This article presents and tests a theory of learning in international politics. Drawing primarily from social psychology and organization theory, the learning theory proposes that lessons tend to be drawn only from high-impact events in world politics, such as large wars and economic depressions. Lessons drawn tend to be simple and are oriented around the question of which policies are likely to be successful and which policies are likely to fail. This learning theory is tested on the alliance choices of small powers in the twentieth century. The predictions of two learning hypotheses are compared with those of a leading realist explanation of alliance choices, balance of threat theory. Quantitative analysis of small powers' alliance choices reveals that a small power's experience in the previous world war is a very powerful explanation of its peacetime alliance choices after that war, whereas the level of threat in the international environment has only marginal effects on the small power's alliance choices. Further, these threat effects may be in the opposite direction of that predicted by balance of threat theory.
1 For works drawing on ideas from social psychology, see Jervis, Robert, Perceptions and Misperceptions in International Politics (Princeton:Princeton University Press, 1976); Snyder, Jack, Ideology of the Offensive (Ithaca, N.Y.:Cornell University Press, 1984); Larson, Deborah Welch, Origins of Containment (Princeton:Princeton University Press, 1985); Vertzberger, Yaacov Y. I., “Foreign Policy Decisionmakers as Practical-Intuitive Historians: Applied History and Its Shortcomings,” International Studies Quarterly 30 (June 1986); Snyder, Jack, Myths of Empire: Domestic Politics and International Ambition (Ithaca, N.Y.:Cornell University Press, 1991); Foong Khong, Yuen, Analogies at War (Princeton:Princeton University Press, 1992). For works drawing on ideas from organization theory, see Steinbruner, John D., The Cybernetic Theory of Decision (Princeton:Princeton University Press, 1974); Lovell, John P., “Lessons of U.S. Military Involvement: Preliminary Conceptualization,” in Sylvan, Donald A. and Chan, Steve, eds., Foreign Policy Decision Making (New York:Praeger, 1984); Posen, Barry, Sources of Military Doctrine (Ithaca, N.Y.:Cornell University Press, 1984); Evangelista, Matthew, Innovation and the Arms Race (Ithaca, N.Y.:Cornell University Press, 1988); Sagan, Scott D., The Limits ofSafety: Organizations, Accidents, and Nuclear Weapons (Princeton:Princeton University Press, 1993). Ideas from information economics have been applied to this question, as well. See, for example, Powell, Robert, “Nuclear Brinksmanship with Two-Sided Incomplete Informaton,” American Political Science Review 82 (March 1988), and Wagner, R. Harrison, “Uncertainty, Rational Learning, and Bargaining in the Cuban Missile Crisis,” in Ordeshook, Peter C., ed., Models of Strategic Choice in Politics (Ann Arbor:University of Michigan Press, 1989). For a review of different ideas about learning in international relations scholarship, see Levy, Jack S., “Learning, and Foreign Policy: Sweeping a Conceptual Minefield,” International Organization 48 (Spring 1994); and Jarosz, William W. with Nye, Joseph S. Jr., “The Shadow of the Past: Learning from History in National Security Decision Making,” in Tetlock, Philip E. et al., eds., Behavior, Society, and International Conflict, vol. 3 (New York:Oxford University Press, 1993).
2 For this study, learning is broadly conceptualized as the application of information derived from past experiences to acquire understanding of a particular policy question. This is distinct from what has been called “corrective” learning, in which experience necessarily improves performance over time. For a discussion of different definitions of learning, see Philip E. Tetlock, “Learning in U.S. and Soviet Foreign Policy: In Search of an Elusive Concept,” in Breslauer, George W. and Tetlock, Philip E., eds., Learning in U.S. and Soviet Foreign Policy (Boulder, Colo.:Westview Press, 1991).
3 Nisbett, Richard and Ross, Lee, Human Inference: Strategies and Shortcomings of Social Judgment (Englewood Cliffs, N.J.:Prentice-Hall, 1980).
4 Taylor, Shelley E. and Crocker, Jennifer, “Schematic Bases of Social Information Processing,” in Higgins, E. Tory, Herman, C. Peter, and Zanna, Mark P., eds., Social Cognition: The Ontario Symposium, vol. 1 (Hillsdale, N.J.:Lawrence Erlbaum Associates, 1981); Schank, Roger and Abelson, Robert P., Scripts, Plans, Goals and Understanding (Hillsdale, N.J.:Lawrence Erlbaum Associates, 1977).
5 Abelson, Robert P., “Script Processing in Attitude Formation and Decision Making,” in Cognition and Social Behavior (Hillsdale, N.J.: Lawrence Erlbaum Associates, 1976); Abelson, and Lalljee, Mansur, “Knowledge Structures and Causal Explanation,” in Hilton, Denis J., ed., Contemporary Science and Natural Explanation (New York:New York University Press, 1988). For a discussion of schemata as applied to international relations, see Larson (fn. 1).
6 Kahneman, Daniel, Slovic, Paul, and Tversky, Amos, eds., Judgment under Uncertainty: Heuristics and Biases (Cambridge: Cambridge University Press, 1982); Nisbett and Ross (fn. 3).
7 Vosniadu, Stella and Ortony, Andrew, eds., Similarity and Analogical Reasoning (Cambridge: Cambridge University Press, 1989); Khong (fn. 1).
8 Read, Stephen J., “Once Is Enough: Causal Reasoning from a Single Instance,” Journal of Personality and Social'Psychology 45, no. 2 (1983).
9 Gilovich, Thomas, “Seeing the Past in the Present: The Effect of Associations to Familiar Events on Judgments and Decisions,” Journal of Personality and Social Psychology 40, no. 5 (1981).
10 Levitt, Barbara and March, James G., “Organizational Learning,” Annual Review of Sociology 14 (1988). See also Argyris, Chris, On Organizational Learning (Cambridge, Mass.:Blackwell, 1993).
11 Sitkin, Sim B., “Learning through Failure: The Strategy of Small Losses,” in Staw, Barry M. and Cummings, L. L., eds., Research in Organizational Behavior 14 (Greenwich, Conn.:JAI Press, 1992); Hedberg, Bo, “How Organizations Learn and Unlearn,” in Nystrom, Paul C. and Starbuck, William H., eds., Handbook ofOrganizational Design (London:Oxford University Press, 1981), 1:17-18; Mirvis, Philip and Berg, David, eds., Failures in Organization Development and Change (New York:Wiley, 1977).
12 Nisbett and Ross (fn. 3).
13 Nystrom, Paul C. and Starbuck, William H., “Managing Beliefs in Organizations,” Journal of Applied Behavioral Science 20, no. 3 (1984); Hedberg (fn. 11), 6; Levitt and March (fn. 10), 324; Jönsson, Sten A. and Lundin, Rolf A., “Myths and Wishful Thinking as Management Tools,” in Nystrom, Paul C. and Starbuck, William H., eds., Prescriptive Models of Organizations (Amsterdam:North-Holland, 1977).
14 Nisbett and Ross (fn. 3), 45. On schemata and belief changes, see Crocker, Jennifer, Fiske, Susan T., and Taylor, Shelley E., “Schematic Bases of Belief Change,” in Eiser, J. Richard, ed., Attitudinal Judgment (New York:Springer-Verlag, 1984).
15 A number of studies have failed to find that vivid events are more persuasive. See Taylor, Shelley E. and Thompson, Suzanne C., “Stalking the Elusive ‘Vividness’ Effect,” Psychological Review 89 (1982); Taylor, Shelley E. and Wood, Joanne V., “The Vividness Effect: Making a Mountain out of a Molehill?” in Bogozzi, Richard P. and Tybout, Alice M., eds., Advances in Consumer Research 10 (Ann Arbor, Mich.:Association of Consumer Research, 1983); Collins, Rebecca L. et al., “The Vividness Effect: Elusive or Illusory?” Journal of Experimental Social Psychology 24 (January 1988), Fiske, Susan T. and Taylor, Shelley, Social Cognition, 2d ed. (New York:McGraw-Hill, 1991). However, Collins et al. found that vivid events are believed to be more persuasive, raising the possibility that vivid events are more persuasive by self-fulfilling prophecy. Additionally, some studies have found that events experienced firsthand have a greater impact on behavior than those experienced vicariously. See Fiske and Taylor, 520-21, and Braver, Sanford L. and Rohrer, Van, “Superiority of Vicarious over Direct Experience in Interpersonal Conflict Resolution,” Journal ofConflict Resolution 22, no. 1 (1978). For examples of international relations research that found that events experienced firsthand were formative of beliefs, see Jervis (fn. 1) and Khong (fn. 1).
16 Sitkin (fn. 11), 238; Levitt and March (fn. 10), 329-31; Mahajan, Viday, Sharma, Subhash, and Bettis, Richard, “Adoption of the M-Form Organizational Structure: A Test of Imitation Hypothesis,” Management Science 34 (October 1988).
17 The formality of an alliance declaration is not a diplomatic triviality. Morgenthau has remarked, “When the common interests are inchoate in terms of policy and action, a treaty of alliance is required to make them explicit and operative.” Morgenthau, Hans, “Alliances in Theory and Practice,” in Wolfers, Arnold, ed., Alliance Policy in the Cold War (Baltimore:Johns Hopkins University Press, 1959), 188.
18 Alliances with great powers are distinguished here from one-way commitments by a great power to defend the minor power, because mutual alliances increase the minor power's risk of entanglement far more than one-way commitments.
19 Singer, J. David and Small, Melvin, “Formal Alliances, 1815-1939,” Journal of Peace Research no. 1 (1966); and Huth, Paul and Russett, Bruce, “Deterrence Failure and Crisis Escalation,” International Studies Quarterly 32 (March 1988).
20 May, Ernest R., “Lessons” of the Past (New York:Oxford University Press, 1973); Neustadt, Richard E. and May, Ernest R., Thinking in Time: The Uses of History for Decision-Makers (New York:Free Press, 1986); Khong (fn. 1).
21 Liska, George, International Equilibrium (Cambridge:Harvard University Press, 1957), 25; Aron, Raymond, Peace and War: A Theory of International Relations, trans. Howard, Richard and Baker Fox, Annette (Garden City, N.Y.:Doubleday, 1966), 58; Waltz, Kenneth, Theory of International Politics (New York:Random House, 1979), 72.
22 Marks, Sally, The Illusion of Peace: International Relations in Europe, 1918-1933 (New York:St. Martin's Press, 1976), 1. See also Gilpin, Robert, War and Change in World Politics (Cambridge: Cambridge University Press, 1981); Levy, Jack S., War in the Modern Great Power System, 1495-1975 (Lexington: University Press of Kentucky, 1983); Thompson, William R., On Global War: Historical-Structural Approaches to World Politics (Columbia:University of South Carolina Press, 1988); and Modelski, George, Long Cycles in World Politics (Seattle:University of Washington Press, 1987). A leading proponent of the argument that systemic wars do not have unique causes agrees that systemic wars do have unique effects. de Mesquita, Bruce Bueno, “Big Wars, Little Wars: Avoiding Selection Bias,” International Interactions 16, no. 3 (1990), 159–60.
23 Neutrality and alliance were not important policy categories for most minor powers after either the Wars of Louis XIV, when the concepts of sovereignty and independence were not as widely applicable as in the twentieth century, or the Napoleonic Wars, when the great powers imposed order via the Concert of Europe.
24 Invasion constitutes the failure of neutrality since the primary reason neutrality is chosen is to avoid participation in war.
25 This is admittedly a crude measure for alliance failure, but population is a good index with which to compare wartime losses with gains made in postwar settlements.
26 It is also assumed that a formative event (systemic war, in this application) has a watershed effect, such that previous formative events are dominated by the most recent formative event in terms of belief formation. This watershed assumption is necessary to avoid infinite regress in determining which event is formative. For 1994 France, for example, there are a number of events that one could deem to be the “primary” event in French learning about international affairs, such as World Wa r II, World Wa r I, the Napoleonic Wars, and so forth. Th e organization theory proposition that crises create new myths that replace old ones supports the watershed assumption.
27 A similar argument was made by Nuechterlein, Donald E., “Small States in Alliances: Iceland, Thailand, Australia,” Orbis 13 (Summer 1969). This paper builds on this past work by constructing a theory of learning, testing its predictions on a broader set of data, and comparing these predictions to those of realism.
28 If a state thinks that its own experience is not idiosyncratic and that the experiences of other nations are relevant, then learning from the experience of all minor powers in a formative event would be more unbiased than learning just from one's own experience.
29 Waltz (fh. 21), is quite blunt on this point (pp. 72—73). There is, however, a considerable literature on the foreign policy problems of non-great powers—termed variously as middle powers, small powers, weak states, and the like—though it is often descriptive rather than predictive. See Fox, Annette Baker, The Power of Small States: Diplomacy in World War II (Chicago:University of Chicago Press, 1959); Rothstein, Robert L., Alliances and Small Powers (New York:Columbia University Press, 1968); Ralston, Jerry Wilson, The Defense ofSmall States in the NuclearAge: The Case of Sweden and Switzerland (Neuchatel:La Baconnière, 1969); Keohane, Robert O., “Lilliputians' Dilemmas: Smal l States in International Politics,” International Organization 23 (Spring 1969); Katzenstein, Peter J., Small States in World Markets (Ithaca, N.Y.: Cornell University Press, 1985); Karsh, Efraim, Neutrality and Small States (London: Routledge, 1988); and Mueller, Karl, “Strategy, Asymmetric Deterrence , and Accommodation: Middle Powers and Security in Modern Europe” (Ph.D. diss., Princeton University, 1991).
30 Christensen, Thomas J. and Snyder, Jack L., “Chain Gangs and Passed Bucks: Predicting Alliance Patterns in Multipolarity,” International Organization 44 (Spring 1990). Interestingly, they note that experiences often drive beliefs abou t th e offense-defense balance. For a diverse list of propositions concerning alliances, see Holsti, Ole, Hopmann, P. Terrence, and Sullivan, John D., Unity and Disintegration in International Alliances: Comparative Studies (New York:John Wiley, 1973).
31 Walt, Stephen M., The Origins of Alliances (Ithaca, N.Y.:Cornell University Press, 1987).
32 This last point, that alliances are likely to break up as the threat recedes or disappears (such as after an aggressor is defeated in war), was made by Walt (fn. 31), 29–30; and Morgenthau (fn. 17), 191–92.
33 Walt (fn. 31); and idem, Testing Theories of Alliance Formation: The Case of Southwest Asia,” International Organization 42 (Spring 1988). A number of quantitative studies using an expected utility approach have found varying empirical support for the general proposition that threats play a part in driving alliance choices, though most of these works have tested models of decision that are more sophisticated than Walt's. See Berkowitz, Bruce D., “Realignment in International Treaty Organizations,” International Studies Quarterly 27 (March 1983); Altfeld, Michael T., “The Decision to Ally,” Western Political Quarterly 37 (December 1984); Lalman, David and Newman, David, “Alliance Formation and National Security,” International Interactions 16, no. 4 (1991); Cusack, Thomas R. and Stoll, Richard J., “Balancing Behavior in the Interstate System, 1816—1976,” International Interactions 16, no. 4 (1991). For an expected utility model that emphasizes the importance of domestic as well as international factors in shaping alliance behavior, see Morrow, James D., “Alliances and Asymmetry: An Alternative to the Capability Model of Alliances,” American Journal of Political Science 47 (November 1991); and idem, “Arms versus Allies: Trade-Offs in the Search for Security,” International Organization 47 (Spring 1993).
34 Walt (fn. 31), 31n.
35 See the works cited in fn. 19.
36 Osgood, Robert E., Alliances and American Foreign Policy (Baltimore:Johns Hopkins University Press, 1968), 19–20.
37 Mueller (fn. 29).
38 Walt (fn. 31), 23–24.
39 Most empirical research in learning focuses on a small group of cases, including May (fn. 20); Snyder (fn. 1,1984 and 1991); Posen (fn. 1); Evangelista (fn. 1); and Khong (fn. 1). In particular, the cases of American and Soviet learning during the cold war are receiving increasing attention. See Breslauer and Tetlock (fn. 2); Fry, Michael G., ed., History, the White House and the Kremlin (London:Pinter Publishers, 1991); Midlarsky, Manus I., Vasquez, John A., and Gladkov, Peter V., eds., From Rivalry to Cooperation: Russian and American Perspectives on the Post Cold War Era (New York:HarperCollins, 1994). Two studies of learning in world politics using larger samples are Leng, Russell, “When Will They Ever Learn?” Journal of Conflict Resolution 27 (September 1983); and Huth, Paul K., Extended Deterrence and the Prevention of War (New Haven:Yale University Press, 1988).
40 Autocorrelation means that there is a systematic relationship between the dependent variable at time t and the dependent variable at time t-1.
41 This technique is known as systematic sampling. Freeman, John R., “Systematic Sampling, Temporal Aggregation, and the Study of Political Relationships,” in Stimson, J. A., ed., Political Analysis (Ann Arbor:University of Michigan Press, 1990).
42 Coding for the dependent variable was made on the basis of primary and secondary historical sources. The absence of alliance between a minor power and any great power does not necessarily indicate the preference on the part of the minor power for neutrality, the minor power may prefer great power alliance, but there is no willing great power. In such a case, the dependent variable was coded asl.
43 Walt (fn. 31), 21–26.
44 Correlates of War data on military spending were used. Using troop levels is problematic for small powers, as some countries rely on reserve or militia forces for defense. It is difficult to assess the contribution of these forces except on a case by case basis, because the quality of these forces and the speed with which they could be committed to the national defense vary widely.
45 The military forces from other countries commited to the security of the defender need to be considered separately from the defender's own military forces, because the ally cannot commit all of its forces to help the defender since it has to defend its own territory as well as that of the defender.
46 On these measures, see Singer, J. David, Bremer, S., and Stuckey, J. in Russett, Bruce, ed., Peace, War, and Numbers (Beverly Hills, Calif.:Sage, 1972).
47 The likelihood ratio test statistic (LRTS) is twice the difference of the log likelihood values for the restricted and unrestricted model specifications. In this case, that value is 5.39, which is greater than the critical value for .20 in a chi-squared distribution, meaning that we can conclude with confidence that there is little systematic relationship between these four variables (STMB, GEOG*STMB, DTMB, and DCOM) and the dependent variable. On this test, see King, Gary, Unifying Political Methodology (Cambridge: Cambridge University Press, 1989).
48 The LRTS for comparing the model in equation (1) with a model including just the two learning variables in 14.8, which is significant at the .05 level with 7 degrees of freedom.
49 See Altfeld (fn. 33); Lalman and Newman (fn. 33); Berkowitz (fn. 33). Lalman and Newman propose that a different global distribution of power encouraged a greater propensity to ally after World War II than after World War I. The learning hypothesis argues that this greater tendency was due to the individual experiences of states in wartime and is preferable to the distribution of global power explanation because it accounts for decisions of individual states for alliance or neutrality as well as for systemwide patterns.
50 de Mesquita, Bruce Bueno, The War Trap (New Haven:Yale University Press, 1981). This paper used an alternative method of directly measuring external threats for two reasons. First, utility for war is an imprecise conceptual proxy for security, partly because it implies that a state is more secure the more it stands to gain from the prosecution ofwar. Second, Bueno de Mesquita's method uses alliances to predict foreign policy preferences, raising falsifiability concerns for an application here, given that alliances would be used as both dependent and independent variables.
51 Modelski, George, “The Asian States' Participation in SEATO,” in Modelski, George, ed., SEATO: Six Studies (Melbourne:F. W. Cheshire, 1962), 90–91.
52 On Belgium, see Helmreich, Jonathan, “The Negotiation of the Franco-Belgian Military Accord of 1920,” French Historical Studies 3 (Spring 1964); and Smets, Paul-F., ed., La pensee europeenne atlantiques de Paul-Henri Spaak (1942–1972), 2 vols. (Brussels:J. Goemere, 1980). Belgium broke off its alliance with France in 1936 due primarily to an array of domestic political factors. Of course, this case constitutes a failure for both learning hypotheses. On the 1936 break, see Kieft, David O., Belgium's Return to Neutrality (Oxford:Clarendon Press, 1972). On Switzerland, see Freymond, Jacques, “The Foreign Policy of Switzerland,” in Foreign Policy in a World of Change (New York:Harper and Row, 1963), 151–52. On the Netherlands, see Vandenbosch, Array, Dutch Foreign Policy since 1815 (The Hague:Martinus Nijhoff, 1959), 289. A more extensive discussion of these and other cases is presented in Dan Reiter, “Learning, Realism, and Alliances: An Empirical Examination of the Causes of Alliances” (Ph.D. diss. University of Michigan, 1994).
53 Abadie-Maumert, F. A., “Le pacifisme norvegian entre 1919 et 1940 et ses conséquences,” Guerres mondialeset conflits contemporains 40 (October 1990); Fox (fn. 29), 81,111–12; Derry, T. K., A History of Modern Norway, 1814–1972 (Oxford:Clarendon Press, 1973), 415; Örvik, Nils, Trends in Norwegian Foreign Policy (Oslo:Norwegian Institute of International Affairs, 1962), 6, 12; Abrahamsen, Samuel, Sweden's Foreign Policy (Washington, D. C:Public Affairs Press, 1957), 14.
54 Aalders, Gerald, “The Failure of the Scandinavian Defence Union, 1948–1949,” Scandinavian Journal of History 15 (1990), 133–34. Nazi Germany invaded Norway and Denmark on April 9,1940.
55 Ralston (fn. 29), 209.
56 Morrow (fn. 33, 1991, 1993).
57 David, Steven R., “Explaining Third World Alignment,” World Politics 43 January 1991); Barnett, Michael N. and Levy, Jack S., “Domestic Sources of Alliances and Alignments: The Case of Egypt, 1962—73,” International Organization 45 (Summer 1991); and idem, “Alliance Formation, Domestic Political Economy, and Third World Security,” Jerusalem Journal of International Relations 14 (December, 1992).
58 Curiously, being a nondemocracy seems to be nearly a necessary condition for not behaving as the individual learning hypothesis predicts. For more on this pattern, see Reiter (fn. 52).
59 Rothstein (fn. 29).
60 Örvik, Nils, The Decline of Neutrality, 1914–11 (Oslo: Tanum, 1953), 177–90.
61 Kieft (fn. 52).
* A version of this paper was presented at the October 1993 Midwest meeting of the International Studies Association in Chicago. This article is based upon work supported under a National Science Foundation Graduate Fellowship. I am also grateful for support from the MacArthur Foundation. I would like to thank Robert Axekod, Nancy Burns, Paul Diehl, Paul Huth, Ted Hopf, Scott Gartner, Karl Mueller, and Jennifer Shulman for their comments on various drafts, though I alone am responsible for any errors.
Email your librarian or administrator to recommend adding this journal to your organisation's collection.
* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.
Usage data cannot currently be displayed