In the last 200 years, the half million Iban living on Borneo's northwest region have undergone a remarkable transformation. When first encountered by colonizers in the 19th century, Iban lived in communal long-houses of 100–200 people and made a living from farming rice and hunting (Freeman, 1970). According to their festivals and mythology, Iban worked toward a community that was harmonious, rich in rice, flush with children, and endowed with an abundance of spiritual energy (Jensen, 1974; Heppell et al., 2005). A key way of fostering such flourishing communities was the taking of human heads—to cure a member of one's group or to rescue a member's soul from limbo or from spiritual slavery in another region (Klokke, 2004). It is important to note here that indiscriminate killing was not acceptable among the Iban. Tribal groupings were defined in part as those people who did not take each other's heads. Killing a fellow group member was considered a major transgression on the order of incest. It could upset the universal order and could lead to sterility in terms of offspring and rice production as well as the future taking of heads (Freeman, 1970; Jensen, 1974; Sutlive, 1992).

Fast forward to today. After the forceful imposition of colonial and state laws banning head-hunting, the practice is effectively dead, and only a few elderly men still wield the hand tattoo used to mark a successful headhunter (Freeman, 1970; Laukien, 2005). Iban engage in far-flung wage labor opportunities alongside members of other ethnic groups with which they have prior histories of war (Lumenta, 2003). They seek formal education, consume Malaysian mass media, and many have converted to dominant world religions, including Christianity and Islam. Many Iban now also identify as citizens of Malaysia in addition to being Iban (Lumenta, 2003; Postill, 2006). At times, violence reminiscent of earlier times flares up (BBC News, 2001), but after two centuries, most Iban have a very different way of defining insiders and outsiders and very different views about appropriate social behavior with other groups.

The Iban transformation illustrates three points. First, the ways that people behave toward others can depend heavily on how those others are classified—as kin, friends, and community members or outsiders, strangers and foreigners. Second, human populations can vary dramatically in: (1) how they define closeness and distance of a social partner and (2) how these qualities of a partner influence social behavior. Third, these population differences are not fixed or static. Populations can change quite dramatically within several generations, in this case, from hunting the heads of neighboring groups to participating relatively peacefully in a much larger nation-state and world system.

How people socially and psychologically construct boundaries between insiders and outsiders or plot gradients of social distance and how these models of boundaries and distance shape behavior toward others are critical questions for a number of fields. Current models for the evolution of human social behavior, and of large-scale cooperation specifically, rely on the construction of groups that can contain the fruits of cooperation, exclude outsiders, and compete with other groups (Boyd et al., 2003; Choi and Bowles, 2007). Paradoxically, the same tribal instincts that may have fostered the human capacity for large-scale cooperation today pose problems for building peaceful and just societies at ever larger scales (Bernhard et al., 2006; Richerson and Henrich, 2012). They also underlay many currently recognized problems in today's world, including favoritism, racial and ethnic discrimination, armed ethnic conflict, and genocide (Levine and Campbell, 1972).

In the past decade, researchers have proposed a number of theories to account for these population differences in parochialism and to explain historical changes like those observed among Iban. However, these diverse approaches are relatively scattered across the social and behavioral sciences, they encompass a wide range of motivations and behaviors under the broad rubrics of in-group favoritism, ethnocentrism, xenophobia, and parochial altruism, and these different theories rarely come into contact in the same paper or analysis. In this paper, we clarify the diverse ways that scholars have operationalized parochialism, we outline and synthesize current hypotheses for cross-population variation in parochialism, and we discuss key methodological challenges in assessing these diverse economic and evolutionary hypotheses.

Humans do not have a general tendency to help, protect, or harm others. Rather, these behaviors are conditioned by many contextual factors (Bekkers and Wiepking, 2011), including the perceived need of the recipient (Taormina and Messick, 1983; Engel, 2011), the legitimacy of the request for help (Bickman and Kamzan, 1973), the degree to which someone deserves harm or help (Skitka and Tetlock, 1992), genetic relatedness or kinship with a person (Rachlin and Jones, 2008; Alvard, 2009), and whether the individual or group are perceived to pose a threat (Semyonov et al., 2004). The degree to which an actor feels socially close to another individual also reliably guides social behavior, whether social closeness is determined by subjective assessments of a spatial metaphor (e.g., closeness or insideness) or by common membership in a group (Leider et al., 2009; Goeree et al., 2010; Mathew and Boyd, 2011; Branas-Garza et al., 2012). Here, we refer to the broad tendency to rely on cues of social closeness in guiding behavior as parochialism, a concept which encompasses a number of related concepts including xenophobia, ethnocentrism, and parochial altruism.

The social and behavioral sciences have a long tradition of studying the proximate mechanisms by which social closeness and group membership influence behavior toward others and how groups emerge in experimental settings (Sherif, 1961; Tajfel et al., 1971; Brewer, 1979; Glaeser et al., 2000; Hewstone et al., 2002; Dovidio et al., 2005; Goette et al., 2006). All of these approaches are united in studying how our decisions to help, protect or harm someone are shaped by perceptions of social closeness. However, these approaches also differ in two key respects: (1) in how social closeness is operationalized, and (2) in what behaviors, preferences and motivations are considered. We review these differences here.

In the last two decades, psychologists and economists have begun to identify key cognitive and neurobiological mechanisms underlying parochialism, including perceptions of threat (Reik et al., 2006) and the role of oxytocin and brain circuits in modulating behavior toward in- and out-group members (De Dreu et al., 2010; Baumgartner et al., 2011; De Dreu, 2012). Researchers have also identified specific kinds of activities which can increase social closeness to others, including focused conversations (Aron et al., 1997), synchronized movement (Vacharkulksemsuk and Fredrickson, 2012), and synchronized multisensory inputs (Paladino et al., 2010). Moreover, it appears that the capacity and propensity to differentiate social groups arises early in development (Kinzler et al., 2007). However, researchers have only recently begun to explore why these psychological capacities for parochialism are recruited differently in different human populations and across different cultural settings (Miller and Bersoff, 1998; Buchan et al., 2009; Gelfand, 2011; Van de Vliert, 2011; Fincher and Thornhill, 2012; Hruschka and Henrich, 2013; Hackman and Hruschka, 2013b).

There are several ways that populations differ in parochialism. First, what counts as a kin tie, a friendship, or an in-group and what counts as appropriate behaviors with different social partners is informed by local cultural categories and norms. For example, most populations in the US do not have a cultural category of blood brother, and so there is no clear set of norms or expectations applied to being in such a relationship (Hruschka, 2010). Second, the social techniques available to organize and maintain in-groups of varying sizes and scales constrain the kinds of in-groups to which people can belong. Mass media and formal schooling makes it much more likely that people can identify with groups as large as those encompassed by modern nation-states. World religions disseminate and enforce common languages, symbols, and rituals which can forge large populations into a single in-group (Atran and Henrich, 2010). These social techniques permit the creation of new in-groups that may have never been possible before. Third, the most salient in-group category can change quickly based on local practices and contexts. Among Enga horticulturalists in Papua New Guinea, rituals aimed at dehumanizing members of another group can swiftly recast allies as enemies (Wiessner, 2006), and among the Nuer of Sudan, changing patterns of competition over resources can re-align in-groups and out-groups (Evans-Pritchard, 1940). Finally, and most relevant to this article, given in-groups of similar scales, individuals from different populations differ remarkably in several crucial ways, including how much they trust and avoid outsiders and how much they favor friends, family, and community members (Fukuyama, 1995; Inglehart et al., 2006; Delhey et al., 2011; Hruschka and Henrich, 2013).

From the perspective of neurobiology, cross-population variability provides an opportunity to establish and distinguish those aspects of human brains and psychology that are reliably developing products of pan-human genes from those that depend on particular culturally-constructed niches (e.g., institutions) or ecological conditions. Grounded in culture-gene coevolutionary theory (McElreath et al., 2003; Henrich and Henrich, 2007), there is now substantial cross-population and developmental evidence suggesting that humans come equipped with cognitive abilities and psychological motivations to preferentially attend to, learn from, and interact with co-ethnics—individuals who share their markers for dress, dialect, language and bodily ornamentation. For example, infants and young children from diverse societies readily use dialect and dress to distinguish in-group members/coethnics (Kinzler et al., 2011, 2012; Mahajan and Wynn, 2012; Corriveau et al., 2013). On the basis of these rather sparse cues, infant and children preferentially learn from these individual (Shutts et al., 2013), seek interaction with them, and punish them for norm-violations (Schmidt et al., 2012). As expected from the theory, such ethnic cues can even trump racial differences in both young children (Kinzler et al., 2009; Corriveau et al., 2013) and sometimes in adults (Kurzban et al., 2001).

However, such reliably developing features of human cognition and motivation have to be understood in the light of two emerging lines of theory and evidence. First, growing up and ontogenetically adapting to very different environments means that different populations of humans have different brains and biologies, even when no genetic differences exist between populations (Henrich et al., 2010b,c). Within neuroscience, both training studies and cross-population research indicates that brains and bodies develop somewhat differently, in different environments, and yield distinct patterns of activation and hormonal responses to identical stimuli (Nisbett and Cohen, 1996; Kitayama et al., 2009; Woollett and Maguire, 2011). Second, mounting evidence indicates that cultural evolutionary processes have systematically shaped the physical and social (institutional) environments that developing humans face. This implies that cultural evolution has shaped our brains ontogenetically and over historical time (Henrich et al., 2012; Richerson and Henrich, 2012). For example, behavioral studies of children from ages 3 to 14 and adults across six diverse societies, ranging from Congo foragers to Westwood Los Angelenos, reveals the emergence of distinct developmental trajectories for social behavior in different places (House et al., 2013). This pattern is broadly consistent with the presence of market institutions in these societies. Several theories suggest that cultural evolution has harnessed and extended aspects of our innate parochialism in forming nations and religions.

These developments suggest that, rather than attempting to make potentially dubious inferences by generalizing from WEIRD undergraduates (Chiao and Cheon, 2010; Henrich et al., 2010c), neuroscientists need to develop collaborations that take advantage of both the existing theories discussed in this paper and then tap the now well-establish psychological diversity in our species.

Several theories have been proposed to account for cross-population differences and historical changes in parochialism. These theories vary along two major axes. First, they vary in the specific mechanisms by which individuals and populations change in response to their environment. Second, they vary in the specific ecological and social conditions which are posited to shape parochialism. We first review proposed mechanisms and then outline the different proposals for relevant environmental conditions, including market integration, religion, and environmental uncertainty.

In the last decade, the observation of substantial between-population differences in parochialism has inspired considerable theoretical work on the possible causes of these between-population differences. This is exciting progress, and this review describes a number of promising theories that may account for cross-population variation.

However, there are serious challenges in efforts to discriminate between these different hypotheses and to identify the specific mechanisms by which parochialism rises and falls in societies. Most studies have relied on observational cross-population designs, raising concerns about causality, identification of specific mechanisms, the direction of effects, and the time-scale of adaptation. Several strategies can provide some check against these issues.

The first task is to begin culling hypotheses through strategic model comparison rather than testing each hypothesis against a straw man null model. This involves identifying different predictions across models and then finding appropriate cross-population data which can discriminate between these predictions. For example, in a recent study of population-level parochialism across countries, Hruschka and Henrich directly compared the parasite stress hypothesis with the material insecurity hypothesis using novel checks against regional autocorrelation, new longitudinal data to assess reverse causation, and an instrumental variable to check for the effects of omitted variables. These results provided consistent support for the material insecurity hypothesis. It also challenged prior studies supporting the parasite stress hypothesis which had not included these methodological checks. In another paper, Hackman and Hruschka re-assessed analyses of US data which had previously found an association between pathogen stress and collectivism. With new data stratified by race, they showed the observed associations across states were due exclusively to substantial differences across US Whites and US Blacks. They also found support for an alternative hypothesis related to the material insecurity hypothesis (Hruschka and Henrich, 2013; Hackman and Hruschka, 2013a). Of course, such model comparison using observational data does not definitively show that the “winning” hypothesis is correct. However, it helps winnow the playing field.

Another important check can come from combining psychological experiments with cross-population studies in order to triangulate between potential psychological processes and the macro-scale correlates of cross-population variation. The findings of experiments alone may not scale up easily to account for cross-population differences, and cross-population correlations without grounding in established psychological mechanisms can easily be explained away as spurious associations. Integrating these two orders of data can ensure that hypotheses are consistent at both the individual and population level. A number of theories, including the market integration, religion, institutional quality, and pathogen stress hypotheses have begun to accrue data at both of these levels.

To mitigate some concerns about causality, mechanism, and directionality, the social sciences offer a number of tools that provide further checks on findings from cross-population observational data. Instrumental variable analysis commonly used in economics provides one additional check by identifying quasi-experimental assignments in observational data. For example, Acemoglu et al. used the mortality rates of early settlers in European colonies (1600–1875) as an instrumental variable which is expected to affect contemporary government effectiveness—an important variable in the material security hypotheses of parochialism. There is ample historical evidence that European colonizers avoided settling in places with high mortality rates, such as in the Belgian Congo (McNeill, 1977; Acemoglu et al., 2001), and instead of settling, they set up extractive systems. In situations of low mortality, on the other hand, colonizers settled in larger numbers and brought with them institutions, such as respect of private property, checks and balances in government, and equality of opportunity, which in turn fostered greater government effectiveness that persisted even after independence (Acemoglu et al., 2001). These measures of settler mortality act in some ways as quasi-experimental assignments of countries to different levels of government effectiveness, and Acemoglu et al. used this quasi-experimental assignment to examine the effect of government institutions on economic growth. More recently, Hruschka and Henrich have used the same reasoning to examine the effect of government institutions on parochialism (Hruschka and Henrich, 2013).

As access to longitudinal data increases with longer running cross-national surveys, it will also be possible to assess the temporal precedence and coincidence of different changes within populations (Inglehart et al., 2006; Hruschka and Henrich, 2013). For example, between 1925 and 2005, US samples have shown steadily decreasing avoidance of other ethnic groups in a number of domains—as in-laws, friends, neighbors, and fellow citizens (Bogardus, 1933; Parrillo and Donoghue, 2005). Long-term longitudinal data like this may provide insights into what factors most readily account for long-term changes in parochialism and how rapidly changes occur. Migration studies, originally developed in epidemiology, but now applied in economics, also show some promise in identifying the time-scale by which different aspects of parochialism change across generations who are put into novel contexts (Guiso et al., 2006; Fisman and Miguel, 2007; Giuliano and Alesina, 2010). For example, Giuliano and Alesina used such a design to show that second generation immigrants carry “cultural baggage” from their home country. Specifically, even after two generations, immigrants from countries with greater stated investment in family ties moved less and lived with their parents longer (Giuliano and Alesina, 2010).

Another approach is to look for natural experiments, as Bauer et al. did with their investigation of the effects of war on parochialism (Bauer et al., forthcoming). They looked around the globe for situations in which the effects of war on individuals, households, and communities were—at least plausibly—random with respect to individuals' own parochial motivations. Refugees and soldiers would be relatively easier to access compared to the approach they took, but both fleeing and being alive might be caused by their particular social motivations (therefore endogenous). As checks on the natural experiment assumption, they also (1) examined whether observables, like ethnicity or age, predicted experiencing war (they did not) and (2) performed their analyses just on those who were children at the time of the conflict (and thus have less control). These analyses support the idea that the experience of war was imposed exogenously, and thus provides a natural experiment.

Despite all of these possible checks and triangulations, observational data is still plagued by concerns about endogeneity and non-random assignment of cases which can threaten causal interpretations. Thus, once hypothesis are culled and honed through the above-mentioned techniques, a growing body of field experiments in economics, public health, and development currently used to understand health and development holds promise in assessing specific mechanisms by which economic, social, and environmental conditions inhibit or foster parochialism (Banerjee and Duflo, 2011). With this combination of model comparison, cross-level confirmation, statistical checks on temporal precedence and causality, and ultimately field experiments of different hypotheses, this exciting and crowded field of theories for parochialism will hopefully lead to a clearer understanding of the specific mechanisms and time scales by which population differences in parochialism emerge and sustain themselves.