University of Birmingham The Empathic Foundations of Security Dilemma De escalation

Security dilemma theorists have long recognized the importance of empathy to the de-escalation of conflict between actors caught in security dilemma dynamics, but they have left empathy undertheorized and have neglected to recognize its deeply contested nature. This article responds to this omission by bringing multidisciplinary literature on empathy to bear on security dilemma thinking. Contrary to some contemporary empathy research that draws attention to its automatic, unconscious, and intuitive properties, the article highlights the deliberate, effortful, and reflexive capacity to empathize across complex social contexts such as security dilemma dynamics. It shows how empathy of this kind can lead actors to moderate their positions on key issues at the heart of a conflict, reinterpret their interests, and broaden the zone of possible agreement between themselves and an adversary. The article demonstrates these notions empirically by locating empathy within the de-escalation of tensions between the United States and Iran between 2009 and 2016. Drawing on primary interview material with former U.S. officials, the argument is made that the development of specific empathic capacities by key U.S. officials played an important and unrecognized role in the de-escalation of security dilemma dynamics between the United States and Iran.

1 effects. 1 This sentiment has been echoed by others who have sought to explain how cooperation can be sparked within the context of security dilemma dynamics (e.g., Booth & Wheeler, 2008;Holmes, 2018;Jervis, 1976Jervis, , 1978Wheeler, 2018). But while these scholars have in various ways alluded to empathy in unpacking this puzzle, they have for the most part left it largely untheorized. In this sense, the security dilemma's treatment of empathy resembles what Hutchinson and Bleiker (2014) describe as the "somewhat paradoxical situation where emotions have been implicitly recognized as central [to IR] but, at the same time, remained largely neglected" (p. 494). In drawing out this connection between empathy and the security dilemma, the article also hopes to go some way towards reclaiming the psychological foundations of security dilemma theorizing. As Shiping Tang (2009, pp. 621-622) has observed, ever since Robert Jervis placed psychology front and center in security dilemma thinking (Jervis, 1976), the field has turned away from psychological factors in favor of studying material variables. (Re)locating empathy in security dilemma theorizing goes some way towards redressing this imbalance.
This article attempts to address these omissions by drawing upon diverse literatures in order to argue that the development and expression of empathy is pivotal to the de-escalation of security dilemma dynamics both prior to and during interaction. This is done in two steps. The first is to bring contemporary empathy research to bear on security dilemma thinking. Dominant understandings of empathy have emerged, particularly from neuroscience, that define empathy as an automatic neural activity which facilitates mind reading and emotional contagion with little need for conscious agency. This view of empathy is often described as basic, automatic, or affective empathy as it happens at the intuitive and unconscious level (Iacoboni, 2008;Rizzolatti, Fadiga, Gallese, & Fogassi, 1996). Research of this kind has been seized upon by some IR scholars as a means of explaining how adversaries can come to read one another in face-to-face settings, understand each other's intentions, and potentially overcome conflicts (Hall & Yarhi-Milo, 2012;Holmes, 2013Holmes, , 2018Wong, 2016). While there is value in these perspectives, not least in improving understandings of face-to-face diplomatic interaction, this article shows how they alone cannot explain the complexities of empathizing across security dilemma dynamics. It contrasts these perspectives with those that emphasize the deliberate, motivated, and conscious elements of empathy, which it argues are profoundly necessary to account for how empathy can develop in the complex context of actors ensnared in security dilemma dynamics. In making this distinction, the article argues that empathy development, which it locates primarily at the individual level, can positively impact upon de-escalation by leading actors to moderate their positions on key issues of contestation, reinterpret their interests, and broaden the zone of possible agreement between themselves and their adversary.
The second step is to illustrate these dynamics empirically. A "plausibility probe" (George & Bennett, 2005, p. 75) is conducted to highlight the relevance of empathy for security dilemma de-escalation. This is achieved by exploring the role of empathy in the de-escalation of tensions between the United States and Iran between 2009 and 2016; a case that is predominantly explained by prevailing material factors, such as multilateral sanctions levied against Iran (e.g., Fabius, 2016;Miller, 2018;Nephew, 2018;Solomon, 2016). Drawing on primary interview material with Obama administration officials, I highlight the development of specific empathic capacities by key U.S. officials towards Iran during this period and link them to U.S. policy changes that unlocked the standoff with Iran over the latter's nuclear program. The aim here is not to present a definitive causal account of empathy's importance to this episode, but rather to highlight gaps in existing explanations and to encourage further research on empathy in security dilemma de-escalation.
The article proceeds in four parts. It first outlines existing perspectives on empathy in the security dilemma literature and highlights various contributions and limitations. Second, it examines how "automatic" accounts of empathy fare in the context of the security dilemma. Third, it argues that empathy can lead to moderation of views and reinterpretation of interests. Finally, the article explores the development of empathy by key U.S. officials prior to and during the Iran nuclear negotiations in order to demonstrate the presence of empathy within an important contemporary example of security dilemma de-escalation.

Locating Empathy in Security Dilemma Theorizing
For a concept that is frequently characterized by fear and mistrust, it could seem counterintuitive to highlight the centrality of empathy to security dilemma thinking. Indeed, some scholars of the security dilemma have viewed empathy as futile due to the intentions of others being "inscrutable" and known only to the holder (Rosato, 2015; see also Mearsheimer, 2001). Others have similarly argued that the impossibility of accurately empathizing with one's adversary is itself what causes security dilemma dynamics in the first place, leading to an "irreducible" level of insecurity (Butterfield, 1951). Less fatalistic perspectives, however, reject this notion. Instead, they argue that it is entirely consistent to accept that one may never have absolute certainty about another's mental state-known in the empathy literature as "empathic accuracy" (Ickes, 1993)-while rejecting the idea that actors will therefore always fail in their efforts to understand and act upon another's motives and intentions (Booth & Wheeler, 2008;Herz, 1959;Jervis, 1976Jervis, , 1978. This was first articulated by John Herz (1959), who wrote, in a pioneering passage, that: For, if it is true… that inability to put oneself into the other fellow's place and to realise his fears and distrust has always constituted one chief reason for the dilemma's poignancy, it would then follow that elucidation of this fact might by itself enable one to do what so far has proved impossible-to put oneself into the other's place, to understand that he, too, may be motivated by one's own kind of fears, and thus to abate this fear. (p. 249) In this passage, Herz recognized the counterfactual logic underpinning fatalist accounts of the security dilemma; namely, that if these dynamics create conflict, it would then follow that understanding their workings would allow policymakers to take the steps necessary to mitigate their effects. Robert Jervis (1976), similarly saw an empathic understanding of the security dilemma as pivotal to overcoming its effects. As he argued, "The first step must be the realisation, by at least one side but preferably by both, that they are or at least may be, caught in a dilemma that neither desires" (p. 82). It is, for Jervis (1978), "failures of empathy" (p. 181) that create security dilemma dynamics and "empathy and skilful statesmanship" (p. 212) that contribute to how actors can chart a course out of them.
While Herz and Jervis did begin to orientate security dilemma thinking towards empathy, they did little to advance its explicit theorisation. Neither recognized the greatly contested nature of the concept, instead seeing it as an unproblematic part of everyday language. The first effort by security dilemma theorists to take empathy seriously and recognize its contested nature came from Booth and Wheeler (2008;see also Wheeler, 2018). Building on Herz and Jervis, they introduced the concept of security dilemma sensibility (SDS), which they described as "a particular expression of the general concept of empathy" (Booth & Wheeler, 2008, p. 237). SDS, they explained, is: an actor's intention and capacity to perceive the motives behind, and to show responsiveness towards, the potential complexity of the military intentions of others. In particular, it refers to the ability to understand the role that fear might play in their attitudes and behaviour, including, crucially, the role that one's own actions may play in provoking that fear. (p. 7) SDS's key contribution is its implicit focus on reflexivity. If one comes to understand that an adversary's behavior is motivated by fear, and that oneself has played a role in provoking that fear, then, by association, actors also learn something about themselves. SDS challenges the intuitive notion that empathy is solely orientated towards understanding others, instead suggesting that self-understanding can and should stem from this. Booth and Wheeler (2008) say little, however, about the relationship between the initial empathic understanding of one's adversary, and how actors then "show responsiveness towards" (p. 7) them. Indeed, the security dilemma literature more broadly provides no clear explanation for how this responsiveness stems from the "ability to understand the role that fear might play" (p. 7). It has shown the importance of an actor's initial empathic realization of the nature of their predicament, but it gives little indication of the role empathy may play in continuing down the path of de-escalation. Empathy at its most basic level is an internal mental process, and thus it should not be assumed that this internal process will automatically get externalized into empathically motivated behavior. If empathy is to be understood as a state of "immanent potential" (Head, 2016a, p. 102), then a more nuanced understanding needs to be advanced of how this potential can be realized and externalized communicatively towards the target of one's empathy. In short, how empathy is communicated, if it is even communicated at all, will likely be crucial to whether or not processes of de-escalation succeed.
What security dilemma theorizing has missed, therefore, is that the communication of one's empathy-as well as other practices of reassurance that follow from an initial realisation-is itself an empathically determined practice. Holmes and Yarhi-Milo (2017) capture this in part, arguing that "conveying empathy to the other may be as important as actually possessing it" (p. 109). While the reflexive inference that the other is acting from fear is vital, they recognize that for empathy to be effectual to deescalating security dilemma dynamics "actors must convey it to others" (p. 110). But they also remain unhelpfully wedded to the idea that empathy is an internal process that can be analytically separated from its external expression or conveyance. Discussing the failure of the 2000 Camp David Summit, they argue that Bill Clinton was "an empathic person, but on this occasion, he did not express it" (p. 107). However, an individual cannot be said to have strong empathic capacities in a relational setting if they are unable to communicate these capacities effectively. Empathy is a relational capacity (Main, Walle, Kho, & Halpern, 2017) of which communication should be considered a central indicator of the strength of an individual's empathy. If an actor cannot express their empathy towards their adversary in an effective manner, this is not just a failure of conveyance or communication but a failure of empathy itself. Empathy and communication are two sides of the same coin; empathy entails internalizing and appreciating the context in which a counterpart will receive a particular signal and then constructing that signal accordingly. Empathy in security dilemma thinking so far has lacked this communicative aspect, effectively ceasing to theorize empathy beyond its initial internal manifestations.
Despite these limitations, the security dilemma literature does provide a number of key points for thinking about empathy in this context. First, it cautions that while the search for "empathic accuracy" (Ickes, 1993) as the antidote for conflict-promoting misperceptions is a futile one, this need not rule out the place of empathy in security dilemma thinking. While not all empathic efforts will result in starting down the path of de-escalation, empathy is nevertheless a necessary if not a sufficient condition for processes of de-escalation between actors caught in security dilemma dynamics. And second, it has identified reflexivity as critical to empathy in this context as empathy entails not solely taking the perspective of another but also considering how the other experiences oneself. The next two sections build upon these propositions by exploring them in relation to the wider empathy literature.

"Automatic" Perspectives on Empathy
IR's neglect of empathy is not mirrored in other disciplines, where many are in the midst of an empathy boom. These debates are dominated by contemporary neuroscientific research that tends to define empathy in "automatic" terms (Iacoboni, 2008;Preston & de Waal, 2002;Rizzolatti et al., 1996). Automatic empathy can be understood as how individuals are able to automatically and instantaneously make sense of the observable actions and expressions of others. When one person sees another experience physical pain, it is the process that causes them to recoil and "feel" their pain. It is how individuals are able to recognize another's visual displays of emotion, such as facial expressions of sadness or anger. These processes happen almost instantaneously and are not the product of active deliberation or theorization. As leading neuroscientist Marco Iacoboni (2008) argues, empathy "is not an effortful, deliberate pretence of being in somebody else's shoes. It is an effortless, automatic, and unconscious inner mirroring" (p. 120). Instead of relying on folk psychological theories that are accumulated throughout life-a school of thought often labeled "Theory Theory" (Goldman, 2008)-automatic accounts propose that humans mirror the emotions and actions of others in their brains through complex processes of neural simulation. They contend that our brains are unconsciously and actively trying to work out what others are thinking and feeling through this shared neural architecture (Goldman, 2008;Holmes, 2018;Iacoboni, 2008).
Ideas about this mirroring system and other forms of "fast thinking" have become increasingly influential in IR. Most pertinent for the security dilemma literature, mirroring systems have been invoked in order to explain how adversaries can come to understand one another and how this can facilitate the transformation of adversarial relations. At its core, this literature argues that the discovery of the mirroring system, along with other similar phenomena, should prompt IR scholars to fundamentally rethink their approach to the other minds problem, particularly in face-to-face settings (Hall & Yarhi-Milo, 2012;Holmes, 2013Holmes, , 2018Wong, 2016). Contrary to the fatalist belief that another's intentions in international politics are inscrutable, these perspectives state that face-to-face diplomacy is a uniquely information-rich environment where diplomats are able to unconsciously discern the intentions of their counterparts while also inadvertently signaling their own intentions. The claimed implications of this are vast, with Holmes (2018) arguing that these mirroring systems provide "sophisticated, precise, and reliable access to the minds of others… [which] severely undercuts the problem of intentions… and more broadly, the security dilemma itself" (p. 241). Wong (2016), drawing on a similar body of research, similarly argues that face-to-face interaction "enables practitioners to exchange individual-level expressions of intentions" (p. 145). These views are prevalent in the neuroscience literature also, where Iacoboni (2008) argues that "By helping us recognize the actions of other people, mirror neurons also help us to recognize and understand the deepest motives behind those actions, the intentions of the other individuals" (p. 6). If these insights are correct, they fundamentally transform current understandings of the security dilemma as constituted by "existential uncertainty" (Booth & Wheeler, 2008) and suggest that IR theorists have overstated the difficultly of empathic accuracy.
While this body of work has done much to advance our understanding of the importance of face-to-face interaction in international politics, there are good reasons to question the applicability of these findings to the security dilemma. For one, "automatic" perspectives at times lack an appreciation of how empathy can be deliberate, motivated, and conscious, all of which are arguably necessary components for the development of reflexive understandings of self and other. Second, and relatedly, the exclusive focus on interpersonal interaction and "fast thinking" has obscured the fact that in certain contexts the development of empathy may well preclude this particular modality and develop at a distance and over time. These two points are related: If we are to account for how reflexive understandings of an adversary can develop in the absence of face-to-face contact, then we need to be able to theorize empathy as deliberate and motivated as well as automatic and intuitive.
These objections find support from a wide-ranging literature that attests to empathy being more than the process of automatically simulating the minds of others in face-to-face settings, particularly in complex social contexts. For the most part, scholars do not deny the existence of a mirroring system, nor the core claims of a simulationist theory of mind. Rather, they simply claim that automatic accounts are limited in the type of empathy they can explain. The neuroscientist, Jamil Zaki (2014), argues that "Although empathy can be automatic, by no means is it always automatic. Instead, this phenomenon is deeply context dependent and varies along with numerous situational features" (p. 1608 emphasis in original; see also Decety & Lamm, 2006;Singer & Lamm, 2009). The philosopher Karsten Stueber (2010) similarly contends that "One can grant the centrality of basic empathy to interpersonal relations while claiming that explanation, prediction, and interpretation of an agent's complex social behaviour requires knowledge of a folk-psychological theory as proposed by a theory-theorist" (p. 219; see also Mitchell, 2009, p. 130). This paints a broad picture of empathy as distinctly multifaceted with different systems being utilized for different contexts and social complexities. If we take instances of security dilemma dynamics as a particular social context, it should be clear that its unique characteristics and complexities will influence how, and indeed whether, empathy develops and how it is expressed. This follows recent research calling for greater acknowledgement of the ways in which unique contexts and political realities mediate and modify the function of emotion and emotional processes (Halperin & Pliskin, 2015;Hutchison & Bleiker, 2014;Tang, 2011, p. 727).
This fusing of politics, context, and psychology is particularly important for our present purposes as some scholars have argued that the development of empathy in cases of intergroup conflict is less likely to be of the automatic kind, due in large part to the cognitive effort required to take the perspective of those with radically different views to oneself (Coplan, 2011;Zaki, 2014Zaki, , p. 1611. Security dilemma dynamics, consequently, as a form of intergroup conflict, contain a number of distinctive characteristics that may directly moderate how-or indeed if-empathy develops and leads to de-escalation. These characteristics include-but are not limited to-(1) the protracted and often intractable nature of many security dilemma dynamics; (2) that the belief systems of actors embroiled within security dilemma dynamics that are often highly resistant to change (Finlay, Holsti, & Fagan, 1967;Larson, 1994); (3) the fact that these actors often hold peaceful/defensive self-images of themselves while imputing malign motives and intentions to the other (Booth & Wheeler, 2008;Butterfield, 1951;Wheeler, 2018); (4) the prevelance of negative images of the other that will often be institutionally and habitually embedded within state bureaucracies and organizations; and (5) that actors on both sides that will likely hold vastly differing interpretations of the key features and issues that constitute the conflict (Kelman, 1978(Kelman, , 1987. Taken together, these characteristics make instances of security dilemma dynamics hard cases for empathy development and emphasize just how difficult it can be for actors to take an adversary's perspective. This corresponds with research that shows empathy to operate most strongly in ingroup settings, effectively meaning that empathy is more commonly found between those who share similar traits and characteristics (Cikara, Bruneau, & Saxe, 2011;Crawford, 2014, p. 541). The implication of this is that empathy towards those in outgroups, where difference is a pervasive characteristic, is scarcer and more difficult to achieve than it is towards ingroups (Head, 2016b, p. 171). As the philosopher Amy Coplan (2011) has put it, "the more unlike a target we are, the more difficult it is to reconstruct her subjective experiences"; the consequence being that "with whom we fail to identify, we must work harder" (p. 58). Understanding the deeper meaning behind another's expressions, actions, or language therefore requires not solely unconscious simulation, but an active reflection of the broader social context from which these emotions, expressions, and actions emerge. While an account of automatic empathy is crucial to understanding the function of empathy at a basic level, there are good reasons to think that neuroscience "is not sufficient, in and of itself, to understand how emotions work in social contexts, especially complex ones" (Reus-Smit, 2014, p. 537).
This effortful and difficult kind of empathy becomes profoundly necessary if an actor is to try and understand the why behind another's behavior. Even where one can be relatively certain that they've accurately understood another's behavior, automatic empathy alone will tell them little about why the other is feeling or acting a particular way. For this, the empathizing actor must consider the broader context that gives rise to their behavior. Dan Zahavi (2014) argues that "We shouldn't forget that emo-tions… are about something… and it is not enough simply to pay attention to their expressions; we also need to look at the context in order to determine what they are about" (p. 163; see also Crawford, 2014, p. 54;Hollan, 2012). This is crucial for our present purposes as an understanding of the "why"or in other words the motivation-behind an expression or action that must be central to theorizing empathy in the security dilemma. Security dilemma scholars have highlighted that attention should be paid not just to how actors' read one another's intentions (how the other intends to behave), but also to their motivations (why the other behaves as they do) (Booth & Wheeler, 2008, p. 300;Glaser, 2010, p. 38). If empathizing with an adversary is to have any impact upon de-escalation, it is insufficient to only consider what the other feels; it is also necessary to try and comprehend why they feel it. If empathy is to instigate a process of de-escalation, then it should not just entail realizing that the other person is angry, but it must entail the act of reflecting upon why they are angry as well. Knowing, for instance, that Iranian leaders feel historically disrespected by the United States (Duncombe, 2016) is related but also fundamentally distinct to actively contemplating why this is the case. For this, an actor must consider the cultural, historical, and political context in which these feelings are embedded and expressed. This will likely entail a reflexive as well as reflective imagining, as oneself or one's group may have played a fundamental role in producing and reproducing that context.
Reducing empathy in adversarial settings to a problem of intentions therefore misses the point that motives-understood as the why behind particular emotion, expression, or action (Booth & Wheeler, 2008, p. 300)-are fundamental to theorizing why other actors behave in particular ways. As I will show later in this article, the development of a reflexive understanding of Iran's perspective on the nuclear issue and their relationship with the United States-which initially developed prior to face-to-face interaction-was crucial to unlocking the stand-off between these two states over Iran's nuclear program.

Empathy, Interests, and Agency
The previous section painted a picture of empathy development in adversarial settings as a difficult and demanding task, both cognitively and emotionally. This should not be taken as confirmation of the fatalist view that empathy in this context is either impossible or futile. Rather, it serves to press the point that empathy of this kind entails more than what "automatic" theories of empathy can account for. Empathy should, instead, be considered as the attempt to take the perspective of other actors in relation to a specific issue or set of issues, and, crucially, the reflexive consideration of the role oneself or one's group may have played in the production and reproduction of the other's perspectives. This section seeks to expound upon various aspects of this definition and to relate empathy of this kind to sympathy, interests, and agency.
The definition of empathy given above demarcates it from similar concepts such as sympathy, which is important given the frequent conceptual slippage in the literature (Wispé, 1986, p. 318). While there is no consensus regarding how to distinguish between these concepts (Zahavi, 2012), Ralph White's (1984) demarcation is highly pertinent. He argues that: Empathy… is distinguished from sympathy, which as defined as feeling with others-as being in agreement with them. Empathy with opponents is therefore psychologically possible even when conflict is so intense that sympathy is out of the question… We are not talking about warmth or approval, and certainly not about agreeing with, or siding with, but only about realistic understanding. (p. 160) Empathy does not, therefore, require the empathizing actor to agree with the worldview, actions, or beliefs of the other in order to try and take their perspective. As White notes elsewhere, "Empathy… does not necessarily imply sympathy, or tolerance, or liking, or agreement with that person-but simply understanding" (1991, p. 292). Sympathy on the other hand denotes some form of agreement or concern with the object of one's sympathy and is therefore different than attempting to understand another individual or group against which you may feel strongly. One may vehemently disagree with the actions and worldview of another, but it is still possible to imaginatively see things from their perspective and understand why, to them at least, their actions might seem justified. This manifested itself in the Obama administration's approach to Iran, with one senior U.S. official stating that "you have to have a certain degree of strategic empathy in any kind of negotiation…. It doesn't mean you have to accept or indulge another narrative, but you've got to understand it." 2 But while the act of empathizing with another does not imply agreement, it can lead to moderation of one's position on various issues and therefore increase the zone of possible agreement between two conflicting actors. In literature on conflict resolution and management, there is widespread recognition that modification of how one perceives their adversary-which entails considering how one's adversary views important elements of the conflict and wider relationship-is a key prerequisite to collaborative problem solving and successful negotiation (Kelman, 1978(Kelman, , 1987Ross, 1993). Marc Ross (1993), for instance, has argued that as "empathy develops, exchanges are more effective, parties are more open to a range of options that speak to each party's interests, and viable agreements become more attractive to all" (pp. 107-108). The development of empathy can therefore impact how an actor perceives their own and the other's interests, while opening up room for agreement on issues that had previously seemed intractable. Interests and preferences, which all too often are seen as the purview of rationalist scholarship, can therefore be considered "largely unmotivated and directionless without affective dynamics" (Hall & Ross, 2015, p. 856; see also Crawford, 2000;Rathbun, 2014) such as empathy. In other words, interests do not exist a priori to emotion and affect but are shaped by these phenomena (Damasio, 1995;Marcus, Neuman, & MacKuen, 2000). The development and expression of empathy by state leaders consequently does not supersede or replace interests with some form of altruistic motivation, but instead gives direction and meaning to interests.
Adopting this understanding of the relationship between interests and empathy opens up space for acknowledging the multiple pathways through which actors understand, articulate, and pursue their interests. While rationalist scholarship in International Relations has focused on coercive pressure and bargaining as the primary means in which actors pursue their interests (e.g., Sechser, 2018), rethinking the relationship between interests and empathy allows for a broader view that interests may also be effectively served through a range of activities such as the "pursuit of fair compromises and win-win outcomes" (Rathbun, 2014, p. 20). In this reading, the empathy Obama held towards Iran was pivotal in his view of how U.S. interests were best pursued. From the 2008 presidential campaign onwards, he rejected the idea, prevalent during the Bush administration, that not talking to and isolating one's adversaries was a viable way to pursue a state's interests. Instead, he consistently argued that only through engagement and dialogue could these be achieved (Obama, 2007(Obama, , 2008(Obama, , 2015. As John Kerry (2016), the U.S. Secretary of State in the second Obama administration, put it in the context of the Iran negotiations, "Trying to understand adversaries is not a favour we do for them; it's in our interests." Viewing empathy in this way refutes narratives of empathy being "soft" or even capitulatory and instead recognised empathy in conflict settings and foreign policy as a "serious, difficult, and important enterprise" (McNamara & Blight, 2003, p. 257).
What, then, are the prospects for empathy's development in cases of deep mistrust and conflict? The first point to make is that successful de-escalation, of which empathy is a prerequisite, requires effective leadership. While empathy is something that can be taught and encouraged, for example through effective mediation (Holmes & Yarhi-Milo, 2017;Kelman, 1987), it is also a dispositional trait. Different individuals have varying abilities and motivations to try and empathize with others. There is therefore significant individual-level variation. If a leader is already "primed for empathy" (Wheeler, 2018, p. 151) and possesses an empathic disposition, then the prospects of that leader recognizing security dilemma dynamics at work and responding accordingly increase. If a leader does not already possess some level of empathy or displays personality tendencies that suggest lower levels of general empathy (e.g., narcissism), then the prospects for de-escalation of this kind are less promising. There are, however, at least two realistic ways in which empathy towards an adversary might develop. The first is because of a particular experience or event that shifts the way a leader perceives their adversary. The classic example here is the Able Archer crisis of November 1983, where U.S. President Ronald Reagan came to believe that Soviet officials had wrongly interpreted a NATO missile simulation as an incoming nuclear attack. Reagan was so shocked that Soviet leaders could misinterpret U.S. intentions and genuinely believe that the United States were capable of such an act that he became "convinced that he had to find a way of reassuring Soviet leaders" (Wheeler, 2018, p. 152).
The second is through iterated rounds of face-to-face interaction, where leaders come to identify with one another through processes of humanization (Wheeler, 2018). This can lead them to reinterpret their individual problems as shared problems (Kelman, 1978). While both of these mechanisms help to explain how empathic capacities can emerge and increase in intensity, in the absence of either mechanism it seems unlikely that leaders once in office will become more empathic towards adversaries without some kind of stimuli. To explain how empathy towards an adversary can develop at a distance, therefore, a great deal of emphasis must be placed on the existing dispositional capacities of particular leaders. Thus, while empathy is something that can indeed be learnt, there is currently little reason to hold out much hope that leaders with low-empathic capacities will be able to develop this while in office. That first Obama, and then Kerry, already possessed high levels of empathy is an important reason why de-escalation with Iran happened when it did. As Tang (2011) has noted in the context of reconciliation, "successful reconciliation requires effective leadership" (p. 734). The same is true for empathically motivated de-escalation.
That de-escalation of this kind is highly dependent upon the right leaders being in office does not preclude the possibility that collectives can take on empathic traits through mechanisms such as contagion and institutionalization (Crawford, 2014). Arguably, effective institutionalization of empathy towards former adversaries is required if de-escalation is to lead to more lasting forms of cooperation. But it is important to recall that a key contextual characteristic of security dilemma dynamics can be embedded bad-faith thinking within state bureaucracies and institutions. Given this, it is logical to conclude that empathy of this kind will likely begin at the individual rather than collective level. If empathy towards adversaries (or former adversaries) is to become institutionalized into wider collectives, then these processes will likely be top down in nature. When individual-level empathic moves are successful in leading to some form of de-escalation, this can over time have a meaningful impact upon collective beliefs concerning former adversaries, and, in particular, lead to a collective form of redressing negative narratives and beliefs. Empathy that remains primarily at the individual level and is insufficiently institutionalized can, however, be fragile, leaving it vulnerable to rupture in the event of changes in leadership. As I clarify below, the case of U.S.-Iran relations is at once an important contemporary case of de-escalation dynamics between entrenched adversaries, while also serving as a cautionary note for the fragility of empathy development that does not become sufficiently institutionalized.

Empathy and De-escalation Between the United States and Iran
The implementation of the Joint Comprehensive Plan of Action (JCPOA)-the nuclear deal reached between Iran and the P5+1 (the United States, the United Kingdom, France, Russia, China, and Germany)-in January 2016 marked the culmination of a dramatic de-escalation of tensions between the United States and Iran. Although this de-escalation has undergone an equally dramatic reversal due to the Trump administration's withdrawal from the JCPOA and reimposition of sanctions (Al Jazeera, 2018), the 2009-16 period nevertheless stands as an important contemporary case of security dilemma de-escalation. These events have thus far primarily been explained as resulting from overwhelming coercive pressure held against Iran in the form of a stringent multilateral sanctions regime (Fabius, 2016;Miller, 2018;Nephew, 2018;Solomon, 2016). While the role of sanctions should not be ignored as they clearly played a role in enabling the conditions for successful de-escalation, insufficient attention has been paid to the development of empathy both prior to and during the diplomatic encounters themselves. This section rectifies this by drawing attention to the development of specific empathic capacities by high-ranking U.S. officials towards Iran and their nuclear program. Using the U.S.-Iran case as a plausibility probe, this section argues that these empathic capacities played a significant role in the de-escalation of tensions as they effectively communicated flexibility on the part of the Obama administration and opened up negotiating space on key issues that previously did not exist. What proved transformative was not just the reflexive inference that oneself may have been complicit in causing the other's insecurity, but that this inference was effectively communicated through concrete policy proposals. These claims are supported by data collected through a series of semistructured interviews with 13 former Obama administration officials involved in the negotiations and foreign policy towards Iran more broadly. The officials are kept anonymous due to the sensitive nature of the material they discussed.
The antecedents of de-escalation are found in the Obama administration's early efforts to reach out to Iran and change the nature of the relationship. From early in the administration's tenure, it was clear that Obama and some, though by no means all, of his key advisors held some level of empathy towards Iran that ran counter to prevailing U.S. foreign policy wisdom. This was born of the reflexive inference that past U.S. administrations were at least partly responsible for causing Iranian insecurity and stoking security dilemma dynamics between the two states. Obama sought to communicate this to Iran's Supreme Leader, Ayatollah Ali Khamenei, both privately and publicly. In a still classified April 2009 letter to Khamenei, which interviewees described to me, Obama explicitly stated that the United States had no malign intentions towards the Iranian government and was not trying to subvert it. 3 Similarly, in a public address to mark Nowruz, the Persian New Year, Obama was the first U.S. president to call Iran by its full name: the Islamic Republic of Iran. Officials have stated that this was an effort to try and convey legitimacy towards the regime while approaching it with respect. 4 While these moves were initially unable to cut through the deep levels of mistrust and were arguably undermined by the administration's later efforts to further sanction Iran , they were an early indication of the empathic capacity to reflexively consider one's own role in stoking insecurity and responding accordingly.
Obama made similar proclamations later in his presidency and directly linked this kind of empathy to the JCPOA. The day after the JCPOA was agreed Obama (2015) stated that: you have to have the capacity to put yourself occasionally in their shoes… the fact is that we had some involvement with overthrowing a democratically elected regime in Iran. We have had in the past supported Saddam Hussein when we know he used chemical weapons… I think that when we are able to see their country… in specific terms, historical terms, as opposed to just applying a broad brush, that's when you have the possibility at least of some movement.
Here, Obama acknowledged both the existence and the legitimacy of some of the grievances Iran has held against the United States and the importance of understanding, if not agreeing with, the narrative they tell about themselves and their relationship with the United States. These reflexive empathic capacities were also highlighted by interviewees, with one former official stating that, a "characteristic of his [Obama's] leadership was that he would ask the analysts and others in the room, how is the other country perceiving our actions?" 5 Another interviewee similarly said that, "I think you had a President and a Secretary of State [John Kerry] who, more so than any of their predecessors, were the type of people who were able to see issues from the perspective of their adversaries, and doing so was helpful." 6 Many of these empathic inferences were not formed through interpersonal interactions but were the product of prior-held beliefs and intuitions alongside sustained internal deliberation. Interviewees attested that this was a top-down process that began with the president, and later with Kerry, who became Secretary of State in January 2013. Current theories of empathy that emphasize its development through face-to-face interaction therefore need to be supplemented by additional perspectives in instances where face-to-face interaction was mostly absent. While there had been various face-toface encounters between U.S. and Iranian officials prior to 2013, they took place in large multilateral plenaries which U.S. officials described as "horrible meetings" where the Iranians were "obstructing any kind of diplomatic process." 7 It is therefore more accurate to say that many initial empathic inferences towards Iran developed in spite of face-to-face interactions rather than because of them.
In addition to a general reflexive recognition of Iranian insecurity and the U.S. role in producing it, these empathic capacities manifested themselves in two interrelated ways that proved pivotal to creating the space needed for productive negotiations. The first related to Iran's uranium enrichment capacity and the second to Iran's domestic political context. Regarding the former, Iran's uranium enrichment capacity was both the jewel in Iran's nuclear crown as well as the primary area of U.S. concern. Official U.S. policy during the Bush administration, which was backed up by multiple UNSC resolutions, was that Iran must suspend all enrichment activities prior to substantive negotiations. This began to change early in Obama's presidency. The administration dropped the precondition (MacAskill, 2009) and began private deliberations which gradually acknowledged that complete suspension of enrichment was an unrealistic goal that would be counterproductive to meaningful negotiations. 8 This capacity to recognize the importance of enrichment and to adjust the U.S. position accordingly was crucially formed before the diplomatic track was established. U.S. Deputy Secretary of State, William Burns (2019), reports in his memoir, which is substantiated by multiple interviews, that the decision to put the enrichment card on the table was one that was discussed privately within the administration long before it was communicated externally to the Iranians. 9 The president and his key advisors fretted over how and when to make this communication to the Iranians, ultimately authorizing Burns to indicate to the Iranians at a March 2013 secret bilateral meeting in Oman that the United States was willing to "explore the possibility of a limited domestic enrichment program as part of a comprehensive agreement" (Burns, 2019, p. 361).
This shift in U.S. policy was the culmination of an acknowledgment early in Obama's presidency that, as one negotiator put it, "If we [the United States] insisted on zero enrichment, we're not going to get a deal, period." 10 It worked to unlock the standoff between the United States and Iran, with Burns (2019) reflecting that "there would have been no agreement if we had insisted on zero enrichment" (p. 384). This perspective coalesced with the second empathic capacity, which was to understand that Iran's leaders needed to be able to sell any future potential deal to their domestic audience and that for them enrichment was a matter of modernity, progress, and national pride. As Abbas Araghchi, the senior Iranian negotiator, described it, their nuclear program and enrichment capability was "our source of national pride, our moon shoot" (quoted in Burns, 2019, p. 384). An inability to recognize Iran's perspective had contributed to dooming past rounds of talks, and there is good reason to believe it would have done so here as well. Demonstrating the importance given to understanding Iran's domestic constraints, one U.S. official expressed that "for both sides doing a deal like this requires… a lot of political capital… If you end up too far away from where the public is, it's not going to work… He [Obama] had to understand [Iranian president] Rouhani's politics as well as his own, in order to come to a place that would allow both to operate, to succeed." 11 American understanding of Iran's perspective therefore played a large role in determining and moderating the U.S. negotiating position. Indeed, U.S. officials recognized that internalizing Iran's narrative was crucial to success: "understanding what the other side's narrative is going to be is es-sential… in getting the agreement in the first place." 12 This narrative perspective became so embedded in the U.S. position that one negotiator claimed, "Secretary Kerry could articulate [Iranian Foreign Minister] Zarif's narrative about the [2015 Iran nuclear] deal about as well as Zarif could." 13 This also contributed to a general sense in the Obama administration that being able to empathize with the Iranians was a valuable commodity that was not only necessary if the negotiations were to be successful, but also if U.S. interests were to be met. As one interviewee reflected, Obama and his key advisors saw "the value of having empathy across foreign policy issues" because "understanding their perception… and how they'll respond to what we're going to do is ultimately going to make you more successful." 14 As a result, negotiators were able to consistently find solutions to issues that had previously seemed intractable. Moreover, it was done in such a way to allow Iran to save face and claim domestic victories while still meeting core U.S. interests concerning the future of Iran's nuclear program. One U.S. official said that "both sides were able to… adjust the details [of the agreement] around their objectives in such a way that allowed the other side to get what they needed." 15 The place of Iran's missile program in the negotiations also underscores this. While the Obama administration would have found it politically and strategically expedient for limits on Iran's missile program to be included in the JCPOA, U.S. negotiators explicitly and empathically recognized the strategic and symbolic importance of Iran's missile program to their sense of security and self. As one negotiator stated, "you have to understand the Iran-Iraq war… and how sanctions meant that Iran cannot have high-quality manned aircraft and that missiles were their only way of having a defence capability." 16 This evidence strongly suggests that the development of specific empathic capacities was an ever present and influential part of the de-escalation of tensions between the United States and Iran that culminated in the JCPOA. While empathy was primarily found at the individual level within senior administration officials, and operated in a top-down manner, there was also in the latter months of Obama's presidency a nascent process of institutionalization underway. After the JCPOA was 10 Interview 9 11 Interview 3 12 Interview 5 13 Interview 5 14 Interview 3 15 Interview 3 16 Interview 9 agreed upon in July 2015, there remained the hard task of implementing and verifying the deal, which required almost daily interaction between multiple levels of the U.S. and Iranian governments. The level of contact from July 2015 to January 2017 was unprecedented given the fraught history between the two states. It became so sustained and ingrained that U.S. officials began to describe their relations with their Iranian counterparts, who had long been considered enemies, as "professional" 17 and even "trusting." 18 One interviewee involved in implementation and verification of the JCPOA even spoke of a sense of "transitive credibility" that existed between officials, whereby the interpersonal connections between individuals-while still important-were beginning to be transcended by institutionalized forms of credibility and confidence. The administration's hope that the JCPOA could prove transformative as opposed to merely transactional was starting to bear fruit. 19 Despite these signs of a nascent institutionalization of empathy-and perhaps even trust-it was not sufficient to withstand a change of leadership in the United States. This is precisely what so concerned John Herz (1959) when he first introduced empathy to security dilemma theorizing. While empathy may in the present allow actors to understand that mutual fear drives conflict, Herz cautioned that there can be little certainty as to the "continuance of good intentions in the case of collective entities with leaders and policies forever changing" (p. 235). In the U.S.-Iran case, all progress towards institutionalizing empathy was abruptly undone when Donald Trump became president in January 2017, withdrew from the JCPOA in May 2018, and instead pursued a policy of "maximum pressure" towards the Islamic Republic (Cooper, 2019). While empathy in security dilemma dynamics may often proceed in a top-down manner, the U.S.-Iran case should sound a cautionary note as to the fragility of empathy that is primarily found at the individual level and that is insufficiently institutionalized. While a nascent process of institutionalization was underway, it was not sufficiently embedded to withstand the policy changes brought by the Trump administration. Individual-level empathy can be powerful in charting an initial course out of conflict. But turning successful de-escalation into more sustainable forms of conflict transformation will require increased attention of how individual-level empathy can become further institutionalized within the habits and practices of state bureaucracies and institutions.

Conclusion
The development of empathy is centrally important to the de-escalation of security dilemma dynamics. Yet, the security dilemma literature has primarily exemplified Crawford's (2000) observation that "Theories of international politics and security depend on assumptions about emotion that are rarely articulated and which may not be correct" (p. 116). The charge made here, therefore, is that in researching how security dilemma dynamics can be overcome, the literature has treated empathy as an unproblematic part of everyday language as opposed to a concept that is deeply contested across numerous disciplines. This neglect is unsurprising because, as others have noted, empathy has received scant attention from IR scholars despite the now well-established turn towards studying emotions (Head, 2016a, p. 95). By bringing the vast empathy literature to bear on security dilemma thinking, this article has attempted to shed further light on the role that empathy plays in instigating and sustaining processes of de-escalation between actors caught in security dilemma dynamics. In doing so, it has cautioned against the rapid embrace of solely "automatic" perspectives on empathy for this context. While merit can be found in researching the automatic, intuitive, and unconscious aspects of behavior, it highlighted the limitations of these perspectives when studying complex social contexts such as security dilemma dynamics. Instead, the article drew upon perspectives on empathy