Human and Global Identities: Different prototypical meanings of all-inclusive identities

The impact of identities encompassing all human beings (e.g., human and/or global identities) on intergroup relations is complex, with studies showing mostly positive (e.g., less dehumanization), but also negative (e.g., deflected responsibility for harm behavior) effects. However, different labels and measures have been used to examine the effects of these all-inclusive superordinate identities, without a systematic empirical examination of the extent to which they overlap or differ in their socio-psychological prototypical content. This study examined whether different labels activate the same contents in laypeople’s conceptualization. 248 participants openly described 1 of 6 labels: “All humans everywhere”, “People all over the world”, “People from different countries around the world”, “Global citizens”, “Citizens of the world” and “Members of world community”. Results from quantitative content analyses showed that the different labels activated different thematic attributes, representing differences in their core prototypical meaning. We propose that a general distinction should be made between labels that define membership based on human attributes (e.g., biological attributes), and those that evoke attributes characteristic of membership in a global political community (e.g., attitudinal attributes), as their effect on intergroup relations may vary accordingly.

In the past decade, research on intergroup relations has shown increased interest in the effects of identification with superordinate categories encompassing all human beings (e.g., identification with all humanity, being a global citizen).However, various labels, definitions, and measures of this concept have been proposed, raising the question of whether they should be treated as a single or as different social psychological realities.Whereas some studies have compared alternative measures (e.g., McFarland & Hornsby, 2015), no empirical research has yet enquired about the spontaneous meanings that people themselves attribute to the different designations of all-inclusive superordinate categories.In this paper, our goal is to fill that gap.Specifically, we aim to look at lay conceptualizations of some of the most commonly used labels, and to enquire as to what extent they are attributed similar or different meanings.
The import of this question largely flows from the fact that, so far, research has identified many positive effects of all-inclusive superordinate identification (McFarland et al., 2019), but also some negative effects (Morton & Postmes, 2011a).However, since studies have often relied on different labels and operationalizations, differences in meanings for participants could partly account for variations in effects.Indeed, the effects of superordinate identification with an all-inclusive category might differ depending on its specific content and meaning (Reese, Berthold, & Steffens, 2016).
In particular, we will argue that different labels might carry different (fuzzy) prototypes, which might become salient when people are explicitly asked to define or think about them.The fact that such distinction can indeed matter may be captured by the following example.Should we ask of someone why they are sending food and clothing to people in need in another part of the world, the answer that it is "because they are human beings like us" would intuitively make perfect sense.By contrast, the answer that it is "because we are all citizens of the world", while perhaps not entirely inappropriate, would probably not fit quite so neatly.Instead, the latter would fit better as an answer to the question of why someone believes that immigrants should be given the right to vote in their host country.In that context, it is the notion of common humanity that may not feel entirely as fitting.This illustrates not only that a distinction is possible in lay conceptualizations, but that such distinction can matter in term of the outcomes of identifying with an all-inclusive category.

Inclusive Social Identities and Intergroup Relations
There is a strong body of social psychological research showing that when intergroup categorization is salient (i.e., "us" vs. "them"), people tend to favor their ingroup, as a default ingroup norm (Allport, 1954;Turner & Reynolds, 2012).Indeed, social identity theory (Tajfel & Turner, 1979) and self-categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) posit that when individuals define themselves in terms of a social identity they experience a psychological depersonalization of the self and categorize themselves as members of particular groups in contrast to others.However, depending on the context, different social category memberships can become salient, and psychological depersonalization can also occur at different levels of abstraction.For instance, people categorized as ingroup vs. outgroup members in one context can be re-categorized as members of a common, higher-order superordinate group (e.g., humans ;Turner, et al., 1987) in another context.
Early work on prejudice reduction proposed precisely that intergroup biases could be reduced by altering the perception of group boundaries, redefining who is perceived as an ingroup member (Dovidio, Gaertner, Shnabel, Saguy & Johnson, 2010;Gaertner & Dovidio, 2000).Specifically, the Common Ingroup Identity Model (Gaertner & Dovidio, 2000) proposed that different forms of inclusive identities (i.e., one-group; dual identity) can be achieved by increasing the salience of existing superordinate memberships or by introducing factors perceived to be shared by these memberships (e.g., common goals; fate).Thirty years after the original proposition of the model, research shows consistent evidence that inducing members of different groups to conceive themselves either as one-group or two groups within a team (i.e., dual-identity), reduces intergroup prejudice and leads to prosocial responses toward former outgroup members, with both laboratory and naturalistic groups (e.g., Dovidio et. al, 2009;Gaertner, Dovidio, Guerra, Hehman, & Saguy, 2016).
In general, research has yielded mixed findings regarding the effects of all-inclusive identities.Some studies revealed positive responses from high identifiers with humans and with all humanity, such as less hostility toward asylum seekers (Nickerson & Louis, 2008), less threat towards religious groups (Dunwoody & McFarland, 2018), less dehumanization towards minority groups (Albarello & Rubini, 2012;Hamer, McFarland & Drogosz, in prep., as cited in McFarland et al., 2019), less ethnocentrism (McFarland et al., 2012), less collective narcissism (McFarland et al., 2019), more solidarity and collective action (Barth et al., 2015), more commitment to human rights (McFarland et al, 2012), and forgiveness of former national enemies (Hamer, Penczek & Bilewicz, 2018).Likewise, high identifiers with the world population and with a global community, also revealed more intentions to act against global inequality (Reese, et al., 2014), promotion of social justice and helping (Reysen, & Katzarska-Miller, 2013), as well as social responsibility and global activism (Reysen & Hackett, 2017).However, endorsement of common humanity has also been found to have potentially detrimental effects.For instance, making salient a common human identity (vs.intergroup identities), led victims of violence to show increased forgiveness of perpetrators, but also lowered intentions to engage in collective action (Greenaway, Quinn & Louis, 2011); and to normalization of intergroup harm, when human nature was perceived negatively (Morton & Postmes, 2011b).Also, members of groups that historically perpetrated harm against other groups deflected feelings of responsibility and guilt by rationalizing the ingoup's actions as a natural expression of human nature (Morton & Postmes, 2011a).
In sum research shows mixed findings of endorsement of all-inclusive identities, suggesting that the effects of these forms of superordinate categorization may be dependent on the specific content and meaning of these identities (Turner & Reynolds, 2012).In that respect, a careful analysis of research examining the impact of all-inclusive identities shows that a variety of labels have been used: "humans" (e.g., "How similar do you feel to other human beings?",Nickerson & Louis, 2008), "all humanity" (e.g., "I identify with all humanity", Barth et al., 2015), "people all over the world" (e.g."How often do you use the word "we" to refer to people all over the world?";McFarland, et al, 2012), "world community" (e.g., "Being part of the world community is an important aspect of my identity"; Reese, et al., 2014); "world as a whole" (e.g., "How strongly do you define yourself as a member of the world as a whole?",Buchan et al., 2011); "global citizen" (e.g., "I strongly identify with global citizens"; Reysen & Katzarska-Miller, 2013) or "world citizen" (e.g.,"I see myself as a world citizen"; Inglehart et al., 2014).On that basis, Reese and colleagues (2016) speculate that ascribing different labels to the superordinate group might activate different content, and thus different behavioral consequences.
However, it has also been proposed that some of these differences in labels might be superficial.In a recent review, McFarland and colleagues (2019) proposed that the constructs of global human identification and global citizenship share much in common, and used the umbrella term of "global human identification and citizenship" to refer to both.Although the authors acknowledge they might represent separate constructs, they state that both can be treated as largely interchangeable in terms of their effects, provided that "measures are strongly related, and each measure has yielded results that are consistent with the other measures" (McFarland et al., 2019, p. 142).
Despite this, there is some evidence suggesting that measures of identification with humanity and global citizenship can lead to different correlates and outcomes.For instance, endorsement of global citizenship was a stronger predictor of prosocial values than other allinclusive identities (e.g., human; Reysen & Katzarska-Miller, 2017).By contrast, McFarland and Hornsby (2015) compared the role of five scales of global human identification on predicting humanitarian concerns and found that measures of identification with all humanity and the psychological sense of a global community were more strongly associated with humanitarian concerns than was a measure of global citizenship.
Overall, while the extent to which different designations of all-inclusive identities can lead to variation in their effects is debated, the fact remains that we still do not know much about their precise content.Different theoretical definitions have been proposed, but, the spontaneous meanings that people themselves attribute to the different designations of allinclusive superordinate categories remains unclear.Considering this lack of evidence, one might question: Do they activate the same content in laypeople conceptualizations, regardless of how they are called?Or do people attribute them significantly different meanings?
In line with Reese et al. (2016), we propose that the different designations of allinclusive identities can indeed activate different prototypical contents.This makes it essential to investigate in more detail what these prototypical contents are, if we are to better understand their potentially varying impact on intergroup relations.As Roccas and Elster (2012, p. 13) put it, it is important to consider "with what people identify" as well as "how much people identify" with a group.It is the purpose of the present study to fill this gap in the literature and carry out such investigation.

Category prototypes
Research shows that language (e.g., labels) strongly affects how we cognitively represent and interact with social groups (Carnaghi & Bianchi, 2017), and that the exact label (e.g., immigrants, strangers) used to invoke social groups drives specific content, which becomes relevant and consequential once the category is activated (Spruyt, van der Noll & Vandenbossche, 2016).This is consistent with self-categorization theory's proposal that people cognitively represent social groups (e.g., Europeans) using category prototypesi.e., fuzzy sets of attributes (e.g., physical, emotional, attitudinal, behavioral) that are meaningfully inter-related, and describe ideal, rather than typical, ingroup members (Hogg & Smith, 2007).These prototypes are context specific, tend to be shared, and prescribe prototype-based attitudes and behaviors of group members.Thus, when we categorize people, "we view them through the lens of the group prototype, assign prototypical attributes to them, and interpret and expect behavior, including their attitudes, to conform to our prototype of the group" (Hogg & Smith, 2007, p. 96).Different social groups have different contents (i.e., attributes), and the degree of ingroup identification is related to the extent to which one endorses the group's normative content (Hogg & Smith, 2007).In this sense, category labels activate category-related contents and evaluative responses that in turn facilitate the gathering and appraisal of subsequent consistent information (Carnaghi & Bianchi, 2017).This perspective is compatible with both a situational and a dispositional approach to category content.On the one hand, it implies that category content can vary as a function of the context in which the category is invoked, as context makes salient different features of the (fuzzy) prototype.On the other hand, "this variability is relatively modest due to the anchoring effect of enduring and highly accessible representations of important groups we belong to" (Hogg & Smith, 2007, p. 95).
One effective method of assessing the lay perspective of a prototype is the prototype approach, which methods have been used to analyze psychological concepts such as emotions (e.g.Fehr & Russell, 1984;Fehr, 1988;Hepper, Ritchie, Sedikides, & Wildschut, 2012;Lambert, Graham & Fincham, 2009), but also social categories, as heroes (Kinsella, Ritchie, & Igou, 2015).In this study we use the first step of a prototype approach to examine the prototypical content of several labels used to refer to all-inclusive identities.

The Present Research
This study examined whether the different labels used for all-inclusive superordinate identities activate the same contents in laypeople conceptualization (e.g., attitudes; emotions; traits; and values).We reviewed instruments used to measure these forms of identification, and identified several keywords used to represent them (e.g., "human", "people", "citizen", "community" and "world").We then selected the most representative labels: "All humans everywhere", "People all over the world", "People from different countries around the world", "Global citizens", "Citizens of the world", and "Members of world community".
These six labels were used as treatment conditions in a between-subjects design.Participants were randomly assigned to one of six conditions and asked to generate attributes, in a free-response format, to describe the assigned label (Fehr, 1988;Hepper, et al., 2012;Kinsella, et al., 2015;Lambert et al., 2009).
Responses were thematically analyzed and organized into representative thematic attributes.Then, we performed chi-square tests of homogeneity to determine whether there were differences in the frequency counts of each attribute between conditions.In general, we expected that some attributes would be more frequently generated to describe some labels than others, pointing to the activation of different contents and, consequently, different prototypes of all-inclusive identities.

Method Participants
Two hundred and forty-eight adults consented to participate and completed the task (29 did not provide demographic information).The mean age was 36.99 years (SD = 12.92; age range: 18 -72), and 55.6% were female; 72.6% had higher education, 23.8% had secondary education, and 3.6% had basic education; 64.8% were employed; 98.6% were Portuguese citizens (194 living in Portugal and 22 abroad), and 1.4% were non-Portuguese living in Portugal.

Procedure
Participants were recruited through online advertisements in social networks, from November 16 th to December 16 th 2017, using Qualtrics platform.Participants were given the opportunity to participate in a lottery to win a 25€ voucher, as compensation for their participation.After consenting to participate, participants were randomly allocated to one of six conditions and asked to write, in 20 lines, characteristics that came to their minds when they thought about one of the six labels.Instructions were adapted from Fehr & Russell (1984, Study 6) 1 .Participants had 10 minutes to write the characteristics/attributes.They then answered demographic questions, and were thanked and debriefed.

Data analysis
In order to obtain and compare a pool of prototypical attributes of each label, the quantitative content analyses involved three major steps: 1) coding participants' responses into thematic attributes, using prototype analysis approach guidelines; 2) selecting the most representative attributes of each label, using a selection formula adapted from Katz and Braly (1933); and 3) determining whether the frequency counts of each attribute differed between conditions, using chi-square tests of homogeneity.
All responses were revised to identify inadequate statements (e.g., off-the-topic statements "acordei agora" [I woke up now]).Five were excluded from the analysis and the Coding procedures.The coding procedures were based on Fehr´s (1988) guidelines for prototype analysis.The first step was to extract a list of attributes for each of the six labels.When participants used full sentences, judgements had to be made about whether they should be treated as a single attibute or divided into several.As a general rule, each word was extracted as a single unit as long as it could stand on its own as an attribute.However, in some cases, group of words were coded as single units, when they possessed no possible relevant meaning on their own, (e.g dão a mão a quem precisa [they give a hand to someone in need] was extracted as a single coding unit), or included mere modifiers of an attribute (e.g., maior respeito pela diferença [more respect for difference] was extracted as respect for difference).The final data set included 3382 coding units (M = 14.09 units per participant, SD = 9.09; Min = 1; Max = 46).
The second step was grouping the extracted coding units into thematic categories.
Following prototype analysis conventions, we first organized the coding units by grouping (a) identical words; (b) word families (e.g.abertura [openness] and abertos [open]); and (c) meaning-related words or sentences (e.g.dão a mão a quem precisa [they give a hand to someone in need] and ajuda [help]).Then, following a bottom-up (inductive) approach, we created two thematic coding levels (Braun & Clarke, 2006).At a manifest level, we created descriptive and conservative thematic categories (e.g., coding units amados [loved], amor Then, at a latent level, we grouped these categories into higher and broader meaning-related macro-categories, designated below as attributes (e.g., amor [love] and amizade [friendship] were grouped into an attribute labeled afecto [affection]).
The third step was testing the reliability of the coding process.A second experienced coder (blind to the project goals') independently screened a sample of representative units (n = 1666).The second coder was given the complete matrix of previously created categories and asked to match them with the original coding units, following a top-down (deductive) approach.According to Cohen's kappa, the intercoder agreement was good at the manifest level (κ = .75)and excellent at the latent level (κ = .93).Coders solved discrepant groupings by agreement.This process allowed the identification of meaning overlaps in some categories, which were reorganized or relabeled.Coders agreed that 68 coding units were doubtful (e.g., explorador [explorer]) or meaningless (e.g., Grand Canyon), which were thus dropped out.
Selection of representative attributes.We analyzed the frequency of participants who generated, at least once, each attribute across the six conditions (coded as: 1 = attribute generated, 0 = attribute not generated).We then selected the most representative or prototypical attributes for each label by using a selection formula adapted from Katz and Braly (1933), i.e., we selected the minimum number of attributes required to account for at least 50% of frequencies sum per condition.
Comparison of representative attributes across conditions.We performed chisquare tests of homogeneity to determine whether frequency counts of each representative attribute were distributed identically across the six conditions (i.e., labels).Post hoc pairwise comparisons were conducted using multiple z-tests of two proportions with Bonferroni correction.Whenever the data violated the sample size adequacy assumption of the chisquare test of homogeneity (i.e., expected cell count less than five), we used Fisher's exact test (2 x c).In these cases, post hoc analysis involved pairwise comparisons using multiple Fisher's exact tests (2 x 2) with Bonferroni correction (statistical significance was accepted at p < .003).

Results
The coding procedures resulted in 170 thematic attributes distributed across the six conditions.A mean of 10.38 attributes were generated per participant (SD = 6.10; min = 1; max = 33).The procedure of selection of the most representative attributes resulted in a final set of 65 attributes 2 , representing the prototypical meaning of each label (Table 1).For example, to describe the label "all humans everywhere" we obtained a total of 124 attributes from the coding procedure, of which 26 attributes were selected, representing 50% of frequencies sum.
Results of omnibus tests revealed a statistically significant difference in the frequency of participants who generated 40 out of the 65 representative attributes depending on the label used (p < .05).Post hoc tests were performed to examine differences between labels.For example (Table 2), 46.7% participants used mobility to describe "citizens of the world" compared to 30.8% participants who used it to describe "global citizens", 27% to describe "people all over the world", 11.4% to describe "people from different countries around the world", 10.8% to describe "all humans everywhere", and 7.9% to describe "members of world community".For this attribute, a statistically significant difference in proportions was shown by chi-square test of homogeneity, p = .000.Post hoc analysis involved pairwise comparisons (between 15 possible pairs, considering 6 conditions) using the z-test of two proportions with a Bonferroni correction (statistical significance was accepted at p < .00333).
As indicated by the different letters in Table 2, mobility was significantly more used to describe "citizens of the world" than "people from different countries around the world", "members of world community" or "all humans everywhere", but no differences were found 2 Exemplars of attributes are available in online supporting information.
on the proportion of participants who use it to describe "citizens of the world", "global citizens" and "people all over the world".However, our interest lies less in the detail of each individual comparisons than in the extent to which labels were overall conceptually similar to or different from each other.This is best assessed by the total number of significant differences in pairwise comparisons: the lower that number, the higher the conceptual similarity between the pair of labels, and viceversa.The number of significant differences in pairwise comparisons between each condition is reported in Table 3.
According to these criteria, we identified the three most conceptually similar pairs of labels: a) "people all over the world" & "people from different countries around the world"; b) "global citizens" & "citizens of the world"; and, c) "global citizens" & "members of the world community".Post hoc pairwise comparisons revealed no significant differences between these pairs of labels in any attribute, suggesting a high conceptual overlap between them.Conversely, using the same criteria, we identified the three most conceptually different pairs of labels: d) "people from different countries around the world" & "citizens of the world", which differed in 10 attributes; e) "all humans everywhere" & "citizens of the world", which differed in 9 attributes; and, f) "people from different countries around the world" & "global citizens", which differed in 7 attributes, suggesting a high conceptual difference between them.
As predicted, then, the analyses showed that all-inclusive labels have different meanings according to laypeople's perception.However, it was also possible to identify some conceptual overlap between some of them.

Testing two new higher-order categories
To help make sense of the data, and based on the pattern of similarities and differences, we decided to aggregate the 6 different labels into two higher-order categories, namely, humanness-oriented labels vs. global citizenship-oriented labels (Figure 1).
We then tested the extent to which this higher-order organization fitted the data by replicating the previously conducted chi-square tests of homogeneity, but this time using only those two higher-order categories (Table 4).
Considering only the significant results, all emotional and physical attributes were significantly more activated by humanness-oriented labels.By contrast, most attitudinal and intellectual attributes, such as openness, cosmopolitanism, mobility, take risks and learning and knowledge were significantly more activated by global citizenship-oriented labels.
Values such as diversity, homogeneity, tradition and spirituality were significantly more activated by humanness-oriented labels, whereas concern for progress, tolerance, valuing technological/scientific development, concern for others' well-being and responsibility were significantly more activated by global citizenship-oriented labels.Overall, as predicted, results showed that the two sets of labels activated significant different sets of concepts, pointing to conceptually different prototypes.

Discussion and conclusions
This study examined whether different labels for all-inclusive superordinate identities activate the same contents in laypeople conceptualizations.
As expected, our data shows that no label had a clear-cut definition shared by all participants (no single attribute was mentioned by every participant, and only two attributes were listed by more than a half, i.e., diversity and multiculturalism to describe "people from different countries around the world", and diversity to describe "people all over the world").
In line with Morton and Postmes (2011b), these results support the idea that all-inclusive identities are very abstract and, therefore, highly fluid.However, this does not mean that allinclusive categories should be understood as empty shells.Indeed, our data also shows that, over and above individual variability, different labels activated substantially different sociopsychological content (or prototype) across participants-although some categories also strongly overlapped.In line with previous prototype analysis of social categories (e.g., Kinsella, et al., 2015), the fact that some attributes were mentioned more frequently than others, might indicate that no rigid boundaries appear to exist within all-inclusive superordinate identities, and thus these might be better represented as a prototype.
Before further discussing these findings, however, some important limitations should be mentioned.First, given the nature of our sample, our results might well be particular to the Portuguese population or to the population of a European country.The topic of endorsement of all-inclusive identities in Portugal is understudied.However, Pichler (2009) portrayed the Portuguese population as the third most cosmopolitan oriented (i.e.openness towards immigration; concern about humankind) in Europe.Certainly, replicating the current study with populations deemed less cosmopolitan would be important to reinforce, or to qualify, our conclusions.It is also worth noting that the different labels used in this study may carry different connotations in different languages.For example, according to experts consulted by McFarland (2017), the term 'world citizen' in English to U.S. citizens carries a "more proactive, participatory connotation of citizenship than does "weltbürger" [in German], which connotes more of a cosmopolitan sense of "wise in the ways of the world"" (p. 7).However, since our main claim is that the meaning of all-inclusive categories may vary, not that it necessarily does so in a systematic way across all contexts and populations, a single case already suffices to make the point.
Second, we freely acknowledge that our method of quantitative content analyses relies on "gross categorization", a feature that has been criticized for leading to a loss of variability and meaning (e.g., Potter & Wetherell, 1987).In particular, splitting statements in different coding units entails losing the potential meaning they possess as a whole, as well as abstracting them from their broader context.Likewise, the decision about what counts as one instance or as several is always in part an interpretative process.Despite these limits, however, we were able to identify meaningful differences between different category labels.
Thus, an alternative method that would preserve more variability and lose less meaning should a fortiori reinforce that conclusion.Undoubtedly, applying such method would prove quite useful to further enlighten the nature of these differences, particularly those within the two macro-categories (humanness-vs, global citizenship-oriented labels).Nevertheless, since our primary goal in this paper was to make a general claim of difference that could be sustained quantitatively, the present method was both less costly and more appropriate to that purpose.
Third, instructions to participants were adapted from Fehr & Russell (1984), which analyzed the concept of emotion.As a result, the example provided to participants was an emotion (fear) rather than a social category.Although this concerned all participants equally and is unlikely to have affected their answers in a significant way without further context, for future studies, we recommend the replication of Kinsella, and colleagues (2015, study 1) procedure, which was designed and adapted to analyze a social category.
Overall, the current study extends previous studies in two ways.First, it provides the first direct comparison of the content of several all-inclusive superordinate categories, as they are psychologically represented by people themselves, rather than defined a priori by scholars and researchers.Second, it challenges the idea that all forms of all-inclusive identities can (or should) be treated as a single social psychological reality.Methodologically, this implies that an undifferentiating use of these labels in research measures may be best avoided.
In particular, while our findings support the statement that identification with humanity and global citizenship partly share a common meaning (McFarland, et al., 2019), they also provide evidence to the claim that they might be better represented as separate constructs.Indeed, our results show that humanness-oriented labels (i.e., "all humans everywhere", "people all over the world", "people from different countries around the world") activated more emotional (e.g., affection; happiness), physical (e.g., human nature of world population; physical appearance) and social-relational attributes (e.g., living around the world; need of family bonds) compared to global citizenship-oriented labels (i.e., "global citizens", "citizens of the world" and "members of world community").By contrast, global citizenship-oriented labels activated more attitudinal (e.g., mobility; cosmopolitanism) and intellectual traits (e.g., learning and knowledge) compared to humanness-oriented labels.
A tempting interpretation of these findings is that humanness-oriented labels mainly evoke biological and socio-cultural aspects that people ostensibly share as members of the human species, and that thereby supersede naturalized (e.g., ethnic) and/or cultural divisions.
By contrast, citizenship-oriented labels generally evoke attitudinal and intellectual aspects that people share as members of a common global political community of citizens, superseding political (i.e., mainly national) divisions.This interpretation mirrors existing distinctions between ethnic and civic nationalisms (Kohn, 1944), civic and cultural identities (Bruter, 2003), essence-based and agency-based groups (Brewer, Hong, & Li, 2004), or heritage-based and project-based identities (La Barbera, 2015).It also implies that, as is the case with those other distinctions, the difference between humanness-and citizenshiporiented labels is likely to impact both on the contexts in which they are invoked and their subsequent effect on intergroup behavior and attitudes.For instance, as our opening example suggested, humanness-oriented labels are more likely to be invoked (and to lead to more effective mobilization) when an ingroup needs help fulfilling perceived basic human needs (e.g., hunger).By contrast, invoking global citizenship-oriented labels may be more appropriate and effective when it is a matter of political rights (e.g.migrants' right to vote).
In support of this view, Reysen and Katzarska-Miller (2017) found that global citizenship identification (using the label global citizen) was related to peace values and attitudes (e.g., concern for human rights, responsibility, support for diplomacy, positive attitudes toward peace) above and beyond human identification (using the label humans) -a finding that is highly consistent with our own (i.e., the attributes "responsibility", and other attributes related to peace, such as "tolerance", "concern for progress", "development and concern for other's wellbeing" emerged more to describe global citizenship-oriented labels than humanness-oriented labels).
Conversely, the more essentialist meaning associated with humanness-oriented labels might contribute to explain some of the negative effects of appealing to common humanity that have been found by studies adopting a situational approach.For instance, essentialist beliefs about human nature and violence have been shown to help harm perpetrators deflecting responsibility and avoid guilting (Morton & Postmes, 2011a), and lead victims to forgive perpetrators while giving up on collective action (Greenaway, et al., 2011).However, our goal is certainly not to suggest a simple equation between global citizenship vs. humanness-oriented labels and positive vs. negative effects on intergroup relations.This would be simplistic insofar as no simple equation exists between essentialism vs. de-essentialism and desirable vs. undesirable outcomes (Morton, Hornsey, & Postmes, 2009;Verkuyten, 2006).Furthermore, Hamer and colleagues (2019) infer that, so far, studies have shown negative effects of endorsing all-inclusive identities only with situational activation, not when it is measured and analyzed as an individual difference.If situational activation is indeed necessary to trigger negative effects, the implication is that such effects should also vary as a function of the particular meaning attributed to the category in that context, whether it focuses on common humanity or global citizenship.Moreover, future studies could investigate further other noteworthy differences between categories, for instance the differential impact of invoking labels that emphasize commonalities/unity (e.g., people all over the world) and those that underline differences/divisions (e.g., people from different countries around the world).
Overall, it is clear that the implications of our results in terms of the outcomes of identification with, or the situational activation of, all-inclusive superordinate categories would require further study.The present study does not-nor did it aim to-provide direct evidence to that question.Instead, it shows that there are potentially significant differences in the semantic universes conjured by the notions of common humanity and global citizenship.
While context might make those differences more or less important compared to the overlap that also exist between the two, it is highly unlikely that they are unimportant in every context in terms of their impact on behavior and attitudes.

Table 1
Number of prototypical attributes selected per labelNote.AHE -All Humans Everywhere; PAOW -People All Over the World; PDCAW -People from Different Countries Around the World; GC -Global Citizens; CW -Citizens of the World; MWC -Members of World Community; Different letters show significant differences between labels; (*) Fisher's exact test.If significant (p < .05;p-values at bold type), post hoc analysis involved pairwise comparisons using multiple Fisher's exact tests with a Bonferroni correction (statistical significance was accepted at p < .003);(**) Chi-square test of homogeneity.If significant (p < .05;p-values at bold type), post hoc analysis involved pairwise comparisons using the z-test of two proportions with a Bonferroni correction (statistical significance was accepted at p < .003).

Table 4
Cross-tabulation of two higher-order categories of labels and percentage of representative attributes