its integration within conventional military strategy. Nuclear planning and strategy in Pakistan were still in a very nascent stage. Significantly, it was this lack of understanding of the meaning of the nuclear revolution that made Kargil planners act as if nothing had changed. Kargil planners at the time still acted as if they lived in a pre-nuclear, conventional world. They were mainly concerned with operational imperatives, restoring honor, and retribution for Siachen.220 I argue that Lavoy and his co-authors are probably wrong. There are three reasons why Pakistani nuclear weapons were probably necessary for the Kargil war. Firstly, if Pakistani nuclear weapons were unnecessary for the Kargil war, it should have occurred in the late 1980s when Pakistan did not have nuclear weapons and began providing extensive political, material and military support for the insurgency.221 The plan that Musharraf authorised and which Sharif signed off on in early 1999 had been floating in senior Pakistani military circles since at least 1987. After a successful Indian attack on June 1987 that dislodged Pakistani troops from a forward outpost overlooking Indian deployments, and an unsuccessful Pakistani attempt to recover these losses in September 1987, alternative Pakistani plans were formulated to gain control of the area. Army General Mirza Alam Beg claimed at a 1989 press conference that two years earlier Benazir Bhutto refused to authorise a plan to occupy the Kargil heights to disrupt the Srinagar-Leh road.222 Some Pakistan military analysts have argued that the operation 220 Feroz Hassan Khan, Peter R. Lavoy and Christopher Clary, “Pakistan's motivations and calculations for the Kargil conflict,” in Peter Lavoy (ed.) Asymmetric Warfare in South Asia: The Causes and Consequences of the Kargil Conflict, 2009. New York: Cambridge University Press, pp.64-91, p.90 221 Wirsing, India, Pakistan and the Kashmir Dispute, pp.114-118 and Ganguly, The Crisis in Kashmir, pp. 14-42. It is of course possible that Pakistan’s imminent nuclear weapon capacity in 1990 emboldened the Pakistani elite to sponsor the Kashmir insurgency then. 222 V. R. Raghavan, Siachen: Conflict Without End, 2002. Delhi: Viking books, pp.92-93. 117 became feasible only after 1995 upon completion of a small road on the Pakistani side of the LoC that allowed forward logistical support for local military operations.223 Why then did the Kargil war occur four years after this road was completed? India’s limited success in suppressing the decade-old insurgency in Kashmir in the 1990s caused growing concern in Pakistan’s politico-military establishment that the Kashmiri cause was losing its international salience.224 The superpowers had been unwilling to be involved in earlier LoC crossings in 1972, 1984 and 1988, and Pakistani leaders worried that this prohibited them from achieving a favourable resolution to the Kashmir dispute. The 1998 nuclear tests isolated India and brought renewed international attention to the Kargil dispute. The long history of estrangement between the United States and India and the intense U.S. pressure and condemnation on New Delhi in relation to its 1998 nuclear tests probably caused the Pakistani belief that the U.S. would favourably intervene. While Clinton offered Sharif many incentives to not reciprocate, the Pakistani tests were widely viewed as a defensive response to Indian belligerence. All of this encouraged the Pakistani belief that the United States might favourably intervene.225 While a reduction of troops at Siachen during the harsh winter months was a common practice, border security in the Kargil sector during the spring and summer of 1999 was particularly sparse.226 Musharraf and his associates would have known this. 223 Waheguru Pal Singh Sidhu, “Of Myths and Realities: The Kargil Experience,” in Kanti Bajpai, Afsir Karim and Amitabh Mattoo (eds.) Kargil and After: Challenges for Indian Policy. New Delhi: Har-Anand Publications, 2001, pp. 107-124, p.109 fn3 224 Sumit Ganguly and Devin T. Hagerty, Fearful Symmetry: India-Pakistan Crises in the Shadow of Nuclear Weapons, 2005. Seattle, University of Washington Press, pp. 193 225 Ashley Tellis, C. Christine Fair and Jamison Jo Medby, Limited Conflicts Under the Nuclear Umbrella: Indian and Pakistani Lessons from the Kargil Crisis. Santa Monica, RAND: 2002, p. 21 226Ganguly and Hagerty, Fearful Symmetry, p. 153. Moreover, Indian defences were particularly sparse because Prime Minister Vajpayee and his associates did not expect Kargil like challenges. 118 The former head of the Inter-Services Intelligence (ISI) said that “the Indian army is incapable of undertaking any conventional operations at present.”227 This pattern of events suggests that Pakistani implementation of the Kargil plan within months of the nuclear tests is more than coincidental. While perceived U.S. support and a tactical opportunity were likely necessary for the Pakistani Kargil intrusion, Pakistan’s nuclear weapons were probably more important. I show below that Indian and Pakistani nuclear tests caused the tactical opening, and recent evidence suggests that the Pakistani Kargil strategy depended on Pakistani nuclear weapons to contain Indian retaliation. Secondly, Paul Kapur has offered strong evidence that nuclear proliferation was necessary to embolden the Pakistani leadership to encroach Indian territory at Kargil. Former Prime Minister Benazir Bhutto stated that Pakistani leaders believed that “having nuclear capabilities would ensure that India could not launch a conventional war.”228 She was presented with a Kargil-like plan in 1989 and 1996 but believed that the United States would force Pakistan to retreat.229 Similarly, former Foreign Ministry Director General for South Asian Affairs Jalil Jilani stated that “since Pakistan’s acquisition of nuclear capacity, Pakistan has felt much less threatened;” nuclear weapons allowed policies that could “put a check on Indian ambition.”230 Other evidence also suggests that Pakistani nuclear proliferation was necessary for the Kargil war. Indian army General at the time V.P. Malik noted that “there was a strong belief that Pakistan’s demonstrated nuclear weapons capability in May 1998 was sufficient to prevent the escalation of the Brajesh Mishra, author interview, Delhi, April 2009 227 Kapur, Dangerous Deterrent, p.123 fn 44 228 S. Paul Kapur, “Revisionist Ambitions, Conventional Capabilities, and Nuclear Instability,” in Scott Sagan (ed.) Inside Nuclear South Asia. Palo Alto: Stanford University Press, pp. 184-216, p.195 229 See Khan, Lavoy and Clary, Pakistan's motivations, pp. 77-79 230Kapur, Nuclear Instability, p.196 119 situation in Kargil to a full-scale conventional war level.”231 India’s Joint Intelligence Committee noted the Pakistani perception in 1991 that low-intensity conflict was feasible against India because Pakistan’s nuclear capability would deter Indian escalation.232 Lavoy and his contributors may be correct that nuclear deterrence theory was a vague notion amongst Pakistani military circles in 1999. Operational imperatives, the restoration of honor and retribution for India's perceived betrayal of the 1984 Simla agreement may indeed have also influenced the Pakistani aggression. But if the Kargil war planners believed that their nascent nuclear arsenal would shield them from potential Indian conventional retaliation and nuclear escalation, Pakistani nuclear weapons were necessary to have caused the war. While they had revisionist preferences long before nuclear missile development, they would not have been prepared to risk Indian escalation before this because their conventional inferiority would have risked a Pakistani defeat. Even after developing nuclear weapons, Pakistan’s nuclear posture since 1998 has been designed to deter nuclear and conventional war.233 Thirdly, the Kargil war exhibited less escalation by India and Pakistan and a different type of resolution to earlier Indo-Pakistani wars.234 In both the 1965 and 1971 wars, India and Pakistan quickly escalated the fighting and opened second fronts. In the 1965 Kashmir war, for example, India escalated the battle into Pakistani territory in 231V.P. Malik, Kargil: From Surprise to Victory, 2006. India: Harper Collins, p.272 232V.P. Malik, “Lessons from Kargil,” Bharat Rakshak Monitor 4, No. 6 (May-June 2002), www.bharat-rakshak.com/MONITOR/ISSUE4-6/malik.html 233 Narang, Pakistan's Nuclear Postures 234All three wars were limited wars. Pakistan sought to dislodge India from Kashmir in 1965. It did not seek to unravel the Indian state and did not harbour visions of wider territorial conquest. In the 1971 war, while India clearly sought and successfully managed to break up Pakistan through its intervention in the East Pakistan crisis, it did not seek to choke off and incorporate any part of Pakistan into its own domain. In the 1999 war, Pakistan only hoped to occupy a small part of Indian occupied Kashmir to use as a bargaining tool in U.S. initiated Kashmir peace talks. 120 response to an attack on Jammu and Kashmir. The Indian Air Force attacked targets deep into Pakistan just as Indian strike corps attacked and seized territory in Pakistan’s Punjab and Sindh provinces. All available weapons systems were employed.235 Offensive operations were conducted in the form of armoured thrusts with air support into Pakistan over a wide front. The Indian navy also attacked Pakistani ports and forced a partial blockade from the sea.236 The Kargil war exhibited significant restraint as Indian forces did not cross the Line of Control and refrained from the opening of additional fronts. Despite pressure from the army, Indian Prime Minister Vajpayee did not enlarge the theatre of operations beyond the Kargil sector or attack Pakistani forces, staging posts or lines of communication beyond the LOC. This defied military logic and entailed the 235 See V. R. Raghavan, “Limited War and Nuclear Escalation in South Asia,” The Nonproliferation Review 8 (3) 2001 pp.82-98, p.88 236In 1965, a far weaker Indian army with untested political leadership lost little time in resorting to horizontal escalation when Pakistan launched Operation Gibraltar in Kashmir. Within a week of the Pakistani attack, India had opened a second front in the Punjab threatening significant Pakistani civilian and military assets. On August 14, Pakistani regular forces made a major incursion across the Cease-Fire line. The next day the Indian army retaliated by crossing the CFL into Pakistani Kashmir. The Pakistanis counter-attacked and shelled Indian positions at Tithwal, Uri and Poonch. The village of Jaurian, 14 miles within Indian controlled Kashmir, fell to advancing Pakistani forces. Because this threatened Indian control of Jammu and Kashmir, Indian forces attacked across the international border near the key Pakistani city of Lahore, and also launched another powerful attack towards the town of Sialkot, a major Punjabi railway and road centre. The Indian forces reached within miles of the major metropolis of Lahore only hours after the commencement of that offensive. Pakistan responded with seventy tanks and two infantry divisions, and India responded with air support. Pakistan called in their air assets, and for the remainder of the war both sides made extensive use of air operations in support of their ground forces over enemy territory. In the 1971 war Pakistan, unable to deal with the growing strength of the insurgents within East Pakistan and frustrated with continuing Indian support of them, launched an air strike at India’s northern military bases on December 3. The IAF responded in kind the following day, and within days established air superiority over Pakistan. The navy bombarded Pakistan’s key port at Karachi. An Indian heliborne and ground forces offensive succeeded in making a small thrust into Pakistan’s desert sector towards Rahimyar Khan. If a larger armoured and mechanised forces thrust had been made in this weakly defended waist of Pakistan, there was a risk of the country being strategically split. By mid December, the Indian navy had established a virtual naval blockade between the two wings of Pakistan. On December 16 Pakistani forces were easily routed and Indian forces entered Dacca. See Ganguly and Hagerty, Fearful Symmetry, pp.28-32, 160, Raghavan, Limited War, p.91 121 acceptance of heavier casualties.237 The Indian air force had strict orders to avoid crossing into or attacking targets in Pakistan-administered Kashmir, although this complicated the task of dislodging the intruders.238 The scale of Pakistan’s covert operation at Kargil was unprecedented in the history of the violent peace in Kashmir; Pakistani troops infiltrated further into Indian Kashmir than they ever had.239 Once they had begun to suffer significant casualties under the onslaught of Indian howitzers, Pakistan did not back them up with reinforcements. Islamabad had increased its official support to the infiltrators sent into India during the 1948 and 1965 operations when they began to fail. The Pakistani Air Force studiously avoided raising the ante and the IAF continued to enjoy local air superiority throughout the conflict.240 Nuclear weapons also profoundly influenced the resolution of the Kargil war; South Asian nuclearisation attracted U.S. interest in the management and resolution of Indo- Pakistan crises. The U.S. had not taken a position on Indian or Pakistani wrongdoing in 1948, 1965, or 1971.241 The 1987 Brasstacks and 1990 Kashmir crises, although they 237 Kapur, Ten Years of Instability in a Nuclear South Asia, International Security 33 (2) pp. 71- 94, pp. 77-80; Author Interview, Brajesh Mishra, Delhi, April 2010 238 See P.R. Chari, Pervaiz Iqbal Cheema and Stephen P. Cohen, Four Crises and a Peace Process: American Engagement in South Asia. Washington D.C.: Brookings, 2007, p.139 239 Since India won a mountaineering race in 1984 to establish military control of the Saltoro mountain range that contains the Kargil region of the Siachen glacier, Pakistan launched several attacks to establish a foothold on the Saltoro ridgeline. There has been a constant exchange until 2003. India and Pakistan have sustained over 3500 casualties since 1984, with many arising from the harsh requirements of glacier living rather than enemy fire. Raghavan, Siachen, p.56; Chari et al, Four Crises, p.21 and Shaukat Qadir, “An Analysis of the Kargil Conflict 1999,” The RUSI Journal, 147 (2) 2002 pp. 24-30 240 V.K. Sood and Pravin Sawhney, Operation Parakram: The War Unfinished, 2003. New Delhi: Sage, p.71 241 President Johnson, consumed with Vietnam, expressed little interest in the 1965 war. The United States deployed two ships from its Middle East Force to Karachi while the U.S. Air Force evacuated Westerners from West Pakistan. The war was resolved under the Brezhnev brokered Tashkent ceasefire agreement in January 1966. During the 1971 war the United States sent the Enterprise carrier battle group and an amphibious-ready group to the Indian Ocean on behalf of Pakistan, but by then the war had ended and East Pakistan had become Bangladesh. Pakistani 122 occurred at a time when India and Pakistan were widely believed to be able to construct nuclear bombs, received relatively little attention by the United States and former Soviet Union. President Clinton however invested significant diplomatic capital in permitting Prime Minister Nawaz Sharif to travel to Washington to contain escalation when Indian forces were poised to cross the LoC.242 Clinton was very concerned that the Kargil crisis could escalate to nuclear war; He believed that “nuclear arsenals made both India and Pakistan less secure.”243 He quickly insisted that Pakistan should withdraw its forces back behind the Line of Control immediately.244 While definitive conclusions remain elusive, the evidence suggests that Pakistani and Indian nuclear proliferation strongly influenced the initiation, conduct and resolution of the Kargil war. forces had surrendered after a comprehensive Indian victory. 242 Most Indians and Pakistanis viewed the 1990 Kashmir crisis at the time as the direct result of Indian management of a Kashmiri insurgency. A senior Bush administration official however noted that he thought that most of his colleagues believed that South Asian nuclear proliferation caused the crisis. By 1990, Pakistan’s covert nuclear weapons program had replaced the Afghan war as the primary focus of America's South Asia policy. See Chari et al, Four Crises, p.82, 106 243 Bill Clinton, My Life. New York: Vintage, 2004 p.597. Clinton aide Bruce Reidel concurred that the President and his colleagues were alarmed from the beginning of the conflict because of its potential for nuclear escalation; “we could all too easily imagine the two parties beginning to mobilize for war, seeking third party support (Pakistan from China and the Arabs, India from Russia and Israel) and a deadly descent into full scale conflict all along the border with a danger of nuclear cataclysm.” See Bruce Riedel, American Diplomacy and the 1999 Kargil Summit at Blair House. Center for the Advanced Study of India, University of Pennsylvania, 2002, pp.3-4 244 He told Sharif that “we can’t get into a position in which India feels that because of what you’ve done, it has to cross the Line of Control itself. That would be very dangerous. I genuinely believe you could get into a nuclear war by accident.” Strobe Talbott, Engaging India: Diplomacy, Democracy and the Bomb. Washington: Brookings, 2004, p.165; Clinton, My Life, pp. 864-66, Reidel, Blair House. 123 4.3. Pakistani Lessons from Kargil Why did Pakistani President Pervez Musharraf refrain from challenging the Kashmiri status quo and commit to the Composite Peace Dialogue by the end of 2002? Indian redeployments to Kashmir after Kargil made a second Pakistani territory grab costly but not impossible, and other variables such as the poor state of the Pakistani economy, Pakistan's political and military leadership after September 1999 and the overall military balance remained constant.245 Pakistani revisionist preferences remained after the Kargil war's failure to achieve a lasting change to the Kashmir status quo. American pressure on Pakistan after the September 11 attacks to refrain from sponsoring terrorism, as Paul Kapur has argued, may have caused Pakistani restraint.246 But Musharraf experienced fear of nuclear escalation during the second peak of the ten month crisis, two years after the Kargil war, and this may have been sufficient to reduce or eliminate further Pakistani challenges in Kashmir and further commit Pakistan to improving its overall relationship with India through the Composite dialogue. Unavailable data on behaviour and cognitive processes limit an assessment of the psychological and rational choice nuclear learning models. Firstly, it is unclear if 245After Pakistani troops retreated behind the LoC, India strengthened its force levels there at Ladakh. A new Corps Headquarters was raised in the region and additional forces were positioned opposite Pakistani positions. These measures changed the comparative balance of forces in the area. Yet Indian forces did not render the LoC impregnable. Malik claimed that despite eyeball to eyeball deployment of troops on both sides and several stretches of minefields, the Saltoro ridgeline is a highly infiltration prone area. Undulating terrain and forest cover present substantial logistical difficulties and make major offensive operations untenable. “It is neither physically possible nor tactically desirable to cover the entire length of any border with manpower…the pattern of deployment with long gaps between defences was followed on either side of the LoC.” See Malik, Kargil, p.61, p.70 and Raghavan, Siachen, pp.202-203 246 Kapur, Ten Years of Instability 124 Pakistani state sponsored revisionism persisted after 2002. Although the intensity of the Kashmir insurgency declined after the ten month crisis de-escalated, it is unclear if Musharraf continued to sponsor different groups after 2002. Musharraf’s role in several small 2003 bus and train bombings and the larger 2006 train bomb attack is unclear. Moreover, the role of Inter-Services Intelligence, the military intelligence unit of the Pakistani Army, and the links between Musharraf and this body are unclear. Secondly, there is no data on the cognitive processes that led Musharraf to authorise the Kargil war and reflect on its lessons. It is unclear if Musharraf systematically analysed historical cases of nuclear coercion or exclusively relied on his own cognitively accessible personal experiences. An interview with Musharraf, conducted in Seattle in March 2009, does not falsify the rational choice model but provides some support for the psychological model. Conclusive evidence on Musharraf’s and his associate’s lessons from the crisis years is probably elusive, but the psychological nuclear learning model explains currently available data as well as the rational choice alternative. Moreover, the psychological nuclear learning model, through suggesting that Musharraf may have experienced fear of imminent nuclear war during the ten month crisis and adopted pessimistic risk estimates of the dangers of nuclear coercion and other unrelated foreign policies, explains the same data that the orthodox wisdom relies on and suggests that nuclear weapons may have had a less dangerous impact on South Asia. Below I show that the evidence suggests that (1) Musharraf`s own personal experiences were important variables in causing the Kargil war, (2) Musharraf did not experience fear of nuclear escalation at Kargil but did during the second peak of the ten month crisis and that (3) he continued to sponsor the 125 insurgency between the Kargil war and ten month crisis. I address each in turn. 4.3.1. Psychological Causes: Availability Rational choice nuclear learning models would lead us to expect Musharraf to have systematically assessed the probability that President Clinton would favourably intervene in the Kargil war. The Kargil war planners could not have expected to prevail over numerically and technologically superior Indian forces but rather must have hoped that the territory grab would usher American intervention that would have allowed Pakistan to keep its gains or perhaps pressured India to negotiate the status of Kashmir. Although the United States has often exhibited an interest in defusing Indo-Pakistani nuclear crises, it has never intervened in a manner favourable to Pakistan. Rational choice models would therefore predict Musharraf to not have initiated the Kargil war because the historical record shows that nuclear coercion is usually ineffective and Pakistan’s allies have never entered a war on its behalf.247 The critical causal process observation of rational choice nuclear learning models is that Musharraf would have been attentive to these factors. Moreover, at a time when nuclear proliferation is an important American security concern, these models might predict Musharraf to have concluded – as his predecessors did – that no American President would favourably intervene and effectively support or tacitly accept a strategy of nuclear coercion and not have authorised the Kargil plan. The psychological nuclear learning model however predicts illusory correlations 247 Sechser and Fuhrmann, The Coercive Limits of Nuclear Weapons. 126 about the coercive utility of nuclear weapons to have caused Musharraf to have exaggerated the likelihood that his Kargil strategy would realise his revisionist Kashmir objectives. Moreover, it would predict that the cognitive accessibility of Musharraf’s personal experiences to strongly influence his preferences. While the Kargil plan had existed since the 1980s, India’s cognitively accessible isolation after the 1998 nuclear tests may have suggested that the United States would intervene in a manner favourable to Pakistan or at least not favourable to India.248Attention to the many other data points on American involvement in the Kashmir peace process, Indo-Pakistani wars and the record of nuclear coercion would have challenged this inference. There is no evidence that Musharraf reasoned as rational choice models predict, but absence of evidence is not evidence of absence. Some evidence however suggests that Musharraf and his associates were preoccupied with India’s isolation after the 1998 nuclear tests.249 Christine Fair conducted the interviews in Pakistan and reported that most of her interviewees believed that India’s nuclear test would cause the United States to favourably intervene. This evidence supports the psychological learning model, but there is no other evidence to evaluate the rational choice learning model. The lack of more data on the sources of Musharraf’s attention and the influence of different beliefs on the decision to authorise 248 The Indian nuclear test would have been highly cognitively accessible for two reasons. Firstly, the simple fact of Pakistan’s primary adversary testing a nuclear bomb, shortly after testing missiles capable of carrying a nuclear warhead that could reach Pakistan, constituted an obvious strategic threat or at least source of strategic change that Pakistani strategists would have immediately noted. Secondly, the Indian nuclear test caused Delhi to suffer economic sanctions and diplomatic isolation. Given that Pakistani strategy on Kashmir has always depended upon favourable intervention from the United States to help achieve what Pakistan could not, the isolation of Pakistan’s primary adversary from the United States would been keenly noticed by Musharraf and his associates. 249Ashley Tellis, C. Christine Fair and Jamison Jo Medby, Limited Conflicts Under the Nuclear Umbrella: Indian and Pakistani Lessons from the Kargil Crisis. Santa Monica, RAND: 2002, p. 21. Christine Fair, author interview, Seattle, September 2, 2011 127 the Kargil plan does not allow a more systematic assessment of the degree to which Musharraf learned from history and his own personal experiences. 4.3.2. Lessons The rational choice learning model’s predictions of Musharraf’s lessons from Kargil would depend on the strength of his prior beliefs about nuclear coercion. One could argue that it would not predict him to systematically overestimate the utility of nuclear coercion against India immediately after developing nuclear missiles. But whether this belief is rational depends on its underlying reasoning, and unavailable data on the sources of his private conclusions of the operation render a test of the rational learning model impossible. The psychological learning model has a stronger prediction. It predicts Musharraf to have continued to believe in the power of nuclear coercion until he experienced fear. Interview data lends plausibility to this hypothesis, although more data on the sources of this belief would be necessary to dismiss the rational choice explanation. Musharraf almost certainly did not experience fear of nuclear escalation at Kargil. He subsequently contrasted his belief that the Kargil war was “controlled and localised” with his belief that the ten month crisis exhibited great dangers of escalation.250 Some other evidence suggests that he continued to believe that nuclear coercion – in the form of sponsoring the Kashmir insurgency – would realise his Kashmir objectives. He 250 Pervez Musharraf, author Interview, Seattle, March 2010. Musharraf may have obvious incentives to deny the fear experienced at Kargil. However, he admitted to the fear he experienced at the ten month crisis. Moreover, many other South Asian observers agree that the Kargil war did not exhibit dangers of nuclear escalation. 128 seems to have learned through economic and diplomatic isolation of the costs of the Kargil operation but continued to believe that less visible nuclear coercion might achieve his goals.251 He stated in his autobiography that by January 2001 he believed the time had come to turn over a new leaf with India and find a non-military solution to their remaining problems.252 Yet this rested on a conviction that sponsoring Islamic insurgents in Kashmir had forced Indian Prime Minister Vajpayee to negotiate the status of Kashmir. Once India had come to the bargaining table, sponsoring the Kashmiri insurgency apparently became less imperative.253 4.3.3. Fear By 2001, the Kashmiri insurgent groups had developed substantial autonomy from their former Pakistani government sponsors and had come to threaten the Pakistani interests that they had once served. In the second Indo-Pakistani nuclear crisis in 2001, the ten month crisis, militants attacked the Indian parliament in Delhi and almost killed several senior politicians. At the crisis’ second peak in May 2002, when another terrorist group killed several dozen civilians at a military camp in Kaluchak, Jammu, Musharraf learned through fear of the dangers of nuclear escalation through sponsoring the Kashmir insurgency. Vajpayee had publically vowed to authorise his mobilised infantry strike forces to invade Pakistan, and Musharraf worried that Pakistani retaliation to this would 251 For similar conclusions about Pakistani strategy in 2000, see Tellis et al, Limited War, p. 7, 30 252Pervez Musharraf, In the Line of Fire: A Memoir, 2006. New York: Free Press, p.299 253 Pervez Musharraf, Author Interview, Seattle, March 2010. There is no data on the sources of Vajpayee’s desire to attend the Agra summit or the sources of Musharraf’s conclusion about Vajpayee’s willingness to meet him. 129 cause nuclear war. He apparently hardly slept for several nights and was deeply worried that the insurgents that Pakistan had once sponsored might cause an Indian response that would warrant a Pakistani response that would cause nuclear war.254 Uncertainty about whether Musharraf continued to sponsor the insurgency after experiencing fear prohibits an assessment of the learning models. The psychological nuclear learning model would lead us to expect that if Musharraf experienced fear of nuclear escalation, he should adopt pessimistic risk estimates in nuclear diplomacy and other foreign policy. It would be falsified if he experienced fear of nuclear escalation but continued to believe that sponsoring the insurgency to achieve his Kashmir goals was worth the nuclear risks. Rational choice learning models could not easily explain such a strong effect of fear on Musharraf’s risk propensity. A stronger test would be the degree to which Musharraf’s personal experiences and a systematic analysis of the historical record influenced his behaviour. These expectations also apply to Musharraf’s successor. Army General Ashfaq Kayani is not the civilian leader of Pakistan but has much control over Pakistani nuclear doctrine and deployment and the uses of Pakistani military assets for insurgency sponsorship. The psychological learning model would be falsified if he experienced fear of nuclear escalation during the ten month crisis and continued to sponsor the Kashmir insurgency. The rational choice learning model would lead us to expect his behaviour to be influenced by a systematic historical analysis rather than his cognitively accessible personal experiences. It is possible that Kayani may not have experienced fear during the ten month crisis. If so, current Pakistani nuclear and foreign policy toward India may be 254 Op. Cit. 130 more dangerous than Soviet policy towards the United States after 1962. Unavailable data does not allow a test of the nuclear learning models on Kayani’s lessons. Nonetheless, while Pakistani sponsored terrorist groups and American nuclear crisis management have differentiated Pakistani nuclear strategy from that of the Soviet Union, Pervez Musharraf’s lessons from the ten month crisis as well as the effect of experience with nuclear weapons in Pakistan are consistent with the same psychological logic. Christine Fair’s interview evidence suggests that Musharraf learned from the Indian nuclear test that nuclear coercion would be effective because Washington would favourably intervene. My interview data suggests that he continued to believe in the power of nuclear coercion, although through sponsoring the Kashmir insurgency rather than Kargil like territory grabs, through summer 2001. He seems to have continued to believe in the power of nuclear coercion until he experienced fear of nuclear escalation at the second peak of the ten month crisis in May 2002. Moreover, while it is unclear what proportion of insurgent attacks in Kashmir were sponsored by Musharraf, the high proportion of time exhibiting Pakistani nuclear coercion in the four years between the nuclear tests in May 1998 and Musharraf’s experience of fear of nuclear war in May 2002 seems too high to be attributed to random error. Rather, it suggests that Musharraf systematically overestimated the power of nuclear coercion as the psychological nuclear learning model would predict. More evidence may call the above analysis into question, but the psychological nuclear learning model is as consistent with the evidence as the orthodox wisdom and suggests that the effects of experience with nuclear weapons in Pakistan may have had a similar pacifying effect as the Cold War. 131 4.4. A Further Test: Indian Lessons from Kargil 4.4.1. Expectations of Kargil Psychological and rational choice models would also predict different Indian expectations of and reactions to Pakistani aggression at Kargil. Rational choice learning models would lead us to expect two outcomes. Firstly, Prime Minister Vajpayee and his associates should have been attentive to enough history or logic to consider the consequences of the Pakistani development of nuclear weapons. The Kargil war was costly to India as well as Pakistan, so rational choice models would predict Vajpayee to have expected Pakistani aggression or at least challenges to the status quo. Secondly, rational choice nuclear learning models would predict Vajpayee and his associates to have systematically considered the historical record of nuclear coercion before concluding that a coercive strategy designed to contain Pakistani terrorist strategy would succeed. Furthermore, rational choice models might predict Vajpayee to have systematically analysed whether he could influence or deter the state sponsorship of terrorism before authorising the Parakram mobilisation during the ten month crisis. The psychological nuclear learning model would predict two different outcomes. Firstly, it would only predict Vajpayee to have expected Pakistani aggression if there were cognitively accessible predictors of it. Vajpayee however faced no cognitively accessible predictors of war. Despite crises in 1986 and 1990, India and Pakistan had not fought a war since 1971. By 1999, India and Pakistan had exhibited twenty seven years of peace; the longest in their otherwise violent history. Perhaps most importantly, since India and Pakistan’s capability of constructing nuclear weapons at short notice in 1990, India and 132 Pakistan engaged in no wars or crises. A post 1990 nuclear related crisis might have taught Vajpayee of the dangers of Pakistani revisionism after the 1998 nuclear tests, but all cognitively accessible data pointed toward peace. The Indian nuclear test was itself motivated by Chinese exports of nuclear technology to Pakistan rather than concerns for Pakistani emboldenment or belligerency. Since India and Pakistan had developed virtual nuclear arsenals in 1998, the rivals had signed a series of confidence building measures prohibiting attacks against nuclear facilities, committing to advance notification of military exercises and troop deployments, and even exchanged of lists of nuclear installations and facilities from January 1993 through January 1999. By the late 1990s the Kashmir dispute had lost the violence and high international salience that it attracted earlier in the decade.255 All cognitively accessible events of the post 1990s period pointed towards regional peace. Two months before the Kargil intruders were discovered, Vajpayee thus claimed that “the nuclear weapon is not an offensive weapon.”256 He and Pakistani Prime Minister Nawaz Sharif signed the Lahore agreement that committed to different confidence building measures in the immediate aftermath of the 1998 nuclear tests. Most cognitively accessible data would not have suggested that a Pakistani challenge in Kashmir was likely, although even a cursory assessment of Soviet behaviour after developing the capability to target its adversary with nuclear weapons, or Pakistan’s long desire to revise the Kashmir status quo, would have suggested otherwise. Secondly, the psychological nuclear learning model would predict him to have 255 Brajesh Mishra, National Security Adviser, Author Interview, New Delhi, April 2010 256 Prime Minister A. B. Vajpayee, Statement of March 15 1999, cited in Prakash Karat, “Kargil and Beyond,” Frontline, www.frontlineonnet.com/fl1614/16140150.htm 133 learned from American intervention in the Kargil war and subsequent ten month crisis that the United States would similarly intervene in future Indo-Pakistani crises. It would thus predict Vajpayee to learn that compellence strategies to contain Pakistani revisionism are safe and effective because Washington would manage and contain nuclear escalation. However, insofar as the belief that the United States would intervene in a South Asian crisis reflects an accurate assessment of current beliefs in Washington, this belief is accurate and therefore explained by rational choice learning models. The psychological nuclear learning model explains Vajpayee and his associates’ beliefs about the consequences of Pakistani nuclear weapons better than rational choice learning models. They did not believe that Pakistani nuclear weapons (and Indian nuclear tests) would cause revisionism. According to Brajesh Mishra, Vajpayee’s national security adviser, the discovery of the Kargil intruders and their Pakistani identity shocked all of the civilian leadership.257 They believed after the nuclear tests that Pakistan would no longer risk a conflict in Kashmir, the strategic status quo would be permanently frozen, and that India would be able to deal with its Kashmir insurgency as an internal problem.258 There is no archival evidence on the sources of Vajpayee’s and his colleague’s beliefs about Pakistani behaviour after the nuclear tests, so the degree to which they learned from history and/or their personal experiences is unclear. Shortly after the second Indian tests but before the Pakistani tests, Home Minister L.K. Advani, a member of Vajpayee's cabinet committee on security, claimed that India's tests would make Islamabad “roll back its anti-India policy, especially with regard to 257 Brajesh Mishra, National Security Adviser, Author Interview, New Delhi, April 2010 258 Rajesh M. Basrur, Minimum Deterrence and India's Nuclear Security, Stanford: Stanford University Press, 2006, p.82 134 Kashmir.”259 Two months before the Kargil intruders were discovered, Vajpayee claimed that “the nuclear weapon is not an offensive weapon...if in the days of the Cold War there was no use of force, it was because of the balance of terror.260 Rational choice models can explain his attention to history but not his neglect of the consequences of Soviet nuclear missiles that offered critical lessons for his country. After the Pakistani nuclear test and increase in number of insurgents in Kashmir, Vajpayee increased his troop deployment there and authorised a conference of the All Parties Hurriyat to address local Kashmiri concerns but continued to believe that Pakistani aggression would not occur.261 Rational choice learning models can explain Indian Army General V.P. Malik affirming the possibility of a limited war before Kargil. But the psychological nuclear learning model explains why “not many people took (his) statement seriously until Kargil happened.”262 Similarly, the psychological learning model explains why Indian strategists who developed Indian nuclear doctrine after the nuclear tests did not take into account the proxy war or sub-conventional conflict in the subcontinent.263 The Kashmir insurgency was not cognitively accessible to Indian strategists in the 1990s thinking about the implications of nuclear weapons. 259 “India Ratchets Up Rhetoric against Pakistan and China,” Agence France Presse, New Delhi, 18 May 1998, cited in George Perkovich, India`s Nuclear Bomb: The Impact on Global Proliferation, 2001. New York: Oxford University Press, p.423 fn. 102 260 Prime Minister A. B. Vajpayee, Statement of March 15 1999, cited in Prakash Karat, “Kargil and Beyond,” Frontline, www.frontlineonnet.com/fl1614/16140150.htm; Ramesh Chandran, “Vajpayee and Musharraf agree on one point - No nuclear war,” The Times of India, 13 March 2000 261Brajesh Mishra, Author Interview, New Delhi, April 2010 262V.P. Malik, “Lessons from Kargil,” Bharat Rakshak Monitor 4, No. 6 (May-June 2002), www.bharat-rakshak.com/MONITOR/ISSUE4-6/malik.html. 263 Sonika Gupta and Arpit Rajain, “An interview with General V. P. Malik,” in P.R. Chari, Sonika Gupta and Arpit Rajain (eds.), Nuclear Stability in Southern Asia. New Delhi: Manohar, 2003, p.154; Basrur, Minimum Deterrence, p.81 135 4.4.2. Lessons from Kargil Vajpayee and his associates learned two lessons from the Kargil war. Firstly, they learned that conventional war is possible after nuclear proliferation. Secondly, they seem to have learned that they could practice nuclear coercion to stop the Pakistani government to from sponsoring the Kashmir insurgency because nuclear war was unlikely just as Musharraf and his colleagues had earlier learned of the effectiveness of the strategy. A test of the two learning models is impossible without data on the sources of these beliefs. However, their timing suggests that they may be better explained by the psychological learning model. If the rational choice model explains them, Vajpayee and his associates should have learned about nuclear proliferation and the use of force after nuclear before the Kargil war. That their beliefs changed after Kargil suggests that they learned from it and not the historical record. Some evidence shows what Vajpayee and his government learned but the sources of these lessons is unclear. Vajpayee and his associates learned from the Kargil war that “there is a strategic space to fight limited conventional war below the strategic threshold.”264 Defence Minister George Fernandes proclaimed in January 2000 that nuclear weapons “can deter only the use of nuclear weapons, but not all and any war;” Indian forces can fight and win a limited war at a time and place chosen by the aggressor.”265 Some evidence suggests that the belief that the United States would contain nuclear escalation may have emboldened Vajpayee to take a hard line towards Pakistani terrorist sponsorship. A senior Indian nuclear strategist noted in 2002 that as long as an 264 Kapur, Dangerous Deterrent, p.129 fn 66; p.133 fn 82 265 C. Raja Mohan, “Fernandes Unveils 'Limited War' Doctrine,” The Hindu, January 25 2000, 136 American naval fleet is present in the Arabian Sea or American troops are located in Afghanistan or Pakistan, Washington would not allow Pakistan to use its nuclear weapons.266 Vajpayee threatened conventional war against Pakistan if Islamabad did not curb cross-border violence in Kashmir. Possible action ranged from attacking terrorist camps and Pakistani military assets within Kashmir to destroying military targets and seizing small pieces of territory within Pakistan.267 Vajpayee and his colleagues unanimously agreed that the December 2001 attacks on the Indian parliament demanded a forceful response.268 He ordered a full military mobilisation that was designed to signal to the Pakistani leadership that Delhi would not tolerate such attacks and would be prepared to invade Pakistan to stop them.269 He also made an end to cross-border terrorism a precondition for future Kashmir negotiations with Pakistan. He demanded that Musharraf turn over twenty criminals suspected to be residing in Pakistan, unequivocally renounce terrorism, shut down terrorist training camps in Pakistani territory, and check militant infiltration into Jammu and Kashmir.270 Within a few weeks, half a million Indian soldiers were mobilised on the border and ready to march into Pakistan. However the causes of Vajpayee’s behaviour here and whether it was caused by lessons from history or his own experiences, learned during the crisis or perhaps earlier, are unclear. Despite American pleas and protests, Pakistan began redeploying its 11th and 12th Army Corps that were sent to Afghanistan only a month earlier to the Indian border. After 266 K. Subrahmanyam, “Indo-Pak Nuclear Conflict Unlikely,” The Times of India, 2 January 2002 267 See Gaurav Kampani, “India’s Compellence Strategy: Calling Pakistan’s Nuclear Bluff over Kashmir,” Monterey Institute of International Studies. 268 Brajesh Mishra, author interview, Delhi, April 2010. 269 Author Interview, Brajesh Mishra, New Delhi, April 2010. Mishra admitted that if the Parliament attacks did not occur something less severe was in the works. 270 See Praveen Swami, “Beating the Retreat,” Frontline, October 26-November 8, 2002 137 the May 14 attacks killed several civilians at Indian army barracks at Kaluchak, Vajpayee visited the front lines and told his troops that “the time has come for a decisive battle, and we will have a sure victory.”271 He almost authorised the 21 strike core, one of three brigades deployed on the border, to penetrate into Pakistani territory and eliminate terrorist facilities.272 However by this time Pakistan’s mobilisation had rendered such an Indian mission infeasible. As a response to the inability of the Indian army to mobilise its forces before Pakistan, The Indian Army has subsequently developed a new “Cold Start” doctrine to enable prompt mobilisation and invasion of small parts of Pakistan to contain Pakistani sponsored terrorist groups before Islamabad can retaliate.273 However there is little evidence on the sources of the motivations for Cold Start and the reasons why it has yet to become integrated into operational doctrine. Organisational inertia in the Indian army may have prevented the Cold Start doctrine from moving into an operational phase. Vajpayee may have authorised Cold Start to remain in the planning phase because of fear he may have experienced during the second peak of the ten month crisis. Manmohan Singh’s lessons from the mobilisation are unclear. He may have ordered Cold Start to not become operational because, consistently with the psychological nuclear learning model, he experienced fear during the ten month crisis, but may also have believed in the limits of nuclear coercion before 2001. 271 Luv Puri, “Be Ready for Decisive Battle, PM Tells Jawans,” The Hindu, May 23, 2002 272 Praveen Swami, ``A war to end a war: The causes and outcomes of the 2001-2 India-Pakistan crisis,`` in Sumit Ganguly and S. Paul Kapur (eds.) Nuclear Proliferation in South Asia: Crisis Behaviour and the Bomb, pp. 144-163. New York, Routledge, 2009. 273 See Walter C. Ladwig III, “A Cold Start for Hot Wars? The Indian Army's New Limited War Doctrine,” in International Security 2008 32 (3) 138 4.5. American Crisis Management in South Asia In this section, I briefly elaborate on the consequences of third party intervention. American intervention in the Kargil war and ten month crisis has created incentives for further Pakistani aggression (and dangerous Indian responses) because it has caused South Asian decision-makers to believe that Washington will contain nuclear escalation. The psychological learning model suggests that this American behaviour will be a cognitively accessible cue that will serve as a basis for predictions of future American behaviour. Rational choice models cannot explain the apparent inattentiveness to history that guides this assessment but are consistent with the fact that the United States may well intervene in future South Asian nuclear crises. This U.S. commitment has been not only ineffective but encouraged (if not caused) what it was designed to prevent. Moreover, terrorist attacks and the experience of fear may have caused the Pakistani restraint that American diplomacy could not achieve. It is impossible to establish what would have occurred at Kargil and the ten month crisis without American intervention. While Sharif quickly became desperate to resolve the Kargil war, Musharraf was resolved to maintain his troops positions at Kargil. Anthony Zinni, the Commander in Chief of U.S. Central Command, travelled to Islamabad and eased Musharraf’s reluctance to order a retreat.274 No American intervention may have led to a Musharraf inspired final Pakistani bid for victory that may have involved further escalation. Vajpayee never intended to cross the LoC because 274 Peter Lavoy, “Why Kargil did not produce general war: the crisis-management strategies of Pakistan, India, and the United States,” in Lavoy (ed.) Asymmetric Warfare in South Asia: The Causes and Consequences of the Kargil Conflict, 2009. New York: Cambridge University Press, pp 171-206, p.196 139 Indian forces began to prevail but did not rule out crossing it if this changed. That a prelude to the future 2001 mobilisation was authorised and called off when Musharraf ordered his troops to withdraw suggests that Zinni’s visit may have contained Indian or Pakistani escalation and postponed learning about the dangers of nuclear escalation.275 During the ten month crisis, President Bush spoke with both Musharraf and Vajpayee on June 5 and issued a statement that committed his administration to the easing of tensions.276 A senior Bush administration official reported that the general American sentiment was “we know how mad you are, but this is not the time to let MAD take over.”277 There was no point at which he and his associates said “these guys (India and Pakistan) can take care of this.”278 Musharraf's two statements condemning terrorism at the crisis's peaks in January and June – both issued under heavy American pressure – provided ample political room for Vajpayee to cancel orders to his Army to march into Pakistan. An Indian use of force after Musharraf`s concession would have been much more unsavoury to concerned international observers than that before it. In both the Kargil war and ten month crisis, no American intervention may have caused further escalation and quicker learning about the dangers of nuclear escalation. President Barack Obama recently issued a secret directive to intensify American diplomacy aimed at easing tensions between India and Pakistan, and asserted that detente between the two rivals was needed to help win Pakistani cooperation in Afghanistan.279 But the American commitment to Afghanistan will make it harder for Obama to credibly 275 Basrur, Minimum Deterrence, p.85 fn 20 276 Chari et al, Four Crises, p.169 277 Nayak and Krepon, U.S. Crisis Management, p.23 fn50 278 Polly Nayak and Michael Krepon, U.S. Crisis Management in South Asia's Twin Peaks Crisis, Henry L. Stimson Centre, Report No. 57, September 2006, p.43 279Wall Street Journal, April 6 2010, p.11. 140 commit to not intervene in a South Asian nuclear crisis. This may undermine the prospects for an India-Pakistan detente by creating incentives for Pakistani sponsorship of the insurgency and aggressive Indian responses. American policies to the Kashmir dispute should balance consistent support of the Kashmir peace process with the credible commitment that Washington will not intervene in a South Asian nuclear crisis. But such a balancing act is very difficult. Obama’s diplomacy might encourage the revisionism that it is designed to avoid if Pakistani decision-makers believe that Washington will contain nuclear escalation. Detente would be best achieved by allowing India and Pakistan to manage their own crises, and this may only be feasible after the American commitment to Afghanistan ends. If the United States cannot commit to not manage a South Asian nuclear crisis and Pakistani Army General Kayani has not learned of the dangers of nuclear coercion through fear, nuclear proliferation may continue to destabilise South Asia. The United States could do more to promote peace in South Asia by not managing regional nuclear crises. If Kayani and his associates have not experienced fear of nuclear escalation, such an experience might provide an impetus to Indo-Pakistani cooperation and make terrorist attacks designed to disrupt this much less likely to succeed. In summary, this chapter has shown that the psychological learning model explains Pakistani behaviour better than rational choice learning models. Christine Fair’s interview data suggests that Indian isolation after the nuclear tests was very cognitively accessible and likely caused the Pakistani Kargil intrusion on the grounds that the United States would favourably intervene. Rational learning models would not predict this data point to have had such a profound effect but rather predict Musharraf and his colleagues to have considered the historical effectiveness of nuclear coercion and not practiced it. 141 Their degree of attention to history is unclear but the claim that Kargil was a case of nuclear coercion is hard to deny. My interview data shows, consistently with the psychological learning model, that Musharraf continued to believe in the power of nuclear coercion through the Summer of 2001 and likely continued to believe this until he experienced fear of nuclear escalation in May 2002. The rational learning model explains neither his continued belief in the power of nuclear coercion nor the impact of fear on subsequent Pakistani policy. 142 Chapter 5: Lyndon Johnson’s Lessons from Cuba 5.1. Lyndon Johnson's Lessons from Cuba When a President makes a decision, he seeks all the information he can get. At the same time, he cannot separate himself from his own experience and memory. This is especially true when his decisions involve the lives of men and the safety of the nation. Lyndon B. Johnson, 1971280 Chapters 3 and 4 argued that learning about the coercive power of nuclear weapons explains the statistical effect of experience with nuclear weapons on state conflict propensity. Moreover, the psychological learning model explains most of the evidence in the Soviet case better than the rational choice alternative. The psychological model is also consistent with the Pakistani case, but unavailable data prohibits a competitive test with the rational choice alternative. This chapter tests a related but distinct part of the nuclear learning models. While the previous two chapters addressed the experience of fear during a nuclear crisis, this chapter addresses the effect of experience with nuclear weapons after those who experience that fear depart from office. Khrushchev and Musharraf, like all leaders, had limited terms in office. To explain the effect of experience with nuclear weapons, we need to know whether their successors learned similar lessons during the crises or wars. If Brezhnev did not experience fear of nuclear escalation or otherwise learn about the dangers of nuclear weapons during the Cuban Missile Crisis or elsewhere, the psychological nuclear 280 Lyndon B. Johnson, The Vantage Point: Perspectives of the Presidency, 1963-69, 1971. New York: Holt, Rinehart and Winston, p.46 143 learning model may not explain the effect of experience in the Soviet case. Similarly, if Kayani did not learn about these dangers in 2002, Pakistani policy toward India today may not be explained by the learning models and might be more destabilising than Soviet policy was to the United States forty years earlier. Data on the foreign policy beliefs of Brezhnev and Kayani that would be sufficient to test the nuclear learning models is however not available. We cannot know whether Brezhnev and Kayani experienced fear of nuclear escalation and thus cannot know whether the nuclear learning models explain the longer term effect of experience with nuclear weapons. To address this problem, in this chapter I address the sources of cognitive accessibility to gain inferential leverage on the question of the origins of these decision-makers foreign policy beliefs. Data that would address whether Brezhnev and Kayani learned their foreign policy beliefs from cognitively accessible experiences or a more systematic assessment of the historical record would provide a plausibility probe of the nuclear learning models. It would suggest whether they likely learned through fear about the dangers of nuclear coercion from their predecessor’s crisis or war, or alternatively learned perhaps similar lessons from other less cognitively accessible events. Another possibility would be that these leaders did not learn anything about nuclear coercion from their predecessors crises and wars and adopted nuclear policies that they would have pursued if the war or crisis did not occur. But the available data on the beliefs of Brezhnev and Kayani also prohibit an answer to this question. In this chapter, I aim to establish inferential leverage over the question of the effect of experience with nuclear weapons over the longer term by using a different value of the independent variable of personal involvement in a nuclear crisis or war to examine 144 whether leaders generally learn about foreign policy from (1) their predecessors and an examination of the historical record or (2) from their own cognitively accessible personal experiences. Perhaps leaders who do not experience fear of nuclear escalation in close proximity to decision-making during a nuclear crisis might nonetheless learn of the dangers of nuclear coercion from their predecessors. But this raises a general question: do leaders learn from their predecessors, or to they tend to learn most from their own cognitively accessible experiences? Do leaders who were not closely involved in their predecessors’ war or crisis, and who therefore were unlikely to experience fear during this time, learn about such dangers of nuclear coercion? Another way of expressing this question is whether leaders who found a different crisis or war than their predecessors’ cognitively accessible bring its own lessons to current crises rather than those that their predecessor brought to it? These questions address the origins of beliefs. Rational choice models would not lead us to expect cognitive accessibility to be a strong predictor of the type of events that leaders learn from. They would predict that decision-makers would systematically scan the external environment and learn from all or most of the relevant cues. For example, because the Cuban Missile crisis might have realistically escalated to nuclear war and was a critically important event in American foreign policy during the Kennedy era, these models would lead us to expect Kennedy’s successor, Lyndon Johnson, to reflect on its lessons when assessing policies to the Soviet Union.281 Rational choice models might also expect leaders to learn from their predecessors. Presidential decision-making involves many decisions in short periods of time, and there would be much utility in being 281 Rational choice models would not predict any specific lesson from Cuba that Johnson would have applied to different cases, but rather expect him to have considered the crisis when addressing different foreign policies that he might have believed the crisis was relevant to. 145 attentive to – and learning from – one’s predecessor who probably experienced similar challenges. Psychological models have a different prediction. They would only lead us to expect decision-makers to learn from their predecessor’s important crises and wars and the historical record if these experiences were cognitively accessible. Moreover, they would lead us to expect decision-makers to apply lessons that were learned from cognitively accessible wars to different cases without a systematic consideration of the similarity of the two cases. Extensive data on the beliefs of President Lyndon Johnson allows a test of the origins of his beliefs that addresses these questions. More specifically, we can assess the degree to which any lessons from the Cuban Missile crisis influenced his decisions to escalate the Vietnam war. We can also assess the role of cognitively accessible experiences in his foreign policy beliefs. Psychological models would predict him to have been influenced only by cognitively accessible experiences. Johnson was marginalised from the Cuban Missile Crisis debates that were dominated by Kennedy and his brother but was heavily influenced by the lessons of Munich. Psychological models would therefore lead us to expect him to apply the lessons of Munich to Vietnam and not learn much from the Cuban Missile crisis. Rational learning models would not predict that the experiences of Munich would loom so large in Johnson’s beliefs about Vietnam. They would also expect a consideration of the validity of other models in addressing such important issues, and perhaps expect him to reflect on the Cuban Missile Crisis. An analysis of Johnson’s beliefs allows me to make two propositions that should be tested if evidence on Brezhnev and Kayani becomes available. Firstly, I propose that leaders will only learn similar lessons about the danger of nuclear escalation from their 146 predecessors if they also experienced fear of nuclear escalation, presumably during the crisis or war. More generally, I predict that they will, on average, only learn similar lessons if similar events were similarly cognitively accessible.. Short of their own experience of fear during their predecessor’s crisis, I predict that leaders will not learn about the dangers of nuclear coercion from their predecessors.282 Secondly, successors in new nuclear powers who did not experience fear of nuclear will make policies that are based not on their predecessor’s lessons but on their own cognitively accessible experiences.283 I examine Lyndon Johnson’s foreign policy beliefs and his lessons from the Cuban Missile crisis and their influence on his escalation of the war in Vietnam as a plausibility probe of the above hypotheses. I address which lessons caused what aspects of Johnson’s foreign policies. I do not address the Soviet and Pakistani cases. However if Johnson formulated his foreign policy through lessons learned from experiences which were cognitively accessible to him and not from his successor or history, we should expect Brezhnev and Kayani to have learned from similar sources. This suggests that if the Cuban Missile and ten month nuclear crises were cognitively accessible to Brezhnev and Kayani the psychological nuclear learning model might explain the effect of experience with nuclear weapons in the Soviet and Pakistani cases. 282 Decision-makers who experience fear during crises or wars might make institutional changes or international commitments based on their lessons that commit their successors to particular policies. But such institutional lock-in that constrains subsequent leaders is unlikely in authoritarian states such as the Soviet Union, Pakistan, North Korea and Iran. Democratic states have political institutions that constrain elite behaviour and which are largely beyond the control or manipulation of the ruling elite. In these cases, policies that lock particular policy trajectories in might not be able to be overcome without challenging the rule of law. But authoritarian states exhibit weaker political institutions. Elites are unlikely to be constrained by their predecessors in such weakly institutionalised environments. Thus Janice Stein concluded her study of Gorbachev’s learning in the late 1980s that the evidence suggests institutionalised change in the Soviet general staff was not a necessary condition for policy change.” Stein, Learning by Doing, p. 244 283In the concluding chapter, I briefly address the implications of this and offer counterintuitive hypotheses on inter-generational nuclear learning. 147 TABLE 3: KENNEDY’S AND JOHNSON’S BELIEFS Experienced WWII at formative time/ personally involved in relevant decisions? Experienced Bay of Pigs crisis at formative stage/personally involved in relevant decisions? Lyndon B. Johnson Yes (learn lessons of Munich) No (no learning) John F. Kennedy Yes (learn lessons of Munich) Yes (learning) As table 3 demonstrates, I argue that sustained personal involvement in an event or the experience of it at a formative time in the development of one’s own political beliefs are better predictors of whether one will learn from an event than if one does not. Both Kennedy and Johnson learned the lessons of Munich from World War Two; it was very cognitively accessible to them and occurred when they were developing their formative political beliefs. But Kennedy’s experience at the Bay of Pigs taught him of the danger of trusting the preferences of his military generals and hawkish military preferences more generally. Johnson, on the other hand, was isolated from these early decisions – his involvement in the Missile Crisis itself was also minimal – so his lessons from Munich were not qualified by Bay of Pigs lessons and strongly influenced his Vietnam decisions. Had Johnson been personally involved in the Bay of Pigs or Missile Crisis, he may have come to the Vietnamese insurgency in late 1963 with more moderate policies. I suggest below that there was a strong probability that Kennedy would not have escalated the Vietnam war to the extent that Johnson did if he was not assassinated because of lessons that he learned from the failed Bay of Pigs invasion. Below I briefly elaborate on the importance of the origins of Lyndon Johnson’s beliefs for understanding 148 the causes of the perpetuation of the Cold War and the role of nuclear weapons in it before examining how they speak to the nuclear learning models. Finally, it is worth reflecting on the fact that Johnson’s lessons from Munich and the issues at stake in the Vietnam war have little to do with nuclear coercion. But as I have argued above, this chapter tests a more general learning mechanism that should operate in contexts beyond nuclear coercion. . 5.2. Lyndon Johnson and the Cold War The Cold War exhibited a fundamental transformation after the Cuban Missile Crisis. Before the end of October 1962, Khrushchev was prepared to risk nuclear war on at least three occasions to achieve his objectives in Berlin. Stalin believed that nuclear war with Truman and Eisenhower was inevitable, and Eisenhower and Kennedy also expected that nuclear war, or at least a nuclear crisis, was likely before October 1962. After the crisis, Khrushchev’s risk aversion moderated his policies in Berlin and elsewhere and the intensity of Cold war hostilities declined. This moderation persisted into the post- Khrushchev era; despite minor war scares in 1973 and 1981, nuclear war was much more likely to have occurred between 1945 and 1962, and especially after 1958 when the Soviet Union developed nuclear missiles capable of targeting the United States, than between 1963 and 1991. The superpower wars and crises in the first era occurred in the strategically important areas of Western Europe (Berlin, Turkey, Greece) and East Asia (Korea and Taiwan). Those in the second occurred in the mostly peripheral areas of Southeast Asia (Vietnam), Central Asia (Afghanistan), Africa and the Caribbean. 149 The first major use of force in the periphery was Lyndon Johnson’s 1965 escalation of the Vietnam war. This suggests an important question. Did Johnson learn lessons from the Cuban Missile Crisis that caused him to escalate the Vietnam war? Did he learn from it that fighting in Vietnam was a more effective means to demonstrate credibility in Western Europe than fighting there? Was he emboldened to escalate the Vietnam war because he learned from the Cuban Missile crisis that nuclear war is unlikely? More generally, did Johnson learn anything about nuclear weapons, nuclear coercion or the use of force during the Cuban Missile crisis that influenced his subsequent foreign policy or Vietnam decisions? In this chapter, I show that Johnson likely learned nothing from the Cuban Missile crisis. At most, he may have learned from it the importance of establishing resolve and not permitting adversaries to challenge the status quo as Khrushchev did at Cuba, but he had already learned this from Munich and the Korean war. There is no evidence that Johnson learned this. He was isolated from the Executive Committee deliberations, and the extensive archival holdings and secondary studies contain no evidence that he learned anything from the crisis. He escalated the Vietnam war because of lessons he learned during the second World War. Johnson’s long held beliefs about international politics, not the lessons of the Cuban Missile Crisis or a systematic analysis of the dynamics associated with the Vietnam war, caused Johnson to escalate the Vietnam war in 1965. Psychological models that emphasise cognitive accessibility and that learning rarely involves comprehensive scanning and assessing of environmental stimuli explain his beliefs. There is little evidence that Johnson’s lessons from the Cuban Missile crisis and Vietnam decisions are consistent with the rational learning model. The American escalation of the Vietnam war within three years of the Cuban Missile crisis 150 had more to do with Kennedy's assassination – and the collapse of the Diem government in South Vietnam – than Johnson’s lessons from the Cuban Missile crisis. The Vietnam war was not inevitable in the aftermath of the Cuban Missile crisis; there is no evidence that Johnson learned anything about nuclear coercion or the use of force from the crisis. Indeed, although he was Vice President, he was relatively uninvolved in the major decisions in that and other crises during Kennedy's Presidency. Rather, Kennedy's assassination brought a man with different salient experiences to the Presidency. Both Kennedy and Johnson, as well as Eisenhower and Nixon, were members of the World War II generation where the experience of Hitler and Munich was cognitively accessible. But while Kennedy's experience at the Bay of Pigs taught him a caution toward the preferences of his military advisers that would moderate his Vietnam decisions, Johnson was not closely involved with the Bay of Pigs mission. Below, I summarise the evolution of Johnson’s beliefs before addressing their origins and evolution with reference to the nuclear learning models in greater detail. Johnson learned the classic lessons of Munich during World War II and applied them to most important foreign policy decisions, including the civil war in Vietnam, twenty years later. These lessons included the belief that it is better to defeat foreign aggression earlier rather than paying more to defeat it later, because appeasing Hitler-like aggression only encourages more of the same. Fighting to establish American resolve, demonstrating the value of American alliance commitments and avoid falling dominoes, according to this worldview, serves American interests more than letting the dominoes fall, even if there is good evidence that they inevitably will. The documentary record on Johnson’s beliefs about nuclear weapons and nuclear coercion is still sparse but he seems to have believed 151 that nuclear superiority – having more nuclear weapons that the Soviet Union – was as important as superiority in conventional armaments. Hitler's are rare, dominoes rarely fall, and American intervention would probably not have stopped German aggression. But the cognitive accessibility of World War II exerted a powerful and constant influence on his thinking throughout the remainder of his life. Rational learning models would predict Johnson to have considered the lessons of other relevant cases in assessing Vietnam and other policies. They would also lead us to expect him to have assessed the frequency of Hitler's in international politics and examine whether the Hitler analogy was the most appropriate model to apply to the insurgency in Vietnam. But as Yuen Foong Khong has shown, he did not reason this way.284 He maintained his belief that the insurgency and civil unrest throughout Vietnam reflected Hanoi's directed and Chinese and Russian sponsored aggression, and neglected the desire of many Vietnamese to be independent of French and American control. From his first day as President Johnson was determined to not allow South Vietnam to fall to the communist North, call into question existing international treaty obligations or otherwise undermine American credibility. He believed that the fall of Saigon would lead to a third world war just as Hitler's unchecked aggression in Europe caused World War II. These commitments reflected the cognitive accessibility of particular aspects of World War II and Johnson's refusal to seriously consider the validity of other models. Johnson’s beliefs are better explained by the availability heuristic and the cognitive accessibility of the lessons of Munich than expected utility assumptions of rational choice models. Rational choice models might also predict Johnson to have considered Kennedy's lessons from his Bay of Pigs disaster when contemplating escalating the Vietnam war. Both cases 284Khong, Analogies at War 152 involved decisions to deploy American forces to undermine weak communist states that were perceived to be deeply hostile to American interests. Yet while the American failure at the Bay of Pigs had a strong influence on Kennedy, Johnson was marginalised from those decisions such that it did not exert much influence on him. There is little evidence that Johnson considered Kennedy’s lessons from the Bay of Pigs when thinking about Vietnam or other decisions. One might also expect that Johnson's commitment to defending the United States from foreign aggression would render Kennedy's rapprochement with the Soviet Union more important than wars in the third world. But the cognitive accessibility of Munich and the deterioration of South Vietnam in late 1963 made him neglect the momentum from the Cuban Missile Crisis and LTBT toward detente in favour of authorising over half a million American soldiers to defend against perceived Chinese and Soviet aggression in Southeast Asia and an expected third world war. The availability heuristic explains Johnson's learning about the use of force better than rational choice alternatives. Rational choice assumptions of carefully assessing the challenges that other data pose to his prior beliefs and them in the face of new information are not borne out in Johnson’s case. The following sections address his beliefs about nuclear weapons and the use of force from his Congressman and Senator days from 1945 to 1960 to the Vice-Presidential and early Presidential periods from 1961 through 1965.285 285Johnson was, of course, a congressman from 1937, but because nuclear weapons were first used in war 8 years later, i start at 1945. There is little evidence that Johnson thought about nuclear weapons until the Korean war in 1950. 153 5.3. Johnson's Beliefs: Congressman and Senator, 1945-1960 Johnson's beliefs and behaviour throughout this period were highly consistent with the central elements of the Munich analogy. There is much evidence that he believed in the central elements of the Munich analogy from his formative political years and little evidence that he gave serious consideration to alternative models. I aim in this section to show that Johnson had developed strong beliefs about Munich long before the 1960s. At several points I show that his statements to his colleagues in Congress and the Senate closely resemble those to his biographer decades later. Although he told isolationist Texas constituencies that American involvement in World War II was unlikely, he supported and voted for Roosevelt's appeals to expand American military power as Hitler conquered Denmark, Norway, Holland, Belgium and France. Johnson worried that Great Britain might also be defeated and that this would leave all of Europe open to German domination and render eastern approaches to the United States undefendable.286 In the Summer of 1941 he thought American involvement in the war was essential; like 85% of the American public, he viewed British defeats in the Balkans, North Africa, and the Atlantic as compelling American participation. Unlike most Americans however, he thought it imperative to enter the conflict as soon as possible or before it was too late to aid Britain. While some American's saw Hitler's invasion of the Soviet Union as easing the need for American involvement, Johnson called it further evidence of America's need to combat Nazi plans for world conquest.287 His account to his biographer Doris Kearns in 1970 summarises his lessons from World War Two; before the war he believed in the 286 Robert Dallek, Lone Star Rising: Lyndon Johnson and His Times, 1908-1960, 1991. New York: Oxford University Press, p.197 287Dallek, Lone Star Rising, p.225, p.228 154 importance of credibility, but his reflection upon signing the Ludlow Resolution seemed to strongly reinforce his belief in the central elements of the Munich analogy. What I learned as a boy in my teens and in college about World War I was that it was our lack of strength and failure to show stamina that got us into that war. I was taught that the Kaiser never would have made his moves if he hadn't been able to count Uncle Sam out because he believed we'd never come in. Then I was taught in Congress and in committees on defense preparedness and by FDR that we in Congress were constantly telegraphing the wrong messages to Hitler and the Japanese – that the Wheelers, the Lindberghs, the La Follettes and the American Firsters were letting Hitler know he could move without worrying about Uncle Sam. I remember those days in Congress. The liberal debate almost got to me. I even signed a petition for something called the Ludlow Resolution, calling for a popular vote before a war. But then I came to my senses and recognized that Hitler could take over America while we were holding our election, and I felt so silly I ran down and took my name off. I firmly believe we wouldn't have been involved in World War II if it hadn't been for all the vacillation.288 He consistently backed strong Truman, Eisenhower and Kennedy actions against Communism throughout the Cold War and often supported greater military spending and defence efforts.289 In May 1947, before the first Berlin blockade, Soviet atomic bomb test and communist takeover in China increased American suspicions of Stalin's intentions and exacerbated Cold War tensions, Johnson believed in the importance of displaying resolve through the use of force. He told his Congressional colleagues that: 288 Doris Kearns Goodwin, Lyndon Johnson and the American Dream, 1991. New York: St Martin's, pp.329-331. One might argue that Johnson’s lessons from Munich are an example of learning from something other than direct experience. The impact of the Ludlow resolution and Johnson’s personal role in it, however, suggest that his lessons from Munich did stem from this personal experience. 289 Dallek, Lone Star Rising, p.245 155 Whether communist or fascist or simply a pistol-packing racketeer, the one thing a bully understands is force and the one thing he fears is courage. I want peace. But human experience teaches me that if I let a bully of my community make me travel the back streets to avoid a fight, I merely postpone the evil day. Soon he will chase me out of my house.290 He would repeat this claim throughout his life, and repeated to Kearns in 1970 that: if you say to him at the start, 'Now, just hold on, wait a minute,' then he'll know he's dealing with a man of courage, someone who will stand up to him. And only then can you get along and find some peace again.291 He informed his Congressional colleagues of his belief that an American use of force could have deterred German and Japanese aggression in the early 1940s. The sentiment was very similar to his statements to Kearns three decades later: We have fought two world wars because of our failure to take a position in time. When the first war began, Germany did not believe we would fight....Thus the Kaiser was led to believe we were complacent and lacked courage. Unrestricted submarine warfare began and we went to war. During earlier stages of World War II, Roosevelt enunciated the doctrine of quarantining aggressors. But there were protests...the tactics of these ostriches and their fellow travellers encouraged indeed if they did not induce Hitler to ignore us and the Japs to attack us...The Truman doctrine will be costly, [but if we] sit back with our arms around our money bags...I have no doubt that the smoldering fires will burst into flame and consume us – dollars and all.292 Johnson did not consider other explanations of German and Japanese aggression that are 290May 7 1947 Congressional record, cited in Kearns, Lyndon Johnson, p.95 291Kearns, Lyndon Johnson, p.95 292Kearns, Lyndon Johnson, pp.329-331; Johnson, Vantage Point, pp. 46-47 156 inconsistent with the Munich model. The Japanese leadership, for example, fought not because of American inaction but American sanctions that were perceived as too costly for Japan to accept and the belief that the United States would fight a limited war commensurate with Japanese aims. Hitler was unlikely to have influenced by American public opinion and probably always intended to fight with the United States but wanted to wait until his control of Europe was consolidated. Johnson could not know that his understanding of history was flawed, but the salience of Munich and the widely drawn lessons from it caused him to ignore other critical aspects of these cases and believe that the Munich model accurately explained them. Like Khrushchev and Musharraf, he ignored other explanations and variables and assumed that this cognitively accessible case was the best model for other ones that varied on multiple dimensions. Johnson maintained these beliefs throughout his life. His statement in his 1971 autobiography closely resembles his earlier remarks to Congress over thirty years before: I felt strongly that World War II might have been avoided if the United States in the 1930s had not given such an uncertain signal of its likely response to aggression in Europe and Asia...With the end of fighting in 1945 we almost repeated the errors of the past by dismantling too quickly the huge military force we had developed. Our haste provided an irresistible temptation to Stalin, who moved to consolidate control over Eastern Europe by 1948 and simultaneously exerted increased pressure on Western Europe.293 Shortly after the communist takeover of Czechoslovakia in March 1948, Johnson elaborated his beliefs about falling dominoes to Congress: If Italy is lost, Greece will be cut off and Turkey isolated. The bell 293Johnson, Vantage Point, pp.46-47 157 has tolled for Rumania, Yugoslavia, Czechoslovakia. It is tolling for Finland, Norway, Sweden. Each toll of the bell brings closer the day when it could toll for you and for me.294 States usually balance against rather than bandwagon with threatening rising powers; The falling domino model is usually incorrect.295 Johnson did not have access to this research, but the cognitive accessibility of Munich caused him to believe that dominoes fall more often than they usually do. He ignored many variables or alternative explanations that would have seriously challenged his beliefs and held them with a high degree of certainty. There is much evidence of his belief in the Munich analogy and no evidence that he seriously considered qualifications or corrections to the model. The fall of China to communism, for example, did not cause other states to succumb to similar forces, and American losses in the Korean war did not cause other American allies to fall to communist forces. But Johnson did not interpret these cases in this way. World War II and the falling domino theory was the decisive lesson that resonated in his mind whenever he thought about international affairs.296 From the late 1940s through his Presidency he described the lessons of World War II in language strikingly parallel to his explanation of the reasons that led him into his own war in Asia: From the experience of World War Two I learned that war comes about from two things – by a lust for power on the part of a few evil leaders and by a weakness on the part of the people whose love for peace too often displays a lack of courage that serves as an 294 Congressional record March 9 1948, cited in Kearns, Lyndon Johnson, p.97 295 See Colin Elman and John Vasquez (eds.), Realism and the Balance of Power: A New Debate, 2002. New York: Prentice Hall; Mercer, Reputation and International Politics; Press, Calculating Credibility 296 Kearns, Lyndon Johnson, p.95 158 open invitation to all the aggressors of the world.297 Staff memos to Johnson in 1948 reveal him to have made many enquiries on issues including manpower strength, stockpiling and particularly airpower.298 In April he wrote to President Truman that his “shoulder always has been and always will be at the wheels of preparedness.”299 There were advantages to Texas and himself from this position, but it also characterised Johnson's genuine anxiety about the Communist threat.”300 Preparedness was also a central theme of the speech that launched his Senate campaign in May 1948. Johnson viewed the North Korean invasion of South Korea as a Russian sponsored plan for world domination. North Korean leader Kim Il Sung had to exert a great deal of pressure on Stalin for the Russian's permission to invade the South, and stipulated his support on the provision of Chinese, rather than Soviet, troops. Johnson could not know this, but while ample evidence was available to him that suggested that Stalin was no Hitler – it was not clear that Stalin desired to control Europe he privately described the North Korean attack as part of a Soviet design for world domination.301 He wrote that “the survival of the ruling clique in Russia depends upon conflict” and that “Russia needs...and wants war.”302 After 200000 Chinese troops crossed the Yalu river and routed American forces back below the 38th parallel on November 26, Johnson proposed a full mobilisation of the nation's manpower and economic resources to avoid 297 op. Cit., 298 See Memos to LBJ in Pre-Presidential Papers – House of Representatives Papers, Box 329, Memos to Johnson Folder, Lyndon Baines Johnson Presidential Library (hereafter referred to as LBJL). 299 Johnson to Truman, April 5 1948, Pre-Presidential Papers – Senate Papers, Box 345, Preliminary Organization, Preparedness Subcomittee folder, LBJL. 300Dallek, Lone Star Rising, p.294 301Johnson to Si Cassidy, 28 June 1950, LBJL, cited in Dallek, Lone Star Rising, pp.383-4. 302op. cit. 159 “being engulfed by a gathering darkness.”303 In a typical formulation, he said that “it is foolish to talk of avoiding war; we are already in a war – a major war. The war in Korea is a war of Soviet Russia.” Two decades later he wrote in similar language that I could never forget the withdrawal of our forces from South Korea and then our immediate reaction to the Communist aggression of June 1950...I could see us repeating the same sharp reversal once again in Asia, or elsewhere – but this time in a nuclear world with all the dangers and possible horrors that go with it. None of the very few who opposed the (Vietnam) decision gave me facts or arguments that broke or even weakened this chain of conclusions.304 Of particular importance here is that Johnson was exposed to many who disagreed with him – such as George Ball – but these arguments did not influence his high degree of faith in his own beliefs. I have thus far shown that Johnson's beliefs before 1950 were consistent with his beliefs after 1960. In this section I show that these beliefs remained constant through the 1950s. . He scribbled during a meeting that: World looks upon us as strong power – we must not appear to be afraid – must appear to be confident – know we are going to win – totalitarian states appear strong.305 In 1957 he wrote to a constituent that “we are now in a real, hot war.” He agreed with the writer’s similarly pessimistic conclusion and added that “we are in trouble – very grave 303Dallek, Lone Star Rising, p.389 304Johnson, Vantage Point, p.152 305 Pre-Presidential Handwriting File, LBJ-Notes, White House 1953-1959, LBJL 160 trouble – and only intelligent, all-out effort will save us.”306 Fourteen months later he wrote that: The provision of an adequate defence for our Nation is the paramount problem which gives me greatest concern. It has been my belief, which I have expressed on many occasions, that we could proceed at a faster rate and provide even more defence than we are now providing.307 In February 1960 he wrote that: I certainly agree with you that a great deal of money is being spent on the arms race with Russia. However, I certainly don't believe we ought to let our guard down as long as Communism maintains its policy of world conquest and exploitation of freedom loving nations.308 Eleven months later he wrote that: I believe that the greatest challenge facing the new Administration is the Cuban situation because that does involve the prospects of halting the spread of communism in this hemisphere.309 On April 2 1954 Eisenhower had Dulles ask congressional leaders of both parties for discretionary authority to use American air and sea power against Communist aggression in Southeast Asia. Johnson asked which allies supported the intervention, and Dulles had 306 Clyde Emery, December 13 1957 letter to LBJ, LBJ response to Emery, LBJA Subject File, Armed Services – Preparedness; Production – Weapons, Box 64, LBJL 307 LBJ to Harold J. Rahm, January 26 1959, 1959 Subject Files, Armed Services, Production – Weapons (Dr. Clayton), Senate Papers, Box 649, LBJL 308 LBJ to Mrs Evelyn Morris, Feb 20 1960, Senate Papers, 1949-61, 1960, Subject Files, Foreign Relations – Cuba, January-March, LBJL 309 LBJ to Claude C. Wild, December 17 1960, Senate Papers 1960, Subject Files, Foreign Relations – Cuba, July - , Box 769, LBJL 161 to reply that none did. The congressmen, led by Senator Richard Russell and supported by Johnson, rejected the idea unless America's allies joined in and the French promised independence to Vietnam. Johnson argued that the Korean war showed that the United States should avoid unilateral involvement in Asian wars, and he is often credited with helping prevent American intervention in Vietnam at this time. Does his rejection of an American commitment to salvage the almost defeated French garrison at Dienbenphu reflect more contingent beliefs about the Munich analogy and the domino theory?310 His hesitation to resist saving an ally from pending defeat in Vietnam in the 1950s might suggest that he did not always believe that allowing one’s allies to fall would encourage future challenges. But Johnson was already sympathetic to intervention and apparently asked his question in a “pro forma” way that was “exactly what the Eisenhower administration expected of him.”311 In a letter to a constituent on May 18, Johnson argued that it would be a mistake to “fight alone,” implying that he opposed unilateral US intervention in Vietnam without the backing of allies.312 But he saw a need to stand firm, and wrote to another constituent that “the real question involved here, of course, is where, when and how to make a stand.”313 He repeatedly told constituents, the majority of whom opposed intervention, that America had been “caught bluffing,” and that “the only language [communists] 310 Dallek, Lone Star Rising, p.444; Fredrik Logevall, Choosing War: The Lost Chance for Peace and the Escalation of the War in Vietnam, 2001. Berkeley: University of California Press, p.22; Tom Wicker, JFK and LBJ: The Influence of Personality upon Politics, 1991. Chicago: Ivan R. Dee, p.202; Khong, Analogies at War, p.73, 76 311John Prados, The Hidden History of the Vietnam War, 1995. Chicago: Ivan R. Dee, pp. 13-16 312 LBJ to R. L. Lindley, May 18, 1954, Pre-Presidential Papers – Senate Papers, Box 1194, 1954 Case and Project Files, Indo-China, 3 of 4 folder, LBJL. 313LBJ to Paul D. Balbin, May 19 1954, op. Cit. 162 understand and respect is the language of strength.”314 As Prados notes, in the absence of diaries, notes, or interviews, Johnson's newsletters to constituents are the most “authoritative sources” on his position on Vietnam, and they “consistently and repeatedly made the case for U.S. intervention.”315 In a letter to a friend on April 23, he wrote that the Vietnam situation was a “very grave threat that we must resist.”316 After the fall of Dienbenphu he wrote that intervention would be “a tragedy of great magnitude,” but usually tended to speak to the importance of achieving credibility through resolve and the dangers of appeasement.317 Despite some misgivings from many Democrats in 1958, he supported Eisenhower's intervention in Lebanon and spoke on the Senate floor that the United States would make it clear to the aggressors that this country is determined to maintain freedom in this world, whatever the cost. While Kennedy wrote to a constituent that this was not a case of outside aggression, Johnson wrote that “all of the evidence seems to indicate strongly that external influences, as well as military assistance, provoked the uprisings in Lebanon.318 In the next section I show that Johnson maintained these beliefs about credibility, the Munich analogy and falling throughout his Vice-Presidency and Presidency in the 1960s. He did not learn anything about foreign policy or nuclear weapons from the Bay of Pigs or Cuban Missile Crisis. 314LBJ to Phil Hopkins, May 13, 1954; LBJ to Omar N. Braddock, May 11, 1954; LBJ to Gus B. Michel, May 5, 1954 and LBJ to Jack Burrus, May 8, 1954, all in Pre-Presidential Papers – Senate Papers, Box 1194, 1954 Case and Project Files, Indo-China, 2 of 4, folder, LJBL. 315John Prados, “Assessing Dienbienphu,” in Mark Atwood Lawrence and Fredrik Logevall, The First Vietnam War: Colonial Conflict and Cold War Crisis, 2007. Cambridge: Harvard University Press, p.234 316LBJ to Claude E. Carter, April 23, 1954, Pre-Presidential Papers-Senate Papers, Box 1194, 1954 Case and Project Files, Indo-China, 1 of 4, folder, LBJL. 317See footnote 26 above. 318LBJ to Henry Grawunder, Sr., July 18, 1958, Pre-Presidential Papers – Senate Papers, Box 602, Foreign Relations, Middle East (Situation), 3 of 3 folder, LBJL. 163 5.4. Johnson's Beliefs: Vice President, 1961- November 1963 Johnson's few serious forays into foreign policy as Vice President reveal beliefs that are clear and consistent with his earlier ones. Kennedy sent him to several Southeast Asian states in May 1961 to demonstrate support for South Vietnamese President Diem. Upon returning from Saigon on May 23, Johnson pressed the case for a greater commitment to the defence of South Vietnam to the President. He stated that “I cannot stress too strongly the extreme importance of following up this mission with other measures, other actions, and other efforts.” Johnson was under no pressure to write this letter to Kennedy, and it was consistent with his beliefs about containing communist aggression. The trip had sharpened and deepened his basic convictions. He argued, consistently with his beliefs about the causes of the expansion of communism and the dangers of appeasement, that “the battle against Communism must be joined in Southeast Asia with strength and determination...or the United States, inevitably, must surrender the Pacific and take up our defences on our own shores.”319 He elaborated that others would infer from the fall of South Vietnam that “we don't live up to our treaties and don't stand by our friends.”320 He said similar words to the House Foreign Affairs Committee upon his return. He wrote to Kennedy upon returning from Berlin in August 1961 amidst the second Berlin crisis that “if we failed to rise to the level of these sombre events, all would be lost, for there would be no one who could remove the sense of failure created by our 319 Robert Dallek, Flawed Giant: Lyndon Johnson and His Times, 1961-1973, 1998. New York, Oxford University Press, p.17 320 LBJ to JFK, “Mission to Southeast Asia, India and Pakistan,” May 23, 1961, VPSF, Box 1, Vice-Presidential Travel, Vice President's Visit to Southeast Asia, May 9-24, 1961 (I) [1 of 2] folder, LBJL: 1-6. 164 default.”321 In a statement about the costs of inaction quite similar to his beliefs about inaction in Vietnam, he stated that: It is not that doing these things (selective ban on the issuance of Temporary travel documents to representatives of the Ulbricht regime) is terribly important, but not doing them would be, because our allies – particularly the Germans – would not understand our failure to take these steps.322 Johnson was uninvolved in the Bay of Pigs decisions that occurred early in Kennedy’s first term. He was mostly uninvolved in the Cuban Missile crisis, and only sat in on the Ex Comm meetings on the one or two rare occasions that Kennedy could not attend. He favoured an unannounced air strike rather than the blockade, although by the end of the Crisis he had come around to the blockade camp. McCone wrote a memo on October 22 1962 stating that “the Vice President finally agreed reluctantly but only after learning...of the support indicated by General Eisenhower.”323 There is no evidence in the extensive primary and secondary sources that Johnson learned anything about nuclear coercion or the use of force from the crisis. In the next section I explore Johnson’s beliefs about nuclear weapons to address whether he learned anything about them from the crisis. 321 Memorandum, Vice President to President, August 21 1961, Vice Presidential Travel, Berlin, Germany, Berlin Papers for the Vice President, p.2 VP Security File, Box 2, LBJL. See also Report by VP Johnson on his visit to Germany, August 19-20, 1961, ibid. 322Memorandum, Vice President to President, August 21 1961, Vice Presidential Travel, Berlin, Germany, Berlin Papers for the Vice President, p.5 VP Security File, Box 2, LBJL, underlining in original 323 Mary S. McAuliffe (ed.), CIA Documents on the Cuban Missile Crisis, 1992. Diane Pub. Co., p.245, McCone, October 22 1962 165 5.4.1. Johnson's Beliefs on Nuclear Weapons There is only scattered evidence of Johnson's beliefs about nuclear weapons and nuclear coercion, and most of it is his letters to constituents. He seems to have not thought much about them. In December 1950 he published an article in Preview Magazine where he stated that “we will have to spend more time and effort...on developing...guided missiles...(which) might become powerful deterrents against aggressors.”324 That same month he wrote that the danger of using nuclear threats to force a withdrawal of Communist forces from Korea was that the Soviet Union might respond with a first nuclear strike.325 In September 1961 he wrote that he did not “favor a nuclear war regardless of what happens to the people of the world” but did “favour our building our military strength to deter war.”326 These statements suggest an awareness of the transformative effects of nuclear weapons on international conflict and a belief in the importance of using nuclear weapons to deter war. But much more numerous are Johnson's claims about the importance of having more nuclear weapons than the Soviet Union and statements that the use of conventional force to display resolve and prevent falling dominoes remains relevant in the nuclear age. Johnson seems to have maintained his belief in the falling domino model and the danger of Soviet nuclear coercion and viewed nuclear superiority as a means to deter aggression. 324 Lyndon B. Johnson, “We Are Ahead of Russia – We Can Stay Ahead,” Preview Magazine, December 1950, Houston, Texas, pp. 12, 40. 325 LBJ To Shelburne H. Glover, December 5, 1950, Leg. Atomic Bomb (General), Senate Papers, 1949-1961, Legislative Files, 1950-52, Box 227, LBJL. See also LBJ to J.W. Thomas, December 5 1950, ibid. 326 LBJ to Walter S. Cahall, September 27, 1961, Armed Services, Disarmament, VP Papers, 1961 Subject File, Box 58, LBJL. See also LBJ To Hy Buschelmann and E. J. Steck, August 4 1959, in 1959 Subject Files, Armed Services, Production – Weapons, 1959 Subject Files, Senate Files, Box 648, LBJL. 166 In October 1961 he wrote that “we cannot allow an enemy to build up atomic superiority.”327 About five weeks later he claimed that: the U.S.S.R. has repeatedly made clear its aggressive intent against the free world, and their recent tests have undoubtedly increased their capability to carry out their hostile plans. If we refrain from all testing, we automatically give the Russians a weapons system advantage which we cannot afford to let them have.328 One week later he wrote that We cannot afford to rely upon rational behaviour on the part of an aggressive nation which wants to control the world. Consequently, we must be strong, while striving toward mutual disarmament and world peace.329 That month he insisted that American unilateral nuclear arms reductions would be dangerous without Soviet reciprocity, and repeated his belief that “we cannot allow an enemy to build up nuclear superiority” numerous times.330 In March 1962, he wrote that “we cannot permit the security of this nation to be jeopardized through failure to maintain strength achieved through further development and testing.”331 Two months later, he wrote that a Soviet violation of a nuclear testing moratorium had endangered American national security such that “we had no rational choice but to improve our own 327 LBJ to Terence Grieder, October 31 1961, VP Papers, Box 61, 1961 Subject File, Armed Services – Production, Weapons, LBJL 328LBJ to Royal Davis, December 8 1961, op. Cit. 329LBJ to Clyde L. Wilson, December 15, 1961, Op. Cit. 330LBJ to R. G. Smith, December 19 1961, LBJ to James W. Hillesheim, December 19 1961, op. Cit. 331 LBJ to Elna Van Essen, March 17 1962, Op. Cit. 167 competence through testing.”332 In October 1963, after the Cuban missile crisis, he wrote that “it is my judgement that the (Nuclear Test Ban) treaty was achieved because of our strength and not because of our weakness.333 Johnson's beliefs about nuclear weapons and their impact on international politics remains unclear, but his beliefs about the revolutionary power of the weapon are less numerous than his pre-nuclear thinking that assumes that nuclear weapons have not changed warfare and which emphasises nuclear superiority and falling dominos.. There is little evidence that he believed that the nuclear weapons revolutionised interstate warfare and none that he learned anything about them from the Bay of Pigs or Cuban Missile crisis. At most, the crisis may have reinforced his beliefs about the importance of credibility since he may have believed that Kennedy’s weakness at Vienna caused the crisis. But there is no evidence to support this claim, and even if it were true the crisis would only have reinforced his earlier beliefs. 5.5. Johnson's Beliefs: President, November 1963-1965 In this section, I show that Johnson’s lessons from World War Two influenced his Vietnam decisions. This is important for the purposes of this dissertation because it shows that the Bay of Pigs, Cuban Missile crisis and John Kennedy had no influence on his beliefs and policies. Vietnam dominated the foreign relations of Johnson’s presidency, and his policies toward the Soviet Union, the two Germany's and Cuba during the remainder of 1963 and most of 1964 were similar to Kennedy's. However his beliefs 332 LBJ to Victoria P. Nichols, May 18 1962, VP Papers, 1962 Subject File, Armed Services – Production, Weapons, Box 125, LBJL 333 LBJ to Albert M. McNell, Jr., October 8 1963, Vice Presidential Papers, 1963 Subject File, “Treaties: Nuclear Test Ban Treaty, 3 of 3,” VP, 1961-1963, Box 239, LBJL 168 about Cuban subversion in the Caribbean and the German status quo were more hawkish than Kennedy’s. Johnson did not learn about the dangers of using force in the third world from Kennedy. His fundamental beliefs about international politics did not change after he became the President, and his policies seem to have been mostly influenced by his own lessons from Munich. While Kennedy had learned about the dangers of trusting his military’s advisers hawkish policies, and more specifically aggressive military policies to stave off defeat in third world conflicts from his Bay of Pigs disaster, Johnson was marginalised from this decision and did not learn similar lessons. The more cognitively accessible Munich lessons seem to explain Johnson’s policies. In his first address before the Joint Session of Congress on November 27, 1963, Johnson stated that “this nation will keep its commitments from South Vietnam to West Berlin.”334 Kennedy had seriously considered and made steps toward recognition of and rapprochement with Castro. Johnson, on the other hand, told the director of the CIA John McCone on November 28 that Cuban sponsorship of radicalism in the Caribbean was “one that we could not live with” and that “we had to evolve more aggressive policies.” On December 2 he told William Fulbright, Chairman of the Senate Foreign Relations Committee, that “I'm not getting into any Bay of Pigs deal,” but on February 8 1964 appointed his National Security Adviser McGeorge Bundy and others to develop a policy for Cuban exportation of subversion, and on May 30 discussed with McCone the feasibility of an invasion of the island.335. Johnson did not receive any Soviet ultimatum’s on Berlin, and on January 1 1964 he told 334 Johnson, Vantage Point, p.xi 335John Dumbrell, President Lyndon Johnson and Soviet Communism, 2004. Manchester: University of Manchester Press, p.138; Michael R. Beschloss, Taking Charge: The Johnson White House Tapes, 1963-64, 1997. New York: Touchstone, p. 87, 234 169 former intimate Helen Douglas that “I'm cutting down the military bases, some of these archaic, old, out-fashioned things, and Khrushchev's cutting his down too...and I’m taking that money and putting it into poverty.”336 This policy carried Kennedy's detente initiatives forward and responded to Khrushchev's decision announced on December 13 1963 to cut Soviet military spending and the size of his armed forces. Johnson also explained to a reporter that an American plane shot down over East Germany was not a serious incident.337 Yet he continued to believe, consistently with the Munich analogy, that if NATO did not “tie the Germans in, there was some 17-year old right now in Germany who would be a 20-year-old little Hitler in another three years.”338 Johnson's beliefs prior to Kennedy's assassination were consistent with making a strong commitment to stave off defeat in South Vietnam. He probably committed himself to use force to save South Vietnam from defeat within the first forty-eight hours of his Presidency.339 The relative quiet in foreign affairs offered no other challenges to distract his attention, and he never seriously challenged the validity of the Munich analogy or reframed questions or alternatives in the course of policy deliberation.340 Questions were not asked about the goals that the United States sought to pursue, what military strategy 336Beschloss, Johnson White House Tapes, p.137 337 Kent Germany et al (eds.) The Presidential Recordings: Lyndon B. Johnson, Volume III, 2005. New York: Norton. p.1069. 338 Date unspecified, in Frank Costigliola, “Lyndon B. Johnson, Germany and the End of the Cold War,” in Nancy Bernkopf Tucker and Warren I Cohen, (eds.) Lyndon Johnson Confronts the World, American Foreign Policy, 1963-68. 1994, pp. 173-210. New York, Cambridge University Press, p.174 339 The literature on the causes of the Vietnam war is extensive, and here i only address Johnson's beliefs about escalating the war in Vietnam from the start of his Presidency in November 1963 to the decision to escalate with over one hundred thousand ground troops in mid 1965. The best two accounts of the American escalation of the Vietnam war are David Kaiser, American Tragedy: Kennedy, Johnson, and the Origins of the Vietnam War, 2000. Belknap: Cambridge, Harvard University Press and Fredrik Logevall, Choosing War: The Lost Chance for Peace and the Escalation of War in Vietnam, 1999. Berkeley: University of California Press. 340 John P. Burke and Fred I. Greenstein, How President's Test Reality, Decisions on Vietnam, 1954 and 1965, 1989. New York: Russell Sage 170 would be consistent with attaining the goals, whether these goals could be attained and what other steps – especially political, social and economic reforms in South Vietnam – were needed to bring stability to the region. Johnson's commitment to the Munich model and his belief that Hitler like aggression required a forceful response inhibited him from questioning how autonomous the Vietcong were from the North Vietnamese government and whether his commitment to Saigon undermined the autonomy and political stability of the Saigon government and the will to win on the part of the South Vietnamese army.341 On his first full day as President on November 24 1963, Johnson happily remarked that international relations was relatively quiet, but admitted that “only Vietnam gave me real cause for concern.”342 That same day, he remarked to his ambassador to South Vietnam Henry Cabot Lodge that I am not going to lose Vietnam. I am not going to be the President who saw Southeast Asia go the way China went. I don't think Congress wants us to let the Communists take over South Vietnam.343 McGeorge Bundy noted Johnson's high priority to South Vietnam at the time: the judgement he's going to have rendered on him is not going to be that he lost Vietnam. He's going to hold it with everybody, and he's going to organise – he's not going to command it – he's going to Senate Majority lead it.344 Johnson's Press Secretary Bill Moyers recalled him expressing recognition of the dangers 341op. cit., p.233 342Johnson, Vantage Point, p.22 343Dallek, Flawed Giant, p.99, italics added; Wicker, JFK and LBJ, p.205 344Dallek, Flawed Giant, p.100 171 in saving South Vietnam and establishing his beliefs about the consequences of its fall: They'll (Chinese and Soviets) think with Kennedy dead we've lost heart. So they'll think we're yellow and don't mean what we say. They'll be taking the measure of us. They'll be wondering just how far they can go. I'm not going to let Vietnam go the way of China. I told them to go back and tell those generals in Saigon that Lyndon Johnson intends to stand by our word, but by God, I want something for my money. I want 'em to get off their butts and get out in those jungles and whip hell out of some Communists.345 Johnson confirmed in his autobiography that he was committed to seeing things through in Vietnam...I was convinced that the broad lines of his [Eisenhower's] policy in Southeast Asia and elsewhere had been right. They were consistent with the goals the United States had been trying to accomplish in the world since 1945.346 Kennedy, who unlike Johnson was influenced by the cognitively accessible Bay of Pigs experience, was also committed to avoid the fall of Saigon during his Presidency. By 1963 he had expanded the number of Americans in Saigon from 685 when he took office to 16700, and shifted their role from advice to assistance. Yet repeated requests from military chiefs for combat troops in Vietnam throughout 1960 and 1961 received a blanket refusal from the President, and he told a number of political associates that he intended a total withdrawal of American military personnel from Vietnam after the 1964 election. In the fall of 1963 he had ordered 1000 advisers home by the end of the year, although this may have been primarily designed to pressure Diem into reforms. While Johnson was prepared to commit American troops independently of the pace of Saigon's 345op. cit. 346Johnson, Vantage Point, p.42 172 domestic reforms that were themselves necessary for the legitimacy and autonomy of the South Vietnamese government, Kennedy viewed such reforms as necessary for further troop commitments. Johnson and Kennedy agreed that a greater South Vietnamese effort was necessary to shore up the weak regime in Saigon. But while Kennedy had always refused to allow American troops to orchestrate combat missions before Saigon showed a greater commitment to do so, Johnson had pushed Kennedy to consider the commitment of “major United States forces to the area” after his May 1961 Vice Presidential visit.347 While Johnson's November 26 1963 National Security Action Memorandum 273 reaffirmed the objective of withdrawing 1000 advisers by the end of 1963, it did not make the levels of American economic and military assistance conditional on domestic reforms, and called for planning “possible increased activity.”348 On January 27 1965, shortly before authorising operation Rolling Thunder, he remarked that “stable government or no stable government, we'll do what we ought to do...we will move strongly.”349 Johnson’s commitment to stave off the fall of South Vietnam is consistent with his long held beliefs about falling dominos, and there is no evidence that he learned from Kennedy about this conflict.350 The question of whether Kennedy would have escalated the Vietnam war as Johnson did is complicated by the sudden deterioration of the status quo in South Vietnam after 347Johnson, Vantage Point, p.54 348Dallek, Flawed Giant, p.101 349op. cit. pp.247-8 350 If Johnson was closely involved in the Cuban Missile crisis (or Bay of Pigs) or otherwise learned from it, perhaps because he did not learn from Munich, it is unclear both what these lessons would have implied for the Vietnam war. He could have concluded, for example, that the crisis showed that Vietnam style counterinsurgencies are an effective means of demonstrating resolve because nuclear war is unlikely. But he also may have concluded that the Vietnam war could have escalated to a nuclear crisis between the United States and Soviet Union (or China). The critical variable here is whether he experienced fear of nuclear escalation (missile crisis) or perhaps resounding conventional defeat (Bay of Pigs). 173 Diem's assassination on November 2 1963, three weeks before his own assassination. Kennedy may have been more averse to escalation than Johnson before Diem's assassination, but he may have escalated the war well beyond sixteen thousand troops if he survived his trip to Dallas. John Prados claimed that on the day Kennedy departed for Texas in November 1963, he asked National Security Council aide for Southeast Asia Michael Forrestal to warn Cambodian leaders and others that he wanted to reduce the American role in the war.351 This could have been designed to pressure reforms in Saigon, and Kennedy may have escalated after further challenges from Republicans or elsewhere. Yet his closest advisers later agreed that Kennedy's response to South Vietnam would have been less than Johnson's. McGeorge Bundy believed that “it would have been different and less than Johnson,” and Robert McNamara thought it “highly probable that he would have pulled us out of Vietnam.”352 Kennedy's most intimate adviser, his brother Robert, stated that if it “hadn't been for Cuba, we probably would have sent large numbers of troops into Laos and Vietnam.”353 Throughout the first eighteen months of his Presidency, he consistently evinced beliefs in the Munich analogy. There is no evidence in the extensive archival and secondary record that any lessons from the Bay of Pigs or Cuban Missile Crisis influenced his beliefs. Much evidence shows that Johnson was of two minds about escalating the Vietnam war. But consistent with psychological learning models, he was not of two minds about the causes and mechanisms of the crisis in Vietnam. His belief that the war would be long 351Prados, Hidden History, p.25 352 Dallek Flawed Giant, pp.98-99, Robert McNamara, In Retrospect: The Tragedy and Lessons of Vietnam, 1995. New York: Vintage, p.96 353 Edwin Guthman and Jeffrey Shulman, RFK in his own words, 1991. p.289. There is also good evidence that Under Secretary of State George Ball's personal experience of the Korean war caused his preference to not escalate in Vietnam. See Kaiser, American Tragedy, p. 349, 357, 375, and Khong, Analogies at War, pp. 150-173 174 and costly and that doing nothing would embolden American adversaries are both consistent with the Munich analogy. Had he entertained beliefs such that the violence in Vietnam was largely unrelated to Beijing and Moscow and that American withdrawal would not undermine his credibility as much as he worried about the costs of action/inaction, he may have not escalated the war. But he did not consider other explanations as rational learning models would predict. In the first week of his presidency, Johnson told the President of the American Electric Power Company that “we've got to either get in or get out, or get off.”354 Two days later he wrote a memo to Maxwell Taylor, Chairman of the Joint Chiefs of Staff, stating that South Vietnam is “our most critical military area right now.”355 Many influential and trusted advisers and associates warned Johnson of the dangers of escalation, but while he was aware of these he always maintained that the costs to his and America's reputation of letting Saigon fall was worse than a long and costly ground war.356 He told his biographer Doris Kearns that he “realised that doing nothing was worse than doing something.”357 On December 23 he defended his beliefs against the warnings of the Secretary for the Majority in the Senate; Vietnam could be “another China,” with its dangerous foreign and domestic repercussions, and he did not want “people around the world worrying about ... whether we've got a weak President or a strong President.”358On February 3 1964, he told the Chairman of the Miami Herald that 354 Beschloss, White House Tapes, p.74 355 Memorandum from President Johnson to the Chairman of the Joint Chiefs of Staff (Maxwell Taylor), Washington, December 2 1963, F.R.U.S. 1961-63, Volume IV, Vietnam, August- December 1963, Document 337 356 Senate Foreign Relations Chair Fulbright and his Senate mentor Richard Russell both explicitly warned of the dangers and costs of escalation. See Beschloss, 1997, p.88, p.95 357Kearns, Lyndon Johnson, p.263 358Beschloss, White House Tapes, p.124 175 leaving Vietnam would “let the dominoes start falling over.” The other options would be “to fight, as we are doing,” or to “sit down and agree to neutralise all of it.” But Johnson knew that “nobody is going to neutralise North Vietnam, so that's totally impractical.”359 Seventeen days later, he informed McNamara of his worry about falling dominoes in Southeast Asia and noted that “I always thought it was foolish for you to make any statements about withdrawing...but you and the President thought otherwise and I just sat silent.”360 On March 4 1964, Johnson noted to Bundy that “we haven't got any Congress that will go with us, and we haven't got any mother that will go with us in a war.”361 But one month later he and McNamara told congressional leaders that they were “considering taking the war to North Vietnam.” Johnson was aware of the challenge and dangers of fighting a counterinsurgency in Vietnam. But he never seriously believed other explanations that challenged the Munich analogy. He rejected that the Vietnamese conflict was a civil war with important local motivations and rather maintained that Hanoi, Beijing and Moscow were in full control of the violence throughout the country. He worried that the cost of the Vietnam war would undermine his prized domestic economic and health care reforms. But he was more worried that defeat in Vietnam would embolden China and the Soviet Union to control Southeast Asia and cause a third world war. But, as Khong as shown, the salience of Korea taught him to not escalate in such a way that would cause a Chinese response. He thus felt committed to a strategy that would demonstrate resolve but not cause China to enter the war. On May 27 1964, he thus told his senate mentor Richard Russell that 359Beschloss, White House Tapes, p.214 360op. cit., (or p.258?) 361Beschloss, White House Tapes, pp.266-7, p.365 176 every time I think about making this decision and think about sending that father of those six kids in there. And what the hell are we going to get out of his doing it? And it just makes the chills run up my back. I just haven't got the nerve to do it, and I don't see any other way out of it.362 One day, lying in bed with the covers almost over his head, Johnson told his Press Secretary Bill Moyers that he felt as if he was in a Louisiana swamp “that's pulling me down.”363 His wife Lady Bird remarked in her autobiography that “Lyndon lives in a cloud of troubles, with few rays of light...in talking about the Vietnam situation.”364 On June 1, after his democratic colleague George Smathers told Johnson that nobody he spoke to in congress thought “we ought to fight a war in that area of the world,” the President replied that we can't retreat because...it'll be very dangerous to the United States to give up all of our interests in Southeast Asia. The Chinese are very aggressive, and, if you run from 'em, why, we'd of run out of Greece and Turkey a long time ago...If we show some strength, we may have a chance to hold on.365 The next day, after he had commented to his advisers on his upcoming speech where he would declare that he had no plans to escalate the Vietnam war, he instructed his top advisers in Honolulu to discuss plans for “selected and carefully graduated military force against North Vietnam after appropriate diplomatic and political warning.”366 On June 11, he again confided in Russell: 362 Beschloss, White House Tapes, pp. 363-370 363 Michael Beschloss, Reaching for Glory: Lyndon Johnson's Secret White House Tapes, 1964-65 (hereafter referred to as Whitehouse Tapes II) 2001. New York: Touchstone, p.378 364Lady Bird Johnson, A White House Diary, 2007. Austin: University of Texas Press, pp.247-8. 365Dallek, Flawed Giant, p.146 366Beschloss, White House Tapes, p.381 177 I'm confronted. I don't believe the American people ever want me to run. If I lose it, I think that they'll say I've lost. I've pulled in. At the same time, I don't want to commit us to a war. And I'm in a hell of a shape....I've got a study being made now by experts...whether Malaysia will necessarily go and India'll go and how much it'll hurt our prestige if we just got out and let some conference fail or something...A fellow like A.W. Moursund said to me last night “Goddamn, there's not anything that'll destroy you as quick as pulling up stakes and running. America wants, by God, prestige and power.” I said, “Yeah, but I don't want to kill these folks.” He said, “I don't give a damn. I didn't want to kill em in Korea, but if you don't stand up for America, there's nothing that a fellow in Johnson city - or Georgia or any other place – they’ll forgive you for anything except being weak.”367 When Russell responded that he shared Mansfield's fears that this would cause war, Johnson replied that “I do too but the fear the other way is more.”368 In November when Johnson asked his advisers to discuss the appropriate American response to Hanoi's next attack, the discussion addressed not whether to act but what to do. There was little, if any, discussion about leaving Vietnam, or whether the U.S. could succeed where the French had failed. It was generally assumed that the U.S. was already committed to stopping the Communists, and that this required the use of U.S. Forces.369 Throughout March, as public calls for negotiations and avoiding another Asian war increased, Johnson reiterated his belief in the domino theory and the dangers of appeasement through negotiations.370 On April 29 Johnson reaffirmed his belief that the violence throughout Vietnam was less the result of widespread yearnings for political 367Beschloss, White House Tapes, pp.401-2 368op. cit., 369 Gibbons, U.S. Government and Vietnam, p.366, Dallek, Flawed Giant, p.241 370Beschloss, White House Tapes II, p.238 178 autonomy after more than a century of colonial rule and more Russian and Chinese sponsored aggression. He claimed that “we could solve the Vietnam thing if the North Vietnamese and the Chinese really felt that we were there to stay.”371 A few days later he defended his 700 million dollars appropriations bill to a congressman with the familiar argument that a failure to stop the Communists in Vietnam would “show that American commitment is worthless, and “the road is open to expansion and to endless conquest.”372 William Bundy, Assistant Secretary of State for East Asian and Pacific Affairs, noted Johnson's thinking behind the July escalation: Withdrawal would be a disaster, a harsh bombing program would not win and could easily bring a wider war, and standing pat with existing forces was only slow defeat. Only...doing what McNamara urged was left...It was the end of the debate on policy, and the beginning of a new debate on tactics and above all on the presentation to the country. The President had decided to put in his stack.373 He told his advisers on July 21 1965 that “we know it is dangerous and perilous, but the big question is, can it be avoided?”374 Johnson recalled in his autobiography that while he was aware of the dangers of sending American troops to fight in Vietnam, his great fears of communist emboldenment after the fall of Saigon and the start of a third world war overcame his caution: We discussed Ball's approach (that we could not win a protracted war against local guerrillas in Asian jungles, and should cut losses 371Op. cit., p.251 372 Dallek, Flawed Giant, p.269 373Dallek, Flawed Giant, p.275 374Khong, Analogies at War, p.124 179 and pull away) for a long time and in great detail. I think all of us felt the same concerns and anxieties that Ball had expressed, but most of these men in the Cabinet room were more worried about the results, in our country and throughout the world, of our pulling out and coming home. I felt the Under Secretary of State had not produced a sufficiently convincing case or a viable alternative. Dean Rusk expressed one worry that was much on my mind. It lay at the heart of our Vietnam policy. “If the Communist world finds out that we will not pursue our commitments to the end,” he said, “I don't know where they will stay their hand.” I could see trouble ahead in every part of the globe...I was convinced that our retreat from this challenge would open the path to World War III. I have no doubt, the North Koreans were only waiting for us to be thrown out of Vietnam before launching their own offensive against South Korea. Thus what we saw taking shape rapidly was a Djakarta- Hanoi-Peking-Pyongyang axis, with Cambodia probably to be brought in as a junior partner and Laos to be merely absorbed by the North Vietnamese and Chinese. The members of this axis were undoubtedly counting on South Vietnam's collapse and an ignominious American withdrawal. The entire region would then have been ripe for the picking. From all the evidence available to me it seemed likely that all of Southeast Asia would pass under Communist control, slowly or quickly, but inevitably, at least down to Singapore but almost certainly to Djakarta.375 His statements to Kearns were consistent with this: I knew from the start that I was bound to be crucified either way I moved. Yet everything I knew about history told me that if I got out of Vietnam and let Ho Chi Minh run through the streets of Saigon, then I’d be doing exactly what Chamberlain did in World War II. I'd be giving a big fat reward to aggression. Then I would be seen as a coward and my nation would be seen as an appeaser and we would both find it impossible to accomplish anything for anybody anywhere on the entire globe.376 This evidence shows that the salience of World War Two influenced Johnson's beliefs and caused his subsequent Vietnam preferences. As Johnson's statement to Kearns suggests, the younger generation may also have supported escalation in Vietnam if they 375Johnson, Vantage Point, p.147, pp.135-6, p.151 376Kearns, Lyndon Johnson, pp. 251-3 180 experienced World War II and if Hitler was as cognitively accessible to them as it was to him. Why should I listen to all those student peaceniks marching up and down the streets? They were barely in their cradles in the dark days of World War II; they never experienced the ravages of Adolf Hitler; they were only in nursery school during the fall of China; they were sitting in grammar school during the Korean War; they wouldn't know a Communist if they tripped over one. They simply don't understand the world the way I do. The professors believe you can get peace by being soft and acting nice. But everything I know about history proves this absolutely wrong. It was our lack of strength and failure to show stamina, our hesitancy, vacillation, and love of peace being paraded so much that caused all our problems before World War I, World War II, and Korea.377 5.6. Conclusion Johnson learned much more from his own cognitively accessible World War Two experiences than from the Cuban Missile Crisis, Bay of Pigs and John Kennedy. His beliefs were remarkably consistent through the three decades assessed above and reflected a serious and consistent commitment to the central aspects of the Munich analogy. As argued above, this suggests two hypotheses. Firstly, leaders will only learn similar lessons about the danger of nuclear escalation from their predecessors if they also experienced fear during a crisis or war. Short of the experience of fear, leaders will not learn about the dangers of nuclear coercion from their predecessors. Secondly, successors to new nuclear power decision-makers who learned through fear of the dangers of nuclear coercion will make policies that are based on their own cognitively accessible 377Kearns, Lyndon Johnson, pp. 312-13, italics added 181 experiences. These cognitively accessible experiences may be their predecessor’s nuclear crisis. Soviet restraint after Khrushchev may be explained by Brezhnev also experiencing fear during the Cuban Missile crisis. The stability of nuclear South Asia likely depends on what Army General Ashfaq Kayani and his ISI associates learned from the ten month crisis. This chapter suggests that if the war scare during the ten month crisis was as cognitively accessible to Kayani as it was to Musharraf, subsequent Pakistani policy in Kashmir and the region may be much more moderate than it was immediately after the nuclear tests. 182 Chapter 6: Conclusion 6.1. Conclusion This dissertation originated with a simple claim. The most common consequence of nuclear proliferation, despite much scholarly attention to the impact of nuclear weapons on international politics since the end of the Second World War, has been neglected. Nuclear proliferation has been followed by crises and conventional wars, not nuclear wars. Moreover, while earlier research aggregated temporal variation and found no relationship between nuclear proliferation and war, disaggregating temporal effects establishes a robust result. There is a robust relationship between nuclear weapons, war and time. The effect of experience with nuclear weapons was established in two multivariate statistical studies. War prone states are more likely to acquire nuclear weapons, and nuclear weapon possession has no effect on their dispute initiation; nuclear powers fight about as much as the average of all other states. But new nuclear powers have a much higher probability of initiating and reciprocating crises and wars than before they develop nuclear weapons. These same experienced nuclear powers subsequently fight much less, and become about as war prone as they were before nuclear proliferation. These two effects washed out in statistical tests that do not model time. But correlation is not causation. We do not know why experienced nuclear powers do not fight. How experience with nuclear weapons pacifies new nuclear weapons – the causal mechanism – has not been established. More specifically, the literature lacks an explanation of why new nuclear powers fight that also explains why experienced nuclear powers do not. This is further problematic because different mechanisms would require 183 different policies, and without knowing which mechanism explains the effect of experience we cannot formulate policies to engage new nuclear powers. If the United States and its allies are to develop policies that will ensure that Iran and North Korea behave like other experienced nuclear powers, we must understand how experience with nuclear weapons causes restraint. The end of World War II and the beginning of the Cold War may explain why powerful new nuclear powers such as the United States, France and Britain fought against weaker non-nuclear adversaries in Korea and Egypt. But these structural changes cannot explain why weak revisionist states challenge their more powerful principal adversary after about 1960. Other plausible explanations based on decision-maker rivalry, territorial defences or preventive motivations for war are also deficient. We cannot explain why Pakistan and the Soviet Union – similar in important ways to Iran and North Korea – found themselves in wars or major crises when they were new nuclear powers but behaved very differently after experience with nuclear weapons. Paul Kapur’s theoretical model specified when new nuclear powers should start crises or wars. States that are conventionally inferior to their principal adversary and which have revisionist objectives over them will use nuclear weapons to challenge the regional status quo without fear of retaliation. Their more powerful nuclear weapon equipped adversaries could conventionally defend against such territorial revisionism before they (the revisionist) acquired nuclear weapons. But the revisionist state’s nuclear posture often deters nuclear and conventional retaliation. Kapur’s model thus explains why weak revisionist new nuclear powers fight, but does not explain why the same 184 experienced nuclear powers do not. His ad hoc explanation for Pakistan may be incorrect, and certainly does not explain the effect of experience with nuclear weapons in other cases. If Kapur’s theoretical model tells the whole story of the relationship between nuclear proliferation and war, we should observe experienced nuclear powers behaving like new nuclear powers. But we do not. This dissertation developed rational choice and psychological nuclear learning models to explain why new nuclear powers fight and experienced ones do not. Both models address decision-making under uncertainty. These approaches are prescient because civilian and military decision-makers who have developed nuclear weapons would face uncertainty about what they can do with them, as well as what other allies and adversaries think they could and will do. They might believe that the possibility of nuclear escalation makes aggression dangerous, but they might also believe that the low probability of nuclear escalation makes well calculated conventional challenges strategically effective. Rational choice learning models have not been applied to explain the specific relationship between nuclear proliferation and war, but two common assumptions in recent game theoretic research stipulate what such models would assume. Firstly, rational choice learning models assume that decision-makers would be attentive to enough information, history and logic to accurately estimate the diagnostic value of each observation and estimate the correct or near correct probabilities of different causal relationships. This would lead to an accurate understanding of the relevant empirical relationships that would generate optimal or near optimal foreign policies. Secondly, 185 given the obvious incentives to be attentive to the fact that the historical record shows that nuclear weapons have hardly ever offered coercive advantages, rational choice learning models would predict decision-makers to be attentive to this. They would predict that new nuclear power decision-makers would not pursue nuclear coercion, and after accounting for error they would expect them to practice it no more than half the time. Expressed differently, after accounting for error, rational choice models would predict that decision-makers would overestimate the power of nuclear coercion as often as they should accurately estimate it because inferential errors are, by definition, random. A second nuclear learning model is based on most robust findings from over three decades of research in social psychology. This research suggests that rather than perceiving and processing information optimally as rational choice learning models predict, the sheer volume and complexity of these procedures force people to rely on rough and ready heuristics that simplify decision making. Such procedures often produce near optimal decisions but also cause systematic and lasting biases. Psychological research has not been applied to address the consequences of nuclear proliferation, but two biases are relevant for the psychological nuclear learning model that I develop in this dissertation. Firstly, the availability heuristic, and its associated cognitive accessibility biases, illusions of control and attribution biases, will cause new nuclear power decision-makers to learn that nuclear coercion is more effective and less dangerous than a systematic analysis of the historical record warrants. This will cause them to challenge the status quo more than rational choice models would predict. The psychological nuclear learning 186 model predicts decision-makers of weak revisionist new nuclear powers to systematically bias the efficacy of nuclear coercion and underestimate the dangers of nuclear coercion. This will cause persistent challenges to the status quo. But rather than learning about the error of their strategy from its failure, only experiencing the second bias, fear, will cause decision-makers to learn about the dangers of nuclear coercion and refrain from it. In the second bias, the experience of fear causes decision-makers to adopt pessimistic risk estimates in their nuclear diplomacy and unrelated foreign policies. The psychological nuclear learning model thus offers an explanation of the statistical effect of experience with nuclear weapons. It counter-intuitively suggests that nuclear proliferation is dangerous when decision-makers believe it is safe and safe when they believe that it is dangerous. The Soviet Union in 1959 and Pakistan in 1999 are cases of weak revisionist new nuclear powers challenging more powerful nuclear adversaries after approximately 1960. Limited data in both cases does not allow definitive conclusions about which learning model better explains them. But biographies of Khrushchev’s son Sergei who intimately observed the making of key decisions, recently released Soviet archival data gathered by Alexandr Fursenko and Timothy Naftali and other secondary sources strongly suggest that the psychological learning model better explains the data than the rational choice alternative. Khrushchev did not consider the other causes of the British and French retreat from Egypt in 1956, and continued to believe in the efficacy of nuclear coercion after two failed attempts until he experienced fear at the end of the Cuban Missile crisis. As the psychological nuclear learning model would suggest, Khrushchev did not consider what other variables may have caused his third world successes. Moreover, he did not learn 187 about the dangers of inadvertent escalation from nuclear coercion from history but the experiences of his own crisis or war. Finally, the experience of fear caused him to adopt pessimistic risk estimates in nuclear diplomacy and other foreign policies. The psychological nuclear learning model explains these lessons much better than the rational choice alternative. The data on Pakistani nuclear decision-making is less reliable, but there is some evidence that the cognitive accessibility of India’s isolation after the May 1998 nuclear test caused Musharraf and his colleagues to believe that the United States would intervene favourably on his behalf. He seems to have learned that Pakistani nuclear weapons would defend his intrusion at the Kargil sector of Kashmir from immediate Indian escalation and attract favourable American intervention. Consistently with the psychological nuclear learning model, Musharraf did not experience fear at Kargil and continued to underestimate the dangers of nuclear coercion until he experienced fear during the second peak of the ten month crisis. But an inability to determine whether he authorised future Pakistani sponsorship of terrorism in Kashmir, as well as ambiguity regarding what Musharraf’s successor, Ashfaq Kayani, learned from the crisis years precludes a more definitive assessment. While there is no evidence that Musharraf considered the historical record on nuclear coercion, this does not prove definitively that he did not do so. Nonetheless, Musharraf’s systematic underestimation of the dangers of nuclear coercion until he experienced fear and his possible restraint thereafter exhibited in the Composite dialogue is more consistent with the psychological nuclear learning model than the rational choice alternative. Finally, Chapter 5 showed that while we cannot test the learning models on Brezhnev and Zardari, the experience of another 188 successor in another case strongly suggests that it is plausible that their lessons are likely to be more consistent with a psychological learning model than a rational choice alternative. Available evidence severely undermines the empirical validity of rational choice learning models. The Soviet, and to a lesser extent Pakistani, case show that cognitively accessible data influences the attention and information processing much more than rational choice models predict. Decision-makers systematically overestimate the power of nuclear coercion until they experience fear of nuclear war. Soviet and Pakistani decision- makers had strong incentives to accurately estimate the low validity of nuclear coercion but they failed to do so. In both cases, the experience of fear of nuclear war was necessary for learning about the limits of nuclear coercion. Rational choice learning models that predict actors to accurately estimate the power of nuclear coercion and not pursue it predict the behaviour of experienced nuclear powers; over time, their predictions mirror those of the psychological nuclear learning model. But they miss a critical part of the story. The claim that these models are good enough to explain how new nuclear powers behave is incorrect – they did not predict the causes and consequences of the most dangerous crises in the Cold War and South Asia. Rational choice learning models do not explain how new nuclear powers become more experienced and are a weak guide to predicting how new nuclear powers behave and when the effects of experience will kick in because they cannot explain the strong effect of fear of nuclear war. The psychological learning model explains the effect of experience with nuclear weapons and offers much more to policy-makers. 189 The evidence also suggests that both states in an enduring rivalry need not experience fear of nuclear war for cooperation to occur. The key variable seems to be whether challengers to the status quo learn through fear of nuclear war about the limits of nuclear coercion. It is not clear that Kennedy experienced fear of nuclear war during the Cuban Missile Crisis. It is also unclear if Vajpayee and his associates experienced fear of nuclear war during the second peak of the ten month crisis. Future research should address whether Kennedy and Vajpayee experienced fear of nuclear war. 6.2. Problems and Rejoinders This section addresses four important problems with the application of the psychological nuclear learning model developed in this dissertation. After showing that they do not undermine the central findings, I address the policy implications for addressing nuclear proliferation in Iran and North Korea. The final section concludes with suggestions for further research for addressing the role of experience in international politics and political science more generally. Firstly, one potential challenge to the thesis of this dissertation is that the effect of experience with nuclear weapons could be explained by other stochastic variables. Either decision-makers could learn of the dangers of nuclear coercion through variables other than fear of nuclear escalation or variables other than the potential for nuclear escalation could induce fear. The small number of cases examined here makes this hypothesis hard to definitively dismiss, but the data in the Soviet and Pakistani cases 190 suggest that Khrushchev and Musharraf learned of the dangers of nuclear coercion through experiencing fear of nuclear escalation. Secondly, there are problems with defining and measuring revisionism. The psychological nuclear learning model predicts new nuclear states that are conventionally inferior to their principal adversary and which have revisionist territorial preferences over them to be conflict prone. Revisionism was defined as dissatisfaction with major aspects of the status quo and a willingness to risk high costs to change it. But there are important ambiguities in the concept of revisionism such that developing rigorous coding rules is difficult. Neither Khrushchev nor Musharraf would have accepted that their preferences were revisionist. Khrushchev told his colleagues in 1959 that “all we want to do is to secure the status quo.” Musharraf claimed that Pakistani claims on Kashmir directly resulted from multiple Indian injustices since partition.378 Revisionism can be motivated not by the desire to expand but to maintain a deteriorating status quo. Khrushchev’s policies in Berlin and Cuba can fairly be viewed as designed not to increase his power but to maintain his security. Coding his policies as anti-status quo and revisionist therefore seems problematic. But even though the status quo moved against Khrushchev and was not static, the Soviet leader was nonetheless motivated to revise it. This is not the same as challenging a static status quo but revisionism nonetheless; a coding of revisionist says nothing about whether the status quo is moving or static or the origins of these territorial preferences. Who qualifies as revisionist often depends on where in time or on a means-ends chain one looks. Khrushchev’s decision to challenge Western access rights in Berlin and deploy 378 Fursenko and Naftali, Khrushchev’s Cold War, p.220; Pervez Musharraf, Author Interview, March 2010. 191 strategic nuclear missiles in Cuba appears revisionist, but it was mostly motivated by defensive motivations in West Berlin that originated in the immediate settlement to World War Two.379 But this Soviet sphere of influence was at least partly motivated by offensive motivations in the strategic competition with the United States. Nonetheless, coding Khrushchev in 1958 and Musharraf in 1999 as revisionist is not the same as coding them as always revisionist. At these junctures in the Cold War and Indo-Pakistani rivalry, the Soviet Union and Pakistan were revisionist. Further research should examine how strategic interaction after exogenous shocks influences adversaries dyadic motivations and objectives over time. The rough distinction between states that are, in a given period, relatively satiated with the prevailing situation and those that are dissatisfied and strongly motivated to change it can be applied to Khrushchev’s Russia and Musharraf’s Pakistan. Coding these states as revisionist says nothing about the origins of these territorial preferences.380 Thirdly, the dissertation begs the question of whether status quo new nuclear powers are subject to the same psychological biases that afflict weak revisionist new 379 Moreover, often the terms offensive and defensive do not capture the complexity of a state’s movitations. Soviet West Berlin motivations were both defensively orientated insofar as they maintained the territorial integrity of East Germany and, more generally, the legitimacy of a Soviet sphere of influence in Eastern Europe. 380 For more on the problem of revisionism see Robert Jervis, “Was the Cold War a Security Dilemma?” Journal of Cold War Studies 3 (1) 2001 pp. 36-60, pp. 38-40. More coding problems arise if we look further back to the end of the second World War; American plans to create a new post-war order after 1945 were surely more revisionist than those of the Soviet Union. One could plausibly argue that Soviet revisionism was caused by American policies. But American policies were also motivated by the at least partly defensive desire to integrate or contain the Soviet Union. Similarly, Pakistan’s Kashmir views have long been viewed as revisionist, but the years immediately after partition witnessed aggressive Indian policies designed to acquire Kashmir. This is surely revisionist but can also be explained by at least somewhat defensive motivations. While coding revisionists in the aftermath of great structural changes is difficult, who qualifies as revisionists after this depends on where one decides to look. 192 nuclear powers. Stronger states that are more content with the status quo such as India and Israel may be subject to the same biases in attention and information processing. But their lack of motivation to challenge the status quo would presumably make them less motivated to do so. One might therefore argue that motivated biases are doing most of the causal work that the psychological model, in explaining new nuclear power behaviour, attributes to cognitive factors. But “people’s cognitive and affective systems are tightly interconnected and emotions typically have much of their effects through their impact on cognitive processes.”381 Further research should examine whether decision-makers in status quo new nuclear powers such as France, Israel and South Africa suffer from similar biases in assessing the power of nuclear coercion and whether some decision-makers are emboldened to develop preferences to challenge or revise the status quo. A final problem is whether leader’s successors who may have not experienced a nuclear crisis learned through fear of the dangers of nuclear coercion. Chapter 5 suggests that they will if the nuclear crisis was cognitively accessible to them. It is likely that the Cuban Missile crisis was cognitively accessible to Brezhnev; the ten month crisis may have been cognitively accessible to Kayani. Future research might examine the hypothesis that nuclear crises that are sufficient to cause fear of nuclear war and restraint in the crisis might be so cognitively accessible that subsequent leaders in the weak revisionist state learn similar lessons about nuclear coercion. This raises issues of intergenerational learning that I address below. In the next section, I apply the model to predict the consequences of Iranian and North Korean nuclear proliferation. 381 Thomas D. Gilovich and Dale W. Griffin, “Judgement and Decision Making,” in Susan T. Fiske, Daniel T. Gilbert and Gardner Lindzey (eds.), Handbook of Social Psychology: Volume 1, (Fifth edition), 2010. New York: Wiley, pp. 542-588, p.561 193 6.3. Iran and North Korea Iran’s military is large, but its conventional weapons are obsolete for military engagements with the United States; it could not defeat the American military if U.S. forces were sent to defend the Gulf States from Iranian aggression. Israel also has superior conventional military power in terms of ground, sea, air and space capabilities.382 Supreme leader Khamenei and the most senior Revolutionary Guard Commanders may be content with the regional status quo. Iran’s nuclear program may be motivated by the desire to deter coercion and regional instability. In this case, Iranian nuclear missiles would be used for deterrence.383 But one or several of the Iranian civilian and military decision-maker may be dissatisfied with some aspect of the status quo. Iran has long desired to reduce American influence in the region, or at least increase the cost of this influence, through support for proxies such as Hezbollah and Hamas, efforts to subvert its conservative Sunni neighbours and provocative naval manoeuvres in the Persian Gulf.384 Alternatively, some Iranian decision-makers may be content with the regional status quo before nuclear weapons development but emboldened to challenge it after nuclear proliferation. 382 Anthony Cordesman, Iran, The Gulf, and Strategic Competition: The Conventional Balance, Center for Strategic and International Studies, August 2010; Mark Fitzpatrick and Michael Elleman, Iran’s Ballistic Missile Capabilities: A net assessment, 2010. London: International Institute for Strategic Studies. 383 For such views, see Mohsen M. Milani, “Tehran’s Take,” Foreign Affairs July/August 2009 Vol. 88 Issue 4 pp. 46-62; Kayhan Barzegar, Iran’s Foreign Policy Strategy after Saddam; Daniel Byman, Shahram Chubin, Anoushiravan Ehteshami and Jerrold Green, Iran’s Security Policy in the Post-Revolutionary Era, 2001. Santa Monica: RAND. 384 Daniel Byman, “Iran, Terrorism and Weapons of Mass Destruction,” in Studies in Conflict and Terrorism, 2008, 31 pp. 169-181; Caitlin Talmadge, “Closing Time: Assessing the Iranian Threat to the Strait of Hormuz,” International Security 33 (1) 2008 pp. 82-117. 194 The orthodox wisdom assumes that Iran is a revisionist state and predicts that an Iranian nuclear weapon or small nuclear arsenal will cause more of these strategies intended to raise the political, economic and human costs of American influence in the Middle East. These include the harassment of Persian Gulf tanker traffic, damage to oil infrastructure and attacks on U.S. forces in Afghanistan and Iraq by Iranian operatives or proxies. More generally, this wisdom asserts that an Iranian bomb would challenge American extended deterrence commitments to Israel, Saudi Arabia, Egypt and weaker conservative Gulf states. In this logic, the Iranian development of a nuclear weapon would cause regional instability. President Barack Obama thus stated that an Iranian bomb would be “profoundly destabilising and extraordinarily dangerous.”385 Kapur suggested that Iranian decision-makers could increase their support for terrorism or even engage in outright conventional aggression to challenge objectionable territorial or political arrangements while insulated from large-scale U.S. or Israeli retaliation.386 The first stage of the psychological learning model predicts that if Iran has revisionist ambitions, an Iranian bomb would cause such short term instability. Biased Iranian learning may be sufficient for a crisis or war without Israeli preventive motivations.387 385 Joint Statement by United States President Obama and Israeli Prime Minister Netanyahu, May 2009 386 Kapur, Ten Years of Instability, p.94; see also Scott D. Sagan, “How to Keep the Bomb from Iran,” Foreign Affairs, Vol. 85, No. 5, Sep-Oct 2006, pp. 45-59, p.53; Shahram Chubin, “Iran’s Risk Taking in Perspective,” IFRI Security Studies Centre, Proliferation Papers, 2008; Anthony Cordesman, Iran’s Evolving Threat, Center for Strategic and International Studies, 2010. Iran could either develop nuclear missiles to realise revisionist ambitions or develop nuclear missiles for defensive purposes and be emboldened to use them for coercion. 387 See William J. Broad, John Markoff and David E. Sanger, “Israeli Test on Worm Called Crucial in Iran Nuclear Delay,” New York Times, January 15, 2011; Eli Lake, “Operation Sabotage: Our secret war against Iran,” The New Republic, July 22 2010, Vol. 241 Issue 12 pp. 16-17; Whitney Raas and Austin Long, “Osirak Redux? Assessing Israeli Capabilities to Destroy Iranian Nuclear Facilities,” International Security 31 (4) 2007 pp. 7-33 195 But this dissertation has shown that the orthodox wisdom’s pessimism has not accounted for the effect of experience with nuclear weapons. James Lindsay and Ray Takeyh hinted at its effects when they noted that a nuclear Iran would be most dangerous “at first, when it would likely be at its most reckless...like other nuclear aspirants before them, the guardians of the theocracy might discover that nuclear bombs are simply not good for diplomatic leverage or strategic aggrandizement.”388 The psychological nuclear learning model developed here provides a causal explanation grounded in empirical research for this speculation, and suggests that if Iranian decision-makers are revisionist and develop nuclear weapons, they will continue to believe that nuclear coercion is effective until it causes a war or crisis significant enough to induce fear of nuclear escalation. They may learn from early crises, especially if the United States intervenes, that nuclear coercion is effective and safe and continue to challenge the status quo. The historical record suggests that if Iranian civilian and/or military decision- makers challenge the regional status quo or otherwise pursue nuclear coercion, they may cause a nuclear crisis and experience fear of nuclear escalation that will resolve the crisis or war and moderate future Iranian nuclear and foreign policies. An Iranian bomb need not cause long term regional instability. If the United States can let Iranian decision- makers learn through fear about the dangers of nuclear coercion in a nuclear crisis, they will probably behave like Musharraf after 2002 and perhaps Khrushchev after 1962. American intervention to contain nuclear escalation might cause some Iranian decision- makers to learn that nuclear coercion is neither costly nor dangerous because the United States will manage nuclear escalation. But if Washington lets Iranian decision-makers 388 James M. Lindsay and Ray Takeyh, “After Iran Gets the Bomb: Containment and its Complications,” Foreign Affairs , March-April 2010, pp. 33-50, pp. 37-38. 196 learn of the dangers and costs of nuclear coercion through not intervening in their own crisis or war, Iran is more likely to behave like other experienced nuclear powers. This suggests a conclusion: risking the very unlikely outcome of nuclear war in the short term may promise stability in the Persian Gulf in the long term. Finally, the model developed here suggests that the ease of extended nuclear deterrence depends on where one looks. Thus Lebow and Stein concluded that “nuclear deterrence is robust when leaders on both sides fear war and are aware of each other’s fears.”389 Extended nuclear deterrence will be relatively easy after Iranian decision- makers have experienced fear of nuclear escalation. But extended deterrence will be much harder before Iranian decision-makers experience fear. The challenge for the United States will be to construct an extended deterrence policy that establishes credible commitments to both secure its regional allies and not manage a nuclear crisis involving a new nuclear Iran. The psychological nuclear learning model offers a similar prediction about the consequences of North Korean nuclear missile development. North Korea currently has enough plutonium for between six and eight nuclear bombs and is believed to be rapidly increasing this stockpile. Pyongyang also recently revealed extensive uranium enrichment facilities. The 2006 nuclear test was widely viewed as a failure, but the 2009 test had a yield roughly the size of the Hiroshima and Nagasaki blasts. A further nuclear test, widely anticipated to occur in 2012, may give North Korea the expertise to construct a nuclear bomb that could be mated with its short and medium range missiles. It is unclear 389 Lebow and Stein, We All Lost the Cold War, p. 366. See also Richard Ned Lebow, Between Peace and War: The Nature of International Crisis, 1981. Baltimore, Johns Hopkins University Press. 197 whether North Korea’s principal adversary is South Korea, Japan or the United States, so it is hard to specify what technical threshold Pyongyang has to reach for the psychological model’s predictions about biased nuclear learning to occur. While Iranian decision-makers want to challenge American influence in their region, or perhaps expand their influence in the wake of the 2003 Iraq war and recent popular uprisings throughout the Middle East, Kim Jong Un and his senior military advisers likely want to simply preserve a deteriorating status quo. Pyongyang’s economic and political isolation renders it increasingly weak in the face of rising regional competitors. The North Korean economy has had serious economic shortfalls since Soviet support ceased two decades ago, and its heavy dependence on Chinese support almost renders it a Chinese protectorate. These preferences provide incentives for the regime to be weak enough to attract support but strong enough to survive. Beijing is worried about a potential refugee exodus across the border if the regime collapses, and would prefer to feed North Koreans in North Korea than China. South Korea, Japan, China and Russia do not want the regime to collapse. China and Russia like the security buffer it offers against American influence, South Korea does not want to pay the high price for reunification and Japan and China worry about a reunified Korean peninsula. North Korean nuclear coercion is thus likely to be ultimately aimed at extracting economic concessions from regional competitors. The United States will loom large in this effort because only Washington can end the state of war on the Korean peninsula that has remained since 1953. American diplomatic recognition of North Korea would offer a significant diplomatic and hence economic windfall to the regime, and may cause others to recognise the isolated regime. 198 Kim Jong Il may have learned that committing to denuclearise for promises of economic and political concessions is effective. The psychological nuclear learning model indeed predicts that Kim Jong Un and the military leadership will not learn from history that nuclear coercion is rarely effective. When North Korea has a nuclear missile capability capable of targeting either South Korea or the United States mainland, the model predicts that they will attempt a more dangerous strategy that involves coercing Washington to negotiate the diplomatic status of North Korea. This will likely take the form of further provocations against South Korea or missile tests near the United States followed by requests for two party negotiations. The psychological model predicts that this behaviour will continue until the relevant decision-maker experience fear of nuclear escalation or nuclear war. If the latter does not intervene to manage nuclear escalation during a crisis or war, the North Korean leadership will likely learn about the costs and dangers of nuclear coercion. But American intervention to manage escalation might cause Kim and his associates to learn that nuclear coercion offers some benefits. The United States can do more to achieve long term stability when engaging nuclear powers by doing very little. The best American policies will strike the balance between committing to South Korea and Japan’s defence while also committing not to manage a North Korean nuclear crisis. Finally, some might conclude that the psychological learning model has another potential policy implication that is more disturbing than the prescription to not manage new nuclear power crises. There is no a priori theoretical reason, this argument would run, why the effects of fear must be caused by a crisis initiated by a weak revisionist state. Why would we not observe similar effects of fear if the crisis was initiated by the 199 experienced nuclear power or a third party? The possibility that a crisis caused by either of these actors could cause similar effects to crises caused by weak revisionist states suggests a counterintuitive (and disturbing) policy: experienced nuclear powers and third parties should cause a crisis involving their weaker adversaries when they develop nuclear weapons to ensure that the pacifying effects of experience kick in. There are however at least two reasons why such a policy would not follow from this dissertation’s premises. Firstly, if a new nuclear power was not revisionist we would not expect them to challenge the status quo and cause regional instability. Starting a crisis to moderate a new nuclear power’s revisionism where the new nuclear power would not have challenged the status quo is bad policy. It may well cause or intensify the revisionism or hostility that it was designed to deter; weak nuclear power decision-makers may become more aggressive or destabilising to deter perceived imminent aggression. Secondly, even if there was largely unambiguous evidence that a decision-maker or regime had intentions to revise the status quo after developing nuclear weapons, it is not clear that a crisis initiated by a third party would have similar effects. It is not clear that fear of imminent invasion would produce the same degree of risk aversion than fear of nuclear war caused by one’s own coercion. In the former case, backing down in a crisis leads to peace, whereas in the latter there is no guarantee that it will not lead to defeat. We might therefore expect third party induced crises to cause less risk aversion. One might argue that in the face of impending invasion or regime change, decision-makers could adopt pessimistic risk estimates of the probability of nuclear escalatin yet also use tactical or even strategic nuclear weapons to deter the aggression. Such a crisis might therefore risk 200 the nuclear war that fear of nuclear escalation moderates. Third party induced crises would only make sense if the new nuclear power was revisionist and the third party could not only deny involvement but plausibly make decision-makers in the new nuclear power believe that their subordinates were in fact responsible for initiating the crisis. Such a scenario would usually be too implausible. The high probability that the weak new nuclear power would learn who initiated the crisis, as well as the unknown effects of this type of risk aversion, caution against this policy. Chapters 3 and 4 showed that the effects of fear will not be caused by any crisis. Nuclear crises instigated by experienced nuclear powers may have very different effects to those unintentionally sparked by weak revisionists. 6.4. Managing Nuclear Proliferation I have argued that the consequences of nuclear proliferation have a cognitive and temporal logic. Weak revisionist new nuclear powers are war prone but they learn through fear about the dangers of nuclear escalation and exhibit restraint in their nuclear and foreign policies. This challenges the conventional wisdom in two ways. Firstly, nuclear war is less likely than usually thought because fear of nuclear escalation that most people would have to experience before pushing the nuclear button would make them much less likely to do so. Secondly, long term conventional instability is unlikely if the United States can refrain from managing nuclear crises and allow Iranian and North 201 Korean decision-makers to learn that nuclear coercion is dangerous. Nuclear proliferation to states like Iran and North Korea is likely to cause, at most, short term crises or conventional wars. These are obviously undesirable, and most states would be wise to commit large resources to prevent nuclear proliferation and the resulting short term instability. But this dissertation has suggested that the costs of containing this proliferation and short term instability must be weighed against the longer term benefits that arise from the experience of fear of nuclear escalation. This dissertation suggests that if proliferators would have behaved more recklessly in the longer term without nuclear weapons, proliferation may be less undesirable in these cases. As Richard Betts has noted, Waltz’s argument that ‘more may be better’ cannot simply be “brushed off,” yet “surprisingly few academic strategists have tried to refute it in detail.”390 Orthodox wisdom assumes that nuclear proliferation causes long term instability. Inadvertent escalation or organisational biases could, the wisdom asserts, cause nuclear use or war in the short and long term. Preventive wars, nuclear proliferation cascades and nuclear terrorism are other highly undesirable consequences. Moreover, nuclear weapons might embolden revisionist states to pursue regional aggression that would cause substantial long term instability. These consequences are certainly dangerous and undesirable. However their frequency warrants much less alarm than the current wisdom generates; these allegedly destabilising causes of nuclear proliferation are either incorrect or overblown. 390 Richard K. Betts, “Universal Deterrence or Conceptual Collapse? Liberal Pessimism and Utopian Realism,” in Victor Utgoff (ed.) The Coming Crisis: Nuclear Proliferation, U.S. Interests, and World Order, 2000. Cambridge, MA: MIT Press, pp. 51-85, p.60. 202 The conventional wisdom has been preoccupied with nuclear war. The destructive consequences of one occurrence make this a serious threat, but its likelihood may be lower than the wisdom assumes. Nuclear war is always possible, but the absence of the use of nuclear weapons since 1945 suggests that either extreme luck or, more likely, strong pressures that do not subside during international crises have prevented this outcome. This dissertation has suggested that fear of nuclear war may explain why no nuclear crisis has escalated to nuclear war. Preventive motivations to destroy nuclear programs or arsenals before their operational readiness, much heralded as a destabilising cause of nuclear proliferation, have rarely occurred and never – the only possible exception is Israel in 1967 – escalated to war. Much of the orthodox wisdom asserts that the use of a nuclear weapon or dirty bomb by terrorist groups presents an unprecedented threat. But the acquisition of enough enriched uranium or plutonium and the construction of a warhead are much harder for non-state groups to achieve than is often assumed.391 Al Qaeda has now had over a decade to carry out a nuclear September 11 and have failed to do so. Moreover, nuclear forensics deters states from passing nuclear weapons to non- state groups or proxies; the state origins of a nuclear bomb or its enriched uranium could be detected from radiation after the blast. Many have argued that nuclear proliferation causes nuclear proliferation: it threatens other states and causes them to acquire their own arsenals. This logic does explain why many current nuclear powers have nuclear weapons, but it does not explain why many more do not. Most of the orthodox wisdom’s concern is misdirected. Long term initiatives to combat nuclear theft and nuclear proliferation remain necessary, but the costs of such measures should reflect the severity 391 For an argument along these lines, see John E. Mueller, Atomic Obsession: Nuclear Alarmism from Hiroshima to Al-Qaeda, 2009. New York: Oxford University Press. 203 of the threat. The orthodox wisdom tends to assume that the main consequence of nuclear proliferation is nuclear war. But the historical record shows that new nuclear powers are much more likely to find themselves fighting conventional wars that do not escalate to nuclear war. Nuclear weapons have not been used since 1945, but John Mueller’s claim that nuclear weapons have had little impact on international politics is incorrect.392 Nuclear weapons have caused weak revisionist states to fight wars. Indeed, the early post- proliferation years when states learn about nuclear weapons are deeply destabilising and dangerous. The Cold War crisis years and Kargil war period were the most dangerous years in those two enduring rivalries. But nuclear weapons did not produce long-term instability. In the two cases examined in this dissertation, civilian and military decision- makers responsible for authorising nuclear coercion learned of the dangers or costs of nuclear coercion from their own war or crisis and subsequently exhibited restraint. The orthodox wisdom is at its strongest when worrying about emboldenment causing massive long term regional stability. But this concern is also misguided because such instability tends to be a short term but not long term phenomenon. Explicitly modelling the role of time when addressing the consequences of nuclear proliferation offers a framework that moves beyond debates about whether nuclear proliferation is tolerable or intolerable. This dissertation suggests that American nuclear crisis management that contains nuclear escalation will likely cause leaders of weak and revisionist new nuclear powers to learn of the efficacy of nuclear coercion, rather than its dangers, and further destabilise the region. American policy that permits 392 See Mueller, Atomic Obsession. 204 new nuclear powers to learn through fear of the dangers of nuclear coercion will cause them to behave like other experienced nuclear powers. The United States faces an inter- temporal trade-off when engaging new nuclear powers. Washington could refuse to manage a new nuclear power crisis and risk a slightly higher probability of nuclear war in the short term for longer term stability after the effects of fear kick in. The alternative is to reduce the probability of nuclear war in the short term but concomitantly increase the probability of longer term conventional aggression and instability. If the United States established a reputation for managing nuclear crises, weak revisionist nuclear powers as well as their status quo adversaries will be emboldened to challenge the status quo and respond to such challenges aggressively because the US would contain nuclear escalation. The best American policies to address Iran and North Korea after they develop nuclear weapons capable of targeting their principal adversaries will involve robust extended deterrence guarantees to regional allies and a credible commitment to not manage new nuclear power crises. Further research should utilise historical case studies to develop public policies that address how this might be done. Subsequent research should also examine what types of allies of weak revisionist new nuclear powers are also emboldened to pursue risky behaviour. This dissertation has shown that new nuclear powers are emboldened but has not addressed the effects on their allies and other adversaries. 205 6.5. Other Implications In this section I outline some implications of the psychological nuclear learning model for future research in international security and political science. Firstly, research should examine how third parties have influenced the conflict and war propensity of enduring nuclear rivalries. I argued that American crisis management in South Asia may have caused Indian and Pakistani decision-makers to learn that nuclear coercion was effective because Washington would bail them out of crises. Permitting decision-makers to learn through fear would be much more likely to cause long term stability. The obvious policy implication of this is not only to refrain from such crisis management but to abandon extended deterrence guarantees that render such commitments incredible: the United States should wind down its alliances in East Asia and the Middle East so that American adversaries can learn through fear.393 Further research should examine whether extended deterrence policies might be compatible with a desire to allow new nuclear powers to learn through fear of nuclear escalation. Secondly, the findings raise questions of intergenerational learning. If people tend to learn from cognitively accessible events, and the passage of time makes historical nuclear crises less cognitively accessible, the effects of experience with nuclear weapons on revisionist state’s conflict propensity might be undermined by generational change. If weak revisionist states maintain this geopolitical position as generations change, younger 393 For an argument that allowing rebels to win civil wars will have a higher likelihood of producing lasting peace than outside intervention, see Monica Duffy Toft. "Ending Civil Wars: A Case for Rebel Victory?" International Security 34, no. 4 (Spring 2010): 7-36. 206 generations of decision-makers may need to experience the effects of fear of nuclear escalation in their own crisis to perpetuate the effects of experience. Many Japanese citizens who were born in the last three or four decades, for example, tend to be more willing for their government to consider developing nuclear weapons or deploy military forces in regional operations. Older generations tend to be more resistant to these policies. On the other hand, there may be a relationship between the cognitive accessibility of a crisis and the probability that later generations will learn from it. For example, insofar as the Cuban Missile Crisis was a very cognitively accessible crisis, subsequent generations of Russian decision-makers may never again consider nuclear coercion if they faced similar structural challenges to Khrushchev. Thirdly, students of international relations have not established how states deal with and adjust to unacceptable situations. Realism suggests that states should not only fit their resources with their interests but also adjust their interests to their resources. This dissertation has shown that psychological variables can explain how leaders come to adjust to undesirable situations. Decision-makers, forced into a corner, pursue dangerous policies with recently developed nuclear weapons that they experience fear and learn from. Fear causes the adjustment of strategies and often preferences. But we know less about how states generally adapt to untenable strategic environments. Future research should examine whether elites in other contexts undergo similar vicarious learning to adjust to challenging environments.394 394 Robert Jervis refers to this as a “yawning gap in our knowledge.” See Jervis, Deterrence and international relations theory, p.383 207 For example, sudden shocks of economic or even conventional military resources to weak revisionist states could produce initial instability that decision-makers similarly learn from. Increases in economic interdependence could cause the relatively less vulnerable state to coerce the more vulnerable state. If economic coercion is as weak a strategic tool as nuclear coercion, this might cause a crisis, and the crisis may cause fear. Some evidence suggests that China’s recent economic growth and economic interdependence with the United States has emboldened Beijing to challenge American influence in East Asia and pressure Washington to accept Chinese monetary policy.395 What have Chinese decision-makers learned from this? Would further Chinese coercion cause a crisis and fear that Chinese decision-makers would learn from and give rise to a more stable Sino-American balance of power in East Asia? Much of the answer to this question may depend on whether Chinese civilian and military decision-makers are motivated to challenge the East Asian status quo. But increases in economic power or interdependence could cause revisionist preferences. Further research that addresses these questions might also explain when and how rising powers do and do not accommodate their regional status quo. More generally, the psychological nuclear learning model suggests that if people in inter-state conflicts, civil wars or terrorist groups find themselves with no viable options to achieve pressing objectives, they may pursue a risky policy that they, prior to a fear inducing crisis, learn will be successful. This will be especially likely if the group has acquired new technologies or resources. If the resources are weak coercive tools, the psychological learning model would predict dangerous challenges to the status quo until 395 Daniel W. Drezner. "Bad Debts: Assessing China's Financial Influence in Great Power Politics." International Security 34, no. 2 (Fall 2009): 7-45 208 the experience of fear causes learning about the dangers of coercion. This pattern need not be confined to international and intra-state security affairs but might also be useful in explaining cycles of instability and stability in any strategic interaction involving competition over scarce resources, the uneven diffusion of resources, diffuse or weak authority structures and the possibility of very costly outcomes. One obvious context is business competition in market economies. The psychological learning model developed here suggests that if failing firms hit upon a strategy that it unlikely to work, executives will not learn from history of the limits of the strategy but rather from their own experiences of fear. More generally, political scientists should pay more attention to the effects of time and experience with particular technologies or strategies. We should not expect revisionist actors to accurately estimate the coercive power of different resources from systematic or even attentive assessments of the historical record. Moreover, it is likely that status quo actors will suffer from the same biases when assessing the power of different coercive tools. Their increasing power may cause them to pursue the same revisionist coercive strategies that weak revisionists have pursued with nuclear weapons. The effects of the diffusion of weapons, technology and information on states and actors facing similar structural challenges should not be constant but vary according to cognitive predispositions and lessons learned. Future research should examine whether this causes cycles of instability and stability in different interstate and intra-state contests similar to that observed with new and experienced nuclear powers. The experience of fear of destructive or highly undesirable outcomes could transform competitive rivalries and constitute the foundation of long term cooperation. 209 Bibliography: Books and Journal Articles Appelbaum, Anne Gulag: A History, 2003. New York: Doubleday Arkin, R. H. Cooper and T. Kolditz, “A statistical review of the literature concerning the self-serving attribution bias in interpersonal influence situations,” Journal of Personality, 48 1980 pp. 435-448 Basrur, Rajesh M. Minimum Deterrence and India's Nuclear Security, Stanford: Stanford University Press, 2006 Beardsley, Kyle and Victor Asal, “Nuclear Weapons as Shields,” Conflict Management and Peace Science 2009 (26) pp. 235-255 -“Winning with the Bomb,” Journal of Conflict Resolution 53 (2) 2009 pp. 278- 301 Bennett, Andrew O. Condemned to Repetition? The Rise, Fall and Reprise of Soviet- Russian Military Intervention, 1973-1996. MIT Press: Cambridge, Massachusetts, 1999. Bennett D. Scott and Allan C. Stam, The Behavioural Origins of War, 2003. Michigan: University of Michigan Press Bernstein, Barton J. “Reconsidering the Missile Crisis: Dealing with the Problems of the American Jupiters in Turkey,” in James A. Nathan (ed.) The Cuban Missile Crisis Revisited. New York, St Martin’s, 1992 Beschloss, Michael R. Reaching for Glory: Lyndon Johnson's Secret White House Tapes, 1964-65, 2001. New York: Touchstone -Taking Charge: The Johnson White House Tapes, 1963-64, 1997. New York: Touchstone -The Crisis Years: Kennedy and Khrushchev, 1960-1963. Harper Collins, 1991 Betts, Richard K. “Universal Deterrence or Conceptual Collapse? Liberal Pessimism and Utopian Realism,” in Victor Utgoff (ed.) The Coming Crisis: Nuclear Proliferation, U.S. Interests, and World Order, 2000. Cambridge, MA: MIT Press, pp. 51-85 -Nuclear Blackmail and Nuclear Balance, 1987. Washington D.C..: Brookings Institute Press 210 Blight, James G. Bruce J. Allyn and David A. Welch, Cuba on the Brink: Castro, The Missile Crisis, and the Soviet Collapse. New York, Pantheon, 1993 Blight, James G. and David A. Welch, On the Brink: Americans and Soviets Reexamine the Cuban Missile Crisis. New York: Hill and Wang, 1989 Broomfield, Lincoln Walter C. Clemens, Jr., and Franklyn Griffiths, Khrushchev and the Arms Race: Soviet Interests in Arms Control and Disarmament 1954-1962, 1966. Cambridge: Massachusetts, M.I.T. Press Bundy, McGeorge Danger and Survival: Choices About the Bomb in the First Fifty Years, 1988. New York: Random House Burke, John P. and Fred I. Greenstein, How President's Test Reality, Decisions on Vietnam, 1954 and 1965, 1989. New York: Russell Sage Byman, Daniel “Iran, Terrorism and Weapons of Mass Destruction,” in Studies in Conflict and Terrorism, 2008, 31 pp. 169-181 Byman, Daniel, Shahram Chubin, Anoushiravan Ehteshami and Jerrold Green, Iran’s Security Policy in the Post-Revolutionary Era, 2001. Santa Monica: RAND Catudal, Honore M. Kennedy and the Berlin Wall Crisis: A Case Study in U.S. Decisionmaking, 1980 Berlin: Berlin-Verlag Chapman, L. J. “Illusory correlation in observational report,” Journal of Verbal learning and Verbal Behaviour, 6 1967 pp.151-55 Chapman, L.J. and J. P. Chapman, “Illusory Correlation as an Obstacle to the Use of Valid Psychodiagnostic Signs,” Journal of Abnormal Psychology, Vol. 74, No. 3 June 1969 pp. 271-80 Chari, P.R. Pervaiz Iqbal Cheema and Stephen P. Cohen, Four Crises and a Peace Process: American Engagement in South Asia. Washington D.C.: Brookings, 2007 Chubin, Shamram “Iran’s Risk Taking in Perspective,” IFRI Security Studies Centre, Proliferation Papers, 2008 Clinton, Bill My Life: The Presidential Years. New York: Vintage, 2004 Cohen, Stephen P. India: Emerging Power, 2001. Washington, DC: Brookings Institution Press Cordesman,Anthony Iran’s Evolving Threat, Center for Strategic and International Studies, 2010 211 -Iran, The Gulf, and Strategic Competition: The Conventional Balance, Center for Strategic and International Studies, August 2010 Costigliola, Frank “Lyndon B. Johnson, Germany and the End of the Cold War,” in Nancy Bernkopf Tucker and Warren I Cohen, (eds.) Lyndon Johnson Confronts the World: American Foreign Policy, 1963-68. 1994, pp. 173-210. New York, Cambridge University Press Cousins, Norman The Improbable Triumvirate: John F. Kennedy, Pope John, Nikita Khrushchev, 1972. New York, Norton Crail, Peter, Daniel Horner and Daryl G. Kimball, “Pursuing the Prague Agenda: An Interview with White House Coordinator Gary Samore,” Arms Control Today, May 2011. http://www.armscontrol.org/act/2011_05/Samore, accessed May 19 2011 Crescenzi, Mark J. C. “Reputation and Interstate Conflict,” American Journal of Political Science, 51 (2) April 2007 pp. 382-396 Dallek, Robert, Flawed Giant: Lyndon Johnson and His Times, 1961-1973, 1998. New York, Oxford University Press -Lone Star Rising: Lyndon Johnson and His Times, 1908-1960, 1991. New York: Oxford University Press DeSteno, David, Richard E. Petty, Duane T. Wegener and Derek D. Rucker, “Beyond valence in the perception of likelihood: The role of emotion specificity,” Journal of Personality and Social Psychology, Vol. 78 (3), March 2000, 397-416 Dobrynin, Anatoly In Confidence: Moscow’s Ambassador to America’s Six Cold War Presidents, 1962-1986, 1995. New York: Times Books Drezner, Daniel W. "Bad Debts: Assessing China's Financial Influence in Great Power Politics." International Security 34, no. 2 (Fall 2009): 7-45 Dumbrell, John President Lyndon Johnson and Soviet Communism, 2004. Manchester: University of Manchester Press Edelstein, David “Time Horizons and the Causes of Conflict and Cooperation,” American Political Science Association 2010 Annual Meeting paper Elman, Colin and John Vasquez (eds.), Realism and the Balance of Power: A New Debate, 2002. New York: Prentice Hall Enthoven, Alain C. and K. Wayne Smith, How Much is Enough? Shaping the Defense Program, 1961-1969. Harper and Row, 1971 212 Evangelista, Matthew, “Why Keep Such an Army? Khrushchev's Troop Reductions,” Working Paper 19, Cold War International History Project (hereafter referred to as CWIHP) Bulletin, December 1997 -“Stalin’s Postwar Army Reappraised,” International Security, Vol. 7, No. 3, (Winter 1982-83), pp. 110-138 Fearon, James D. “Rationalist Explanations for War,” International Organization 1995 (Summer) pp. 379-414 -“Domestic Political Audiences and the Escalation of International Disputes,” American Political Science Review, Volume 88 (3) 1994 pp. 577-592 -“Signalling Versus the Balance of Power and Interests: An Empirical Test of a Crisis Bargaining Model,” Journal of Conflict Resolution 1994 (February) pp. 236-69 Fearon, James and Alexander Wendt, “Rationalism v. Constructivism: A Skeptical View,” in Walter Carlsnaes et al, Handbook of International Relations, 2005. Sage: California, pp. 52-72 Feaver, Peter D. “Optimists, Pessimists, and Theories of Nuclear Proliferation Management,” Security Studies 4 (4) 1995 pp. 754-772 -Guarding the Guardians: Civilian Control of Nuclear Weapons in the United States, 1992. Ithaca: Cornell University Press Feaver, Peter D. and Emerson Niou, “Managing Nuclear Proliferation: Condemn, Strike or Assist?,” International Studies Quarterly, 1996 40 pp.209-234 Filson, Darren and Suzanne Werner, “A Bargaining Model of War and Peace,” American Journal of Political Science 2002 46 pp. 819-38 Finnemore, Martha and Kathryn Sikkink, “International Norm Dynamics and Political Change,” International Organisation 52 1998 pp. 887-917 Fischhoff, Baruch, Roxana M. Gonzalez, Jennifer S. Lerner, and Deborah A. Small, (2005) “Evolving judgments of terror risks: Foresight, hindsight, and emotion,” Journal of Experimental Psychology: Applied, 11, 124-139. Fiske, Susan T. and Shelley E. Taylor, Social Cognition: From Brains to Culture, 2008. New York: McGraw-Hill Fitzpatrick, Mark and Michael Elleman, Iran’s Ballistic Missile Capabilities: A net assessment, 2010. London: International Institute for Strategic Studies. 213 Fursenko, Alexsandr and Timothy Naftali, Khrushchev’s Cold War, 2007. New York: Norton. - One Hell of a Gamble: Khrushchev, Castro and Kennedy, 1958-1964, 1998. New York: Norton Ganguly, Sumit “Nuclear Stability in South Asia,” International Security 33 (2) Fall 2008 pp. 45-70 -The Crisis in Kashmir: Portents of War, Hopes of Peace, 1997. New York: Cambridge University Press Ganguly, Sumit and Devin T. Hagerty, Fearful Symmetry: India-Pakistan Crises in the Shadow of Nuclear Weapons, 2005. Seattle, University of Washington Press Gartzke, Erik “Nuclear Proliferation Dynamics and Conventional Conflict,” available at http://dss.ucsd.edu/~egartzke/papers/nuketime_05032010.pdf Gartzke, Erik and Dong Joon Jo, “Bargaining, Nuclear Proliferation and Interstate Disputes” Journal of Conflict Resolution 53 (2) 2009 pp. 209-233 Geller, Daniel “Nuclear Weapons, Deterrence and Crisis Escalation,” Journal of Conflict Resolution 34 (1990), pp. 291-310 George, Alexander L. and Andrew Bennett, Case Studies and Theory Development in the Social Sciences, 2005. Cambridge: MIT Press Germany, Kent et al (eds.) The Presidential Recordings: Lyndon B. Johnson, Volume III, 2005. New York: Norton Gilovich, Thomas D. and Dale W. Griffin, “Judgement and Decision Making,” in Susan T. Fiske, Daniel T. Gilbert and Gardner Lindzey (eds.), Handbook of Social Psychology: Volume 1, (Fifth edition), 2010. New York: Wiley, pp. 542-588 Ginor, Isabella and Gideon Remez, Foxbats Over Dimona: The Soviets’ Nuclear Gamble in the Six-Day War. New Haven: Yale University Press, 2007 Gupta, Sonika, and Arpit Rajain, “An interview with General V. P. Malik,” in P.R. Chari, Sonika Gupta and Arpit Rajain (eds.), Nuclear Stability in Southern Asia. New Delhi: Manohar, 2003 Guthman, Edwin and Jeffrey Shulman, RFK in his own words, 1991. Gribkov, Anatoli I. “The View from Moscow and Havana,” in Operation ANADYR: U.S. And Soviet Generals Recount the Cuban Missile Crisis, ed. Alfred Friendly, Jr, 1994. Chicago: Edition Q, pp. 3-76 214 Goldstein, Lyle J. “Return to Zhenbao Island: Who Started Shooting and Why it Matters,” The China Quarterly 2001 No. 168 pp.985-997 Goodwin, Doris Kearns Lyndon Johnson and the American Dream, 1991. New York: St Martin's Hall, Peter “Policy Paradigms, Social Learning and the State: The Case of Economic Policy Making in Britain,” Comparative Politics 1993 25 (3) pp.275-296 Harrison, Hope M. Driving the Soviets up the Wall: Soviet-East German Relations, 1953- 1961, 2003. New Jersey: Princeton University Press - Harrison, “Ulbricht and the Concrete Rose: New Archival Evidence on the Dynamics of Soviet-East German Relations and the Berlin Crisis,” CWIHP Working Paper 5 Hilsman, Roger To Move a Nation: The Politics of Foreign Policy in the Administration of John F. Kennedy. New York: Doubleday, 1967 Hopf, Ted “The Promise of Constructivism in International Relations Theory,” International Security 23 (1) 1998 pp. 171-200 Horowitz, Michael “The Spread of Nuclear Weapons and International Conflict: Does Experience Matter?” Journal of Conflict Resolution 2009 53 No. 2 (April) pp. 234-257 Huth, Paul D. Scott Bennett and Christopher Gelpi, “System Uncertainty, Risk Propensity, and International Conflict among the Great Powers,” Journal of Conflict Resolution, September 1992, Vol 36 No. 3 pp. 478-517 Hymans, Jacques E. C. The Psychology of Nuclear Proliferation: Identity, Emotions, and Foreign Policy, 2006. Cambridge: Cambridge University Press Janis, Irving L. and Leon Mann, Decision Making: A Psychological Analysis of Conflict, Choice and Commitment, 1977. New York: Free Press Jarosz, William W. and Joseph N. Nye, Jr, “The Shadow of the Past: Learning from History in National Security Decision Making,” in Philip E. Tetlock et al, (eds.) Behaviour, Society and International Conflict, 1993, p. 130, 180 New York: Oxford University Press Jervis, Robert “Unipolarity: A Structural Perspective,” in World Politics 2009 61 (1) pp. 188-213 - “Kargil, Deterrence and International Relations theory,” in Peter Lavoy (ed.) Asymmetric Warfare in South Asia: The Causes and Consequences of the Kargil Conflict,” 2009, pp. 377-397. New York: Cambridge University Press 215 -“Was the Cold War a Security Dilemma?” Journal of Cold War Studies 3 (1) 2001 pp. 36-60 -“Why Nuclear Superiority Doesn't Matter,” Political Science Quarterly 1980 94 (4) -Perception and Misperception in International Politics, 1976. New Jersey: Princeton University Press Johnson, Eric J. and Amos Tversky, “Affect, Generalization, and the Perception of Risk,” Journal of Personality and Social Psychology, 1983, Vol. 45, No. 1, 20-31 Johnson, Lady Bird A White House Diary, 2007. Austin: University of Texas Press Johnson, Lyndon B. The Vantage Point: Perspectives of the Presidency, 1963-69, 1971. New York: Holt, Rinehart and Winston Kahneman, Daniel, Paul Slovic and Amos Tversky, Judgment under Uncertainty: Heuristics and Biases, 1982. Cambridge, U.K.: Cambridge University Press Kaiser, David American Tragedy: Kennedy, Johnson, and the Origins of the Vietnam War, 2000. Belknap: Cambridge, Harvard University Press Kampani, Gaurav “India’s Compellence Strategy: Calling Pakistan’s Nuclear Bluff over Kashmir,” Monterey Institute of International Studies Kapur, S. Paul “Revisionist Ambitions, Capabilities, and Nuclear Instability,” in Scott D. Sagan (ed.), Inside Nuclear South Asia, 2009. Stanford: Stanford University Press, pp.184-218 -“Ten Years of Instability in Nuclear South Asia,” International Security 33 (2) 2008 pp. 71-94 -Dangerous Deterrent: Nuclear Weapons Proliferation and Confllict in South Asia. Stanford: Stanford University Press, 2007 Karabell, Zachary Parting the Desert: The Creation of the Suez Canal, 2003. New York: Knopf Karl, David J. “Proliferation Pessimism and Emerging Nuclear Powers,” International Security 21 1997 pp. 87-119 Katzenstein, Peter (ed.), The Culture of National Security, 1996. New York: Columbia University Press 216 Keltner, Dacher, Phoebe C. Ellsworth and Kari Edwards, “Beyond Simple Pessimism: Effects of Sadness and Anger on Social Perception,” Journal of Personality and Social Psychology (1993) Vol. 64 No. 5 740-752 Kennedy, Robert Thirteen Days: A Memoir of the Cuban Missile Crisis, 1969. New York, Norton Khan, Feroz Hassan, Peter R. Lavoy and Christopher Clary, “Pakistan's motivations and calculations for the Kargil conflict,” in Peter Lavoy (ed.) Asymmetric Warfare in South Asia: The Causes and Consequences of the Kargil Conflict, 2009. New York: Cambridge University Press Khong, Yuen Foong Analogies at War, 1992. New Jersey: Princeton University Press. Khrushchev, Nikita Memoirs of Nikita Khrushchev, Volume 3: Statesman, 1953-1964, 2004. Edited by Sergei Khrushchev, Translated by George Shriver. University Park: Pennsylvania State University Press Khrushchev, Sergei Nikita Khrushchev and the Creation of a Superpower, 2000, trans. by Shirley Benson. University Park: Pennsylvania State University Press -Khrushchev on Khrushchev: An Inside Account of the Man and his Era, 1990. Little, Brown and Company Knopf, Jeffrey “The Importance of International Learning,” Review of International Studies (2003) 29, pp. 185-207 Kroenig, Matthew “Nuclear Superiority or the Balance of Resolve? Explaining Nuclear Crisis Outcomes,” http://www.matthewkroenig.com/Kroenig_Nuclear%20Superiority_July2010.pdf Kuisong, Yang “The Sino-Soviet Border Clash of 1969: from Zhenbao Island to Sino- American rappochement,” Cold War History, No. 1 August 2000, pp. 21-51 Kunz, Diane The Economic Diplomacy of the Suez Crisis, 1991. Chapel Hill: University of North Carolina Press Ladwig, Walter C., III, “A Cold Start for Hot Wars? The Indian Army's New Limited War Doctrine,” in International Security 2008 32 (3) Lake, Eli “Operation Sabotage: Our secret war against Iran,” The New Republic, July 22 2010, Vol. 241 Issue 12 pp. 16-17 Lavoy, Peter “Why Kargil did not produce general war: the crisis-management strategies of Pakistan, India, and the United States,” in Lavoy (ed.) Asymmetric Warfare in South 217 Asia: The Causes and Consequences of the Kargil Conflict, 2009. New York: Cambridge University Press, pp 171-206 Lebow, Richard Ned Between Peace and War: The Nature of International Crisis, 1981. Baltimore, Johns Hopkins University Press Lebow, Richard Ned and Janice Gross Stein, We All Lost the Cold War, 1994. New Jersey: Princeton University Press Leng, Russell J. “Realpolitik and learning in the India-Pakistan rivalry,” in T.V. Paul, (ed.) The India-Pakistan Conflict: An Enduring Rivalry, 2005. New York: Cambridge University Press, pp. 103-127 - Bargaining and Learning in Recurring Crises: The Soviet-American, Egyptian- Israeli and Indo-Pakistani Rivalries, 2000. Ann Arbor: University of Michigan Press Lerner, Jennifer and Dacher Keltner, “Fear, Anger and Risk,” Journal of Personality and Social Psychology, Vol. 81 No. 1 2001 pp. 146-159 -“Beyond valence: Toward a model of emotion-specific influences on judgment and choice,” Cognition and Emotion, (2000) 14, 473-493 Lerner, Jennifer, Roxana M. Gonzalez, Deborah A. Small and Baruch Fischhoff, “Effects of fear and anger on perceived risks of terrorism: A national field experiment,” Psychological Science, (2003) 14, 144-150 Levy, Jack S. “Learning and Foreign Policy: Sweeping a Conceptual Minefield,” International Organization 1994 48 (2) Lindsay, James M. and Ray Takeyh, “After Iran Gets the Bomb: Containment and its Complications,” Foreign Affairs , March-April 2010, pp. 33-50 Logevall, Fredrik, Choosing War: The Lost Chance for Peace and the Escalation of the War in Vietnam, 2001. Berkeley: University of California Press McDermott, Rose “Arms Control and the First Reagan Administration: Belief-Systems and Policy Choices,” Journal of Cold War Studies, Volume 4, Number 4, Fall 2002, pp. 29-59 -Risk Taking in International Politics: Prospect Theory in American Foreign Policy, 1998. Ann Arbor: University of Michigan Press McNamara, Robert In Retrospect: The Tragedy and Lessons of Vietnam, 1995. New York: Vintage 218 MacFarquhar, Roderick and Michael Schoenhals, Mao’s Last Revolution, 2008, Harvard University Press Malik, V.P. Kargil: From Surprise to Victory, 2006. India: Harper Collins -“Lessons from Kargil,” Bharat Rakshak Monitor 4, No. 6 (May-June 2002), www.bharat-rakshak.com/MONITOR/ISSUE4-6/malik.html May, Ernest R. and Philip D. Zelikow, eds., The Kennedy Tapes: Inside the White House During the Cuban Missile Crisis, 2002. New York: W. W. Norton Mearsheimer, John J. “Numbers, Strategy and the European Balance,” in International Security, Vol. 12, No. 4, (Spring 1988) pp. 174-185 Mercer, Jonathan Reputation and International Politics, 1996. Ithaca: Cornell University Press Mikoyan, Anastas Tak Bylo [As It Was], 1994. Moscow: Vagrius Milani, Mohsen M. “Tehran’s Take,” Foreign Affairs July/August 2009 Vol. 88 Issue 4 pp. 46-62 Miller, D. and M. Ross, “Self-serving biases in the attribution of causality: Fact or fiction?,” Psychological Bulletin 82 1975 pp. 213-225 Montgomery, Alexander H. and Scott D. Sagan “The Perils of Predicting Proliferation,” Journal of Conflict Resolution, 2009 Vol. 53 (2) April pp. 320-328 Mueller, John E. Atomic Obsession: Nuclear Alarmism from Hiroshima to Al-Qaeda, 2009. New York: Oxford University Press. Mullen B. and C. A. Riordan, “Self-serving attributions for performance in naturalistic settings; A meta-analytic review,” Journal of Applied Social Psychology 18 1988 pp. 3- 22 Musharraf, Pervez In the Line of Fire: A Memoir, 2006. New York: Free Press Naftali, Timothy “Author's Response,” in Khrushchev's Cold War: The Inside Story of an American Adversary, H-Diplo Review, p.37. http://www.h-net.org/~diplo/roundtables/PDF/KhrushchevsColdWar-Roundtable.pdf; accessed August 19, 2010 Narang, Vipin “Posturing for Peace? Pakistan's Nuclear Postures and Nuclear Stability,” in International Security 34 (3) 2009 pp. 38-78 219 Nash, Philip The Other Missiles of October: Eisenhower, Kennedy and the Jupiters, 1957-1963, 1997. Chapel Hill: University of North Carolina Press Nawaz, Shuja Crossed Swords: Pakistan, Its Army and the Wars Within, 2008. Oxford: Oxford University Press Nayak, Polly and Michael Krepon, U.S. Crisis Management in South Asia's Twin Peaks Crisis, Henry L. Stimson Centre, Report No. 57, September 2006 Nisbett, Richard and Lee Ross, Human Inference: Strategies and Shortcomings in Social Judgement, 1980. Prentice Hall Ohman, Arne “Fear and Anxiety: Overlaps and Dissociations,” in Michael Lewis, Jeannette M. Haviland-Jones and Lisa Feldman Barrett (eds.), Handbook of Emotions, Third Edition, 2008. New York, Guilford Press, pp. 709-729 Ohman, Arne and Susan Mineka, “Fears, Phobias and Preparedness: Toward an Evolved Module of Fear and Fear Learning,” Psychological Review, Vol. 108 No. 3 2001 pp. 483- 522 Paul, T.V. “Nuclear Weapons and Asian Security in the Twenty-first Century,” in N. S. Sisodia, V. Krishnappa and Priyanka Singh, Proliferation and Emerging Nuclear Order in the Twenty-first Century, 2009. New Delhi: Institute for Defence Studies and Analyses, pp. 29-43 - The Tradition of Non-Use of Nuclear Weapons, 2009. Stanford: Stanford University Press -Power Versus Prudence: Why Nations Forgo Nuclear Weapons, 2000. Montreal and Kingston: McGill-Queens University Press. Perkovich, George India`s Nuclear Bomb: The Impact on Global Proliferation, 2001. New York: Oxford University Press Pleshakov, Constantine “Studying Soviet Strategies and Decisionmaking in the Cold War Years,” in Odd Arne Westad (ed.) Reviewing the Cold War: Approaches, Interpretations, Theory, 2000. Frank Cass: London, pp. 232-241 Podvig, Pavel (ed.) Russian Strategic Nuclear Forces. Massachusetts: The MIT Press, 2001 Poirier, Dale “Frequentist and Subjectivist Perspectives on the Problems of Model Building in Economics,” Journal of Economic Perspectives, Vol. 2, No. 1 (Winter 1988) pp. 121-144 220 Posen, Barry “Is NATO Decisively Outnumbered?,” in International Security Vol. 12 No. 4 (Spring 1988), pp. 186-202 -“Measuring the European Conventional Balance: Coping with Complexity in Threat Assessment,” in International Security, Vol. 9, No. 3 (Winter 1984-1985), pp. 47-88 Potter, Phillip “Does Experience Matter? American Presidential Experience, Age and International Conflict,” Journal of Conflict Resolution 2007 51 (3) pp.351-379 Powell, Robert “Bargaining and Learning While Fighting,” American Journal of Political Science, 48 (2), April 2004, pp.344-361 -In the Shadow of Power, 1999. Princeton: Princeton University Press Prados, John “Assessing Dienbienphu,” in Mark Atwood Lawrence and Fredrik Logevall, The First Vietnam War: Colonial Conflict and Cold War Crisis, 2007. Cambridge: Harvard University Press -The Hidden History of the Vietnam War, 1995. Chicago: Ivan R. Dee Press, Daryl Calculating Credibility: How Leaders Assess Military Threats, 2005. Ithaca: Cornell University Press Price, Richard M. “Reversing the Gun Sights: Transnational Civil Society Targets Landmines,” International Organisation 1998 52 pp. 613-644 Price Richard M. and Nina Tannenwald, “Norms and Deterrence: The Nuclear and Chemical Weapons Taboos,” in Peter Katzenstein, ed. The Culture of National Security, 1996. New York, Columbia University Press Qadir, Shaukat “An Analysis of the Kargil Conflict 1999,” The RUSI Journal, 147 (2) 2002 pp. 24-30 Raas, Whitney and Austin Long, “Osirak Redux? Assessing Israeli Capabilities to Destroy Iranian Nuclear Facilities,” International Security 31 (4) 2007 pp. 7-33 Raghavan, V.R. Siachen: Conflict Without End, 2002. Delhi: Viking books -“Limited War and Nuclear Escalation in South Asia,” The Nonproliferation Review 8 (3) 2001 pp.82-98 Rapport, Aaron and Ronald R. Krebs, “International Relations and the Psychology of Time Horizons,” American Political Science Association 2010 Annual Meeting paper 221 Rauchhaus, Robert “Evaluating the Nuclear Peace Hypothesis: A Quantitative Approach,” Journal of Conflict Resolution 53 (2) 2009 pp. 258-277 Reiter, Dan How Wars End, 2009. New Jersey: Princeton University Press -Crucible of Beliefs: Learning, Alliances and World Wars, 1996. Ithaca: Cornell University Press Richter, James G. Khrushchev's Double Bind: International Pressures and Domestic Coalitional Politics, 1994. Baltimore: Johns Hopkins University Press Riedel, Bruce American Diplomacy and the 1999 Kargil Summit at Blair House. Center for the Advanced Study of India, University of Pennsylvania, 2002 Sagan, Scott D. “How to Keep the Bomb from Iran,” Foreign Affairs, Vol. 85, No. 5, Sep-Oct 2006, pp. 45-59 -The Limits of Safety: Organisations, Accidents and Nuclear Weapons, 1993. Princeton: Princeton University Press Salinger, Pierre With Kennedy, 1966. Garden City, N.Y.: Doubleday Salisbury, Harrison E. A New Russia?, 1962. New York: Harper and Row Schecter, Jerold L. (ed.) Khrushchev Remembers: The Glasnost Tapes, 1990. Boston: Little Brown Schlesinger, Arthur, Jr., Robert Kennedy and His Times, 1978. Boston: Houghton Mifflin Schultz, Kenneth “Do Democratic Institutions Constrain or Inform?” International Organization 1999 53 (Spring) pp. 23-66 Seaborg, Glenn T. Kennedy, Khrushchev and the Test Ban, 1981. Berkeley, University of California Press Sechser, Todd and Matthew Fuhrmann, “The Coercive Limits of Nuclear Weapons,” presented at the International Studies Association Annual Meeting, Montreal, 2011 Selvage, Douglas “The End of the Berlin Crisis: New Evidence from the Polish and East German Archives,” CWIHP Bulletin 11, 1998, pp. 218-229 -“Khrushchev's November 1958 Berlin Ultimatum: New Evidence from the Polish Archives,” CWIHP Bulletin 11, 1998, pp.200-203 Short, Philip Mao: A Life. 2004, John Murray 222 Sidhu, Waheguru Pal Singh “Of Myths and Realities: The Kargil Experience,” in Kanti Bajpai, Afsir Karim and Amitabh Mattoo (eds.) Kargil and After: Challenges for Indian Policy. New Delhi: Har-Anand Publications, 2001, pp. 107-124 Slantchev, Branislav “The Principle of Convergence in Wartime Negotiations,” American Political Science Review 97 (4) November 2003, pp. 621-632 -“The Power to Hurt: Costly Conflict with Completely Informed States,” American Political Science Review 2002 (47) pp. 1-11 Slovic, Paul, Baruch Fischhoff and S. Lichtenstein, “Cognitive processes and societal risk taking,” in J. S. Carroll and J. W. Payne (eds.), Cognition and social behaviour, 1976, pp. 165-184. New Jersey: Erlbaum Smith, Alastair “Fighting Battles, Winning Wars,” Journal of Conflict Resolution, 1998 42 (June) pp. 301-320 -“International Crises and Domestic Politics,” American Political Science Review 92 (3) 1998 pp. 623-638 Smith, Alastair and Allan C. Stam, “Bargaining and the Nature of War,” Journal of Conflict Resolution 2004 48 (6) pp. 783-813 Smith, Craig A. and Phoebe C. Ellsworth, “Patterns of cognitive appraisal in emotion,” Journal of Personality and Social Psychology, 48 (1985) 813-838 Snyder, Glenn H “The Balance of Power and the Balance of Terror,” in Paul Seabury (ed.) Balance of Power, 1965 Chandler: San Francisco, pp. 184-201 Solingen, Etel Nuclear Logics: Contrasting Paths in East Asia and the Middle East, 2006. Princeton: Princeton University Press Sood, V.K. and Pravin Sawhney, Operation Parakram: The War Unfinished, 2003. New Delhi: Sage Stein, Janice Gross “Political Learning and Political Psychology: A Question of Norms,” in Kristen Renwick Monroe (ed.) Political Psychology, 2002. New Jersey: Lawrence Erlbaum Associates, pp. 107-115 -“Political Learning by Doing: Gorbachev as Uncommitted Thinker and Motivated Learner,” International Organization 1994 48 (2) Streich, Philip and Jack S. Levy, “Time Horizons, Discounting, and Intertemporal Choice,” Journal of Conflict Resolution, 2007 51 (2) pp. 199-226 223 Swami, Praveen “A war to end a war: The causes and outcomes of the 2001-2 India- Pakistan crisis,” in Sumit Ganguly and S. Paul Kapur (eds.) Nuclear Proliferation in South Asia: Crisis Behaviour and the Bomb, 2009, pp. 144-163. New York: Routledge Talbot, Ian Pakistan: A Modern History, 1999. New Delhi: Oxford University Press Talbott, Strobe Engaging India: Diplomacy, Democracy and the Bomb. Washington: Brookings, 2004 -(ed.) Khrushchev Remembers, 1970. Boston: Little Brown. Talmadge, Caitlin “Closing Time: Assessing the Iranian Threat to the Strait of Hormuz,” International Security 33 (1) 2008 pp. 82-117 Tannenwald, Nina The Nuclear Taboo: The United States and the Non-Use of Nuclear Weapons since 1945, 2008. New York: Cambridge University Press Taubman, William Khrushchev, 2003. New York: Norton Tellis, Ashley, C. Christine Fair and Jamison Jo Medby, Limited Conflicts Under the Nuclear Umbrella: Indian and Pakistani Lessons from the Kargil Crisis. Santa Monica, RAND: 2002 Tetlock, Philip Expert Political Judgement: How Good Is It? How Can We Know?, 2005. New Jersey, Princeton University Press -“Social Psychology and World Politics,” in Daniel T. Gilbert et al, The Handbook of Social Psychology, Fourth Edition, 1998, pp.868-914. New York: McGraw-Hill -“Learning in U.S. And Soviet Foreign Policy: In Search of an Elusive Concept,” in George Breslauer and Philip Tetlock (eds.) Learning in U.S. And Soviet Foreign Policy, 1991. Boulder: Westview Press Thompson, William J. Khrushchev: A Political Life, 1997. New York: St Martin’s Griffin Tiedens, Larissa Z. and Susan Linton, (2001) “Judgement under emotional certainty and uncertainty: The effects of specific emotions on information processing,” Journal of Personality and Social Psychology, 81, 973-988 Toft, Monica Duffy "Ending Civil Wars: A Case for Rebel Victory?" International Security 34, no. 4 (Spring 2010): 7-36 Tomz, Michael Reputation and International Cooperation: Sovereign Debt Across Three Centuries, 2007. New Jersey: Princeton University Press 224 Trachtenberg, Marc A Constructed Peace: The Making of the European Settlement, 1945- 1963, 1999. New Jersey, Princeton University Press Trager, Robert F. “Diplomatic Calculus in Anarchy: How Communication Matters,” American Political Science Review 104 (2) 2010 pp. 347-368 Uhl, Matthias and Vladimir I. Ivkin, “‘Operation Atom’: The Soviet Union’s Stationing of Nuclear Missiles in the German Democratic Republic, 1959,” CWIHP Bulletin, issue 12/13 (Winter-Spring 2001), pp. 299-307 Volkogonov, Dmitri Autopsy for an Empire: The Seven Leaders Who Built the Soviet Regime, 1998. New York: Free Press Wagner, R. Harrison “Bargaining and War,” American Journal of Political Science 2000, 44 (July) pp. 469-484 -“Peace, War and the Balance of Power,” American Political Science Review 1994 88 (September) pp. 593-607 -“Uncertainty, Rational Learning and Bargaining in the Cuban Missile Crisis,” in Peter C. Ordeshook (ed.) Models of Strategic Choice in Politics, 1989. Ann Arbor: University of Michigan Press, pp.177-205 Waltz, Kenneth N. “Nuclear Myths and Political Realities,” American Political Science Review 84 (3) 1990 -Theory of International Politics, 1979. New York: Mcgraw-Hill Waltz, Kenneth N and Scott D. Sagan, The Spread of Nuclear Weapons: A Debate Renewed, 2003. New York: Norton Wendt, Alexander Social Theory of International Politics, 1999. New York: Cambridge University Press Wicker, Tom JFK and LBJ: The Influence of Personality upon Politics, 1991. Chicago: Ivan R. Dee Wirsing, Robert G. India, Pakistan and the Kashmir Dispute: On Regional Conflict and Its Resolution, 1998. New York: St Martin's Zaloga, Steven J. The Kremlin's Nuclear Sword: The Rise and Fall of Russia's Strategic Nuclear Forces, 1945-2000. Washington: Smithsonian Institute Press, 2002 Zubok, Vladislav A Failed Empire: The Soviet Union in the Cold War from Stalin to Gorbachev, 2007. Chapel Hill: University of North Carolina Press 225 -“The Mao-Khrushchev Conversations, 31 July – 3 August 1958 and 2 October 1959,” CWIHP Bulletin, 2001, Issue 12-13, pp.244-272 Zubok, Vladislav M. and Constantine Pleshakov, Inside the Kremlin's Cold War: From Stalin to Khrushchev, 1996. Cambridge, MA: Harvard University Press 226 Newspaper Articles William J. Broad, John Markoff and David E. Sanger, “Israeli Test on Worm Called Crucial in Iran Nuclear Delay,” New York Times, January 15, 2011 David E. Sanger and William J. Broad, U.N. Says Iran Has Fuel for 2 Nuclear Weapons, New York Times, May 31 2010 Wall Street Journal, April 6 2010, p.11 Barry R. Posen, “We Can Live With a Nuclear Iran,” New York Times, February 27, 2006 K.J.M. Varma, “Pak Not Ready to Sideline Kashmir: Musharraf,” Press Trust of India, June 17, 2003 Praveen Swami, “Beating the Retreat,” Frontline, October 26-November 8, 2002 Luv Puri, “Be Ready for Decisive Battle, PM Tells Jawans,” The Hindu, May 23, 2002 K. Subrahmanyam, “Indo-Pak Nuclear Conflict Unlikely,” The Times of India, 2 January 2002 Peter Popham, ‘The World’s most dangerous place’ is already at war,’ The Independent: Asia, March 18 2000 Ramesh Chandran, “Vajpayee and Musharraf agree on one point - No nuclear war,” The Times of India, 13 March 2000 C. Raja Mohan, “Fernandes Unveils 'Limited War' Doctrine,” The Hindu, January 25 2000 Prakash Karat, “Kargil and Beyond,” Frontline, www.frontlineonnet.com/fl1614/16140150.htm Harry Schwartz, “Soviet Wipes out Promise to Raise Meet Allowances,” New York Times, March 4, 1963 Lyndon B. Johnson, “We Are Ahead of Russia – We Can Stay Ahead,” Preview Magazine, December 1950, Houston, Texas, pp. 12, 40 227 Archives National Security Archive, Washington D.C. Lyndon Baines Johnson Presidential Library, Austin, Texas Archival Collections Miller Centre for Public Affairs, Kremlin Decision-Making Project United States Department of State, Foreign Relations of the United States -1955-1957: Vol. V, The Summit Conference Vol. XVI: Suez Canal Crisis -1958-60: Vol. VIII: The Berlin Crisis -1961-63 Vol. IV: Vietnam Vol. VI: Kennedy-Khrushchev Exchanges Vol. XV: Berlin Crisis -1964-68 Vol. XIV: The Soviet Union U.S. Department of State, Documents on Germany, 1944-1985. Office of the Historian, Bureau of Public Affairs (Washington, D.C.: Government Printing Office, 1985 Lawrence Chang and Peter Kornbluh (eds.) The Cuban Missile Crisis, 1962: A National Security Archive Documents Reader, 1992. New York: New Press McAuliffe, Mary S. (ed.), CIA Documents on the Cuban Missile Crisis, 1992. Diane Pub. Co 228 Interviewees Pervez Musharraf, former President of Pakistan, Seattle, March 2010 Brajesh Mishra, former Chief National Security Adviser to the Prime Minister, Delhi, April 2010 Christine Fair, Assistant Professor, Georgetown University, Seattle, September 2011 Several other interviews in India (April 2010) requested and were granted anonymity
UBC Theses and Dissertations
Nuclear proliferation and the use of force : nuclear coercion and nuclear learning Cohen, Michael David 2012
Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
- 24-ubc_2012_spring_cohen_michael.pdf [ 1.18MB ]
- JSON: 24-1.0072740.json
- JSON-LD: 24-1.0072740-ld.json
- RDF/XML (Pretty): 24-1.0072740-rdf.xml
- RDF/JSON: 24-1.0072740-rdf.json
- Turtle: 24-1.0072740-turtle.txt
- N-Triples: 24-1.0072740-rdf-ntriples.txt
- Original Record: 24-1.0072740-source.json
- Full Text