Making stability a problem

By the early 1990s, two of the main currents in comparative politics—historical and rational choice institutionalism—each ran into theoretical problems.¹¹

While historical institutionalists had made great strides in understanding the sources of policy stasis and institutional stability, their theories had difficulty in accounting for change.¹² So these scholars began turning to ideas.¹³ Rationalists had a different challenge: they needed to more adequately explain stasis, for reasons internal to their models.

In line with their methodological individualist foundations, rationalists make agents' interests the basis of their theories. In short, what people want drives politics. To elevate such a claim beyond truism, rationalists add to their model assumptions about agents' preferences (the ordering of what people want) and behavior (some postulated function that agents maximize, such as “utility”). So far, so good. But as Mancur Olson informed us, left alone, self-interested maximizing agents will suffer endemic collective-action problems. And when one adds into the mix uncertainty over the possible outcomes of actions (multiple equilibria), then the set of possible choices facing agents becomes too complex for stability to occur naturally.¹⁴

Indeterminacy, defection, and a lack of successful collective action seem to be the unavoidable outcomes of a rationalist world.¹⁵ Such a conclusion pointed to a complex problem for rational choice. Its theory derived from a model that focused on stasis but predicted indeterminacy; and while the real world was always changing, it did not appear to be nearly as much “in flux” as the theory would predict. Therefore, because of this internal theoretical problem, rational choice theorists invoked institutions to explain stability. Institutions were invoked as instruments that help agents realize their structurally given interests. They were the glue that made the social world stick.

It quickly became apparent, however, that this approach created a second-order difficulty. For if supplying institutions was itself a collective action problem, then it made little sense to appeal to institutions to provide stability, since institutions, like other public goods, would be undersupplied.¹⁶

Given this problem, rational choice theorists needed to search for another source of stability; hence, the turn to ideas. But rationalism's individualist ontology and its understanding of interest meant that the adoption of ideas threatened to open a Pandora's box of complications.

A bridge too far?

The exemplar—and to some extent, inventor—of the ideational agenda in rational choice theory is Douglass North.¹⁷

Dissatisfied with the inability of existing rationalist theories to deal with the issue of institutional emergence, North developed a theory of institutional supply based on the concepts of transaction cost, uncertainty, and ideology. For North, institutions were still doing the “stabilizing,” but ideologies were deployed to solve the collective action problems that so complicated their construction. Basically, North argued that the ideologies individuals hold cheapen the price of having and acting on one's convictions.¹⁸ Therefore, by developing and deploying an ideology, agents can overcome the collective action problems inherent in supplying institutions, while still adhering to individualist microfoundations.

While attractive, North's adoption of ideas rests upon a paradox: for while he argues that ideas make collective action for institutional supply possible, he also argues that “institutions, by reducing the price we pay for our convictions, make ideas…important sources of institutional change” (my emphasis).¹⁹

In other words, ideas allow agents to create institutions by overcoming barriers to collective action, while existing institutions make ideas powerful—by doing the same. This seems contradictory, for if institutions make ideas “actionable,” then one cannot appeal to ideas to create institutions.²⁰ Conversely, if ideas create institutions, then one cannot appeal to institutions to explain ideational and thus institutional change. There is, in short, a circularity in the microfoundations of the model. Appealing to ideas does not solve the problem of supply, but simply pushes it one step back on the causal chain.²¹

Unsatisfied with this attempt to incorporate ideas into a rationalist framework, North next developed a theory of how individuals’ “shared mental models” aggregate together through communication to form ideologies. In this new approach, shared ideologies are the building blocks that make coherent collective action and institutional construction possible. This theory, however, engendered an even more serious problem than the first one did: it effectively reduced rationality to various individual psychological states. If one admits that each individual has unique perceptions of the world, then although one may aggregate these perceptions into common ideologies through learning, this is a far cry from any notion of substantive rationality that agents behave in accordance with their structurally derived interests. Arguably, the theory constitutes its own undoing.²²

To see why this is the case, recall that in rational choice theory, at least in its “thin” version, agents are assumed to be consistent in their choices and how they make them.²³

What people want is usually posited as structurally given, by virtue of a hypothesized material interest, with the realization of interests limited by the payoffs the agent faces (what the agent thinks she will get) and the informational structure of the environment.²⁴ In short, preferences are things with which to do the explaining, not things to be explained. Yet if one takes ideas seriously as causal elements, as “subjective mental models,” then one must see them as having an effect on the content of what agents want, and not just the order of what they want; otherwise, ideas would simply be the residue of preexisting interests and thus unimportant.

Yet if acquiring a new idea means changing one's conception of self-interest rather than just reordering one's preferences, and if different agents can hold different mental models regardless of the similarities of their structural positions, then the hard core of rationalist theory comes up for grabs.²⁵

One is no longer talking about ideas as an information surrogate that helps agents realize their existing interests or as a source of institutional supply.²⁶ Instead, one must admit the possibility, as ideationalist scholars have increasingly done, that ideas can in fact change an agent's interests.

But then, on a fundamental level, rational choice is no longer rational choice. When ideas are allowed to give content to interests, the sparse, elegant, predictive, and parsimonious structure of rational choice theory becomes compromised since one can no longer assume transitive preferences, given interests, or a coherent methodologically individualist ontology.²⁷

The explanatory work in such a theory necessarily occurs offstage, in the ideas that constitute interests, and these cannot simply be “given” by structural location or other material factors.

There's something about interests …

Rational choice theory encounters this difficulty because of a conceptual confusion in its notion of interest—that is, conceiving what agents want apart from the ideas agents have about what they want. Positing that an agent did something because his or her interest lay in x over y ignores the fact that the concept of interest presupposes unacknowledged, but important, cognates of interest, such as wants, beliefs, and desires. As decision theorists have demonstrated, however, these cognates are not separate from interests and must be considered as part of the concept itself.²⁸

Interest is therefore a “cluster” concept: one whose intention, or core meaning, is intimately bound with its extension, or its cognates (such as beliefs and desires).

Recognizing the cluster composition of interest is important. For example, to suppose an agent has an interest in achieving outcome y presupposes that she has a want for y given her beliefs and desires. However, if interests are constituted by beliefs and desires, and if agents are confused about what they should want, perhaps owing to uncertainties over likely future states of the world, then their interests may be unstable too. Transitivity—and hence, predictability—goes out the window, since the stability of this cluster cannot be taken for granted.²⁹

Holding ideas apart from interests, even analytically, makes little sense. But if rational choice theorists were to accept such reasoning, then specifying interests, and hence what people do, would become less about a priori structural determination and more about the construction of wants as mediated by beliefs and desires (i.e., ideas).

Consequently, turning to ideas threatened to rob rational choice theory of its essence: the ability to construct a model based on a thin notion of rationality that takes preferences as given and explains outcomes as a function of preexisting interests. After all, the point of the theory in the first place was to allow the analyst to do away with culture, psychology, ideology, history, and context, and to concentrate on what really mattered—a materially specified notion of self-interest as the basis of a predictive theory.³⁰

In sum, if one allows ideas to constitute interests—to determine the content of preferences and not simply their ordering—then rational choice theory risks losing what makes it distinctive in the first place. This is why, after a strong turn toward ideas in the early 1990s, rational choice theorists turned away from them rather quickly.³¹

Ideas and regime change

While rational choice theorists were making their ideational turn, another group of scholars also turned to ideas, norms, and culture, in order to extend the explanatory reach of their theories. Although disparate in their precise theoretical foci and conceptual arsenals, what united them all was the desire to challenge the notion of self-interest so central to rational choice's success.

Scholars in comparative politics, particularly those writing in the historical institutionalist tradition, have contributed much in this regard.³²

For example, Sheri Berman's book The Social Democratic Moment exemplifies how comparativists have used ideas to problematize interests and give better accounts of large-scale historical change. Berman seeks to explain why, despite having the largest and most powerful social democratic party in the world at the time, the interwar German Social Democratic Party (SPD) not only capitulated before Nazism, but, when in government, did not even attempt to fight Germany's economic crisis through remedial policies. In comparison, the Swedish Social Democrats (SAP), although a much smaller and weaker party in terms of both size and parliamentary influence, managed to avoid fascism and laid the foundations for the world's most successful and long-lived experiment in social democracy. The difference between the two paths taken, argues Berman, was “each party's long held ideas and the distinct policy legacies these ideas helped to create.”³³

Rather than relying on structurally derived notions of interests, Berman uses archival data to reconstruct the ideas that constituted both parties’ interests. She focuses on the “programmatic” beliefs of party leaders to explain policy choices. Though derived from the broader ideological context within which agents operate, “programmatic beliefs provide guidelines for practical activity and for the formulation of solutions to everyday problems.”³⁴

Such beliefs are not reducible to a priori interests. Instead, party leaders are ideational entrepreneurs who actively modify agents' beliefs about what their interests are. This does not mean that structures are irrelevant—far from it—but such structures do not come with an instruction sheet. There is still plenty of room for agents to make history apart from their structurally given interests.

The German case provides a robust challenge to any notion of materially derived self-interest. Berman reminds the reader that a peculiar thing about the interwar German SPD was its special relationship to Karl Marx and Marxism. Marx saw revolution coming first to Germany and anticipated a special role for the SPD in producing it. After his death, the SPD became a kind of defender of the faith for Marxism, which made policy innovation highly problematic. But the Swedes, unencumbered by such an ideational legacy, were free to interpret Marxism as a statement of ends rather than means; they reinvented socialism around the concepts of an inclusive “people's home” and reflationary macroeconomics, instead of adhering to the doctrines of historical materialism. When in power, the SAP was able to implement these ideas and advocate radical policies to stabilize capitalism. The SPD, the heirs of Marx, could hardly advocate saving capitalism, despite being in charge of the Parliament. Because of this ideational straightjacket, a trade union proposal to adopt compensatory spending measures that could have at least ameliorated the collapse of the economy was defeated in 1932.³⁵

In 1933, through ideational inflexibility rather than structural weakness, the SPD laid the ground for Hitler to come to power.

One could reject such a reading and model the interwar period as a game among the SPD, the Communists, and the National Socialist German Workers’ Party, with the SPD attempting to maximize the long-term benefits of the collapse of capitalism over the short-term benefits of stabilization, given the relative probabilities of each. Yet doing so would be to restate the question rather than answer it.³⁶

The beauty of Berman's analysis is that it shows how similarly placed actors, in similarly placed states, with similar problems, reacted in such utterly different ways.

Ideas, legitimacy, and distribution

On a more meso level, Hilary Appel seeks to explain why different postsocialist economies chose such radically different privatization strategies.³⁷

Given that privatization institutions, once chosen, have very different distributional effects, rationalist theories would predict that agents with common interests would attempt to form coalitions and build institutions in order to direct the stream of benefits toward themselves. What determines their success in doing so is their relative bargaining power, which itself is structurally given. However, viewing interests in this way, according to Appel, leads to erroneous predictions since such approaches cannot explain variation across cases. In contrast, an approach that takes ideology seriously gets it right.

Appel questions the notion of groups having different relative bargaining power because of structural factors such as asset specificity or factor position. She argues that while such a position may give an approximation of groups’ potential interests, it cannot be used as a basis to predict actual behavior. Doing so, “without bringing ideology and legitimacy into our analysis, would [allow us to] realize the bargaining position of…groups…only post hoc, namely, with the benefit of knowing how property had in fact been distributed.”³⁸

Appel argues that a focus on ideology avoids such circularity since paying attention to the ideological context of choice gives agents' interests content.

To make this case, Appel compares the Czech and Russian privatizations. While in each country a post hoc analysis could be constructed, such an approach would identify the wrong actors and posit nonexistent coalitions. Instead, a focus on ideology reveals why groups with ostensibly similar material interests could not form distributional coalitions, even though the groups’ relative bargaining power should have enabled them to do so. In the Czech case, the rapid delegitimation of the socialist political order weakened the prestige of industrial managers and organized labor. New parties were thus unwilling to ally with these groups, for fear of being tarred with the same brush as the old regime. For the same reason, labor and managers were unable to join with the old Communists. Both sets of actors found themselves devoid of possible coalition partners, so they had no significant input into the privatization process, despite their ostensible bargaining power and materially dictated common interests.³⁹

In Russia, though, the former Communist legislature was not delegitimated in the transition. Workers and managers could thereby form coalitions to channel the proceeds of privatization toward themselves. Like Berman, Appel uses ideology as an explanatory concept precisely because it goes beyond notions of given interests. By examining the ideological context of action—what is perceived as legitimate or not in a given context—Appel explains why distributive coalitions formed in one state but not in another.⁴⁰

Ideas and monetary politics

Bridging international relations and comparative politics is the central question of why European monetary integration has taken the form that it has. At first glance, this furrow would seem rather barren for ideational analysis to plow, since sophisticated rationalist theories of optimal currency areas and credibility already seek to explain monetary cooperation. Kathleen McNamara argues that such a conclusion is not warranted. For McNamara, ideas that give content to interests are crucial to any understanding of what channeled European monetary cooperation into the precise institutional forms we see today.⁴¹

How would it be possible for states to have preferences over a set of institutions with which they have no experience? After all, in rational choice theory, learning is Bayesian; and in a situation with no prior examples to draw upon, no priors could be ranked. Rational agents could then have no preference for x over y set of monetary institutions—hence, the difficulty of choosing monetary institutions.⁴²

Applying this logic to European monetary cooperation, McNamara asks why cooperation converged on the European monetary union, a common platform of exchange rate stability and low inflation. After all, if the agents could have no materially derived preference for such arrangements—and if the arrangements could not be a function of strategic bargaining, given this lack of prior interests—then where did they come from? For McNamara, as for Berman and Appel, ideas are critical explanatory factors.

McNamara posits a three-step learning model that does not depend upon any notion of structurally derived interests and substantive rationality. First, a period of policy failure—the supply shocks and inflation of the 1970s—threw existing institutions into doubt. Second, new monetarist ideas about the role of the state in the economy created a neoliberal consensus that saw fighting inflation as the top policy priority. Third, European policy elites generalized from the low-inflation and high-employment experience of postwar Germany, a country whose outcomes were attributed to having an independent central bank. Ideas drawn from this exemplar were combined with neoclassical ideas about credibility; the end result was a period of policy emulation and innovation in which new institutions of monetary cooperation were constructed.

Note that once again the outcomes McNamara seeks to explain are underdetermined by structural factors and rationalist logics. For example, if institutions were as rationalists perceived them (instrumental products designed to help agents realize their interests), the imprint of those interests should have been present in the new institutions. Yet neither sectoral nor factoral logics could indicate which actors were in fact important in the creation of these new institutions.⁴³

Moreover, neither the complexion of a state's governing party (left or right) nor a particular state's production profile could predict support for neoliberal cooperation. Ideas in this case cannot be mere reflections of underlying interests. Without attention being paid to the shared ideas that made up the neoliberal policy consensus, the precise shape of the institutions that emerged could not be predicted. Once ideas are taken seriously, such choices can be seen as reasonable and explicable, but they cannot be seen as rational.

Beyond neorealism and neoliberalism

Similar to what was under way in comparative politics at this time, real-world changes—specifically, the unexpected and peaceful end to the Cold War—opened the door to internal theoretical challenges already issued against the two dominant theories of IR in the late 1980s: neorealism and neoliberal institutionalism.⁴⁴

While the end of the Cold War seemed to cast doubt upon the predictive capacity of neorealism, for neoliberal institutionalists the problem was different. Because their body of theory had the same conceptual framework as did the rational institutionalist school in comparative politics, it was argued that many of the issues that IR had to grapple with in the post–Cold War era (the growth of transnational actors and advocacy networks, the importance of international organizations, et cetera) only had so much to do with increasing transparency and information flows.⁴⁵

As a result, while rationalist scholarship still flourished, especially in the international political economy, it began to do so with new competitors, particularly those who grouped themselves under the rather heterogeneous rubric of constructivism. Norms, identity, and culture—instead of ideas or ideologies—are the weapons of choice for constructivist theorists.⁴⁶

What unites them all is a desire to problematize interests as a basic category and move beyond rationalist explanations.

Ideas and (national) interests

Peter Katzenstein's edited volume The Culture of National Security constitutes a major statement in this new tradition of scholarship. Rather than use ideas or ideologies as explanatory concepts, the authors in this book “adhere to the sociological use of such concepts as norms, identity, and culture…to characterize the social factors they are analyzing.”⁴⁷

The authors argue that the international environment's cultural specificity engenders different identities among states. Power becomes as much about culture as it is about structure. Therefore, since “material power and coercion often derive their causal power from culture,” it makes little sense to take state interests as given.⁴⁸ Instead, interests develop from states’ identities, with materialism playing second fiddle to meaning. In such a world, norms are endowed with causal properties and are viewed as “collective expectations about the proper behavior for a given identity.”⁴⁹ Norms can have either constitutive effects that define who a state is—holder of the balance, lender of last resort—or regulative effects that define appropriate behavior given a specific identity.⁵⁰ Putting it plainly, if all states are not the same, then who you are will say a lot about how you will probably act, irrespective of material capabilities.

For example, Martha Finnemore demonstrates how the norm of humanitarianism has shaped state interests, and thus state policies, in ways that rationalist theories would not predict. She asks why states intervene in areas where there seems to be little or no material payoff for doing so. Finnemore finds the answer in the expansion of a regulative norm of humanitarianism. Beginning in the nineteenth century, the norm of who counted as human (and who should be protected) expanded from native subjects to extranational white Christians, encompassing practically all mankind by the late twentieth century. This expansion, Finnemore argues, did not simply constrain state behavior; it enabled state actions in areas where self-interest arguments would see no basis for action, by creating the sufficient cultural conditions for such actions. Seen in this way, interventions in regions where a material payoff is hard to discern, such as Cambodia and Somalia, becomes far more explicable as the response to an increasingly powerful regulative norm of state behavior. As Finnemore puts it: “As shared understandings about who is human…change, behavior shifts…in ways not correlated with standard conceptions of interests.”⁵¹

For other theorists in Katzenstein's volume, norms and identity offer more explanatory purchase when deployed as complementary concepts. For example, Michael N. Barnett argues that Arab alliance politics in the 1950s and 1960s are best explained by differing state identities rather than distributional capabilities. Barnett argues that “inter-Arab politics largely concerns the debate over the norms that should govern Arab politics…[and] that are directly related to issues of identity.”⁵²

In contrast to what rationalist and materialist theories would predict, Arab alliance politics was not based upon who made a good military match with whom, relative to external threats. Rather, it was based upon how Arab states garnered domestic legitimacy by appealing to a pan-Arab identity that superseded the Western-drawn and inherited boundaries of their nation-states. Yet Arab alliance politics was not “interests” seeking a justification in pan-Arab “ideology,” since the deployment of such ideas constrained the courses of action that the states who wielded these ideas could in fact take. While appealing to pan-Arabism may have enhanced domestic legitimacy, it also created norms of engagement as to how Arab states should act vis-à-vis one another.

For example, when Iraq and Turkey signed the Baghdad Pact in 1955, they undermined the domestic legitimacy of the Iraq government. While such an alliance may have made military sense, it flew in the face of the norm of pan-Arabism since alliances outside the pan-Arab core, especially with former colonial powers, were intersubjectively understood as being beyond the pale. Consequently, while states such as Jordan may have wished to join this alliance (given that it made military sense), the fact that it contradicted this norm of appropriate association undermined any possible accession to the treaty. Indeed, states such as Jordan and Lebanon were forced into alliances that made little security sense, but made sense when seen as a response to Iraq's violation of the norm. By contesting the normative basis of the Iraq-Turkey alliance, Egypt positioned itself as the defender of pan-Arabism; other states, not wishing to antagonize their domestic populations, were forced to support the Egyptian line.⁵³

In sum, both identity and norms give content to state interests and direction to state actions. A focus on material factors and assumptions of self-help and rationality cannot predict who allies with whom and why.

Ideas and state identities

A related exploration of the role of ideas and norms in world politics comes from Alexander Wendt. Rather than see ideas as giving content to interests, Wendt radically reduces interests to ideas. As he puts it, “[T]he meaning of power and the content of interests are largely a function of ideas.”⁵⁴

So the main factor structuring international politics is neither the distribution of capabilities, as neorealists would have it, nor the informational imperfections pointed to by neoliberal institutionalists. Rather, “the distribution of ideas” in the system is paramount.⁵⁵

Like Barnett and Finnemore, Wendt sees anarchy in the international system as underdetermining. Anarchy may lead to being a status quo state or to being a revisionist state. The core insight here is that all notions of interest rest upon assumptions about motivation that cannot simply be “read off” the structural context. As Wendt puts it, “only a small part of what constitutes interests is actually material…. The rest is ideational.” As such, we have to pay attention, once again, to “how preferences are constituted.”⁵⁶

Drawing on the work of cognitive psychologists, Wendt claims that in order to specify interests, one must previously specify the beliefs an agent has about what is desirable in the first place. The need to consider “what is desired” as a sociological construction rather than as a material given, argues Wendt, lies in political science's continuing acceptance of a dualism discussed earlier: analysts tend to concentrate on interests while bracketing beliefs and desires. In this view, “desire is constitutionally unrelated to belief.

Desire is a matter of passion, not cognition; and while beliefs activate and channel desires, they cannot be desires.” For Wendt, such a separation leads analysts to argue that if desires are wholly separate from beliefs, then they can be treated “in rationalist fashion as a means for realizing exogeneously given interests.”⁵⁷

Yet such a position is problematic, as noted above, since it confuses a distinction in theory between ideas and interests with a synthetic distinction in the real world. In contrast, Wendt contends that in the real world “we want what we want because of how we think about it” and not because of any innate properties of the object desired.⁵⁸

Seen in this way, the distinction between interests and ideas about our interests collapses. Wendt writes that “biology [and structure] matters little. Human nature does not tell us whether people are good or bad, aggressive or pacific, even selfish or altruistic. These are all socially contingent, not materially essential.”⁵⁹ In light of such an analysis, the contention that ideas are epiphenomenal to—or, at best, an adjunct to—materialist explanations seems all the less convincing.⁶⁰

Ideationalism and constructivism

Unlike rational choice's turn to ideas, a turn that threatened to throw the predictive baby out with the reductionist bathwater, the work of Berman, Appel, and McNamara marks a progressive extension of earlier historical institutionalist scholarship. In fact, these scholars have transformed this body of theory. Whereas earlier historical institutionalist analysis saw institutions as sources of agents' preferences—and therefore as sticky and path-dependent—ideational institutionalism changes institutions into something more dualistic.

This conception of institutions is consonant with that developed by William Sewell, who sees institutions as composed of both schemas and resources.⁶¹

In this view, ideas (schemas) and institutions (resources) are mutually supportive and antagonistic. Agents can instrumentally use ideas to delegitimate, contest, and refashion existing institutions.⁶² Yet such institutions, as the outcome of social action, give meaning to the material environment in which agents find themselves—and thus give content to what agents want in the first place.

This interactive effect is seen most powerfully in the work of Robert Lieberman.⁶³

Like many scholars in the American political development tradition, Lieberman is interested how institutions can be seen as sources of political change as well as stability.⁶⁴ For Lieberman, “any political…outcome is situated within a variety of…institutional and ideological patterns” that together comprise a political order, and such an order is neither a stable equilibrium nor a path-dependent set of structures. Rather, the context of political action is composed of multiple institutional and ideological locations that are layered together. These orders can generate frictions for change when their elements embody contradictory logics. Instead of viewing politics as path-dependent along a single vector, Lieberman argues that “at any given moment, politics is situated upon multiple paths.” He continues: “[W]hen these paths are consonant…the result may be stability; when they are not…the result will more likely be instability and uncertainty.”⁶⁵

Lieberman powerfully illustrates the analytic purchase of this conception of institutional change in his explanation of American civil rights politics. He focuses upon how the Equal Employment Opportunity Commission (EEOC), a weak bureaucracy with little enforcement capacity, was created at the juncture of several institutional and ideological orders as a result of the civil rights struggle. Friction among these orders (the imperatives of electoral politics, the demands of the civil rights movement, the contradiction between color-blind liberalism and race-conscious affirmative action ideas) built up to produce an outcome (strong enforcement of color-conscious policies by the EEOC) that neither institutional nor ideational approaches could predict on their own.⁶⁶

In international relations, constructivist scholars who use identity, norms, and culture as explanatory variables have similarly challenged many of the basic assumptions of rationalist theory. However, given their disparate conceptual arsenal, a question remains as to how these new concepts fit together. Do they add explanatory value, or do they simply label the same concepts with different words? I argue that these concepts—norms, identities, and culture—do in fact fit rather well together, but are nonetheless different in kind from the concepts deployed by comparativists.

In the work of such theorists, one can view culture, norms, and identities as operating on different levels of analysis. However, whereas traditional international relations theory tends to prioritize one level at the expense of others, in a constructivist framework the interaction across levels becomes all important. For example, if norms operate on the meso level of analysis, culture can be seen as its macro correlate. Culture can be conceptualized as being composed of multiple competing norms and identities that set the evaluative and cognitive standards of world politics as a whole. Evaluative standards of behavior are governed by the norms of the system, while cognitive standards are the rules by which actors are recognized as such.⁶⁷

Thus, specific norms are a function of, and embedded in, particular cultures. Following this logic, state identity refers to the micro level of theory, to the socially constructed interests of states defined by this wider normative (meso) and cultural (macro) context. Within this framework, scholars have developed a research agenda that offers a fundamentally different way of viewing world politics from that offered by rationalist accounts.⁶⁸

These concepts are, however, not simply the international relations analogues of those developed by ideationalists in comparative politics, since the two sets of concepts refer to different worlds, both theoretically and empirically. But what they do share is striking: a focus on how agents' interests are specified and how nonmaterial factors constitute those interests. That these two bodies of scholarship appeared at the same time and voiced essentially similar concerns tells us something about how the discipline evolves—a point I now turn to in conclusion.

Plus ça change, or evolution, in political science?

In this article, I have argued that, like the world around it, political science is undergoing another round of change. The dominant theory of our discipline, rational choice, built its justified reputation on being able to explain statics with concepts such as equilibrium, information, veto points, and path-dependence. Yet it did so with a theory that predicted a world where stability was hard to maintain. Indeed, some of the most successful applications of rational choice theory in political science were so impressive because they made the discipline think about how achieving stability was a problem. Olson explained why collective action does not happen easily, Robert Bates explained why Africa failed to develop, and Douglass North and Robert Thomas explained why growth is so difficult to achieve.⁶⁹

All of these analyses, milestones though they are, focus on statics rather than dynamics. Change is not explained easily within such frameworks.

Yet events inside and outside the discipline in the past decade have made the search for adequate models of change all the more pressing; and in an effort to be responsive—while paradoxically addressing internal theoretical problems concerning stability—some rationalists turned to ideas. But the cost of doing this was to threaten the hard core of rational choice theory itself, given interests, thin rationality, prediction, and generalizability. In turning toward ideas to explain stability, rational choice began to encounter its own limits. That this is the case should not be a surprise, for something similar has occurred in political science before.

At the moment victory was declared in the behavioral revolution, real-world changes and internal theoretical developments together signaled the limits of the behavioralist framework.⁷⁰

Its central concepts of positivism, pluralism, and modernization were called into question. Consequently, the field splintered into a variety of approaches, one of which was rational choice theory.⁷¹ Analogous to what we saw at the end of the behavioral revolution, some of the most innovative works in contemporary political science are not being developed within what is arguably the mainstream disciplinary approach. They are instead constructed because of it and in opposition to it.

One area of growth lies in the turn to ideas by nonrationalist scholars. The fact that these works have appeared at a time when rational choice's claims are so hotly contested, and that they have done so in direct opposition to its basic tenets, is significant. It suggests that progress in political science is perhaps dialectic rather than paradigmatic. Without the very real and important advances that rational choice theory made, and therefore the problems it left unattended, new scholarship that seeks to advance beyond it could not have come into being. Whether or not such scholarship on ideas does in fact constitute an advance beyond rational choice theory is obviously contentious, but I have made a case here for why it does.

The limits of ideational explanations

Legitimate methodological concerns remain in this nonrationalist turn to ideas. Such scholarship does not come with the elegant tool kit that rational choice offers, and some scholars may be justifiably wary about trading in what they have for an uncertain future.⁷²

Indeed, while the scholarship reviewed in this article is interesting precisely because it is willing to tackle the issues with which rational choice has problems, there is a price to be paid in generalizability, conceptual clarity, and rigor. However, stark opposition and all-or-nothing alternatives need not be the choice facing the discipline.

As the scholarship reviewed here demonstrates, ideas give content to preferences and thus make action explicable, but they need not be conceptualized in this way. Ideas can also been seen as power resources used by self-interested actors or as weapons in political struggles that help agents achieve their ends. Such an approach obviously is compatible with rational choice and is of great theoretical importance.⁷³

It would be a mistake, though, to limit ideas to such a role and to assign analytical priority to structurally given interests as a matter of course. The work of comparativists shows what would be lost to us in terms of explanatory power if we did limit ideas in this way, while IR constructivists demonstrate that the price paid for ignoring such factors may be to foreclose entire research agendas.

Nonetheless, rationalists may object that my analysis of the evolution of political science draws all-too-dark distinctions. After all, some rationalist scholars have attempted to apply “thick” notions of rationality to political problems that get at exactly the same issues as the approaches used by ideational theorists. But such a response is fundamentally unconvincing. Regarding thick rationalist approaches: the point of rational choice theory was to do away with the need to posit unobservables as causes, so bringing them back in merely expands the theory beyond its own epistemological limits and robs it of its distinctiveness. As Wendt points out, “[A] key assumption of the traditional rationalist model is that beliefs have no motivational force on their own; they merely describe the world.”⁷⁴

While one can indeed treat ideas as instruments, they are also much more than that. Ideas are not simply surrogates for information, nor are they shortcuts to structures that are somehow lodged between the ears. Ideas are powerful because they are intersubjective. They constitute our interests. They do not simply alter our strategies, and they do not come with an instruction sheet.⁷⁵

Critics of my approach may note that thick versions of rationality need not posit egoism as an assumption; altruism would do just as well, and rationality can be bounded as well as instrumental.⁷⁶

Yet such responses run into their own problems. Just as people are never always selfish, they are never always altruistic. The point, then, is to understand how and why different behaviors pertain, which unavoidably assumes that interests are something to be explained and not something with which to do the explaining. Positing alternative motivations a priori keeps the concept of a given self-interest intact by simply substituting one preference function for another, thereby making anything and everything consistent with a rationality assumption. But the works under discussion here show both how productive it can be to do away with such an assumption and how limiting it is to maintain strict methodological individualist foundations.

Some may also argue that I overstate what rationalists in fact claim. In practice, no rationalists assert that models of strategic calculation based upon preexisting preferences always provide useful explanations of politics. Rather, most would argue that such models sometimes provide useful explanations of politics. This more moderate stance runs into the problem that Donald Green and Ian Shapiro, and Gerardo Munck, have called “segmented universalism,” in which the restriction of a theory to particular domains “where it is sometimes useful” weakens its distinctive claims to superior generalizability and scope.⁷⁷

Moreover, if the point of a theory is to abstract from context, then such abstractions can offer no guide as to where the appropriate context for such a theory is.

This is not to say that models of strategic calculation based upon preexisting preferences never provide useful explanations of politics; they do. But surely it costs the discipline as a whole to specify one approach as intrinsically better than another before even asking what the research question is, especially when a priori domain specification is such a difficult task. Taking ideas seriously does not mean abandoning social science; it means accepting that the limits of one set of theories open up space for others to move forward and enrich the discipline.

I am not suggesting that we embrace a naive, “the best theory eventually wins out” notion of progress.⁷⁸

However, the field evolves when we discover the limits of what has gone before. Exploring those absences is path-dependent and uneven, not linear and progressive. All theories have their biases, and political science as a field may well swing from one theoretical support to another. It may not be linear progress, but I would suggest that progress is still being made.

Rational choice theory is far from finished. Reports of its death are surely both premature and exaggerated. However, its very success has created a situation where it again has to share the field with other approaches. That this is the case should not be a cause for concern. For if what happened after the behavioral revolution is anything to go by, running into intrinsic theoretical limits is nothing to fear, since this is exactly how political science moves forward. The literature on ideas, norms, identity, and culture has given theoretical voice to the limits of existing theories. Such a development should not be rejected, for it is only through a dialectic process that the field evolves.