Abstract
Monin, Sawyer, and Marquez (2008) demonstrated that a rebel who refused to participate in a racially prejudiced decision task was derogated by actors who had themselves taken part in the task, but exalted by uninvolved observers. These authors argued that this rejection resulted from actors who expected and resented imagined moral reproach from the rebel. Two new studies address alternative explanations for such rejection of moral rebels. Study 1 rules out both (a) self-presentation and (b) a more cognitive alternative wherein anchoring on one’s first impression of a task leads to do-gooder derogation. Study 2 breaks down the principled rebellion into its two components, a principled stance and rebellion itself, to show that both seem to play a part in derogation. Individuals imagine that moral others would reject them, which in turn leads to rejecting those deviants.
Observing other people’s selfless behavior may sometimes elevate us to pursue similarly noble courses of action (Schnall & Roper, 2012; Schnall, Roper, & Fessler, 2010), but at other times it makes us feel morally inadequate (Monin, 2007). How do we contend with such threats to our moral identities? Imagine learning that a coworker took a stand against a sexist joke distributed on departmental email, or that a colleague refused industry gifts out of concern that they might compromise the integrity of her work—after you have failed not only to take a similar stand but even to recognize the opportunity to do so. In such instances, you might feel a pang of irritation at the implicit reproach levied by the rebel. Whether or not they ever learn of your own choice and whether or not you even agree with their choice, there is something especially antagonizing, we propose, about the fact that they could have judged your behavior and find it morally wanting. In the current paper we endeavor to better understand this rejection and demonstrate the role of this imagined reproach by ruling out potential alternatives.
The Resentment of Moral Rebels
Perhaps paradoxically, principled deviance—taking a stand against a morally problematic situation or decision—can be judged simultaneously as virtuous behavior by some and derided as self-righteousness by others (Minson & Monin, 2012; Monin, 2007; Monin & O’Connor, 2011; Monin et al., 2008). Consider individuals outside an organization who might praise an employee for refusing to forward emails of an offensive joke—calling that person virtuous, upstanding, and strong—while coworkers who received the joke and did not do anything about it might deride the same individual as self-righteous, overly sensitive, or politically correct. Monin and colleagues demonstrated such rejection of moral rebels in the domains of prejudice (Monin et al., 2008) and meat eating (Minson & Monin, 2012), in each case attributing this negative reaction to imagined moral reproach—individuals reject a principled deviant because they resent the moral reproach of their own behavior implied by the deviant’s stand.
Principled deviants who take the high road threaten others’ moral self-worth (Allison, Messick, & Goethals, 1989; Aquino & Reed, 2002; Cramwinckel, van Dijk, Scheepers, & van den Bos, 2013; Monin & Jordan, 2009). To protect their own sense of moral self-worth (e.g., Tesser, 1991), individuals sometimes reject members who attempt to change immoral ingroup practices (e.g., combatting out-group discrimination; Chan, Louis, & Jetten, 2010), exclude them for behaving more unselfishly than other group members (Parks & Stone, 2010), and even shun them for forgiving others because they appear to be acting self-righteously (Adams, Zou, Inesi, & Pillutla, 2015).
Monin et al. (2008), for example, showed that the same moral rebels who refused to participate in a racist task were embraced as principled heroes by participants assigned to be mere observers, but vilified by actors who themselves had already taken part in the same task: In three studies, participants read about a “whodunit” scenario where the evidence strongly pointed to a lone African American suspect as the culprit for a burglary. Participants also rated a target, ostensibly a previous participant, who either accused the African American suspect as the situation dictated (an obedient target), or refused to do so and commented that the task was “offensive” (a rebel target). Some participants (actors) had been asked to solve this same crime task first and, sure enough, they overwhelmingly accused the African American suspect, whereas others (observers) simply rated their fabricated peer’s answer. Relative to an obedient target, the moral rebel was liked and respected by observers who had not first completed the task, but disliked by actors who had already complied with it. Further, these initial studies showed that (a) this resentment was mediated by actors’ perceptions that the rebel would dislike them, and (b) prior self-affirmation reduced resentment: Self-affirmed participants still believed that the rebel would dislike them, but this belief no longer triggered rejection of the rebel. These results support the interpretation that participants’ resentment is a preemptive strike against the rebel who would presumably dislike them.
Upon closer reflection, these results are puzzling: Why would individuals lash out against moral rebels, mediated by their fear that the rebel would reject them, when they know that the rebel will never actually know anything about them? The rebellion witnessed by participants in Monin et al. (2008) was on a tape recording (Study 1) or a handwritten questionnaire (Studies 2–4) allegedly from a “previous participant” who had already left and unlikely to ever see the current participant’s responses. Thus, it is unlikely that participants reacted literally to the rebel’s expected judgment. Could merely imagining the rebel’s hypothetical judgment be enough to threaten individuals’ self-worth? This is the possibility we explore in the current paper.
The Phantom Rebel: Whose Reproach Motivates Resentment?
Whereas Monin et al. (2008) found that perceived rejection by the moral rebel statistically mediated the resentment towards them, suggesting a fairly interpersonal and maybe even self-presentational process, this resentment was also found to be alleviated by the more individual and self-focused process of self-affirmation (Sherman & Cohen, 2006; Steele, 1988), suggesting a potential paradox. If rejection expresses a threat to one’s sense of self-worth (as the self-affirmation data suggest), then why would the opinion of some unknown previous participant matter and even predict rejection (as the mediation data suggest)?
The goal of this paper is to tackle this potential paradox and to support the hypothesis that individuals reject the moral rebel because they resent the imagined rebel’s judgment of their behavior. Thus, we argue that it is specifically the imagined reproach of the long-gone, “phantom” rebel that is irksome enough to trigger rejection. To support this interpretation, in the present paper we attempt to rule out five possible alternative explanations for the rejection of moral rebels: (1) self-presentation to the experimenter, (2) anchoring on the first response to the task, (3a) judgmental language (without actual rebellion), (3b) rebellion (without judgment), and (4) projection of self-reproach on any imagined audience. We attempt to rule out these five alternative explanations in two studies to clarify the process of moral threat and the derogation of principled deviants.
Alternative 1: Self-presentation to the experimenter
Perhaps participants respond to the expected reproach not of the rebel (who is not present and whom they do not expect to encounter), but rather of the experimenter—the only person physically present to witness the participants’ compliance in the context of the rebel’s principled deviance (Leary, 1995). Despite the fact that the rebel’s stand challenges the experimenter and his study, participants may have suspected that the decision task was not about decision-making but instead a test of one’s moral and racial attitudes. According to this alternative, participants may have felt vulnerable to social judgment after seeing the previous participant’s response, assumed their own choice would be compared with the rebel’s and found wanting, and sought to restore their own esteem in the eyes of the experimenter by putting down the rebel and his response. In fact, the experimenter was the only person in participants’ immediate environment who had access to their responses and the rebel’s. Individuals may therefore have rejected the rebel primarily to regain status in the eyes of the experimenter. We tested this alternative in Study 1 by manipulating whether the experimenter was a possible source of threat. If individuals derogate moral rebels even when their own task compliance is unknown to the experimenter, these private reactions would suggest that the source of moral threat does not result from fear of the experimenter’s disapproval.
Alternative 2: The first choice is an anchor
Faced with an ambiguous task, such as the police lineup scenario, individuals may be highly sensitive to informational and social cues indicating the appropriate response (Deutsch & Gerard, 1955; Kahneman, 1992; Kahneman & Miller, 1986). Those participants initially asked to complete the task themselves may have used their own response as an anchor for acceptable behavior (Epley & Dunning, 2000; Epley, Keysar, van Boven, & Gilovich, 2004). For the vast majority who comply with the task, compliance appears to be the appropriate response—making the rebel’s response seem inappropriate. Conversely, observers, who witness the rebel’s response first without having given their own, may encode rebellion as the appropriate response through the same primacy process. According to this interpretation, if participants saw a peer comply with the task before seeing a rebel, they should express the same sort of rejection of the rebel as actors, whose first encounter with the police task was their own compliance with it. However, if self-threat is necessary to trigger rejection, then seeing another actor comply first should not lead to rejection of a later rebel. We also tested this alternative in Study 1.
Alternatives 3a and 3b: A vocal stance or the act of rebellion?
When a principled deviant is rejected, is it because of his refusal to go along or because of his invocation of moral principles and condemnation of the experiment (calling the task “offensive”), and by extension, of others who go along? By confounding rebellion with verbal condemnation, Monin et al. (2008) left ambiguous which component triggers the sting of moral threat and subsequent rejection of principled deviants. We test Alternatives 3a and 3b together in Study 2 by creating two new conditions to clarify whether judgmental language and nonjudgmental rebellion are, independently, sufficient triggers to reject the rebel.
Alternative 3a: Judgmental language alone produces derogation
According to Alternative 3a, participants reject moral rebels simply because they verbally condemned the task, not because they actually refused to complete it. This alternative predicts that participants would reject someone who calls the task offensive, even if they end up selecting the same African American suspect as other participants. Participants would reject such a judgmental, yet obedient, target because he expressed a verbal stance even without the accompanying behavior.
Alternative 3b: Rebellion alone produces derogation
Conversely, the mere act of refusing to participate in the task may be sufficient to elicit rejection without ever vocally condemning it (Alternative 3b). According to this alternative, rebellion itself is the trigger to produce moral threat and subsequent rejection simply because the rebel’s behavior implies judgment. Actors reject rebels because of their choice, and nothing more. Even if the rebel explicitly expresses understanding and sympathy for those who went along with the task (i.e., they are nonjudgmental rebels), participants might still reject them because of the judgment that their rebellion insinuates.
Alternative 4: Rebel reproach is entirely projective
If self-presentation to the experimenter (Alternative 1) cannot account for the rejection of moral rebels, another possibility is that participants feel bad about their own behavior upon witnessing the rebel’s principled deviance, and they expect to be rejected by any imagined audience who comes across both their behavior and the rebel’s. This model is consistent with Monin and colleagues’ (2008) interpretation of their mediation results. They found that participants rejected the moral rebel because they expected the rebel to reject them on account of their nonmoral behavior. The authors also proposed that “participants would anticipate a similar rejection from a random bystander who witnessed their obedience and the rebel’s stance and that this generic imagined rejection would similarly mediate the effect” (p. 90). This claim offered a resolution for the seeming discrepancy between the self-affirmation (suggesting a self-threat process) and mediation results (suggesting an interpersonal process involving the rebel), but Monin and colleagues did not actually test the proposition that imagined reproach from any audience would similarly elicit rebel derogation. We explored this possibility in Study 2 by testing whether rejection by any imagined audience of one’s peers would lead participants to reject the rebel.
Overview of the Current Studies
Two studies using variations of the Monin et al. (2008) paradigm tested these five alternatives. Study 1 tested the effects of self-presentation (Alternative 1) by manipulating the privacy of responses: if individuals are motivated to reject moral rebels because of impression management, responding privately should alleviate fear of rejection by the experimenter. Study 1 also tested the encoding/anchoring explanation (Alternative 2) by manipulating “observer” participants’ first encounter with the task: if individuals anchor on their first exposure to a context to determine the most appropriate response, they should reject moral rebels after observing another’s compliance with the task even if they were not asked to comply with it first. Study 2 explores two unique features of moral threat to clarify whether any moralized language (Kreps & Monin, 2011), even when it is accompanied by the same choice that participants made (Alternative 3a), and whether rebellion, even when it is explicitly nonjudgmental (Alternative 3b), can evoke threat and subsequent derogation. Finally, Study 2 also tests whether imagined rejection from any random third party, and not just from principled deviants, is sufficient to produce moral threat and subsequent rejection (Alternative 4).
Study 1
In Study 1, we used a similar design as the Monin et al. (2008) police lineup task to explore two central alternative explanations for the rejection of moral rebels. We first wanted to rule out that derogating the rebel was an attempt to save face in the experimenter’s eyes, fueled by self-presentational concerns (Alternative 1). To test this alternative, we created a condition wherein participants’ responses to the police decision task remained private. If these private actor participants still rejected the rebel, this effect could not be explained by self-presentation, since the experimenter never would see their decisions and could, therefore, not pass judgment about their own task behavior.
Study 1 also tested Alternative 2, that rejecting the rebel has little to do with being threatened by the rebel’s choice, but results instead from the only reference point that participants have. Specifically, actor’s own compliance with the task makes the rebel appear deviant, whereas observers have no prior reference point and consider the rebel’s response to be the normal one. We tested this by creating a new observer condition in which participants viewed another target comply with the task (providing them with a reference point of obedience) before judging the rebel.
To test both Alternatives 1 and 2, we created four experimental conditions: two actors (public, private) and two observers (rebel first, obedient first). If anchoring on first impressions can explain why individuals reject moral rebels, observers (like actors) should reject the rebel after first seeing an obedient participant. Conversely, if observers react positively to the rebel even after viewing another’s compliance, results would suggest that a different process produces rejection.
Method
Participants
Seventy-two undergraduates at a large U.S. private university (31 Whites, 25 Asians, 10 Latinos, 3 Native Americans, 1 Middle Eastern, 1 African American, and 1 non-Black multiracial; 48 women, 24 male; Mage = 19.5, SD = 1.74) came to the laboratory for individual research sessions to complete the research participation requirement of an introductory psychology class or in exchange for $8.00. The experimenter was a White male.
Procedure
Design
We randomly assigned participants to one of four conditions: public actor, private actor, observer-rebel first, or observer-obedient first. Participants in the first two conditions were actors who completed a police decision task before seeing a previous participant’s response. We instructed private actor participants to place their response to the decision task in a sealed folder, which they would take home with them. In contrast, to increase the salience of the experimenter’s presence, we instructed public actor participants to leave their completed decision tasks face-up on their desk for the duration of the study, consistent with the procedure of Monin et al. (2008, Study 4). Observer participants did not complete a police decision task. Instead, they began the study by viewing and rating both a rebel and an obedient other, the order of which was counterbalanced. Further, because the key alternatives addressed in this study related to comparisons within the actor and observer conditions, we decided a priori not to test actor-obedient cells captured by Monin et al. After viewing, but before rating the targets, all participants also completed an “impression essay” about what kind of person they thought the other was like.
Materials
We used identical stimuli as Monin et al. (2008). A police lineup task was presented on a single page with three photos of male suspects and information about each suspect’s name, alibi, and previous police record. Public and private actor participants were instructed to imagine that they were the detective in charge of the robbery case, and to carefully consider all of the information, circle the photo of the suspect they thought was most likely to be guilty, and provide written justification for their choice.
The task was designed to incriminate the only African American suspect, Steven Jones. Jones had no alibi, had a previous record, was carrying cash and a screwdriver, and was unemployed. The other two suspects, who were White, were unlikely to be guilty: they both had alibis, clean records, and no suspicious possessions. Public and private actor participants encountered this task twice, once when completing it themselves and once when viewing the “previous participant’s” response, always a rebel (see following lines). Observer participants did not complete the task themselves and rated two tasks—the same rebel that actor participants rated, and an obedient target—in counterbalanced order.
Target stimuli
In the obedient condition, on the completed survey, the ostensible previous participant had circled the African American face and provided the reasons for doing so. In the rebel condition, no face was circled, and the handwritten text communicated that the participant had refused to make a choice (see Table 3). The other participant was always White, age 19, and matched the participant’s gender.
Measures
Consistent with Monin et al. (2008), participants in all conditions indicated how much they would like, respect, and want to work on a project with the other participant on 11-point bipolar scales, ranging from −5 (dislike very much) to +5 (like very much). They also indicated the personality trait word that they thought best described the other participant (open-ended). Finally, participants rated the other participant on 14 semantic differential items, including stupid–intelligent; strong–weak; unpleasant–pleasant; funny–not funny; confident–insecure; immature–mature; active–passive; cruel–kind; nice–awful; cold–warm; honest–dishonest; unfair–fair; moral–immoral; dependent–independent. Participants in the two observer conditions completed these measures twice, once for the rebel other and once for the obedient other. Participants in the actor condition rated only the rebel other.
Results
Preliminary Analyses
One participant refused to choose anyone, calling the task biased, 1 did not complete the ratings, 3 expressed suspicion at debriefing, and 1 was African American. We excluded these six participants, leaving 66 cases for the analyses that follow. We created an attraction scale by averaging scores of the liking, working with, and respect items for the rebel in all four conditions (α = .90) and for the obedient target in the observer conditions (α = .80).
Between-Condition Comparisons of Reactions to Rebels
An omnibus ANOVA on attraction for the rebel revealed that the four conditions differed significantly, F(3, 62) = 3.61, p = .018, ηp2 = .15. Comparing the two observer cells to the two actor cells yielded the predicted difference, consistent with Monin et al. (2008): actors approved of the rebel less (M = 0.33, SD = 2.54) than observers (M = 2.13, SD = 2.09), F(1, 64) = 9.91, p = .002, ηp2 = .13. Exploring these four cells more closely and to test Alternative 1 (impression management), each actor condition differed from the two observer conditions, public actor: t(62) = 2.28, p = .026; private actor: t(62) = 2.76, p = .008 (means in Table 1). By contrast, the private actor—whose task was kept hidden from the experimenter—did not differ from the public actor—whose task was visible to the experimenter—t(62) = 0.505, p = .615, ruling out Alternative 1, that actors are motivated by social approval to reject rebels. Further, the two observer ratings of the rebel—completed first or after an obedient target—did not differ significantly from each other, t(62) = 0.90, p = .372, ruling out Alternative 2, that rebels rated first would anchor observers’ first impressions and thereby garner more approval than rebels rated after obedient targets.
Means and standard deviations of reported attraction to target by condition, Study 1.
Note. Attraction items (liking, respect, want to work with) were measured on 11-point scales (−5 to +5) and averaged to create composite measures.
Note that although our study was somewhat underpowered with an average of 16.5 participants in each of the four cells, the predicted actor–observer difference in derogating the rebel obtained at p = .002 and the lack of support for Alternatives 1 and 2 did not come from lack of power. In both cases the pattern of means is in the opposite direction as what the alternative would predict: private actors were, if anything, more critical of the rebel (M = 0.11) than public actors (M = 0.53), ns, and observers actually liked the rebel more after rating an obedient other (M = 2.47) than when they encountered the rebel first (M = 1.75), though again ns. Thus, our rejection of Alternatives 1 and 2 does not seem to result from a lack of statistical power.
Open-ended responses also reflected resentment toward rebels in both public (e.g., stubborn, neurotic, self-righteous) and private actor conditions (e.g., ignorant, self-righteous, uptight), while observers seemed to express more respect for rebels (e.g., assertive, strong-willed, confident, intelligent) as well as for obedient targets (e.g., independent, mature, sensible, logical). We provide all open-ended responses in Table 2, giving us some insight into the flavor of the rejection (for observers there were a few nonresponses, explaining the different number of traits listed). Additionally, when one of the authors (blind to condition) coded these spontaneous descriptions of the targets as either positive or negative, we found that observers used primarily positive words to describe either an obedient other (27/32, or 84% positive) or a rebel (22/32, or 69% positive). By contrast, when actors described a rebel, they used instead a majority of negative words (16/34, or 47% positive), χ2(2) = 10.40, p < .01. Providing further evidence undermining Alternative 1, public (53% positive) and private actor (40% positive) conditions did not ascribe significantly more critical attributions to the rebel, χ2(1) = .54, ns. Further, observers in this analysis actually tended to be more favorable of the rebel (81% positive) when rated after an obedient target than before (56% positive; Alternative 2 would predict the opposite), although this difference was not significant here, χ2(1) = 2.33, p = .13. Thus, the analysis of spontaneous open-ended trait descriptors reinforces the more constrained data presented before.
Open-ended trait responses volunteered by participants to describe the rebel and obedient targets, Study 1.
Note: Observer responses on the same row are from the same participant. Post-hoc coded responses blind to condition are noted after each trait.
Discussion
Study 1 explored two previously untested alternative explanations for the rejection of rebels. Efforts to protect one’s self-worth from moral threat are driven at least in part by social motives, such as approval and acceptance by others (Leary, 2007; Leary & Baumeister, 2000). When actors reject moral rebels, we reasoned that perhaps they are motivated less by the rebel than by social approval from those in the immediate context, such as the experimenter. This theorizing would suggest that impression management is responsible for the derogation of the rebel (Alternative 1). This alternative explanation, however, was not supported by Study 1. The evidence suggests that participants did not reject moral rebels because they were motivated to preserve their impression in the presence of the experimenter; instead, participants were just as likely to reject the rebel even when the experimenter was unaware of their private responses to the police task.
A second alternative explanation derives from the power of anchoring and first impressions to determine task norms (Alternative 2). When judging others, individuals tend to rely on their own perspective as an egocentric starting point, adjusting their judgments accordingly when others interpret the same stimulus differently than they do (Epley et al., 2004). Consistent with this theorizing, Alternative 2 suggests that actors reject rebels because they first encounter the police task and anchor on their own egocentric perceptions of compliance as the “normal” response. As a result, they see the rebel as odd, whereas observers who first see the task in the context of the rebel’s refusal may encode rebellion as the appropriate response. Findings from Study 1 do not support this alternative either. Replicating previous studies, observers who had not completed the task themselves did not reject the rebel. Furthermore, we found that observers who first saw a target comply with the task (thus setting the “normal” response as compliance according to the anchoring interpretation) were not more critical of a subsequent rebel (this was in fact the cell where the rebel was liked most, as reflected also in the trait analysis). Contrary to Alternative 2, this evidence suggests that for actors to derogate rebels it is essential that the self be involved in one’s first experience with the task.
Rather than impression management or egocentric anchoring, Study 1 results provide evidence consistent with previous theorizing that posits an intrapersonal mechanism of imagined reproach. Study 2 explores this internal mechanism more deeply. It is still unclear whether the resentment results from the rebel’s pronouncements (Alternative 3a) or actions (Alternative 3b), or if the imagined reproach could have been attributed to any observer and yielded the same rejection (Alternative 4). Study 2 tests these possibilities.
Study 2
In Study 2, we parsed two forms of principled deviance that have been previously conflated. In prior studies, behavior on the decision task was combined with a verbal justification: obedient targets chose the African American suspect and justified it (e.g., “he was the most likely suspect”), while rebels abstained and provided a moral justification for their abstention, in somewhat condemning terms (e.g., “. . .this task is obviously biased . . . offensive. . .”). The conflation of rebellion and of the principled comments in past studies left ambiguous what feature triggered derogation. If the key component for participants was the rebel’s verbal judgmentalism (calling the task and therefore the experimenter “offensive”), then such language should be enough to trigger rejection even if the target ends up completing the task (Alternative 3a). Conversely, if the key component is the actual refusal to comply regardless of the judgment of others, then targets who refuse to go along, even if their language implies no judgment of those who comply, will trigger rejection by actor participants (Alternative 3b). To test these questions in Study 2, we created two new targets: a nonjudgmental rebel who refuses to choose a suspect but expresses understanding of those who do; and, a judgmental obedient target who chooses the most likely suspect consistent with the vast majority of participants, but who also expresses disapproval of the task itself. Table 3 summarizes these stimuli.
Manipulation of target responses by condition, Studies 1 and 2.
Additionally, it remains possible that the fear of imagined rejection captured by the mediation results in Monin et al. (2008) has little to do with the rebel him- or herself (who, participants knew, would never actually have a chance to see the participant’s answer). Instead, perhaps fear of imagined rejection actually captures a form of self-reproach that is projected on the rebel, just as it would be projected on anyone else whom the participants are asked to imagine (Alternative 4), as Monin and colleagues theorized in their paper. Study 2 also tests this possibility by measuring anticipated reproach by a random observer.
Method
Participants
One hundred sixty-three individuals at a private West Coast university (67 White, 62 Asian, 18 Latino, 4 Multiracial, 4 Indian, 4 Native American, 2 African American, and 2 opted not to report their ethnicity; 104 females, 59 males; M = 20.26, SD = 2.34) came to the laboratory to participate in individual experimental sessions in exchange for $8.00.
Procedure
To explore Alternatives 3a and 3b, we created a 2 (Target: Rebel; Obedient) x 2 (Response Language: Judgmental; Nonjudgmental) factorial design, creating four conditions that varied by which target participants rated (see Table 3), including the two original targets from Study 1—a verbally judgmental rebel and a nonjudgmental obedient target—in addition to the two new targets designed to test Alternative 3a (judgmental obedient target) and Alternative 3b (nonjudgmental rebel). We matched all targets with participant age and gender. All participants were actors and completed the decision task before (a) writing a brief first-impression description of the target as in Study 1, (b) rating the target on the same items and scales (liking, respect, work with; α = .89), (c) reporting how they expected the target to rate them on the same scales, and (d) to explore Alternative 4, reporting how they expected a random other participant to rate them if they could see both the participant’s and the target’s task responses, specifically they read:
Imagine we showed your and the other participant’s responses on the police task to a random student at [participant’s institution]. After seeing your answers on the “Criminal Decision Making Task,” how much would a random student like to work on a project with you, like you as a friend, and respect you? [three scales measured separately]
In addition, we included two new measures to explore future behavioral intentions, asking participants whether they would like to sign up for future dyadic studies in our lab: “Please contact me to participate in your upcoming studies” (yes, no), and whether they would like to work with the same target person they rated in the current study, “I would rather be paired with the same/a different student” (same, different).
Results
Excluded Participants
Of the initial 163 participants, 12 reported suspicion at the funneled debriefing, 5 chose someone other than the African American suspect as the likely culprit, 2 were African American, and 1 participant’s condition information was missing due to experimenter error. The following analyses are based on the remaining 143 participants.
Attraction
See Table 4 for all means and standard deviations discussed in what follows. As before, an omnibus ANOVA revealed that participant ratings of the target varied as a function of target condition, F(3, 139) = 10.37, p < .001, ηp2 = .18. Consistent with Monin et al. (2008), participants liked the original (judgmental) rebel less than the original (nonjudgmental) obedient target, t(139) = 5.38, p < .001, MSE = 4.50.
Means and standard deviations of reported attraction to and anticipated attraction by target;percentages of those interested in follow-up experiments and preference to work with a different person than the target, Study 2.
Note. Cells with different superscripts connote significantly different means at p < .05 using post-hoc pairwise comparisons (df = 139, attraction to target MSE = 4.50; imagined attraction by target MSE = 3.05). Attraction items (liking, respect, want to work with) were measured on 11-point scales (−5 to +5) and averaged to create composite measures.
Testing Alternative 3a (moral language alone leads to rejection), actors disliked the judgmental obedient target more than the original nonjudgmental obedient target, t(139) = 2.12, p = .035, suggesting that judgmental moralized language is enough to trigger some rejection of an obedient target. Still, participants liked this target more than the original judgmental rebel, t(139) = 3.21, p = .002. Thus, moralized language is enough to produce some rejection, but cannot fully account for the full derogation of judgmental rebels.
Testing Alternative 3b (rebellion alone leads to rejection), participants liked the new nonjudgmental rebel less than the original obedient target, t(139) = 3.59, p < .001, so rebellion alone seemed enough to trigger some rejection despite the target’s claim to be nonjudgmental of those who comply. And yet participants did like this nonjudgmental rebel marginally more than the original judgmental rebel, t(139) = 1.81, p = .073. Similar to our findings pertaining to Alternative 3a, isolating the behavioral component of rebellion invites some rejection, but again, cannot fully account for the level of derogation the original rebel faces for both moralized language and action.
Exploring the relative strength of reactions to moralized language and behavior of the rebel, these two new targets (judgmental obedient and nonjudgmental rebel) did not differ significantly from one another on attraction, t(139) = 1.43, p > .15.
Anticipated Attraction
In the next two sections, we test Alternative 4: how participants expect others to judge them and whether these anticipated reactions—from each of the four targets and from random peers—account for participants’ subsequent rejection of them. To begin, we explore the first step in this model (imagined reproach) and report the full mediation paths in the following section.
Anticipated attraction from each target mirrors how participants predicted that target would rate them (again, see means and standard deviations in Table 4). Consistent with Monin et al. (2008), participants expected to be rejected by judgmental rebels, but accepted by nonjudgmental obedient targets, t(139) = 9.08, p < .001, MSE = 3.05. We next explored how participants expected to be judged by obedient targets who used moralized language to judge the task (related to Alternative 3a). As intended, compared to the original nonjudgmental obedient target, participants expected the new judgmental obedient target to like them less, t(139) = 3.30, p = .001, but more than the imagined reproach from the original judgmental rebel, t(139) = 5.70, p < .001.
Related to Alternative 3b (rebellion alone), actors expected the nonjudgmental rebel to approve of them less than the nonjudgmental obedient other, t(139) = 4.78, p < .001, but more than the judgmental rebel, t(139) = 4.31, p < .001. Interestingly, the fact that participants expected a nonjudgmental rebel to like them less than a nonjudgmental obedient target suggests that they did not quite believe the rebel’s claim that he or she would not judge others who participated in the racist task. This may explain why participants still resented the seemingly nonjudgmental rebel: despite his or her claim to the contrary, they still expected him or her to judge them based on what they knew of his or her own decision.
Mediation: Imagined Rejection by the Target Versus by a Random Peer
Why do participants derogate any individual who takes a moral stance—is it because individuals expect to be rejected by moralizing others or, as Alternative 4 suggests, because individuals expect an audience of their peers to reject them when their actions are compared to a morally superior other?
First, we dummy coded the four targets to create a set of three contrasts on which we regressed attraction and both imagined attraction ratings (Baron & Kenny, 1986). We tested for this mediation between our two original targets, the judgmental rebel and the nonjudgmental obedient target, while controlling for the other two dummy coded contrasts. This contrast predicts attraction ratings of the target, B = −2.69, t(139) = 5.38, p < .001, imagined attraction by the target, B = −3.74, t(139) = 9.08, p < .001, and imagined attraction by a random other peer, B = −1.59, t(139) = 5.36, p < .001.
However, when we regress attraction, imagined attraction by the target, and imagined attraction by a random other on the contrast variable, the contrast is no longer significant, B = −.96, t(137) = 1.63, p = .11, while imagined attraction is still significant, B = .42, t(137) = 3.76, p < .001, and importantly, imagined attraction by a random other is not, B = .09, t(137) = 0.60, p = .55. Using the bootstrapping procedures outlined by Preacher and Hayes (2004) for multiple mediators and the guidelines for mediation with a multicategorical independent variable (Hayes & Preacher, 2014), the indirect effect of imagined attraction by the target is significant with 5,000 resamples (−1.58, SE = 0.65), as the bias-corrected bootstrapped 95% confidence interval does not include zero [−2.88, −0.34], while the indirect effect around imagined attraction by a random peer is not (−0.02, SE = 0.08; CI [−0.30, 0.07]; see Hayes, 2013, Model 6). Thus, consistent with what Monin and colleagues predicted, participants seem to reject the rebel because they imagine the rebel will reject them. However, expected judgment by a random peer did not mediate this effect (Alternative 4), contrary to what Monin et al. (2008) had also predicted. It seems that the imagined reproach needs to come from the rebel to trigger rejection of the rebel.
Finally, we tested for mediation with the two new types of targets using the same procedures outlined before (see Table 5 for dummy codes and regression coefficients). In both cases, participants reported more attraction for the original obedient target (which served as the comparison in the dummy coding), which was mediated by their imagined attraction ratings of that target. Importantly, this relationship was not statistically mediated by imagined attraction by a random peer. Furthermore, in all instances, participants expected to be accepted less by the target, which drove their reduced approval of that target—a process specific to the imagined audience of the target claiming moral superiority, and not driven by any generic audience (contrary to Alternative 4).
Dummy codes and regression coefficients for multicategorical multiple mediation analyses, Study 2.
Note. a1 = dummy predicting imagined attraction by the target; b1 = imagined attraction by target predicting attraction to target; a2 = dummy predicting imagined attraction by a random other; b2 = imagined attraction by a random other predicting attraction to target; c = total effect of dummy on attraction to target; c’ = dummy predicting attraction, controlling for imagined attraction by the target and by a random other.
p < .05. **p < .01. ***p < .001.
Behavioral Intentions
There was no difference between our four conditions in participants’ willingness to take part in (yes/no) future follow-up experiments, χ2(3) = .73, p = .87 (see Table 4 for percentages by condition). However, there was a difference between groups in their willingness to work with the same target, χ2(3) = 13.36, p = .04 (see Table 4). Only 17% of participants (n = 6 of 35) coupled with a nonjudgmental obedient other asked to be paired with a different person in a future study, but this number nearly tripled for the judgmental rebel (45%; n = 15 of 33), χ2(1) = 6.38, p = .01. In fact, when the rebel claimed not to be judgmental, this did not reduce the number of participants who expressed a preference to work with a different person in the future (47%, n = 16 of 34), so the nonjudgmental claim did not make the rebel more likeable as far as this behavioral choice measure is concerned. Results were more ambiguous for peers who called the task offensive but ended up going along with it (32%, n = 11 of 34): these judgmental obedient targets were not significantly more likely to be rejected than rebels, χ2(1) = 1.21, ns, but also not significantly less likely to be rejected than obedient targets, χ2(1) = .49, ns, essentially falling in the middle. The rates of rejection for the two new targets were also not significantly different, χ2(1) = 1.54, ns.
Discussion
Study 2 aimed to address two central questions: What is the precise trigger that leads participants to reject the rebel (the verbal condemnation or the actual refusal—via Alternatives 3a and 3b), and do participants reject the rebel because they imagine that anyone would condemn them (Alternative 4)? Results from Study 2 suggest that actors reject moral rebels for merely expressing a moral stance, even if they still go along with the racist task, consistent with Alternative 3a. Conversely, actors also reject moral rebels who take a stand by refusing to complete the task, even if they attempted to express sympathy and understanding for those who would have taken part themselves, consistent with Alternative 3b. Thus, it seems that both the verbal condemnation (3a) and the actual refusal (3b) may have been active ingredients in the rejection reported by Monin et al. (2008). Further, rejection is strongest when both elements are combined, as in our judgmental rebel, but either is enough to trigger some rejection.
Importantly, Study 2 addressed two mechanisms for such rejection: Is it because actors imagine the target will reject them or because any imagined audience would reject them? Mediation results suggest the former: Individuals expect to be rejected by the target, which leads them to make a preemptive strike against him or her. It is not because any imagined audience might judge them for participating in the task, contrary to Alternative 4. However, it is interesting to note that individuals do expect a random other peer to like them less when comparing their responses to a judgmental rebel who refused to participate in the task; yet, this anticipated reaction from third parties is not the motivating trigger that leads them to reject the rebel.
Finally, Study 2 provides the first evidence for actual social choice in this paradigm: when asked whether they would be willing to be paired with the target for a subsequent experiment, participants were nearly 3 times as likely to request a switch when they had seen the rebel (45%) than when they had seen an obedient other (17%).
General Discussion
How do individuals respond to another’s moral stance? While others’ principled deviance may elevate us to pursue virtuous actions ourselves, it can also make us feel inadequate. Previous studies by Monin et al. (2008) demonstrated that another’s moral stance can cause involved others to feel a pang of moral threat, leading them to reject these rebels, while uninvolved observers extol the very same rebels. Building on this work, the two studies in this paper explored whether individual actors are threatened more by the interpersonal, public scrutiny they might face when their amoral behavior is contrasted with another’s principled stand, or whether the threat is driven by more intrapersonal processes.
Two studies ruled out alternative explanations that account for derogation of deviants who take a principled stand. Study 1 provided evidence that threat from morally superior others, not the threat of censure from others in the immediate environment (the experimenter, Alternative 1), drives individuals to reject moral rebels. Even when no one else witnessed their behavior, private actor participants rejected rebels as much as actors whose own behavior was public. Study 1 also ruled out the possibility that rejection was driven by one’s first encounter with the task (Alternative 2): individuals who first witnessed another’s compliance felt no less positively toward the newcomer rebel than those who first observed rebellion. Finally, what seems to matter is whether someone else expresses moral dissatisfaction with the task either verbally (Alternative 3a) or by refusing to participate (Alternative 3b), which leads participants to anticipate social rejection and to derogate the source of that threat. Contrary to predictions by Monin et al. (2008), Study 2 demonstrated that this process of rejection is driven by imagined reproach specifically from moralizing others and not from an audience of random peers (Alternative 4). Finally, Study 2 also showed for the first time with an actual social choice measure that many participants would actually prefer to avoid the rebel target in subsequent interactions.
These findings rule out five alternative explanations for rebel rejection that rely on external influences of moral threat. Participants did not derogate the rebel due to anticipated reactions from any external audience (random bystanders [Alternative 4] or real experimenters [Alternative 1]). In fact, we suspect that participants are likely to believe that others would have behaved as they did (Ross, Greene, & House, 1977), and would therefore support their behavior. Rather, rebel derogation arose from the imagined reproach uniquely elicited by the rebel’s response even though that person was distant from the study context and unlikely ever to meet participants. These studies thus support the notion of a more self-involved process leading to the rejection of rebels.
Our Mind’s Imagined Juries
Some of our most important audiences live on in our heads. Individuals work hard to elicit the pride of parents or grandparents who are no longer living, and experience private guilt when they imagine what friends and loved ones would think of their missteps, even when those audiences are not physically present to witness them. Imagining or being primed with the faces of a private audience (e.g., older family members) affects individuals’ experience of the self and their own self-evaluations (Baldwin, 1997; Baldwin & Holmes, 1987). In one study, graduate students primed with the face of their department chair were more critical of their own research ideas than those primed with the familiar face of a lower status colleague (Baldwin, Carrell, & Lopez, 1990). Like the two studies in this paper, these findings suggest that individuals’ self-concepts can be shaped less by others’ actual perceptions of them (e.g., the “looking glass” self; Cooley, 1902; Mead, 1934), than by what individuals imagine others think of them (Bem, 1967; Kenny & DePaulo, 1993; Shrauger & Schoeneman, 1979). The present studies provide further evidence that anticipated rejection by moral rebels stems less from the rebel’s actual judgment of individuals and more from how they imagine the rebel reproaches them in “the back of their minds” (Baldwin et al., 1990, p. 437).
Limitations and Future Directions
In this paper, we propose that the internalized moral rebel, like members of both groups of imagined audiences, can be absent from our immediate environment, but can become a private audience, evoking shame and disapproval because individuals know that they would condemn them if given the opportunity. Though the rebel’s judgment is not real, its imagined presence can cause a very real sting of moral reproach, leading to resentment. Our studies suggest an intriguing possibility: our social lives may be less literally social than we think. We may react more often than we recognize to our imagined audiences than to the physically real audiences in our environment. At the same time, our more private moments may be more social than we recognize, shared with different forms of imagined others.
While evidence from Study 2 is consistent with an internalized process of projection, we acknowledge important limitations of these studies. Actors imagined what several audiences (e.g., judgmental rebel, random peer) would think of their own behavior—none of whom were physically present or would ever meet participants—and these imagined reactions drove participants to reject the rebel. Nonetheless, we should not overstate these mediation results nor overemphasize the nuanced conclusions we are able to draw without more concrete data about the projection process. One concern with Study 2, for example, is that the imagined rejection by the random other might have mediated rejection of the random other, something we do not measure, since our focus is on the rejection of the rebel target. More research is necessary to fully understand the intrapersonal processes at work in rejecting moral others, especially with care to more directly measure and test some of the mechanisms we begin to explore in the present studies, such as individuals’ self-perceptions in the process of psychological projection and imagined rejection.
In fact, one exciting possibility related to the predictions of others’ attitudes is the intrapersonal root cause of imagined rejection by rebels: it is possible that individuals who witness rebellion feel a pang of regret at the realization that they are not as moral as they could have been (Gilovich & Medvec, 1995), which leads them to believe that others will condemn their behavior. Thus, self-criticism could be the first step in the ultimate rejection of moral others. On the one hand, sensing one’s own moral shortcomings could highlight the threat imposed by social comparison with the rebel, leading individuals to reject him or her. On the other hand, admitting and acknowledging one’s moral shortcomings might reduce the likelihood that threat leads to derogation (e.g., why reject the rebel if individuals agree they are not as moral as him or her?), consistent with the self-affirmation evidence in Monin et al. (2008, Study 4). Future work could carefully test the paths from threat to self-criticism and ultimate rejection.
Last, the present studies contribute to the broader literature on deviance and subjective group dynamics. On one hand, individuals tend to prefer in-group members who deviate from group norms by upholding them even more strongly (“pronorm deviants”) compared to those who deviate by violating them (Abrams, Marques, Bown, & Dougill, 2002; Abrams, Marques, Bown, & Henson, 2000; Marques, Abrams, Paez, & Martinez-Taboada, 1998). On the other hand, our results suggest that in some instances individuals respond with antipathy toward pronorm deviants, such as moral rebels in the present studies who arguably do more of the right thing than actor participants from the same in-group. These possibly divergent results underscore the importance of the self in evaluating other in-group members. When in-group members are uninvolved, and therefore not at risk to threatening social comparison (e.g., observer participants in the present studies; see also Abrams et al., 2002, Study 2), pronorm deviants who uphold and bolster the in-group identity for all members will earn support. However, the exemplary behavior of pronorm deviants, such as the rebels in the present studies, can also threaten involved in-group members by highlighting what those individuals could have or should have done. Thus, when the self is at stake in-group members may resent pronorm deviants’ ideal behavior because it exposes their own shortcomings. Future work could extend this contribution, especially testing the presence or absence of self-involvement in a full model of pro- and antinorm deviance with both in-group and out-group members.
Conclusion
Others who take a moral stance on an issue reframe a possibly ambiguous context as a moralized one (Kreps & Monin, 2011). In this case, making a different, moral choice compared to others invites criticism, as does expressing a moralized opinion while making the same complicit choice as others. Reactions to both targets imply a pessimistic future for those considering the high road: any form of moralization appears to threaten others’ moral self-regard. Because others anticipate being rejected by those who took the high road, they lash out and reject those would-be juries. Future research should investigate alternative routes to moralized action and speech, exploring options to pursue virtue without the risk of threatening others and inviting subsequent resentment.
The possibility that we ourselves generate the threatening moral comparisons that lead us to reject others is an intriguing one. When another’s principled stance contradicts our own past behavior and opinion, it matters less what that other person actually thinks about or says to us than what we imagine they say in our own minds. While moral others can be in the position to castigate us for our immoral or amoral behaviors in the past, these findings suggest that they need not ever say anything to us directly; our own imaginations are sufficient to cause threatening feelings of moral inadequacy. Thus, we may reject moral rebels for the things they could have said to us, rather than any admonitions they issue themselves.
Footnotes
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.