Content uploaded by Jennifer Mitzen
Author content
All content in this area was uploaded by Jennifer Mitzen on Mar 03, 2015
Content may be subject to copyright.
Ontological Security in World Politics:
State Identity and the Security Dilemma
JENNIFER MITZEN
Ohio State University, USA
This article proposes that in addition to physical security, states also
seek ontological security, or security of the self. Ontological security is
achieved by routinizing relationships with significant others, and actors
therefore become attached to those relationships. Like its physical
counterpart, the ontological security motive is a constant. But states
may adhere to routines rigidly or reflexively, and variation in attachment
style has implications for security-seeking. This article conceptualizes the
individual-level need for ontological security, scales it up to states, and
applies the ontological security-seeking assumption to the security
dilemma. Realists argue that states want to escape security dilemmas but
uncertainty prevents them. Ontological security-seeking suggests that
states may not want to escape dilemmatic conflict. Because even danger-
ous routines provide ontological security, rational security-seekers
could become attached to conflict. Ontological security-seeking sheds
new light on seemingly irrational conflict, and suggests lines of research
into the stability of other outcomes in world politics.
K
EY
W
ORDS
♦ attachment ♦ intractable conflict ♦ ontological security
♦ routines ♦ security dilemma ♦ uncertainty
The security dilemma is the heart of structural realist theory: in anarchy,
actions taken for one’s own security can threaten the security of others,
leading to arms races, conflict and war. The fundamental cause of the
security dilemma is uncertainty. As Randall Schweller (1996: 119–20) points
out, a world of known greedy states generates no ‘dilemma’, but insecurity,
while a world of known security-seekers generates no ‘dilemma’, but security
(also Glaser, 1997: 191ff.). But states’ intentions (or ‘type’) are hard to
know and easily misperceived. Moreover, even accurate perceptions today
give little information about intentions in the future, so states always must
European Journal of International Relations Copyright © 2006
SAGE Publications and ECPR-European Consortium for Political Research, Vol. 12(3): 341–370
[DOI: 10.1177/1354066106067346]
be on guard. Uncertainty thus generates the tragedy of world politics, where
a world of security-seekers can be a world at war. Structural realists disagree
on the extent to which uncertainty can be mitigated and unnecessary conflict
avoided. But all assume states seek security and all recognize a causal link
between uncertainty and conflict that looms in every state interaction.
An important premise of security dilemma theory is that the security states
seek is physical, the protection of their territory and governance structure
from others who can cause material harm. This assumption is plausible and
productive — indeed, the security dilemma seems to capture something
deep about anarchy — but in my view it also is limiting. Building on the
work of Jef Huysmans (1998), Bill McSweeney (1999) and others (e.g.
Manners, 2002), who in turn build from Anthony Giddens (1991), in this
article I argue that physical security is not the only kind of security that states
seek, and show how this changes our thinking about the dilemma states face
and the tragedy of world politics.
Specifically, I propose that states also engage in ontological security-
seeking. Like the state’s need for physical security, the need for ontological
security is extrapolated from the individual level. Ontological security refers
to the need to experience oneself as a whole, continuous person in time —
as being rather than constantly changing — in order to realize a sense of
agency (Giddens, 1991; Laing, 1969: 41–2). Individuals need to feel secure
in who they are, as identities or selves. Some, deep forms of uncertainty
threaten this identity security. The reason is that agency requires a stable
cognitive environment. Where an actor has no idea what to expect, she
cannot systematically relate ends to means, and it becomes unclear how to
pursue her ends. Since ends are constitutive of identity, in turn, deep
uncertainty renders the actor’s identity insecure. Individuals are therefore
motivated to create cognitive and behavioral certainty, which they do by
establishing routines.
Importantly, for theorists of ontological security individual identity is
formed and sustained through relationships. Actors therefore achieve
ontological security especially by routinizing their relations with significant
others. Then, since continued agency requires the cognitive certainty these
routines provide, actors get attached to these social relationships.
Applied to states, ontological security-seeking reveals another, second,
dilemma in international politics: ontological security can conflict with
physical security. Even a harmful or self-defeating relationship can provide
ontological security, which means states can become attached to conflict.
That is, states might actually come to prefer their ongoing, certain conflict
to the unsettling condition of deep uncertainty as to the other’s and one’s
own identity. The attachment dynamics of ontological security-seeking thus
turn the security dilemma’s link between uncertainty and conflict on its
European Journal of International Relations 12(3)
342
head, suggesting that conflict can be caused not by uncertainty but by the
certainty such relationships offer their participants.
1
Specifically, the assumption of ontological security-seeking generates
insights into two important problems. First, it offers a structural explanation
for the apparent irrationality of conflicts among security-seekers that persist
for long periods of time — ‘intractable conflicts’ or ‘enduring rivalries’.
2
Realists tend to look to first and second image causes, drawing on, e.g.,
psychology (Jervis et al., 1985), national myths (Snyder, 1991; Snyder and
Jervis, 1999) and bureaucratic politics (Van Evera, 1999). While many of
these explanations are plausible and my argument does not preclude them,
all focus on factors inside of states, exogenous to interaction. In contrast, I
specify a source of conflict persistence that is located ‘in-between’ states, i.e.
at the third image and endogenous to the logic of competition itself. Where
conflict persists and comes to fulfill identity needs, breaking free can
generate ontological insecurity, which states seek to avoid. No realist
argument fully captures the identity effects of persistent conflict, because
none acknowledges the social construction of state identity.
Second, the assumption of ontological security-seeking helps address the
problem of ending such conflicts. The distinct research implication is that
inter-state routines must be attended to in attempts to end recurrent
conflict. Indeed, practices that appear epiphenomenal — truth commissions,
Cold War history panels, etc. — can be seen as essential components of
change because they help groups achieve reflexive routinization.
It is important to note that ontological security is a basic need, and as such
a constant that cannot explain variation. In this sense it is like the desire
for physical security. The purpose of positing basic needs in social science
is not to explain behavioral variation, but rather to help uncover processes
by which continuity is produced (Baumeister and Leary, 1995). In partic-
ular, ontological security sheds light on the stability of social relationships,
cooperative or conflictual, and the difficulty of effecting change. That
said, while all actors routinize, they vary in what I call their mode of
attachment: some actors rigidly repeat routines, while others participate
more reflexively. This variation has implications for the potential to trans-
form dilemmatic conflicts.
I develop these ideas in three parts. While the desire for physical security
might seem obvious, it may not be clear that ontological security-seeking
even exists, much less has consequences for world politics. With this in mind,
I first explain and defend the assumption of ontological security-seeking at
the individual level. I then scale up this argument, justifying the assumption
that not just individuals but also states seek ontological security. In the third
part of the article I illustrate the assumption’s theoretical fruitfulness by
applying it to the security dilemma. I show that assuming states seek
Mitzen: Ontological Security in World Politics
343
ontological security provides new theoretical leverage on the phenomenon
of persistent conflict among security-seekers. To make the argument
concrete, I refer especially to two cases, the Cold War and Israel–Palestine
after Oslo, both of which can be described as security dilemmas.
3
Impor-
tantly, these empirical references are not meant to test the ontological
security-seeking assumption.
4
This article is an exercise in conceptualization,
and for that it is necessary to show that ontological security-seeking coheres
theoretically, accounts logically for important phenomena, and fits with
existing theory. Testing is obviously equally important, but is a separate
endeavor that necessarily comes afterward. Without adequate conceptualiza-
tion theory cannot be properly operationalized, and poor operationalization
might lead us to reject concepts prematurely (Goldberg, 1963: 30).
Ontological Security
The need for ontological security is not part of the conceptual repertoire we
customarily bring to IR scholarship, and as such some groundwork must be
laid before exploring its implications for international politics. In this section
I conceptualize the individual’s need for ontological security. Several
theoretical and empirical traditions capture aspects of this need, including
psychoanalysis, object relations theory, anxiety/uncertainty management
theory and terror management theory. Using Giddens (1991) as my
reference point (cf. Butler, 1997), I have integrated their insights into a
single logic.
Identity, Action and Uncertainty
Ontological security is security not of the body but of the self, the subjective
sense of who one is, which enables and motivates action and choice.
5
To say
that individuals need security of this self is to say that their understandings of
it must be relatively stable. Needing stability does not mean that self-
understandings must be forever unchanging; indeed such changes are
essential for learning and personal development. The idea is rather that
individuals value their sense of personal continuity because it underwrites
their capacity for agency. A crucial requirement of a stable self-under-
standing is that one’s actions can sustain it over time. The consequences of
action will always either reproduce or contradict identities, and since identity
motivates action its stability over time depends on it being supported in
practice. Another way to say this is that identity is a dynamic process from
which action flows and in turn sustains identity.
Of course, there are many ways to be agentic, including choosing
rationally, matching appropriately, or varying a performative act. The
European Journal of International Relations 12(3)
344
application of ontological security to these differing conceptions is at this
point unclear. However, since my goal is to engage realist IR theory, which
treats states as rational actors, I develop the concept of ontological security
with respect to rational agency (this also is broadly consistent with Giddens’
focus). Rational agents make purposive choices in consequentialist terms,
weighing alternatives and directing action toward a set of internally
consistent ends. In most IR scholarship this capacity is taken-for-granted.
The concept of ontological security allows us to see rational agency instead
as an effect of practices.
Specifically, the claim that ontological security is a basic need begins with
the proposition that actors fear deep uncertainty as an identity threat. Such
uncertainty can make it difficult to act, which frustrates the action–identity
dynamic and makes it difficult to sustain a self-conception. Ontological
insecurity refers to the deep, incapacitating state of not knowing which
dangers to confront and which to ignore, i.e. how to get by in the world.
When there is ontological insecurity, the individual’s energy is consumed
meeting immediate needs. She cannot relate ends systematically to means in
the present, much less plan ahead. In short, she cannot realize a sense of
agency. Ontological security, in contrast, is the condition that obtains when an
individual has confident expectations, even if probabilistic, about the means–
ends relationships that govern her social life. Armed with ontological security,
the individual will know how to act and therefore how to be herself.
Normally, we do not consider uncertainty as posing a problem for action,
much less identity. From a rationalist perspective, faced with uncertainty
actors will assign probabilities and maximize their expected utility. They then
update probabilities in a Bayesian fashion, i.e. by adjusting their initial beliefs
about the relative plausibility of an event in light of new evidence, making
optimal use of all available information available (Morrow, 1994: Ch. 6).
This model has the demanding preconditions that actors must know, at least
probabilistically, the alternative courses of action, the causal relationship
between action and outcomes, and the consequences of possible outcomes
(March, 1999: 14–15; Morrow, 1994). Most rationalist theory assumes that
this knowledge is either relatively easy to come by or that actors can
compensate for its absence rather unproblematically. Decision-makers might
have different understandings of causal relationships or states of the world,
which means they could have different probability distributions over likely
outcomes (for example, reasonable people in the US disagree about the
threat posed by China). But despite subjective differences the theory
assumes that actors always have sufficient knowledge to assign probabilities
and act rationally.
However, this assumes that the actor has confidence in the fundamental
cognitive stability of her environment, and such confidence is not automatic.
Mitzen: Ontological Security in World Politics
345
Everyday life is so full of potential dangers that individuals cannot possibly
process them all. Threats are both physical — your neighbor might attack, a
tornado might strike — and social — you might be fired, your spouse might
leave you. Moreover, novel or infrequent events are simply impossible to
know in advance. Some theorists call these latter cases, situations where
assigning probabilities is impossible, ‘hard’ or ‘fundamental’ uncertainty
(Knight, 1971(1921); Ellsberg, 1961).
6
Hard uncertainty reduces —
objectively — the confidence actors can have in probability assignments.
Pulling these points together, Giddens (1991: 36) argues that all social
actors intrinsically know that behind the routines of daily life, ‘chaos lurks’.
Constant awareness of such chaos would generate tremendous anxiety,
making it extremely difficult to reconcile competing threats and take any
action at all. Even if an actor could imagine every possible contingency, the
attempt to hold all threats at bay would be exhausting. Knowing she cannot
possibly imagine the universe of contingencies only compounds the anxiety,
paralyzing any remaining capacity to act.
In order to be themselves and to act, therefore, individuals need to bring
uncertainty within tolerable limits, to feel confident that their environment
will be predictably reproduced. Importantly, this confidence is independent
of the objective level of uncertainty, which might remain high. It is an
internal, subjective property.
Routines and Basic Trust
Ontological security-seeking is the drive to minimize hard uncertainty by
imposing cognitive order on the environment. Actors do this by developing
a cognitive ‘cocoon’ that ‘bracket(s) on the level of practice [knowledge of
the] possible events which could threaten the bodily or psychological
integrity of the agent’ (Giddens, 1991: 39, 40). Because this cocoon enables
actors to trust that their cognitive world will be reproduced, following
Giddens I call it an actor’s basic trust system.
7
Because actors cannot respond
to all dangers at once, the capacity for agency depends on this system, which
takes most questions off the table. Importantly, this happens outside the
level of conscious choice. On a day-to-day basis identity is not ‘held in
mind’; actors concentrate on the ‘task at hand’ and the need to stabilize
one’s ends is cognitively set aside (Giddens, 1991: 36). That is, self-
integration is maintained at the level of ‘practical’ consciousness while
purposive choice occurs at the level of ‘discursive’ consciousness.
The mechanism generating basic trust is routinization, which regularizes
social life, making it, and the self, knowable. Routines are internally
programmed cognitive and behavioral responses to information or stimuli.
Some are strictly personal,
8
but social relationships are an important source
European Journal of International Relations 12(3)
346
of routinization. Whether personal or social, by definition routinized
responses are unthinking or habitual — options are not weighed; informa-
tion is not updated. Unlike rational action, in short, which implies a
conscious decision to do A but not B, routines are not chosen in any
meaningful sense, but taken-for-granted; reflection is suppressed (March,
1999: Ch. 5). In fact, this suppression is the source of their security-
generating power. By giving actors automatic responses to stimuli, routines
pacify the cognitive environment, bounding the arena of deliberative choice.
Routines thus serve the cognitive function of providing individuals with
ways of knowing the world and how to act, giving them a felt certainty that
enables purposive choice. They also serve the important emotional function
of ‘inoculating’ individuals against the paralytic, deep fear of chaos.
Attachment and Recognition
Because routines sustain identity, actors become attached to them. Individ-
uals like to feel they have agency and become attached to practices that make
them feel agentic. Letting go of routines would amount to sacrificing that
sense of agency, which is hard to do.
Importantly, attachment can develop to dangerous as well as safe routines
— ontological security is perfectly compatible with physical insecurity. Think
of the battered wife who resists efforts by social workers to make her leave
her husband. One possible explanation for her reluctance is that powerful
ontological forces induce her to stay. The identity of ‘wife’ means that at
least she knows who she is and how to be herself through the couple’s
routines. To break from those would cause great anxiety. Because routinized
social relations stabilize our identities, individuals become attached to the
self-conceptions their routines support, regardless of their content.
Equally important, routines depend on others responding predictably,
which of course individuals cannot always control. Powerful external actors
and hostile environments help determine how an individual’s intentional
actions are received. This suggests that the self-conceptions that motivate
intentional action cannot always be reinforced. When this happens, an
individual may well develop a basic trust system that supports a less desired
self. In other words, ontological security is compatible with achieving
‘second best’ identities, or compromising one’s goals. Think of the aspiring
actor who waits tables. He may see himself as an actor, and take classes,
audition and talk constantly about theater. But until he gets the break-
through role, in an important sense he cannot ‘be’ an actor. There is simply
no way for us to know him as such; to society he is a waiter. Moreover, if his
acting attempts are poorly received, over time he may become attached to
Mitzen: Ontological Security in World Politics
347
the waiter identity, because that is the identity his daily routines actually
sustain. He did not intend to be a waiter, but once ontological security
needs are met through the relationships that sustain it, it becomes difficult
to let go. Ten years on that breakthrough role looks less essential, and the
effort of continually trying out for roles hardly seems justified. The old self-
understanding, the aspiring actor, withers away or becomes mere fantasy.
The need for ontological security is so deep, and our attachment to
routines so profound, that we rarely see ontological insecurity in daily life. At
the individual level it only emerges when we cannot help it, when our
cognitive-affective organization of the environment is ruptured, as in cases
of trauma. A rape victim might ask herself how this could happen to me —
which becomes who am I, what kind of person am I that this could happen?
A survivor of 9/11 might ask why he survived and not his partner, which
becomes who am I without her? In such cases, questioning and insecurity
can be associated with an almost catatonic state or, conversely, an acting out,
both of which indicate an inability to ‘go on’ or ‘be’ one’s normal self.
Returning to routines is therefore a crucial step toward recovery. Of course,
trauma is by definition exceptional and most of the time we do not obsess
over potential identity threats. But that is precisely the point — our routines
keep ontological fears out of discursive consciousness. By exposing onto-
logical insecurity, trauma is the exception that proves the rule of how deeply
individuals rely on routines.
In one sense, society solves its members’ ontological security problem for
them, since society is a shared cognitive ordering of the environment
(Giddens, 1991). When trauma happens, the individual’s fall is cushioned by
the social order, which reproduces a general ontological security until she
can pick herself up again. But society is no more than the social practices its
members engage in, which means that its continuation depends on the
constant reproduction of those practices. Individual-level routines thus
constitute society, which in turn stabilizes each individual’s sense of self.
Empirical Support
Empirical research in various areas of social psychology confirms that
uncertainty generates identity insecurity, which is resolved through routines.
The basic insight of anxiety/uncertainty management theory (AUM) for
example, supported by experimental work, is that uncertainty is both a
cognitive and affective problem (Ball-Rokeach, 1973; Gudykunst and
Nishida, 2001). Humans need to ‘make sense of their world’, and when
there is insufficient information or meanings are unsettled, individuals suffer
European Journal of International Relations 12(3)
348
anxiety. The first response to uncertainty is thus not necessarily ‘information
search’, but to reduce psychological stress.
Another line of support comes from Harold Garfinkel’s (1967) ethno-
methodological work on ‘breaches’ of the social order. Garfinkel placed
graduate students in a variety of situations — a subway morning commute,
a family dinner — and instructed them to violate social conventions. He
found that ‘(w)hen “normal” expectations are not met . . . reactions are
anomic and demonstrate confusion’. People first try to normalize discrep-
ancies and preserve predictability, by ignoring or trying to change the
disruption; but if they cannot they experience increasing negative affect, and
social interaction comes to a halt (1967: 236).
Terror management theory (TMT) provides a third convergent set of
findings (Solomon et al., 1991). TMT breaks down the security motive into
two components, fear of physical harm and fear of mortality, where the latter
is understood as the fear of not being, which is as much an identity loss as a
physical one. TMT argues that humans use society to buffer the anxiety
associated with awareness of mortality. Because society is non-corporeal, it
survives personal death; therefore attachment to the symbols and practices
that constitute a social order defends us psychologically against mortality.
TMT experiments have found that when mortality is made salient,
individuals become more attached to group identity, and exhibit increases in
stereotypic thinking and nationalistic bias (Nelson et al., 1997; Schimel et
al., 1999). Importantly, these empirical findings specifically distinguish
between fear of physical harm and fear of death; only the latter taps into
identity needs (Arndt et al., 2001).
Finally, the empirical work particularly of TMT can be seen as supporting
a developmental story, rooted in psychoanalysis and self-psychology, which
stresses the centrality of interpersonal relations for the development of the
self.
9
That story begins by noting that the infant at first does not even know
herself as distinct from the surrounding world, experiencing others instead
as extensions of the self. Over time, however, others show themselves to be
not part of the self, which causes disappointment or trauma and forces the
infant to differentiate from them. A primary way the infant copes with this
differentiation is by routinizing relations with significant others. She
becomes attached to these relations because they enable feelings of
autonomy without trauma.
Taken together, these diverse streams of scholarship support the claim that
individuals fear uncertainty as an identity threat and suppress that fear
through routines to which they become attached. Attachment to routines
and the social order they implicate is thus connected to, indeed a
precondition for, identity and therefore the capacity for rational action.
Mitzen: Ontological Security in World Politics
349
Varieties of Basic Trust
All individuals are attached to their routines, but not all basic trust systems
are equally functional. Broadly speaking there are two modes of attachment,
flexible and rigid, which give rise to differing capacities for agency.
10
With healthy basic trust the individual’s attitude toward routines is
reflexive, in the sense that she can take a critical distance toward them
(Willmott, 1986: 113–14). When uncertainty arises, the individual com-
pensates through various decision rules; when needs go unmet or routines
are disrupted, these lacks are perceived as temporary. Disruption certainly
generates uncertainty, potentially tapping into the domain of ontological
insecurity. But an actor with healthy basic trust can tolerate the uncertainty
of small disruptions because she trusts routines will be re-established, or that
the need eventually will be met through new routines. In other words, the
individual does not treat routines as ends in themselves or consciously direct
her action toward maintaining them. Instead she takes for granted — trusts
— the stability of the environment as she pursues other goals.
This enables her to do two things. First, she can learn. James Rosenau
(1986) points out that learning itself is a habituated response. My argument
is that insofar as this habit (learning) requires a capacity to accept some
uncertainty, it rests on healthy basic trust. If, for example, a friend breaks a
promise, or a rival acts generously, actors with healthy basic trust can
respond to the new information flexibly, even if it means restructuring the
relationship. Healthy basic trust is thus necessary for Bayesian updating.
Second, the actor with healthy basic trust can more readily go beyond
physical security-seeking and pursue ‘higher’ needs such as sociation,
development and self-esteem.
11
This is why Giddens (1991: 38) says that
hope, courage and the capacity for creativity all derive from how an actor’s
ontological security is organized.
In contrast, the individual with rigid or maladaptive basic trust is unable
to maintain distance from her routines. She treats routines as ends in
themselves rather than as a means toward realizing her goals. When an
individual becomes too heavily invested in routines as such rather than the
self-conception they protect, then even temporary disruptions feel highly
threatening and the response is to cling to routines. This is true even if the
routines reproduce physical threat (Willmott, 1986: 118). The problem is
that the actor identifies with routines to such an extent that disruption
threatens the cognitive-affective boundaries of the self; and she knows
extreme uncertainty lurks just beyond those boundaries. Compounding the
problem, the more aware the actor is of her need for the routines, the more
anxiety she feels, and the more compulsively she clings to routines.
When basic trust systems are dysfunctional in this way, action still is
European Journal of International Relations 12(3)
350
possible, but it is internally constrained. This actor cannot respond flexibly
to dissonant information; indeed, dissonant information causes anxiety and
a retreat into routines. Cognitive distance is necessary for both the
deliberation we associate with rational choice and the processing of new
information we associate with learning. Rigid basic trust inhibits these. Rigid
basic trust, in short, can drive decision-makers to act in irrational ways, i.e.
to display what Robert Jervis, Richard Ned Lebow and Janice Stein (1985)
call motivated bias.
Of course, by definition, learning and human development require
disrupting routines, and disruption always generates some anxiety (Johnson,
1990: 111). But importantly, disruption alone does not make one onto-
logically insecure. Only if routines are rigidly held does disruption open up
the person to deep insecurity. With healthy ontological security, actors can
‘encounter all the hazards of life from a centrally firm sense of their own and
other people’s reality and identity’ (Willmott, 1986: 113). In short, basic
trust systems condition our ability to tolerate change. If we want to
understand why some individuals seem invested in dysfunctional relation-
ships while others are able to grow out of them, one place to look is the
individual’s mode of attachment to routines.
States and Ontological Security
Thus far I have argued that human beings need ontological security and
achieve it through cognitive and behavioral routinization. But states are not
human beings and their behavior might be subject to different logics. To
make the concept of ontological security relevant to world politics,
therefore, the argument that states seek ontological security requires
justification. I offer three defenses.
First, IR scholars routinely assume that states seek physical security, which
upon close inspection is no less problematic than ontological security.
Physical security-seeking assumes that states have something like ‘bodies’
that can die. What exactly is the state’s body? Territory? True, like the body,
the state’s territory gives it a spatial boundary; but certain body parts are
essential to human functioning — brain, heart, etc. — whereas it is not
obvious that any particular piece of the state is similarly essential. Another
possibility is that the state’s body is the aggregate of its individual members’
bodies. But then we could ask what percentage of the population must die
in order to conclude that the state is dead —
1
⁄
2
?
3
⁄
5
? What about a case like
East Germany, which died even though not a single member died? East
Germany simply lost sovereignty. Is sovereignty the state’s body, then?
Perhaps, but that is hardly a physical criterion. Of course, my point is not to
challenge the assumption that states seek physical security, but to point out
Mitzen: Ontological Security in World Politics
351
that it raises many questions — as many, in my view, as the assumption that
states seek ontological security. Thus, the real issue here is not physical
versus ontological security, but state personhood more generally. Any
theorist who wants to attribute anthropomorphic qualities to states must
grapple with that (Wendt, 2004). In my view, the ‘state as person’ has
heuristic value insofar as it indexes real aspects of the ways in which states
operate in world politics. As such, assuming ontological and physical
security-seeking alike can be theoretically productive.
The fact that everyone else treats states as people, however, does not in
itself justify my doing so. Thus, a second rationale for assuming that states
seek ontological security points to the ontological security needs of their
members. I have argued that society must be cognitively stable in order to
secure the identities of individuals, and as such individuals will become
attached to these stable group identities. A key part of society is its identity
and distinctiveness vis-a-vis other societies (e.g. Brewer, 1999; Mercer,
1995). One important way that groups maintain distinctiveness is by
routinizing their relationships to other groups. Such inter-societal (in our
case inter-state) routines help maintain identity coherence for each group,
which in turn provides individuals with a measure of ontological security.
From here it is only a short step to argue that the states themselves act at
least ‘as if’ they are ontological security-seekers. Because losing a sense of
state distinctiveness would threaten the ontological security of its members,
states can be seen as motivated to preserve the national group identity and
not simply the national ‘body’. Grounding group needs for ontological
security in individual needs in this way suggests that state institutions are not
just an aggregate of leaders’ decisions, and also that states project self-images
to which their members will be attached in complicated ways.
12
A final reason to assume that states seek ontological security is that this
micro-foundational assumption helps us explain certain macro-level patterns,
organizing anomalies in current theory into an overarching analytical
framework. Consider, for example, Jervis’ argument (1976; Jervis et al.,
1985) that misperceptions and motivated biases frequently impede rational
state action, or Deborah Larson’s (1997) argument that cognitive biases and
errors led US and Soviet leaders repeatedly to misperceive one another’s
actions and miss opportunities for cooperation, even when preferences
overlapped. John Foster Dulles, John F. Kennedy, Richard Nixon and
Ronald Reagan had very different personalities and were separated by
generations, and yet, as Larson points out, they reacted to Soviet actions in
similarly distrustful, irrational ways. Although Jervis and Larson are both
talking about individuals and not states, the fact that such impediments to
rationality persist across decision-makers and give rise to a consistent macro-
level outcome — distrustful state behavior — suggests that something about
European Journal of International Relations 12(3)
352
the nature of states as corporate actors could be reproducing that mistrust.
Assuming that states seek ontological security provides a sociological basis
for understanding why we might see different decision-makers acting
similarly over time.
Similarly, consider the fact that most states follow international law most
of the time, and that they do so unreflexively or routinely, irrespective of
regime type, leader personality, or position in the balance of power.
Assuming that states seek ontological security helps explain this macro-level
pattern, despite micro-level variation.
Critical theorists and methodological individualists alike might still
problematize attributing needs to collectives, even on the ‘as if’ basis I am
proposing. However, in my view, taken together, the above arguments show
that assuming that states seek ontological security is at least as plausible as
other anthropomorphic assumptions made in IR theory. Of course, like any
assumption, positing that states need ontological security is only as useful as
the knowledge it produces. So let me turn to how this assumption might
inform our thinking about world politics.
The Dilemma of Ontological Security
As a fundamental need, ontological security operates in all social contexts,
cooperative or conflictual. Here I explore its implications in just one context:
the security dilemma.
As we have seen, the core problem animating realist theorizing about the
security dilemma is uncertainty. Because states cannot know one another’s
present and especially future intentions with certainty, security-seekers can
be forced to take measures that threaten other states’ security, and thus lead
to war. The crucial variable here is a state’s motive structure or type.
13
Realists assume that the system is populated by states of two potential types,
states with satiable security needs — variously called ‘security-seekers’,
‘status quo’ or ‘conditionally aggressive’ states — and states with insatiable
security needs — ‘greedy states’, ‘revisionists’ or ‘aggressors’. Without the
possibility that some states want more than security, a state’s own quest for
security could not be dilemmatic. These type classifications are not without
problems
14
and actually existing states rarely can be identified as purely
either type. But the parsimonious assumption is made to facilitate theory-
building, and it does seem to capture some real-world dynamics. For
example, take the later Cold War, or the Israeli–Palestinian conflict after
Oslo, where it can be argued that each side mainly wanted reassurance that
the other would not invade or destroy it; both were willing to fight but
neither intrinsically valued aggression. These parties saw themselves as only
‘conditionally aggressive’ (Christensen, 2002), and as such knowing the
Mitzen: Ontological Security in World Politics
353
other’s motives with confidence would have mitigated their security
dilemma. Yet despite constant interaction over many years, they were unable
to learn one another’s type.
Security dilemma theory tells us that in an anarchic context, successfully
communicating intentions is difficult, since efforts at self-protection often
threaten others. But for many realists, the problem of communicating
intentions can in principle be overcome. The process tends to be understood
as one of Bayesian learning, where one state sends a costly signal of its type,
the other interprets and updates, then responds with costly signals of its
own, and so on (Glaser, 1994/5; Kydd, 1997). Such learning is not easy,
quick or inevitable. But among security-seekers who expect continued
interaction, communicating type is possible (cf. Roe, 2000: 389).
However, in many instances, parties who might think of themselves as
security-seekers seem to act as if they want conflict: rejecting overtures,
missing opportunities, exploiting the other, and so on. From a realist
perspective, such persistent competitive behavior is pathological not rational,
and realists tend to locate the pathologies at the first or second images,
inside individual states. I propose instead that one cause of such dysfunc-
tional behaviors is ‘in between’ states, in the inter-state routines that satisfy
each state’s need for ontological security. Ontological security tells us that
rational agency relies on a platform of routines, which suppresses uncertainty
and make the world knowable. Because routines that perpetuate physical
insecurity can provide ontological security, states can become attached to
physically dangerous relationships and be unable, or unwilling, to learn their
way out.
Understanding attachment to dilemmatic conflict requires two steps. The
first is to show that state identities or types are constituted and sustained by
social relationships rather than being intrinsic properties of the states
themselves. Socializing type is important for my argument because if types
did not depend on social relationships, then states could not become
attached to those relationships and ontological security would not give
purchase on the security dilemma. I show that as realists themselves use the
term, type is a role identity, which means that in trying to secure their types
states will secure the relationships that make those roles meaningful. The
second step is an internalization story, where I unpack how interaction over
time transforms state identity and generates attachment. Even realists who
acknowledge that repeated interaction can reinforce conflict among security-
seekers do not fully develop the identity aspects of this process, namely that
states become attached to their conflict because its routines sustain identity.
Once this happens, transforming dilemmatic conflict will be harder than
even realists recognize.
European Journal of International Relations 12(3)
354
Socializing Type
Realists assume that type is self-organized — in game-theoretic terms ‘given
by Nature’ — rather than constituted by relationships. This means that a
state’s type does not depend on other states but is internally generated and
upheld. Type identity, which is subjective, does not require a system of
shared meanings among states to sustain it. In this sense, realists treat type
like other intrinsic aspects of the state such as whether it is mountainous or
has representative institutions. No realists call into question this basic
atomism. For example, Snyder and Jervis (1999) argue that group identity
can include perceptions of the other and that salient negative perceptions
intensify security dilemmas. But here, (mis)perceptions are subjectively held
states, wholly independent of interaction with, much less recognition by, the
other. Assuming that type is self-organized has two implications.
First, type identity has no associated behavioral requirement, which means
nothing about a state’s identity is at stake in interaction.
15
Interaction is
driven by physical security needs, and is not linked to identity. Realists
usually express this as a caution: behavior is not a reliable indicator of
intentions so states must be cautious. But it is also a conceptual point. Many
different behaviors can be consistent with a single type, and there are few
clear separating behaviors between types. In Jervis’ (2001: 37–8) words, ‘the
same hostile actions can flow from the hope for gain or the fear of loss, from
offensive drives or defensive responses’. A security-seeker might take
aggressive actions, while a greedy state might lie low and even appease its
neighbors. This ambiguity is what makes the search for security tragic,
making states prone to behaviors that reinforce the tendency toward conflict
(Jervis, 1976: Ch. 3).
The second implication of atomism is that states know their own types
with certainty, even if they cannot act on them in the way they would like.
For realists type is an aspiration, a cognitive conception of what the state
would like to be if conditions were right, in short, a ‘possible self’
(Cinnirella, 1998; Markus and Nurius, 1986). These aspirations are
important because they set an actor’s goals and make possible certain futures
that otherwise would not be attainable. As internally held aspirations,
possible selves are known to the actor, even if they often must be suppressed
in daily behavior and may never be realized in practice. Thus, for realists, to
be a security-seeker means that if conditions are right, the state would not
invade or otherwise threaten others’ core values. Indeed, if conditions are
right it would be able to reveal its true identity. But conditions might not be
right, and then the state might have to act as if it is aggressive, even though
it really wants nothing more than security.
Mitzen: Ontological Security in World Politics
355
While I agree that type is distinct from behavior, I think realists take this
too far. My intuition is the following. If a state feels compelled by the logic
of the security dilemma continually to defect from cooperation even while
defining itself as a security-seeker, then in what sense is the state actually
‘seeking’ security? If others cannot relate to the state as one that holds
benign intentions, then for all intents and purposes, within the system states
share, the state ‘is’ the type of state others are interacting with.
For example, in the later Cold War the US and USSR both might have
seen themselves as security-seekers. On this view, the US would have liked
to act as a security-seeker, but was not certain whether the Soviet Union
was greedy. Given such uncertainty, the US could not afford to express its
true identity and was forced repeatedly to take actions that could be seen as
aggressive (and indeed were aggressive). On this realist account, the US
conception of itself, not the character of its interactions, defined its type.
And since the Soviet Union could not relate to the US as a security-seeker,
the two states found themselves in arms races, disputes over missiles in
Cuba, and proxy wars in Afghanistan and elsewhere. In Jervis’ words, ‘US
efforts to exploit opportunities that arose were indistinguishable in their
effect from expansionism’ (2001: 53). Similarly, in the Oslo process Israel
and the Palestinians both saw themselves as security-seekers. Both would
have liked to act on that identity, but neither was sure that the other was
satiable. Thus the actions of both were indistinguishable from actions of
greedy states, with Israel exploiting loopholes in the agreement to build
settlements and the Palestinians insufficiently curbing terrorism. In Emanuel
Adler’s (2005) evocative image, their sense of themselves as security-seekers
remained in a ‘bubble’ closed off from interaction. This is not to say the
actors in either example were greedy, but to suggest the limits of a purely
subjective account.
These limits raise two questions. First, how do we know a type when we
see it? On one hand, states (and social scientists) face a revealed preference
problem if they infer type directly from behavior. If a state’s behavior simply
revealed its preferences the theory would be tautological. Good states can do
bad things. But on the other hand severing type from behavior altogether is
equally problematic. If a good state keeps doing bad things, at what point
can we say it just isn’t good anymore?
This widely recognized epistemic difficulty is rooted in an under-
appreciated ontological one: What is a type? In fact, on a closer look the
assumption that type is wholly self-organized is difficult to sustain.
16
While
the basic need to survive might be self-organized, successfully meeting that
need in a given social system requires defining the need as a goal in terms of
the meanings and practices of that system. An island society unaware of
other human life on earth, for example, would not need to define itself as
European Journal of International Relations 12(3)
356
security-seeker relative to an other, since no other beyond itself would exist.
Its survival would be wholly self-organized, reproduced through internal
processes. Contemporary states, on the other hand, can only survive as
members of the states system; as such their survival motive is intrinsically
relational and can only be expressed as a positioning of the self vis-a-vis other
states. Greedy states covet things from others, like territory; status quo states
do not. In other words, the translation from a basic need into a motive
structure, much less to specific preferences that can guide action, is an
intrinsically socializing move (Wendt, 1999: Ch. 4).
This suggests that as realists use the term, type is actually a role identity,
not an intrinsic one.
17
Like intrinsic identities, role identities are subjective;
they are in the actor’s mind and motivate behavior. But unlike intrinsic
identities, role identities get their meaning from role positions in the social
order and therefore are not understandable in terms of qualities individuals
have alone. Roles locate and define the individual with respect to a social
context; they are clusters of practices that constitute actors as objects of
social experience (the ‘Me’). For example, Paul Schroeder (1984) highlights
the role of ‘intermediary states’ in inhibiting great power war in 19th-
century Europe’s balance of power. Role identities, in turn, are internalized
roles, aspects of an actor’s sense of self that reflect the appropriation of roles
and motivate behavior (the ‘I’). Lisbeth Aggestam (2004) explores how role
identities such as leader and civilized power guided foreign policy prefer-
ences of British, French and German leaders in the 1990s. In my view, the
aspect of state identity that drives realist security dilemma logic must be a
role identity because it refers to the state’s internalized attitude toward
others in anarchy, and anarchy is a social environment. Role identities are
formed and sustained relationally; they depend on others to be realized.
Specifically, the fact that type is a role identity has two implications. First,
types are intersubjective at the level of knowledge. States do not have the
final say in whether they are security-seekers; other states play a crucial role.
There are parameters to the practices that can constitute security-seeking,
which are defined at the system level not by the state itself. To be sure, state
behavior can be ambiguous and particular actions are open to interpretive
dispute (e.g. arms racing). But the ambiguity is not infinite — Hitler may
have called Germany a security-seeker but other states (and we) know Nazi
Germany was not. While realists might agree that states themselves do not
have the final say in whether a given situation is a security dilemma, the
realist point is an epistemic one: states might lie about their intentions, so
only detached observers with access to memoirs and classified documents
can make a determination (e.g. Jervis, 2001; Butterfield, 1952). Here,
security-seeking motives are intrinsic properties of states, and the question is
how to access that knowledge. In contrast, my point is ontological. Security-
Mitzen: Ontological Security in World Politics
357
seeking is not a property of the state, like being mountainous or having
representative institutions, but a property of the shared social order. Thus,
the detached observer must access not only memoirs but the contemporary
shared understandings of behaviors other states acknowledge as security-
seeking versus greedy. This point is not recognized in extant realist theory.
The second implication is that type is intersubjective at the level of
practice, meaning that type identities must be in some sense shared to
sustain interaction over time. Whatever a state’s private aspirations, the social
meaning of its type depends on whether other states represent that state in
a similar way. Intrinsic identities can be maintained alone. But no state can
be a security-seeker for very long without others recognizing it as such. In
interaction, others infer from behavior to the role it constitutes and treat the
actor as if she fulfills that role. Thus, for a role to constitute an actor and
motivate behavior over time, it must be expressed in behavior and that
behavior must be recognized by others as fulfilling the role. Note that this
argument is constitutive (Wendt, 1999: 77–88), resting on the notion of
identity as a social relationship. A state cannot ‘be’ or sustain its type without
its strategic partner acting in a certain way (recognizing it). The other’s
recognition is a condition of possibility for the self to retain its type.
Some realists might recognize aspects of this point in Jervis’ argument
(1976: 75ff.) that treating a state as if it is greedy can become a self-fulfilling
prophecy, transforming that state’s preferences over time. Jervis’ argument
does have similarities to the internalization story I tell below. But the self-
fulfilling prophecy argument is strictly causal: state A affects state B’s
preferences over time. My argument that type is a role identity highlights
that interaction does not only affect others, it also constitutes the self. Both
parties’ identities are constituted by the relationship. Jervis’ argument thus
overlooks how both states might depend on their relationship to maintain
who they are.
Returning to the Cold War example, realists might interpret its end as a
process of type revelation, a series of interactions in which the two states
were able to overcome mutual mistrust and progressively reveal their true,
security-seeking nature (Kydd, 2000; Larson, 1997). Indeed, in the mid-
1980s, the Soviet Union did try to communicate a status quo type through
glasnost and costly behavioral signals such as the INF Treaty and withdrawal
from Afghanistan. But the US did not need to recognize these efforts as a
sign of security-seeking, and indeed at first it saw them as tactics or tricks.
Had that interpretation prevailed, Soviet troop withdrawals would not have
become a step in a process of type revelation and the conflict would have
persisted. What the revelation story overlooks is that US recognition of
Soviet actions as a security-seeker is what ultimately constituted the Soviet
European Journal of International Relations 12(3)
358
Union as a security-seeking state. Neither private aspiration nor public
behavior was enough; social recognition was necessary as well.
In sum, the realist variable of type is a role identity; and because role
identities depend on the social structure states share, they require recogni-
tion. If types only exist as private aspirations they will be socially unstable
and hard to sustain. Role identities that remain in a bubble severed from
practice are fantasy identities. The potent insight of security dilemma theory,
that interaction has dynamics of its own that can compromise physical
security, is certainly still true; but these interaction dynamics have unex-
plored identity effects with implications for the search for security.
Attachment and Recognition in Anarchy
Ideally, internally held role identities and externally recognized roles
correspond, but as security dilemmas persist, by definition they will not.
Each state sees itself privately in the role identity of a security-seeker, but
each is recognized publicly in the role of a potential aggressor. From an
ontological security standpoint this mismatch between subjective identity
and recognized role is unstable. My argument is that as interaction persists,
this mismatch resolves itself in favor of the latter: states get invested in
socially recognized identities. That is, recognized roles feed back on role
identities at the level of routines, creating a new role, of competitor or rival.
The competitor is not a security-seeker, i.e. it is not conditionally aggressive
but aggressive all the time. But nor is the competitor greedy or revisionist,
i.e. it does not pursue any specific object vis-a-vis other states. Competitive
states are simply motivated to compete, whatever the object, and always
ready to fight.
To understand how security-seekers become competitors, consider how
security dilemmas unfold, which many realists analyze as a Prisoner’s
Dilemma (see Jervis, 1976; Glaser, 1994/5; cf. Kydd, 1997). The logic
begins with two states, each of which sees itself as a security-seeker but is
uncertain about the other’s type and as such feels compelled to defect. Now,
with every subsequent round of interaction, the states gain knowledge. If
both continue to defect, then of course they do not gain knowledge about
one another’s true type. They do, however, gain knowledge about the
partner’s behavior. With this practical knowledge, each state can more easily
make inferences about its partner’s future actions, if not its intentions.
After many rounds of this, while these states may be no closer to knowing
one another’s true types, on a social level many things have been resolved.
States have acquired a crucial, behavioral kind of certainty. They ‘know’
whom they face in the sense of knowing how the other will respond to their
actions, which means they know how to act. By repeatedly defecting, each
Mitzen: Ontological Security in World Politics
359
state knows it will not get suckered. In short, these states have overcome
type uncertainty by imposing the behavioral certainty of competitive
routines. Even if conflict and war cannot be prevented, they at least can be
foreseen, which means the states have some control over situations and can
plan. These states may fear and mistrust one another, but the fear is not
paralytic. It is focused on the other and linked to particular cognitive and
behavioral responses, thus giving these states a degree of cognitive mastery
over a complex, hostile environment. While it is hard to say the states are
actively seeking security, since neither is acting like a security-seeker would in
a world where types were known, this ersatz security-seeking behavior is
mutually recognized. It continues without disruption, neither calling the
other’s bluff. In other words, both states begin as security-seekers, but
uncertainty causes each to act as if the other is hostile, which, in turn,
prompts the other to respond accordingly.
18
Thus far my account parallels Jervis’ (1976: 64ff.) argument that states
tend to resolve their initial uncertainty with worst-case thinking and then get
trapped into conflicts. Even more consonant is Badredine Arfi’s argument
(1998: 159, 169ff.), which locates the causes of ethnic conflict in the need
social groups have to resolve that initial uncertainty. Like me, Arfi stresses
that social identities depend on routines and that groups fear the identity
effects of disruptions of routines.
But from here, the concept of ontological security isolates something
neither Jervis’ realist nor Arfi’s constructivist argument captures: there is an
appropriative moment, where both states take on the identity that is
embodied in the competitive routines and therefore become attached to the
competition as an end in itself. That is, physical security aspirations cannot
be made salient for interaction because they are not recognized by the
other; but because actors also need ontological security, as competitive
practices are repeatedly recognized and reinforced, the routines supporting
the identity of a competitor likely will feed back on the states’ self-concepts.
At this point, continuing to be a rational actor requires repeating one’s own
competitive routines. Moreover, because these role identities are co-
constituted, the states are profoundly dependent on each other. Each needs
the other’s competitive routines as much as it needs its own. Maintaining
the security dilemma is a joint activity, a social structure sustained by mutual
recognition. Indeed, the entrenched security dilemma is a type of collective
identity.
19
Of course, since the relationship is premised on competition and
physical insecurity, an entrenched security dilemma is a rather dysfunctional
collective identity where each feels as if he is acting alone. The sense of
being part of a greater whole, i.e. the ‘we’ dimension of their relationship,
remains implicit or submerged, making this type of conflict particularly
difficult to overcome.
European Journal of International Relations 12(3)
360
At this point, what began as a means of security-seeking, cautious
competition, has become an end in itself. That is, attachment to the conflict
is rigid not reflexive. For each, maintaining the capacity for agency requires
constantly reproducing the conflict. Because each needs the competitive
routines, neither can make its private aspiration for security salient in
interaction. Recognized roles have crowded out private aspirations; and,
severed from practice, aspirations for security retreat into fantasy. Impor-
tantly, their dependence on routinized competition is at a practical rather
than a discursive level, meaning it is not held in mind by either state during
interaction. Because these states might not be aware of their dependence,
they cannot loosen it. Even if a state wishes it could be a security-seeker, it
has become attached to the identity that is reinforced through competi-
tion.
From here it becomes impossible to acquire reliable information about the
other’s putative identity as a security-seeker. That information might exist,
but it cannot be processed. Recalling the earlier discussion of the possible
self, I am arguing that if the cocoon of routines providing a state with basic
trust does not support its possible self, it will develop a basic trust system
that supports a less desired self, the competitor rather than the security-
seeker. It did not intend to be a competitor, and indeed may maintain
security-seeking aspirations in principle, but once ontological security needs
are met through relationships that sustain competition, those aspirations are
effectively insulated from practice. One generation later potential peace
overtures seem threatening, the risk involved not worth the sacrifice in
stability. The difficulties of choosing peace are as much ontological as
physical, since neither party knows any longer even how to be a security-
seeker.
In short, where ontological needs are met by routinized competition, it is
no longer accurate to say that states face a physical security dilemma. A state
in a true dilemma would prefer cooperation to defection but cannot be sure
it won’t be exploited. States in routinized competition, on the other hand,
are quite sure. On a deep level, they prefer conflict to cooperation, because
only through conflict do they know who they are.
Theoretical and Empirical Implications
The assumption that states seek ontological security sheds light on the
empirical phenomenon of conflict where states have no apparent conflict of
interest. Specifically, it yields a critique of the defensive realist argument that
security-seeking states can learn their way out of conflicts, and suggests
particular, novel strategies that might address ontological security needs.
Mitzen: Ontological Security in World Politics
361
Defensive realist arguments assume that rational states have the capacity to
engage in Bayesian updating in response to new information. But as we have
seen, this type of rational agency rests on healthy basic trust, the ability to
step back and reflect on routines. Such adaptive actors are unlikely to
populate longstanding conflict. These states might very well be rational, but
they consistently reason on the basis of worst-case possibilities, responding
to every action with a simple decision rule: always defect (Brooks, 1997).
The cognitive certainty that anchors rationality is maintained anxiously, by
rigidly reproducing conflictual routines. These states cannot acquire or
process new information about type, and thus cannot act on their aspirations
for security.
In short, this is the offensive realist world as articulated by John
Mearsheimer (2001), or the tension-filled post-structuralist world of David
Campbell (1998), where states ward off existential uncertainty through
boundary-constituting ‘discourses of danger’. This suggests that Bayesian
learning arguments for ending security dilemmas are not applicable to
longstanding conflict. I would hypothesize that situations in which states have
climbed their way out of entrenched conflicts through processes of rational
updating are few and far between, because the conditions of rationality
require leaving previous, identity-stabilizing relationships behind.
Indeed, the failure of Oslo — a peace process explicitly based on learning
through incremental confidence-building measures — is instructive. Israelis
and Palestinians did not have the key prerequisite for learning their way out
of conflict, healthy basic trust; and the peace process did not address this
lack. Michael Barnett (1999) shows that Rabin constructed a discursive
space within Israeli politics that legitimated withdrawal as a means of
security-seeking. But an inter-group public space was not similarly fostered;
no Oslo strategies compelled the actors to reconcile action and identity and
support mutual security-seeking (Adler, 2005). Because the relationship
between aspirational role identities and recognized roles was not nurtured,
security-seeking could not become a salient joint aspiration.
Of course, not all conflicts persist and where they do, attachment may not
be the only reason — decision-makers may perpetuate conflict for instru-
mental reasons as well. But where conflicts persist among security-seekers,
even when decision-makers want to end security competition, that desire will
remain insulated from practice unless it is explicitly linked to new routines of
interaction. This is because even where competitive routines get disrupted,
the habit of conflict is easily restored. Leaving old routines behind generates
ontological insecurity. Each party previously had known itself through
routines of enmity, and it is hard to relate ends and means without those
routines as a baseline. Cognitively and affectively, it is easier to act on old,
European Journal of International Relations 12(3)
362
concrete fears than on new, untested hopes (Bar-Tal, 2001). To break with
routines would be fraught with ontological fear.
Attempts to end intractable conflict should therefore focus on breaking
down the rigid attachment to routines and creating routines of interaction
that permit parties to reveal aspirations and learn from interactions. No
state can be a security-seeker alone. The ability of former adversaries to
rationally seek security is constituted in the space between them, not just
in their individual ‘heads’. In other words, since privacy helps cause
attachment to dilemmatic conflict, strategies to end such conflict should
focus on publicity.
Specifically, ontological security-informed research could examine the role
of international society and transnational public spheres in keeping salient
security-seeking identities (Risse, 2000). For example, research could
investigate the role of routinized public meetings and public commitments
for solidifying mutual recognition of security-seeking. Repeatedly articulat-
ing commitments, and calling on states to justify publicly their actions
against shared aspirations, should foster habits of reflection, which offer the
potential to de-rigidify attachment to competition. Indeed, the regular
meetings of the Concert of Europe powers and the Security Council can be
read this way. Arguably, Europe’s ability to overcome its conflictual past and
develop an identity which does not depend on ‘discourses of danger’ is
linked to its consultation reflex and other practices through which member
states publicly reaffirm and perform their identity as security-seekers to one
another (Mitzen, 2006).
One also could inquire whether recognition must come directly from the
security partner, or whether other states and transnational non-state actors
can grant recognition. In the late 1980s, the US initially rejected the
Soviet identity as security-seeker. Only once others began to recognize the
USSR’s security-seeking identity did the US move toward recognition
(Evangelista, 1999).
Note that these implications of state ontological security-seeking do not
‘rival’ so much as complement or extend existing explanations of why
security dilemmas persist and how to overcome them. Realist work treats
important dimensions of the problems states face when attempting to
overcome entrenched conflict. But this work tends to treat the post-conflict
case narrowly, as a problem of physical security only (e.g. Posen, 1993;
Walter, 1997). In my view, realizing that security-seeking is a social practice
that implicates identity is a crucial step toward getting out of what seem to
be externally imposed logics. Thus far, practitioners more than theorists have
stressed public sphere oriented strategies for preventing and ending conflict,
and no clear theoretical justification has been provided. Ontological security
provides that justification, and a framework for further research.
Mitzen: Ontological Security in World Politics
363
Conclusion
Analytical attention in social science often is drawn to the phenomenon of
change, but world politics also is characterized by powerful homeostatic
tendencies. Indeed, the phenomenon of inertia we call order is itself produced
through social processes; changes are no more than the disruptions that
overcome or alter those processes (Wight, 2001). This means that theories of
change are improved by better understanding order. To do so, I have
proposed a new motivational assumption, that states need ontological
security, and have theorized its relationship to physical security-seeking. The
argument is only a beginning; much work remains to be done.
Some of this work must obviously be empirical, beginning by operational-
izing the modes of routinization in security dilemmas. If states seek
ontological security then they should develop routines with other states and
be attached to them. If there are two forms of basic trust then routines, and
their effects, should vary systematically. Rigid routines should be associated
with an inability to learn; we should not see the states searching for ways out
of the conflict or engaging in debates about the other’s type. Flexible inter-
state routines that permit reflection, on the other hand, should be associated
with learning and transformative change. These states search for ways out of
conflict and jointly attempt to reconcile interaction with security-seeking
goals. Once operationalized, it also will be important to specify conditions
under which each mode might form and which state behaviors tend toward
routinization.
There is also theoretical and empirical work to be done on other
applications of this general proposition. I have focused on the security
dilemma, but the dynamics of ontological security-seeking should be broadly
applicable. For example, the assumption could be applied to cooperative
outcomes such as security communities (Adler and Barnett, 1998), inter-
national society and international institutions. Indeed, while regime theory
assumes that states cooperate to pursue mutual gain, ontological security
suggests a different starting point — fear of chaos. The fact that cooperation
might be rooted in fear-avoidance rather than gain-seeking may lead to
counter-hypotheses regarding the durability and shape of institutions.
Moreover, like conflict, in principle cooperation could be characterized by
either rigid or healthy basic trust. This raises questions — Does cooperation
generate attachment? How does the form of basic trust affect the durability
of cooperation? — that may have implications for institutional design.
Physical security-seeking has been a powerful, unquestioned assumption
in IR theory. Assumptions usefully organize our thinking, but also constrain
it — in this case making us unable to see how states might become attached
to conflict. With this in mind, in this article I have proposed supplementing
European Journal of International Relations 12(3)
364
the physical security assumption with attention to ontological security needs,
and have shown that these needs can help explain how stubborn conflict can
be. Assuming that important aspects of state identity are constituted
relationally provides new theoretical leverage on interaction dynamics and
can lead to practical implications for important problems in world politics.
Of course, treating ontological security needs might not be sufficient to
secure stable peace, but it is necessary. If we ignore these needs and focus
only on physical fears, we overlook the attachment dynamics that can
underwrite cycles of enmity and conflict. International politics need not be
tragic. But overcoming conflict requires hard work, and the first step is to
acknowledge in theory and practice that conflict may benefit a state’s
identity even as it threatens its body.
Notes
For helpful comments on previous drafts, I am grateful to Emanuel Adler, Michael
Barnett, Deborah Boucoyannis, Steve Brooks, Alex Cooley, Cosette Creamer, Alex
Downes, Dan Drezner, Bud Duvall, James Fearon, George Gavrilis, Charles Glaser,
David Hayes, Rick Herrmann, Ted Hopf, Peter Katzenstein, Morten Kelstrup, Zack
Kertcher, Matt Kocher, Ian Manners, Patchen Markell, Amanda Metskas, Michelle
Murray, Dan Nexon, Vincent Poulios, Pradeep Ramamurthy, John Schuessler, Randy
Schweller, Mira Sucharov, Leslie Vinjamuri, Srdjan Vucetic, Lisa Wedeen, Nick
Wheeler, Mike Williams and especially Alex Wendt.
Earlier versions were presented at APSA 2001, the Program on International
Security Policy (PISP) Workshop at the University of Chicago, the University of
Minnesota, the University of Wales, Aberystwyth, Ohio State University and the
CIDEL Workshop, ARENA, Oslo.
1. By drawing attention to an uncertainty deeper than that stressed in extant realist
theory, my proposition that states seek ontological security reveals a conflictual
logic that realists have overlooked. However, my argument might be made
consonant with realism if we adopt Michael Williams’ (2005: 25ff.) reinterpreta-
tion of the depth of the uncertainty problem in Hobbes’ state of nature.
Williams argues that not simply physical insecurity but epistemic indeterminacy
characterizes the state of nature, and actors feel a need to escape this radical
unknown. Even fear of the other, i.e. conflict, can be a remedy for the deeper
fear.
2. Of course, not all persistent conflicts are security dilemmas. But realists draw on
the security dilemma to understand this phenomenon (e.g. Snyder and Jervis,
1999), while enduring rivalries theorists also acknowledge the relevance of the
security dilemma model — e.g. Mor and Maoz (1999) analyze the strategic
evolution of rivalry as a game specified similarly to the security dilemma.
3. Whether either conflict is a security dilemma is debated. See Jervis (2001) on the
Cold War; Posen (1993), Snyder and Jervis (1999) and Roe (2000) on ethnic
conflict.
Mitzen: Ontological Security in World Politics
365
4. See Steele (2005) and Mitzen (2006) for empirical applications of state
ontological security-seeking.
5. Brubaker and Cooper (2000: 17–18) discuss ‘self-understanding’ as a mode of
identity talk that is sufficiently precise to be analytically useful. I use ‘identity’
and ‘self-understanding’ interchangeably.
6. For discussion in an IR context see Wendt (2001).
7. Giddens notes that the term ‘basic trust’ comes from Erik Erikson.
8. On stereotyping as routinization, see Schimel et al. (1999). On routinization of
cognitive-affective responses to others as motivated bias, see Jervis et al.
(1985).
9. This paragraph combines TMT with insights from object relations and self-
psychology, e.g. Bacal and Newman (1990); Willmott (1986).
10. Positing this variation distinguishes my ontological security theorization from
that of, e.g. Kinnvall (2004) and McSweeney (1999). The existence of two
forms of basic trust are supported by experimental work on modes of adaptation
to mortality fears, which finds that attachment style matters and varies
systematically — only individuals with low self-esteem cling unreflexively to
routines (e.g. Mikulincer and Florian, 2000). Research in cognitive psychology
confirms that individuals vary in their need for certainty or cognitive closure
(e.g. Kruglanski, 1989), although that work does not specify that the self/
personality is relationally constituted.
11. The need hierarchy is generally accepted in social theory. See Johnson (1990);
Wendt (1999: Ch. 5).
12. TMT and AUM find experimental support of the linkage between uncertainty,
identity, and routines at the inter-group level (Ball-Rokeach, 1973; Gudykunst
and Nishida, 2001; Solomon et al., 1991). One might reasonably ask why
ontological security implies specifically inter-state dynamics, since the state might
not always be the most salient locus of corporate subjectivity and identification.
Indeed, the dynamics I develop might be found at other levels of inter-group
relations.
13. Although some realists argue that the offense–defense balance is the crucial
cause of security dilemmas, buffering effects of state type. See Jervis (1976: 64);
Van Evera (1999); cf. Mearsheimer (2001).
14. For critical discussions of the type terminology, see Jervis (2001: 39ff), Glaser
(1997) and Buzan (1991: 298–303).
15. For some realists (e.g. Copeland, 2000: 206) states need not even be aware of
one another for a security dilemma to exist. Wendt (1999: Ch. 7) discusses the
stakes of interaction in rationalism versus constructivism.
16. This discussion relies on Wendt (1999: Chs 3–4), although Wendt does not
analyze security dilemmas as social structures.
17. On the distinction between role and role identity see Wendt (1999: Ch. 6) and
Aggestam (2004: 56ff.). For an earlier treatment of roles in IR, see Walker
(1987).
18. For support of the tendency among cooperators to behave aggressively when
European Journal of International Relations 12(3)
366
treated as if aggressors, see Kelley and Stahelski (1970); although they do not
consider identity effects.
19. A persistent security dilemma could also be interpreted as a form of collective
intentionality. See Searle (1995: 23–6).
References
Adler, Emanuel (2005) Communitarian International Relations. New York: Rout-
ledge.
Adler, Emanuel and Michael Barnett (eds) (1998) Security Communities. Cam-
bridge: Cambridge University Press.
Aggestam, Lisbeth (2004) ‘A European Foreign Policy? Role Conceptions and the
Politics of Identity in Britain, France, and Germany’. PhD Thesis, Stockholm
University.
Arfi, Badredine (1998) ‘Ethnic Fear: The Social Construction of Insecurity’, Security
Studies 8(1): 151–203.
Arndt, J., J.J.B. Allen and J. Greenberg (2001) ‘Traces of Terror: Subliminal Death
Primes and Facial Electromyographic Indices of Affect’, Motivation and Emotion
25: 253–77.
Bacal, Howard A. and Kenneth M. Newman (1990) Theories of Object Relations:
Bridges to Self Psychology. New York: Columbia University Press.
Ball-Rokeach, Sandra J. (1973) ‘From Pervasive Ambiguity to Definition of the
Situation’, Sociometry 36: 378–89.
Barnett, Michael (1999) ‘Culture, Strategy and Foreign Policy Change: Israel’s Road
to Oslo’, European Journal of International Relations 5: 5–36.
Bar-Tal, Daniel (2001) ‘Why Does Fear Override Hope in Societies Engulfed by
Intractable Conflict, as it Does in the Israeli Society?’, Political Psychology 22:
601–27.
Baumeister, Roy F. and Mark R. Leary (1995) ‘The Need to Belong: Desire for
Interpersonal Attachments as a Fundamental Human Motivation’, Psychological
Bulletin 117(3): 497–529.
Brewer, Marilynn (1999) ‘The Psychology of Prejudice: In-Group Love or Out-
Group Hate?’, Journal of Social Issues 55(3): 429–44.
Brooks, Stephen G. (1997) ‘Dueling Realisms’, International Organization 51:
445–77.
Brubaker, Rogers and Frederick Cooper (2000) ‘Beyond Identity’, Theory and
Society 29: 1–47.
Butler, Judith (1997) The Psychic Life of Power. Stanford, CA: Stanford University
Press.
Butterfield, Herbert (1952) History and Human Relations. New York: Macmillan.
Buzan, Barry (1991) People States and Fear (2nd edn). Boulder, CO: Lynne
Rienner.
Campbell, David (1998) Writing Security: United States Foreign Policy and the
Politics of Identity (revised edn). Manchester: Manchester University Press.
Christensen, Thomas (2002) ‘The Contemporary Security Dilemma: Deterring a
Taiwan Conflict’, The Washington Quarterly 25(4): 7–21.
Mitzen: Ontological Security in World Politics
367
Cinnirella, Marco (1998) ‘Exploring Temporal Aspects of Social Identity’, European
Journal of Social Psychology 28: 227–48.
Copeland, Dale (2000) ‘The Constructivist Challenge to Structural Realism’,
International Security 25: 187–212.
Ellsberg, Daniel (1961) ‘Risk, Ambiguity, and the Savage Axioms’, Quarterly
Journal of Economics 75: 643–69.
Evangelista, Matthew (1999) Unarmed Forces: The Transnational Movement to End
the Cold War. Ithaca, NY: Cornell University Press.
Garfinkel, Harold (1967) Studies in Ethnomethodology. New York: Polity Press.
Giddens, Anthony (1991) Modernity and Self-Identity. New York: Polity Press.
Glaser, Charles (1994/5) ‘Realists as Optimists: Cooperation as Self-Help’, Inter-
national Security 19: 50–90.
Glaser, Charles (1997) ‘The Security Dilemma Revisited’, World Politics 50:
171–201.
Goldberg, Arthur (1963) ‘Political Science as Science’, in Nelson Polsby, Robert
Dentler and Paul Smith (eds) Politics and Social Life, pp. 26–36. Boston, MA:
Houghton Mifflin.
Gudykunst, William B. and Tsukasa Nishida (2001) ‘Anxiety, Uncertainty, and
Perceived Effectiveness of Communication across Relationships and Cultures’,
International Journal of Intercultural Relations 25: 55–71.
Huysmans, Jef (1998) ‘Security! What do you Mean? From Concept to Thick
Signifier’, European Journal of International Relations 4: 226–55.
Jervis, Robert (1976) Perception and Misperception in International Politics.
Princeton, NJ: Princeton University Press.
Jervis, Robert (2001) ‘Was the Cold War a Security Dilemma?’, Journal of Cold War
Studies 3: 36–60.
Jervis, Robert, Richard Ned Lebow and Janice Gross Stein (1985) Psychology and
Deterrence. Baltimore, MD: Johns Hopkins University Press.
Johnson, Doyle Paul (1990) ‘Security versus Autonomy Motivation in Anthony
Giddens’ Concept of Agency’, Journal for the Theory of Social Behaviour 20:
111–30.
Kelley, Harold and Anthony Stahelski (1970) ‘Social Interaction Basis of Coop-
erators’ and Competitors’ Beliefs about Others’, Journal of Personality and Social
Psychology 16(1): 66–91.
Kinnvall, Catarina (2004) ‘Globalization and Religious Nationalism: Self, Identity,
and the Search for Ontological Security’, Political Psychology 25(5): 741–67.
Knight, Frank (1971/1921) Risk, Uncertainty and Profit. Chicago, IL: University of
Chicago Press.
Kruglanski, A.W. (1989) Lay Epistemics and Human Knowledge: Cognitive and
Motivational Bases. New York: Plenum Press.
Kydd, Andrew (1997) ‘Sheep in Sheep’s Clothing: Why Security-Seekers do not
Fight Each Other’, Security Studies 7: 114–54.
Kydd, Andrew (2000) ‘Trust, Reassurance, and Cooperation’, International Organi-
zation 54: 325–58.
Laing, R.D. (1969) The Divided Self. Harmondsworth: Penguin.
European Journal of International Relations 12(3)
368
Larson, Deborah (1997) Anatomy of Mistrust. Ithaca, NY: Cornell University
Press.
McSweeney, Bill (1999) Security, Identity and Interests: A Sociology of International
Relations. Cambridge: Cambridge University Press.
Manners, Ian (2002) ‘European [Security] Union: From Existential Threat to
Ontological Security’, IIS Working Papers, 05/2002, Copenhagen Peace Research
Institute.
March, James G. (1999) The Pursuit of Organizational Intelligence. Malden, MA:
Blackwell.
Markus, Hazel and Paula Nurius (1986) ‘Possible Selves’, American Psychologist 41:
954–69.
Mearsheimer, John J. (2001) The Tragedy of Great Power Politics. New York: W.W.
Norton.
Mercer, Jonathan (1995) ‘Anarchy and Identity’, International Organization 49:
229–52.
Mikulincer, M. and V. Florian (2000) ‘Exploring Individual Differences in Reactions
to Mortality Salience’, Journal of Personality and Social Psychology 79: 260–73.
Mitzen, Jennifer (2006) ‘Anchoring Europe’s Civilizing Identity: Habits, Capabil-
ities, and Ontological Security’, Journal of European Public Policy 13(2):
270–85.
Mor, Ben and Zeev Maoz (1999) ‘Learning and the Evolution of Enduring
International Rivalries’, Conflict Management and Peace Science 17(1): 1–48.
Morrow, James (1994) Game Theory for Political Scientists. Princeton, NJ: Princeton
University Press.
Nelson, L.J., D.L. Moore, J. Olivetti and T. Scott (1997) ‘General and Personal
Mortality Salience and Nationalistic Bias’, Personality and Social Psychology Bulletin
23: 884–92.
Posen, Barry (1993) ‘The Security Dilemma and Ethnic Conflict’, in Michael E.
Brown (ed.) Ethnic Conflict and International Security, pp. 103–24. Princeton,
NJ: Princeton University Press.
Risse, Thomas (2000) ‘“Let’s Argue!”: Communicative Action in World Politics’,
International Organization 54: 1–39.
Roe, Paul (2000) ‘Former Yugoslavia: The Security Dilemma that Never Was?’,
European Journal of International Relations 6(3): 373–93.
Rosenau, James N. (1986) ‘Before Cooperation: Hegemons, Regimes and Habit-
Drive Actors in World Politics’, International Organization 40: 849–94.
Schimel, J., L. Simon, J. Greenberg, T. Pyszczynski, S. Solomon, J. Waxmonsky and
J. Arndt (1999) ‘Stereotypes and Terror Management’, Journal of Personality and
Social Psychology 77: 905–26.
Schroeder, Paul (1984) ‘The Lost Intermediaries: The Impact of 1870 on the
European System’, International History Review VI(1): 1–17.
Schweller, Randall L. (1996) ‘NeoRealism’s Status-Quo Bias: What Security
Dilemma?’, Security Studies 5: 90–121.
Searle, John (1995) The Construction of Social Reality. New York: The Free Press.
Mitzen: Ontological Security in World Politics
369
Snyder, Jack (1991) Myths of Empire: Domestic Politics and International Ambition.
Ithaca, NY: Cornell University Press.
Snyder, Jack and Robert Jervis (1999) ‘Civil War and the Security Dilemma’, in
Barbara Walter and Jack Snyder (eds) Civil Wars, Insecurity, and Intervention, pp.
15–37. Ithaca, NY: Columbia University Press.
Solomon, Sheldon, Jeff Greenberg and Tom Pyszczynski (1991) ‘A Terror Manage-
ment Theory of Social Behavior’, in Mark Zanna (ed.) Advances in Experimental
Social Psychology, Vol. 24, pp. 93–159. New York: Academic Press, Harcourt Brace
Jovanovich.
Steele, Brent (2005) ‘Ontological Security and the Power of Self-Identity: British
Neutrality and the American Civil War’, Review of International Studies 31(3):
519–40.
Van Evera, Stephen (1999) Causes of War: Power and the Roots of Conflict. Ithaca,
NY: Cornell University Press.
Walker, Stephen (ed.) (1987) Role Theory and Foreign Policy Analysis. Durham, NC:
Duke University Press.
Walter, Barbara F. (1997) ‘The Critical Barrier to Civil War Settlement’, Inter-
national Organization 51: 335–64.
Wendt, Alexander (1999) Social Theory of International Politics. Cambridge:
Cambridge University Press.
Wendt, Alexander (2001) ‘Driving with the Rearview Mirror: On the Rational
Science of Institutional Design’, International Organization 55: 1019–49.
Wendt, Alexander (2004) ‘The State as Person in International Theory’, Review of
International Studies 29: 289–316.
Wight, Colin (2001) ‘The Continuity of Change, or a Change in Continuity?’,
International Studies Review 3(1): 81–9.
Williams, Michael (2005) The Realist Tradition and the Limits of International
Relations. Cambridge: Cambridge University Press.
Willmott, Hugh C. (1986) ‘Unconscious Sources of Motivation in the Theory of the
Subject’, Journal for the Theory of Social Behaviour 16: 105–21.
European Journal of International Relations 12(3)
370
