Self-Interest Masquerading as Ingroup Beneficence

Abstract

An altruistic rationalization explanation of the interindividual–intergroup discontinuity effect proposes that intergroup interactions are more competitive than interactions between individuals because group membership creates an opportunity to rationalize selfishly motivated competitiveness as being enacted for the ingroup’s sake. To test this explanation, we compared participants whose decisions did not influence the earnings of other participants, and who therefore had no opportunity for altruistic rationalization of competitiveness, with participants whose decisions did influence the earnings of other participants, and who therefore did have an opportunity for altruistic rationalization. Competition in a prisoner’s dilemma game was greater when opportunities for altruistic rationalization were present, but this effect was significant only for participants who were low in dispositional proneness to guilt, a trait that motivates prosocial behavior. In other words, we found that dispositionally selfish individuals were most likely to become more competitive when opportunities for altruistic rationalization were present. These findings provide evidence for an additional mechanism driving the discontinuity effect.

Keywords

interindividual–intergroup discontinuity guilt proneness intergroup conflict

Social scientists have long been concerned with understanding differences in the behavior of interacting individuals and groups. Research on the interindividual–intergroup discontinuity effect (see Wildschut & Insko, 2007; Wildschut, Pinter, Insko, Vevea, & Schopler, 2003) has approached the problem by examining the differential competitiveness of individuals and groups in the context of social dilemmas, such as the prisoner’s dilemma game (PDG; Figure 1). This work has tested and supported four explanations for the greater competitiveness of intergroup interactions compared with interindividual interactions. The schema-based distrust, or fear, explanation suggests that distrust is elevated in intergroup interactions because actual or anticipated interaction with a group activates generalized beliefs that groups are competitive, deceitful, and aggressive (Insko & Schopler, 1998; Pemberton, Insko, & Schopler, 1996; Wildschut, Insko, & Pinter, 2004). The social support explanation proposes that group members can provide mutual social support for a competitive choice whereas single individuals cannot (Insko, Schopler, Hoyle, Dardis, & Graetz, 1990; Schopler et al., 1993; Wildschut, Insko, & Gaertner, 2002). The identifiability explanation suggests that a group context shields members from personal responsibility for a competitive choice (Schopler et al., 1995). Finally, the ingroup-favoring norm explanation suggests that group membership implies normative pressure to benefit the ingroup (Cohen, Montoya, & Insko, 2006; Pinter et al., 2007; Wildschut, Insko, & Gaertner, 2002). The key objective of the present research was to test a possible fifth explanation for the discontinuity effect: altruistic rationalization.

Figure 1.

A prisoner’s dilemma game matrix

Altruistic Rationalization

The altruistic rationalization explanation proposes that intergroup interactions are more competitive than interindividual interactions because only in intergroup interactions can group members rationalize selfishly motivated competitiveness as being enacted for the ingroup’s sake. Altruistic rationalization was the first explanation offered for the discontinuity effect but was abandoned after it failed to garner support in an initial experiment (Insko et al., 1987). In that experiment, 3 participants seated on one side of a central corridor interacted with 3 participants seated on the other side of the corridor. The altruistic rationalization explanation was tested by contrasting an individuals condition with an outcome interdependence condition. In both conditions, 2 participants, seated directly opposite each other across the central corridor, interacted in the context of a PDG. However, only participants in the outcome interdependence condition were told that they would share their outcomes with the 2 other participants seated on their side of the central corridor. As only participants in the outcome interdependence condition could rationalize their competitiveness as being enacted for the sake of the persons with whom they would share outcomes, Insko et al. predicted more competition in the outcome interdependence than in the individuals condition. The results, however, revealed that competition was uniformly low in both conditions. Insko et al. concluded that altruistic rationalization was not a viable explanation for the discontinuity effect, and subsequent attention shifted to other explanations.

However, recent findings by Pinter et al. (2007) suggested that the altruistic rationalization explanation was discarded prematurely. In the Pinter et al. experiment, 3 participants seated on one side of a central corridor completed PDG interactions with 3 participants seated on the other side of the corridor. In an unaccountable leaders condition, participants were led to believe they had been randomly (and secretly) appointed as the leader of their three-person group. Furthermore, they were instructed that their decisions alone would determine their group’s outcomes and that other members of that group would never know the true basis for the outcomes they would receive. In contrast, participants in an individuals condition were not assigned to groups or to leadership positions. Pinter et al. found that unaccountable leaders, who could rationalize competitiveness as being enacted for their group’s sake, were more competitive than individuals, who had no such opportunities for altruistic rationalization. However, this effect emerged only among participants who were low in dispositional proneness to guilt, a trait that motivates prosocial behavior (see Tangney, 2003). Unaccountable leaders who were high in dispositional guilt proneness were highly cooperative and did not differ significantly from individuals. On the basis of their findings, Pinter and his colleagues proposed that low-guilt persons are more likely than high-guilt persons to capitalize on opportunities for altruistic rationalization because low-guilt persons are more self-interested and less concerned with the welfare of others (Baumeister, Stillwell, & Heatherton, 1994; Tangney, 2003; Tangney & Dearing, 2002).

Reconciling Inconsistent Findings

Although the altruistic rationalization explanation failed to receive support in an early study by Insko et al. (1987), a more recent study by Pinter et al. (2007) found evidence for altruistic rationalization among unaccountable leaders who were low in guilt proneness. How can these apparently inconsistent findings be reconciled? We focused on two possible explanations. The first explanation assumes that low-guilt (compared with high-guilt) persons are more likely to capitalize on opportunities for altruistic rationalization. This explanation implies that evidence for altruistic rationalization should be observed both for unaccountable leaders (as studied by Pinter et al.) and for participants sharing outcomes in a set (as studied by Insko et al.) but only (or primarily) when guilt proneness is low. This pattern would be consistent with the idea that Insko and his colleagues failed to find support for altruistic rationalization because they did not measure guilt proneness.

The second explanation also assumes that low-guilt (compared with high-guilt) persons are more likely to capitalize on opportunities for altruistic rationalization. However, it further assumes that such opportunities are more salient to unaccountable leaders (as studied by Pinter et al., 2007) than they are to participants who share group outcomes (as studied by Insko et al., 1987). This additional assumption is based on the fact that leaders usually have more control over group outcomes than do other members, who merely share those outcomes. The second explanation thus implies that evidence for altruistic rationalization should be stronger among low-guilt unaccountable leaders than among low-guilt participants sharing group outcomes. Such a result would suggest that if Insko and his colleagues had studied leaders, and measured their guilt proneness, then they might have found more support for altruistic rationalization. The present research sought to identify which, if either, of the two explanations we have identified is most accurate.

We compared three conditions in the context of a PDG: an individuals condition (individuals not sharing outcomes within a 3-person group), an interdependence condition (individuals sharing outcomes within a 3-person group), and an unaccountable leaders condition (group leaders completely determining the outcomes of a 3-person group). The first two conditions were modeled after the individuals and interdependence conditions of Insko et al. (1987), and the latter condition was modeled after the unaccountable leaders condition of Pinter et al. (2007). Our analyses involved two orthogonal contrasts and the interactions of these contrasts with guilt proneness. The first contrast compared the individuals condition with the pooled interdependence and leaders conditions. Hence, it compared participants who had an opportunity for altruistic rationalization to those who did not have this opportunity. This contrast was relevant to the possibility that opportunities for altruistic rationalization are salient for both leaders and participants sharing outcomes in a group (compared with individuals), but only low-guilt persons capitalize on these opportunities. We refer to this as the altruistic-rationalization contrast. The second contrast compared the interdependence and leaders conditions. This contrast was relevant to the additional possibility that altruistic rationalization is a more salient option for leaders, who have complete control over ingroup outcomes, than for participants, who have only partial control over those outcomes. We refer to this as the leaders contrast.

Method

Participants

Participants were 114 undergraduate students (84 women) enrolled in an introductory psychology course at a college in the Eastern United States. Participants received extra credit for their participation. Each experimental session involved 6 same-sex, unacquainted participants. They were collectively assigned to one of three conditions: individuals, interdependence, or unaccountable leaders. When participants were recruited, they were told only that we were studying social interactions.

Independent Variables

The design included two independent variables: interaction type and guilt proneness. The first variable was manipulated and the second variable was measured. The interaction type variable had three conditions: individuals, interdependence, and unaccountable leaders. The experiment was conducted in a laboratory containing two sets of three adjoining rooms located on opposite sides of a central corridor. Each session involved 6 participants seated in separate rooms. In the individuals and interdependence conditions, participants interacted with the person seated opposite to them across the central corridor. However, only participants in the interdependence condition expected they would share their outcomes with the 2 other participants seated on their own side of the central corridor. Participants seated on the same side of the corridor were not informed of each other’s individual decisions or outcomes. Hence, participants were unaccountable to persons in their own group. In the leaders condition, participants also interacted with the person seated opposite to them across the central corridor. They learned privately that they had been randomly selected to be the leader of the group on their own side of the laboratory and that the person with whom they were interacting was the (also randomly determined) leader of the group seated on the opposite side of the central corridor. Participants expected that their decisions would completely determine the outcomes that their own group would receive (e.g., if they earned US$5, then each person in their group would also receive US$5). Furthermore, participants learned that the people in their own group would not be aware of the leader’s influence on their outcomes (or, in fact, that there would be no leaders at all) and that the money they received would be described as bonus money for participating. It was emphasized to these participants that ingroup members “will never know your decisions had anything to do with the money they receive.” Hence, leaders were unknown and unaccountable to persons in their own group.

Guilt proneness was assessed at the beginning of each session with the Test of Self-Conscious Affect (TOSCA; Tangney, Wagner, & Gramzow, 1989). The TOSCA is a reliable and extensively validated (Tangney, 2003) measure in which individuals are asked to imagine a situation (e.g., “You make a mistake at work and find out that a coworker is blamed for the error”) and then rate on 5-point (1 = not likely; 5 = very likely) scales how likely it is that they would react in certain ways. For each of the 16 scenarios, there is a response that indicates guilt proneness (e.g., “You would feel unhappy and eager to correct the situation” when a coworker was blamed unfairly). We calculated a guilt proneness score by averaging across these 16 guilt responses (α = .72).

Procedure

After signing consent forms, participants completed a short questionnaire, which included the TOSCA and some filler items. The experimenter then described the interactions that would take place and trained participants on a sample PDG matrix. Next, participants received additional written instructions regarding their roles during the experiment. Participants in the individuals condition were simply reminded of the basic features of the interaction sequence (e.g., that they would interact with the person seated opposite to them on the other side of the corridor). Participants in the interdependence condition received a similar reminder, along with an instruction concerning the distribution of their PDG earnings. Specifically, they read as follows: “The three persons seated on your side of the room, including yourself, will share their earnings equally. That is, each will receive 1/3 of the total amount of money earned by the three-person set.” Finally, in the leaders condition, participants were instructed that they would serve as the leader of the 3 participants on their side of the room and that the entire amount of money that they would earn during the interactions with the other group’s leader would be divided evenly among the 3 members of their own group at the end of the experiment. Participants in the leaders condition were further instructed that the other members of their group would not know their role as leader and that these persons would also complete PDG interactions, but those interactions would involve points rather than money. All participants in the leaders condition were under the impression that they served as leaders and earned money. Participants then completed a short exercise to check their understanding of the additional instructions. The experimenter reviewed the written answers on that exercise and corrected participants on the few occasions (3%) where this was necessary.

Next, the experimenter described the trial sequence to all participants. Each trial was divided into three 1-min phases. Participants first considered their decisions privately, then met with the person seated directly opposite to them, and finally returned to their rooms to make a decision. On each trial, the experimenter recorded the decisions and returned the decision records to the participants, thereby informing them of their opponent’s decision. Participants anticipated completing 10 to 12 trials, but after 5 trials, the experimenter stopped the interaction and administered a postexperimental questionnaire. On completing that questionnaire, all participants were awarded US$3 and debriefed.

Dependent Variables

The main dependent variable was the proportion of competitive (Y) choices across the five PDG trials. Although each trial involved a dichotomous choice, the proportion of competition across trials could assume one of six equidistant values (0, 0.2, 0.4, 0.6, 0.8, 1), so no transformation of the proportion scores was required. Multiple motives may underlie PDG choices (e.g., a Y choice could reflect a concern for maximizing own outcomes or a concern for maximizing relative outcomes). Following previous research, we therefore examined participants’ reasons for their PDG choices with a postexperimental questionnaire (see Cohen, Wildschut, & Insko, 2010; Insko, Kirchner, Pinter, Efaw, & Wildschut, 2005; Pinter et al., 2007). Reasons were assessed using 10 items that were each rated on a 7-point scale (1 = not at all, 7 = very much). Recall that, in the individuals condition, participants’ choices affected only their own outcomes and those of their interaction partners. Accordingly, for participants in this condition, the items were phrased with reference to self and interaction partner. In the interdependence and leaders conditions, participants’ choices affected the outcomes of other persons on both sides of the laboratory. Hence, in these conditions, the items were phrased with reference to own and other group. There were two items to assess each of five reasons: concern for maximizing own absolute outcomes, max own (e.g., “to maximize my [my group’s] earnings”); concern for maximizing own relative outcomes, max rel (e.g., “to earn more than the other person [group]”); distrust (e.g., “did not trust the other person [group]”); concern for maximizing the joint outcomes of both persons (groups), max joint (e.g., “to maximize the joint outcomes of both persons [groups]”); and concern for minimizing the difference between the outcomes of both persons (groups), min dif (e.g., “to minimize the difference in outcomes between both persons [groups]”). We estimated the reliability of these two-item assessments via the Spearman-Brown split-half coefficient. Spearman-Brown split-half reliabilities for the reasons assessments were acceptable: .78 for max own; .78 for max rel; .62 for distrust; .91 for max joint; and .66 for min dif.

What pattern of results might be expected on the reasons assessments? Given the scarcity of evidence regarding altruistic rationalization, this was an open question. The notion of altruistic rationalization implies that unpalatable selfish competition will be attributed to concern for the ingroup. Such concern can be expressed in a number of ways: It could be indicated by increased concern with maximizing absolute ingroup outcomes (i.e., increased max own) or by increased concern with maximizing ingroup outcomes relative to outgroup outcomes (i.e., increased max rel; H1). It could also be expressed as reduced concern for maximizing the joint outcomes of both groups (i.e., reduced max joint) or reduced concern for equality between the groups (i.e., reduced min dif; H2). Finally, concern for the ingroup might be expressed in terms of distrust of the other group (i.e., increased distrust; H3).

Unit of Analysis

Within each session, participants interacted with only one other person. We used hierarchical linear modeling (HLM; Raudenbush & Bryk, 2002) to account for the nonindependence of PDG choices and choice reasons of interacting participants. Specifically, the analysis treated individual participants (Level 1) as units of observation nested with pairs (Level 2). PDG choices and choice reasons were modeled as a function of one Level 1 variable (guilt proneness) and one Level 2 variable (interaction type). We included a random intercept representing differences in the mean level of the dependent variable between pairs. The HLM analyses were implemented with SAS PROC MIXED. There were 22 pairs of participants in the individuals condition, 15 pairs of participants in the interdependence condition, and 20 pairs of participants in the unaccountable leaders condition.

Results

Our analyses focused on two orthogonal contrasts and their interactions with guilt proneness. The altruistic-rationalization contrast compared the individuals condition (opportunity for altruistic rationalization absent) with the pooled interdependence and leaders conditions (opportunity for altruistic rationalization present). The leaders contrast compared the interdependence and leaders conditions. All analyses included the two contrasts, guilt proneness, and the interactions between the contrasts and guilt proneness. Guilt proneness was mean centered. As guilt proneness is not a discrete variable with predefined low and high levels, we calculated predicted means and associated standard errors for each of the dependent variables at conditional values of guilt proneness (Table 1). A conditional value of one standard deviation below the mean of guilt proneness was denoted as low-guilt proneness and a conditional value of one standard deviation above the mean was denoted as high-guilt proneness (Aiken & West, 1991). Preliminary analyses revealed that there were no significant gender differences and that gender did not qualify the statistically significant findings reported below. We therefore omitted gender from the final analyses.

Table 1.

Predicted Means (Standard Errors) for Proportion Competition and Choice Reasons as a Function of Interaction Type and Guilt Proneness

	Low-guilt proneness			High-guilt proneness
	Individuals	Interdependence	Leaders	Individuals	Interdependence	Leaders
PDG choices
Competition	0.11 (0.06)	0.20 (0.07)	0.33 (0.06)	0.19 (0.06)	0.14 (0.07)	0.31 (0.06)
Choice reasons
Max own	5.28 (0.37)	5.16 (0.45)	6.22 (0.37)	5.03 (0.38)	5.47 (0.44)	5.94 (0.36)
Max rel	2.19 (0.40)	3.04 (0.49)	4.19 (0.41)	2.13 (0.42)	2.18 (0.48)	3.43 (0.40)
Distrust	1.89 (0.33)	3.15 (0.41)	3.60 (0.34)	2.16 (0.35)	1.87 (0.40)	3.09 (0.33)
Max joint	6.65 (0.34)	5.78 (0.42)	4.39 (0.35)	5.91 (0.36)	6.60 (0.41)	5.33 (0.34)
Min dif	5.25 (0.37)	4.72 (0.45)	4.13 (0.37)	5.55 (0.39)	6.12 (0.44)	4.97 (0.36)

Note: PDG = prisoner’s dilemma game. Entries are predicted means (standard errors) conditioned at one standard deviation below (low-guilt proneness) and above (high-guilt proneness) the mean of guilt proneness (Aiken & West, 1991). Max own = maximizing own absolute outcomes; Max rel = maximizing own relative outcomes; Max joint = maximizing the joint outcomes of both sides; Min dif = minimizing the difference between the outcomes of both sides.

Altruistic-Rationalization Contrast

Competition

Analysis of competition across trials revealed a significant interaction between the altruistic-rationalization contrast and guilt proneness, F(1, 54) = 4.34, p = .041. This interaction is depicted in the left panel of Figure 2. Among low-guilt participants, the altruistic-rationalization contrast was significant: There was more competition when the opportunity for altruistic rationalization was present than when it was absent, F(1, 54) = 4.09, p = .048. Among high-guilt participants, this contrast was not significant, F(1, 54) = 0.22, p = .64. These results provide a conceptual replication of findings by Pinter et al. (2007) and indicate that relatively selfish individuals (i.e., those low in guilt proneness) are most likely to capitalize on opportunities for altruistic rationalization.

Figure 2.

Altruistic-rationalization contrast × Guilt proneness interaction effects for competition across trials (left panel), distrust (middle panel), and max joint (right panel)

Reasons

The altruistic-rationalization contrast was significant for three choice reasons: max rel, F(1, 54) = 6.87, p = .011; distrust, F(1, 54) = 8.07, p = .006; and max joint, F(1, 54) = 10.71, p = .002. Compared with participants in the individuals condition, those in the interdependence and leaders condition scored higher on max rel (H1) and distrust (H3), and lower on max joint (H2; see Table 1). However, the effects on distrust and max joint were qualified by guilt proneness. There was a significant Altruistic-rationalization contrast × Guilt proneness interaction on distrust, F(1, 54) = 4.08, p = .048 (Figure 2, middle panel). Among low-guilt participants, the altruistic-rationalization contrast was significant: Participants expressed more distrust when the opportunity for altruistic rationalization was present (vs. absent), F(1, 54) = 12.18, p = .001. Among high-guilt participants, this contrast was not significant, F(1, 54) = 0.54, p = .464.

There was also a significant Altruistic-rationalization contrast × Guilt proneness interaction on max joint, F(1, 54) = 7.24, p = .001 (Figure 2, right panel). Among low-guilt participants, the altruistic-rationalization contrast was significant: Participants expressed less concern for maximizing joint outcomes when the opportunity for altruistic rationalization was present (vs. absent), F(1, 54) = 12.66, p = .001. Among high-guilt participants, this contrast was not significant, F(1, 54) = 0.01, p = .907.

With one exception, analyses of max own and min dif did not reveal statistically significant effects involving the altruistic-rationalization contrast or guilt proneness. The exception was a significant positive association between guilt proneness and min dif, B = 0.42, SE = 0.16, F(1, 54) = 6.73, p = .012. High-guilt (relative to low-guilt) participants expressed stronger concern for equality, thus corroborating the positive link between guilt proneness and concern for the welfare of others.

Mediation analyses

As results for distrust and max joint paralleled those for competition (Figure 2), we tested whether these two choice reasons mediated the Altruistic-rationalization contrast × Guilt proneness interaction on competition (Baron & Kenny, 1986).

We entered distrust as a covariate in a model that also included the two contrasts, guilt proneness, and their interactions. The results are summarized in the top panel of Figure 3. With distrust in the model, there was a significant positive association between distrust and competition, B = 0.04, SE = 0.01, F(1, 54) = 17.96, p = .00001, and the previously significant Altruistic-rationalization contrast × Guilt proneness interaction was no longer significant, B = −0.02, SE = 0.02, F(1, 54) = 1.41, p = .240. We then used the PRODCLIN program (MacKinnon & Fairchild, 2009) to calculate asymmetric confidence intervals for the indirect effect on competition via distrust. The point estimate for that effect was −.016 and the 95% confidence interval did not include zero [−.034, −.001], indicating that the indirect effect was significant.

Figure 3.

Distrust (top panel) and max joint (bottom panel) mediation of the Altruistic-rationalization contrast × Guilt proneness interaction effect on competition across trials

Next, we repeated the analyses just described with max joint as the mediator (Figure 3, bottom panel). With max joint in the model, there was a significant negative association between max joint and competition, B = −0.03, SE = 0.01,F(1, 54) = 13.12, p = 10⁻⁴, and the previously significant Altruistic-rationalization contrast x Guilt proneness interaction was no longer significant, B = −0.01, SE = 0.02, F(1, 54) = 0.50, p = .482. The point estimate for the indirect effect was −.016 and the 95% confidence interval did not include zero (−.034, −.003), indicating that the indirect effect via max joint was significant.

Finally, we conducted an analysis in which distrust and max joint were entered simultaneously as covariates. In this analysis, both distrust, B = 0.03, SE = 0.01, F(1, 54) = 9.50, p = .003, and max joint, B = −0.02, SE = 0.01, F(1, 54) = 4.99, p = .030, shared a significant association with competition, independently of the other potential mediator.

Taken together, results of the mediation analyses are consistent with the possibility that low-guilt participants who had an opportunity for altruistic rationalization (compared with low-guilt participants who had no such opportunity) linked their increased competitiveness to a perceived threat to the ingroup (distrust) and reduced concern for maximizing the outcomes of both groups (max joint).

Leaders Contrast

Competition

Analysis of competition across trials revealed a marginal leaders contrast, F(1, 54) = 3.16, p = .081. There was a tendency for participants in the leaders condition (M = .32, SD = .29) to be more competitive than participants in the interdependence condition (M = .17, SD = .22). The interaction between the leaders contrast and guilt was not significant, F(1, 54) = 0.47, p = .497.

Reasons

The leaders contrast was marginal or significant for all five choice reasons: max own, F(1, 54) = 3.13, p = .083; max rel, F(1, 54) = 5.75, p = .020; distrust, F(1, 54) = 4.42, p = .040; max joint, F(1, 54) = 10.71, p = .002; and min dif, F(1, 54) = 4.39, p = .041. Compared with participants in the interdependence condition, participants in the leaders condition scored higher on max own, max rel, and distrust and lower on max joint and min dif (Table 1). The analyses revealed no significant interaction effects between the leaders contrast and guilt proneness, Fs(1, 54) < 1.31, ps > .257.

To summarize, findings for the leaders contrast are suggestive but inconclusive. For the focal dependent variable—competition—the leaders contrast was marginal. This ruled out mediation analyses because we did not meet the first condition for mediation specified by Baron and Kenny (1986), namely that there must be a statistically significant effect of the independent variable (the leaders contrast) on the dependent variable (competition). For choice reasons, however, there was a consistent pattern of results indicating a more competitive (and less cooperative) orientation in the leaders condition than in the interdependence condition. Yet we found no evidence that the leaders contrast was stronger when guilt proneness was low (compared with high).

Discussion

We evaluated the altruistic rationalization explanation of the discontinuity effect by comparing the PDG choices of participants whose decisions did not influence the earnings of ingroup members (and who therefore did not have an opportunity for altruistic rationalization) with participants whose decisions did influence the earnings of ingroup members (and who therefore did have an opportunity for altruistic rationalization). Within the latter category, we made a further distinction between an interdependence condition, in which participants’ decisions only partially determined the outcomes of ingroup members, and an unaccountable leaders condition, in which participants’ decisions completely determined the outcomes of ingroup members.

Contrasting these conditions allowed us to examine two potential explanations for the discrepant findings from earlier research by Insko et al. (1987), who did not find evidence for the altruistic rationalization explanation, and Pinter et al. (2007), who found evidence for that explanation among low-guilt persons. The first explanation assumed that low-guilt (compared with high-guilt) persons are more likely to capitalize on opportunities for altruistic rationalization. According to this explanation, there should be more competition when opportunities for altruistic rationalization are present (vs. absent) among low-guilt (but not high-guilt) persons. The second explanation also assumed that low-guilt (compared with high-guilt) persons are more likely to capitalize on opportunities for altruistic rationalization but further proposed that altruistic rationalization is a more salient option when participants completely determine ingroup outcomes (as in the unaccountable leaders condition) than when participants only partially determine ingroup outcomes (as in the interdependence condition). According to this explanation, low-guilt (but not high-guilt) persons should be more competitive in the unaccountable leaders condition than in the interdependence condition.

Our findings were consistent with the first alternative explanation. When guilt proneness was low, participants in the interdependence and leaders conditions (who had an opportunity for altruistic rationalization) were more competitive than participants in the individuals condition (who did not have an opportunity for altruistic rationalization). When guilt proneness was high, the presence versus absence of opportunities for altruistic rationalization did not affect competition. An examination of choice reasons further revealed that distrust was higher, and concern for maximizing joint outcomes lower, when opportunities for altruistic rationalization were present (vs. absent) but only when guilt proneness was low (Figure 2). Mediation tests were consistent with the possibility that low-guilt participants who had an opportunity for altruistic rationalization (compared with low-guilt participants who had no such opportunity) linked their increased competitiveness to a perceived threat to the ingroup (distrust) and reduced concern for maximizing the outcomes of both groups (max joint).

These results provide preliminary evidence that low-guilt persons who have opportunities for altruistic rationalization will seek to rationalize selfishly motivated competition in terms of an alleged threat to the ingroup and a reduced concern with maximizing the outcomes of both groups. We speculate that these “indirect” altruistic rationalizations are preferred because they more effectively disguise selfishness than do “direct” expressions of concern for the ingroup (max own or max rel).

There was no support for the prediction, based on the second alternative explanation, that low-guilt (but not high-guilt) persons would be more competitive in the unaccountable leaders condition than in the interdependence condition. Instead, we found suggestive evidence that, regardless of guilt proneness, unaccountable leaders were more competitive than participants in the interdependence condition. Whereas we anticipated that low-guilt unaccountable leaders might be particularly competitive, we did not expect that high-guilt unaccountable leaders would show comparable levels of competitiveness. Recall that in the Pinter et al. (2007) experiment, high-guilt unaccountable leaders were very cooperative and did not differ significantly from individuals. It is unclear why high-guilt unaccountable leaders in our experiment were more competitive than in this previous research. That is an issue that calls for more research.

Three additional concerns are worth noting. The first concern relates to the issue of identifying whether altruistic rationalization takes place before or after a competitive choice is made. Our assessment of choice reasons followed choice behavior, which might suggest that rationalization took place after the choice. However, the ordering of the assessments need not conform to the actual sequence of rationalization and behavior. We were unable, based on the available evidence, to determine if altruistic rationalization occurred before or after the choice. However, logic dictates that opportunities for altruistic rationalization can only exert an influence if they are detected before the choice. In all likelihood, then, the altruistic rationalization process is at least initiated (but perhaps not completed) before a competitive choice is made. This issue could be explored further by assessing reasons both before and after the choice.

The second concern relates to the issue of whether the purpose of altruistic rationalization is to reassure the competitive person that he or she is doing something for others (and not just for the self), or to convince others that the competitive person is doing something for them (and not just for the self), or whether it serves both purposes. Competition is incompatible with strong societal norms of fairness and trust (Cohen et al., 2010) and norm violation begets social disapproval. We propose that altruistic rationalization serves to shield the self from such social disapproval. Thus, one purpose of altruistic rationalization is almost certainly to convince others that one is doing something for them (and not just for oneself). However, norm violation also begets self-criticism, so altruistic rationalization may also serve to reassure oneself that one is doing something for others (and not just for oneself). In this way, group membership can provide an important basis for self-deception regarding unpalatable behavior.

The third concern is whether our results imply altruistic rationalization at all. Perhaps participants were more competitive in the outcome interdependence and unaccountable leaders conditions (compared with the individuals condition) because they felt genuine concern for their groups. Although we cannot rule out this explanation, it seems likely that if our findings represented real love for the ingroup, then they would have been stronger for high-guilt persons, because high-guilt (compared with low-guilt) persons are more likely to experience genuine concern for the ingroup (Cohen et al., 2006; Pinter et al., 2007; Wildschut & Insko, 2007). The fact that it was low-guilt participants with an opportunity for altruistic rationalization who were most competitive supports the idea of selfish, rationalized conduct.

Conclusion

The key message from research on the discontinuity effect is that intergroup conflict is multiply determined. Previous work has supported a number of mechanisms—social support, fear, reduced identifiability, and ingroup-favoring norms—that underlie the greater conflict and competition observed between groups rather than individuals. The main contribution of the present research is to clarify the conditions under which yet another mechanism—altruistic rationalization—undermines intergroup harmony.

Footnotes

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Aiken

L. S.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: SAGE.

Baron

R. M.

Kenny

D. A.

(1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182. doi:10.1037/0022-3514.51.6.117310.1037/0022-3514.51.6.1173

Baumeister

R. F.

Stillwell

A. M.

Heatherton

T. F.

(1994). Guilt: An interpersonal approach. Psychological Bulletin, 115, 243-267. doi:10.1037/0033-2909.115.2.24310.1037/0033-2909.115.2.243

Cohen

T. R.

Montoya

R. M.

Insko

C. A.

(2006). Group morality and intergroup relations: Cross-cultural and experimental evidence. Personality and Social Psychology Bulletin, 32, 1559-1572. doi:10.1177/014616720629167310.1177/0146167206291673

Cohen

T. R.

Wildschut

Insko

C. A.

(2010). How communication increases interpersonal cooperation in mixed-motive situations. Journal of Experimental Social Psychology, 46, 39-50. doi:10.1016/j.jesp.2009.09.00910.1016/j.jesp.2009.09.009

Insko

C. A.

Kirchner

J. L.

Pinter

Efaw

Wildschut

(2005). Interindividual–intergroup discontinuity as a function of trust and categorization: The paradox of expected cooperation. Journal of Personality and Social Psychology, 88, 365-385. doi:10.1037/0022-3514.88.2.36510.1037/0022-3514.88.2.365

Insko

C. A.

Pinkley

R. L.

Hoyle

R. H.

Dalton

Hong

Slim

R. M.

. . . Thibaut

(1987). Individual versus group discontinuity: The role of intergroup contact. Journal of Experimental Social Psychology, 23, 250-267. doi:10.1016/0022-1031(87)90035-710.1016/0022-1031(87)90035-7

Insko

C. A.

Schopler

(1998). Differential distrust of groups and of individuals. In Sedikides

Schopler

Insko

C. A.

(Eds.), Intergoup cognition and intergroup behavior (pp. 75-107). Hillsdale, NJ: Erlbaum.

Insko

C. A.

Schopler

Drigotas

S. M.

Graetz

Kennedy

Cox

Bornstein

(1993). The role of communication in interindividual–intergroup discontinuity. Journal of Conflict Resolution, 37, 108-138. doi:10.1177/002200279303700100510.1177/0022002793037001005

10.

Insko

C. A.

Schopler

Hoyle

R. H.

Dardis

G. J.

Graetz

K. A.

(1990). Individual–group discontinuity as a function of fear and greed. Journal of Personality and Social Psychology, 58, 68-79. doi:10.1037/0022-3514.58.1.6810.1037/0022-3514.58.1.68

11.

MacKinnon

D. P.

Fairchild

A. J.

(2009). Current directions in mediation analysis. Current Directions in Psychology, 18, 16-20. doi:10.1111/j.1467-8721.2009.01598.x10.1111/j.1467-8721.2009.01598.x

12.

Pemberton

M. B.

Insko

C. A.

Schopler

(1996). Memory for and experience of differential competitive behavior of individuals and groups. Journal of Personality and Social Psychology, 71, 953-966. doi:10.1037/0022-3514.71.5.95310.1037/0022-3514.71.5.953

13.

Pinter

Insko

C. A.

Wildschut

Kirchner

J. L.

Montoya

R. M.

Wolf

S. T.

(2007). Reduction of interindividual–intergroup discontinuity: The role of leader accountability and proneness to guilt. Journal of Personality and Social Psychology, 93, 250-265. doi:10.1037/0022-3514.93.2.25010.1037/0022-3514.93.2.250

14.

Raudenbush

Bryk

(2002). Hierarchical linear models. Thousand Oaks, CA: SAGE.

15.

Schopler

Insko

C. A.

Drigotas

S. M.

Wieselquist

Pemberton

Cox

(1995). The role of identifiability in the reduction of interindividual–intergroup discontinuity. Journal of Experimental Social Psychology, 31, 553-574. doi:10.1006/jesp.1995.102510.1006/jesp.1995.1025

16.

Schopler

Insko

C. A.

Graetz

K. A.

Drigotas

Smith

V. A.

Dahl

(1993). Individual–group discontinuity: Further evidence for mediation by fear and greed. Personality and Social Psychology Bulletin, 19, 419-431. doi:10.1177/014616729319400710.1177/0146167293194007

17.

Tangney

J. P.

(2003). Self-relevant emotions. In Leary

M. R.

Tangney

J. P.

(Eds.), Handbook of self and identity (pp. 384-400). New York, NY: Guilford.

18.

Tangney

J. P.

Dearing

R. L.

(2002). Shame and guilt. New York, NY: Guilford.

19.

Tangney

J. P.

Wagner

Gramzow

(1989). The Test of Self-Conscious Affect (TOSCA). Fairfax, VA: George Mason University.

20.

Wildschut

Insko

C. A.

(2007). Explanations of interindividual–intergroup discontinuity: A review of the evidence. European Review of Social Psychology, 18, 175-211. doi:10.1080/1046328070167654310.1080/10463280701676543

21.

Wildschut

Insko

C. A.

Gaertner

(2002). Intragroup social influence and intergroup competition. Journal of Personality and Social Psychology, 82, 975-992. doi:10.1037/0022-3514.82.6.97510.1037/0022-3514.82.6.975

22.

Wildschut

Insko

C. A.

Pinter

(2004). The perception of outgroup threat: Content and activation of the outgroup schema. In Yzerbyt

V. Y.

Judd

C. M.

Corneille

(Eds.), The psychology of group perception: Contributions to the study of homogeneity, entitivity, and essentialism (pp. 335-359). Philadelphia, PA: Psychology Press.

23.

Wildschut

Pinter

Vevea

Insko

C. A.

Schopler

(2003). Beyond the group mind: A quantitative review of interindividual–intergroup discontinuity effect. Psychological Bulletin, 129, 698-722. doi:10.1037/0033-2909.129.5.69810.1037/0033-2909.129.5.698