This article reviews major decision-making models with an emphasis on basic theoretical perspectives as well as on how these models explain foreign policy decision making and national and international security decisions. Furthermore, we examine how these models have been utilized in explanations of various international crises. Specifically, for each model, we present examples drawn from the literature on applications of the respective model to foreign policy and national security decisions. The theories we have reviewed are: rational choice, cybernetic model, prospect theory, poliheuristic theory, organizational and bureaucratic politics, groupthink and polythink, analogical reasoning, Applied Decision Analysis (ADA) and biases in decision making.
How do national security level decision makers actually make foreign policy decisions? How do they process information? Which decision rules do they utilize in making decisions? How do the rules and strategies adopted by decision makers affect the policies they enact? More importantly, how do scholars of foreign policy and national security decision making utilize decision models in explaining real world decisions? In this article, we attempt to answer these questions by reviewing various decision making models and what they tell us about how national security level policy makers process information en route to choice and how this information processing affects the policies they choose. We begin with rational choice theory and then proceed to review the contributions of cybernetic theory, prospect theory, poliheuristic theory, organizational process and bureaucratic politics models, groupthink, polythink, analogical reasoning, Applied Decision Analysis (ADA) and biases in decision making. We chose these theories and models of decision making because they are generally regarded as the most prominent in the literature on foreign policy decision making and have been examined and applied in numerous cases as demonstrated below. We first briefly summarize each theory and then review how each has been applied in the scholarly literature to actual foreign policy and national security cases.
The term "rational actor," broadly defined, refers to an individual "who makes choices by taking the following steps: (1) searching for relevant information regarding the conditions of choice; (2) integrating that information so as to discover existing alternatives for action; (3) drawing upon empirical generalizations to deduce the likely results each alternative will yield; (4) judging which will best satisfy his or her wants; (5) choosing a course of action accordingly." Rosenberg (1995, 111).
The rational model is parsimonious and elegant. Few straightforward assumptions explain a wide range of policy decisions. Perhaps the simplest conceptualization of decision making according to rational choice theory is that put forth by Stein and Welch (1997, 52) wherein they state that,
Rational choice theory proceeds from a simple and intuitive idea: If we know what people want, we can explain and predict what they do. It assumes merely that people are cognitively competent to match means to ends and to rank options accordingly. In standard formulations, a rational analysis simply needs information about what people want, what alternatives are open to them, and what they know (or can reasonably be expected to figure out) about the likely costs and benefits of alternatives.
The key assumption of the rational choice school in international relations is that "nations are led by rational, forward-looking, expected-utility-maximizing leaders" (ibid., 751). Scholars distinguish between thin and thick rationality (Mintz and DeRouen 2010, 59). Thin rationality refers to the strategic pursuit of stable and ordered preferences (ibid). Thick rationality assumes, in contrast, that actors have specific preferences. Thus, in politics, for most politicians, the preference is typically perpetuation in office (ibid).
Rational choice approaches in foreign policy analysis and international relations have centered around several important contributions made by Bueno de Mesquita (1981, 1983, 1984, 1989), Bueno de Mesquita and Lalman (1990, 1992), and others (see e.g., Morgan and Bickers 1992; Morrow 1985, 1994; Wittman 1979). A review of this literature shows that the authors use the rational choice approach to explain and predict outcomes in foreign policy and international conflict. In general, the analytic, rational model, should lead to better decisions, although not always to better outcomes (Mintz and DeRouen 2010, 58).
Rational choice theory has been used to analyze many different historical foreign policy and national security choices. Bueno de Mesquita and Lalman (1992) found that rational choice theory was more persuasive in explaining domestic leaders' decisions pertinent to the Sino-Indian War and the Seven Weeks' War, just to name a few, than power preponderance theory or balance-of-power theory. Szalai (2008) used rational choice theory to examine the creation and evolution of U.S. nuclear defense policy in the 1940s. Pollack (2007) applies rational choice to a study of European Union politics including decision making with respect to integration; legislative, executive, and judicial politics within the context of the EU; and public opinion and Europeanization.
Kim and Bueno de Mesquita (1995) used a game theoretic approach to show that decision makers take situational perceptions into account in war and peace decisions. More specifically, they were able to specify the conditions under which differences in perception were more or less likely to lead to war. Fearon (1995, 400) explained war as a function of rationalist explanations arguing that "the combination of private information about relative power or will and the strategic incentive to misrepresent these afford a tenable rationalist explanation for war." Fearon used this focus on private information and deliberate misrepresentation to explain decisions in July 1914 in World War I on the part of the principal actors, as well as on the part of Russian and Japanese leaders in the Russo-Japanese War of 1904.
Mintz and DeRouen (2010) examined New Zealand's decision to defy the ANZUS Treaty and showed that New Zealand's decision, while appearing to be perhaps irrational in the sense that it made decisions that caused it to lose important allies in the United States and Great Britain during the Cold War, was actually quite rational. As the authors state (60), "by establishing itself as a promoter of international peace, New Zealand had conceivably increased its security and saved money that might otherwise have gone to expensive defense systems." Understanding New Zealand's decision as rational, however, can only be properly understood by taking domestic politics into account (Bueno de Mesquita and Lalman 1992), particularly the role of public opinion that focused on antinuclear policies (Mintz and DeRouen 2010).
Perhaps the chief critique of rational choice theory was originally put forth by Herbert Simon (1957) in his notion of bounded or procedural rationality, which states that decision makers are still rational, only boundedly so. Specifically, he posited that rational choice models were insensitive to the cognitive limitations of individuals and organizations (Simon 1992). By claiming that individuals were boundedly rational, Simon recognized the limitations of the information-processing capacities of both individuals and organizations in decision making (Simon 1957, 1990).
The basic argument is that individuals face constraints that limit decision makers' computational capabilities, their memory and recall abilities, etc. Because of these constraints, individuals develop decision procedures that enable them to deal more effectively and decisively with both their own cognitive limitations as well as with the demands imposed by the decision environment (March 1986; Simon 1955, 1957). Simon (1957) used the term 'satisficing' to denote these decision procedures. Satisficing implies that decision makers stop searching for information once they have found a satisfactory alternative; moreover, this alternative need not be an optimal one, merely one that satisfies some a priori minimum threshold (Monroe 1991; Zey 1992).
Steinbruner (1974, 12-13) in his influential book, The Cybernetic Theory of Decision states that, "Critics have long noted that rational theory assumes such sophisticated processing of information that it strains credulity to impute such procedures to real decision makers. The mind of man, for all its marvels, is a limited instrument." Steinbruner attempted to explain decision making as it occurs in reality, i.e., under conditions of complexity and uncertainty. Decision makers operate under conditions of "structural uncertainty" wherein an individual is not able to ascertain the state of the environment, locate available alternatives, or even assess the consequences of a chosen alternative (ibid., 18). Basically speaking, he argued that the cybernetic processes of "incrementalism" and "satisficing" used by individuals explained simple and "routine" decisions. Steinbruner (ibid., 86) summarizes the cybernetic paradigm by stating that "[its] major theme is that the decision process is organized around the problem of controlling inherent uncertainty by means of highly focused attention and highly programmed response. The decision maker in this view does not engage in alternative outcome calculations or in updated probability assessments."
Ostrom and Job (1986; see also James and Oneal 1991), using several key variables to represent the effect of domestic politics, the economy and international tension, constructed a cybernetic model of national security decision making for the political use of military force. Ostrom and Job (ibid., 543) refer to a decision maker's tendency to "[formulate] simple and manageable decision algorithms." Specifically, they argue that presidents monitor a limited number of critical factors and then consider a restricted set of decision options (ibid.).
Ostrom and Job (ibid.) constructed their cybernetic model to account for three factors originally suggested by Simon (1959): (1) the cognitive structure of the president-as-decision-maker, (2) the content and formulation of his decision premises, and (3) the logic of his inference process (i.e., the decision rule) used to reach a decision. Specifically, with respect to cognitive structure, the authors (Ostrom and Job 1986, 544) posit that decision makers focus on a small and relatively fixed number of environmental factors but that they perceive these various inputs in "terms of quite gross distinctions." In other words, a decision maker will not exhaustively catalog the actual state of the environment as well as the available options. In addressing decision premises, Ostrom and Job (ibid.) quote Simon (1959, 274) wherein he states that each premise specifies the "computational procedures for assessing the state of the environment and its implications for action" (italics in original). Ostrom and Job (ibid.) state that the president forms his decision premises on the basis of his three major functional responsibilities: commander-in-chief (international dimension), chief executive (domestic dimension), and political leader (political dimension).
Ostrom and Job (ibid.) note that the president must decide how to choose among the available alternatives in a given decision task, i.e., he must implement a decision rule. Again, the authors refer to Simon (1959) and his notion of satisficing wherein a decision maker attempts to find solutions that are "good enough" rather than those that maximize expected utility.
Prospect theory was first introduced by Kahneman and Tversky (1979), and has since become one of the leading alternatives to rational choice (expected utility) as a theory of decision under conditions of risk (see also Levy 1992a, 1992b, 1997a, 1997b; McDermott 1992,2004; Mercer 2005). The work by Kahneman and Tversky and their colleagues forms the foundation upon which other scholars have built as they have applied prospect theory's basic decision-making concepts to other subfields such as political science.1
Prospect theory posits that individuals evaluate outcomes, not from net asset levels, but instead as a function of deviations from a reference point. They also overweight losses relative to comparable gains, and are risk-acceptant in the domain of loss but risk-averse in the domain of gain. Their identification of this reference point is a critical variable, and they react to probabilities in a nonlinear fashion (Kahneman and Tversky 1979; Levy 1992a, 1992b, 1997a, 1997b).
Instead of evaluating net asset levels, individuals tend to think in terms of gains and losses, specifically choosing among options in terms of deviations from a reference point (ibid.). Reference dependence is the central analytic assumption of prospect theory (Levy 1997a; Tversky and Kahneman 1991). This reference dependence is in violation of the expected utility assumption of an individual utility function that is defined in terms of net asset levels. Reference dependence manifests itself when we see that an individual "may prefer x to y when x is currently a part of her endowment but prefer y to x when y is part of her endowment" (Levy 1997a, 35). The reference point is usually the status quo, but is not necessarily so. Levy (1992a) speaks of deviations from an aspiration level or even some other reference point that is not synonymous with the status quo.
Because of the differential treatment of gains and losses and the importance of the reference point in determining outcomes associated with these domains, identifying the reference point, i.e., framing the choice problem, can have a critical effect on choice (Kahneman and Tversky 1979; Levy 1992a, 1997a; Tversky and Kahneman 1981, 1986). Framing the choice problem in terms of gains vs. losses has a significant impact on preferences regardless of whether or not the two representations are mathematically equivalent. Framing of the reference point is often predetermined by the situation, i.e., in static environments the status quo equals the reference point. However, under dynamic conditions the reference point is not so well defined and differences in the way individuals accommodate to gains and losses may affect the framing of the reference point (Levy 1992a, 1997a).
Levy (1992a, 1997a) notes that prospect theory divides choice processes into two phases: (1) an editing phase which includes a preliminary analysis of the choice problem wherein actors identify options, the possible consequences or outcomes of each, and the values and probabilities associated with each of these outcomes, and (2) an evaluation phase in which the edited prospects are evaluated and then the preferred prospect is chosen. However, Levy (1997a, 42) states that "these parameters [identifying options, possible outcomes, and values and probabilities associated with each] are taken as given and treated exogenously. In its current form, therefore, prospect theory is a theory of the evaluation of prospects but not a theory of the editing of choices" (see also Levy 1992a). Levy (1997a, 1997b, also see McDermott 1992) notes that this is a limitation of prospect theory especially if one considers the plausible argument that much of the explanatory power in international relations exists in the specification of the problem, the available options, the values and probabilities associated with possible outcomes, and reference point framing.
However, this is a problem for expected utility and many other decision theories as well (ibid.). Both expected utility and prospect theory are structural theories rather than process theories. "Given certain parameters of the choice problem, they attempt to explain choices or outcomes, not the processes through which those choices come about" (Levy 1997, 42; see also Abelson and Levi 1985).
Prospect theory has been applied in various studies of national and international security and foreign policy behavior (see e.g., Boettcher 2004; Farnham 1992; Jervis 1988, 1989, 1991, 1992; McDermott 1992).
Farnham (1992) examined FDR's decision-making behavior with respect to the Munich crisis and found that while FDR was initially hesitant to intervene in the crisis, he changed his mind once the crisis was framed as a loss. Specifically, Farnham (1992, 233) states,
"Roosevelt's decision-making behavior shows that his reversal of preferences about the desirability of American intervention in the crisis was not the result of a reassessment of the expected utility of intervening on the basis of new information. Rather, in the midst of the crisis, despite the fact that the objective situation had not changed materially from an American point of view, Roosevelt suddenly reframed the outcome of war as a loss and became anxious to prevent it, even to the point of incurring risks he had previously judged unacceptable."
Farnham (1992, 227) further notes that FDR's change to a loss frame was most likely due to the prospect of impending war in Europe becoming "emotionally compelling to him."
McDermott (1992) used prospect theory to explain Carter's decision making in the Teheran hostage crisis. Specifically, she argued that Carter was willing to undertake the risky hostage rescue mission because he had been operating in the domain of political loss as evidenced by low public approval numbers and challenges from within his own political party. Biglaiser and DeRouen (2004) used prospect theory to explain how Latin American leaders, when operating in the domain of loss (economic problems centering around high inflation), were more likely to choose risky liberal economic reforms (see also Weyland 1996). Fanis (2004) showed how several economic groups in Chile in the mid-1970s were motivated to cooperate with others, even though doing so went against their own self-interest, and they did so because they were operating in the domain of loss.
Poliheuristic theory concentrates on the "why" and "how" of decision making which makes the theory relevant to both the contents and the processes of decision making. The term poliheuristic can be subdivided into the roots poly (many) and heuristic (shortcuts), which refers to the cognitive mechanisms decision makers utilize in attempts to simplify complex decision tasks (Geva, Redd, and Mintz 2000; Mintz and Geva 1997; Mintz et al. 1997).
The poliheuristic theory of decision making proposes that policy makers employ a two-stage decision process where in the first stage decision makers initially screen available alternatives utilizing cognitive-based heuristic strategies. In the second stage, when the decision matrix has been reduced to a more manageable number of alternatives and dimensions, policy makers resort to analytic, expected utility, or lexicographic rules of choice in an effort to minimize risks and maximize rewards (Mintz and Geva 1997; Mintz et al. 1997; Payne et al 1993). The first phase in the decision process typically involves a nonexhaustive search wherein decision makers process information across dimensions in an attempt to select "surviving" alternatives before the completion of the consideration of all alternatives along all dimensions (Mintz 1993; Mintz and Geva 1997; Mintz et al. 1997; Payne et al 2003). The second phase, then, consists of a lexicographic or maximizing decision rule used in selecting an alternative from the subset of "surviving" alternatives.2
Another key premise of the poliheuristic theory is its reference to the political aspects of decision making in a foreign policy context. The assumption is that the policy maker measures costs and benefits, risks and rewards, gains and losses, and success and failure in terms of political ramifications above all else (Mintz 1993). Furthermore, politicians are concerned about challenges to their leadership, their prospects of political survival, and their level of support (Kinne 2005). Domestic politics is the essence of decision. Because loss aversion (Kahneman and Tversky 1979; Levy 1992a, 1992b) outweighs all other considerations, leaders are driven more by avoiding failure than by attaining success (Anderson 1983). As Mintz and Geva (1997, 84) assert "the political dimension is important in foreign policy decisions not so much because politicians are driven by public support but because they are averse to loss and would therefore reject alternatives that may hurt them politically." The theory, then, suggests procedures for eliminating alternatives by adopting or rejecting courses of action based on this political heuristic in a two-stage decision process (Mintz et al. 1997).
The theory also posits that different decision heuristics may be employed in response to different decision tasks as a function of environmental and personal variations. This premise implies that these decision heuristics and strategies may be suboptimal (i.e., not always the "best"). Again, decision makers not only use different strategies depending on various environmental and/or cognitive constraints (Geva, Redd, and Mintz 2000; Mintz and Geva 1997; Mintz et al. 1997), they also resort to the use of different strategies en route to a single choice (Mintz and Geva 1997; Mintz et al. 1997). Brulé (2008, 266) extensively reviewed the poliheuristic literature and found that it was "progressive in the Lakatosian sense" and superior to other decision models.
Both expected utility and cybernetic theories of decision making assume that decision makers typically utilize compensatory decision rules in the first stage of information processing. In contrast, poliheuristic theory of decision making, posit that decision makers employ noncompensatory rules of decision making.
Freedman and Karsh (ibid., 35) state that during the Persian Gulf crisis of 1990-1991, Iraqi president Saddam Hussein rejected outright the option of withdrawing from Kuwait because "there was an absolute certainty in [his] mind of what could not be sacrificed—his political survival." This description fits poliheuristic theory which asserts that policy makers will use an attribute, or dimension-based process instead of an alternative-based approach for processing information. A dimension-based (intradimensional) strategy signifies that an individual focuses on a given dimension and then reviews information within that dimension across alternatives and then continues the process for another dimension (Payne 1976). Russo and Dosher (1983) specifically state that intradimensional, or attribute-based, processing is cognitively easier and hence more likely to be employed in cognitively demanding conditions.
The noncompensatory principle suggests that "a low score on one dimension cannot be compensated for by a high score on another dimension" (italics added) (Ford et al. 1989; see also Billings and Marcus 1983; Billings and Scherer 1988; Hogarth 1987; Payne, Bettman, and Johnson 1988, 1993). In other words, decision makers do not make tradeoffs between high and low scores. Hogarth (1987, 77) states that, psychologically, tradeoffs are difficult to make because decision makers find them difficult to execute as a result of "information-processing limitations."
Mintz (1993), in his study of the Persian Gulf War, began the extension of the noncompensatory principle to the field of foreign policy decision making. Specifically, Mintz (ibid.) argued that the decision by the United States to attack Iraq followed the noncompensatory principle of elimination by aspects (EBA). Mintz (ibid., 598) specified the noncompensatory principle for use in foreign policy situations by arguing that the political dimension is the paramount attribute. Moreover, "in a choice situation, if a certain alternative is unacceptable on a given dimension (e.g., it is unacceptable politically), then a high score on another dimension (e.g., the military) cannot compensate/counteract for it, and hence the alternative is eliminated" (italics in original) (see also Mintz and Geva 1997). By examining historical documents and written accounts of the deliberations leading up to the conflict, Mintz (1993, 606-7) was able to show that the political dimension was the most salient in President Bush's decision calculus with the military/strategic dimension also playing a critical role. He further points out that there was no comprehensive, i.e., search for compensatory tradeoffs, evaluation of all the alternatives and in fact it was quite obvious that President Bush did not consider the withdrawal option (ibid., 607).
Redd (2005, 129) examined Clinton's decision making in the Kosovo crisis and found that "President Clinton's decision was influenced by noncompensatory domestic political calculations and the strong influence of his Secretary of State, Madeleine K. Albright." More specifically, President Clinton's decision was heavily influenced by his concern over how Congress and the public would react to his initial decision not to act and then secondly to any possible casualties from a use of force. He greatly feared the loss of American lives and how such losses would influence his political fortunes. Similarly, Brulé (2005) examined Carter's decision making with respect to the Iran hostage rescue decision and found that Carter eliminated in a noncompensatory fashion all options that threatened his chances of reelection. Once he narrowed the choice set down to a few politically-palatable alternatives, he engaged in more of a maximizing process that focused on military and strategic concerns (Brule 2005, 99).3
Taylor-Robinson and Redd (2003, 95) examined Eisenhower's decision-making surrounding the 1954 U.S.-led coup in Guatemala and found that he "quickly eliminated the do-nothing option as too politically costly. . . . Because of United Fruit's framing, the U.S. public and press, as well as the U.S. Congress and key actors within the administration, were very worried about communism in Guatemala and were pressuring President Eisenhower to take action. Thus, in the first stage of the decision process, Eisenhower employed a noncompensatory decision rule that caused him to eliminate the option of taking no action because it was politically infeasible" (see Mintz and Redd 2003 for a more lengthy discussion of framing in the context of international relations). Similarly, Sathasivam (2003, 55) found that Pakistan had "no other choice" than to build and test a nuclear device in response to India's decision to do the same. "Politically, Pakistan's prime minister could not choose the do-nothing option . . . and realistically expect to survive in government" (70).
DeRouen (2003) even used poliheuristic theory to explain decisions "not" to use force, specifically in the U.S. decision not to intervene in Dien Bien Phu in 1954. Again, DeRouen (2003, 24) found that choosing not to use force was the politically expedient, noncompensatory option and "domestic politics constrained Eisenhower from using force." Astorino-Courtois and Trusty (2003, 47) investigated Israel-Syria peace decisions and found that the "sensitivity of the poliheuristic model to the political aspect of foreign policy choices added clarity and explanatory power to the analysis of apparent changes in Assad's interest in a peace accord, especially from the Rabin to Peres governments."
Below (2008) used poliheuristic theory to examine how Latin American leaders processed information with respect to ratifying the Kyoto Protocol. She found that leaders used cognitive heuristics that eliminated various policy options en route to choice and that their choices surrounding this environmental treaty were often not motivated by environmental concerns, but by domestic political calculations. James and Zhang (2005) and Sandal et al. (2011) examined crisis decision making in China, and comparatively in China and Turkey, respectively, using poliheuristic theory and found that it does quite well in explaining leaders' crisis decision making.
Groupthink is a theory of defective decision making (Janis 1982). More specifically, groupthink addresses defective decision making on the part of a cohesive decision-making group in which loyalty to real or perceived group norms takes precedence over independent, critical judgment. Conformity is the result of two possible factors: (1) Conformity from group pressure on the individual. Direct pressure comes from members of the group against dissenters, usually from the emergence of self-appointed "mindguards"; (2) Conformity from stress-induced cohesion. Time pressure may cause members of the group to withhold dissenting opinions for the sake of reaching consensus on a decision.
The consequences of groupthink and its resulting defective decision making include an incomplete survey of alternatives and objects, failure to examine the risks of preferred alternatives and a failure to reappraise initially rejected alternatives, poor information search, selective bias in processing information, and failure to work out contingency plans (Janis 1982).
Mintz and colleagues (Mintz, Mishal, Morag 2005; Mintz and DeRouen 2010, chapter 3) introduced the concept of Polythink, which is essentially the opposite of Groupthink: A plurality of opinions, views and perceptions of group members.
Polythink means "Poly (many) ways of perceiving the same decision problem, goals and solutions" (ibid 2005; 2010). Polythink can be contrasted with homogenous, uniform, monolithic worldview of group members.
Some of the consequences of Polythink are similar to those of Groupthink. "This is the case not because the group is thinking alike or sharing the same views but because the group is failing to carry out any significant collective thinking" (ibid). However, there are a number of important consequences that are unique to Polythink.
As is the case with Groupthink, Polythink is likely to lead to:
However, there are several consequences of Polythink that are very different from those of Groupthink (listed in Mintz, Mishal and Morag 2005):
Mintz, Mishal, and Morag (2005) examined whether Polythink (or Groupthink) existed at the Camp David 2000 Summit and concluded, based on in-depth interviews with all members of the Israeli delegation to the summer, including former Israeli Foreign Minister and former Chief of the IDF, that there was evidence of Polythink among members of the Israeli delegation to Camp David. Mintz and Wayne (2013, forthcoming) apply Polythink in an analysis of the 9/11 decision and a series of US foreign policy decisions (vis a vis Iraq, Afghanistan, Iran).
The organizational process model is, in some respects, similar to the cybernetic model discussed above. It views foreign policy as organizational output. One of the key concepts of the organizational process model is that of Standard Operating Procedures (SOPs), which are routinized courses of action. In every organization there are SOPs, which help determine how the actors in that organization may/will behave. Organizations do not attempt to estimate the probability distribution of future outcomes, as rational choice theory states. Instead, they avoid uncertainty by having and resorting to these SOPs. Therefore, the best explanation of an organization's behavior at time t is t-1, and the best prediction of behavior at t+1 is t.
This model is clearly grounded in the notion that policy makers are boundedly rational, as noted in the reference above to cybernetic models. Because of numerous constraints such as time pressure, cognitive effort, and cognitive skills, which result in an individual's or group's inability to maximize, they instead satisfice (instead of optimizing, they look for alternatives that are good enough).
Perhaps the most well-known example of the application of the organizational process model is Allison's examination of ExComm decision making during the Cuban Missile crisis (1969, 1971). However, Allison concluded that the bureaucratic politics model (see discussion below) performed better in explaining decision making during the Cuban Missile Crisis than did the organizational process or rational choice models.
The development of the bureaucratic politics model can be attributed to Graham Allison's The Essence of Decision, and his further work with Morton Halperin (Allison 1969, 1971; Allison and Halperin 1972). This work pioneered the conceptualization of the theory, and developed its structure as a way to explain government action. This in no respects means that there is wide acceptance of this model and its implications, and the text was met with both praise (Holsti 1972; Rourke 1972; and Wagner 1974) and criticism (Caldwell 1977; Krasner 1972). More recently scholars have questioned both the internal logical consistence (Bendor and Hammond 1992), and its generalizability to other political systems (Kasza 1987).
There are two fundamental aspects in understanding decision-making in this approach: 1) how decisions are arrived at, and 2) what beliefs the decision maker has about his/her advisors. Allison makes it quite clear that "the name of the game is politics: bargaining along regularized circuits among players positioned hierarchically within the government. Government behavior can thus be understood according to a third conceptual model, not as organizational outputs but as results of these bargaining games" (Allison 1971, 144). Government actors bargain over outcomes due to their different policy goals. The high level positions that the actors possess in the foreign policy environment allow them to participate in the bargaining game (Allison 1971, 164). Furthermore, Allison argues that "bargaining games are neither random nor haphazard" where "action channels structure the game by pre-selecting the major players, determining their usual points of entrance into the game, and distributing particular advantages and disadvantages for each game" (1971, 169-170).
As such actors will bargain over policy to maximize the influence they have in implementing a particular policy. They do so to "promote the positions their organizations have taken in the past" that "are consistent with the interests their organization represents" (Feldman 1989, 13). "Each player's probability of success depends upon at least three elements: bargaining advantages, skill and will in using bargaining advantages, and other players' perceptions of the first two ingredients" (Allison and Halperin 1972, 50). One of the ways in which organizational/bureaucratic actors may increase their bargaining advantage is by altering the group's membership to increase the number of similar advisors. Frequently, problems arising from group composition become the driving force behind restructuring executive agencies. For example, Roman and Tarr (1998) describe the restructuring of the JCS as a way to provide a unified, and more powerful, organizational voice to the president by getting everyone "on the same page." This illustrates how various group compositions affect bargaining advantages. Similar evaluations of information by organizationally align advisors increases the bargaining weight of the preferred option.
Smith (1984/5) examines how the composition of advisors to President Carter affected policy choice in the Iran hostage crisis. Smith divides advisors into groups according to perceived and generalized "stances" in dealing with the crisis. Smith concludes that group composition limited the evaluation of the alternatives. The multiple "hawks" at the table outmaneuvered the single "dove," Under Secretary of State Warren Christopher. We could expect a bargaining advantage to exist in such cases of multiple "similar" advisors wherein they would outweigh other single advisors. For example, Smith (1984/5) illustrates that while Under Secretary of State Christopher illustrated a number of dovish alternatives initially proposed by Secretary of State Cyrus Vance, "Brown immediately dismissed these as ‘not impressive,' and he was supported by Brzezinski, Jones, Turner, Powell and Jordan [the hawks], all of whom wanted to go ahead. Christopher was alone in his opposition to the plan" (19). Even when Secretary Vance returned to the bargaining table to present his objections to the rescue mission "his objections were met by ‘a deafening silence'" (Smith 1984/5, 20).
The second fundamental aspect of the bureaucratic politics model is that actors within the bargaining game represent organizationally formed preferences. Many authors have addressed whether choices are arrived at through interaction of organizational beliefs (Drezner 2000; Feldman 1989; Simon 1997; Smith 1984/5), whether manipulation affects the choice outcomes (Maoz 1990), or how bureaucracies may help prevent such manipulation (George 1980).
Many authors (see e.g., Allison and Halperin 1972; Downs 1994; Drezner 2000) argue that organizations will try to maximize their influence over a particular policy choice. A policy stance will develop according to either the real or perceived "organizational" mission. In recent work Drezner (2000) argues that "idea based" organizations will have a consistent set of goals and attempt to propagate these goals over policy maintaining their preferences over both the means and outcomes of policy. Organizational actors will have policy preferences that are in line with their stated organizational policy preferences. These organizational interests are "often dominated by the desire to maintain the autonomy of the organization in pursuing what its members view as the essence of the organization's activity" (Allison and Halperin 1972, 49). Experimental findings presented by Hermann, Geva, and Bragg (2001) indicate that actors that have a high sense of organizational commitment are more likely to prefer their organizational goals within groups having unanimous decision rules.
Hollis and Smith (1986, 275), in applying the bureaucratic politics model to President Jimmy Carter's decision to pursue a rescue mission in Iran, argue that organization "allegiances are so striking that one might even surmise that, had the participants switched positions, they would also have switched preferences.
Allison and Halperin (1972) argue that individuals in positions within organizations have preferences over alternatives that are determined by the nature of the position itself. "Given the face of the issue that he sees, each player must calculate how the resolution of the issue may affect his interests. This defines his stakes in the issue at hand. In light of these stakes he then determines his stand on the issue" (Allison and Halperin 1972, 49). Furthermore, "participants define national security according to the interests of the organization to which they belong. Career officials naturally come to believe that the health of their organization is vital to the nation's security" (Halperin 1972, 66).
Are policy makers beholden to their perceptions of the bureaucratic "actor as advocate?" Research by Tetlock (1999) indicates that experts often utilize defense mechanisms to reinforce beliefs when presented with disconfirming evidence. These actors' beliefs can play a significant role in choice selection as illustrated by Lamb's (1985) research of the Mayaguez Crisis, or Jervis' (1976; Jervis 1988) work on misperception of both intentions and consequences of decisions. Geva and Skorick (1999) also find that inconsistency of information plays a role in choice selection (for further discussion see Billings and Hermann 1998, Herrmann 1988, Taber 1998, and Vertzberger 1990).
Analogical reasoning implies that individuals operate by "transferring knowledge from past problem-solving cases to new problems that are similar to the past cases" (Sage 1990, 1). Sage (ibid.) notes that this type of decision process is fundamentally different from deductive inference or inductive inference-based reasoning. Interestingly, cybernetic theory relies on analogies in that information is funneled through a feedback mechanism and human behavior in general is therefore highly dependent on previous experiences. Vertzberger (1990) divides analogical reasoning into two stages: (1) establishing an analogy between two events and another analogy between their causes, and (2) inferring that similar causes lead to similar outcomes. Instead of relying solely on the information at hand, individuals resort to historical analogies as a guide in making decisions. The implication is that individuals are "less-than-rational" in their reliance on historical analogies, analogies which may or may not be accurate. Obviously, the success in using analogical reasoning depends on the accuracy of a given analogy to a particular foreign policy problem.
Prior to the Persian Gulf War, many opponents of the impending conflict attempted to portray the situation as "another Vietnam" and that the United States was being dragged into another foreign policy quagmire. Conversely, President George Bush repeatedly referred to Saddam Hussein as another "Hitler" and that the United States could not afford to "appease" another dictator as Chamberlain had prior to World War II.
Khong (1992) has written a book on the influence of analogies on decisions dealing with war. His primary goal was to demonstrate "how historical analogies, once invoked, influence the actual selection of policy options" (9). Khong (1992) develops what he calls the AE (Analogical Explanation) framework. The AE framework "suggests that analogies are cognitive devices that ‘help' policymakers perform six diagnostic tasks" (10). These diagnostic tasks aid the decision maker in dealing with complex decision environments and allow the cognitively limited decision maker to rely on "knowledge structures" to assist in decision making (13). Khong then uses the AE framework to examine decision making with respect to Korea, Dien Bien Phu, and Munich.
Applied Decision Analysis is a procedure for developing descriptive (and predictive) decision profiles of individual decision makers such as political leaders, financial decision makers, other policy makers (Mintz 2005). The procedure consists of two key steps. First, the analyst must identify the decision matrix---the alternatives, dimensions and implications of the alternatives corresponding to each dimension---of the policy maker. The second step involves the analysis of each decision through the use of multiple decision rules (e.g., elimination by aspect, lexicographic, poliheuristic, satisficing or utility maximizing) to understand how the policy maker made the decision and which decision rule and model have been used (ibid).
In identifying the decision matrix of the policy maker, the analyst/researcher should:
A decision matrix consists of a set of alternatives, the dimensions (or criteria) for selecting among these alternatives, and an assessment of the implications of each dimension for each alternative. Weights (or levels of importance) can be assigned to each dimension, if the analyst observes that dimensions should receive unequal weight in the analysis. The analyst can also use counterfactual scenarios to analyze potential actions and reactions of leaders.
The set of alternatives includes the likely courses of action a decision maker may reasonably consider when faced with some decision problem. For example, leaders may consider using force, applying sanctions, do nothing.
A dimension or a decision criterion, is an organizing theme relevant in evaluating the alternatives. Thus, if the leader considers using applying sanctions, using force or doing nothing in response to another country's behavior, relevant dimensions could be: military, diplomatic, political, economic.
The implications consist of a description of the likely consequences of an alternative for a given dimension. Each alternative has implications corresponding to each dimension. For example, in the case of applying sanctions, using force or doing nothing, each of these alternatives has economic, military, political implications.
Implications can be rated by the analyst, for example, from -10 (very bad) to +10 (very good). For example, if choosing the alternative "Do Nothing" is likely to result in a loss of public support for the decision maker, the analyst should assign a negative rating (very bad, -7 or -8) to the political implications of "Doing Nothing". In contrast, if "Apply Sanctions" is likely to lead to an increase public support, then this alternative should receive a positive rating (e.g., very good, or +8).
Weights indicate the importance level of each dimension, for example from 1 (not important at all) to 10 (very important). Thus, in the Apply Sanctions example, the analyst assigns different weights to the military, economic, diplomatic, and political dimensions.
Once a policy maker's decision matrix is constructed, it can then be analyzed to uncover the leader's decision rule.
In the second stage of the ADA procedure, the analyst determines the decision rule used by the policy maker to make a particular choice (ibid). For example, if it is clear that the policy maker stopped before considering all of the alternatives and selected a "good enough" option, then the decision rule may be consistent with a satisficing rule, such as in cybernetic theory.
It is possible to discern the decision rule used by an individual in making multiple decisions. Mintz et al (2006) termed this the Decision DNA of leaders. The analyst can examine multiple choices made by the policy maker and classify them accordingly. This will reveal a particular decision pattern, which can be further refined by collecting additional observations.
Data to be inserted into the decision matrix can be obtained by interviewing experts (often accompanied by the use of a Delphi technique), by using actual data, conducting content analysis of publications, or by relying on a key expert.
Previous research has used Applied Decision Analysis to understand decisions made by US presidents (Brulé 2005, Redd 2005), non-democratic leaders (Kinne 2005) as well as leaders of terrorist organizations—Bin Laden, leaders of Hamas and Hezbollah (Mintz et al. 2006, Chatagnier et al 2012). ADA has also been utilized for predictions of decisions of terrorist leaders (ibid).
National security and foreign policy decision makers are often prone to serious biases in decision making, such as a focus on one alternative, the acquisition of information which is "supportive" of this alternative while ignoring critical information which is contradictory to this option, susceptibility to the "preference over preference" bias and so on. Scholars of international relations, organizational behavior, judgment and decision making processes point to common biases and errors in decision making (Jervis 1976, 1988). Kahneman, Lovallo, and Sibony (2011) and Thaler and Sunstein (2009) write extensively on how biases influence decision making in strategic choices as well as in everyday life. They also review strategies for improving decision making and eliminating biases from decision making. Sage (1990, 227-29) has identified and listed close to thirty such biases ranging from adjustment and anchoring biases, primacy and recency effects, to "wishful thinking". His general conclusion is that "human decision behavior systematically deviates from (or is biased when compared to) a normative model that is assumed to be the optimal way to make the decision under investigation." (see also Jervis' 1976 work on motivated and unmotivated biases).
Markus and Zajonc (1985) state that bias occurs when individuals become victims of their own cognitive limitations. They divide biases into three general categories: (1) input biases, (2) output biases, and (3) operational biases. An input bias occurs when inference relies on data selectively to the extent that some classes of data are given more weight than others. A common input bias is the availability heuristic wherein an individual tends to interpret the frequency or probability of an event based upon what instances or occurrences are brought to mind (Hogarth 1987; Stein and Welch 1997; Tversky and Kahneman 1982).
Output biases, in contrast, reflect response preferences such as the acquiescence bias wherein there is a tendency to give positive rather than negative answers. Operational biases are generally rules of thumb for inference. Hindsight, attributing a causal structure to past events and blaming ourselves and others for not having foreseen these events, is an example of operational bias (Markus and Zajonc 1985).
Other biases frequently mentioned in the literature by scholars such as Forman and Selly (2001), Kahneman and Tversky (1979, 1984), Mintz and DeRouen (2010), and others are: Shooting from the Hip, Frame Blindness, Lack of Frame Control, Wishful Thinking, Plunging in (Forman and Selly 2001), misperception (Jervis 1976, 1988), gain/loss framing (Kahneman and Tversky 1984), the Poliheuristic Bias (wherein decision makers often reject attractive alternatives because they are damaging politically or personally, resulting in a suboptimal choice), the preference over preference bias, and the "locking" on a preferred alternative rather than examining a broad range of alternatives.
"Plunging in" is defined by Forman and Selly (2001, 4) as, "gathering information and reaching conclusions without thinking about the crux of the issue or how decisions like this one should be made." "Frame blindness" is defined by Forman and Selly (4), as "setting out to solve the wrong problem because your framework causes you to overlook attractive options or lose sight of important objectives." "Lack of Frame Control" is defined as "failing to define the problem in more ways than one, or being unduly influenced by the frames of others."(4) "Shooting from the Hip" is defined as "trying to keep straight in your head all the information relating to the decision rather than relying on a systematic procedure" (4). "Preference over preference" occurs when the decision maker has a clear preference for a course of action that affects his choices (Mintz and DeRouen 2010). Closely related to the preference over preference bias is the locking on a preferred alternative bias.
There is no central theory of cognitive bias. Scholars and researchers have identified numerous biases and many are incorporated into various theories of decision making while many scholars focus on a particular bias and examine its impact on information processing and choice (see e.g., Frisch 1993, Hogarth 1987, Sage 1990). Moreover, while many scholars discuss biases in information processing and choice (see e.g., Fiske and Taylor 1991, Forman and Selly 2001, Sage 1990), generally speaking little effort is made to systematically compare biases to each other with respect to origin, similarity, and impact on decision making. Moreover, there is no review of biases in decision making as applied to foreign policy and national security decisions.
For example, while frame blindness, as noted above, refers to the tendency for decision makers to solve the wrong problem because one's mental framework is focused on a particular piece of information or objective to the exclusion of others, how does this bias relate to other biases such as the preference over preference bias? We would submit that the preference over preference phenomenon refers to a decision maker identifying a preferred option regardless of the information presented whereas the frame blindness bias implies that the decision maker has misidentified the decision task at hand because of a fixation on a particular goal or objective, not necessarily a course of action per se.
During the July-August 2007 second war in Lebanon, Israel "locked in" on the air campaign course of action. Its chief of staff largely advocated the use of this option even though the air attack was ineffective in shutting off short term the shelling of Katyusha rockets from southern Lebanon into northern Israel. There was clearly a "preference over preference" bias in the Israeli military thinking pertaining to the air campaign as the preferred mode of fighting against Hezbollah (see the Winograd Committee preliminary report 2007).
Mintz and Redd (2007) have conducted an experimental analysis using the Decision Board simulator with high ranking military and Pentagon officials, focusing on selecting a counter-terrorism technology. The study revealed many of the cognitive biases reported above in decision on combating terrorism.
Decisions in Pearl Harbor, the Bay of Pigs, Operation Barbarossa, the Yom Kippur War of 1973, the 2003 Invasion of Iraq, and many other foreign policy decisions, suffered from motivated and unmotivated biases in decision making. Ironically, leaders are continually influenced by biases and even repeat the same mistakes.
In this article we have reviewed the major decision-making models that have been used in the literature examining foreign and national security level decision making. To begin, we have spelled out basic tenets of each theory, highlighting key concepts, assumptions and processes. We then illustrated how each theory has been applied in specific case studies of foreign and national security decision making.
Rational choice theory is the baseline standard against which other theories are often compared. Decision making is purposeful, with decision makers maximizing utility with respect to choice. However, rational choice theory is better adept at explaining decision making outcomes than foreign policy processes.
Cybernetic theory, developed precisely to respond to the rational choice assumption of utility maximization, instead argues that decision makers "satisfice" en route to choice. Policy makers select an alternative which is "good enough". Prospect theory counters rational choice theory assumptions, positing that decision makers evaluate gains and losses not in absolute terms, but based on a reference point. As a consequence, decision makers tend to be more risk-averse in the domain of gain and risk-acceptant in the domain of loss. Poliheuristic theory combines elements of rational choice theory with cognitive approaches to decision making. As such, it can account for both the outcomes and processes of foreign and national security decision making. Decision makers use multiple heuristics en route to choice in response to both cognitive constraints and situational factors, such as time pressure, information overload, etc. The theory also emphasizes the political calculations that policy makers often face, arguing that political calculations are often noncompensatory considerations, i.e., other factors cannot compensate, or override, the political dimension. Finally, according to Poliheuristic theory, decision makers tend to process information in two stages, with the first stage focusing on heuristics and cognitive shortcuts, and the second stage being more analytical in nature.
Organizational process and bureaucratic politics models concentrate primarily on the assertion that groups cannot be "rational" in all aspects that rational choice theory prescribes. Instead, groups employ standard operating procedures and/or make decisions based on bureaucratic interests and preferences. Groupthink, as the name implies, argues that individuals participating in group decision making suppress individual evaluations of the merits of a particular course of action. Unlike the other theories presented here, this theory focuses on particular problems in foreign policy decision making, i.e., groupthink is a theory of defective decision making. Thus, its utility may be narrower in application.
Polythink's central premise is that rather than a group erroneously focusing on one particular option, instead, too many policy options and alternatives are presented, which results in defective decision making because the group is unable to engage in effective collective decision making.
Applied Decision Analysis (ADA) is an analytic procedure aimed at understanding and forecasting decisions of national leaders. It consists of two key stages: uncovering the decision matrix of the leader and then uncovering the decision rule and decision "DNA" of leaders.
Analogical reasoning posits that policy makers use analogies from previous successful/unsuccessful decisions and decision making processes to help guide them as they deal with their own foreign and national security crises. However, decision makers often misapply analogies, which results in problems in the decision-making process as well as for the final outcome. Finally, biases and mistakes are common in political decision making.
We have highlighted many different historical and contemporary cases in which these various theories have been applied. In many instances, different theories have been applied to the same case, in which different facets of either the theory or case are manifest. Often times the manner in which a theory is applied to a particular real-world policy decision is a function of the objectives of the researcher as well as the specific strengths and weaknesses of the theory. Researchers may wish to illuminate the processes that led to a particular decision outcome, identify the decision unit, focus on how members in a decision making group interacted with one another, how previous decisions influenced policy deliberations, how information processing went "wrong," and how biases influenced the decisions. As discussed above, each of these models focuses on different aspects of decision making and can thus be applied to different features of the decision environment and/or the policy maker himself or herself. Perhaps these considerations help to explain the diversity of decision making theories and the lack of a central theme in how they are applied to real-world events.
Future applications of decision-making theories in the policy realm may benefit from reconsidering Allison's (1969) seminal approach wherein he applied three different theoretical models to one case, the Cuban Missile Crisis, and compared and contrasted their explanatory value. By so doing, scholars will be better able to examine the unique contributions of each theory in a comparative manner, rather than attempting to do so in isolation of other theoretical considerations.
1See, for example: Kahneman and Tversky (1979, 1984); Kahneman, Knetsch, and Thaler (1990, 1991); Kahneman, Slovic, and Tversky (1982); Tversky and Kahneman (1981, 1982a, 1982b, 1986, 1991); Tversky, Slovic, and Kahneman (1990).
2Mintz and Geva (1997) distinguish between the adoption or acceptance of alternatives vs. the rejection of options. These lead to the utilization of different decision rules.
3Note that Brulé (2005), McDermott (1992), and Smith (1984/5) each attempt to explain Carter’s decision making in the Iran hostage crisis, albeit from three different decision-making perspectives: poliheuristic theory, prospect theory, and the bureaucratic politics model, respectively.