Ingroup projection as a challenge of diversity: Consensus about and complexity of superordinate categories

A theoretical perspective that may help explain conflict between diverse groups is the ingroup projection model (Mummendey & Wenzel, 1999). Ingroup projection is an ethnocentric representation of a superordinate group that renders the ingroup relatively more prototypical (thus more normative, positive and deserving) than an outgroup believes is the case. Ingroup projection can lead to ingroup bias and prejudice against outgroups by portraying them as deviant. We review theory and research on the ingroup projection model, demonstrating its applicability to a wide range of intergroup contexts. We specifically discuss research on the multi-faceted cognitive, motivational, strategic and ideological underpinnings of ingroup projection, as well as threats resulting from beliefs that outgroups engages in such projection. We then focus on representational properties of superordinate identities that might limit ingroup projection, such as complexity or diversity, which however in themselves can become matters of contention between the involved groups. We argue that a precondition for positive intergroup relations is not only one of consensus between groups about their relative prototypicalities, which would foster harmony, but also acceptance of more equal prototypicality, which implies social change. Despite limitations and obstacles, superordinate identity complexity or diversity may hold the key for obtaining this goal.

As often we could begin such a chapter in a volume on intergroup conflict with a dire description of the state of human society and the continuing menace of social discrimination, prejudice, injustice, and ethnic violence. However, a children's book by the Austrian writer Edith Schreiber- Wicke (1990), whose title may be translated as "When the crows were still colourful", provides a fable that is more fun, yet insightful. It describes the story of the crows when they still came in all sorts of colours and patterns -orange with blue stripes, green with yellow spots, and so on -until one day a snowman asked the fateful (and probably spiteful) question of what a real, true crow looked like. Now the yellow-with-blue-spotted crows declared yellow with blue spots was the true colour of crows, but the lilac crows argued the ur-crow was lilac-coloured, and all the other crows also claimed their colours were the real ones. There was arguing and quarrelling; the crows began to fly with like-coloured others only. The fighting ended only when one day a black rain turned all animals black.
Afterwards, only the crows stayed black and no longer had a reason to argue. The moral of the story? Obviously: ingroup projection is a challenge of diversity! And if we do not want to buy social harmony with dull sameness, we had better think of a more creative way to appreciate and enjoy differences.
Ingroup projection is the perception or claim that one's own group is more prototypical for a higher-order superordinate identity, hence more normative and positive, than a relevant comparison outgroup is; or more prototypical at least than the outgroup thinks the ingroup is. In the present chapter we will briefly outline the ingroup projection model (IPM; Mummendey & Wenzel, 1999;Wenzel, Mummendey, & Waldzus, 2007), discuss its key concepts, relevant recent findings, and essentially argue for two ways in which we need to construe our superordinate identities in order to reduce tension between diverse and divergent groups included in them: we need to advance consensus about the superordinate identity in question, and about the complexity of its representation.

Outline of the Ingroup Projection Model
Ingroup projection may be understood as a contemporary version and specification of an early concept in psychology and sociology: ethnocentrism (Gumplowicz, 1879;Sumner, 1906). Ethnocentrism describes a tendency of members to use their own group, its values and norms, as a judgmental reference standard for their social world, including other groups.
Judging others by one's own group's standards is likely to lead to a sense of superiority of one's own group and the devaluation of others. However, whereas in ethnocentrism the imposition of one's own standards is somehow direct from one group to others (see Hegarty & Chryssochoou, 2005), in the concept of ingroup projection the process is mediated via the generalization of such standards to a higher-order identity in which certain other groups are included too; and qua their inclusion in the higher-order, superordinate identity these standards are seen as valid and applicable to those other groups. As a consequence, in contrast to ethnocentrism as an almost natural unfettered tendency, the process in ingroup projection is constrained by the social context and how individuals represent it.
The notion of ingroup projection is basically a corollary of self-categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). This theory assumes that individuals use social categories to structure and give meaning to their social world, including their selves.
They classify themselves as members of one category in contrast to others; more specifically, depending on the social context, as members of categories at various levels of inclusiveness: at an individual level (or even sub-individual level, such as the real vs. ideal self), a group level (with groups obviously also differing in their breadth and inclusiveness), up to the human level (or even supra-human level, such as living beings). These self-categorizations form the basis of an individual's contextually defined self-concept; self-categorization as a member of a group is the cognitive component of social identity (Tajfel & Turner, 1986).
The evaluative connotations of one's social identity derive from comparisons with relevant outgroups (although they may also form through intragroup interactions alone; Postmes, Spears, Lee, & Novak, 2005). Self-categorization theory assumes that comparisons between ingroup and outgroup imply that both are included in a superordinate, more inclusive category; only their shared inclusion makes two groups comparable (Turner et al., 1987).
Their more abstract similarity at a superordinate level (e.g., humans have skin with pigmentation) allows their comparison at a group level (e.g., the ingroup's skin tone is lighter than the outgroup's). Moreover, the superordinate category may be represented in form of a contextually defined prototype (see Oakes, Haslam, & Turner, 1998), which constitutes the normative and positively valued position on the comparison dimension (e.g., the prototypical human skin tone is light olive). The groups are evaluated in terms of their relative closeness to that prototypical position; their relative prototypicality. The more prototypical a group is relative to a relevant comparison group, the more positively valued it is, and the higher in status and more deserving it is judged (see Wenzel, 2004).
When individuals self-categorize as members of an ingroup, they depersonalise and adopt their group's goal and values (Turner et al., 1987). Their commitment to their ingroup's goals and values means they are motivated, and may experience ingroup pressures, to advance their group's goals; its positive distinctiveness, status, power or deservingness. This is where ingroup projection comes in, namely as the vehicle through which committed group members may seek to advance their group's standing, by perceiving or portraying it as more prototypical for a salient and identity-relevant superordinate category, compared to a salient outgroup (Wenzel et al., 2007). On the flipside, to the degree that the ingroup claims to represent the prototype of the superordinate group (specifically, in terms that render the ingroup distinct from the outgroup), the comparison outgroup is perceived to be less prototypical, more deviant, less valued and less deserving. Thus, ingroup projection should be related to ingroup favouritism (i.e., a relatively more positive evaluation and treatment of the ingroup). Conversely, to the extent that the outgroup is seen as deviating from the superordinate norm or questioning it, ingroup projection should be related to outgroup derogation and hostility. Hence, diversity (i.e., the presence of group differences) poses challenges when groups engage in projection and make their own the norm of the superordinate group.

Fundamental Processes
There is good empirical evidence for these fundamental predictions of the IPM (for an earlier review, see Wenzel et al., 2007). For example, as processes of ingroup projection should generally apply to the perspectives of both groups in a salient intergroup context, this would lead to a perspective-divergence between the groups about their relative prototypicality. Consistent with this, business administration students and psychology students disagreed about their relative prototypicality for students generally, with business administration students perceiving themselves as more prototypical than psychology students thought they were, and vice versa . Similar patterns were found for chopper and sports bikers, primary and high-school teachers, and East and West Germans (Waldzus, Mummendey, Wenzel, & Boettcher, 2004). Further evidence shows that members tend to engage in greater projection when they identify with both their ingroup (implying commitment to advance their group's goals) and the superordinate group (implying recurrence to it is a relevant normative referent), as the model predicts (Ullrich, Christ, & Schlüter, 2006;Wenzel et al., 2003;. Likewise, group members claim greater prototypicality specifically for ingroup attributes that are distinct from those of the outgroup in the given context. When Britons were the salient outgroup German participants projected attributes on which Germans were stereotypically different from Britons, whereas when Italians were the salient outgroup they claimed greater prototypicality on attributes on which they were stereotypically different from Italians (Waldzus, Mummendey, & Wenzel, 2005). This is consistent with the idea that ingroup projection is most functional when distinctive and contextually group-defining features are considered prototypical.
Similarly, there is supportive evidence that perceived relative prototypicality is related to negative attitudes and behavioural intentions towards the outgroup (e.g., Wenzel et al., 2003;Waldzus et al., 2003Waldzus et al., , 2005, perceived legitimacy of higher ingroup status (Weber, Mummendey, & Waldzus, 2002), and perceptions of entitlement (Wenzel, 2001), including support for policies limiting the rights and opportunities of ethnic minorities (Huynh, Devos, & Altman, 2015). However, consistent with the model, relative ingroup prototypicality was related to negative outgroup attitudes only when the superordinate identity was positively valenced; when it was negatively valenced the meaning of prototypicality reversed, with it being positively related to outgroup attitudes (Wenzel et al., 2003, Study 3). As a further theory-consistent qualification, the perceived relative prototypicality of the ingroup was only related to negative outgroup attitudes when the superordinate category was indeed inclusive of the outgroup. For example, Germans' perceived relative prototypicality for West Europeans (a category that does not include Poland) was not related to attitudes towards the Polish, but relative ingroup prototypicality for Europeans, which includes the Polish, was ; see also Imhoff, Dotsch, Bianchi, Banse, & Wigboldus, 2011).
A meta-analysis reported by Wenzel et al. (2007) indicated a significant relationship of moderate size between relative prototypicality and intergroup attitudes. Two studies by Berthold, Mummendey, Kessler, Luecke and Schubert (2012) found this relationship to be stronger for relative prototypicality measured in relation to the ideal, rather than actual, superordinate category, probably because the ideal implies a more prescriptive reference standard. Further, using cross-lagged regressions in a two-wave study on native Germans' attitudes towards immigrants, Kessler et al. (2010) found indications for causal effects of relative ingroup prototypicality on intergroup emotions and attitudes; however, they also found signs of the reverse causality, suggesting that ingroup projection may also serve a justification of existing prejudices. This should not surprise: if, as assumed, motivational and normative pressures elicit ingroup projection based on one's identification with one's ingroup, then such motivations and norms may equally well stem from one's 'antiidentification' with, and one's group's negative views towards, the outgroup. Below, we return to the question of the different social and epistemic functions of ingroup projection.

History, Humanity and Holiness
Despite the identity-driven tendency to cast one's ingroup as the more prototypical group for a salient superordinate group, this is conditioned by the given social, historical and ideological context. For example, in re-unified Germany both East and West Germans rated West Germans as the more prototypical subgroup (even if they disagreed about the extent to which this was the case), which is not surprising as it was the East German system that collapsed whereas the West German one prevailed and integrated East Germany -going hand in hand with West Germans having economic and political dominance , as well as constituting the numerical majority. Similarly, while both Protestants and Catholics agreed on Northern Ireland being a relevant superordinate category, they both also perceived Protestants as having greater overlap with (prototypicality for) Northern Ireland (Noor, Brown, Taggart, Fernandez, Coen, 2010). Protestants, while identifying more strongly with Northern Ireland than Catholics did, were less prepared to forgive the outgroup, the more strongly they identified with the superordinate group. This is contrary to research showing a common superordinate categorization is beneficial for intergroup forgiveness (Wohl & Branscombe, 2005; see also Gaertner & Dovidio, 2000). Protestants might have considered their group as higher-status and more deserving because of their greater prototypicality, and therefore might have been less willing to make concessions to, and sympathize with, the outgroup in particular when they identified strongly with the superordinate group. Ingroup projection may thus be an important qualifier of the effects of common ingroup identity.
Indeed, even the most abstract of human-social categories, widely considered the most sacrosanct, is not immune to these processes: humanity. Certainly, identification with humanity has regularly been found to be positively related with a number of prosocial outcomes such as the reduction of global inequality and the advancement of human rights (e.g., McFarland, Brown, & Webb, 2013;Reese, Proch, & Finn, 2015). However, the human category also seems to be a platform for ingroup projection processes, where distinct attributes of the ingroup in relation to a salient outgroup tend to be considered more typically human (Paladino & Vaes, 2009). Such a process mirrors the phenomenon of infrahumanisation where an ingroup is perceived to be able to experience complex, uniquely human emotions more than a salient outgroup is (Leyens et al., 2000), which has been shown to underpin or express intergroup prejudices (e.g., Demoulin et al., 2004). Koval, Laham, Haslam, Bastian and Wheelan (2012) found that such seeming projection held more specifically for negatively valenced attributes, suggesting that ingroup members may humanize their weaknesses to protect their group identity. Further evidence has supported the view that such ingroup projection processes may limit, or even pervert, the positive benefits commonly expected from a human identity. Reese, Berthold and Steffens (2012) found that members of an economically developed country rated developed countries (the ingroup) as more prototypical for all humans than developing countries. There was furthermore evidence that ingroup prototypicality was positively related to perceived legitimacy of global inequality and, mediated by this, negatively related to behavioural intentions to advance the situation for developing countries and the idea of global equality. Reese, Berthold, and Steffens (2015) replicated these findings with regard to actual donation behaviour: Claiming to be more human can make one less humane.
Finally, processes of ingroup projection do not stop at religion either (or, perhaps at religion least of all!). Lie and Verkuyten (2012) referred to Islam as superordinate category and surveyed Turkish-Dutch Sunni Muslims and their feelings towards Alevi and Shiite Muslims. The authors found that Sunnis regarded their ingroup as more prototypical for Islam than Alevis and Shiites; perceived relative prototypicality was in turn negatively related to intergroup feelings towards these groups measured on a feeling thermometer.
There is thus good evidence for the basic principles of the ingroup projection model across a range of intergroup contexts; from the mundane to the politically charged, from the quintessentially human to the holy.

Advances in our Understanding of Ingroup Projection
The label ingroup projection should be understood as a metaphor. It is not meant as an expression of hidden drives and suppressed desires in a psychodynamic sense; and it does not privilege an understanding of the process as motivated and unconscious, rather than cognitive and heuristic, or strategic and deliberate; it may be all of the above. Likewise, the process can be literally projective/inductive in that the representation of the higher-order group is infused by that of the lower-level ingroup, but it can also be introjective/deductive where the ingroup representation is assimilated to that of the superordinate category (e.g., classical infrahumanization where the ingroup is described in terms of quintessentially human emotions; Paladino & Vaes, 2009). Ingroup projection basically only describes the observation that people indicate their ingroup to be relatively more prototypical for a common superordinate category than other observers, such as members of a relevant outgroup, indicate it to be (Wenzel et al., 2007; see Figure 1).

Implicit Ingroup Projection
Recent research, however, has made some considerable advances in our understanding of the processes. First, ingroup projection can indeed be observed at a subliminal level and can occur spontaneously without cognitive control. Bianchi, Mummendey, Steffens, and Yzerbyt (2010) used a semantic priming technique and found that the superordinate group label presented as subliminal prime prior to lexical decisions (i.e., whether a presented letter string is a word or not a word) facilitated the recognition of typical ingroup attributes as words compared to typical outgroup attributes. The results suggest that ingroup traits were associated with the superordinate group more than outgroup traits.
Another interesting technique by Imhoff et al. (2011) used the visual representation of the superordinate group as a measure of ingroup projection. A sample of Portuguese and German participants decided over hundreds of trials which one of two presented faces (randomly generated from one base image) was more European. The average of the faces chosen by each sample were then rated by two independent Portuguese and German samples.
Both groups rated the average face resulting from the Portuguese sample as more typically Portuguese, and the face resulting from the German sample as more typically German. In other words, both Portuguese and Germans represented Europeans more in their own image.

Ingroup versus Social Projection
Ingroup projection has obvious similarities to the older concept of social projection, an individual's tendency to expect others to be similar to self (Allport, 1924;Krueger, 2007).
While this tendency is likewise constrained by social categorization and applies in particular to others who are part of the same category as self (Clement & Krueger, 2002) -similar to ingroup projection applying to an inclusive, superordinate category -the reference point or anchor is the individual self rather than the ingroup (see also Otten & Wentura, 2001). Thus, the question is whether ingroup projection is indeed, as the theory presumes, a group-level phenomenon or whether it is a redundant concept, where the process is due to self-tosuperordinate category projection. Bianchi, Machunsky, Steffens, and Mummendey (2009) investigated this question by having participants rate self, ingroup, outgroup and superordinate group on a series of attributes (ingroup-typical, outgroup-typical and irrelevant attributes), and calculating intraindividual partial correlations between ingroup and superordinate ratings (controlling for self), and between self and superordinate ratings (controlling for ingroup), as measures of ingroup and social projection, respectively. Ingroup projection was greater than social projection for ingroup-typical traits, and the former uniquely related to a measure of ingroup bias (Study 1); and ingroup projection but not social projection was greater when the ingroup was manipulated to have positive rather than negative value. These findings suggest that ingroup projection is distinct from, and not reducible to, social projection (for further evidence see Imhoff & Dotsch, 2013;Machunsky & Meiser, 2009).

Cognitive-Heuristic and Identity-Motivated Ingroup Projection
Theoretically, we understand ingroup projection as a multi-faceted process that can have cognitive-heuristic, identity-motivational, politically-strategic or ideological dimensions. A number of studies have contributed to the delineation of these different meanings (even if some of the researchers set out to privilege one meaning over others). Machunsky and Meiser (2009), for example, regard abstract, superordinate categories as rather ill-defined, yet individuals are often quite familiar with one particular part of it: their ingroup. Therefore, for questions of representation of a superordinate category the ingroup constitutes a reasonable heuristic for inference. Illustrating their point, Machunsky and Meiser (2014a) argue and found that groups represented as prototypes (rather than exemplars) are easier, and hence more likely, to be used to infer the representation of a superordinate category, irrespective of whether the perceiver is implicated as a member or not. The authors argue that this cognitive mechanism alone may explain why ingroup projection occurs, as on positive attributes ingroups (and on negative attributes outgroups) tend to be represented in form of prototypes. However, on top of representation effects the authors also found a main effect of group membership (Study 3), suggesting that identity motives conjointly with cognitive mechanisms may account for ingroup projection.
There are further empirical clues that point to identity-motivational underpinnings of ingroup projection. As mentioned above, ingroup projection has been found to be positively related to ingroup identification only when the evaluation of the superordinate category is positive (e.g., Wenzel et al., 2003), implying a motivation to maintain a positive social identity that ingroup projection may thus help satisfy (Tajfel & Turner, 1986). Moreover, highly prototypical groups often show stronger identification with superordinate categories and stronger correlations between subgroup and superordinate identification than subgroups that are less prototypical (Devos, Gavin, & Francisco, 2010). In line with an identity-serving function, ingroup projection was further enhanced when an outgroup posed a threat to the ingroup (Ullrich et al., 2006). Indeed, whether ingroup projection is more of an identity-motivated or merely cognitive-heuristic process may depend on how much is at stake for the ingroup's identity.
Consistent with this argument, Rosa and Waldzus (2012) observed in status-secure intergroup contexts, where the ingroup's higher status was stable, elevated levels of ingroup projection when participants responded under time pressure, cognitive load, or spontaneous responding instructions, thus presumably in a heuristic processing mode. In contrast, when the ingroup's status was insecure ingroup projection was more pronounced when there was no time pressure or when highly identified participants were instructed to respond thoughtfully (a systematic processing mode; see Chaiken & Trope, 1999). Machunsky and Meiser (2014b) report equivalent results using manipulations of positive versus negative mood to induce heuristic and systematic processing, respectively. Hence, ingroup projection can be both, a cognitive heuristic for efficient responding and a socially motivated, deliberate act of defending one's social identity.

Politically Strategic and Ideological Ingroup Projection
Not always, however, are group members, even those highly identified and committed, concerned with the positive portrayal of their group; sometimes the group's goals require a longer-term perspective than the immediate gratification through a positive identity.
For example, members may present their group as poor in order to point to an inequity and demand redistribution (van Knippenberg & van Oers, 1984). Similarly, ingroup projection may not always be in the service of a positive evaluation of the ingroup, but may rather be used strategically to promote other (longer-term) goals. Sindic and Reicher (2008) advanced this position in two studies set in the context of Scottish aspirations for independence from the British. They argued that Scots who support independence would seek to bolster their case by portraying the Scottish as unprototypical for Britain, thus emphasizing a lack of fit or belonging that supports the aim of secession. Indeed, the results showed that, when the issue of independence was salient (and only then) support for independence was related to perceived relative prototypicality in a curvilinear way, with the latter declining at higher levels of support for independence. Further, when support for independence was relatively low Scottish identification was positively related to relative prototypicality (ingroup projection); however when support for independence was relatively strong Scottish identification was negatively related to relative prototypicality. Strotmann (2007) found similar results for Catalan and Basque people in Spain, two regional groups with strong political aspirations for independence. This shows that ingroup projection is not only a function of group identification, but rather depends on the group's goals and values. Ingroup projection (or the opposite: claimed unprototypciality) can be used strategically to advance the group's goals.
Moreover, perceptions of relative prototypicality can also be in the service of higherorder values and ideologies. For example, they could be used by both low and high-status groups to justify existing social inequalities (Peker, 2009), for example in line with the presumed palliative function of system justification (Jost, Gaucher, & Stern, 2015), or as an expression of commitment to values of the superordinate group that provides members with a positive social identity (Spears, Jetten & Doosje, 2001). For example, in the context of salient economic status differences between African and European Americans in the US, lowerstatus African Americans could describe European Americans as more prototypical than themselves on attributes such as entrepreneurship, justifying the status differences but, at the same time, elevating the attribute to a value of the superordinate group, Americans, that might distinguish it positively from other countries, benefitting also African Americans with a positive American identity (see Wenzel, 2002). On the other hand, ingroup projection could give expression to high-status group members' members' endorsement of an ideology that justifies group-based hierarchy and their own dominant position in it (Sibley, 2013). In sum, ingroup projection can be seen as having cognitive, motivational, strategic and ideological underpinnings and functions.

Perceived Projection by Outgroups
The ingroup projection process implies that the outgroup is relatively unprototypical for a superordinate group that serves as a shared evaluative reference standard; to the extent that this is a positive standard, the outgroup is likely to be valued negatively, as deviant or inferior. More than that, the outgroup -just by being different, but probably in particular when seen as choosing to be different -may be regarded as challenging the representation of the superordinate group and what the ingroup perceives to be its defining values. Thus, outgroup members could be accused of engaging in projection on their part, trying to impose their values onto the representation of the superordinate category. Such an accusation could be made by dominant and dominated groups alike: A dominant majority may regard a minority's stubborn difference and unwillingness to assimilate as subversion, an attempt at undermining traditional values. Conversely, a low-power minority may regard a majority's dominance and insensitivity to difference as cultural imperialism, imposing its values on everyone else. From either perspective, perceived projection by outgroups could be an identity threat that triggers negative reactions (see Figure 1).
First, this reasoning can account for what may otherwise seem paradoxical effects of the endorsement of common identity, observed by Gómez and colleagues (Gómez, Dovidio, Huici, Gaertner, & Cuadrado, 2008;Gómez, Dovidio, Gaertner, Fernández, & Vázquez, 2013). Upon learning that ingroup members categorized ingroup (e.g., Spaniards) and outgroup (e.g., Eastern European immigrants) as being part of a common identity (e.g., Europeans), participants showed more positive orientations towards the outgroup; this is in line with the common ingroup identity model (Gaertner & Dovidio, 2000) which posits that a common, superordinate identity turns an outgroup into a higher-order ingroup to whom ingroup-favourable attitudes are extended. However, upon learning that the outgroup categorized both groups in the same way as part of a common identity, participants responded with less positive orientations. Thus, the outgroup's endorsement of a common identity did not have the same positive effect and indeed backfired. Gómez et al. (2013) showed that this backfiring effect was mediated by symbolic threat, measured as the perceived incompatibility of the groups' values. We can understand these results in terms of perceived projection by the outgroup: Outgroup members' statements of belonging to a shared superordinate category may be seen as a claim that their group (with its different values) is prototypical of, and equally fitting into, that group; an attempted subversion of values that the ingroup is likely to object to.
Indeed, measuring such perceptions directly, Von Oettingen (2012; Von Oettingen, Mummendey, & Steffens, 2015) found evidence that ingroup and outgroup not only objectively diverge in their relative prototypicality perceptions (as in Wenzel et al., 2003), but group members also tend to hold subjective perceptions of a divergence, believing that the outgroup thinks the ingroup is less relatively prototypical than they, the ingroup members, think it is. In the majority-minority contexts studied by Von Oettingen, it was in particular the prototypicality of the minority that was subject to such meta-perceptive divergences; for example, minority members tended to believe the majority considered the minority to be less prototypical than the minority members thought was the case. Moreover, there was evidence, above and beyond the effects of ingroup projection, that beliefs that the outgroup engaged in projection were predictive of negative intergroup attitudes (in particular among majority members, and minority members who maintained a separate subgroup identity). Indeed, suspected projection by an outgroup may elicit ingroup projection tendencies in response (see also Finley & Wenzel, 2003). Even when a contest over prototypicality is only in group members' minds, it is no less real in terms of the intergroup tensions it can cause.

Representations of the Superordinate Identity to Curb Ingroup Projection
It needs to be said that ingroup projection as such is not necessarily problematic. If members see their ingroup as highly prototypical they may commit more strongly to the relevant superordinate category. Any project trying to promote a superordinate identity for the sake of large scale cooperation, such as environmentalism mobilizing the global human community, might gain from such processes. However, if humanity, for example, is framed as an extension of Western, Christian, economically developed societies, groups that are minorities or just less powerful in the public discourse might feel they are not sufficiently represented. Under some circumstances ingroup projection by dominating groups may condemn less dominant subgroups to a stigma of deviance, which in turn undermines their access to resources and power that prototypical members of the superordinate category are entitled to. Thus, ingroup projection may at times be a means to legitimizing injustice and dysfunctional inequalities that prevent people from developing their full potential (Weber et al., 2002). Moreover, when groups disagree about their prototypicalities, whether or not there is an asymmetric status relation involved, the seed is sown for misunderstandings and tension which all too easily escalate in outright conflict. In such circumstances it would be desirable to curb ingroup projection.
As ingroup projection implicates the prototype of superordinate categories, the theorizing about remedies has focussed above all on how such superordinate categories are represented, cognitively and in discourse, and in particular how much scope there is in these representations for groups that are different from the mainstream. Broadly, one could think of two options (Mummendey & Wenzel, 1999;Wenzel et al., 2007): superordinate categories that are only vaguely or minimally defined, thus otherwise undefinable, so that no subgroup can claim to represent the prototype better than others; or, superordinate categories that are represented as complex so that different subcategories can be equally prototypical at the same time.

Vagueness, Complexity, Coherence and Indispensability
As an example for the first option, vagueness, consider Bilewicz and Bilewicz's (2012) argument against the recently proposed concept of omniculturalism (Moghaddam, 2012), which suggests focussing on human commonalities before taking into account intercultural differences, as a tool to improve intercultural relations. Such omniculturalism, Bilewicz and Bilewicz argue, carries the risk that humanity is characterized by the distinctive features of one's own culture, which has historically served to justify western colonial hegemony. Instead they propose a more undefined notion of humanity, as "[w]ithout a clear definition of what makes someone human, people would not deny humanity to outgroups" (Bilewicz & Bilewicz, 2012, p. 340).
Whether a representation of the superordinate category as undefined can reduce ingroup projection has been tested experimentally. In a study by Waldzus et al. (2003 However, the study by Waldzus et al. (2003) also points to limitations of undefined representations as a way to reduce ingroup projection: Participants who identified strongly with both Germans and Europeans showed relatively high scores of ingroup projection even in the undefined representation condition. It is possible that an undefined representation appears open to be defined for those with a strong motivation to do so.
The second option for curbing ingroup projection can be achieved by increasing the superordinate category's complexity (see Figure 2). That way, superordinate categories are created that make prototypicality of outgroups not only possible, but rather likely or even mandatory. Originally defined as a representation in which "the distribution of representative members on the prototypical dimension is […] multimodal" as compared to "unimodal" (Mummendey & Wenzel, 1999;p. 167), the more general aspect of complexity is that the superordinate category is represented as diverse, allowing different instantiations to be prototypical. It carries the notion of a group that would not be what it is without the recognition of intra-category differences between similarly representative yet distinct positions. A complex superordinate category implies accepting differences within this superordinate category as normative, which makes it unlikely that a subgroup that is different from the others will be stigmatized because of this difference. Indeed, there is a larger tradition of research studying how the appreciation of difference and diversity impacts intergroup relations, for instance regarding effects of multiculturalist ideology (Park & Judd, 2005) or diversity beliefs in organizations (van Dick, van Knippenberg, Hägele, Guillaume, & Brodbeck, 2008).
Testing this experimentally, Waldzus et al. (2003, Study 2) asked German participants to describe either the diversity (complexity condition) or unity (control condition) of Europe.
As predicted, while participants in the control condition indicated that Germans were more prototypical for Europe than the Polish outgroup, in the complexity condition participants considered Germans and Polish as equally prototypical. Waldzus et al. (2005)  Still, the findings suggest that, to be effective, complexity should not be chaos but rather a meaningful constellation.
An alternative way of thinking about positive implications of superordinate complexity is in terms of the subgroups' indispensability for the superordinate category.
Verkuyten and colleagues (e.g., Verkuyten, Martinovic, & Smeekes, 2014) propose the concept of indispensability as an alternative to relative prototypicality. Prototypicality conventionally implies a graded category structure (see Turner et al., 1987) by which minorities would have little chance to be considered equally prototypical, but when thinking of a superordinate group as a combination of complementary subgroups (akin to a mosaic) even the smallest subgroup (mosaic piece) can be considered necessary and indispensable for the overall picture. Complexity can thus be thought of as complementarity of groups toward an overall identity. The ensuing perceived indispensability of outgroups has been found to be positively related to intergroup attitudes in Mauritius (Ng Tseung-Wong & Verkuyten, 2010), positive interethnic feelings in Malaysia (Verkuyten & Kahn, 2012), and support for immigrant rights in the Netherlands (Verkuyten et al., 2014).

Second-Order Conflict about Superordinate Representation
Two important implications have to be taken into account when inducing such complex superordinate categories. First, complex representations of superordinate categories do not necessarily reduce an ingroup's relative prototypicality. Rather, they lead to the perception of more equal prototypicality between subgroups. Depending on whether the ingroup started off from a high or low level of relative prototypicality, complexity might imply either a decrease or increase in relative ingroup prototypicality, respectively (Waldzus, 2010). For instance, where a minority group has shared in a consensual notion that it is less prototypical than a dominating majority (e.g., , the idea of superordinate category complexity leads its members to perceive increased relative ingroup prototypicality (Alexandre, Waldzus & Wenzel, 2015). Moreover, such a shift towards more equal prototypicality between subgroups does not necessarily mean that relations between subgroups become more harmonious. Rather, a complex framing of superordinate categories may represent a challenge to the status quo that could be seen as threatening by established higher-status groups, while it may facilitate cognitive alternatives to the status quo among minority groups (Tajfel & Turner, 1986) that foster dissatisfaction with its inferior position and tendencies to seek social change.
This idea leads to the second implication: Whether or not superordinate categories are seen as complex (or, more generally, represented in some form that implies the possibility for minority groups to be prototypical) might in itself become a battleground of dispute.
Established dominating subgroups might reject such representations exactly because they put into question the legitimacy of their dominance. Indeed, Strotmann (2007) found that the induction of superordinate complexity (by triggering elaboration on the cultural richness and diversity in Spain) led Andalusians, a highly prototypical group in Spain, to identify less with this superordinate group, essentially withdrawing from a shared common ground. In contrast, powerless groups (often minorities) might frame superordinate categories as complex as part of their striving for social change (Saguy & Kteily, 2014). For instance, in Strotmann's (2007) research, groups that saw themselves as non-prototypical for Spain and that had strong social change ambitions (i.e., Catalans and Basques) perceived Spain to be more complex than highly prototypical subgroups did (i.e., Andalusians). Similarly, in an Irish context Joyce, Stevenson and Muldoon (2012) found in interviews that Irish Traveller participants proactively displayed and claimed what the authors called 'hot' national identity (Reicher & Hopkins, 2001) in order to establish their Irishness, whereas Irish majority participants tended to disparage proactive displays of Irishness and seemed to endorse a 'banal' (Billig, 2005) Irish identity that represents an assumed and unquestioned backdrop for their own unproblematic prototypicality. Thus, whereas minorities may seek to contest, majorities may seek to normalize a superordinate identity representation. More generally, the representation of superordinate categories as more or less open for minority subgroups' prototypicality depends on its functional role in actual intergroup relations.

Dominant Groups' Resistance to Complexity Interventions
The possibility that dominant groups regard the notion of superordinate identity complexity as a threat, questioning their greater prototypicality and status, is particularly problematic considering that interventions designed to break down prejudices and increase intergroup tolerance are typically aimed at dominant groups. It could be expected that members of dominant groups would generally not be very receptive to programs that seek to redefine superordinate identity as more complex, particularly if such interventions carry not only the abstract idea of diversity but also make salient its implementation in the concrete context (Yogeeswaran & Dasgupta, 2014).
Indeed, majority group members and participants who consider their ingroup prototypical of the superordinate group have been shown to report more uneasiness or threat after thinking about the diversity rather than unity of the superordinate group (Ehrke & Steffens, in press;Steffens, Reese, Ehrke & Jonas, 2015). In one study, U.S. Americans were asked to think about the role of the U.S. in North America or in the world (Steffens et al., Experiment 3). As expected, U.S. Americans thought they were highly prototypical of North Americans, but less so for the world population. They were then asked to think about either diversity or unity (either within North America or within the world). The results indicated more reported threat and less positive outgroup attitudes following diversity instructions in the North American context than in all other conditions. In other words, in a context in which members more (than less) strongly feel their subgroup represents the prototype, they may respond negatively to suggestions of superordinate diversity.
Similarly, there may be levels of diversity that are more (vs. less) taxing for subgroups who consider themselves prototypical, with diversity that includes particularly counter-prototypical groups likely to provoke more negative responses from a majority subgroup. Dieckmann, Steffens, Mummendey, and Methner (2015) investigated this in the context of metal music. The participants were death metal music fans, who were a highly prototypical group among metal music fans. All of them were asked to think about the diversity of metal fans. However half of them were presented with the examples of black metal fans and thrash metal fans, which are two other prototypical groups of metal fans; the other half were given the examples of white metal fans and industrial metal fans, which are peripheral groups that are not regarded as "true" metal fans by some. After thinking about the diversity of metal fans and seeing peripheral rather than prototypical groups as examples, death metal fans disliked another peripheral group, nu metal fans, more. Further findings indicated that too much diversity appears to threaten the groupness of a superordinate group, and members of prototypical subgroups react to this threat by becoming more conservative, less inclusive, and with negative attitudes towards peripheral groups.
The same can be true for "natural experiments" in diversity, such as increased diversity due to demographic shifts. In the US, for example, there are predictions that in coming decades non-Hispanic Whites will lose their numerical majority status and will represent less than 50% of the population. Danbold and Huo (2015, Study 1) found for White American participants such a perceived demographic shift to be related to prototypicality threat (i.e., the feeling that their status as prototypical Americans was threatened), particularly for those respondents who perceived prototypicality to be unevenly distributed (i.e., seeing Whites as more prototypical). Moreover, in an experimental Study 2, White American participants reacted to such information with prototypicality threat, and mediated by this threat, they were less embracing of diversity in the US. In other words, increased diversity (whether naturally occurring or as an intervention strategy) can pose a threat to dominant groups and lead to the rejection of diversity (complexity) as an identity-defining property (see also Craig & Richeson, 2014;Outten, Schmitt, Miller, & Garcia, 2012). Hence, while complex superordinate identity might be an answer to the challenge of ingroup projection, it is not a simple one.

Consensus versus Complexity: Harmony and Social Change
As we are struggling with an answer, let us review what the question is (see Figure 3, for an overview of the model). Ingroup projection processes imply that members represent their own group as relatively more prototypical for a superordinate identity that also includes the outgroup; not necessarily more prototypical than the outgroup in absolute terms, but relatively more prototypical than the outgroup grants (or is seen to grant) the ingroup to be.
Indeed, because both sides of a given intergroup context can engage in ingroup projection, the groups are likely to show a perspective divergence in their perceptions of superordinate prototypicalities Wenzel et al., 2003). A perceived divergence may be all the more likely as members may suspect outgroups of engaging in projection (Von Oettingen et al., 2015). The result would be disagreement, either tacit misunderstanding or explicit dissent (Kessler & Mummendey, 2009), which causes or reflects negative intergroup relations, discord and discrimination. Based on this analysis the question would be how to create greater consensus between the groups about their relative prototypicalities. With greater consensus we may expect more intergroup harmony.
Towards this aim, one could think of approaches meant to overcome an ingroupcentric view and promote a better understanding of the outgroup's view. For example, Berthold, Leicht, Methner and Gaum (2013) investigated the effects of perspective-taking and found that its positive effects on attitudes towards outgroups were mediated by reduced perceptions of relative ingroup prototypicality. Hence, perspective-taking may be a means for reducing ingroup projection (as well as possibly perceived outgroup projection), thus reducing disagreement about prototypicality and increasing intergroup harmony. Likewise, superordinate identity complexity could be seen as an approach to facilitate intergroup consensus about the superordinate group's representation and the subgroups' prototypicalities (or indispensabilities), in that it portrays such perceptions as not being zero-sum. With one group's prototypicality not being at the cost of another group's prototypicality, each group's claims for prototypicality might be more agreeable to the other.
However, as we have seen, this is in fact not quite true, as members of dominant groups tend to resist such notions of complexity (e.g., Steffens et al., 2015). The reason is, there is more at stake than mere agreement and harmony; namely equality and respect (see Simon, Mommert, & Renger, 2015). Indeed, too much emphasis on consensus and the implied harmony may only enshrine social inequalities and foster a misguided contentment with the status quo (see Jost et al., 2015). A similar criticism has been raised against the contact hypothesis as an approach to improve intergroup relations, namely that its narrow focus on the reduction of prejudice and promotion of intergroup harmony helps cement the status quo and reduce motivations to seek social change towards greater equality between the groups (Dixon, Levine, Reicher, & Durrheim, 2012;Wright & Lubensky, 2009). If the goal is social change towards greater equality and justice, intergroup contact (and other approaches) would have to aim at greater complexity in the representation of superordinate identities, which affords included subgroups more equal prototypicality as a path to equal respect, status and entitlements (Wenzel, 2013).
Complexity is therefore not necessarily a fluffy agreeable notion, but can be rather inconvenient and hard work. However, it is also not an impossible one, as demonstrated in research by Ehrke, Berthold and Steffens (2014) who found positive and lasting effects of diversity trainings on intergroup attitudes, mediated by the perceived complexity of a superordinate identity. Comparable findings were demonstrated in the context of a simulated United Nations conference to mitigate climate change, where diversity training increased the willingness to engage in collective action (Knab, 2015). Theoretical sophistication likewise helps, as demonstrated by Ehrke and Steffens (2015) who devised an approach that led autochthonous Germans to feel they themselves were not prototypical Germans in all respects; they were subsequently more receptive to a diversity manipulation and in response to it showed reduced ingroup bias.
Moreover, while a more complex superordinate identity representation implies social change that tends to be resisted by those who fear to lose their dominance, it might still be a more constructive path, for both winners and losers, compared to the alternative: seeking to overthrow dominance by converting others to one's group. For instance, Prislin and Filson (2009) found that minority members achieving social change (i.e, acceptance for their position) by advocating tolerance for diversity within a superordinate group showed greater levels of identification with the group compared to when they managed to convert majority members to their position. This was mediated by their perception that future differences would be regulated in a more conciliatory manner if the change came about through tolerance for different views rather than conversion of others to a new single view. Even majority members remained more strongly identified with the superordinate group after having lost their majority status, when the minority had advocated for greater tolerance of diversity.
Hence, while complexity implies and requires social change, it might also help to build common ground for it.
Above all, however, superordinate identity complexity is more likely to grow from a concerted approach, including consistent policy frameworks and institutions, as indicated in research by Sibley and Barlow (2009). These authors found that European Australian participants implicitly associated the superordinate category Australians more with their ingroup, European Australians, than with Indigenous Australians (mirroring US American=White findings by Devos & Banaji, 2005). In contrast, European New Zealanders associated the superordinate category New Zealanders equally strongly with Maori as with European New Zealanders. Importantly, the fact that this result was found on an implicit measure rules out the suspicion that answers were just driven by social desirability within an ideological climate demanding political correctness. While the quasi-experimental comparison with Australia defies clear conclusions, it is plausible that the historical, institutional and cultural acknowledgement of Maori in New Zealand (e.g., official bilingualism), fosters a more complex representation of New Zealanders -one that makes its subgroups more equally prototypical for their shared superordinate identity.

Coda
If it is mere consensus we want then the fable of the crows, with which we started this chapter, would seem to provide the answer: a rain that turns all crows black -if there were such thing. In the human world, cultural homogenization would seem a close equivalent to black rain: all would be the same, and there would be nothing more to quarrel about. And, yet, as much as their sameness and blackness left the crows grumpy, the human world would be considerably bleaker, too. Colourful diversity is a more difficult project, requiring more creative responses as to what joins us all together, and probably a continuous recreation of such responses too. One may wonder what would have happened if one crow, to the question what a real crow is like, had answered "colourful". Perhaps many more crows would have joined in, reaching consensus that no single crow colouring is the part that represents the whole; instead, the whole diversity of colours is needed to understand what crows are like.