Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
NATIONAL LEADERS IN INTERDEPENDENT Sll~lJATIONS Interdependence in a Crisis Situation: Simulating the Caribbean Crisis . . . VIKTOR M. SERGEEV, V.P. AKIMOV, VADIM B. LUKOV, AND PAVEL B. PARSHIN THEORETICAL BACKGROUND The paper summarized below follows what the authors would describe as an "empirical cognitive modeling" approach. That is, in flying to elu- cidate the problem of interdependence in crisis situations, we first turn to concrete historical analysis, specifically, the paradigmatic, well-documented case of the Caribbean crisis of 1962. Second, we join a tradition of coun- terfactual modeling which allows us to remove an event from its historical context and to ascribe to it certain "coordinates." By changing the coor- dinates, we may conduct quasi-experimental tests of hypotheses about the event, thereby identifying typical features of crisis situations and explaining them. Third, we assume that anyone who is going to design a model of crisis interdependence should be allowed to refer to such entities as world models, goals, plans, values, intentions, desires, and so on, as well as rela- tions, both static and dynamic, among them. In particular, we claim that the principal feature of the political crisis is a rapid mental restructuring of participants' beliefs. 1b handle such entities, we turn to '~weak" cognitive models of the type elaborated in the fields of artificial intelligence and cognitive science (see Sergeev and Parshin, 1987~. What such models do is (1) transform large masses of data about the modeled system's behavior into a more convenient and observable form (the reconstructive stage, or modeling proper), and (2) create and use research tools to simulate the modeled system's behavior (simulation stage). The study is based on two different bodies of empirical data. One consists of texts per se, that is, verbal records; the other is the set of events A more detailed version has been recently published (Sergeev et al., 1989~. 47
48 SOYIET-AMERICAN DIALOGUE IN THE SOCIAL SCIENCES during the crisis, which can also be considered as a text sui genesis an action text. Making use of a computer algorithm, we constructed game- theoretic matrices from the set of events in order to describe the structure of the interaction. The strategic matrix approach, however, does not allow us to make situational representations that are detailed enough to study how actors achieve their concrete goals; neither does it allow us to evaluate the intensity of the crisis. Therefore, the strategic matrix model of the crisis should be supplemented by modelers) of the actors' perceptions of the situation. The knowledge representation technique used in our perception model goes back to the ideas of cognitive mapping. Once a model of perception is constructed, it is possible to simulate the process of decision making. The most natural way to do so is to transform the perception structure into a dynamic system that is able to react to changes in the situation. AN ACTION PERSPECTIVE: EVOLUTION OF THE CRISIS 1b construct a strategic 2x2 matrix (each side has two strategic choiceseither conflictual or cooperative behavior), one needs to Mow the preference structure that links the conflictual or cooperative strategies of the actors to payoffs. In essence, the construction of a strategic matrix is the construction of an interdependence structure, not on the grounds of normative beliefs about actors' priorities and their modes of thinking, but on the grounds of actual actions. A set of conflict or crisis events may be thought of as a chronologically ordered description of participants' actions in a process of conflict or crisis development. We considered a set of 34 actions of the two main participants in the crisis (the USSR and the United States). The time covered is July 1962 to January 1963. Since every action of the participants could be regarded as the outcome of a choice between cooperative and conflictual strategies based on their evaluation of the alternatives, inferring these evaluations from their actions defines a special class of mathematical tasks the inverse problem of conventional game theory, in which the values of the alternatives are given. The flow of events may be represented as multiple plays of a 2x 2 matrix game. Supposing that both players are rational (that is, that they are guided by strict systems of preferences that are defined on matrix elements), we can define, with the help of logical procedures and a Bayesian approach, the likelihood that the given sequence of moves is produced by one or another game-theoretic matrix. By analyzing event-row segments of varying length, we can identify variations of the situation structure, represented as a game- theoretic matrix, over time. The distributions of likelihoods defined on the
NATIONAL LEADERS IN INTERDEPENDENT SITUATIONS 49 set of 576 strategic matrices were calculated) for the segments of the event sequences, at cut points in time corresponding to the basic events. As the participants advance to crisis regulation, we see decreasing uncertainty in their interpretations of the situation. Up to October 22, possible developments of the situation were considered by the participants to be rather uncertain (more than 160 strategical matrices were equally possible). At the moment of final crisis resolution (January 7, 1963), analysis shows the presence of only one substantially possible variant of the participants' perception of the situation, represented in the matrix below: USSR Cooperation Conflict United States: Cooperation 3,3 2,4 Conflict 4,2 1,1 At the same time, well-known facts and eye-witnesses' accounts indicate that there was one more "peak of certainty" in the course of the crisis just at the moment of maximal aggravation (October 26-27~. Our analysis reveals this peculiarity also: for the segment marked by the October 26 cutting point the program gives two not-very-different matrices, each having a rather substantial likelihood: USSR Cooperation Conflict United States: Cooperation 1,4 3,3 Conflict 4,2 - 2,1 USSR Cooperation Conflict United States: Cooperation 1,3 3,4 Conflict 4,2 2,1 The peaks of certainty are of different natures: the matrices for October 26-27 presuppose conflict escalation on the part of the United States, while for January 7 we see the well-lmown Chicken matrix. The analysis of the Caribbean crisis shows that by no means all con- flicts associated with global strategical problems should be described by matrices of the Prisoner's Dilemma type. In fact, a very different type of interdependence occurred in the Caribbean crisis. 1An algorithm has been described by Akimov and Sergeev (1987~. An English translation is available from Scripta Technica, Inc., New York, NY.
so SOKIET-AMERICAN DIALOGUE IN THE SOCIAL SCIENCES - Let us return~to the Chicken matrix. If we restrict ourselves to the traditional criteria clef the game theory approach, consideration of the estimated matrices' dynamics can lead to rather pessimistic conclusions. The last matrix is characterized by a sharply conflictual correlation of priorities; that is, the participants, at least outwardly, "didn't learn any lessons." Is that true? To answer this question, we turn to the model of the participants' thinking, confining ourselves for the time being to an analysis of the U.S. leadership.2 AN INTENTIONAL PERSPECTIVE: COGNITIVE MAPPING ANALYSIS Cognitive maps were made from the following documents by John F. Kennedy: a speech in the U.S. Senate on June 14, 1960; statements on the situation in the Caribbean area at news conferences of September 5 and 13, 1962; a television address to the people on October 22, 1962, and a statement at the news conference of November 22, 1962. Our main research goals were to display Kennedy's perception of the strategic situation before the crisis; to reproduce his perception of the sequence of tactical situations of September-November 1962, in order to shed light on the restructuring of his thinking and to identify trends in its evolution; and to implement (using computer devices at our disposal) the model of Kennedv's nercention of one tactical situation, namely that of October 22, 1962. ~ r ---~- We found in Kennedy's cognitive maps the following evidences of entering the crisis (and hence, of an evolving crisis pattern of thinking) (see Lukov and Sergeev, 1982~: 1. An apparent move from a favorable (due to prepared "victorious" strategies) to a rather unfavorable (1:2.5) ratio of controlled to uncontrolled factors perceived in the situation: an environment acquires a developmental logic of its own. 2. An emergence of mature crisis tangles that is, structures formed by multiple negative effects of the environment on U.S. goals. Viewed through the lens of U.S. interests, the main crisis tangles are represented by the substructures centering around the "urgent transformation of Cuba into an important strategic base" and the "insincerity of the Soviet leader- ship." Both are out of U.S. control; they can be resisted, but not unilaterally manipulated, canceled, or ignored. The former factor, with its multiple neg- ative effects, reflects an overlapping of two principal dangers to the security 21be fist version of the analysis of the Soviet position was presented at the 2nd NASUSSR Workshop by Sergeev and Parshin [1989~.
NATIONAL LEADERS IN INTERDEPENDENT SllUATIONS 51 of the United States the "domestic instability in the underdeveloped coun- try" and the "Soviet nuclear threat." Those dangers were anticipated by Kennedy; however, earlier he had seen them separately, and their melding contributed much to his perception of the situation surrounding Cuba as a . . crisis. 3. A distinctly more careful (although rather biased regarding the motives of Soviet actions) analysis of the situation, including the goals of the USSR, its capabilities, the situation in Cuba, and consequences of various U.S. options for action. Interdependence makes a powerful breakthrough into Kennedy's belief system. AN INTENTIONAL PERSPECTIVE: COMPUTER SIMULATION There are many models of decision making. In this study we simulate only one of the subprocesses postulated by the easiest rational decision model, namely, that of a comparative assessment of options available to a decision maker in a concrete situation. We treat a cognitive map as a set of interconnected states of the world, which can be activated to a certain degree. Quasi-causal relation is interpreted as a relation of activation, coded as positive, negative, or zero. By discretely changing (to 1 or 0) the activation of input nodes (the switching-on of factors), one can produce different "waves of activation" and analyze how this or that combination of input possibilities influences the system of interest of the text's author. Our simulation made it possible to infer that the crisis dynamic was perceived by the U.S. leadership in the form of two scenarios: (1) when "sincerity" and "restraint" are absent from the pattern of Soviet actions and (2) when the Soviets display sincerity and restraint. In the second scenario, the effectiveness of the control measures instituted by Kennedy increases significantly: the measures secure a 98 percent attainment of U.S. goals. Besides, the heretofore "secondary" measures of the first scenario acquire greater significance (their sum total increases from 0 to 38 percent). Under the second scenario it is no longer necessary to escalate the crisis beyond measures declared already by the White House. We conclude from our study that interdependence transforms sincerity into an important resource in promoting one's goals. In addition, in a conflict situation, political actions that are considered quite natural and not very significant by one side can be extremely dangerous if they change entirely the other side's perception of the situation. In other words, a way of thinking that recognizes interdependence presupposes not only sincerity, but also a great deal of reflection.
52 SOVIET-AMERICAN DIALOGUE IN THE SOCIAL SCIENCES REFERENCES Akimov, V.P., and V.M. Sergeev 1987 Determination of preferences on the basis of post-game analysis of events (in Russian). TeAhnicheskay Kibemetika 2:188-193. Lukov, V.B, and V.M. Sergeev 1982 Patterns of crisis thinking. An analysis of the governing circles in Germany, 1866-1914. In Managing; Intemational Cnses, D. Frei, ed.,pp. 47~0. Beverly Hills, CA: Sage. Sergeev, V.M., V.P. Akimov, V.B. Lukov, and P.B. Parshin 1989 Modelling the Caribbean crisis- An essay. USA: Econonucs, Politics, Ideology 5:36-49. Sergeev, V.M., and P.B. Parshin 1987 Ideas and methods of artificial intelligence in the study of political thinking (in Russian). Yearbook of the Soviet Association of the Political Sciences. Political Sciences and the Scientist and Technological Revolution, pp. 206-221.
