To Confront or to Avoid

Abstract

What do people want to do in response to witnessing someone violate a moral norm? Prior research posits that violations of distinct norms elicit specific emotions, specifically anger and disgust. We examined whether moral violations analogously elicit distinct behavioral responses, focusing on desires to confront and avoid moral violators. Participants read scenarios depicting harmful and impure actions (Study 1) or violations of all six content domains proposed by Moral Foundations Theory (Study 2). Bayesian inference revealed that participants expressed distinctively high levels of desire to avoid (vs. confront) violators of purity norms. Violations of other moral norms did not similarly elicit unique patterns of avoidance or confrontation. Thus, behavioral responses to moral violators depend in part on which norm was violated, with impure acts eliciting a uniquely strong avoidance response. Moral judgment can serve as a precursor to strategic action in the face of perceived immorality.

Keywords

moral judgment moral foundations avoidance confrontation Bayesian inference

What do people want to do in response to witnessing someone violate a moral norm? Seeing a person kicking a puppy, for instance, might elicit anger, whereas seeing a brother and sister passionately kissing might instead elicit disgust. Why do specific types of moral violations elicit anger and disgust? According to the social functionalist perspective (Hutcherson & Gross, 2011), distinct emotional states map onto context-relevant behaviors. For example, threatening stimuli elicit fear, which activates the appropriate fight-or-flight response. Does witnessing moral violations similarly elicit context-specific behavioral responses? Specifically, we examined whether violations of distinct moral norms uniquely motivate people to confront or avoid the violator.

Moral violations elicit emotional reactions due to their social relevance (Hutcherson & Gross, 2011). For example, witnessing perpetration of physical or emotional harm (e.g., seeing someone kick a puppy or insult a friend) typically activates the emotional response of anger (Graham et al., 2013; Whitton, Henry, Rendell, & Grisham, 2014). Depending on its severity, anger may lead one to want to confront the perpetrator. Because confrontation is often instrumental in correcting behavior (Martin & Cushman, 2016), we investigated desire to confront as one potential response to moral violators. By contrast, witnessing behavior that appears impure or degraded (e.g., incest, cannibalism) tends to activate disgust (Graham et al., 2013), the purpose of which may be to prompt avoidance (Curtis, Aunger, & Rabie, 2004; but see Landy & Goodwin, 2015). Accordingly, we examined desire to avoid as another potential response to moral violators. Although confrontation and avoidance may share a common desired end state (e.g., punishment), they are psychologically distinct for the perceiver (e.g., confrontation may be more costly; Ohtsuki, Iwasa, & Nowak, 2009) and may be chosen based on contextual cues.

Prior work has primarily examined emotional responses to moral violators. We propose that focusing only on emotions is insufficient for methodological and conceptual reasons. Methodologically, there is no consensus concerning the measurement of moral emotions. This is especially the case with regard to disgust and anger, which interact and overlap substantially (Salerno & Peter-Hagene, 2013; Yoder, Widen, & Russell, 2016). Indeed, the word “disgust” may encompass instances of anger, and people may use it to denote general dislike rather than disgust more specifically (Kayyal, Pochedly, McCarthy, & Russell, 2015). Thus, people may not be able to report and may not reliably experience distinct emotional reactions to moral violations. Moreover, disgust and anger are highly similar in terms of valence and arousal (Cameron, Lindquist, & Gray, 2015). Accordingly, conceptually, understanding emotional responses may not straightforwardly allow one to predict how people want to behave toward moral violators (Cameron et al., 2015). Thus, we assessed behavioral intentions directly.

Research on moral emotions suggests specific predictions regarding the types of behaviors that violations of distinct moral norms may elicit. For example, the specific correspondence model (Chapman & Anderson, 2013) posits that specific moral emotions map onto specific violation types. Thus, any action in which an entity was harmed should predictably elicit anger and any action in which a bodily norm was violated should predictably elicit disgust (see also Rozin & Haidt, 2013; Russell & Giner-Sorolla, 2013). The specific correspondence model therefore suggests that witnessing harmful actions should uniquely activate a desire to confront the violator, whereas witnessing impure actions should uniquely elicit avoidance. By contrast, constructionist models of moral emotion argue that there are no exclusive relationships between moral content domains and emotions (Cameron et al., 2015). Rather, contextual cues and conceptual knowledge (e.g., who was harmed and how; Gray & Schein, 2012) inform the interpretation of core affect in specific emotional terms, and no specific stimulus will reliably elicit the same emotion across contexts (Cameron et al., 2015). Constructionist models thus suggest that no differences in desires to confront versus avoid will reliably emerge from distinct moral violation types.

We used a Bayesian approach to examine whether desires to confront and avoid uniquely map onto violations of specific moral norms. Unlike frequentist approaches, Bayesian inference allows for directly comparing the likelihood of two models, given observed data (Rouder, Morey, Verhagen, Swagman, & Wagenmakers, 2017). Frequentist methods are not ideal for the present research question, because failing to reject the null hypothesis does not yield evidence that it is true. However, evidence for a null effect would disconfirm specific correspondence patterns. For example, a lack of difference in desires to confront versus avoid violators of distinct moral norms would provide evidence against the notion that specific violations uniquely map onto specific behavioral responses.

Study 1

In Study 1, we examined desires to confront and avoid violators of norms related to harm and purity, because prior research suggests alternative predictions for violations of these two norms in particular. We utilized the framework of Moral Foundations Theory, a social intuitionist model of moral psychology that proposes that salient aspects of moral violations activate different domains, which determine key features of subsequent moral appraisal (Graham et al., 2013; Haidt, 2012). In Study 1, we focused on violations of care norms (harmful actions) and sanctity norms (impure actions).

Based on the specific correspondence model of moral emotion (Chapman & Anderson, 2013), we derived the hypotheses that desire to confront should be greater in response to care (vs. sanctity) violations, and the opposite should be the case for desire to avoid. Additionally, desire to confront (vs. avoid) should be greater in response to care violations, and the opposite should be the case for sanctity violations. Any other pattern would disconfirm predictions based on the specific correspondence model. Participants read scenarios describing harmful and impure acts, reported their desire to confront and avoid each violator, and provided ratings of moral wrongness. We included nonmoral social norm violation scenarios as a control condition. Preregistered hypotheses can be found at https://osf.io/zuwxk/.

Method

Participants

Our a priori goal was to recruit at least 250 participants; planned procedures are preregistered at https://osf.io/u8k92/. In total, 278 participants (48.6% women; mean age = 18.85, standard deviation [SD] = 1.34) were recruited via an Introduction to Psychology research pool in return for class credit. For race/ethnicity (participants could check multiple options), 62.2% responded White, 12.9% responded Black, 14.0% responded Latino/Latina, 12.9% responded Asian, 2.5% responded multiracial, and 6.5% responded other. For class rank, 55.8% were freshmen, 26.3% were sophomores, 12.6% were juniors, and 5.4% were seniors.

Procedure

Participants completed the study individually at desktop computers in groups of 1–4. Participants completed the Care, Sanctity, and Social Norm subscales of the Moral Foundations Vignettes (Clifford, Iyengar, Cabeza, & Sinnott-Armstrong, 2015) and provided demographic information.¹ Measures were preregistered (https://osf.io/8bh2c/).

Materials

The Moral Foundations Vignettes contain one-sentence scenarios that depict people committing moral violations. Seven subscales represent violations of six moral foundations and nonmoral social norms. Pilot testing by Clifford et al. (2015) confirmed that all scenarios were interpreted as violations of the intended moral foundation. For Study 1, participants read scenarios from the Care (27 items; e.g., “You see a boy telling a woman that she looks just like her overweight bulldog”), Sanctity (9 items; e.g., “You see a man having sex with a frozen chicken before cooking it for dinner”), and Social Norm (16 items; e.g., “You see someone at the school gym lifting free weights in nice business clothes”) subscales, for a total of 52 scenarios presented in a randomized order. We excluded one scenario from the Sanctity subscale because it described reading about a distant tribe, making confrontation and avoidance impractical. For each scenario, participants provided three ratings (1 = not at all to 5 = extremely): (a) How wrong is this action?, (b) How likely would you be to want to avoid this person?, and (c) How likely would you be to want to confront this person? Ratings were averaged within each subscale (αs = .74–.95).

Results

We employed Bayesian inference (using the JASP software; JASP Team, 2016). The key statistic in Bayesian inference is the Bayes factor, which is a ratio of the likelihood of one model compared to another, given observed data. The Bayes factor can be stated either as a comparison of alternative model likelihood to null model likelihood (BF₁₀) or a comparison of null model likelihood to alternative model likelihood (BF₀₁). Bayes factors ranging from 3 to 10 are considered “some evidence” for the likelihood of the model of interest over the alternative; Bayes factors between 10 and 30 are considered “strong evidence,” 30 and above are considered “very strong evidence,” and below 3 are “inconclusive” (Jeffreys, 1961; Rouder, Speckman, Sun, Morey, & Iverson, 2009). We used these guidelines to interpret our findings, acknowledging that they can be sensitive to subjective interpretation and choice of prior (Lavine & Schervish, 1999). Most of our Bayes factors were orders of magnitude above the cutoff for very strong evidence and robust to prior choice. We report effect sizes in terms of δ (mean or total effect of difference divided by SD; Rouder et al., 2009), with 95% credible intervals (CIs). The CIs indicate that one can be 95% confident that the true effect size falls within the interval (Morey, Hoekstra, Rouder, Lee, & Wagenmakers, 2016; Wagenmakers, Morey, & Lee, 2016). No participants were excluded. The analysis plan was preregistered (https://osf.io/5krht/).

Avoidance and Confrontation

A 3 (violation type: care, sanctity, social norm) × 2 (behavior: avoid, confront) Bayesian repeated-measures analysis of variance (ANOVA) revealed very strong evidence for an interaction, BF₁₀ = 1.48e⁺³⁴⁵ (see Figure 1). Follow-up paired-samples t tests compared behavioral responses across violation type (for Bayes factors, see Table 1); a default Cauchy prior width of 0.707 was used for all t tests. Participants reported greater desire to avoid violators of sanctity norms (M = 4.01, SD = 0.64) than violators of care norms (M = 3.15, SD = 0.83). Participants also reported greater desire to confront violators of care norms (M = 3.23, SD = 0.71) than violators of sanctity norms (M = 2.32, SD = 0.89). Finally, participants reported less desire to confront (M = 1.71, SD = 0.64) and avoid (M = 1.66, SD = 0.53) social norm violators compared to both care and sanctity violators.

Figure 1.

Mean desire to avoid and confront the violator across the three violation types in Study 1. Possible range: 1–5. Error bars represent standard errors of the mean.

Table 1.

Bayes Factors (BF₁₀), Strength of Evidence Indicating a Difference Between Conditions (Relative to a Null Difference), and Effect Sizes (Median δ with 95% Credible Interval [CI]) for Comparisons of Desires to Avoid and Confront the Violator Across the Three Violation Types in Study 1.

Behavior	Comparison	BF₁₀	Evidence	Median δ [95% CI]
Avoidance	Care vs. sanctity	2.51e⁺⁵¹	Very strong	−1.18 [−1.33, −1.02]
	Care vs. social norm	1.47e⁺⁷⁷	Very strong	1.64 [1.46, 1.82]
	Sanctity vs. social norm	1.63e⁺¹⁴⁰	Very strong	3.08 [2.81, 3.38]
Confrontation	Care vs. sanctity	2.15e⁺⁴⁴	Very strong	1.06 [0.92, 1.21]
	Care vs. social norm	1.41e⁺⁷⁶	Very strong	1.62 [1.44, 1.80]
	Sanctity vs. social norm	8.37e⁺¹⁹	Very strong	0.65 [0.52, 0.77]

We next compared desires to avoid and confront within violation type. Participants reported greater desire to avoid than confront violators of sanctity norms, BF₁₀ = 2.55e⁺⁷², median δ = 1.55, 95% CI [1.38, 1.73]. By contrast, participants reported similar levels of avoidance and confrontation toward violators of care norms, with the Bayes factor indicating some evidence of a null difference, BF₀₁ = 4.96, median δ = −0.09, 95% CI [−0.20, 0.03]. As expected, participants reported similar levels of avoidance and confrontation toward social norm violators, with the Bayes factor indicating some evidence of a null difference, BF₀₁ = 6.14, median δ = −0.08, 95% CI [−0.20, 0.04].

Perceived Wrongness

A one-way Bayesian repeated-measures ANOVA on perceived wrongness revealed very strong evidence for differences across violation types, BF₁₀ = 2.12e⁺²⁶⁶. Social norm violations (M = 1.91, SD = 0.61) were perceived as less wrong than care violations (M = 3.85, SD = 0.53), BF₁₀ = 3.97e⁺¹¹⁹, median δ = 2.54, 95% CI [2.30, 2.80], and sanctity violations (M = 4.06, SD = 0.57), BF₁₀ = 2.47e⁺¹³⁹, median δ = 3.06, 95% CI [2.79, 3.36]. Sanctity violations were perceived as more wrong than care violations, BF₁₀ = 2.02e⁺⁸, median δ = −0.41, 95% CI [−0.53, −0.29].

Discussion

Three findings provided support for predictions derived from the specific correspondence model of moral emotion. First, participants wanted to avoid sanctity violators more than they wanted to avoid care violators. Second, desire to avoid sanctity violators was stronger than desire to confront them. These findings suggest that there is something special about individuals who violate moral purity norms that elicits avoidance. People may perceive these violators as contaminating and thus seek to avoid contact (Curtis et al., 2004). Third, participants wanted to confront care violators more than they wanted to confront sanctity violators. Thus, something about witnessing harmful actions elicits a desire to confront, more so than witnessing “impure” actions does. This response may prepare the observer to intervene or prevent further infractions (Hutcherson & Gross, 2011; Whitton et al., 2014). The pattern across these three findings suggests that people are tuned into information about which norm is being violated when responding to moral violators.

By contrast, a fourth finding was not in line with the specific correspondence model: People were equally likely to want to confront and avoid violators of care norms. Witnessing harmful acts may elicit a mixture of reactions, contrary to the notion of a one-to-one mapping. People may be torn between the motivations to reprimand perpetrators of harm to prevent further offense (Martin & Cushman, 2016) and to express disapproval that a moral rule was violated by avoiding the violator (Kupfer & Giner-Sorolla, in press). Although Study 1 revealed that care violations were perceived as less wrong than sanctity violations (notably, this finding was not replicated in Study 2), it is possible that care violators were perceived to cause more directly harmful outcomes. Participants may thus have been less inclined to want to directly confront sanctity violators, given that people pay particular attention to outcomes when making judgments about appropriate punishment (Martin & Cushman, 2016).

We acknowledge that the specific correspondence model concerns moral emotion (specifically anger and disgust), whereas we assessed desires to confront and avoid. We may have failed to find evidence for predictions derived from this model simply because we did not measure emotion. Accordingly, we conceptually replicated Study 1 in an independent sample recruited from the same university research pool (N = 313; see https://osf.io/925xv/). Participants were exposed to the same care, sanctity, and social norm violations and indicated how angry at and disgusted by the violators they were. Mirroring Study 1, participants were more disgusted by sanctity violators than care violators, were angrier at care violators than sanctity violators, and were more disgusted by than angry at sanctity violators. Notably, participants were also more disgusted by than angry at care violators—the opposite of what the specific correspondence model predicts. This finding may be due to the tendency to use the word disgust to encompass instances of anger (Kayyal et al., 2015).

Our results thus far support the notion of a unique mapping of desires to confront and avoid solely in the case of sanctity violations. However, because our comparisons were limited to care and sanctity violations, our results may simply suggest that something unique about impure acts elicits distinct responses. That is, Study 1 might speak more to the unusual nature of sanctity violations than to how behaviors map onto violation types (see Gray & Keeney, 2015). We addressed this issue in Study 2.

Study 2

In Study 2, we examined reactions to violations of all six domains from the Moral Foundations Theory—care, sanctity, fairness, liberty, loyalty, and authority (Graham et al., 2013). Fairness norms represent violations of reciprocity (e.g., lying about hours worked on a timesheet); violations of liberty norms involve encroachment on personal freedom (e.g., dictating how one’s family should vote in the election); violations of loyalty norms involve betrayal of group values (e.g., cheering for the opposing sports team); and violations of authority norms involve affronts to legitimate leaders and institutions (e.g., calling one’s professor a fool during class). We considered two alternative predictions. First, sanctity violations may uniquely elicit a distinct pattern of desire to avoid (compared to violations of other norms and to desire to confront), with all other violations eliciting confrontation and avoidance to equal degrees. This pattern would be consistent with Dehghani et al. (2016), who found that moral purity concerns were uniquely predictive of social distance. We contrasted this hypothesis with the alternative that violations of all “binding” norms (sanctity, loyalty, and authority), which represent group-level violations (Graham, Haidt, & Nosek, 2009), show a distinctive pattern of avoidance, whereas violations of “individualizing” norms (care, fairness, and liberty) do not.

Method

Participants

Our a priori aim was to collect data from at least 250 participants (see https://osf.io/26mfc/). We analyzed data from 282 participants (42.2% women; mean age = 33.72, SD = 10.35) recruited via Amazon Mechanical Turk in return for US$0.50. Of the 412 participants who began the study, we excluded 130 participants who responded to ≤50% of the scenarios. This exclusion criterion was determined a priori before analyses were conducted (https://osf.io/w6jau/). Of these 130 participants, 101 (78%) responded to only 0–5 of the 104 scenarios; thus, data were excluded primarily because participants did not provide sufficient responses.

Procedure and Materials

We preregistered our procedures (https://osf.io/fk9ch/), measures (https://osf.io/e9v2q/), and analysis plan (https://osf.io/26mfc/). Study 2 followed the same procedure as Study 1, with the exception that participants responded to the complete Moral Foundations Vignettes (with randomized presentation order). Participants read scenarios from the Care (27 items), Fairness (12 items; e.g., “You see a referee intentionally making bad calls that help his favored team win”), Liberty (11 items; e.g., “You see a man telling his girlfriend that she must convert to his religion”), Loyalty (16 items; e.g., “You see a college president singing a rival school’s fight song during a pep rally”), Authority (13 items; e.g., “You see a student stating that her professor is a fool during an afternoon class”), Sanctity (9 items), and Social Norm (16 items) subscales, for a total of 104 scenarios. We excluded one scenario from the Authority subscale because it described a group (rather than an individual). Participants responded to the same three questions as in Study 1 for each scenario (αs = .84–.96).

Results

Avoidance and Confrontation

A 7 (violation type: care, fairness, liberty, authority, loyalty, sanctity, social norm) × 2 (behavior: avoid, confront) Bayesian repeated-measures ANOVA revealed very strong evidence for an interaction, BF₁₀ = 1.01e⁺⁵⁷⁵ (see Figure 2) that we followed up with paired-samples t tests.

Figure 2.

Mean desire to avoid and confront the violator across the seven violation types in Study 2. Possible range: 1–5. Error bars represent standard errors of the mean.

Comparing desires to avoid and confront across violation types (see Table 2), we found that participants wanted to avoid sanctity violators (M = 3.83, SD = 0.84) more than they wanted to avoid care violators (M = 3.32, SD = 0.88). Participants also wanted to confront care violators (M = 3.18, SD = 0.86) more than they wanted to confront sanctity violators (M = 2.39, SD = 0.94). These findings replicate Study 1. The majority of the remaining Bayes factors also indicated strong support for differences in desires to avoid and confront based on violation type. This was not the case in the comparison of loyalty and authority violations, which elicited equal desires to confront (loyalty: M = 2.23, SD = 0.90; authority: M = 2.24, SD = 0.81) and avoid (loyalty: M = 2.53, SD = 0.85; authority: M = 2.49, SD = 0.79). Additionally, fairness and liberty violations elicited equal desire to avoid (fairness: M = 3.12, SD = 0.88; liberty: M = 3.08, SD = 0.90), although desire to confront differed across these violations (fairness: M = 3.13, SD = 0.89; liberty: M = 2.66, SD = 0.92). Moreover, care and fairness violations elicited equal desire to confront. Social norm violations elicited less desire to avoid (M = 1.48, SD = 0.65) and confront (M = 1.34, SD = 0.66) than all other violations.

Table 2.

Bayes Factors (BF₁₀), Strength of Evidence Indicating a Difference Between Conditions (Relative to a Null Difference, Except Where the Opposite Contrast Is Indicated), and Effect Sizes (Median δ with 95% Credible Interval [CI]) for Comparisons of Desires to Avoid and Confront the Violator Across the Seven Violation Types in Study 2.

Behavior	Comparison	BF₁₀	Evidence	Median δ [95% CI]
Avoidance	Care vs. fairness	1.81e⁺⁶	Very strong	0.36 [0.24, 0.48]
	Care vs. liberty	5.92e⁺⁹	Very strong	0.44 [0.32, 0.56]
	Care vs. authority	1.25e⁺⁵³	Very strong	1.20 [1.04, 1.35]
	Care vs. loyalty	2.58e⁺⁴⁰	Very strong	0.99 [0.85, 1.13]
	Care vs. sanctity	6.41e⁺¹⁶	Very strong	−0.58 [−0.71, −0.45]
	Care vs. social norm	1.08e⁺⁸⁴	Very strong	1.75 [1.56, 1.93]
	Fairness vs. liberty	0.12	Some (for null)	0.06 [−0.06, 0.18]
	Fairness vs. authority	9.30e⁺⁴²	Very strong	1.03 [0.88, 1.17]
	Fairness vs. loyalty	1.29e⁺³⁰	Very strong	0.82 [0.68, 0.95]
	Fairness vs. sanctity	4.60e⁺²⁶	Very strong	−0.76 [−0.89, −0.62]
	Fairness vs. social norm	1.97e⁺⁷⁸	Very strong	1.64 [1.46, 1.82]
	Liberty vs. authority	7.94e⁺²⁸	Very strong	0.80 [0.66, 0.93]
	Liberty vs. loyalty	1.34e⁺¹⁹	Very strong	0.62 [0.50, 0.75]
	Liberty vs. sanctity	7.63e⁺²⁷	Very strong	−0.78 [−0.91, −0.65]
	Liberty vs. social norm	4.91e⁺⁷⁴	Very strong	1.57 [1.39, 1.75]
	Authority vs. loyalty	0.13	Some (for null)	−0.07 [−0.19, 0.05]
	Authority vs. sanctity	6.26e⁺⁶⁸	Very strong	−1.46 [−1.63, −1.30]
	Authority vs. social norm	2.35e⁺⁶⁰	Very strong	1.32 [1.16, 1.48]
	Loyalty vs. sanctity	5.79e⁺⁶⁷	Very strong	−1.45 [−1.62, −1.28]
	Loyalty vs. social norm	1.13e⁺⁵⁸	Very strong	1.28 [1.12, 1.44]
	Sanctity vs. social norm	1.15e⁺¹⁰⁵	Very strong	2.17 [1.96, 2.38]
Confrontation	Care vs. fairness	0.16	Some (for null)	0.08 [−0.03, 0.19]
	Care vs. liberty	2.06e⁺³¹	Very strong	0.83 [0.70, 0.97]
	Care vs. authority	5.32e⁺⁵⁹	Very strong	1.31 [1.15, 1.46]
	Care vs. loyalty	5.94e⁺⁵⁰	Very strong	1.16 [1.00, 1.30]
	Care vs. sanctity	1.56e⁺³⁹	Very strong	0.97 [0.83, 1.11]
	Care vs. social norm	4.08e⁺⁸⁶	Very strong	1.79 [1.61, 1.99]
	Fairness vs. liberty	4.00e⁺¹⁹	Very strong	0.63 [0.51, 0.76]
	Fairness vs. authority	2.02e⁺⁵⁵	Very strong	1.23 [1.08, 1.39]
	Fairness vs. loyalty	1.01e⁺⁵⁰	Very strong	1.14 [1.00, 1.29]
	Fairness vs. sanctity	1.95e⁺³³	Very strong	0.87 [0.73, 1.00]
	Fairness vs. social norm	5.17e⁺⁸²	Very strong	1.72 [1.53, 1.91]
	Liberty vs. authority	2.83e⁺¹⁵	Very strong	0.56 [0.44, 0.68]
	Liberty vs. loyalty	3.82e⁺¹¹	Very strong	0.48 [0.36, 0.60]
	Liberty vs. sanctity	40,543.63	Very strong	0.31 [0.20, 0.43]
	Liberty vs. social norm	1.52e⁺⁶³	Very strong	1.37 [1.21, 1.53]
	Authority vs. loyalty	0.07	Strong (for null)	0.02 [−0.10, 0.13]
	Authority vs. sanctity	32.08	Very strong	−0.21 [−0.33, −0.09]
	Authority vs. social norm	2.14e⁺⁵⁵	Very strong	1.23 [1.08, 1.39]
	Loyalty vs. sanctity	20.15	Strong	−0.20 [−0.32, −0.09]
	Loyalty vs. social norm	2.44e⁺⁴⁷	Very strong	1.10 [0.95, 1.25]
	Sanctity vs. social norm	8.32e⁺⁴⁹	Very strong	1.14 [0.99, 1.29]

We next compared desires to confront and avoid within violation type. Replicating Study 1, participants wanted to avoid sanctity violators more than they wanted to confront them, BF₁₀ = 1.07e⁺⁵³, median δ = 1.19, 95% CI [1.04, 1.35]. We observed an inconclusive difference in desires to avoid versus confront care violators, BF₁₀ = 1.41, median δ = 0.15, 95% CI [0.03, 0.27]. Notably, this pattern suggested a greater desire to avoid than confront care violators (the opposite of what specific correspondence models predict). Strong evidence was revealed for no difference between desires to avoid and confront fairness violators, BF₀₁ = 14.95, median δ = 0.00, 95% CI [−0.12, 0.11]. Bayes factors for all other within-group comparisons yielded very strong evidence for desire to avoid (vs. confront) being greater (liberty violations: BF₁₀ = 4.57e⁺⁹, median δ = 0.44, 95% CI [0.32, 0.56]; authority violations: BF₁₀ = 142,006.32, median δ = 0.33, 95% CI [0.21, 0.45]; loyalty violations: BF₁₀ = 3.10e⁺⁸, median δ = 0.41, 95% CI [0.29, 0.53]; social norm violations: BF₁₀ = 2.74e⁺⁷, median δ = 0.39, 95% CI [0.27, 0.51]).

Individualizing Versus Binding Norms

We created composites of avoidance and confrontation for violations of individualizing (care, fairness, and liberty) and binding (loyalty, authority, and sanctity) norms, following Clifford et al. (2015). A 2 (violation type: individualizing, binding) × 2 (behavior: avoid, confront) Bayesian repeated-measures ANOVA revealed very strong evidence for an interaction, BF₁₀ = 4.87e⁺⁷⁴. Paired-samples t tests revealed very strong evidence that desire to avoid violators of individualizing norms (M = 3.22, SD = 0.83) was greater than desire to avoid violators of binding norms (M = 2.82, SD = 0.71), BF₁₀ = 2.19e⁺²², median δ = 0.68, 95% CI [0.55, 0.81]. Similarly, desire to confront violators of individualizing norms (M = 3.05, SD = 0.81) was greater than desire to confront violators of binding norms (M = 2.27, SD = 0.80), BF₁₀ = 6.00e⁺⁵⁸, median δ = 1.29, 95% CI [1.13, 1.45]. Although desire to avoid (vs. confront) was greater in both cases, this difference was larger for violations of binding norms.

Perceived Wrongness

A one-way Bayesian repeated-measures ANOVA revealed very strong evidence for a difference in perceived wrongness across violations, BF₁₀ = 3.76e⁺⁴³⁵ (see Table 3). Fairness violations (M = 3.79, SD = 0.63) were considered more wrong than all other violations (except for sanctity violations, for which there was only some evidence of a difference), followed by care (M = 3.68, SD = 0.65) and sanctity (M = 3.66, SD = 0.82) violations (which were considered equally wrong), liberty violations (M = 3.40, SD = 0.75), authority violations (M = 3.06, SD = 0.71), and loyalty violations (M = 2.86, SD = 0.84). Social norm violations (M = 1.43, SD = 0.63) were considered the least wrong. Violations of individualizing norms were considered more wrong (M = 3.64, SD = 0.60) than violations of binding norms (M = 3.12, SD = 0.70), BF₁₀ = 5.50e⁺³⁹, median δ = 0.98, 95% CI [0.83, 1.12].

Table 3.

Bayes Factors (BF₁₀), Strength of Evidence Indicating a Difference Between Conditions (Relative to a Null Difference, Except Where the Opposite Contrast Is Indicated), and Effect Sizes (Median δ with 95% Credible Interval [CI]) for Comparisons of Perceived Wrongness Across the Seven Violation Types in Study 2.

Comparison	BF₁₀	Evidence	Median δ [95% CI]
Care vs. fairness	152.89	Very strong	−0.24 [−0.35, −0.12]
Care vs. liberty	7.18e⁺¹²	Very strong	0.51 [0.38, 0.63]
Care vs. authority	5.24e⁺⁴⁵	Very strong	1.07 [0.93, 1.22]
Care vs. loyalty	4.80e⁺⁴⁷	Very strong	1.10 [0.96, 1.25]
Care vs. sanctity	0.07	Strong (for null)	0.02 [−0.10, 0.14]
Care vs. social norm	1.38e⁺¹¹⁶	Very strong	2.42 [2.19, 2.65]
Fairness vs. liberty	5.07e⁺¹⁹	Very strong	0.64 [0.51, 0.76]
Fairness vs. authority	7.79e⁺⁶⁰	Very strong	1.33 [1.16, 1.49]
Fairness vs. loyalty	1.59e⁺⁵⁴	Very strong	1.21 [1.06, 1.37]
Fairness vs. sanctity	6.08	Some	0.18 [0.06, 0.30]
Fairness vs. social norm	6.27e⁺¹¹⁸	Very strong	2.48 [2.24, 2.72]
Liberty vs. authority	1.15e⁺¹¹	Very strong	0.47 [0.34, 0.59]
Liberty vs. loyalty	1.11e⁺¹⁸	Very strong	0.60 [0.48, 0.73]
Liberty vs. sanctity	7,869.78	Very strong	−0.29 [−0.41, −0.18]
Liberty vs. social norm	1.27e⁺¹⁰⁰	Very strong	2.06 [1.86, 2.27]
Authority vs. loyalty	173,049.16	Very strong	0.33 [0.21, 0.45]
Authority vs. sanctity	1.28e⁺³²	Very strong	−0.85 [−0.99, −0.71]
Authority vs. social norm	1.93e⁺⁹⁵	Very strong	1.96 [1.76, 2.17]
Loyalty vs. sanctity	1.14e⁺⁴⁹	Very strong	−1.13 [−1.27, −0.98]
Loyalty vs. social norm	4.35e⁺⁷⁹	Very strong	1.66 [1.48, 1.85]
Sanctity vs. social norm	2.19e⁺¹⁰⁸	Very strong	2.24 [2.02, 2.47]

Discussion

Even with a greater variety of moral norms considered, violations of sanctity norms elicited distinctively high levels of desire to avoid the violator, both relative to other violations and to desire to confront. This pattern aligns with the notion that impure behaviors are unique in terms of the response they elicit (Dehghani et al., 2016). Additionally, violations of liberty, authority, and loyalty norms elicited a similar (though less extreme) preference for avoidance over confrontation as sanctity violations, whereas care and fairness violations elicited equal desires to avoid and confront. Moreover, care and sanctity violations were judged to be equally wrong in Study 2, whereas in Study 1 sanctity violations were judged more wrong. Despite this discrepancy, patterns of avoidance and confrontation remained consistent across the two studies, indicating that something beyond perceived wrongness (e.g., perceived contamination) is driving behavioral responses.

By contrast, Study 2 did not support the alternative prediction that the broader category of violations of binding (vs. individualizing) norms elicits distinctive avoidance. Participants rated violations of individualizing (vs. binding) norms as more wrong, and violations of individualizing norms elicited stronger desires to avoid and confront. People may place greater moral weight on individualizing (vs. binding) norms. For example, violations of individualizing norms may be perceived as more representative of immorality (Gray & Schein, 2012). However, we advise caution in comparing individualizing versus binding norms, given substantial within-category variation. Whereas liberty norms are theorized to fall in the individualizing domain (Haidt, 2012), in Study 2 reactions to liberty violations more closely resembled reactions to violations of binding norms. Similarly, reactions to sanctity violations were distinctly different from reactions to authority and loyalty violations. Thus, the individualizing/binding distinction may miss important nuance.

General Discussion

People have varying reactions to moral violations. Our participants wanted to confront perpetrators of harm more than perpetrators of impure acts, and they wanted to avoid perpetrators of impure acts more than perpetrators of harm. Additionally, participants reported a much stronger desire to avoid than confront perpetrators of impure acts, compared to other kinds of perpetrators. Thus, whether people prefer to avoid or confront moral violators depends partially on which norm was violated.

However, we found no evidence for a specific correspondence mapping of behavioral response to violation type. A unique mapping emerged only in the case of purity violations, which distinctively elicited avoidance. Although this preference to avoid was most pronounced in the case of purity violations, it held for all violations except those involving care and fairness norms. Thus, faced with two options, people prioritize avoiding aberrant behavior over correcting wrongful actions, perhaps because they are considering potential third-party observer evaluations. Expressions of disgust and anger in response to moral violations convey different social signals, with disgust being perceived as a signal of general moral condemnation and anger as a signal of indignation at a self-directed offense (Kupfer & Giner-Sorolla, in press). If confrontation and avoidance are behavioral reflections of anger and disgust (Hutcherson & Gross, 2011), people may be tuned into how these behaviors are perceived by others. This possibility could explain the overall preference for avoidance we observed, as avoidance may be less likely than confrontation to be perceived as self-interested. Avoidance may thus be more effective than confrontation because it ostracizes the violator while being less costly to the perceiver (Ohtsuki et al., 2009). Perhaps people must view moral infractions as severe enough to warrant the cost of confrontation. Indeed, in the present research, care and fairness violations were perceived to be especially morally wrong and also elicited the greatest desire to confront.

Why do violations of purity norms uniquely elicit avoidance? One possibility is that people seek to prevent contamination. This explanation aligns with the notion that moral disgust is psychologically and physiologically analogous to physical disgust (Rozin & Haidt, 2013). Another possibility is that people simply view impure acts as abnormal (Gray & Keeney, 2015). This interpretation presents a potential issue for comparing purity violations to other moral violations, as the perceived abnormality of purity violations may lead to qualitatively different judgments. For example, people make harsher character judgments of purity (vs. harm) violators, even though harmful (vs. impure) acts are judged more severely (Chakroff & Young, 2015). Indeed, people may attribute purity violations to a flawed moral character (Rozin, Haidt, & McCauley, 2000). Such person-based attribution may lead people to assume that impure behavior could not easily be corrected via confrontation, causing a preference for avoidance. A third possibility is that purity violations are perceived as unlikely to affect people other than the perpetrator, whereas perpetration of harm typically impacts others. Perhaps people see less need to reprimand or confront perpetrators of “victimless” purity violations and prefer to simply avoid targets engaging in abnormal behavior. By contrast, there may be more of an impetus to confront perpetrators of harm in order to prevent future harm to other victims. Future research should further investigate the underlying motivations for these action tendencies (e.g., wanting to avoid targets to ostracize them or to protect oneself or others) to better understand when and why people prefer confrontation or avoidance.

Our findings may have implications for moral regulation, blame, and punishment. For example, a distinction has been proposed between proscriptive (what one should not do) and prescriptive (what one should do) morality (Janoff-Bulman, Sheikh, & Hepp, 2009). Proscriptive morality is avoidance-motivated and concerned with inhibiting harm, whereas prescriptive morality is approach-motivated and concerned with assigning praise. People view proscriptively immoral acts (failing to avoid harm) as more blameworthy than prescriptively immoral acts (failing to do good; Janoff-Bulman et al., 2009). However, our results suggest that forms of approach-oriented behavior are not limited to the positive domain of praise: Witnessing violations of care and fairness norms led participants to want to not only avoid but also confront (i.e., approach) the violator. Future research would benefit from exploring how desires to confront (and other approach-oriented behaviors) versus avoid moral violators relate to punishment. For example, whereas avoidance may serve to punish violators indirectly via ostracism, approach-oriented behaviors may serve to restore justice via overt punitiveness.

We acknowledge a key limitation to our findings: The unique desire to avoid sanctity violators may reflect differences in perceived wrongness and typicality across violation types, rather than reflecting distinctive responses to impurity-related behaviors per se. We conducted exploratory analyses to address this issue (for details, see https://osf.io/w6jau/). First, we identified three care and three sanctity violations that were more closely matched on perceived wrongness and typicality (based on validation data from Clifford et al., 2015). When we restricted the Study 1 analysis to these scenarios, the pattern remained intact and reflected a distinctive preference to avoid sanctity violators. Second, we curated a subset of three authority, three loyalty, and three sanctity violations that were more closely matched on perceived wrongness and typicality (again based on Clifford et al., 2015). When we restricted the Study 2 analysis to these scenarios, we again found that only sanctity violations elicited a distinctive preference for avoidance; thus, the pattern did not generalize to all binding moral foundations. Although we were unable to definitively match all scenarios, these exploratory results are consistent with the interpretation that responses to sanctity violators are indeed unique in a way that cannot be fully explained by differential perceptions of wrongness or typicality.

To conclude, our results help inform the study of moral psychology in general. First, not all kinds of moral violations are equal. Affronts to purity, care, and reciprocity elicit more severe moral judgments and stronger behavioral reactions than affronts to personal liberty, authority, and group cohesion. Thus, responses to purity violations do not match authority or loyalty violations, suggesting that categorizing these three violation types together as group-directed affronts disregards important distinctions. Caution should be used when comparing responses to purity violations to other moral infractions, as purity violations may elicit disproportionately skewed action tendencies. Finally, our findings provide insight into behaviors that may follow from witnessing moral violations. Moral judgment is not merely a cognitive and emotional exercise but more importantly a precursor to strategic action in the face of perceived immorality.

Footnotes

Declaration of Conflicting Interests

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

Funding

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

Note

References

Buss

A. H.

Perry

(1992). The aggression questionnaire. Journal of Personality and Social Psychology, 63, 452–459.

Cameron

C. D.

Lindquist

K. A.

Gray

(2015). A constructionist review of morality and emotions: No evidence for specific links between moral content and discrete emotions. Personality and Social Psychology Review, 19, 371–394.

Chakroff

Young

(2015). Harmful situations, impure people: An attribution asymmetry across moral domains. Cognition, 136, 30–37.

Chapman

H. A.

Anderson

A. K.

(2013). Things rank and gross in nature: A review and synthesis of moral disgust. Psychological Bulletin, 139, 300–327.

Clifford

Iyengar

Cabeza

Sinnott-Armstrong

(2015). Moral Foundations Vignettes: A standardized stimulus database of scenarios based on moral foundations theory. Behavior Research Methods, 47, 1178–1198.

Curtis

Aunger

Rabie

(2004). Evidence that disgust evolved to protect from risk of disease. Proceedings of the Royal Society of London B: Biological Sciences , 271, S131–S133.

Dehghani

Johnson

Hoover

Sagi

Garten

Parmar

N. J.

… Graham

(2016). Purity homophily in social networks. Journal of Experimental Psychology: General, 145, 366–375.

Graham

Haidt

Koleva

Motyl

Iyer

Wojcik

S. P.

Ditto

P. H.

(2013). Moral foundations theory: The pragmatic validity of moral pluralism. Advances in Experimental Social Psychology, 47, 55–130.

Graham

Haidt

Nosek

B. A.

(2009). Liberals and conservatives rely on different sets of moral foundations. Journal of Personality and Social Psychology, 96, 1029–1046.

10.

Gray

Keeney

J. E.

(2015). Impure or just weird? Scenario sampling bias raises questions about the foundation of morality. Social Psychological and Personality Science, 6, 859–868.

11.

Gray

Schein

(2012). Two minds vs. two philosophies: Mind perception defines morality and dissolves the debate between deontology and utilitarianism. Review of Philosophy and Psychology, 3, 1–19.

12.

Haidt

(2012). The righteous mind: Why good people are divided by politics and religion. New York, NY: Pantheon.

13.

Haidt

McCauley

Rozin

(1994). Individual differences in sensitivity to disgust: A scale sampling seven domains of disgust elicitors. Personality and Individual Differences, 16, 701–713.

14.

Hutcherson

C. A.

Gross

J. J.

(2011). The moral emotions: A social-functionalist account of anger, disgust, and contempt. Journal of Personality and Social Psychology, 100, 719–737.

15.

Janoff-Bulman

Sheikh

Hepp

(2009). Proscriptive versus prescriptive morality: Two faces of moral regulation. Journal of Personality and Social Psychology, 96, 521–537.

16.

JASP Team. (2016). JASP (Version 0.8.0.0 Beta) [Computer software]. Amsterdam, the Netherlands: JASP Team.

17.

Jeffreys

(1961). Theory of probability (3rd ed.). Oxford, England: Oxford University Press.

18.

Kayyal

M. H.

Pochedly

McCarthy

Russell

J. A.

(2015). On the limits of the relation of disgust to judgments of immorality. Frontiers in Psychology, 6, 951.

19.

Kupfer

T. R.

Giner-Sorolla

(in press). Communicating moral motives: The social signaling function of disgust. Social Psychological and Personality Science.

20.

Landy

J. F.

Goodwin

G. P.

(2015). Does incidental disgust amplify moral judgment? A meta-analytic review of experimental evidence. Perspectives on Psychological Science, 10, 518–536.

21.

Lavine

Schervish

M. J.

(1999). Bayes factors: What they are and what they are not. The American Statistician, 53, 119–122.

22.

Martin

J. W.

Cushman

F. A.

(2016). The adaptive logic of moral luck. In Sytsma

Buckwalter

(Eds.), The Blackwell companion to experimental philosophy (pp. 190–202). Malden, MA: Wiley Blackwell.

23.

Morey

R. D.

Hoekstra

Rouder

J. N.

Lee

M. D.

Wagenmakers

E.-J.

(2016). The fallacy of placing confidence in confidence intervals. Psychonomic Bulletin & Review, 23, 103–123.

24.

Ohtsuki

Iwasa

Nowak

M. A.

(2009). Indirect reciprocity provides only a narrow margin of efficiency for costly punishment. Nature, 457, 79–82.

25.

Rouder

J. N.

Morey

R. D.

Verhagen

Swagman

A. R.

Wagenmakers

E. J.

(2017). Bayesian analysis of factorial designs. Psychological Methods, 22, 304–321.

26.

Rouder

J. N.

Speckman

P. L.

Sun

Morey

R. D.

Iverson

(2009). Bayesian t tests for accepting and rejecting the null hypothesis. Psychonomic Bulletin & Review, 16, 225–237.

27.

Rozin

Haidt

(2013). The domains of disgust and their origins: Contrasting biological and cultural evolutionary accounts. Trends in Cognitive Sciences, 17, 367–368.

28.

Rozin

Haidt

McCauley

(2000). Disgust. In Lewis

Haviland-Jones

(Eds.), Handbook of emotions (pp. 637–653). New York, NY: Guilford Press.

29.

Russell

P. S.

Giner-Sorolla

(2013). Bodily moral disgust: What it is, how it is different from anger, and why it is an unreasoned emotion. Psychological Bulletin, 139, 328–351.

30.

Salerno

J. M.

Peter-Hagene

L. C.

(2013). The interactive effect of anger and disgust on moral outrage and judgments. Psychological Science, 24, 2069–2078.

31.

Wagenmakers

E.-J.

Morey

R. D.

Lee

M. D.

(2016). Bayesian benefits for the pragmatic researcher. Current Directions in Psychological Science, 25, 169–176.

32.

Whitton

A. E.

Henry

J. D.

Rendell

P. G.

Grisham

J. R.

(2014). Disgust, but not anger provocation, enhances levator labii superioris activity during exposure to moral transgressions. Biological Psychology, 96, 48–56.

33.

Yoder

A. M.

Widen

S. C.

Russell

J. A.

(2016). The word disgust may refer to more than one emotion. Emotion, 16, 301–308.