Societal security and social psychology

The concept of societal security as developed by the Copenhagen school has three underlying weaknesses: a tendency to reify societies as independent social agents, a use of too vague a definition of ‘identity’, and a failure to demonstrate sufficiently that social security matters to individuals. This article shows that applying social identity theory to the societal security concept helps remedy these weaknesses and closes the theoretical gaps that the Copenhagen school has left open. It enables us to treat ‘society’ as an independent variable without reifying it as an independent agent. It also suggests a much sharper definition of identity, and a rationale for the Copenhagen school's claim that individuals have a psychological need to achieve societal security by protecting their group boundaries. Social identity theory thus supports the societal security concept in its central assumptions while giving it stronger theoretical foundations and greater analytical clout.


Introduction
The concept of societal security first came to prominence in Barry Buzan's classic People, States and Fear. 1 From the early 1990s, a series of writings by Buzan, Ole Waever and a number of collaborators revived and significantly reinterpreted it, turning societal security into a conceptual cornerstone of what some have labelled the 'Copenhagen school' in International Relations (IR). 2 By the end of the decade, the Copenhagen school had developed additional concepts and societal security had lost some of its erstwhile centrality. Nonetheless, it has remained the most innovative element in the Copenhagen school and the one that sets it apart most clearly from other approaches in International Relations.
It thus comes as no surprise that many critics of the Copenhagen school make societal security their main target. They accuse it of being too fuzzy and 'soft' a concept to be either theoretically convincing or analytically useful. And they argue that its concern with issues such as identity, culture, ethnicity and nationalism signifies more a desire by members of the Copenhagen school to jump on trendy academic bandwagons than the basis for a credible challenge to classical security studies. 3 This article focuses on the concept of societal security and on the criticisms it has attracted. It contends that many of these criticisms thrive on theoretical ambiguities and gaps which overshadow the societal security concept and which the Copenhagen school has not adequately addressed. But it shows that applying social psychological theory to the societal security logic helps resolve these ambiguities and close the gaps. As a result, the concept of societal security acquires stronger theoretical foundations and lives up better to its analytical potential.
Social psychology stands far apart from most other areas of the humanities and social sciences in terms of methodology, terminology and level of analysis. This helps explain why very few students of IR have systematically engaged with it, even those who are otherwise receptive to 'human nature' explanations. The Copenhagen school is no exception. Yet as this article shows, many assumptions that underlie the societal security concept have a strong bearing on social psychology, and the two ultimately make very similar claims. Social psychology, however, corroborates these claims in a much more coherent and rigorous fashion than does the Copenhagen school. Incorporating the insights of the former into the latter thus makes the notion of societal security more precise. Beyond this, it shows how IR can learn from a body of knowledge which it has too often overlooked.

Societal security
A good way to illustrate the concept of societal security is by contrasting it with more traditional approaches to security in IR. The latter are primarily concerned with the security of the state. Since the state is a legal and political construct built on the concept of sovereignty over a defined territory and population, it achieves security by ensuring its continued sovereignty and territorial integrity. And since traditionally the main threat to these has come from military aggression by other states, conventional approaches to security are preoccupied with broadly defined military issues, such as deterrence, balances of power and alliance formation.
This conventional security agenda leaves little room for a systematic treatment of society, defined as a social, cultural and psychological formation distinct from the political and legal construct that is the state. Instead, implicitly drawing on an idealtypical notion of the nation-state as a unit in which the political, social and cultural converge, it conceives of society as the social and cultural 'substance' that the hard political and military shell of the state encloses: each society has 'its' state and each state 'its' society. Societal security -defined as the cultural, linguistic and identitive survival of a particular social group -then becomes the logical extension of state security.
In Identity, Migration and the New Security Agenda in Europe and a series of subsequent writings, Buzan, Waever and their collaborators refuse to subsume societal security under the heading of state security. Instead, they treat society both as an object of security in its own right and as a potential security actor.
What, for the Copenhagen school, is 'society' and what is 'societal security'? Society is defined as the social unit that provides the primary locus of identification for its members. It has both an 'objective' and a 'subjective' dimension: objectively, a society is signified and differentiated from other societies by markers such as language and customs; subjectively, it is the repository of shared meanings and identifications for its members who possess what Karl Deutsch has termed a shared 'we-feeling '. 4 According to the Copenhagen school, the most important identity communities in modern times are ethnic groups and nations. But at different historical stages they have included various other kinds of social formations, ranging from religious groups to clans and tribes. What characterises every identity community is that its members afford it a claim to survival which is ultimately self-referential. Since it is bound up with their identity, they value the community's preservation as an end in itself rather than just as a means to achieve other ends, and hence give it priority over other potential objectives. For the Copenhagen school, it is this quality that turns identity communities into security objects.
Societal security, then, pertains to the perceived ability of an identity community to survive. It refers to 'the sustainability, within acceptable conditions for evolution, of traditional patterns of language, culture, association, and religious and national identity and custom' in a given society. 5 This implies that societal security, too, has an 'objective' and a 'subjective' dimension. In the former sense, it is linked to the preservation of group markers such as language and customs; subjectively, it entails the community's survival as a locus of identification for its members.
Though they hesitate to do so explicitly, most Copenhagen scholars would probably accept that some security threats are not mere discursive constructs but instead have an underlying 'realness' to them -the Nazi threat to Jews, for example, or Iraq's threat to Kuwait. All the same, for the Copenhagen school societal security and perceived threats to it typically have a strong discursive dimension. The crucial variable is securisation. Securisation represents a speech act with a 'specific rhetorical structure'. 6 To securitise is to identify an alleged threat to the survival of the community and to the shared identity it sustains, its presumed origins and perpetrators, as well as a strategy to ward off that threat and thereby render society secure again. Given that these are perceived to be existential threats to something whose survival is sought as an end in itself and is afforded absolute priority, effective securisation often leads to defensive measures that go beyond the limits of what qualifies as politically or morally acceptable conduct in normal circumstances. Perceived threats to societal security generate a sense of Societal security and social psychology 251 emergency and a corresponding willingness to take extraordinary emergency measures. 7 For the Copenhagen school, a society that seeks to fend off a perceived identity threat becomes a security actor. What kind of defensive measures it takes depends on the nature of the presumed threat: they can range from restrictions on immigrants or foreign films to the launching of a pre-emptive war to defend the society's 'way of life' against foreigners. But as was argued, in this rendering societies will take such defensive measures only if the 'objective' developments they are intended to ward off have been securitised to the point where they are perceived to pose an identity threat. For only then is societal security seen to be in danger.
How does societal security relate to state security? For as long as state and society -or nation -are fused, the two typically converge. Threats to state boundaries are at once threats to societal boundaries and thus constitute potential identity threats. The two also converge in the sense that societies facing perceived identity threats will call on 'their' states to act on their behalf -by initiating immigration controls, cultural content quotas, pre-emptive wars or whatever.
Yet what interests the Copenhagen school above all are cases where state and society are not congruent. They focus on two particular situations.
First, many states are internally divided into several societal units -be they selfconceived nations, ethnic communities or cultural or linguistic sub-groups. In these cases, societal security and state security may well be negatively correlated, as one may seek to enhance its own security at the other's expense. State governments often seek to ensure the allegiance of minority groups by trying to culturally assimilate them into the dominant culture. Conversely, many minority groups have sought to secure their cultural and identitive survival by demanding a state of their own. According to the Copenhagen school, the longing by insecure societies to obtain their own states contributed to, among other things, the disintegration of Yugoslavia and the Soviet Union in the 1990s.
The second example is the European Union (EU). There, according to Copenhagen scholars, we witness a gradual transfer of some political functions traditionally exercised by the state to the supranational level. But so far this has not triggered the parallel formation of a 'European nation' to match. As a result, Europe's primary societal units -its nations -persevere, but are increasingly unable to rely on their respective states for security. Seen against this backdrop, the EU faces a range of different possible scenarios.
First, European integration may progress relatively smoothly and without threatening the societal security of its constituent populations. This is most likely to happen if the EU convinces its member populations that it will not culturally assimilate them and if it leaves some essential 'protective functions' at the state level, for instance in cultural policy and education. In the second scenario, European integration does elicit a security response from its member populations, but this does not translate into hostility against the EU. Instead, it leads to a growing 'culturis-

252
Tobias Theiler ation' of the different member nations: they seek to strengthen, reaffirm and 'rediscover' their social and cultural distinctiveness to ensure their survival in a Europe of weakening political and economic boundaries. In the third scenario, fears for societal security lead to backlashes against the EU itself, possibly even to a reversal of European integration. For societal security theorists, the widespread popular opposition to the Maastricht Treaty in the early 1990s was an example of such a development: once the EU's member populations started to fear that further integration would threaten their national identities, they sought to 'call back' their states for protection. 8 According to Copenhagen scholars we cannot yet predict which of these scenarios will ultimately prevail in the EU. What is more, all three could become redundant if the EU manages to evolve into a security object in its own right, either complementing or even usurping the security dignity currently enjoyed by its constituent units. This is theoretically possible but far from inevitable. But whatever the end-result, for the Copenhagen school it is the EU's impact on the societal security of its member populations that will help determine its ultimate fate, not just the interests of state élites and EU bureaucrats, economic conditions, 'functional pressures' and related factors on which most students of European integration have traditionally focused.

Reification
The most serious charge against the societal security concept is that of reification. Bill McSweeney, for instance, claims that for the Copenhagen school both ' ''society'' and ''identity'' are [. . .] projected as objective realities, out there to be discovered and analyzed' 9 while it fails to approach communities 'from a deconstructionist sociological angle, which focuses on the processes and practices by which people and groups construct their self-image'. 10 According to this line of criticism, societal security theorists reify communities by treating them as timeless entities in search of security and protection rather than as culturally and historically contingent products of the social world. By extension, they treat them as fixed rather than as being subject to perpetual redefinition and reimagination.
Yet before the societal security concept can be tried on the reification charge, the concept of reification itself needs some clarification. In particular, we need to distinguish between two types of reification.
The first type equates to what in the research on nationalism and ethnicity is known as primordialism or essentialism. This is based on the notion that nations, ethnic groups and the like represent ahistorical entities whose origins are not social -Societal security and social psychology 253 rather like rivers, mountains and animal species. Such a conception thus naturalises human communities; it obfuscates their socially constructed and historically contingent character. Buzan, Waever and their collaborators, however, explicitly reject primordialism and throughout their writings emphasise the socially constructed origins of nations and other identity communities. What they do imply is that once a community has become socially constructed it can become so stable and psychologically entrenched that it acquires a widely accepted self-referential claim to survival and thereby becomes a potential security object.
With this in mind, some critics might accuse the Copenhagen school of treating societies as de facto fixed variables rather than as entities that are constantly reconstructed, reimagined, renegotiated and so on, and of failing to endogenise these social (re)construction processes into its account of societal security dynamics. Yet this line of criticism is not compelling. As was argued, Buzan Waever et al. readily admit to treating social construction processes as exogenous to the societal security logic and societies as potentially stable, but they manage to mount a persuasive defence. In particular, they correctly argue that social constructivism does not imply that the social world is always fickle. Once constructed, many social practices, beliefs and institutions become deeply sedimented and thereby congeal and change only very slowly. 11 In the very long term, of course, the assumption of stability is unlikely to hold, but then the insight that in the very long run all aspects of social reality always change beyond recognition is often no more useful to the social analyst than the notion that in the long term we are all dead. The central point to bear in mind is that the Copenhagen school uses societal security as a conceptual tool to try to account for specific events that have occurred at specific points in time in specific contexts -events which range from Denmark's initial rejection of the Maastricht Treaty to the war in Bosnia and the rise of anti-immigrant movements in parts of western Europe. In line with this, it stands on solid methodological ground when it treats the social constructs concerned as de facto stable and fixed over the limited periods relevant to the analysis of these events, as long as there is no evidence to the contrary. This point is further illustrated, and to some extent qualified, below.
If the charge of reification in the sense of primordialism or unduly ignoring the fluidity of social reality thus does not seriously undermine the societal security concept, allegations that it suffers from a second type of reification are much more difficult to refute. This type of reification is compatible with the claim that human groups and institutions -be they states, nations, societies, armies, political science departments or whatever -are socially constructed rather than somehow 'essential' or primordial. But it is reifying because it treats these groups and institutions as independent actors with preferences and the ability to shape social outcomes. As Rogers Brubaker has argued, in contemporary nationalism research this type of reification often leads to a 'substantialist' approach to nations and ethnic groups. It treats them as 'categories of analysis' rather than as mere 'categories of practice' and as social agents with needs and aspirations and the capacity to act in a purposive manner. 12 Buzan, Waever and other members of the Copenhagen school, too, subscribe to such a conception. This leads them to describe society as a 'social agent which has an independent reality and which is different and more than the sum of its parts'. 13 And it allows them to talk about societies as actors that are able to form preferences regarding this or that issue, of communities looking to defend their culture, and of nations striving for security.
At the same time, Buzan, Waever and their colleagues admit that their treatment of societies as 'units of analysis' and independent social agents goes 'against sociological orthodoxy' 14 and they anticipate the charges of reification it is bound to attract. They insist, however, that this view is 'essential' 15 to their model. In places, they suggest that this is so because the very concept of societal security presupposes the existence of societies as real entities. At other times, they associate their 'societies as social agents' approach with a stated preference for methodological collectivism over methodological individualism, and with the alleged necessity of treating 'society as more than the sum of its parts' 16 so as to avoid sliding into an 'atomistic, aggregate view of security'. 17 Yet, paradoxically, in some places Copenhagen scholars themselves seek to break with such reifying notions of society. Take the concept of 'securisation'. It is in essence an attempt to deconstruct the intersubjective processes that ultimately lead to 'society' perceiving and doing certain things in certain ways. At times, Buzan and Waever travel even further down the de-reification route, suggesting that society should not be treated as a 'thing' at all but rather as a mental representation and discursive referent object. This is the case, for instance, when they insist that ' ''[s]ociety'' never speaks, it is only there to be spoken for'. 18 When arguing in this mode, they come close to a conception of groups that predominates in social psychology, as is discussed further below.
In sum, the Copenhagen school fluctuates between reifying conceptions and dereifying ambitions, between treating societies as 'categories of analysis' and independent social agents on the one hand, and treating them as products of the human imagination and mere discursive referent objects on the other. 19  the social sciences experience this tension, and it is very difficult to avoid. Using language to think and talk about social groups, institutions and practices inevitably entails a need to 'pin them down' and thereby a measure of objectification and reification. 20 The many 'scare quotes' in the contemporary social science literature epitomise the resulting quest to signal distance and reflexivity.
All this is bound up with many important metatheoretical questions and it is hard to fault Copenhagen scholars for not addressing them at greater length. All the same, for the societal security logic to be potentially plausible it must conceptualise society as something other than a social agent. The most important reason is this: it is only by shedding the idea that societies strive for societal security and defend their identity that we can focus fully on individuals as security-seeking and identifying agents. And as is shown below, this in turn is the prerequisite to successfully tackling the question of motivation and therefore of what drives the societal security dynamic in the first place. But before this, the next section turns to the second problem that plagues the societal security concept, which is its failure to espouse a sharp enough notion of identity.

Identity
As was noted, it is identity rather than society or security that ultimately constitutes the key variable in the societal security literature: people want societal security because society helps sustain their identity. This preoccupation with identity must lead one to ask how well the Copenhagen school defines identity, and whether this definition has solid theoretical grounding.
When the Copenhagen school came into being in the early 1990s, 'identity' had already experienced a surge in popularity throughout the social sciences and humanities. But some had also started to voice doubts regarding its usefulness. 21 They assert that throughout the social science literature 'identity' is overused and poorly defined and that this contributes to reifying and 'substantialist' approaches to social formations (for example, 'the Danish identity' or, worse still, 'Denmark's identity'). Moreover, critics accuse much of this literature of depicting identity as a 'thing' that people 'possess' rather than as a concept that signifies a process of permanent location and relocation of the self in relation to the identity object. In this way, identity, too, is often reified.
The Copenhagen school's use of identity is vulnerable to many of these criticisms. To be sure, in places Buzan and Waever claim to treat identity as an 'intersubjectively constituted social factor', 22 and Waever in particular sometimes alludes to a Derridian notion of identity as epitomising a kind of infinite longing and search for closure. 23 But ultimately, he and his colleagues revert to a more prosaic concept of identity, as something rather fixed that people possess and want to defend. What is more, combined with their conception of society as a social agent discussed earlier, they often slide into a double reification scenario in which societies possess identities which they seek to protect. 24 Here too, the danger of reification owes much to the dilemma that the very enunciation of a concept such as 'identity' (much like 'society' and 'nation') to some extent fixes, congeals and 'thingifies' its referent. But Copenhagen school theorists compound this problem. For, like much of the identity-centred literature in comparative politics and IR, they rely on an unelaborated notion of identity that is oblivious to the large body of psychological and social psychological scholarship in which the identity concept originated. As is shown below, it is precisely by engaging with this literature that one can obtain a sharper and more reflective definition of identity that treats it as something other than a possession. The same engagement also leads to the at first puzzling conclusion that people derive their communal identities more from the community's boundaries than from its cultural 'content'.

Motivation
The third problem with the societal security concept as proposed by the Copenhagen school is its failure to account for what drives the alleged societal security logic. As was noted, in places it makes the reifying claim that societies seek to protect their societal security. Elsewhere it implies that it is ultimately individuals who do so, but then fails adequately to explain what motivates them. The question that Copenhagen scholars largely fail to address is thus the following: why do people strive for societal security and why do they want to preserve their group identity?
This question is crucial. After all, from anti-Maastricht and anti-immigrant movements in western Europe to the various east European state breakups, even events that are the empirical jewels in the crown of the societal security theorists have attracted competing explanations. Many of them focus on economic rather then identity-centred motivations, and consider elite calculations rather than mass sentiments to be the deciding factor. 25 These alternative accounts may well suffer from a myriad of shortcomings in their own right. But without theoretical evidence for the assertion that in some circumstances individuals value societal security more than other objectives, they enjoy no less prima facie plausibility than do societal security-centred explanations. In short, proponents of the societal security concept must justify their claim that societal security matters to individuals.

Societal security and social psychology 257
Such justifications must be of a social psychological nature in the broadest sense of the term. That is, they must centre on the relationship between individuals and society, and on the dialectic between psychological processes on the one hand and social processes on the other -in much the same way that the societal security logic itself implies some kind of interchange between these two levels. Moreover, given its core subject manner, social psychology has the potential to advance the other two issues raised in this section: the need for a non-reified concept of society and for a sharper definition of identity.
So far, the Copenhagen school has failed to engage with social psychology, as have almost all other approaches in IR. It is against this backdrop that the next section peers over the disciplinary fence in order to glean from contemporary social psychology what is most relevant to the societal security concept. It concentrates on one approach that over the past twenty years has become dominant in the social psychological study of group behaviour, called social identity theory. 26 After outlining its main features, the essay goes on to apply social identity theory to the societal security concept and shows how it helps resolve the problems just discussed.

Social identity theory
Social psychology is a broad field. This is true for both the subject matter it espouses and for the range of approaches that have flourished within it. Yet as for the particular sub-field of social psychology that is concerned with group behaviour, the past twenty years have seen a growing concentration on one approach called social identity theory (SIT). Social identity theory originated with the pioneering work of Henry Tajfel, John Turner and their collaborators in the 1970s. Since then, its core assumptions have achieved a widespread following in the mainstream literature on group psychology. This section briefly summarises these, starting with a legendary psychological experiment that led to their formulation. 27

The minimal group experiment
The minimal group experiment involves a number of volunteers who have not previously met. 28 The experimenter starts out by asking them to estimate the number of dots shown on a slide and to write down their estimate. Immediately afterwards the subjects are led into individual cubicles. There they learn that they have been assigned into one of two groups (the 'A group' or the 'B group') depending on whether they under-or overestimated the number of dots on the slide. In reality, the separation is completely random. Having been assigned to a group and without meeting their fellow group members, the subjects receive the task of distributing money in the form of points between members of their own group and members of the other group (but never to themselves). To do so they must select from a range of pre-defined distribution matrices. Some matrices allocate the money equally between members of both groups while others allocate much more to one group than to the other. In these circumstances the commonsense expectation is that the subjects will be indifferent as to how they distribute the money. After all, their group categorisation is based on a single trivial criterion and completely random and ad hoc, and they are equally unfamiliar with in-and outgroup members. However, their actual behaviour turns out to be very different. First, they display consistent ingroup favouritism: they give more money to members of their own group than to those in the other group. Second, they are prepared to see their own group worse off in absolute terms if this improves its relative position vis-à-vis the other group: beating the outgroup by a wide margin is more important than maximising ingroup profit. In short, the arbitrary, ad hoc and minimal nature of the groups in the experiment does not prevent the participants from internalising their group categorisation and from discriminating against the outgroup in favour of the ingroup. Even though the groups in the experiment are objectively minimal they become subjectively meaningful.
A cascade of subsequent experiments refined the basic minimal group setting and produced similar findings. For instance, they show that group discrimination occurs even when the experimenter makes it clear to the subjects that the group division is completely arbitrary ('ultraminimal'), for example, by throwing a coin in order to determine their group placement. 29 Moreover, similar results obtained in a series of cross-cultural experiments that involved participants from a wide range of social, cultural, geographic and religious backgrounds. This, in turn, suggests that they reflect innate human dispositions rather than cultural conditioning. 30 How should one explain such findings and the underlying propensity for 'groupness' they suggest? Social identity theory focuses on two main factors: our tendency to divide the world into categories and our quest to maximise self-esteem. These are now discussed in turn.

Categorisation
Humans inhabit a world in which there are few clear lines of division. Social and physical reality is fluid and continuous rather than neatly divided into different Societal security and social psychology 259 categories. But this threatens to overwhelm our cognitive apparatus with a mass of seemingly chaotic and unstructured stimuli. To counteract this, we seek to order, systematise and simplify our field of perception by dividing the social and physical world into categories. This makes it more manageable and renders our 'experience of the world subjectively meaningful'. 31 For instance, when we look at a rainbow we detect relatively discrete bands of colour, even though its colours are in reality continuous. In the same way, we perceive the world to contain races, generations, personality types, historical epochs and so forth, all of which reflect attempts to impose structure upon confusion and order upon chaos. Categorisation, then, 'is a fundamental and universal process precisely because it satisfies a basic human need for cognitive parsimony. 32 Categorisation always entails accentuation. We perceive items placed into the same category as more similar to one another than they are in reality but more different from those in other categories than is objectively the case. As a result, the cognitive divisions we impose on the world do not seem arbitrary to us but instead a reflection of seemingly objective breaks and discontinuities.
Social identity theory builds on these premises. It asserts that human beings, too, are subject to being categorised and that this leads to the formation of groups. This has two main corollaries.
First, just as we internalise other divisions of the world we internalise our group categorisations, too. Through this process the group 'installs itself in the mind of the individual' 33 and becomes a part of his or her self-concept. As the individual becomes part of the group, so the group becomes part of the individual; the self and the group partially merge.
Second, in line with the principle of accentuation discussed earlier, we tend to see those who are placed in the same category -including ourselves -as more similar to each other than is actually the case while we overestimate the differences that separate us from members of other groups. In other words, we conceive of others and ourselves partially, in a 'depersonalised' manner, as 'undifferentiated items in a unified social category'. 34 For social identity theory, such 'depersonalisation' is at the heart of phenomena such as group stereotyping and prejudice which are discussed below. But first, I turn to the second major factor which, according to social identity theory, conditions group behaviour.

Social comparison
In common with almost all other psychological theories, social identity theory asserts that individuals have a need to maintain a positive self-image, as this is linked to their sense of self-worth and their self-esteem. One's self-image, in turn, is always formed in part through social comparison. For instance, to maintain my perception of myself as a virile and brave individual I must implicitly or explicitly compare myself to people who I believe do not qualify for these characteristics. All selfperceptions depend in part on how we perceive others.
Social identity theory draws on the concept of social comparison and argues that it, too, has a group dimension. This is implicit in the logic of group categorisation and internalisation discussed above. By internalising a group category it becomes part of our self-concept. As a result, a favourable image of the group reflects favourably upon the self. Moreover, like self-perceptions, group perceptions are always relational; they are created and sustained through comparison with outgroups. Consequently, according to social identity theory people distinguish their ingroups from outgroups in ways that they perceive favourably to reflect upon the ingroup (and thus upon themselves) and, by extension, negatively upon the outgroup. And since the quest for a positive self-image is ongoing, they continuously compare their ingroups to outgroups along dimensions that make their groups look good to them. 35

Group behaviour
For social identity theory, the two basic mechanisms of categorisation and social comparison thus account for group formation and group perception. But in addition, they also account for group behaviour -or, differently put, for the behaviour of individuals when acting in 'group mode'.
First, group members seek to align their actions and beliefs with what they perceive to be dominant group norms, with what social identity theory calls the 'group prototype'. For example, if the characteristics ascribed to the group include the attributes 'virile and brave', its members will seek to adjust their actions so as to appear to others and to themselves as 'virile and brave'. This tendency for behavioural conformity logically grows out of the categorisation, accentuation and internalisation principles discussed earlier. Individuals seek to level ingroup differences by conforming to the 'group prototype' and they apply this not only to how they perceive other ingroup members but also to how they perceive and, crucially, act themselves. As a result, people gravitate towards conformity with group norms even in the absence of hierarchies or institutions that seek to make them conform. 36 Second, social identity theory suggests that we can enhance our valuation of the ingroup and thus of the self not only by perceiving outgroups in a discriminatory fashion but also by treating them in this way, just as the subjects in the minimal group experiment improve their self-esteem by giving more to the ingroup than to the outgroup. In real world situations, of course, such discriminatory behaviour Societal security and social psychology 261 35 Social identity theorists disagree about the relative importance of evaluative factors (and thus social comparison) in eliciting group behaviour. Some go as far as to argue that cognitive mechanisms alone can be sufficient. 36 Though these may of course exist and reinforce the tendency for conformity inspired by group categorisation. The same logic suggests that the power to help shape the 'group prototype' brings with it the ability to influence the behaviour of group members. By extension, it offers a rationale for why actual and aspiring group leaders often try to present themselves as the embodiments of the 'group prototype'. often leads to intergroup conflict. Yet, according to social identity theory, even conflict itself can improve individuals' self-esteem in so far as it offers an opportunity to try and heighten the perceived superiority of the ingroup over the outgroup. For this reason, conflicts between groups are always about more than just the ostensible object of the dispute (such as territory or resources). Instead, they are in part about winning per se, to the point where winning can become an end in itself and where the defeat of one's group is often linked to a sense of personal humiliation. Finally, social identity theory asserts that once we have internalised a group category and thereby acquired the relevant social identity, we generally want to preserve it. We want to do so because of the cognitive importance of categorisation and because the group is now part of us and bound up with our sense of self-worth. Consequently, perceived threats to the group -such as to its relative status or to its very existence -are also threats to the self, and for the individuals concerned protecting the former means protecting the latter as well.
In sum, social identity theory rests on two pillars: categorisation and social comparison. These operate together [. . .] to generate a specific form of behaviour: group behaviour. This involves intergroup differentiation and discrimination, ingroup favouritism, perceptions of the evaluative superiority of the ingroup over the outgroup, stereotypic perception of ingroup, outgroup and self, conformity to group norms, affective preference for ingroup over outgroup, and so on. Categorization leads to stereotypic [. . .] perceptions of self, ingroup and outgroup, and also a degree of accentuation of intergroup differences. Social comparison accounts for the selectivity of the accentuation effect [. . .] and the magnitude of the exaggeration of intergroup differences and intragroup similarities. 37 For social identity theory, these elementary processes of categorisation and social comparison can enter into every group situation, ranging from nuclear arms races and relations between different communities in multicultural societies to clashes between rivalling football supporters or feuding family clans at a wedding. At the same time, social identity theory is self-consciously abstract. It makes no assumptions regarding the nature of groups, the signifiers used to demarcate group boundaries or the group norms that prevail at any given time. All these factors are socially constructed and therefore culturally specific and historically contingent. Accounting for them is the responsibility of historians, sociologists, anthropologists and political scientists. Their inquiries must focus on the role of social, economic and political power and interest in defining particular categorisation agendas and group norms, on the formation of intra-group hierarchies and so forth. The task of social psychology, by contrast, is to identify the underlying psychological parameters within which the social construction of groups and the processes within and between them transpire. To be sure, as many social theorists from Durkheim onwards have argued, attempts even indirectly to relate 'human nature' variables to particular social phenomena are fraught with dangers, not least that of pyschological reductionism. 38 These difficulties, however, cannot further concern us here. Instead, the next section returns to the Copenhagen school and its notion of societal security. It tries to show that social identity theory helps resolve the three major difficulties that are linked to the societal security concept as they were discussed earlier.

Societal security and social identity
As was argued, the first weakness of the societal security concept as proposed by the Copenhagen school is its tendency to reify societies -not by essentialising or 'primordialising' them, but by treating them as social actors with preferences and the potential for agency. At the same time, non-reified conceptions of social groups and institutions are hard to come by. The search for them can lead one dangerously close to the slippery slope of theorising the social world out of existence and of sliding into the kind of reductionistic individualism which the Copenhagen school is right to reject.
Social identity theory suggests a way out of this dilemma. It saves us from having to choose between adopting reifying notions of groups as purposive social agents or, alternatively, rejecting society as an analytical variable altogether and thereby ipso facto repudiating the societal security concept in the process. Groups, according to social identity theory, are internalised categories that give rise to representations of 'groupness' and to corresponding identifications, preferences and behaviour. These group representations are psychological facts that shape individuals' beliefs and actions, but they are not social actors themselves. Recall the minimal group experiment. The groups invented by the experimenter and internalised by the subjects are in the first instance mental representations. But qua mental representations they exist and account for the behaviour of the subjects in the experiment. In much the same way, nations, ethnic groups and other communities exist in the first instance inside the individuals who identify with them, as representations of 'groupness', 'nationness' and so forth.
All this, to be sure, does not relegate groups to the status of mental hallucinations. Instead, to remain psychological facts in the first place groups must manifest themselves 'out there' in individuals' everyday social experience. Consider the countless practices and artefacts, the many ways of thinking, acting, talking, relating, knowing, producing and feeling that are shaped by one's belonging to a particular group. Together, these inscribe the group in social and physical space; they sustain its perceived 'entitivity' and make it subjectively apprehended as real.
Seen against this background, the reproduction of identity communitiesnations, ethnic groups and the like -ultimately hinges on an ongoing 'dialectic of the internalization of externality and the externalization of internality'. 39 The community's members continuously 'act out' -or externalise -their internalised representations of the group by thinking, acting and feeling in line with prevailing group norms and the 'group prototype'. As individuals externalise their represent-Societal security and social psychology 263 ations of the community in this way and experience other individuals doing the same, the community becomes intersubjectively real. It appears as an object 'out there' in social space, liable to be observed and experienced. 40 To use the earlier terminology, it becomes a 'category of practice'. But as soon as individuals experience the group as objective reality they internalise it anew, with the resulting internalisations bound to be externalised again, and so forth. In reality, this dialectic of internalisation and externalisation is of course concurrent rather than sequential. People internalise and externalise simultaneously and continuously as part of a cycle of group reproduction that must continue for as long as the group exists. 41 The logic of societal security as postulated by the Copenhagen school must be understood in light of this dialectic. Individuals seek to defend their group boundaries because, as social identity theory postulates, group belonging satisfies basic cognitive and emotional needs. Sparked by securisation processes discussed earlier, this can lead to attempts to ward off a perceived threat to the group by, say, building a fence to keep out immigrants or fighting a war to defend the group's 'way of life'. But the very act of defending the group (through fences, wars or whatever) is itself a group-signifying and group-affirming one. By engaging in it, people externalise and affirm the existence of the group and their membership in it to each other and to themselves. Intersubjectively they come to experience the group more intensely and as more real. In the language of social identity theory, the group becomes more strongly internalised and more salient, and occupies a still larger part of the self. As a result, it becomes still more apt to elicit societal security responses, which in turn makes it still more salient and so on. In this way, the group as social representation that -mediated by securisation -conditions conflictual behaviour on the one hand, and intergroup conflict as a social process marked by this behaviour on the other, can become mutually constitutive. 42 In some circumstances, this cycle of group identification, group salience and group defence can accelerate to the point where it spirals out of the control of all social actors concerned. Once it has gathered enough momentum it has the potential to 'overcome' them and to contribute to the intensity and ferocity that characterises some group conflicts yet is often incomprehensible to those who are not caught in that cycle themselves. The many eyewitness descriptions of violence in places such as Rwanda and the former Yugoslavia in terms of 'suddenness', 'delirium' and 'ethnic intoxication' are, I think, an attempt to capture the effects of this dynamic on the participants. 43 On a more analytical level, a similar notion shines through in (still rare) attempts to conceive of such phenomena through the prism of a 'sociology of the event'. It takes seriously the notions of 'critical moments', sudden fluctuations, ruptures and 'overcome-ness' as factors that impact on the course of group conflict. 44 For the societal security approach, one important theoretical insight that flows from this is that the intensity of a given inter-group conflict does not just depend on factors such as the strength of initial securisation moves and the discursive power of the 'securitisers'. Instead, once the dialectic between group defence and group affirmation has taken hold, it can acquire a dynamic of its own and escape the control of the 'securitisers' who initially unleashed it. This in turn adds an important dynamic and transformative dimension to the societal security logic which the Copenhagen school largely lacks. Related to this, understanding securisation, group defence, group salience and group identification in dialectical terms enables one to endogenise the construction and transformation of groups into the logic of societal security processes -something, that, as was shown, the Copenhagen school itself also does not do. More specifically, it qualifies the insistence by the Copenhagen school that when analysing societal security processes we should treat communities as de facto fixed entities. As was argued, given the high degree of rigidity and 'sedimentation' of many groups such an approach is often justified. But in some circumstances, the logic of group conflict, group defence and group salience may transform communities so rapidly and so significantly that this transformation should be endogenised into accounts of group identity and group conflict. For example, for many Bosnians the very notion of 'Bosnia' and 'what it means to be Bosnian' might have changed dramatically during the relatively brief period of recent conflict, with these changes feeding back into the course of the conflict itself, and then again into representations of 'Bosnia' and 'Bosnian-ness' and so forth. 45 Finally, the same conception allows us to treat groups as both dependent and independent variables. They are dependent variables because, as is all social reality, they are ultimately human-made. They typically reflect both the particular social, economic and political interests of the initial categorisers but also, following social identity theory, a basic human need for 'groupness'. At the same time, groups are also independent variables, in two particular ways. First, as does all 'sedimented' social reality, they confront most individuals as something 'external' to them and mould their beliefs and behaviour. Second, through the dialectical process discussed above, groups can become literally independent: they can influence behaviour and transform themselves in ways that do not reflect the intentions of any social actor, the initial categorisers and 'securitisers' included.
However, contrary to what societal security theorists at times imply, being an independent social variable is not the same as being an independent social agent: the latter suggests consciousness, purpose and self-control whereas the former does not. Just as a pilot who loses control over his or her airplane in a thunderstorm does not Societal security and social psychology 265 thereby enable the airplane to control itself, so groups and group processes that spiral beyond the influence of those who once controlled them do not get to 'control themselves'; they cannot develop preferences or display agency and they cannot value or protect 'their' societal security. As do many other approaches in the social sciences, the Copenhagen school often blurs the distinction between 'groups as independent social actors' and 'groups as independent social variables'. But this distinction is crucial. The former entails impermissible reification while the latter builds on the valuable insight that social formations and processes can move beyond the control of the individuals who initiated, participate in and are affected by them. And as was shown, this can apply to the logic of societal security as much as it can apply to any other social process.
To summarise: first, for the Copenhagen school to claim that groups want to defend their societal security is broadly equivalent to social identity theorists asserting that individuals want to preserve their internalised group categorisations. The latter, however, is conceptually much more precise and avoids reifying and 'substantialist' language. Moreover, as is shown below, it paves the way for an understanding of the motivational factors that drive the societal security logic in the first place. Second, societal security theorists are mistaken in their claim that treating society as an independent social agent is both essential to the societal security logic and the only alternative to a reductionistic individualism that dispenses with society as an analytical concept. Instead, social identity theory allows us to de-reify societies without thereby theorising them out of existence, to treat them as independent variables but not as independent agents. This leads to the second vulnerability of the societal security concept, namely, its reliance on too vague a definition of 'identity'. Here, too, social identity theory does valuable work. As was shown, it takes social identity to pertain to group categories that have become internalised and thus part of the self. This definition of identity is close to the term 'sameness' which reflects its literal meaning. Following on from this, social identity theory treats group identity as a process in the same way that the dialectic of internalisation and externalisation that sustains group representations is itself ongoing: to 'have' a particular identity means continuously to identify with the identity object and to remain entangled in the process that makes identification possible. However, as does the societal security model, social identity theory does not claim that identities are necessarily fickle and transient. Some groups, such as those in the minimal group experiment, are very ephemeral. But through the process discussed, many 'real world' identity communities come to be so deeply internalised and inhabit such a large portion of the self that they become very resistant to change. Though neither groups nor group identities are things, they can acquire a 'thing-like' texture. It is this characteristic which enables the societal security logic to take hold.
Still on the identity issue, social identity theory offers another important insight. It flows from its insistence on the cognitive importance of group categorisation and is borne out by the minimal group experiments in which subjects identify with their groups even though -being 'minimal' -they lack any cultural characteristics. This implies that an individual's social identity does not derive in the first instance from the cultural attributes of the community in question (such as, its language, values and customs). Societal security is not cultural security. Instead, according to social identity theory, individuals seek to protect group boundaries because group categorisations and group differences are intrinsically valuable to them. They do value particular cultural attributes, but less as ends in themselves than as markers that sustain the boundaries of their groups. In short, societal security theorists sometimes imply that individuals want to defend their groups because these groups contain their culture. Social identity theory, by contrast, suggests that people value their culture because it helps sustain their groups, with 'groupness' serving basic cognitive and emotional needs.
All this echoes Fredrik Barth's well-known conclusion that what is important about groups is their boundaries rather than the cultural stuff they enclose. 46 It also conforms to the observation that groups often change their signifiers while the boundaries they signify and thus the groups themselves remain intact. 47 There is a danger here of sliding into excessive 'social psychologising': not everything people value and want to preserve is valued because it signifies group boundaries. All the same, if one accepts the central claim of social identity theory that group identity and by extension societal security is more about group boundaries than about group 'content', an expanded research agenda for societal security studies comes into view. This agenda brings it into overlap with semiotics and large parts of contemporary cultural theory. It must focus on how particular cultural signifiers become 'attached' to and 'detached' from group boundaries, and thereby become bound up with processes of securisation and desecurisation respectively. A given development constitutes an identity threat if it is perceived to threaten something that signifies group boundaries. But once the thing in question no longer signifies group boundaries a threat to it no longer threatens these boundaries. For instance, if being a 'white country' ceases to be a defining aspect of 'who we are', then 'non-white' immigrants stop threatening 'our identity' and stop eliciting societal security responses. Immigration thus becomes desecuritised.
Among societal security theorists, Ole Waever in particular has increasingly come to focus on the potential for desecurisation. 48 And in some ways this is a happier project by comparison, as it helps offset the bleaker 'tribalist' implications that stick to social identity theory and the societal security concept alike, and that trouble scholars in both traditions. 49 If identity pertains more to group boundaries than to particular group norms, a change of these norms -for instance away from racial into 'civic' directions -becomes more conceivable.
There remains the final problem with the Copenhagen school's concept of societal security, namely its failure to account for motivational factors. As was argued, in places its proponents make the reifying claim that societies seek to protect their societal security -a view that makes the question of human motivation redundant to the extent that it does not treat individuals as the central actors. Elsewhere, they take the sounder view that individuals want societal security, but largely fail to elaborate on what motivates them. As should have become clearer by now, social identity theory offers an answer: people defend their group boundaries because they have an innate need for 'groupness', that is, to maintain self-and othercategorisations and to protect their groups as sources of positive self-evaluations. In this way, social identity theory provides a motivational explanation which the Copenhagen school lacks and renders the concept of societal security sharper and more plausible.