Article Review 45 on “The Inscrutable Intentions of Great Powers.”

International Security coverIn his recent article “The Inscrutable Intentions of Great Powers,” Sebastian Rosato argues that it is far more difficult for states to signal their intentions than existing scholarship recognizes. He claims that the various signaling mechanisms proposed in the IR literature — both domestic-level characteristics and international-level behaviors — “at best…allow for marginal reductions in uncertainty” (51). Thus, Rosato supports the ‘offensive realist’ worldview that “great powers focus on the balance of power” and that “self-help is persistent, balancing is endless, the security dilemma is intractable…competition is the norm and cooperation is both rare and fleeting” (88).

H-Diplo | ISSF Article Review 45

H-Diplo/ISSF Editors: Thomas Maddux and Diane Labrosse
H-Diplo/ISSF Web and Production Editor: George Fujii
Commissioned for H-Diplo/ISSF by Thomas Maddux

Sebastian Rosato. “The Inscrutable Intentions of Great Powers.” International Security 39:3 (2014/15): 48-88. DOI: 10.1162/ISEC_a_00190.

Reviewed by
Brandon Yoder, Old Dominion University and the National University of Singapore
Kyle Haynes, Webster University

Published by ISSF on 27 November 2015

Always Inscrutable or Context-Dependent? A Critique of Sebastian Rosato’s “The Inscrutable Intentions of Great Powers”

In his recent article “The Inscrutable Intentions of Great Powers,” Sebastian Rosato argues that it is far more difficult for states to signal their intentions than existing scholarship recognizes. He claims that the various signaling mechanisms proposed in the IR literature — both domestic-level characteristics and international-level behaviors — “at best…allow for marginal reductions in uncertainty” (51). Thus, Rosato supports the ‘offensive realist’ worldview that “great powers focus on the balance of power” and that “self-help is persistent, balancing is endless, the security dilemma is intractable…competition is the norm and cooperation is both rare and fleeting” (88).

Rosato’s article makes several useful contributions. It comprehensively compiles barriers to signaling intentions that have been identified in the existing literature, and highlights several areas in which existing signaling mechanisms have been underspecified. The article also offers a welcome corrective to academic works that overstate the ease with which intentions can be signaled, and to policy analyses that naïvely treat non-credible signals as informative.

Nevertheless, we find Rosato’s claims unconvincing. Most broadly, his core claim that great powers cannot “discern the intentions of their peers with confidence” (51) is implausible. This would imply, for instance, that Japan is unsure whether the United States harbors aggressive intentions against it, and that Germany is meaningfully uncertain whether France will use force to resolve their next serious dispute. It is uncontroversial among most observers of international politics that these major powers, and many others, are indeed confident about each other’s intentions. Whether this confidence is driven by structural constraints or deeper beliefs about another’s basic goals, great powers regularly act based on information about their peers’ intentions.

Below, we focus on three overarching logical problems that underpin the article’s erroneous conclusions. First, the article mischaracterizes the literature and fails to recognize that existing work has long grappled with the signaling barriers it discusses. Second, the argument exhibits confusion regarding how states signal intentions and how those signals are made credible, leading Rosato to dismiss signaling mechanisms that actually convey significant information. Third, even granting the premise that individual signals convey little information, the article’s conclusions do not follow. States can aggregate many different sources of information to form confident beliefs about others’ intentions. Moreover, if states respond predictably to the distribution of power, as offensive realism claims, this would imply that states are actually completely informed of each other’s intentions as Rosato defines them.

Mischaracterization of Existing Signaling Literature

Rosato’s main argument is that great powers are unable to “reach confident conclusions about the intentions” of others (48). The signaling mechanisms he examines only “allow for marginal reductions in uncertainty” (51). Rosato notes that signals are often somewhat informative (51, 57, 73, 87-88), but argues that they never reveal a significant amount of information. Crucially, however, Rosato fails to specify any logical or empirical criteria that determine whether a signal is “significantly” informative. He therefore cannot demonstrate in this article why the “somewhat informative” signaling mechanisms he examines are incapable of crossing this undefined threshold of significance.

Absent criteria for what constitutes significant information, Rosato is free to treat any signal that is not completely informative as insignificant. Indeed, he conceptualizes beliefs dichotomously as either “confident” or “uncertain” at the outset of the article (48, fn 1). This conceptualization implicitly mischaracterizes any claim that signals are somewhat/sometimes credible to instead hold that signals always fully reveal states’ intentions. If true, the burden of proof would then merely be to show that signals are not always completely informative. This is reflected in Rosato’s empirical strategy. The article’s supporting evidence is consistently selected on the dependent variable: cases of signaling failure are cited while countervailing observations are ignored, an approach that is only valid for refuting the claim that behavioral signals are invariably credible.[1]

However, Rosato’s characterization of the literature creates a straw man argument. Few signaling theorists would claim that signals are uniformly credible under all circumstances, or that they perfectly reveal a state’s intentions. Indeed, the signaling literature has long acknowledged and grappled with the barriers Rosato identifies to show precisely when and how they can be overcome.[2] Rosato repeatedly quotes signaling optimists’ qualifications of their conclusions as evidence that signaling is effectively impossible (58-9, 62, 74, 85). But in fact, these quotes signify recognition by theorists such as Charles Glaser, Andrew Kydd, Robert Jervis and John Owen that credible signals are only possible under certain conditions and through certain strategies.[3] Thus, framing the analysis around the question of whether interstate signals are dichotomously either informative or not mischaracterizes the literature and moves the debate backward. The proper question, which the rest of the field has long been asking, is when signals are informative, and how that information affects states’ strategies.

This failure of the article to engage current debates is largely due to a fundamental confusion about what makes signals credible. Rosato accurately defines a credible signal of peaceful intentions as an action that is “more costly for an aggressive state than it is for a peaceful one” (70). Such signals convey at least some information because “peaceful” states are more likely than “aggressive” states to send them.[4] Yet the article thereafter ignores the logic of costly signaling, and thus radically understates the effectiveness of the signaling mechanisms it examines.

Most prominently, Rosato reproduces Evan Montgomery’s argument that aggressive states might misrepresent themselves as peaceful in order to trick others into letting their guard down. Conversely, peaceful states may have incentives to behave like aggressive ones in the face of onerous external threats.[5] Rosato then assumes that these incentives are insurmountable, such that peaceful and aggressive states will always exhibit identical behaviors, making credible signals impossible.

Yet scholars such as Charles Glaser, James Fearon, and Andrew Kydd have made great progress in resolving this “paradox.”[6] Their nuanced arguments do not claim that cooperative signals are always credible, as Rosato implies. Kydd and Glaser concur with Montgomery that under high stakes, high vulnerability, and low initial trust, peaceful states are likely to behave like aggressive types out of insecurity. However, under other circumstances incentives to misrepresent are reduced and insecurity can be overcome. For instance, Kydd shows using formal models that the efficacy of signaling often depends on the stakes of the interaction. Even with very low levels of initial trust and high vulnerability, states can gradually reassure each other by making small concessions at first, and building toward cooperation on high-stakes issues as trust grows over time.[7] Rosato does not engage with this logic, rejecting Kydd’s argument by simply stating “it is not clear why this should be the case” (74).

Rosato’s dismissal of audience costs as a source of credible signals is similarly unjustified. He argues “[t]here is little reason [for leaders] to state their intentions in an unequivocal fashion, because they understand that doing so will expose them to charges of incompetence if they later reverse course.” Strikingly, Rosato summarizes the precise mechanism that makes incurring audience costs a credible signal: it is costly to leaders that will later reverse course, but costless to highly-resolved leaders who intend to carry out their promises. Thus, highly resolved leaders are more likely than bluffers to incur audience costs, such that doing so provides information about the leader’s actual level of commitment.[8] Rosato articulates this logic, despite the fact that it contradicts his central claim.

In sum, by adopting a dichotomous conceptualization of signals as being either informative or not, Rosato treats the existing literature as being far more optimistic and less nuanced than it actually is. He then replaces this overstated optimism with overstated pessimism without providing any criteria for determining the credibility or “significance” of a signal. In doing so, his argument ignores the progress made by signaling theorists in determining how and when signals are informative.

Flawed Conceptualization of Intentions

The article is also problematic in how it characterizes the nature and scope of signals states send about their intentions. As such, it ignores many sources of information about intentions, and dramatically understates the credibility of those it identifies.

Rosato nicely defines intentions as “the actions states plan to take under certain circumstances” (52). He correctly notes that intentions can concern “any sphere of activity,” including economic, military, and ideological issue areas. Rosato also accurately characterizes intentions as “how states plan to realize their goals” (53). Here, goals are a state’s ultimate ends — the outcomes it would impose in the absence of external constraints. Intentions are the means it will choose to accomplish those ends, given the constraints of the international system. Thus, goals are dispositional qualities or “state-level attributes” (52) that are inherent to the actor and independent of the external environment. In contrast, intentions are a function of both a state’s internal goals and the external constraints it faces.

However, the article fails to acknowledge the relationship between signaling intentions and signaling goals. For Rosato, the main component of the international environment is the distribution of power, which is “easily measurable” (88). By this logic, states should generally be well informed of the external component of others’ intentions. Uncertainty about intentions must then derive primarily from a lack of information about others’ internal goals.[9] Thus, when states signal their intentions, they are actually signaling the degree to which their goals are compatible.

This is crucial because “aggressiveness,” the core dimension of intentions within Rosato’s framework, is not a goal or objective. He defines aggressiveness as the extent of a state’s “plans to threaten or use [military] force” (53). But states are not inherently aggressive. Rather, they selectively employ aggressive strategies in order to achieve their goals under specific circumstances. Thus, aggressiveness is not the construct states infer from another’s signals. States instead form beliefs about the compatibility of their goals and combine these beliefs with assessments of the external environment (which, again, is easily measurable for Rosato) to infer whether others are likely to act aggressively toward themselves. In short, aggression does not indicate incompatible goals. Aggression against a shared threat could even signal compatible goals, making it less likely that the sender has hostile intentions toward the receiver.[10]

This is problematic for Rosato’s argument for two reasons. First, he assumes throughout the paper that states treat aggressiveness as a basic goal or dispositional trait that generalizes across contexts. Second, in defining “aggressiveness” as the degree to which a state intends to use military force, the article immediately abandons its broader, multidimensional definition of intentions on the dubious grounds that great powers focus almost exclusively on security (i.e., military) issues (53). Yet the compatibility of states’ goals necessarily goes beyond military matters to include important issue areas across economic, ideological and security realms. Thus, “aggressiveness” is far too narrow to capture the range of intentions that states signal. Once we correctly conceive of aggressiveness as one dimension of intentions rather than a fundamental goal, and expand the range of signals beyond military behavior, it becomes clear that Rosato’s argument overlooks or wrongly dismisses many behaviors that credibly signal the compatibility of states’ goals.

Most broadly, Rosato argues that past actions are poor indicators of future intentions because external contexts may change over time. States that acted aggressively in the past may stop when their aims are satisfied, and states that have acted peacefully may become aggressive as constraints are lifted and new opportunities for expansion emerge (84). Here, the article clearly assumes that states treat aggressiveness as a dispositional goal, and form beliefs by assigning others a general reputation for being inherently aggressive or peaceful. Were this true, Rosato would be correct that behavioral signals are uninformative. But states do not simply assign general reputations for aggressiveness, ignoring the context in which others’ signals are sent. Instead, rational states will control for the context in which a signal was sent in order to form sophisticated beliefs about the compatibility of the sender’s underlying goals.

This point is nicely illustrated by Rosato’s discussion of Prussian/German aggression in the mid-nineteenth century, which was followed by a long period of peace. If states formed beliefs as Rosato indicates, by naïvely assigning a dispositional reputation for general aggressiveness, Prussian signals would indeed have been uninformative. However, the context in which Prussian actions took place allowed others to infer German Chancellor Otto von Bismarck’s goals of national unification and security from territorial rivals. For France, this certainly meant that Prussian goals were incompatible. But for Britain, Prussian aggression was compatible with its goal of creating a continental balance of power. Moreover, Prussia’s contemporaneous non-military behaviors — e.g., foreswearing of colonial expansion and general adherence to free trade — also signaled that Bismarckian Germany’s goals were largely compatible with Britain’s.[11] By restricting signals to military policies, the article overlooks these important sources of information.

This unjustifiable exclusion of non-military signals is further manifested in Rosato’s argument that institutions do not constrain states’ behavior, and therefore do not credibly signal states’ intentions. The article focuses on famously weak collective security institutions — the Concert of Europe, the League of Nations, and the United Nations — which indeed lack enforcement mechanisms. But it ignores the dense web of highly constraining economic institutions that govern the contemporary international order. Institutions such as the World Trade Organization credibly constrain states’ actions by facilitating beneficial economic cooperation, thereby raising opportunity costs of aggression. Moreover, joining such institutions also signals overlapping goals, since membership raises the costs of non-cooperative behavior. Rosato’s narrow focus on military actions ignores the primary mechanisms by which institutions yield information about intentions.

Logical Incoherence

Finally, the article’s conclusions do not follow from its premises. Even granting Rosato’s individual arguments, we still reject his claims that information about intentions is necessarily scarce and that offensive realist assumptions are empirically accurate.

Rosato considers six classes of signaling mechanisms, three at the domestic level (foreign policy goals, regime type, and ideology) and three at the systemic level (arms control, institutional membership, past actions), with several subcategories within each. In identifying so many different types of signals, the article inadvertently highlights the abundant sources of information available to states. Rosato rightly argues that individual signals occur in a unique context, and may therefore be only “marginally” informative of a state’s broader intentions in other contexts. But states do not respond to individual signals in isolation. Rather, they observe others’ actions over time, across issue areas, and under a wide range of external circumstances.[12] Thus, although any individual signal may be quite “noisy,” by aggregating information from many observations across many contexts states can glean significant information about each other’s intentions.

Rosato also correctly notes that if the domestic determinants of states’ goals are unstable and unpredictable, future intentions will be difficult to infer (86). Yet it is wrong to argue that this barrier is uniformly insurmountable. To do so conflates the potential that goals might change with certainty that goals will be highly unstable. Indeed, domestic attributes are often quite stable, as national culture and socioeconomic structure generally change gradually and/or infrequently. And although leadership change can occur, new leaders will likely face similar domestic constraints as their predecessors.

Finally, and perhaps most problematically, Rosato concludes that if information about intentions is scarce, offensive realism is descriptively accurate. Because states cannot estimate others’ intentions, they must simply assume the worst and “focus on the balance of power, which is more easily measurable” (88). But if states respond only to the readily observable distribution of power, they should be well aware of each other’s incentives, and thus highly confident that others’ intentions are universally “aggressive.” The irony here has been pointed out in numerous critiques of offensive realism: Rosato is describing a world in which states are actually quite certain about each other’s intentions.[13]

Brandon Yoder is an Assistant Professor of political science at Old Dominion University and a Research Fellow at the Centre on Asia and Globalisation at the Lee Kuan Yew School of Public Policy, National University of Singapore. His teaching and research interests include power shifts and interstate signaling, Chinese foreign policy and US-China relations, and comparative democratization.

Kyle Haynes is an Assistant Professor of political science at Webster University. His teaching and research center on international security, US foreign policy, and the interaction between domestic and international politics.

Copyright ©2015 The Authors.

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 United States License



[1] For instance, Rosato cites Germany’s failure to reassure Great Britain of its benign intentions prior to World War I (58). Yet Wilhelmine Germany sent numerous signals that allowed British leaders to update their beliefs that German intentions were hostile, including a sudden zeal for colonial expansion, mercantilist economic policies, and the rapid construction of a short-range battlefleet that was capable only of engaging the British Navy. The article also fails to acknowledge the concurrent Anglo-American rapprochement that occurred as British leaders positively updated their beliefs in response to U.S. foreign economic policy.

[2] Charles L. Glaser, Rational Theory of International Politics: The Logic of Competition and Cooperation (Princeton, NJ: Princeton University Press, 2010); Andrew H. Kydd, Trust and Mistrust in International Relations (Princeton, NJ: Princeton University Press, 2005).

[3] Glaser, Rational Theory, 36, 68; Kydd, Trust and Mistrust, 202; Robert Jervis, “Cooperation under the Security Dilemma,” World Politics, Vol. 30, No. 2 (January 1978), 168; John M. Owen IV, Liberal Peace, Liberal War: American Politics and International Security (Ithaca, N.Y.: Cornell University Press, 1997), 26.

[4] Rosato defines states’ intentions as being “aggressive” to the extent that they intend to use military force, and “peaceful” to the extent that they do not (53). We critique this typology of intentions below.

[5] Evan Braden Montgomery, “Breaking Out of the Security Dilemma: Realism, Reassurance, and the Problem of Uncertainty,” International Security Vol. 31, No. 2 (2006), 151-185.

[6] Glaser, Rational Theory; Kydd, Trust and Mistrust; James D. Fearon, “Signaling Foreign Policy Interests: Tying Hands Versus Sinking Costs,” The Journal of Conflict Resolution Vol. 41, No. 1 (1997), 68-90.

[7] Kydd, Trust and Mistrust, Chapter 7.

[8] Fearon, “Signaling Foreign Policy Interests.”

[9] Glaser, Rational Theory; Kydd, Trust and Mistrust. This problematic understanding of the relationship between intentions and goals is evident in Rosato’s treatment of “foreign policy goals” as a “feature” that states can use to signal their intentions (54, 57-58). Properly conceived, goals are the core component of intentions that states are signaling, not a means of signaling something else.

[10] This point is reflected in Rosato’s critique of democratic peace theory. His discussion mischaracterizes the theory to hold that democracies are generally peaceful. In fact, the theory claims only that democracies harbor peaceful intentions toward other democracies, but often use force against non-democracies. Rosato inadvertently demonstrates this by arguing that democracies may behave aggressively to export their democratic ideals (63). But democracies would have little reason to fear such aggression, making them more trusting of each other’s intentions than the intentions of non-democracies.

[11] Crucially, when subsequent German leaders adopted more expansionist goals, British policymakers observed key shifts in German policy and updated their beliefs about German objectives accordingly. Paul M. Kennedy, The Rise of the Anglo-German Antagonism, 1860-1914 (London: Allen & Unwin, 1980).

[12] David M. Edelstein, “Managing Uncertainty: Beliefs about Intentions and the Rise of Great Powers,” Security Studies Vol. 12, No. 1 (2002), 1-40.

[13] Glaser, Rational Theory; Glenn H. Snyder, “Mearsheimer’s World—Offensive Realism and the Struggle for Security: A Review Essay,” International Security, Vol. 27, No. 1 (Summer 2002), 149–173; Jonathan Kirshner, “The Tragedy of Offensive Realism: Classical Realism and the Rise of China,” European Journal of International Relations, Vol. 18, No. 1 (2012), 53-75.