2. Suicide terrorism as extreme self-sacrifice

In his classic discussion of altruistic suicide, Durkheim detailed many examples of individuals taking their lives for the glorification of the group. Durkheim cited Charlevoix's colourful observations in Japan as exemplars of the category:

In the passage where this quotation occurs, and in many others, Durkheim is at pains to emphasize that altruistic suicide is an expression of “underindividuation” – the sublimation of the personal self to the greater and nobler will of the group. Durkheim associated underindividuation with what he called “the lower societies” – small-scale traditional societies where the division of labour was sufficiently simple that persons seemed eminently replaceable, one person substituting more or less easily for any other. Durkheim famously argued that the emotionally arousing rituals of simple societies served to heighten “collective conscience” to a point of great fervour and excitement, eclipsing or even obliterating any sense of personal agency and individuality. He regarded the military as one of the last refuges of this primitive mentality in otherwise civilized societies, and used this to explain the high incidence of altruistic suicide in various armies and navies, as well as self-sacrifice on the battlefield (Durkheim Reference Durkheim1897/1951, pp. 234–37).

In what follows, however, a somewhat different view is presented. Fusion resulting from the perception of shared essence does not sublimate individual identity but enables group alignments to tap into the motivational power of personal agency, strengthening and emboldening it. Fused individuals see themselves not merely as instruments of the group, but as willing to do more for the group than other members. They see their personal self as encompassed, but not eclipsed, by the group. From this view, altruistic suicide is not caused by “underindividuation” as Durkheim conceived of it, but by a visceral sense of oneness between self and group.

Suicide terrorism, the act of laying down one's own life in an effort to weaken or destroy an out-group, may also be regarded as a form of altruistic suicide (Pape Reference Pape2005). As such, it forms part of a much larger class of extreme behaviours that involve sacrificing oneself for the group, ranging from nonhomicidal forms of suicide (such as taking a bullet for a comrade or setting fire to oneself) through to so-called “suicide missions” in conventional armies. Although suicide terrorism may be shaped by sophisticated strategic considerations (e.g., careful selection of targets to maximally advance terrorist objectives), altruistic motivation is a necessary condition for implementing these strategies, inasmuch as the individual is required to give up his or her own life for the sake of a group. Such actions may be described as extreme because they are (or are most likely to be) fatal. Efforts to explain the phenomenon have often emphasized its seemingly irrational nature, for example, attributing such suicides to extreme religious beliefs or to individual pathology (Pape Reference Pape2005, p. 16).

The notion that religious extremism (e.g., via indoctrination) motivates suicide terrorism may seem a little more than common sense. Indeed, according to some public intellectuals (Harris Reference Harris2004), religious doctrines extolling the righteousness of waging holy wars or of exterminating infidels self-evidently explain extreme self-sacrifice such as suicide bombing. And it is easy to find examples throughout the history of suicide terrorism linked to religious beliefs of various kinds, from the Jewish Zealots and the Sicarii of antiquity to the many Islamist terror groups of today. Less convenient for those who favour this line of argument is the fact that most suicide attacks, at least until 2000 (Atran Reference Atran2006), have been carried out by secular organizations and not by religious extremists (Gambetta Reference Gambetta and Gambetta2005; Post et al. Reference Post, Ali, Henderson, Shanfield, Victoroff and Weine2009). Moreover, studies in the lab and in natural settings suggest that religious beliefs lead to pro-social action because they serve as markers of group alignment. That is, it would seem to be attachment to a collective, forged through shared rituals or other identity markers, and not beliefs per se that motivate pro-group action (Bloom Reference Bloom2012). For example, in a series of studies (Ginges et al. Reference Ginges, Hansen and Norenzayan2009), frequency of participation in collective rituals has been shown to predict support for suicide attacks, whereas frequency of prayer (as a less groupish devotional act) does not. Ginges et al. found that for Palestinian Muslim adults, frequency of mosque attendance predicted stated willingness to die for one's religion as well as support for suicide attacks, but again, frequency of prayer did not.

It is possible that some extreme beliefs become so closely linked to the group that they take on an aura of sacredness (Atran Reference Atran2010), but if that is so, then what connects those values to acts of self-sacrifice may well be fusion with the group rather than commitment to any kind of explicit belief system or ideology per se (Whitehouse Reference Whitehouse2016a). Efforts to disambiguate the effects of sacred values and group alignment are hard to interpret given that the measures of sacred values so far used in such studies are related to similar measures of willingness to sacrifice for sacred values (Gómez et al. Reference Gómez, López-Rodríguez, Sheikh, Ginges, Wilson, Waziri, Vázquez, Davis and Atran2017).

The theoretical framework presented here suggests that willingness to fight and die is not motivated by doctrines and ideologies, religious or otherwise, but by a particularly intense love of the group. For highly fused individuals, the survival of the group constitutes a form of personal immortality in ways that may be more real psychologically than any religious dogma alone could convey.

Nevertheless, those who argue that suicide terrorism is a result of pathology may not regard such behavior as an act of self-sacrifice at all. For example, Lankford (Reference Lankford2014b) has argued that jihadist martyrs may simply be suicidal individuals (see also Merari Reference Merari2010) who choose this method of killing themselves because it is socially approved and even glorified by the groups with which they identify, although also providing conveniently reliable methods (e.g., fatal explosive devices). In this view, the suicide bomber is not embracing death to achieve some greater goal but is actually engaging in a cowardly and selfish act. “For many suicide terrorists, blowing themselves up may feel like the least risky thing they could do – it could offer the greatest certainty that their overwhelming crisis will no longer plague them. For these individuals, the risky thing may be to face their uncertain future, tackle their problems one day at a time, or swallow their pride and ask for help” (Lankford Reference Lankford2014b, p. 360). Lankford (Reference Lankford2014b, p. 360) argues also that acts of suicide terrorism and genuine acts of self-sacrifice differ in the manner of their planning. Suicide attacks are generally planned long in advance, whereas heroic self-sacrifice, such as leaping on a grenade and shielding a comrade in a hail of bullets, is more typically a split-second reaction to unforeseen events. Moreover, the suicide terrorist plans to die, whereas genuine military heroes hope to survive, despite appalling risk to life and limb (Reference Lankford2014b, p. 359).

The distinction between laying down one's life for a group and merely risking one's life may not be easy to draw, however. Expressed in terms of probabilities, it has been calculated that the chances of surviving an act for which a Victoria Cross medal has historically been awarded in the British and Commonwealth forces is just 1 in 10 (Gambetta Reference Gambetta and Gambetta2005, p. 272). Salim Jawha, a former colonel in Gaddafi's army who joined the insurgency in Misrata on the first day of the revolution, put it to me like this: “When the revolution began, there was no compulsion to join. We just called our friends and asked them, ‘Do you want to die or not? If you want to die, come with us. If not, go home and stay out of harm's way’” (Jawha, quoted in Whitehouse Reference Whitehouse2016a). Of course, there was no guarantee that one would die as a result of joining the revolutionary forces. But the chances of dying in action were dauntingly high, and those choosing to fight were well aware of this. Nevertheless, thousands made the decision to go ahead anyway. They were not suicidal but they were surely prepared to lay down their lives.

Lankford's argument that suicide terrorists are suicidal has deservedly triggered much instructive debate but has also prompted criticism, mainly highlighting limitations of the supporting data (Atran Reference Atran2003; Beit-Hallahmi Reference Beit-Hallahmi2014; Egan Reference Egan2014; Funder Reference Funder2014; McCauley Reference McCauley2014; Merari Reference Merari1993; Qirko Reference Qirko2014; Sela & Shackelford Reference Sela and Shackelford2014; Tobeña & Vilarroya Reference Tobeña and Vilarroya2014; Weiss & Weiss Reference Weiss and Weiss2014). Although these debates remain unresolved, according to Post et al. (Reference Post, Ali, Henderson, Shanfield, Victoroff and Weine2009), there is a well-established consensus among researchers that group, social, and organizational factors provide the key to understanding most suicide attacks. Factors frequently implicated in this form of terrorism are collective identity (Post Reference Post, Ali, Henderson, Shanfield, Victoroff and Weine2009), kin psychology (Gray & Dickens Reference Gray and Dickins2014), and out-group hostility (Ginges et al. Reference Ginges, Hansen and Norenzayan2009). New research into the causes of identity fusion suggests that these seemingly disparate theories of suicide terrorism may in fact be highlighting different aspects of a single process. As with some past approaches to suicide terrorism, the fusion theory emphasizes group alignment and psychological kinship, but it also focuses on the role of shared essence combined with perceptions of out-group threat in motivating extreme self-sacrifice. This general theoretical framework is further elaborated and empirically substantiated in the next section.

Nevertheless, studying the role of identity fusion in motivating suicide attackers is fraught with practical difficulties. Successful suicide attackers are, by definition, already dead, and those who are foiled, even if they can be interviewed, may be unable or unwilling to provide accurate information on the psychological processes that drew them to violent extremism in the first place. The studies reported below have focused largely on measuring identity fusion and extreme self-sacrifice among currently serving members of armed groups, ranging from revolutionary insurgents to conventional military forces, especially those who, having witnessed the violent deaths of many of their fellow fighters, nevertheless voluntarily expose themselves to the same high risks. Much recent research also investigates the role of identity fusion among those who strongly endorse the use of violent self-sacrifice to accomplish group goals (Swann et al. Reference Swann, Gómez, Huici, Morales and Hixon2010b; Reference Swann, Buhrmester, Gómez, Jetten, Bastian, Vázquez, Ariyanto, Besta, Christ, Cui, Finchilescu, González, Goto, Homsey, Sharma, Susianto and Zhang2014a). Despite the difficulties of conducting research into these topics, there is growing evidence that fusion can motivate extreme pro-group action (Whitehouse et al. Reference Whitehouse, McQuinn, Buhrmester and Swann2014b; Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017) and that this process could plausibly explain at least some, if not most, instances of suicide terrorism as well as other forms of violent extremism.

3. Shared essence, fusion, and willingness to fight and die for a group

Fusion – a visceral feeling of oneness with a group (Swann et al. Reference Swann, Gómez, Seyle, Morales and Huici2009) – entails an identity configuration such that essential features of one's social identity are also considered to be essential features of one's personal self. Essentialized conceptions of the personal self or of a group presume the presence of nonobvious properties that are necessary for the entity to exist in its current form (Medin & Ortony Reference Medin, Ortony, Vosniadou and Ortony1989). This is an all-or-nothing mode of categorization in that the person or group must have this nonobvious property or would no longer be the person or group in question (Diesendruck & Gelman Reference Diesendruck and Gelman1999, p. 339). But how do people come to attribute essences to persons and groups? One possibility is that the essential personal self derives from inherited biological attributes (Gil-White Reference Gil-White2001), another that it is formed through life-defining experiences (Whitehouse Reference Whitehouse2013; Whitehouse & Lanman Reference Whitehouse and Lanman2014). That is, we can imagine ourselves as being products of inherited traits as well as being moulded by unique personal histories (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017).

A series of recent studies sought to compare the effects on fusion of shared biology versus shared life experiences. In one study (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017), 198 participants were assigned to three priming conditions in which they wrote about either a self-shaping experience, a set of traits transmitted genetically, or the changing seasons (as control). They were then asked to imagine meeting somebody who either had shared their transformative life event, had turned out to be a long-lost sibling, or was a complete stranger (control). Those in the shared experience and shared biology conditions reported higher levels of fusion with the imaginary person, although interestingly, the effects were notably stronger for shared experience. In a further study, actual, rather than imaginary, shared genes and shared experience were experimentally manipulated (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017). Two hundred and sixty monozygotic and 246 dizygotic twins were asked to describe transformative experiences shared with their twin, and measures of fusion with twin were then taken. The results showed that both shared biology (as measured by zygosity) and shared experience predicted fusion levels independently.

Hence, sharing either biological or autobiographical essence with other group members, or both, produces identity fusion. Highly fused individuals report intense family-like connections to other group members, high levels of personal agency, and feelings of invulnerability in their group. Therefore, when the group is felt to be threatened, it feels personal. This may help explain why so many groups committed to violence describe themselves as victims acting to defend themselves against external aggressors (Furedi Reference Furedi2015). The desire to protect the group is experienced by fused individuals as an urgent and compelling act of self-defence. This process may resemble the way in which mortality salience stimulates in-group defence and support in the population at large (Fritsche et al. Reference Fritsche, Jonas and Fankhänel2008) but more acutely in the case of highly fused individuals, as a result of amplified feelings of personal agency (Swann et al. Reference Swann, Gómez, Huici, Morales and Hixon2010b). Highly fused individuals exhibit an urge to make personal sacrifices for their group, ranging from donations of blood to helping victims of terrorist attacks (Buhrmester et al. Reference Buhrmester, Fraser, Lanman, Whitehouse and Swann2015) to fighting on the front lines at grave personal risk (Whitehouse et al. Reference Whitehouse, McQuinn, Buhrmester and Swann2014b).

Efforts to investigate the shared experience pathway to fusion have focused on the role of self-defining episodic memories (Whitehouse Reference Whitehouse2013). Psychologists have long appreciated that memories for transformative experiences have an impact on the development of personal identity (Conway Reference Conway1995; Singer & Salovey Reference Singer and Salovey1993), and considerable empirical research has been conducted into the mechanisms linking memory of and reflection on life-changing experiences to autobiographical narratives and the construction of the personal self-concept (Çili & Stopa Reference Çili and Stopa2014). What fusion theory adds is the insight that some emotionally intense experiences are not only transformative in shaping the personal self but – insofar as these experiences are shared with relevant others – they also define the group (Whitehouse Reference Whitehouse2004), producing a visceral sense of “oneness” or shared essence. Consequently, the personal self and the group self are uniquely conjoined in fused individuals – being forged through the same potent experiences that endure in memory (Whitehouse & Lanman Reference Whitehouse and Lanman2014).

To investigate these processes, recent empirical research has focused on populations that have undergone emotionally distressing experiences together. Events involving psychological or physical suffering appear to have an especially enduring impact on memory and subsequent reflection (Conway Reference Conway1995; Pillemer et al. Reference Pillemer, Koff, Rhinehart and Rierdan1987; Whitehouse Reference Whitehouse2002). In the literature on shared experience and identity fusion, such events are typically described as “dysphoric” (Jong et al. Reference Jong, Kavanagh, Lane and Whitehouse2015; Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017), but they are often also accompanied by feelings of elation, for example, upon surmounting the ordeal, and it may be that an emotional “high” is partly responsible for the long-term memory effects of such experiences and their impact on fusion (Kavanagh et al. Reference Kavanagh, Jong, McKay and Whitehouse2018). The theory elaborated here is concerned primarily with the impact of shared emotional events on the fusion process (via the mediating effects of memory, reflection, etc., as set out in Fig. 1). Some of the research described in the remainder of this section suggests that dysphoric events have a bigger effect on fusion than euphoric ones, but it is not yet clear whether negative valence or merely overall emotional intensity is the key factor.

A cluster of studies focused on the effects of recalling terrorist attacks in New York, Madrid, and London have shown that the simple act of remembering such experiences increases reported levels of fusion with and willingness to die for one's country (Buhrmester et al., Reference Buhrmester, Gómez, Vázquez, Mercedes, Lanman, Swann and Whitehousein preparation). Moreover, the extent to which dysphoric experiences are felt to be transformative or ”self-defining” predicts their effects on fusion. Similarly, the more that nationalists and unionists in Northern Ireland have reflected on their sufferings during the history of sectarian conflict in that region, the more fused they are with their respective communities (Jong et al. Reference Jong, Kavanagh, Lane and Whitehouse2015). Further, to demonstrate that such reflection actually causes elevated fusion, Bostonians affected by the 2013 Marathon bombings were primed with memories of the atrocity (versus controls who were invited to recall memories of running errands in Boston), and those who felt more intense negative emotions as a result of the prime were indeed more fused with their fellow Bostonians (Jong et al. Reference Jong, Kavanagh, Lane and Whitehouse2015).

Similar patterns have been observed among spectators at football matches where fans who felt they had been most shaped personally by their memories of especially emotional club events also reported higher levels of fusion (Newson et al. Reference Newson, Buhrmester and Whitehouse2016). For example, a recent survey of Premier League football supporters in the United Kingdom found that sharing the dysphoric experience of losing soccer matches made fans more willing to sacrifice themselves for each other, an effect that was mediated by identity fusion (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017). Losing seems to fuse fans more effectively than winning, and, consequently, the less successful teams are likely to have the most loyal fan bases. Of course, this also means that the more embattled supporters would show greater eagerness to fight on the terraces, but, although such fervent commitment to club does sometimes spill over into violence, intergroup rivalry is more typically expressed in relatively harmless symbolic acts such as chanting and song.

Some aspects of the process by which shared dysphoric experiences lead to fusion have now been measured physiologically, as well as by using psychometric tests and self-report. For example, a recent longitudinal study of Brazilian fans during the 2014 FIFA World Cup revealed that increased heart rate and cortisol levels during live matches predicted higher levels of fusion over time (Newson et al., Reference Newson, Buhrmester, Jong, Hattori, Shiramizu, Yamamoto and Whitehouseunder review). Efforts have also been made to explore the physiological processes by which out-group threat motivates self-sacrifice in highly fused individuals. For example, studies using brain scans to monitor activity in the medial prefrontal cortex, a brain region associated with group membership processing, have explored the effects of fusion on the brain when deciding to pay a personal cost to harm a rival out-group. In these studies, football supporters playing economic games with rival fans showed a marked tendency to punish fair and unfair offers alike, at cost to self. Activity across ventral and dorsal portions of medial prefrontal cortex, however, was affected by levels of fusion in these participants; the more highly fused players showed the strongest evidence for fairness by group membership interaction in the relevant brain regions (Apps et al. Reference Apps, McKay, Azevedo, Whitehouse and Tsakiris2018).

In the preceding examples, sharing emotionally intense experiences was shown to give rise to fusion and, therefore, willingness to fight and die for the group. Nevertheless, most of these studies rely mainly on self-report measures and hypothetical cooperation problems. Although behavioural measures, such as economic games and trolley problems, have been used in some of the studies reported above, these are not ideal proxies for extreme self-sacrifice, our main interest in the present context. To test this theory properly requires investigation of processes of fusion among people who actually lay down their lives for each other. The ethical and practical impediments to conducting psychological research with suicide bombers undergoing training are severe, however. And, although it may be possible to interview thwarted terrorists serving prison sentences, this approach, too, is fraught with difficulties, including the need to build rapport and trust with interviewees, as well as to overcome incentives for offenders to provide self-serving or misleading responses.

Recent efforts to investigate violent extremism operating outside the rule of law have focused on studies with frontline fighters in military groups such as members of revolutionary battalions in Libya during 2011, the year of the Arab Spring. The vast majority of Libyans who took up arms in 2011 were ordinary civilians, many of whom had never even held a gun prior to the uprising. All of them knew their chances of survival were poor. Many thousands were killed or suffered devastating injuries, and all of them lost friends and family during the conflict.

Interviews were conducted with 179 insurgents in the city of Misrata (Whitehouse et al. Reference Whitehouse, McQuinn, Buhrmester and Swann2014b). Half the sample were frontline fighters, and the others were providers of logistical support to the fighters. All reported near-ceiling levels of fusion with their families, with their closest friends in the battalions, and even with the members of other battalions, but not with pro-revolutionary Libyans who never joined a battalion and therefore did not fully share the intensely dysphoric experiences of participation in the 2011 uprising. In view of these high levels of fusion with multiple groups, a forced-choice question was introduced: If you had to choose only one group as your primary fusion target, which would it be? This produced a striking difference between those who faced the most traumatic ordeals of the war and those who suffered, but not as intensely, by working behind the scenes. Frontline fighters were nearly twice as likely (compared with providers of logistical support) to choose fellow revolutionaries over their families.

A limitation of this study, however, is that there was no way of ascertaining whether shared dysphoric experience in battle led to high levels of fusion, or fusion drove Libyan insurgents to the front line in the first place. To adjudicate on this question, studies have been undertaken with conventional troops that lacked control over their deployment on the front lines. For example, a survey of 380 Vietnam war veterans in the United States found that intensity of dysphoric combat experience predicted fusion with fellow fighters and that this fusion also mediated willingness to make personal sacrifices for other veterans (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017). Although there are many practical impediments to conducting research with groups whose members have sacrificed their lives in appreciable numbers, and the evidence drawn upon here is limited, efforts to access more participants in other troubled regions of the world are ongoing. Data collection using much the same methods as those employed in Libya in the research described above has been undertaken among fundamentalist Islamist groups (Kavanagh et al., Reference Kavanagh, Buhrmester, Yustisia, Rufaidah, Wibisono, Sofianna and Whitehousein preparation; Yustisia et al., Reference Yustisia, Putra, Rufaidah, Whitehouse and Kavanaghunder review), and efforts are also underway to collect data from convicted terrorists in prison.

4. The evolution of fusion and extreme self-sacrifice

From a gene's eye point of view, the vehicle for its transmission (the individual organism) should usually preserve its own life and maximize its reproductive advantage at all costs, but kin selection famously presents an exception. Kin selection causes genes to increase in frequency when the degree of genetic relatedness of individuals benefiting from an altruistic act, multiplied by the benefit thereby procured, is greater than the reproductive cost to the altruist. Self-sacrifice to save fellow group members might make evolutionary sense if all members of the in-group are genetically very similar, as in the case of some eusocial insects. In the case of humans, taking a fatal spear to protect one's kinsmen might be similarly adaptive if, under evolutionarily stable conditions, self-sacrificing individuals stood a better chance of passing on their genes via surviving relatives than by mating successfully themselves.

Fusion may have arisen as a psychological adaptation to facilitate cooperation among kin in the face of extreme adversity, such as lethal out-group threat (Whitehouse & Lanman Reference Whitehouse and Lanman2014), but it may simply be a by-product of the way autobiographical memory evolved in humans. Further evolutionary explanations for fusion are considered below. Nevertheless, even if fusion was a psychological adaptation that arose through kin selection, social institutions could hijack the fusion mechanism in novel ways. For example, male initiations involving extreme privations and sufferings could mimic the trials and tribulations of kin groups struggling to survive in tough environments, producing similar psychological effects and behavioural outcomes. An important difference is that the ordeals of initiation deliberately maximize emotional and sensory impact, so as to create a stable impression of shared essence in an imaginary “brotherhood” (Whitehouse Reference Whitehouse and Corrigan1996; Reference Whitehouse2004). In everyday life, by contrast, the shared sufferings and setbacks of kin groups occur more unpredictably, and their emotional intensity, causal opacity, and consequentiality for group members are more variable, taking many years to assemble and, therefore, to produce fusion and psychological kinship (Whitehouse Reference Whitehouse2013). In short, some social institutions may have arisen through cultural evolution that exploit our biologically evolved systems for supporting and defending kin in the face of adversity.

It is also conceivable that culturally evolved discursive practices exploit intuitions of shared biology to some extent, for example, by referring to priests as “fathers” or group territory as the “motherland.” In a recent survey covering 11 countries on six continents, 86.1% of the 2,438 respondents expressed willingness to die for their families before any other group (Swann et al. Reference Swann, Buhrmester, Gómez, Jetten, Bastian, Vázquez, Ariyanto, Besta, Christ, Cui, Finchilescu, González, Goto, Homsey, Sharma, Susianto and Zhang2014a), but the same survey also showed that priming feelings of shared biology among people already fused with their countries made them more willing to make extreme sacrifices for their fellow countrymen. Mediation analyses showed also that fusion had an impact on willingness to fight and die for country via feelings of kinship. These findings are supported by studies of groups actually (as opposed to hypothetically) experiencing an external threat. For example, in the wake of the 2013 Boston Marathon bombings, the willingness of locals fused with America to give blood or money to help the victims was mediated by feelings of psychological kinship with fellow countrymen, expressed by endorsing statements like “members of my country are like family to me” (Buhrmester et al. Reference Buhrmester, Fraser, Lanman, Whitehouse and Swann2015).

The kin selection account for the evolution of fusion suggests that shared life-shaping experiences, just like biologically inherited traits, may have served as reliable phenotypic markers in ancestral groups composed mainly of closely related individuals (Lieberman et al. Reference Lieberman, Tooby and Cosmides2007; Whitehouse & Lanman Reference Whitehouse and Lanman2014), and this also seems consistent with the finding that sharing core values or attitudes signals genetic relatedness (Park & Schaller Reference Park and Schaller2005; Swann et al. Reference Swann, Buhrmester, Gómez, Jetten, Bastian, Vázquez, Ariyanto, Besta, Christ, Cui, Finchilescu, González, Goto, Homsey, Sharma, Susianto and Zhang2014a). A common cause of shared life-shaping experiences prompting fusion and extreme self-sacrifice may have been chronic intergroup raiding and warfare. There is some debate on the prevalence of warfare in the Pleistocene, but some scholars estimate that up to 40% of all male deaths in human prehistory can be attributed to intergroup conflict (Keeley Reference Keeley1996). Even if this seriously exaggerates war-related mortality rates (Ferguson Reference Ferguson and Fry2013), there is little doubt that intraspecies violence is a widespread and persistent feature of human behaviour (Gómez et al. Reference Gómez, Verdú, González-Megías and Méndez2016b), and ancient foragers probably always faced threats of predation that were best parried by standing together, despite strong temptations to defect or run away.

Previous research has likewise emphasized family-like bonds as a powerful motivator of suicide terrorism (Atran Reference Atran2003; Mandel & Litt Reference Mandel and Litt2013) and of self-sacrifice for comrades in conventional military groups (e.g., Stouffer et al. Reference Stouffer, Lumsdaine, Lumsdaine, Williams, Smith, Janis and Cottrell1949; Vaughan & Schum Reference Vaughan and Schum2001). Gray and Dickens (Reference Gray and Dickins2014) link this urge to protect one's “brothers-in-arms” to perceptions of shared biology based on phenotypic matching. In an interesting application of kin selection theory to the phenomenon of suicide in general, it has been suggested that individuals with little chance of reproducing may constitute a drain on the resources of their kinsmen, such that committing suicide might actually increase their inclusive fitness by improving their chances of passing on genes via their surviving relatives (deCatanzaro Reference deCatanzaro1980; Syme et al. Reference Syme, Garfield and Hagen2015). One might object that the act of suicide could itself cause serious collateral damage, impairing the prospects of bereaved kinsmen, not least because of social stigma. And Joiner (Reference Joiner, Hom, Hagan and Silva2015) has argued that suicide, in general, results from a pathological underestimation of one's own evolutionary worth and overestimation of the burden one is inflicting on others. But Joiner's argument could be turned on its head in the case of suicide bombers, in that martyrdom may indeed serve to improve the circumstances of their families. For example, Blackwell (Reference Blackwell2008) has argued that Palestinian suicide attackers increased their inclusive fitness outcomes by contributing, through the celebrity of their deaths, to the well-being and reproductive prospects of their close kin.

Following a closely argued discussion of these issues, Orbell and Moriwaka (Reference Orbell and Moriwaka2011) consider whether Blackwell's argument can shed light on the motivations of kamikaze pilots in World War II. Based, however, on extensive analysis of letters, poems, wills, and memoirs left behind by 661 Japanese pilots who perished in suicide attacks, the authors conclude that the act of self-sacrifice was not undertaken to increase the welfare of close kin. Rather, the writings of kamikaze pilots placed much greater emphasis on the desire to die for the nation or for the emperor. One possibility, considered by Orbell and Moriwaka, is that an evolved algorithm to sacrifice self for kin had somehow been hijacked by nationalistic ideology, such that the country or its ruler had taken on the status of family. Nevertheless, as these authors also observe, evolution should act strongly against mistaking fellow countrymen for kin, especially where the stakes are so high. Rather, they suggest that when warfare becomes genocidal, a distinct psychological mechanism comes into play that enables kin groups to form coalitions:

Intriguing as this argument undoubtedly is, it is by no means clear that genocidal out-group threat, as envisaged by Orbell and Moriwaka, was sufficiently acute and widespread in human prehistory for such a mechanism to evolve under natural selection. At any rate, it would be wise to consider a range of alternatives to kin selection to explain how extreme self-sacrifice may have evolved under natural selection.

One recent study presents a mathematical model predicting that any group suffering negative experiences, not only kin groups or coalitions of kin groups, should be more cooperative in the future (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017). The model, based on a multilevel selection approach, considered an imaginary population divided into numerous groups whose survival depended on overcoming two kinds of collective problems: “us versus nature” contests (e.g., hunting game, defending against predators) and “us versus them” contests (e.g., conflict with other groups over access to resources). In the real world, failure of any given group in an “us versus nature” contest might have little impact beyond the group itself; by contrast, doing well or badly in “us versus them” contests would likely have a significant impact on surrounding groups participating in conflict. In the model, “doing well” equated to a euphoric experience with a corresponding positive impact on the fitness of group members, as measured by their capacity to leave offspring. By contrast, “doing badly” equated to dysphoric experience and had the opposite effect. Groups that did so badly they died out were replaced by the offspring of more successful groups.

The setup consisted of two rounds. In the first round, euphoric (fitness-increasing) or dysphoric (fitness-reducing) experiences were randomly assigned. If that was all there was to it, then groups undergoing euphoric, fitness-enhancing experiences would obviously do better than those that had dysphoric experiences. But the model also contained a second round in which the consequences of each group's efforts overall to solve collective action problems were measured. Every individual was allocated a pair of genes, one of which was capable of triggering cooperative behaviour only after a euphoric experience and the other of which was capable of doing so only after a dysphoric experience. But only one of these genes could be expressed in any given individual, and all individuals of the same group shared this capacity. Thus, the model was designed to study the evolution of genetic endowments controlling behaviour. Although these genetic endowments made cooperation possible, they certainly did not make it inevitable. In fact, the model was set up so as to make the chances of cooperation close to zero. However, we ran not one, but many simulations, allowing for mutation, recombination, and migration. Each group lineage went through both euphoric and dysphoric experiences at equal frequencies. The question was whether cooperative behaviour would evolve over repeated simulations. The finding was that the gene effects on cooperation resulting from dysphoric experiences evolved to be larger than gene effects on cooperation resulting from euphoric experience. This pattern was even more pronounced with groups engaged in “us versus them” contests than with those engaged in “us versus nature” contests, the former involving more intense intergroup competition.

5. Ritual and local fusion

Throughout most of prehistory, our ancestors lived in hunter–gatherer bands, confronted with both “us versus nature” and “us versus them” contests. If the fusion mechanism originally evolved in small, relational groups of this kind, then it should come as no surprise that terrorists are also usually tied to highly cohesive cells comprising around 10 core individuals on average (Richardson Reference Richardson2006). We refer to this as local fusion (Swann et al. Reference Swann, Bastian, Gómez, Jensen and Whitehouse2012; Whitehouse Reference Whitehouse2013), as distinct from extended fusion discussed in the next section. Whereas fusion in ancestral foraging bands may have come about through exogenously caused dysphoric events, such as enemy ambush and natural disaster, some prehistoric groups probably also engaged in rare, emotionally intense rituals, serving as cultural gadgets to increase levels of fusion among warriors and hunters.

Many scholars have observed that collective rituals are a potent source of cohesion in social groups and that the more physically or psychologically arduous the rituals in question, the more powerfully they bond participants (Durkheim Reference Durkheim1912; Henrich Reference Henrich2009; Irons Reference Irons and Nesse2001; Olivola & Shafir Reference Olivola and Shafir2013; Xygalatas et al. Reference Xygalatas, Mitkidis, Fischer, Reddish, Skewes, Geertz, Roepstorff and Bulbulia2013). Early examples come from evidence for initiation rituals in Upper Paleolithic deep cave sites involving altered states of consciousness and manipulation of the visual and auditory affordances of underground labyrinths (Rossano Reference Rossano2010). In the early stages of the Neolithic too, rare and emotionally intense rituals probably occurred, for example, in the form of hunting ancestor cults at sites like Cataloyuk, where pictorial art suggests the sporadic performance of dramatic feasting events involving the baiting of large and dangerous animals (Whitehouse & Hodder Reference Whitehouse, Hodder and Hodder2010; Whitehouse et al. Reference Whitehouse, Mazzucato, Hodder, Atkinson and Hodder2014a). Emotionally intense rituals, fusing participants into small relational groups, may also have been necessary for ancient foragers coping with natural hazards (Rossano Reference Rossano2010, p. 91). The association between rare, emotionally intense ritual ordeals, on the one hand, and social cohesion and self-sacrifice for the group, on the other hand, is even more readily apparent in the much richer data sets relating to contemporary small-scale societies. For example, based on an analysis of ethnographic data stored in the standard cross-cultural sample, it has been argued that the severity of traumatic initiation rites is positively correlated with warfare intensity (Sosis et al. Reference Sosis, Kress and Boster2007).

The pain and fear evinced by some collective rituals documented by anthropologists are so extreme that they have been described as “rites of terror” (Whitehouse Reference Whitehouse and Corrigan1996). Procedures entailed in such rituals bear comparison with techniques of torture entailing beatings, whippings, mutilation, burning, and evulsion of the fingernails (Barth Reference Barth1987; Strehlow Reference Strehlow1947). Pain is often inflicted on especially sensitive regions of the body, such as the genitals (Barth Reference Barth1975). In addition to direct assault, such rituals commonly inflict suffering by depriving participants of rest, sleep, food, warmth, light, social contact, and other basic needs, often for extended periods (Allen Reference Allen1967). One of the effects of agonizing ritual ordeals is that they bond participants together so tightly that they will stop at nothing to defend each other and their fellow tribesmen from external threats. Often these extreme rituals are associated with warrior cults and may be accompanied by oaths to defend the group and commitments of unwavering loyalty (Weisfeld Reference Weisfeld1979).

Rites of terror have been documented all around the world, and evidence of their existence is present throughout human history and prehistory, so the need for a general explanation has long been recognized. Inspired by Festinger's (Reference Festinger1957) theory of cognitive dissonance, Aronson and Mills (Reference Aronson and Mills1959) proposed that the endurance of painful initiations into the group is inconsistent with disliking the group. Consequently, initiates convince themselves that the group is worthy of their loyalty and affection. Others have suggested that participation in painful rituals serves as a costly signal of commitment to the group, thereby promoting trust and pro-sociality among group members (Bulbulia et al. Reference Bulbulia, Geertz, Atkinson, Cohen, Evans, Francois, Gintis, Gray, Henrich, Jordon, Norenzayan, Richerson, Slingerland, Turchin, Whitehouse, Widlok, Wilson, Richerson and Christiansen2013; Sosis Reference Sosis2003). A drawback with both dissonance and costly signalling explanations of rites of terror, however, is they assume that participation is voluntary. Although that may be true in some cases, very often it is not. Failure to submit to the ritual tortures typically carries heavy penalties, ranging from social exclusion to execution (Cimino Reference Cimino2011). Moreover, these theories often assume that the ritual ordeals are used to mark entry into groups, and often this is not the case; they may be performed by already established groups for a wide range of stated purposes that have little or nothing to do with initiation or the conferment of membership (Atkinson & Whitehouse Reference Atkinson and Whitehouse2011; Whitehouse Reference Whitehouse and Corrigan1996).

An alternative perspective is provided by the theory of “imagistic practices” (Whitehouse Reference Whitehouse2000), which is consistent with the “shared experience” pathway to fusion discussed previously (and which it largely inspired). Unlike cognitive dissonance and costly signalling approaches, the theory of imagistic practices applies to a wider range of intensely emotional rituals and not only voluntary initiations (Whitehouse Reference Whitehouse2004). An early account of imagistic bonding focused on a single case study: a cargo cult in the rainforest of New Britain, Papua New Guinea, known as the Pomio Kivung (Whitehouse Reference Whitehouse1995). Observations and interviews in the field suggested that emotionally intense rituals, particularly those involving negative valence (such as pain, fear, and disgust), produced vivid and enduring memories (Whitehouse Reference Whitehouse, Pachis and Wiebe1992). In view of the causally opaque, seemingly arbitrary nature of ritualized behaviour, episodic memories for cult rituals prompted efforts to interpret the meaning and significance of what had occurred, a process that has been dubbed “spontaneous exegetical reflection” (Whitehouse Reference Whitehouse2001; Reference Whitehouse2004). The sharing of traumatic ordeals and subsequent process of reflection on their significance and value seemed to contribute to group bonding among splinter group members (Whitehouse Reference Whitehouse1995).

Based on all of the real-world research conducted so far, both quantitative and qualitative, it appears that imagistic rituals produce social cohesion through the sharing of exceptionally thought-provoking and life-shaping experiences encoded in episodic memory (Whitehouse Reference Whitehouse, Pachis and Wiebe1992; Reference Whitehouse1995). Because these kinds of memories specify who else was present at the time, the groups they generate have rigid boundaries – members cannot be added if they are not part of the recalled episode nor can anyone who participated be excised from memory. As such, imagistic practices are associated with relatively fixed, small, face-to-face groups, an observation based on both ethnographic data (e.g., Whitehouse & Laidlaw Reference Whitehouse and Laidlaw2004) and historical materials (e.g., Whitehouse & Hodder Reference Whitehouse, Hodder and Hodder2010; Whitehouse & Martin Reference Whitehouse and Martin2005). Fine-grained qualitative case studies have provided useful insights into the nexus of relations between ritual, emotion, memory, reflection, group cohesion, and pro-social commitment, including willingness to fight and die for the group under conditions of intergroup conflict and warfare (Whitehouse & McQuinn Reference Whitehouse, McQuinn, Juergensmeyer, Kitts and Jerryson2012). This qualitative information has been gradually augmented by quantitative correlational studies of large data sets (Atkinson & Whitehouse Reference Atkinson and Whitehouse2011; Gantley et al. Reference Gantley, Bogaard and Whitehouse2018; Reference Gantley, Whitehouse, Bogaard, Watkins, Beuger and Clarein press; Whitehouse et al. Reference Whitehouse2013). Research on these topics has also led to the creation of by far the most ambitious longitudinal data set on ritual and social complexity ever constructed, known as Seshat: Global History Databank. This vast storehouse of historical data is now making it possible to explore the relationship between shared emotion, ritual frequency, group size and structure, warfare, and agricultural intensity over thousands of years and on a global scale (Currie et al. Reference Currie, Bogaard, Cesaretti, Edwards, François, Holden, Hoyer, Korotayev, Manning, Garcia, Oyebamiji, Petrie, Turchin, Whitehouse and Williams2015; Turchin et al. Reference Turchin, Whitehouse, Francois, Slingerland and Collard2012; Reference Turchin, Brennan, Currie, Feeney, François, Hoyer, Manning, Marciniak, Mullins, Palmisano, Peregrine, Turner and Whitehouse2015; Reference Turchin, Currie, Whitehouse, François, Feeney, Mullins, Hoyer, Collins, Grohmann, Savage, Mendel-Gleason, Turner, Dupeyron, Cioni, Reddish, Levine, Jordan, Brandl, Williams, Cesaretti, Krueger, Cecceralli, Figliulo-Rosswurm, Tuan, Peregrine, Marciniak, Preiser-Kapeller, Kradin, Korotayev, Palmisano, Baker, Bidmead, Bol, Christian, Cook, Covey, Feinman, Júlíusson, Kristinsson, Miksic, Mostern, Petrie, Rudiak-Gould, ter Haar, Wallace, Mair, Xie, Baines, Bridges, Manning, Lockhart, Bogaard and Spencer2018; Whitehouse et al. Reference Whitehouse, François and Turchin2015).

Central to the theory of imagistic practices is the idea that rare or unique emotionally intense experiences give rise to vivid, long-lasting memories for episodic details (such as what happened, how it felt, who else was present). These memories in turn become a locus for subsequent reflection, infusing the episodes recalled with meaning and salience (Barth Reference Barth1975; Reference Barth1987; Whitehouse Reference Whitehouse, Pachis and Wiebe1992; Reference Whitehouse, Light and Wilson2003). If experiences of suffering and trauma are especially memorable and thought provoking, this should be all the more so in the case of strongly valenced rituals, which, like all rituals, are by definition “causally opaque,” that is, actions for which nobody would expect there to be a rational physical-causal explanation (Whitehouse Reference Whitehouse2011). Ritual procedures are the way they are because that is the conventional way of performing them. This, in turn, makes them interpretable in a wide range of ways (Humphrey & Laidlaw Reference Humphrey and Laidlaw1994; Whitehouse Reference Whitehouse2004). Emotionally charged rituals, because they are both causally opaque and remembered long afterwards, prompt a protracted search for meaning, a feature frequently observed in mystery cults and other esoteric religious traditions (Chinnery & Haddon Reference Chinnery and Haddon1917; Martin & Pachis Reference Martin and Pachis2009; Whitehouse Reference Whitehouse, Pachis and Wiebe1992; Reference Whitehouse2001; Reference Whitehouse2002; Williams Reference Williams1928). Undergoing especially salient, symbolically charged rituals, as well as the process of revelatory meaning-making ensuing from them, is at once a very personal experience but also one that is shared with the group. For participants, these rituals are not only self-shaping; they are also group defining. In part, this is because the memories for such experiences are unique and unrepeatable, specifying who else was present. Groups formed in this way have somewhat rigid boundaries – those who were not present cannot be inserted into one's memories after the fact, nor can anybody who has been through the ordeals be excluded subsequently (Whitehouse Reference Whitehouse2004).

Early efforts to investigate these processes in carefully controlled experiments used artificial rituals with sufficient emotional intensity to prompt both episodic recall and subsequent reflection on their meaning. In one such study, participants were invited to participate in what they were told was a reconstruction of an ancient Amazonian fertility rite (Richert et al. Reference Richert, Whitehouse, Stewart, Whitehouse and McCauley2005). According to the cover story, anthropologists were seeking to puzzle together the long-forgotten meanings of the ritual acts and artefacts and were eager to learn from the impressions and interpretations of participants in the reconstruction. As part of the ritual sequence, participants were invited to place their hands in a hole in the ground while a drum played in the background. In this between-subjects design, the actions were varied such that participants in a high-arousal condition wore blindfolds when reaching into the hole. This made the experience considerably more frightening, as evidenced by self-reports of emotional intensity gathered immediately afterwards. All participants were asked to describe their ideas about the meaning of the ritual a week later, allowing time for reflection in the intervening period. Those experiencing the ritual as more frightening scored significantly higher on the meaning-making measures, suggesting that dysphoric intensity is linked to spontaneous reflection on the symbolism of ritual actions. These findings have been further supported by experiments using more objective measures of physiological arousal (Richert et al. Reference Richert, Whitehouse, Stewart, Whitehouse and McCauley2005) and other methods of inducing dysphoria, such as the trauma film paradigm (Russell et al. Reference Russell, Gobet and Whitehouse2014). In addition to experiments using made-up rituals, studies of real-world rituals suggest that the more frightening or painful the ritual experience, the more exegetical is the reflection it provokes (Xygalatas Reference Xygalatas2007). Moreover, recent studies of hazing rituals have shown that elevated exegetical reflection on the meanings of collective rituals mediates the relationship between dysphoric intensity, identity fusion, and willingness to sacrifice oneself for the group (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017).

Although modern-day hazing rituals are usually illegal and therefore difficult to study directly, especially in high-security environments such as the military, it is possible to conduct research on the phenomenon in certain contact sports such as Brazilian Jiu-Jitsu (BJJ). BJJ is a martial art based on principles derived from Judo. BJJ practitioners must progress through a system of grades associated with distinctive belts. In some BJJ schools, promotion to a higher grade entails a form of hazing via agonizing belt whippings, producing severe welts and bruises to the back and shoulders. A recent study of 564 BJJ practitioners showed that those who had suffered the most intense whippings reported higher levels of fusion to their school and also expressed greater stated willingness to risk their lives fighting for the club (Kavanagh et al. Reference Kavanagh, Jong, McKay and Whitehouse2018; Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017). In a related study, 146 members of American college fraternities and sororities were asked about their experience with hazing or other initiatory ordeals. The more central such ritual experiences were to the participants’ personal identities, the higher were their reported levels of identity fusion with their fraternity or sorority and the more willing they were to sacrifice themselves for the sake of the group (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017).

In view of the above, it should come as no surprise that a recent exhaustive survey of modern suicide terrorism identified bonding via intense initiation rituals as one of the features common to most suicide groups (Pape Reference Pape2005, p. 8). But whereas imagistic practices in the small-scale societies of prehistory would have supported efforts to fend off other groups of similar size, wielding more or less equivalent levels of lethal force, such practices took on a whole new significance with the emergence and spread of states and empires. Jewish Zealots who set out to assassinate Herod during the Roman occupation of Judea or Ismaili Assassins undertaking suicide missions in eleventh- and twelfth-century Persia, Syria, and Lebanon willingly gave their lives for the sake of a group, but they did so in a highly asymmetric theatre of operations, where the outgroup was more or less certain to overpower them as individual actors (Gambetta Reference Gambetta and Gambetta2005). Through their martyrdom, they intended to instil fear in the powerholders and motivate others to rise up in support of their cause. The same logic, albeit without the element of murderous intent, motivated the hunger strikes of Mahatma Gandhi in 1940s India and the self-immolations of Buddhist monks and nuns in 1960s South Vietnam, both of which resulted in massive movements of popular support. In the same way, suicide terrorists of recent decades have harnessed the motivating power of fusion, whether generated through naturally occurring shared experiences of oppression or artificially induced ritual ordeals (or, more likely, both), to mobilize would-be martyrs to strategic effect (Pape Reference Pape2005).

Not only in the specific case of terrorist cells but also in other kinds of interest groups, embedded in religions, professional guilds, and even schools and universities, imagistic practices have posed a threat to centralized states, empires, and priesthoods and have historically played a prominent role in many civil wars and revolutions as well as religious splintering and reformations (Whitehouse Reference Whitehouse2004; Whitehouse & Martin Reference Whitehouse and Martin2004). Efforts throughout history to suppress, contain, or wipe out imagistic organizations have met with varying success (Whitehouse & Martin Reference Whitehouse and Martin2005). The same may be said of America's global “war on terror,” which in many cases amounts to an effort to eliminate unauthorized imagistic cells, especially those with a revolutionary vision to establish alternative states or empires. But ruling elites have also opportunistically harnessed the imagistic mode by endorsing rituals that foster local fusion among elite groups and crucially also in the military. And this continues today, for example, in the form of secretive hazing practices in the institutions through which ruling classes are recruited, in the fraternities of elite universities and masonic lodges, and in the armed forces. These practices give rise to fusion in much the same way as imagistic rituals have done for many millennia in small-scale societies and, today, in a wide range of non–state-armed groups (Whitehouse & McQuinn Reference Whitehouse, McQuinn, Juergensmeyer, Kitts and Jerryson2012).

6. Extended fusion, identification, and self-sacrifice

So far, we have been focusing on fusion within a local group. The members of such groups fuse as a result of undergoing transformative, self-defining experiences together or sharing biological traits. As such, the bonds of fusion are based on relational ties among particular individuals: family members, fellow fighters in a military unit, co-participants in a ritual, and so on. But there is also evidence that people can fuse with much larger group categories, such as country, ethnic group, and world religion. This has been described as “extended fusion” (Swann et al. Reference Swann, Jetten, Gómez, Whitehouse and Bastian2012; Whitehouse Reference Whitehouse2013).

The notion of extended fusion raises a number of conceptual challenges. One of these is the thorny question of how it relates to the more extensively studied phenomenon of “identification,” which is essentially a measure of the strength of one's commitment to a social category (Tajfel & Turner Reference Tajfel, Turner, Austin and Worchel1979). Another is how extended fusion might fit with the theory of “tribal social instincts” (Richerson & Henrich Reference Richerson and Henrich2012), which includes an evolutionarily grounded account of how ethnic markers arise and spread (Boyd & Richerson Reference Boyd and Richerson1987). Identification is a depersonalizing form of group alignment in which group members perceive themselves to be interchangeable (Swann et al. Reference Swann, Gómez, Seyle, Morales and Huici2009) because they are merely the bearers of prototypical traits that have been socially acquired from others (Whitehouse & Lanman Reference Whitehouse and Lanman2014). By contrast, fused individuals regard their group identities as grounded in personal experience (Whitehouse Reference Whitehouse2013), producing a “strong autonomous self” that is “merged with the group” and, therefore, capable of motivating extreme pro-group action in non-prototypical ways (Swann et al. Reference Swann, Gómez, Seyle, Morales and Huici2009; Reference Swann, Buhrmester, Gómez, Jetten, Bastian, Vázquez, Ariyanto, Besta, Christ, Cui, Finchilescu, González, Goto, Homsey, Sharma, Susianto and Zhang2014a; Whitehouse & McQuinn Reference Whitehouse, McQuinn, Juergensmeyer, Kitts and Jerryson2012). In contrast, identification with a group motivates self-sacrificial behaviour only to the extent that it is endorsed by the group and that one's personal self does not become salient and trigger self-preservation motives that conflict with group values or interests (Whitehouse Reference Whitehouse2013). The first empirical demonstration of the fundamental differences between identification and fusion was a study in which participants were asked first about their willingness to defend themselves personally when threatened (priming personal identity) and then about their willingness to defend their group when threatened (priming group identity). Compared with a control condition that did not prime personal identity, the personal identity prime increased willingness to defend the group at cost to self in strongly fused individuals. But the personal identity prime had no effect on willingness to defend the group among those who were highly identified but not highly fused with the group (Swann et al. Reference Swann, Gómez, Seyle, Morales and Huici2009). Subsequent studies further support the view that identification and fusion are fundamentally distinct forms of group alignment (Gómez et al. Reference Gómez, Morales, Hart, Vázquez and Swann2011b; Swann et al. Reference Swann, Bastian, Gómez, Jensen and Whitehouse2012).

Nevertheless, extended fusion, like identification, entails alignment with group categories, rather than with a network of local, relational ties. Some of these group categories specify enormous populations – far too large for their members to know each other personally. Such groups may be described as “imagined communities” (Anderson Reference Anderson1983) in the sense that one cannot actually perceive one's fellow members directly; instead, one can only imagine the coexistence of others in the group that share the same identity markers. It has been argued that semantic memory for shared beliefs and procedural memory for shared practices both play a crucial role in the formation of imagined communities (Whitehouse Reference Whitehouse2004). Such memories take the form of schemas and scripts in which the slots for actions, actors, patients, and instruments are not populated by particular individuals and artefacts, but are generic representations of prototypical ones (Whitehouse Reference Whitehouse, Milton and Svasek2005). Semantic memory provides us with most of the knowledge required to be a competent member of a large social group, such as a nation or a world religion, but it is a very impersonal kind of knowledge, in which the relevant agents are faceless bearers of social roles. Such memories are not anchored in personal experience; in fact, they are acquired through social learning from others in ways that are rapidly detached from any single episode in time or space (Whitehouse & Lanman Reference Whitehouse and Lanman2014).

Unlike fusion, which taps directly into personal experience, identification is “deindividuating” (Diener et al. Reference Diener, Lusk, DeFour and Flax1980). Activation of social identities makes personal experience (and, therefore, the personal self) less salient – it is as if one loses oneself in the crowd. Because identification does not tap into personal agency in the same way as fusion, we should expect identification to be a comparatively weaker basis for group cohesion (Whitehouse Reference Whitehouse2013; Whitehouse & Lanman Reference Whitehouse and Lanman2014). True, identification motivates many forms of cooperation in society, including submission to higher authority, following and enforcing norms, participating in democratic institutions, and dutifully paying tax or tribute. It may even lead to heartfelt sympathy for those who lay down their lives for the group. But sympathy for the sacrifices of other group members, including the actions of suicide bombers, is not the same as being willing to undertake such actions oneself. Because identification is not a sufficiently powerful social glue to overcome selfish drives and impulses, penal systems are often required to sanction selfishness. This is particularly noticeable where conflicts of interest between individual and group are most extreme and the temptations to defect are high. A case in point would be participation in the military during wartime. But although shooting deserters and punishing criminality in general may have a deterrent effect, these are not the most effective ways of inspiring commitment to the group. Military leaders have long appreciated that combatants motivated by fear of punishment are far less effective in battle than those motivated by love of the group, of the kind that only fusion can produce. This may be one reason why terrorist and guerrilla forces, even when greatly outnumbered, can present such a stubbornly persistent threat to states with only conventional armies at their disposal.

Although identification with large group categories may be a relatively weak motivator, at least when it conflicts with personal self-interest, it is reasonable to ask whether extended fusion fares any better. Extended fusion is thought to entail the extension of bonds of kinship to larger groups (Whitehouse Reference Whitehouse2013), such that the metaphor of brotherhood triggers similar emotional responses, via norm internalization, as genetic relatedness (Richerson & Henrich Reference Richerson and Henrich2012, pp. 62–63). This might be conceptualized as a process of “projection” (Swann et al. Reference Swann, Bastian, Gómez, Jensen and Whitehouse2012) whereby bonds forged in small groups, such as the family, come to permeate our feelings about larger groups, such as church or nation. Consistent with this view is the pervasive use of metaphors of kinship when talking about country (e.g., as motherland, fatherland), fellowship (e.g., as brotherhood), and ethnicity (e.g., emphasizing common genealogical roots). Nevertheless, the bonds forged in small relational groups, such as families, are often rooted in memories quite different from those uniting larger social categories. This is very apparent in the way different kinds of rituals bond participants.

As we have seen, imagistic rituals rely heavily on shared episodic memories deriving from rare, emotionally intense, and personally consequential events, such as initiation and frontline combat (see above). Relying as it does on self-defining episodic memory, fusion taps into the agentic personal self at the same moment it activates social identities (Swann et al. Reference Swann, Bastian, Gómez, Jensen and Whitehouse2012). To the extent that these kinds of episodic memories may be associated with much larger group categories, we can fuse with “imagined communities.” But note that this process of fusion, if correctly conceptualized, must be grounded in concrete personal experience, and consequently, it must be “local” before it can be “extended.”

Although fusion with country has been shown to correlate highly with stated willingness to fight and die to protect one's fellow countrymen (Swann et al. Reference Swann, Gómez, Huici, Morales and Hixon2010b), it remains doubtful whether anyone would prefer to die for an extended fusion target over a local one. The fact that the beliefs and practices defining large groups are acquired from others (Whitehouse Reference Whitehouse2004, pp. 123–24), rather than arising from internal processes of reflection and individual learning, may be one reason why identification fails to tap into personal agency, as noted above. Consequently, even if collective beliefs and practices can be “personalized” via the projection of local fusion onto an extended group, such a process may dilute the authenticity and uniqueness of the episodic memories upon which the fusion of personal and group identities depends. For example, the Christian evangelist might have experienced uniquely episodic and personal revelations even though her conversion narrative is at the same time heavily shaped by socially learned and often quite highly standardized cultural schemas. To the extent that the experience of being “born again,” for example, can be shared with others in one's group, it might really be just the socially learned semantic schemas that are common to conversion rather than the self-defining experiences of converts.

There is also the question of whether sharing life-shaping experiences firsthand produces stronger fusion and associated behavioral consequences than would be the case if the evidence for sharedness is indirect. Tribal initiates or frontline fighters, for example, can actually recall who else was there during the most salient rituals or battles, suffering by one's side. It is possible that this kind of memory fosters the strongest fusion and motivates the most extreme pro-social actions in defending other members of the group. At the other end of the spectrum might be bonds based only on indirect evidence of shared experience, such as wearing the same medals or other insignia. Viewed in this light, war wounds or scars of initiation might serve as evidence of intermediate reliability, a compelling testimony to common suffering but without the episodic “time travel” quality of remembering particular group members actually being present during shared ordeals. In the study of Libyan revolutionaries reported above, participants expressed ceiling levels of fusion both with members of their own battalions (local fusion) and with those they had not met from other battalions who fought bravely (extended fusion), but on a forced-choice question they overwhelmingly chose their relational network over any extended group (Whitehouse et al. Reference Whitehouse, McQuinn, Buhrmester and Swann2014b). Although it is not yet known whether differences in the strength of local and extended fusion are due to the directness of evidence for shared experience or some other factor, it is quite possible that only local fusion is capable of motivating extreme self-sacrifice. All of these topics should be explored more systematically in future research.

7. Conclusions and next steps

Why die for a group? This paper integrates core insights from the literature on suicide terrorism into a novel theory in which identity fusion, combined with perceptions of out-group threat, motivates extreme self-sacrifice. It is argued here that fusion is caused by perceptions of shared essence, whether as a result of shared biology, shared experience, or both (although there may be additional, as yet unknown, factors that give rise to fusion). This theoretical framework results from a synthesis of several decades of research on religious groups, together with a wide range of more recent studies measuring willingness to fight and die for the group in special populations including football fans, martial arts clubs, Islamic fundamentalists, and other highly cohesive organizations, as well as data from groups whose members actually laid down their lives for each other on the battlefield, including non–state-armed groups in Libya and conventional forces serving in Iraq and Afghanistan.

Several theories of the evolutionary origins of the fusion mechanism are considered here. One proposes that fusion is the outcome of kin selection, motivating high levels of cooperation and mutual support among close genetic relatives (Whitehouse & Lanman Reference Whitehouse and Lanman2014). Another proposes that conditioning cooperation on past experience is sufficient to fuse groups of distantly related individuals in the face of adversity (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017). These theories are not mutually exclusive and could both help explain the biological evolution of the fusion mechanism under natural selection. In much of human prehistory, fused groups probably comprised small warring bands bound together in adversity. However, in contemporary complex societies, fused groups are often embedded in much larger organizations, such as armies, religious sects, and terrorist organizations. In many cases, fusion results not only from ordeals triggered by external factors, such as enemy attacks and natural disasters, but also through culturally evolved cohesion gadgets such as traumatic initiations. So-called “imagistic practices” of this kind are found not only in small-scale societies traditionally at war, but also in modern military units and terrorist cells, including those using suicide attacks as a strategic weapon.

The theory of extreme self-sacrifice proposed in this paper is falsifiable, hinging as it does on the following testable hypotheses regarding the psychological causes and behavioural consequences of identity fusion (see Fig. 1): Perceptions of shared essence lead to local fusion; perceptions of shared essence are outcomes of at least two distinct processes (experiencing emotionally intense events with others and/or believing that one shares inherited biological traits); local fusion motivates psychological kinship and self-sacrifice for the group. A number of similarly testable subsidiary hypotheses have also been presented regarding the causal pathways linking shared emotional events to fusion and self-sacrifice. For example, it is proposed that episodic memories for shared events are “group defining” to the extent that they also prompt reflection on the meanings of those experienced events. Causally opaque events are hypothesized to generate more reflection than causally transparent ones. It is also proposed that fusion and psychological kinship motivate violent self-sacrifice only when a plausible out-group threat of sufficient magnitude is present. Evidence has been presented in support of each of these hypotheses, but there remains a need for wider independent replications to validate existing findings.

The pathways to fusion and self-sacrifice proposed here could turn out to be mistaken in some of their details without being completely wrong. What would be fatal for the theory is if it turned out that convictions of shared essence failed to predict high fusion scores, or if fusion (plus out-group threat) were shown to be a poor predictor of actual (as opposed to declared) willingness to fight and die for the group. These two claims are so central to the conceptual framework that, if shown to be false, the entire edifice would collapse. Somewhat less disastrous for the theory, but still a setback, would be a significant reduction in its explanatory provenance. For example, the theory may eventually prove to be applicable only to some armed groups but not all, and perhaps most crucially not to suicide terrorists. Although there would seem to be many similarities between the self-sacrificial acts of armed militia (whose fusion levels with numerous target groups have been measured) and those of suicide bombers (whose fusion levels are unknown), these similarities may turn out to be more apparent than real. If, as some have argued (see sect. 2), most suicide terrorists are motivated by pathology (e.g., suicidal depression) rather than the desire to act in the interests of a group, that would be a serious problem for the theory as articulated in this paper. Decisive evidence on this question may require more extensive research among would-be suicide terrorists and those who have attempted unsuccessfully to carry out such attacks (the previously acknowledged difficulties of conducting such studies notwithstanding).

The theory presented here also raises many new, empirically tractable questions, for example, concerning the relationship between local and extended fusion. Future research should investigate whether perceptions of shared essence are stronger if they are based on direct observation rather than on the testimony of others. Would remembering who else was there alongside you in a decisive battle or a traumatic rite of passage or perceiving shared phenotypic traits in a sibling provide more compelling evidence of shared experience or shared biology than merely displaying the same kind of medals or reciting myths of shared ancestry? Relational ties to a local group often incorporate episodic memories for self-defining events, which other group members indelibly inhabit. By contrast, categorical ties to an extended group are based largely on “knowing that” certain identity markers serve as indirect testimony to shared experience. Indirect evidence of shared experience may not be capable of motivating acts of self-sacrifice to the same extent as bonds forged through episodic memories of shared ordeals within a band of brothers.

Research into the causes of extreme pro-group action is not merely of scientific interest; there is potential also to use the findings in practical ways. For example, deradicalising Islamist militants might be reframed as a process of defusing extremists. Given that we now have a well-substantiated account of the causal pathways to fusion, together with evidence that priming the mediating variables in this pathway increases fusion (Whitehouse et al. Reference Whitehouse, Jong, Buhrmester, Gomez, Bastian, Kavanagh, Newson, Matthews, Lanman, McKay and Gavrilets2017), it may be possible to reduce the effects of mediating variables so as to obstruct or reverse the fusion process. This has yet to be demonstrated in practice, but the general approach is well motivated theoretically. Such an approach should not be confused with the notion of “deprogramming” because the goal would not be to alter people's beliefs or goals against their will. Indeed, the aim would not be to challenge the validity of ideologies or doctrines at all, but only to facilitate a process of reflection on past experiences and their relevance to group alignments. The process would need to engage the wider participation not only of extremists but also of members of their social networks and surrounding communities (such as parents, schoolteachers, and religious leaders), although the ethics of any interventions would require careful scrutiny and monitoring.

Yet another potential application of this new framework would be neither to create nor to obstruct group alignments but to harness existing ones. There are a number of potentially desirable ways in which this could be done, not least in rebuilding societies devastated by conflicts or natural disasters. For example, during the uprising of 2011, many Libyans fought passionately and at huge cost to clear the way for a prosperous future under a more consensual system of governance. The social cohesion needed to build that vision was available in abundance at the end of the revolution, but there was a failure to harness it for the public good both on the part of the international community and on the part of Libyan leaders vying for power at the time. The same pattern repeats itself endlessly in other conflicts around the world. Only by better understanding the underlying causes of pro-group commitment can we benefit from its potential for building trust and cooperation while limiting its capacity to stoke intergroup conflict.