Moral duty or moral defence? The effects of perceiving shared humanity with the victims of ingroup perpetrated harm

Abstract

Previous theory and research suggests that perceiving shared humanity with others should be a positive force for intergroup relations. The present research considers the alternative possibility, that notions of shared humanity might protect people from feelings of guilt over ingroup perpetrated harm by obscuring the ingroup's unique role in these events. Consistent with this idea, Study 1 (N = 58) found that perceiving shared humanity with a harmed outgroup was associated with less guilt and stronger expectations of forgiveness among members of the perpetrator group. Study 2 (N = 52) demonstrated that these effects only occurred when the moral integrity of the ingroup was open to question. When ingroup morality was instead secure, defensive use of humanity was not apparent. Together, these studies suggest that perceiving harmful ingroup actions as ‘only human’ can sometimes be a moral defence that absolves group members of feelings of responsibility for wrongdoing. Copyright © 2010 John Wiley & Sons, Ltd.

History is important. Problematic histories fuel intergroup conflicts (Barkan, 2000; Pennebaker, 1997) and awareness of past victimization can spill over into heightened vigilance in the present (Wohl & Branscombe, 2008). Arriving at a common understanding of history is a crucial step in the process of reconciliation between groups. However, groups are more likely to justify their past actions than to admit to wrongdoing (Baumeister & Hastings, 1997; Doosje & Branscombe, 2003; Doosje, Branscombe, Spears, & Manstead, 1998; Leach, Snider, & Iyer, 2002). The defensiveness that surrounds problematic episodes in history raises the question of how it is possible for groups to move beyond their past and into a more positive future.

Wohl and Branscombe (2005) recently approached this question through the lens of self-categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Specifically, they predicted that representatives of a victimized group should be more forgiving of past harm when they perceive their group and the perpetrator group in terms of their common humanity: A single superordinate category that includes both groups. Consistent with this prediction, Jewish American and Native Canadian participants were more forgiving of Germans and White Canadians, respectively, when their victimization was framed as an example of ‘what humans do to each other’ (i.e. making the human category salient) rather than what one specific outgroup did to the ingroup (i.e. making the intergroup context salient).

We aimed to extend these ideas by considering the effects of perceiving in- and outgroups in terms of shared humanity within groups that have perpetrated harm against others. Just as human categorizations can facilitate forgiveness among members of the victim group, we argue that human categorizations might have implications for forgiveness among the perpetrator group. More specifically, we argue that by seeing past actions as ‘only human’, representatives of the perpetrator group may find it easier to forgive themselves. Self-forgiveness should render the further experience and expression of remorse unnecessary. If this is the case, encouraging groups to focus on their shared humanity may not always promote reconciliation between groups, at least when the perspective of the perpetrator group is considered. Before presenting studies that explore this possibility, we briefly review literature that has discussed the roles of shared group membership and perceptions of humanity in intergroup relations—both positive and negative.

POSITIVE EFFECTS OF SHARED HUMANITY

Self-categorization theory argues that the self-concept is structured hierarchically in terms of overlapping self-categories—ranging from unique personal attributes (personal identity), through to group memberships and social categories (social identities), to an all-inclusive human identity (Turner, 1982). The contextual salience of each level of identity has important implications for thoughts and feelings about the self and others (e.g. Schmitt, Branscombe, Silvia, Garcia, & Spears, 2006). Specifically, when individuals perceive themselves in terms of group membership (i.e. social identity is salient), they are prone to perceptually and behaviourally favour their ingroup over relevant outgroups.

Given that salient intergroup categorizations contribute to intergroup conflict, it has been argued that moving people away from a two-group representation (i.e. ‘us’ versus ‘them’), and toward a single common identity that includes ingroup and outgroup (i.e. ‘we’), should have positive implications for conflict reduction. Focusing on a common identity should lead people to extend the favouritism and trust they previously reserved for the ingroup to all within the new categorical boundary, including the former outgroup (Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993). This basic prediction has been supported in a variety of contexts (Gaertner & Dovidio, 2000), and provides the conceptual basis for predicting positive effects of human categorization (the ultimate super-ordinate category) on intergroup forgiveness.

Extending their work on responses to human categorization among members of victimized groups, Wohl, Branscombe, and Klar (2006) suggested that a focus on shared humanity might also have positive implications for representatives of perpetrator groups. Specifically, they predicted that when members of the perpetrator group categorize themselves and their victims within a common boundary (i.e. seeing them as equally human) more intense feelings of remorse and restorative intentions should follow. The reasoning behind this prediction is two-fold. First, if salience of group membership contributes to defensiveness about ingroup wrongdoing (Branscombe & Miron, 2004; Doosje, Branscombe, Spears, & Manstead, 2004), reducing the salience of intergroup distinctions should reduce this defensiveness. Second, common category membership should inspire the application of common principles of justice to both in and outgroups (Wenzel, 2004). These common standards for behaviour render harm against the outgroup less legitimate and less justifiable (i.e. more immoral).

This analysis accords with work on dehumanization and infrahumanization. Research in this domain has demonstrated that people tend to associate humanity more strongly with their ingroup than outgroups, both subtly (e.g. Leyens et al., 2000; see also Haslam, 2006) and in more explicit ways (e.g. Viki, Winchester, Titshall, Chisango, Pina, & Russell, 2006). Furthermore, perceiving others as less-than-human has been found to serve as one mechanism of moral disengagement (Bandura, 1999) that dis-inhibits aggressive action (Bandura, Underwood, & Fromson, 1975) or otherwise justifies harm done (e.g. Castano & Giner-Sorolla, 2006). Conversely, humanizing perceptions of others are thought to foster empathy and a sense of moral obligation, thereby restraining harmful action and compelling people to help rather than hurt. Thus both research on categorization effects and on dehumanization suggests that perceiving others as ‘equally human’ should be associated with more positive forms of intergroup behaviour.

NEGATIVE EFFECTS OF SHARED HUMANITY

Despite the above theory and research, there are reasons to expect that focusing on shared humanity may not always have positive effects. A number of studies have demonstrated that activating shared superordinate categories does not invariably lead to more harmonious subgroup relations. Instead, people sometimes display heightened intergroup bias following the activation of a shared superordinate category (e.g. Crisp, Stone, & Hall, 2006; Hornsey & Hogg, 2000), or in response to implicit claims to equal humanity when made by outgroup members (e.g. Vaes, Paladino, Castelli, Leyens, & Giovanazzi, 2003). One complication is that as categories become more and more inclusive (Europeans, Westerners and particularly ‘humans’), they also tend to become more abstract. Such abstract categories tend to be ill-defined—and what it means to be ‘human’ is highly malleable. Because of this it is still possible to imagine shared categories in ways that favour the ingroup (e.g. Mummendey & Wenzel, 1999; Wenzel, Mummendey, & Waldzus, 2007). Along similar lines, recent research within the common ingroup identity model suggests that members of dominant groups can draw on ideas of commonality to obscure differences of power and status between them and dominated groups, and thereby undermine challenges to the legitimacy of intergroup inequality (e.g. Dovidio, Gaertner, & Saguy, 2007; Saguy, Dovidio, & Pratto, 2008).

Applying these ideas to the context of the human category, it seems plausible that construing ingroup perpetrated harm in terms of the shared human category might sometimes be an effective means for obscuring the ingroup's unique role in such events, and that this in turn may limit individual feelings of group-based responsibility and guilt. Consistent with this, studies have demonstrated that perceptions of ingroup responsibility, and associated feelings of guilt, are most likely when people are focused on the ingroup's agency in acts of wrong-doing (e.g. Harth, Kessler, & Leach, 2008; Iyer, Leach, & Crosby, 2003; Leach, Iyer, & Pedersen, 2006). When alternative categorizations are available (e.g. humans rather than group members), feelings of responsibility and guilt should be less likely. Put simply, if our actions were ‘only human’ then we alone are not to blame.

Some dehumanization research also raises this possibility. Haslam et al. (Haslam, Bain, Douge, Lee, & Bastian, 2005) demonstrated that while people tend to see their own traits as more human than the traits possessed by others, this tendency is amplified for undesirable personal traits (Haslam & Bain, 2007). They suggest that seeing one's flaws and failings as ‘only human’ might be motivated by a desire to mitigate individual responsibility for these and thus minimize their impact on self-regard. It seems plausible that a similar process could play out in the domain of intergroup perceptions: People may sometimes prefer to see negative actions committed by their ingroup as typical of humans and thereby less worthy of guilt-feelings or further responsibility. This being the case, although humanizing outgroups might facilitate moral engagement (Bandura, 1999), humanizing ingroups might actually lead to moral disengagement. Indeed, in one study where ingroup and outgroup humanity were considered separately, these displayed opposing relationships with positive intergroup orientations (Zebel, Zimmerman, Viki, & Doosje, 2008). Specifically, whereas perceiving the humanity of a harmed outgroup was consistently positively related to the experience of group-based guilt and support for reparations, perceiving ingroup humanity typically displayed weak but negative relationships with these variables.

THE PRESENT RESEARCH

The present research explored how thinking in terms of shared humanity influences responses to harm committed by representatives of one's ingroup. The research reviewed above suggests two possible outcomes. When focused on their shared humanity with a victimized group, members of the perpetrator group might appraise their actions as more illegitimate and therefore experience more remorse over their actions (e.g. Wohl et al., 2006). In this way, common humanity may elicit a sense of ‘moral duty’ to outgroups—laying the foundation for reconciliation by eliciting expressions of remorse from representatives of the perpetrators as well as encouraging forgiveness by representatives of the victims. Conversely, when thinking in terms of the human category, members of a perpetrator group might find it easier to construe negative ingroup actions as reflective of their broader humanity, rather than indicative of the ingroup's own unique motivations. This, in turn, may undermine the experience and expression of remorse for past harm. In this sense, perceptions of shared humanity may act as a ‘moral defence’—that is, a shield that perpetrators can hide behind and use to minimize their own group's unique role in history. To explore these alternative possibilities, we conducted two studies. Our first study examined the effects of perceiving common humanity in the context of ingroup perpetrated harm. Study 2 then explored the role of threats to the moral image of the ingroup in eliciting defensive use of the human category.

STUDY 1

Our first study surveyed British participants' thoughts and feelings about historical wrongs perpetrated by their group, specifically their country's involvement in the trade of African slaves. To highlight questions of ongoing responsibility for these events, we emphasized that although the ingroup had benefited from slavery at the expense of Africans, to date government representatives had refused to either apologize for this aspect of history or attempt any kind of official redress. Against this backdrop, we assessed participants' feelings of guilt about their ingroup's behaviour and their expectations that others should forgive their ingroup for its past. Participants were also asked how they categorized the relationship between their own group and African nations—as two separate groups or as equal members of the human race. If perceptions of shared humanity imply a moral duty to outgroups, these should elicit stronger expressions of remorse over harm done (i.e. increase feelings of guilt). However, to the extent that shared humanity can be used as a moral defence perceptions of shared humanity should instead be negatively correlated with feelings of guilt and positively correlated with the expectation that the ingroup should simply be forgiven for its past.

METHOD

Participants

Fifty eight students were approached on a large university campus in the UK and volunteered to fill in a survey regarding opinions about historical events. All participants were British and not of African descent. This sample comprised 30 men and 28 women who ranged from 16 to 59 years old (mean age = 22.76, SD = 10.00).

Measures and Procedure

The questionnaire began with a summary of Britain's involvement in the slave trade (including information about the extent of the slave trade, the social and economic effect of this on African communities) and the fact that to date the British government had refused to offer any official apology for these events (which was true at the time of data collection). Participants were then asked for their own thoughts about these issues. Specifically we assessed collective guilt about the historical behaviour of ingroup members, expectations of forgiveness and categorizations of the relationship between British and African nations. Responses to all items were measured on 9-point scales (ranging from 1 = Strongly disagree to 9 = Strongly agree)

Guilt

Three items adapted from Branscombe, Slugoski, and Kappen (2004) assessed the extent to which participants experienced guilt based on Britain's involvement in the slave trade: For example, participants indicated their agreement with statements like ‘I can easily feel guilty for the bad outcomes brought about by Britain's involvement in the slave trade’. These items were combined and averaged into single index on which higher scores indicated greater acceptance of collective guilt (α = .87)

Expectations of Forgiveness

Three items assessed the degree to which participants believed that contemporary Britons should be forgiven for the actions of their ancestors: ‘Britain has done enough to repair any harm inflicted on others in the past’, ‘It is not fair for people today to hold negative feelings toward British people for the way their group treated African slaves in the past’ and ‘British people living today should be forgiven for their ancestors’ involvement in the slave trade' (α = .71). Factor analysis confirmed that the items comprising this variable could be empirically distinguished from those assessing guilt.1

Categorization

Finally, two questions assessed participants' categorization of the relationship between British and African nations. Specifically we asked participants the extent to which they saw British and African nations as ‘two distinct groups working separately and against each other’ or as ‘a single group: Both equal members of the human race’. We reverse-scored the former and averaged these into a single index on which low scores indicated a two-groups representation and high scores indicated representation as a single group of equal humans (r = −.40, p = .002).2

RESULTS AND DISCUSSION

Preliminary analyses revealed no relationships between gender and any of the variables, rs < .20, ps > .14. To explore how perceptions of shared humanity influenced guilt and expected forgiveness, we examined the correlations among the variables. As can be seen in Table 1, perceiving British and African nations as equally human was associated with significantly less intense feelings of collective guilt, and a significantly stronger expectation that the ingroup should be forgiven for its past. Put simply, a focus on shared humanity did not seem to inspire feelings of responsibility or expressions of remorse among members of the perpetrator group. Instead, perceptions of shared humanity seemed to function as a moral defence which absolved group members from feelings of guilt and ongoing responsibility.

Table 1. Study 1: Means, standard deviations and variable inter-correlations

	Mean	SD	1.	2.
Note: Variables measured on 9-point scales. * p < .05; ** p < .005.
1. Shared humanity	6.05	1.59
2. Guilt	4.44	2.17	−.30*
3. Expect forgiveness	5.66	1.38	.29*	−.40**

STUDY 2

In contrast to the patterns observed in Study 1, others have suggested that thinking in terms of the shared human category might elicit a sense of moral duty to outgroups—that is, prompting members of a perpetrator group to acknowledge immoral actions and eliciting the desire to make amends with their victims (Wohl et al., 2006). To the extent that the results of Study 1 do, in fact, reflect a defensive use of the human category, this should be particularly apparent when there is something to defend against. Thus context may determine which of these alternative possibilities are observed. In particular, we reasoned that members of a perpetrator group may become motivated to use ideas of humanity defensively when the moral image of their group may be open to question by others. Some work has established that questions of morality are central to how people perceive ingroups (Leach, Ellemers, & Barreto, 2007). Given this, threats to the moral image of one's group should be particularly concerning and particularly likely to prompt a defensive response. Conversely, when the moral image of one's ingroup is secure, or at least unlikely to be questioned by others, shared humanity might operate more like a moral duty rather than a moral defence, along the lines originally suggested by Wohl et al. (2006).

To test this idea, we again asked British participants to contemplate their group's involvement in the slave trade and the effects of this on Africans today and then assessed feelings of guilt, expectations of forgiveness and their categorization of in- and outgroups as equally human. To vary the presence of concerns about the ingroup's moral image, we introduced a manipulation of the ingroup's behaviour in relation to its past. Half the participants were told that their group had refused to apologize for its involvement in slavery, as was the case in Study 1. The other half was instead told that the ingroup would soon be issuing an official apology.

We reasoned that in the current global context, in which intergroup apologies are increasingly common and perceived to be appropriate (Barkan, 2000), information that one's group refused to apologize for wrongdoing should raise participants' concerns about the moral image of the ingroup. Refusing to apologize leaves the ingroup open to the accusation of immoral behaviour by others. Conversely, through issuing an apology, the ingroup is conforming to current trends and is seen to be behaving morally. Regardless of how individuals within the group may feel about apologies, this protects the ingroup from potential criticism by outsiders. Consistent with these ideas, research on interpersonal conflict suggests that not apologizing has negative consequences for perceptions of transgressors (Scher & Darley, 1997). Although there is less direct evidence for the effects of intergroup apologies (Philpot & Hornsey, 2008), research suggests that compared with representatives of the victimized group, representatives of the perpetrator group react more positively once an intergroup apology is issued, resulting in elevated perceptions of their ingroup (see Blatz, Schumann, & Ross, 2009). This suggests that intergroup apologies, though often resisted, can be affirming.

We predicted that the significant correlations between perceptions of shared humanity and less intense feelings of guilt and stronger expectations of forgiveness observed in Study 1 would be replicated in the no apology condition of this study, indicating the use of humanity as a moral defence. In comparison, we expected that these relationships would reverse in the apology condition, indicating the operation of humanity as a moral duty.

METHOD

Pilot Study

To validate our assumptions about the moral implications of refusing to apologize we first conducted a pilot study. British participants drawn from the same population as the main study (N = 67) were given summary information about their ingroup's involvement in the slave trade. At the end of this summary, half the participants (n = 37) were reminded that the British government had refused to offer an official apology for Britain's role in slavery. The other half (n = 30) was told that the government had decided to issue an official apology in the near future. At the time that this study was conducted, both these statements were based in fact.

Immediately after the manipulation, participants were asked to rate their government's behaviour on five 9-point semantic differential scales, anchored incorrect–correct, foolish-wise, immoral–moral, unjust–just, crazy–sane (α = .91). Next, participants were asked to indicate their agreement with three items designed to assess the perceived moral image of their group to the outside world: ‘Overall, Britain is viewed positively within the international community’; ‘As a nation, Britain is respected’ and ‘Other countries look up to Britain as an honourable and principled society’ (1 = Strongly disagree, 9 = Strongly agree; α = .94). The results indicated that not apologizing was rated less positively than apologizing, (means = 4.50 and 6.57; SDs = .26 and .29), F_(1,66) = 28.74, p < .001, η² = .31, and that not apologizing had negative consequences for the perceived moral image of their group (means = 4.97 and 5.87; SDs = .30 and .33), F_(1,66) = 3.97, p = .05, η² = .06. Thus, the pilot study confirms our assumption that refusing to apologize for past harm is both negative and raises concerns about the moral image of the ingroup.

Main Study

Participants

Fifty three students at a large British university responded to an email inviting them to participate in an online survey regarding opinions about historical events. All participants were British and not of African descent. In their written comments, one participant directly questioned the veracity of the apology manipulation. This person was therefore excluded from analyses. The remaining 52 participants comprised 18 men and 34 women who ranged in age from 18 to 26-years-old (mean age = 19.87, SD = 1.47). Participants were entered into a prize draw to win one of four gift vouchers worth £20 each. All responses were given on a 7-point scale (1 = Strongly disagree, 7 = Strongly agree).

Design and Procedure

As in Study 1, participants were provided with a summary of Britain's involvement in the slave trade and were asked the extent to which they felt guilty about their group's past (three items; α = .84) and their expectations that their group should be forgiven for that past (three items; α = .66).3 Finally, we assessed perceptions of British and African nations as two separate groups, or ‘equally human’. Again, these two items were significantly negatively correlated (r = −.48, p < .001) and after reverse scoring the first item, we combined these into a single index on which low scores indicated a two-group representation and high scores a representation in terms of equal humanity.

Between the summary of Britain's past and the questionnaire items we inserted our manipulation of the ingroup's behaviour in relation to its past. As described in the Pilot Study, half the participants (n = 25) were reminded that the British government had refused to offer an official apology for Britain's role in slavery, whereas the other half (n = 27) was told that the government had decided to issue an official apology in the near future. To check that participants had attended to this manipulation, they were asked to indicate their agreement with the statement ‘Britain has apologized for its role in the slave trade’.

RESULTS

Preliminary Checks

Analysis of the manipulation check item revealed that participants in the apology condition (M = 5.12) were more likely to agree that their group had apologized relative to participants in the no apology condition (M = 3.64), t₍₄₉₎ = 3.01, p = .004. Preliminary checks also revealed that perceptions of shared humanity were independent of the apology manipulation, t < 1. Thus it was appropriate to treat these as independent variables in the main analysis. Finally, there were no relationships between gender and any of the variables, rs < .15, ps > .28, and controlling for gender did not affect the pattern of results. Therefore results are reported without controlling for gender. Means, standard deviations and correlations among the measured variables are presented in Table 2.

Table 2. Study 2: Means, standard deviations and variable inter-correlations

	Mean	SD	1.	2.	3.
Note: Variables measured on 7-point scales. *** p < .001.
1. Shared humanity	5.25	1.20
2. Guilt	3.49	1.56	−.07
3. Expect forgiveness	4.50	1.05	.19	−.56***

Main Analysis

Our hypothesis was that perceiving shared humanity would undermine the experience and expression of remorse in the presence of threats to the ingroup's moral image (i.e. the no apology condition), but that this might facilitate expressions of remorse in the absence of moral concerns (i.e. the apology condition). To test this hypothesis, we conducted a series of hierarchical regression analyses in which we entered experimental condition (dummy coded: −1 = no apology, 1 = apology) and human categorization (centered) at the first step, followed by the interaction term at Step 2. Significant interactions were explored according to recommendations by Aiken and West (1991).

The analysis performed on expected forgiveness revealed a significant effect of the interaction term alone, R²Δ = .08, FΔ_(1,48) = 4.59, β = −.29, p = .04. As can be seen in Figure 1, in the apology condition there was no relationship between perceptions of shared humanity and expected forgiveness, β = −.09, t₍₄₈₎ = .46, p = .67. However, consistent with the moral defence hypothesis, in the no apology condition, there was a significant positive relationship between shared humanity and expected forgiveness, β = .50, t₍₄₈₎ = 2.50, p = .02.

Figure 1. Study 2: The interaction between perceptions of shared humanity and condition on expectations of forgiveness

The analysis of guilt also revealed a significant effect of the interaction term alone, R²Δ = .09, FΔ_(1,48) = 4.63, β = .30, p = .04. As can be seen in Figure 2, this was due to the absence of a relationship between perceptions of shared humanity and feelings of guilt in the apology condition, β = .20, t₍₄₈₎ = 1.08, p = .28, and a marginally significant negative relationship between these variables in the no apology condition, β = −.39, t₍₄₉₎ = 1.93, p = .059.4

Figure 2. Study 2: The interaction between perceptions of shared humanity and condition on guilt

DISCUSSION

The results of this study suggest that ideas about shared humanity can be used to deflect notions of guilt, and that this may be especially likely when the morality of the ingroup is under threat or otherwise open to question. When participants were reminded that their group had refused to apologize for its involvement in the slave trade, action that is likely to have compromised the ingroup's moral image (see Pilot Study), there were significant associations between perceptions of shared humanity with Africans and both feelings of guilt and expectations of forgiveness. Specifically, participants who perceived African and British nations as ‘equally human’ were less inclined to experience guilt, significantly more inclined to expect that their group should be forgiven for its past. The patterns replicate the findings of Study 1.

In contrast, when the ingroup's moral image was more secure (i.e. in the apology condition) we did not find this defensive pattern. Although this indicates that defensive use of humanity subsides in the absence of moral threats, contrary to our original expectations perceptions of shared humanity did not function as a moral duty: The results did not show significantly more intense feelings of guilt as a function of perceiving shared humanity, along the lines suggested by Wohl et al. (2006). One reason for the absence of more positive effects may be due to the details of our manipulation. Rather than telling participants that their group had apologized, we told them that their group would apologize in the near future. We chose this form of manipulation because it reflected the political reality of the time. Human categorizations may have been more effective in eliciting positive responses, rather than simply removing negative ones, if participants had been told that the apology had actually been issued, or if they had been presented with a stronger moral affirmation of their ingroup. It would be useful for future research to consider this possibility, or ways in which concerns about the ingroup's moral image might be manipulated more directly.

GENERAL DISCUSSION

The aim of this paper was to examine the effects of perceiving shared humanity on responses to intergroup harm from the perspective of representatives of the perpetrator group. Previous theory and research (Gaertner & Dovidio, 2000; Gaertner et al., 1993; Wohl & Branscombe, 2005; Wohl et al., 2006) suggest that a focus on shared humanity should promote forgiveness and reconciliation between groups. Although this possibility has been supported for members of historically victimized groups (Wohl & Branscombe, 2005), to date research has not explored the effects of focusing on shared humanity among representatives of perpetrator groups. The present research shows that effects on perpetrator group members are not straightforwardly positive. Instead, members of a perpetrator group seem able to hide behind their common humanity.

Rather than encouraging the experience and expression of remorse (Wohl et al., 2006), the two studies presented here demonstrate that a focus on shared humanity can undermine these. Importantly, these negative effects of perceiving shared humanity were evident only when the moral integrity of the group was open to question. When the group's moral integrity was instead assured (Study 2), this pattern of absolution via humanity was not evident. This supports our contention that notions of shared humanity can be a moral defence—that is a means of deflecting responsibility for harm and avoiding feelings of guilt when the moral integrity of the ingroup is likely to be questioned by others.

The present findings seem to run counter to conventional wisdom about the role of humanizing perceptions in intergroup relations. Indeed, previous research has established that dehumanization of outgroups can be a mechanism of moral disengagement (e.g. Bandura, 1999) that facilitates harmful action (e.g. Bandura et al., 1975) or justify previous harm done (e.g. Castano & Giner-Sorolla, 2006). Given this discrepancy, it is useful to consider how the present research differs from the typical focus of research on dehumanization. One important difference is that our research primarily focuses on perceptions of ingroups and responses to ingroup actions. Research on dehumanization, however, typically focuses on perceptions of outgroups and support for action directed towards them. It seems possible that these different foci of judgment and attention (i.e. ingroup versus outgroup) might determine whether thinking in terms of the human category leads to justification of ingroup harm (via humanization of the ingroup) or heightened perceptions of illegitimacy and intentions to repair harm (via humanization of the outgroup). It would be interesting for future research to disentangle precisely who is being humanized when the human category is in focus, and whether this has different implications for intergroup responses (e.g. Zebel et al., 2008).

Additional research is needed to explore the specific process behind the effects we have identified. We assume that the patterns observed in these studies were guided by desires to preserve ingroup morality and/or avoid ingroup responsibility (see also Haslam & Bain, 2007). While the conditional relationships identified in Study 2 support the notion of morality-motivated defensiveness, the precise role of such motives remains to be tested. In addition, although we have identified conditions under which ideas of shared humanity is used defensively, it would be useful for future research consider how this defence is achieved (i.e. the underlying mechanism). If, as we suggest, people are saying ‘we were only human’ and therefore not to blame, their response seems to rest on a particular image of humanity, namely that human beings are flawed creatures and harmful actions are somehow normative. Thus in addition to considering the context within which people consider their shared humanity with outgroups, it would seem useful to examine how they construe the stereotypic content of the human category in order to permit this defensive use. If the human category is associated with negative content, then harmful actions are likely to be seen as normal rather than troubling, perhaps regardless of whether these actions are performed by the ingroup or an outgroup (Morton & Postmes, 2007).

Although our findings qualify some of the more optimistic conclusions from past research in this domain, it is also important to continue to search for conditions under which shared humanity might instead contribute to a sense of moral duty that facilitates the acceptance of responsibility and expressions of remorse for the negative actions of fellow group members. Affirming the moral integrity of ingroups may undermine some of the defensive tendencies observed in this research. However, this alone was not sufficient to fully reverse the negative effects of shared humanity on perpetrator group members. Accordingly, additional factors may be necessary for this to happen—for example, instilling a positive image of human nature that de-legitimizes harmful intergroup actions.

CONCLUSIONS

Our research suggests that attempts to resolve conflict by encouraging antagonistic groups to focus on their shared humanity might be naïve, or indeed counterproductive—particularly if the perspective of perpetrator group members is considered. Shared humanity does not merely impose a moral duty, it can also be a moral defence. If harmful ingroup actions are construed in terms of humanity, then the ingroup itself is no longer to blame. To truly transform a negative relations between groups would seem to require more than each side brushing aside their differences as ‘only human’.

Acknowledgements

Thanks to Colin Leach and Michelle Ryan for helpful feedback on earlier drafts. This research was partially supported by an ESRC research fellowship (RES 000-27-0050) to Tom Postmes.

• 1

  Factor analysis using principal axis factoring with oblique rotation indicated two factors (eigenvalues > 1) which together accounted for 59.10% of variance. Each individual item loaded highest on its respective factor (>.61).

• 2

  Additional analyses revealed no differences between correlations with each dependent measure when the component items were examined separately, zs < 1.31, ps > .19. This further supports treatment of these two items as opposite ends of a single dimension of perceived shared humanity.

• 3

  Factor analysis using principal axis factoring with oblique rotation indicated two factors (eigenvalues > 1) which together accounted for 60.25% of variance. Each individual item loaded highest on its respective factor (>.39). Again, this suggests that guilt and expected forgiveness can be empirically distinguished.

• 4

  We also included two items measuring support for restitution (e.g. ‘British people living today should do more to repair the damages done to Africans in the past’). This measure could not be empirically distinguished from guilt, with which it was highly correlated (r = .72, p < .001). Given this, results for this measure are not reported; however, it is worth noting that the predicted interaction was also significant on this variable (p = .004) reflecting reduced support for restitution as a function of shared humanity in the no apology condition (p = .003).