1 Introduction

Intergroup conflict is ubiquitous across cultures, and is often regulated through some form of negotiation among representatives from both sides (De Dreu et al. 2014). Although representatives often help to resolve intergroup conflicts (e.g., Kelman 1995, 2005) and to establish more positive intergroup relations (Saygi et al. 2014), their behavior may also, deliberately or inadvertently, escalate the conflict and damage the relationships among competing groups.

Using a bottom-up approach to representative negotiation (De Dreu et al. 2014), we examine when and why cooperative behavior by the outgroup representative will improve or harm intergroup relations, including constructive behavioral tendencies defined as group members’ tendency to help and protect the outgroup (e.g., Cuddy et al. 2008; Fiske et al. 1999, 2002). One example of bottom-up influences in representative negotiations is when Kelman (1997) applied the problem solving approach to the Cyprus conflict by bringing together political leaders, negotiators, public-political constituencies through personal contact, speeches and interviews which led participants to change their attitude about the conflict as well as the opposing group in a positive way (Kelman 1995, 2005; Davidson and Montville 1981; Druckman et al. 1988). However recent poll data from Cyprus suggest that even if the outgroup representative is cooperative in the negotiations and make concessions on important issues, citizens would still vote “No” in a likely referendum due to lack of trust towards the outgroup representative’ cooperative intentions (Seeds of Peace 2013). As such, we would like to identify the boundary conditions for the positive effects of cooperativeness on constructive behavioral tendencies.

Our point of departure is that in order for cooperative behavior by the outgroup representative to improve intergroup relations, first the representative needs to be perceived as a typical outgroup member to ensure that the perception of the representative will spill over to outgroup perceptions and influence intergroup relations positively. Second, the cooperative behavior of the outgroup representative should not raise any doubt about the true intentions of the outgroup. We identify two key characteristics of the outgroup representative that block the positive effects of cooperativeness on intergroup relations, namely (1) the representative’s peripherality to the represented group (Oakes et al. 1991) and (2) the representative’s competence.

1.1 Theoretical Model and Overview of Studies

In our overall conceptual model, we define two pathways that explain why the cooperative behavior of an outgroup representative may influence the ingroup members’ constructive behavioral tendencies towards the outgroup members who were not even involved in the negotiations. Furthermore we identified two boundary conditions that may block the positive effects of representative cooperativeness on constructive behavioral tendencies. In two studies we investigated whether or not the representative characteristics, namely representative prototypicality and competence will interact with the cooperative behavior of the outgroup representative to influence constructive behavioral tendencies towards the outgroup. In part 1 we focus on the moderating role of representative prototypicality and in the second part of our model we focused on the moderating role of representative competence. Please see Fig. 1 for our overall conceptual model wherein part 1 and 2 are tested in Study 1 and 2 respectively.

Fig. 1
figure 1

Overall conceptual model that was tested in two empirical studies. Please note that part 1 and 2 were tested in Study 1 and 2 respectively

Full size image

1.2 Bottom-up Effects of Representative Cooperation on Intergroup Relations

Representative negotiation involves a two-level situation, in which the negotiation among individual representatives is embedded in a broader intergroup context, and the broader intergroup relations in turn depend on the way representatives negotiate (e.g., Knowles and Bassett 1976; Lickel et al. 2000; Saygi et al. 2014). Especially the second component, or the degree to which representatives’ behaviors impact broader intergroup relations, is poorly understood (De Dreu et al. 2014).

Current insights rest on relatively indirect lines of inquiry and evidence. First, the study of problem solving workshops suggests that constructive, face-to-face discussions between representatives are vital for the development of constructive intergroup relations (Burton 1969; Druckman et al. 1988; Fisher 2006; Tam et al. 2009). For example, Kelman (1995, 2005) applied this approach to the Israeli–Palestinian conflict by bringing together political leaders, negotiators, and constituencies through personal contact, speeches, and interviews.Footnote 1 Consequently, cooperation between the representatives improved ingroup members’ view of the conflict as well as their perception of the entire outgroup (Davidson and Montville 1981; Fisher 2006). Second, the notion that a cooperative negotiation process among representatives can improve relations between groups resonates with the main principles of the intergroup contact theory (Pettigrew and Tropp 2006). The theory proposes that under four conditions - equal status among group members, common goals, intergroup cooperation, and support from authorities, a positive contact with an outgroup member can improve the perception of the entire outgroup (Allport 1954; Druckman 1968; Pettigrew 1998). For example, in the context of the Northern Ireland conflict between Catholics and Protestants, researchers examined the link between intergroup contact and action tendencies towards the outgroup. In addition to direct contact between members of opposing groups, they also examined “extended contact” which is defined as the cross-group friendship between another ingroup member and an outgroup member (Wright et al. 1997). In line with contact theory and the extended-contact hypothesis (Wright et al. 1997), they found that trust mediated the effect of both direct and extended contact on action tendencies towards the outgroup (Tam et al. 2009), suggesting that positive contact, such as through cooperation, can improve intergroup relations and resolve intergroup conflicts.

Although these related lines of inquiry are not directly concerned with representative negotiation, they suggest that cooperative communication during negotiation by representatives may positively influence intergroup relations. Negotiation processes are usually categorized as competitive or cooperative (Deutsch 1949; Pruitt 1967). When negotiation processes are competitive, parties use cheating, act hostile, d o not exchange information, and try to exploit one another. When negotiation processes are cooperative, there is exchange of information, parties make concessions, and refrain from hostile exchanges and deceitful tactics (Bazerman et al. 2008; De Dreu et al. 2007). Previous research mostly focused on the effect of cooperative/competitive communication on the negotiated outcome or outcome satisfaction. For example, when negotiation processes were competitive, even if the final outcome was the same, negotiators were less satisfied with the negotiated outcome when the negotiation process was marked by competitive rather than cooperative communication processes (e.g., Curhan et al. 2006; Druckman et al. 1988; Schei 2008; also see Van den Bos et al. 1998). Other research showed that communicating competitively increases fear and intergroup anxiety (Stephan and Stephan 1984; Wildschut et al. 2003), the cognitive salience of the intergroup boundary (Wilder 1986), and confirms the categorization of ingroup and outgroup (Tajfel and Turner 1979). Given this we would expect that competitive communication of the outgroup representative will negatively influence ingroup members’ outgroup perceptions regardless of the negotiated outcome.

Conversely, we would expect cooperative communication to positively influence ingroup members’ outgroup perceptions, regardless of the negotiated outcome. Indeed, in a first study of the effects of representative behavior on intergroup relations, Saygi et al. (2014) found that cooperative tendencies by the outgroup representative promoted positive perceptions of, and behavioral tendencies towards, the outgroup as a whole. From this it follows that when the outgroup representative is perceived as cooperative (rather than non-cooperative), ingroup members should engage in more constructive behavioral tendencies towards outgroup members, such as helping, protecting, and associating with the outgroup (e.g., Cuddy et al. 2008; Fiske et al. 1999, 2002).

However, as we argue in this paper, this should only be the case when the outgroup representative’s behavior is taken as a reliable signal of the entire outgroup’s intentions. Cooperation may not always be trusted (e.g., Kramer 1994, 2004a, b). Cooperation under certain conditions may be seen as manipulative and unauthentic, and could potentially backfire on the ingroup. We postulate here that one source of information that groups may use to infer the authenticity of the other group’s intentions is the type of representative the group chooses to send to represent itself. When groups send a representative that looks to be there for strategic purposes and to exact gains—such as a highly competent negotiator, the authenticity of the group’s cooperative behavior may be doubted, and the positive effects of cooperativeness on intergroup relations may be blocked. As such, it’s critical for groups to know which representative to send if they want to cooperate with the outgroup and want this cooperation to be seen as sincere. We therefore offer an important contribution to the literature on intergroup negotiations by identifying characteristics of a representative, which may make groups see cooperative behavior by the representative as being more or less authentic.

1.3 When Cooperation Backfires: The Role of Perceived Negotiator Characteristics

Both problem-solving workshops and contact theory suggest that a group member can create a positive image for the entire group by behaving in a cooperative manner during an exchange with the opposing group, which may in turn ameliorate intergroup relations (e.g., Brown et al. 1999). However, such member-to-group generalization rests on the assumption that the representing group member actually represents cooperative intentions of the entire outgroup. Signs that contradict such an assumption may mitigate generalization processes and abolish the positive effects of cooperative representative behavior on outgroup perceptions. One such representative characteristic that may block generalization processes is the representative’s perceived peripherality to the represented group (Oakes et al. 1991). Another such characteristic is the representative’s perceived competence. We explain below why these two perceived negotiator characteristics may be critical in determining whether or not representative cooperation is seen as authentic and representative of outgroup intentions.

1.3.1 Representative Prototypicality

In Study 1 we examined representative-outgroup generalization effects as a function of the outgroup representative’s prototypicality. According to self-categorization theory (Turner et al. 1987), group members vary in the extent to which they share common characteristics of their group that differentiate the group from other groups. Some group members are more prototypical, in that they possess more of those common characteristics, and hence they can be considered better examples of the group than others (Hogg et al. 1993). Group members who strongly match group prototypes can be referred to as prototypical group members and those who are less prototypical examples of their group can be referred to as peripheral group members (e.g., Jetten et al. 2002, 2003, 2006; Van Knippenberg 2000). Accordingly, representatives in intergroup negotiations may vary in the degree to which they are prototypical or peripheral members of the group they represent (Van Kleef et al. 2007).

Previous research has shown that positive contact with an individual outgroup member can generalize to the entire outgroup (e.g., Johnston and Hewstone 1992; Saygi et al. 2014; Weber and Crocker 1983), and this effect has been found even in the absence of direct contact (e.g., Turner et al. 2007). However, this generalization process is most likely when group membership is made salient (Vivian et al. 1997), such as by the prototypicality of the outgroup members (Oakes and Turner 1986; Oakes et al. 1991). For example, Brown et al. (1999) showed that positive attitudinal generalization was facilitated by encounters with prototypical outgroup members. Similarly, in stereotype change studies, stereotype-disconfirming information had a stronger effect on generalization when the information was associated with a prototypical outgroup member (Johnston and Hewstone 1992) or with an outgroup member who was a good fit to the category prototype (Rothbart and Lewis 1988).

The above suggests that, in representative negotiations, the cooperative behavior of the outgroup representative may spill over to influence the perceived cooperativeness of the entire outgroup, but only when the outgroup representative is a prototypical member. If the outgroup representative is a peripheral member, such generalization may be mitigated, and hence outgroup cooperativeness may not lead to increased outgroup cooperativeness (Hypothesis 1). Furthermore, we propose a moderated mediation model wherein cooperativeness of the outgroup representative indirectly (through perceived outgroup cooperativeness) influences constructive behavioral tendencies towards the outgroup. These mediated effects are expected to be stronger when the outgroup representative is a prototypical member rather than a peripheral member (Hypothesis 2). However, prototypicality of the outgroup representative can potentially influence the spread of any kind of representative characteristic to the entire outgroup. The effect is not specific to the cooperativeness of the outgroup representative. For example, if a prototypical outgroup representative is acting unethically in the negotiations, it would influence the perceived unethicality of the entire outgroup. Hence, we aimed to identify a moderator that specifically influences the effect of cooperativeness on intergroup relations.

1.3.2 Representative Competence

In Study 2, we examined the second boundary condition, which concerns the fact that intergroup negotiations are mostly marked by ample opportunities for distrust and hostile attributions given the mixed motive feature of negotiation (e.g., Fein and Hilton 1994; Kramer 1994, 2004a, b; Messick and Mackie 1989). Hence, even if a representative is a cooperative prototypical member of his group, the opposing party first needs to trust the cooperative intentions of the representative as well as the outgroup in order to improve outgroup perceptions. For example, in a recent report prepared for policy makers in Cyprus it was suggested that one of the main constraining factors for conflict resolution in Cyprus is that both Turkish and Greek Cypriots do not trust in the cooperative intentions of each other’s groups. They believe that there is an ulterior motive behind the outgroup representative’s cooperative behaviors. Hence, even if representatives behave in a cooperative way, due to lack of trust in the outgroup, cooperation during the negotiations fails to facilitate conflict resolution (Seeds of Peace 2013).

Along similar lines, Kramer developed a construct called intergroup paranoia which is defined as “beliefs-either false or exaggerated-that are held by members of one group that cluster around ideas of being harassed, threatened, harmed, subjugated, persecuted, accused, mistreated, wronged, tormented, disparaged, or vilified by a malevolent outgroup” (Kramer 2004a, b). He suggests that intergroup paranoia in the context of negotiations leads negotiators and their constituents to overly examine the meaning of each move by the other party and also to obsessively reanalyze both past and anticipated interactions with the other party. In fact, previous research shows that a history of hostility is not even essential to have a more biased and negative view of the outgroup. Brewer and her colleague showed that even a mere categorization of individuals into distinct groups can create an ingroup bias which translates into negative perceptions of the outgroup members such as untrustworthy, dishonest, and competitive (Brewer 1979; Brewer and Silver 1978). In a negotiation context such negative outgroup perception leads representatives and their constituents to be distrustful and suspicious of outgroup members and also to anticipate competitive negotiation behaviors from them (Insko and Schopler 1997). Ironically, this means that cooperative negotiation behaviors by an outgroup representative are counter-intuitive and rather unexpected (Benton and Druckman 1974; Steinel et al. 2010), and such counter-intuitive and unexpected behavior by the negotiators may lead outgroup members to overly examine the meaning of each move. Because cooperation compared to competition by the out-group representative is more unexpected and counter-intuitive, it should trigger additional processing of information and a deeper search for possible explanations. If true, additional cues that may explain the (unexpected) event should be picked up more and integrated better in judgments and decisions.

One such cue is whether the out-group representative appears competent. Highly competent people are in control of their behavior, they know what they are doing, and they pursue their goals effectively (Fiske et al. 2002). When unexpected and counter-intuitive behavior is performed by highly competent individuals, internal attributions to deliberate strategizing become more likely, and a highly competent outgroup representative may mitigate the positive effect of cooperative representative behavior on outgroup trust by creating a sneaky and untrustworthy image for his group rather than favorable perceptions (Boon and Holmes 1991). Thus, we hypothesized that the outgroup representative’s cooperation increases outgroup trust only when the outgroup representative has low competence, and the positive effect of representative cooperativeness on outgroup trust may be mitigated when the outgroup representative has high competence (Hypothesis 3). Furthermore, we predicted that cooperativeness of the outgroup representative leads to higher constructive behavioral tendencies through trust when the outgroup representative has low competence, but not when the outgroup representative has high competence (Hypothesis 4).

2 Study 1: Method

2.1 Sample and Design

Sixty-nine undergraduate students (39 female; \(M=22.42\,years, { SD}=5.8\)) at the University of Amsterdam participated for monetary compensation. Using a double blind procedure, participants were randomly assigned to a 2 (cooperative vs. competitive outgroup representative) \(\times \) 2 (peripheral vs. prototypical outgroup representative) between-subjects factorial design. Main dependent variables were perceived outgroup cooperativeness and constructive behavioral tendencies towards the outgroup.

2.2 Procedure and Materials

Upon arrival at the laboratory, participants were seated in separate cubicles in front of a computer. They read that they would be divided into two three-person groups based on their responses to a personality survey. After they completed the survey, all participants were told that they belonged to Group O, together with two other participants who responded similarly to the survey, and that they would engage in an intergroup negotiation with three other students who formed Group P. Participants were further told that one of the members in each group would be randomly chosen to represent their group in the intergroup negotiation, and that the other two members in each group would be able to observe the negotiation (see Van Kleef et al. 2007).

All participants were then shown the instructions for a multi-issue negotiation between Group O and Group P (adopted from De Dreu et al. 2006, see Appendix 1 for the pay-off schedule), in which they had to negotiate about the salary, salary increase, and duration of the contract. After they read the instructions, all participants were in fact placed in the observer role, and informed that they would receive feedback from their representative about the negotiation with the outgroup representative.

To manipulate the prototypicality of the outgroup representative, we gave participants additional information after the representative selection procedure (Van Kleef et al. 2007; Van Knippenberg and Van Knippenberg 2005). Participants in the prototypical [peripheral] condition received the following message: “A member of Group P is randomly selected by the computer to be the representative of Group P. There is one more thing about this group member that we want to share with you. According to his score on the personality test, he is [not] a typical member of Group P, which means that he thinks and acts [does not think and act] like other members of Group P. He is a very good example [not a good example] of a person with a P-type personality”. Finally we added an additional sentence to the peripheral condition to prevent participants from perceiving the peripheral outgroup member as pro-outgroup deviant. The sentence was “According to his score, he also does not have an O-type personality”. Next, participants learned that they could earn up to 30 euro as a group (10 euro maximum per person depending on the total number of points they earned on each issue) if the representatives managed to reach an agreement on all three issues within 10 min. If not, neither of the groups would receive any extra money. Finally, participants read that they would see the offers made by their own representative but not the offers made by the outgroup representative. However, they would be shown any messages from the outgroup representative to their representative. Participants did not actually interact with another person; rather, the outgroup representative’s behavior was simulated via pre-programmed messages sent over the computer (see Van Kleef et al. 2007).

The negotiation started and messages sent by the outgroup representative were shown to the participants. We manipulated the cooperativeness of the outgroup representative through these messages. Participants observed three cooperative/competitive messages by the outgroup representative in each round (e.g., “I not only want my group to be satisfied, I also want your group to be satisfied”, “I hear often that I am very nice”, and “Not everything is about money, I do not want anyone to be hurt in this negotiation” for cooperation; “I only want my group to be satisfied, it does not interest me how your group likes it”, “I hear often that I can be tough”, “Everything is about money here, not about whether we have a good time or not” for competition). Following the messages, they were told that their ingroup representative made an offer to the other group. The ingroup representative’s offers were linear concessions and were kept constant across the conditions. (Round 1: 20,000 salary/2.5 years contract/ 7 % salary increase; Round 2: 20,000 salary/2.5 years contract/6 % salary increase; Round 3:19,000 salary/ 2.5 years contract/ 6 % salary increase). In round 3, participants were told that while the representatives would continue to negotiate, they would answer a series of questions about their impressions thus far (see below). Because we were not interested in their reactions to the negotiated outcome but rather the effect of communication during the negotiation, we measured their behavioral tendencies towards the outgroup as well as perceived outgroup cooperativeness before the negotiations ended.

Outgroup cooperativeness was measured by five items (“cooperative”, “friendly”, “nice”, “trustworthy”, “honest”) (1 \(=\) Not at all, 7 \(=\) Very much) (Cuddy et al. 2008; \(\alpha =0.90\)). Constructive behavioral tendencies were measured with three items from the facilitation tendencies scale of Cuddy et al. (2007). Participants indicated to what extent they thought their group members would have (1) a tendency to help the outgroup, (2) to protect the outgroup, and finally (3) to associate with the outgroup (1 \(=\) Not at all, 7 \(=\) Very much). We asked participants about their group members’ tendencies, rather than their own tendencies, to reduce social desirability concerns (Cuddy et al. 2008). Please note that outgroup was defined as Group P members excluding the Group P representative.

Finally, participants completed a manipulation check for the cooperativeness and prototypicality of the outgroup representative. First, participants indicated to what extent they perceived the outgroup representative as “cooperative”, “friendly”, “nice”, “trustworthy”, and “honest”(1 \(=\) Not at all, 7 \(=\) Very much), \((\alpha =0.87)\). Secondly, we used nine items from Van Kleef et al. (2007) to check for the prototypicality manipulation (1 \(=\) Not at all, 7 \(=\) Very much), \((\alpha =\,0.92)\). All items are listed in Appendix 2.

3 Study 1: Results and Discussion

3.1 Manipulation Checks

Analysis of variance (ANOVA) showed that participants in the cooperative conditions perceived the outgroup representative as more cooperative \((M=4.95,\,SD=1.03)\) compared to participants in the competitive condition \((M=3.77,\,SD=1.20), F(1, 65)=19.02,\,p\,<\,.001, \upeta _{\mathrm{p}}^{2}=0.23\). ANOVA on perceived typicality of the outgroup representative showed that participants in the prototypical condition perceived the outgroup representative as more typical \((M=4.20,\,SD=0.89)\) compared to participants in the peripheral representative condition \((M=3.14,\,SD=1.28), F(1, 65)=16.05,\,p\,<\,.001, \upeta _{\mathrm{p}}^{2}=0.20\). No other effects reached significance, all \(F\hbox {s} < 2\), all \(p\hbox {s} > .05\). Thus, our manipulations were successful.

3.2 Hypothesis Testing

Table 1 shows means, standard deviations and correlations between the central variables in the current study. To test our proposed model (see Fig. 1), we used a regression based path analysis with the aid of the PROCESS macro developed by Hayes (2013) in order to estimate interactions and conditional indirect effects. Figure 2 includes three sub-models that we tested separately. Although our primary interest is the moderating role of prototypicality on the relationship between representative cooperativeness and outgroup cooperativeness, for the preciseness of our predictions (see Edwards and Lambert 2007) we also tested the moderating role of prototypicality on the relationship between representative cooperativeness and constructive behavioral tendencies (see Fig. 2 panel A). In Fig. 2 panel B, we tested Model 2 and 3 wherein we examined whether or not the prototypicality of the outgroup representative moderates the relationship between representative cooperativeness and outgroup cooperativeness (Hypothesis 1). Furthermore we estimated the conditional indirect effect of representative cooperativeness on constructive behavioral tendencies (via outgroup cooperativeness), contingent on representative prototypicality, which is calculated by multiplying a paths by path b. (Hypothesis 2; Hayes 2013).

Table 1 Means, standard deviations, and correlations among dependent variables (Study 1)
Full size table
Fig. 2
figure 2

The three ordinary least squares regressions estimated and reported in Table 2

Full size image

Consistent with Hypothesis 1, we found a significant interaction effect of representative cooperativeness and prototypicality on perceived outgroup cooperativeness (see Table 2). In line with our predictions, the conditional effect of representative cooperativeness on perceived outgroup cooperativeness in the prototypical representative condition was significant and estimated to be 1.867 with a bias adjusted and accelerated 95 % CI ranging from 1.066 to 2.666. The conditional effect of representative cooperativeness on perceived outgroup cooperativeness in the peripheral representative condition was significant and estimated a negative effect of \(-\)1.041 with a bias adjusted and accelerated 95 % CI ranging from \(-\)1.876 to \(-\)0.207. This confirms that a prototypical cooperative representative can improve perceived outgroup cooperativeness compared to a competitive prototypical representative. However, a peripheral cooperative outgroup representative actually produced lower perceived outgroup cooperativeness than a peripheral competitive outgroup representative. Furthermore, testing model 3 in Fig. 2 showed that in fact the conditional indirect effect of representative cooperativeness on constructive behavioral tendencies via outgroup cooperativeness was positive and significant in the prototypical representative condition, and negative and significant in the peripheral representative condition. These results support Hypotheses 1 and 2 (Table 3).

Table 2 Ordinary least squares regression model coefficients (standard errors in parentheses) (Study1)
Full size table
Table 3 Model coefficients for the conditional process model in Fig. 1
Full size table

Taken together, cooperation by the outgroup representative evoked higher constructive behavioral tendencies towards the entire outgroup because of improved cooperative perceptions of the entire outgroup, but only when the outgroup representative was a prototypical outgroup member. When the outgroup representative was a peripheral member, cooperativeness of the outgroup representative actually decreased perceived outgroup cooperativeness and hence lowered constructive behavioral tendencies. Clearly, cooperative negotiation strategies by out-group representatives do not always boost intergroup relations, but can also undermine favorable intergroup perceptions and constructive behavioral tendencies. It depends, first of all, on the representative’s prototypicality. In Study 2, we tested representative competence as the second boundary condition for the positive effects of cooperative behavior by the outgroup representative on constructive behavioral tendencies. In the study 2 we tested Hypothesis 3 and 4.

4 Study 2: Method

4.1 Sample and Design

Eighty-six undergraduate students (60 females; M \(=\) 21.12 years, SD \(=\) 4.5) at the University of Amsterdam participated in the study for monetary compensation. Using a double blind procedure, participants were randomly assigned to a 2 (cooperative/competitive outgroup representative) by 2 (low/high competence outgroup representative) between-subjects factorial design. Main dependent variables were outgroup trust and constructive behavioral tendencies towards the outgroup.

4.2 Procedure and Materials

The procedure was the same as Study 1, with one exception: In addition to cooperative/competitive messages, participants also saw 3 high/low competence related messages. These messages were: “my friends always tell me that I should be in politics, because I am really good at negotiations”, “people always tell me that I am really efficient because I can do multiple things at the same time”, “I do not easily lose control in this kind of situations” for high competence, and “people always tell me that I should not be in politics because I am not good at negotiations”, “people always tell me that I am not efficient because I cannot do multiple things at the same time”, “I easily lose control in this kind of situations” for low competence. We assessed constructive behavioral tendencies \((\alpha =0.73)\) as before and also included a 3-item outgroup trust scale (“To what extent do you believe the outgroup has a strong sense of justice?”, “To what extent do you believe the outgroup is trustworthy?”, “To what extent do you believe the outgroup is sneaky?” (reversed item) \((\alpha =0.77)\). We checked for the manipulation of cooperativeness of the outgroup representative as before \((\alpha =0.91)\), and for competence, by asking to what extent participants perceived the outgroup representative as experienced, skillful, competent, intelligent, efficient, and confident \((1=Not\, at\, all,\, 7=Very\, much; \alpha =0.85)\).

5 Study 2: Results

5.1 Manipulation Checks

Participants perceived the outgroup representative as more cooperative in the cooperative \((M=4.78,\,SD=0.96)\) compared to the competitive condition \((M=3.97,\,SD=1.29), F(1, 82)=11.17, p=.001, \upeta _{\mathrm{p}}^{2 }=.12\). Furthermore, participants in the high competence condition perceived the outgroup representative as more competent \((M=4.89,\,SD=0.76)\) than those in the low competence condition \((M=4.07,\,SD=1.07), F(1, 82)=16.63, p\,<\,.001, \upeta _{\mathrm{p}}^{2}=0.17\). No other effects reached significance, all \(F\hbox {s} < 2\), all \(p\hbox {s} > .05\). Thus our manipulation of cooperation and competence was successful.

5.2 Hypothesis Testing

Table 4 presents means, standard deviations and correlations between the central variables in Study 2. To test our proposed model in Fig. 1 we used the same analysis as we did in Study 1. Figure 3 shows the three sub-models that we tested separately. Similarly to Study 1, as a stringent test of our predictions (see Edwards and Lambert 2007), we tested the moderating role of competence both on the relationship between representative cooperativeness and outgroup trust and the relationship between representative cooperativeness and constructive behavioral tendencies despite the fact that our primary interest is the former. In Fig. 3 panel B, we tested Model 2 and 3 wherein we examined whether or not the competence of the outgroup representative moderates the relationship between representative cooperativeness and outgroup trust (Hypothesis 3). Furthermore we estimated the conditional indirect effect of representative cooperativeness on constructive behavioral tendencies (via outgroup cooperativeness), contingent on representative competence (Hypothesis 4).

Table 4 Means, standard deviations, and correlations among dependent variables (Study 2)
Full size table
Fig. 3
figure 3

The three ordinary least squares regressions estimated and reported in Table 5

Full size image

Consistent with Hypothesis 3, we found a significant interaction effect of representative cooperativeness and competence on outgroup trust (see Table 5). As predicted, the conditional effect of representative cooperativeness on outgroup trust in the low competence condition was significant and estimated to be 1.911 with a bias adjusted and accelerated 95 % CI ranging from 1.228 to 2.594. The conditional effect of representative cooperativeness on outgroup trust in the high competence condition was non-significant. This confirms that a highly competent cooperative outgroup representative did not improve trust compared to a competitive outgroup representative. Furthermore, testing model 3 in Fig. 3 showed that in fact the conditional indirect effect of representative cooperativeness on constructive behavioral tendencies via outgroup trust was positive and significant under low competent representative conditions, and non-significant under high competent representative conditions (see Table 6). These results support Hypotheses 3 and 4.

Table 5 Ordinary least squares regression model coefficients (standard errors in parentheses) (Study 2)
Full size table
Table 6 Model coefficients for the conditional process model in Fig. 1
Full size table

6 General Discussion

The idea that cooperation improves intergroup relations is widespread (e.g., Brown et al. 1999). Qualifying this, however, the current paper shows that cooperation by a peripheral outgroup representative decreased perceived outgroup cooperativeness and lowered constructive behavioral tendencies (as opposed to a competitive peripheral representative), and cooperation by a highly competent representative (as opposed to a low-competent representative) mitigated the positive effects of representative cooperativeness on outgoup trust and constructive tendencies. Furthermore, we observed that outgroup representative cooperativeness increased constructive behavioral tendencies via perceived outgroup cooperativeness (Study 1) and trust (Study 2) only when the outgroup representative was a prototypical member of the outgroup (Study 1) or had low competence (Study 2). Finally, outgroup representative cooperativeness lowered constructive behavioral tendencies via perceived outgroup cooperativeness when the outgroup representative was a peripheral member of the outgroup (Study 1).

In Study 1, we predicted that the effect of cooperative representative behavior on constructive behavioral tendencies would be weaker when the outgroup representative is a peripheral member of the group. But in fact we found that the effect of cooperative representative behavior on constructive behavioral tendencies even reversed and led to decreased constructive tendencies when the outgroup representative was a peripheral member of the outgroup. This suggests that perhaps there is a contrast effect between the perception of a peripheral representative and the perception of the outgroup (e.g., Simpson and Ostrom 1976). This indicates that appointing an external representative to a group to which he does not necessarily belong to may create a contrast effect on the perceptions of the outgroup and hence cooperative behavior of the representative may backfire.

In sum, our results show that cooperative behavior in intergroup negotiations can both improve and undermine intergroup relations. Intriguingly, it is especially when cooperative outgroup representatives are seen as peripheral and/or highly competent that intergroup relations tend to sour rather than flourish. Implications for negotiation and intergroup relations as well as some limitations of our studies are discussed below.

First, the present study provides first evidence that cooperative behavior by the out-group representative can have negative effects on in-group members’ perceptions of that outgroup, the extent to which they trust out-group members, and their inclination to approach out-group members in a constructive way. These findings are at odds with the common assumption that cooperation improves intergroup relations (e.g., Pettigrew and Tropp 2006), and indicate that cooperation by out-group representatives may backfire when the representative is perceived as peripheral and a-typical, and/or highly competent. Whereas cooperative negotiations may promote intergroup relations by creating more benign perceptions of, and behavioral tendencies towards, uninvolved out-group members, cooperative strategies may backfire and deteriorate intergroup relations when representatives are not seen as typical for their group, or when their high competence allows for alternative attributions. Future work in this area could seek to further illuminate the underlying mechanisms that explain when and why cooperative behavior by out-group representatives promotes, or blocks, the development of constructive intergroup relations.

Second, our findings also contribute to extent research that investigated the relationship between warmth (cooperative) and competence in person or group perceptions (for a review please see Kervyn et al. 2010). In this line of research, some researchers showed that warmth and competence are positively related (eg., Kelley 1950; Asch 1946; Anderson 1974), whereas others demonstrated that there is an orthogonal relationship between them (for a review, see Fiske et al. 2007) or that they are negatively related to each other (e.g.,Yzerbyt et al. 2005; Cuddy et al. 2004). Differently from previous research we examined the interactive relationship between cooperativeness and competence and found that higher competence mitigates the positive effects of representative cooperativeness on behavioral tendencies of the ingroup towards the outgroup. Our findings call for more research to explore the interactive relationship between cooperativeness and competence on person or group perception.

As such, our findings may also have relevance for representative selection. Whereas in-group members want competent leaders and representatives (Sears and Kinder 1985), our results suggest that communicating such competence while also seeking to cooperate with the out-group may actually undermine the positive effects of cooperation on intergroup relations. Likewise, Teixeira et al. (2011) uncovered that ingroup members sometimes prefer a representative who is pro-outgroup deviant member and, accordingly, an a-typical in-group member. Whereas such a pro-outgroup deviant may indeed help negotiations with the out-group, our results suggest ironic adverse effects on intergroup relations. Relatedly, although competence, power and status are sometimes used interchangeably in the literature, we believe that a clear distinction must be made when interpreting our results. As status may derive from respect to the person (e.g., Bendersky and Hays 2012), higher status of the outgroup representative may even facilitate trust and hence improve intergroup relations. Future research should investigate whether or not the cooperative behavior of high status outgroup representative would facilitate trust and hence improve intergroup relations.

Given the contrived laboratory settings, studies reported in this article may have limited ecological validity due to the lack of real groups or decisions with high stakes in the laboratory. However, both these challenges can be addressed: actual groups or their representatives can be brought into the laboratories; negotiating representatives can be confronted with high-stakes scenarios as developed by Cuhadar and Druckman (2014). In some real life contexts, ingroup members would be able to see their representative and be in contact with their fellow group members, which may potentially influence their perception of the outgroup (e.g., Castelli et al. 2003; Cialdini and Goldstein 2004). The ecological validity of research on representative negotiations would be improved by taking these factors into account. The trade off is between the internal validity obtained through control over variables and the external or ecological validity obtained through field studies. We chose to emphasize the former in our experiments but encourage further studies that achieve a balance between these validities. (See Druckman 2005, for a discussion of these issues in conflict research.)

For if the current findings hold, cooperation by a representative can indeed have counterproductive effects on intergroup relations. Having a-typical, or highly competent representatives may not be the best strategy to create constructive exchange among competing groups.