To Defer or to Stand up? How Offender Formidability Affects Third Party Moral Outrage

Abstract

According to models of animal behavior, the relative formidability of conspecifics determines the utility of deferring versus aggressing in situations of conflict. Here we apply and extend these models by investigating how the formidability of exploiters shapes third party moral outrage in humans. Deciding whether to defer to or stand up against a formidable exploiter is a complicated decision as there is both much to lose (formidable individuals are able and prone to retaliate) and much to gain (formidable individuals pose a great future threat). An optimally designed outrage system should, therefore, be sensitive to these cost-benefit trade-offs. To test this argument, participants read scenarios containing exploitative acts (trivial vs. serious) and were presented with head-shot photos of the apparent exploiters (formidable vs. non-formidable). As predicted, results showed that, compared to the non-formidable exploiter, the formidable exploiter activated significantly more outrage in male participants when the exploitative act was serious. Conversely, when it was trivial, the formidable exploiter activated significantly less outrage in male participants. However, these findings were conditioned by the exploiters' perceived trustworthiness. Among female participants, the results showed that moral outrage was not modulated by exploiter formidability.

Introduction

Over the course of human evolutionary history, our ancestors have repeatedly been confronted with exploiters – individuals trying to harvest benefits at the expense of others. As demonstrated by a number of laboratory and field studies, this adaptive problem has selected for cognitive and motivational counter-mechanisms including activation of moral outrage and punitive sentiments in second and third parties in response to acts of exploitation (Fehr and Gächter, 2002; Greene and Haidt, 2002; Kahneman and Frederick, 2002; Price, Cosmides, and Tooby, 2002; Wilson and O'Gorman, 2003). Importantly, however, all exploitative acts are not sanctioned equally. For example, research shows that kin, physically attractive, socially valuable and trustworthy individuals are treated more lenient when engaging in acts of exploitation (Ahola, Christianson, and Hellström, 2009; Feather and Atchison, 1998; Lieberman and Linke, 2007; Petersen, Sell, Tooby, and Cosmides, 2010; Stewart, 1980). In this paper, we add to this research agenda by investigating how the formidability of exploiters is factored into third party reactions to exploitation.

According to the concept of resource holding power (Parker, 1974) derived from evolutionary biology of animal conflict (Archer, 1988; Hammerstein and Parker, 1982; Maynard Smith and Price, 1973), the relative formidability (the capacity to inflict costs on others) of conspecifics determines the utility of deferring versus aggressing in situations of conflict. If an individual is more formidable than oneself, one is better off deferring and, conversely, if an individual is equal or less formidable than oneself, one is better off aggressing. This model has been validated across a range of non-human animal species (Archer, 1988; Dugatkin, 1997; Huntingford and Turner, 1987), just as recent studies of humans have produced results consistent with the model. First, studies have demonstrated that humans across cultures are very adept at gauging physical strength and aggressiveness by using a number of subtle cues such as facial and vocal features (Carré, McCormick, and Mondloch, 2009; Sell, Cosmides, Tooby, Sznycer, von Rueden, and Gurven, 2009; Sell et al., 2010). Second, formidability has the predicted behavioral effects on the self. Hence, an association between physical size and dominance has been established in children and in adolescent boys (Pellegrini et al., 2007; Tremblay et al., 1998) as well as in adult men (Archer and Thanzami, 2007, 2009; Felson, 1996; Gallup, White, and Gallup Jr, 2007). Furthermore, Sell, Tooby, and Cosmides (2009) found that men's weight lifting ability predicts feelings of entitlement, proneness to anger, history of fighting, utility of personal aggression, and finally, success in conflict.

Importantly, these previous studies on humans have all focused on how the formidability of the self shapes behavior. Fewer studies have focused on the role of others' formidability in situations of exploitation. The implication of traditional models of animal behavior is that relative formidability matters and that the effects of others' formidability simply mirror the effects of own formidability (see Archer and Benson, 2008; Cummins, 1999, 2005). Yet, while this might apply to most non-human species, it is not necessarily so in the human animal (Boehm, 2000). Rather, a dissection of the selection pressures involved in the presence of formidable exploiters suggests that we may – under specific conditions – be psychologically designed to react with greater outrage in the face of formidable relative to non-formidable exploiters.

Selection for psychological mechanisms designed to produce moral outrage when exploitive acts have occurred is predominantly governed by the adaptive problem of hindering further exploitation (Petersen et al., 2010). Accordingly, in each instance, these mechanisms should be designed to carefully modulate their output by weighing the potential costs of future exploitation against the potential costs of aggressing against the exploiter here and now. In this case, formidable exploiters provide an interesting mix of cross-cutting pressures. On the one hand, as argued above, aggressing against a formidable individual is a risky strategy. On the other hand, a formidable exploiter poses a far greater future threat than a non-formidable exploiter. As argued above, past research shows that formidability increases exploitive tendencies, and thus a formidable exploiter could be more likely than a non-formidable to repeat an offense. Furthermore, given the larger physical strength, a formidable exploiter is more capable of acting based on a given exploitive tendency. These observations imply that formidable exploiters, on average, will inflict greater future costs than non-formidable exploiters.

The first selective gradient implies that it is adaptive to defer to formidable exploiters. The second selective gradient implies it is particularly adaptive to punish formidable exploiters – especially in the face of serious exploitive acts. For most non-human animals, the first selective gradient dominates because they cannot buffer against the costs of punishing formidable opponents. As argued by Boehm (1999; 2000), however, this does not necessarily apply to humans, given their extraordinary abilities to build and mobilize coalitions. By forming large coalitions, non-formidable humans are able to buffer against the costs associated with aggressing against formidable exploiters. In line with this, hunter-gatherer-bands across continents view alpha-male-like dominant behavior (e.g., rape, theft, unprovoked aggression, undue competition) as deviant (Boehm, 1993; Cashdan, 1980; Furer-Haimendorf, 1967; Gardner, 1991) and broad coalitions from below have been observed to take pre-emptive measures to maintain social control in the group through sanctions. In particular, people have been observed to be vigilant and attentive towards controlling high-power individuals (Boehm, 1999; Diamond, 1997; see also Overbeck, Tiedens, and Brion, 2006).

In this respect, moral outrage seems to be of fundamental importance. Moral outrage is a particular aversive arousal state experienced when others violate social norms, encompassing specialized cognitive (harsh character attributions), affective (anger, contempt, and disgust), as well as behavioral (support for infliction of punitive damages) elements (Darley and Pittman, 2003; Haidt, 2003; Tetlock, Kristel, Elson, Green, and Lerner, 2000). A number of studies have demonstrated that moral outrage fuels punitive sanctions towards norm violators (Fehr and Gächter, 2002; Greene and Haidt, 2002; Kahneman and Frederick, 2002; Price, Cosmides, and Tooby, 2002; Wilson and O'Gorman, 2003). In addition, the activation of outrage also involves more social functions. As Tooby and Cosmides argue (2010; see also Tooby, Cosmides, and Price, 2006), broadcasting moral outrage coordinates the attention of coalitional members towards a common problem, thereby allowing for collective actions. Similarly, Tetlock et al. (2000) argues that moral outrage facilitates meta-norm enforcement; that is, support for punishment of those who do not punish the initial transgressor. In the face of formidable exploiters, these features of moral outrage would help mobilize coalitional behavior and reduce the risks and costs of punishments. ¹ Given this, we propose that exploiter formidability influences moral outrage conditioned by the seriousness of the exploitive act. For less serious acts, the risks of aggressing against formidable individuals outweigh future threats and, hence, such exploitive acts are more acceptable when performed by a formidable individual than by a non-formidable individual. For more serious acts, the cost-benefit ratio at some point shifts and more moral outrage is produced against the formidable individual; not necessarily to motivate the individual to engage in punitive behavior on his own, but to mobilize coalitional aggression against the formidable exploiter.

While experimental studies show that victims of exploitative acts (e.g., non-reciprocation, free-riding) in particular are angered and willing to punish exploiters, studies document that unaffected third parties also feel angered and are willing to sacrifice personal winnings in order to punish exploiters (Carpenter and Matthews, 2005; Fehr and Fischbacher, 2004; Fehr and Gächter, 2000, 2002; O'Gorman, Wilson, and Miller, 2005; Ostrom, Walker, and Gardner, 1992). Also, norm enforcement through third party sanctioning is common in most small-scale hunter-gatherer societies (Cronk, Chagnon, and Irons, 2000; Sober and Wilson, 1998). Third parties mobilized to engage in the physical punishment of exploiters will often face the same cost-benefit-ratios as second parties. For example, direct physical aggression against a formidable exploiter is equally costly for a second or a third party. Furthermore, a formidable exploiter may also pose a problem for individuals who were not direct victims of the exploitive act – either because of involvement with the victim or because they could themselves be the next victim (for discussion, see Petersen et al., 2010). In this way, we expect the above selection pressures to not only apply to second parties but also shape how third parties factor formidability into their reactions to exploitation.

Studies in criminology have already provided some evidence for the proposed argument, that exploiter formidability influences third party moral outrage conditioned by the seriousness of the exploitive act. An experimental study with simulated civil trials found that male baby-faced defendants in crime cases involving negligent actions (i.e., in our perspective, less serious exploitative acts) were more likely to be convicted and, if convicted, were punished more compared to male mature-faced defendants (Berry and Zebrowitz-McArthur, 1988). In crime cases involving intentional actions (i.e., more serious exploitative acts), male baby-faced defendants were, in contrast, less likely than mature-faced male defendants to be convicted. A field study of 506 cases in small claims court in USA somewhat corroborated these results (Zebrowitz and McDonald, 1991). Similarly, other judicial studies suggest that high status individuals are punished less for minor transgressions, but conversely punished more for large transgressions (Levine, 1989; Rosoff, 1989; Wiggins, Dill, and Schwartz, 1965).

To corroborate the evolutionary interpretation of these effects, additional predictions can be offered. For second as well as third parties, an exploiter's formidability is a factor of particular importance when punishment is physical (or, at least, when one runs the risk of direct physical confrontation). Cross-cultural evidence suggests that, over evolutionary time, males have engaged much more than females in both coalitional and individual aggression, and that the aggression has predominately been directed against other males (e.g., Chagnon, 1983; Daly and Wilson, 1988; Van Vugt, De Cremer, and Janssen, 2007; Wrangham and Peterson, 1996; see also Tooby and Cosmides, 2010). Therefore, the predictions specified above only apply in a straightforward manner to males. It is males that should fear the formidable male exploiter and be responsible for mobilizing against him. In the case of females, different possibilities suggest themselves from an evolutionary perspective. First, females could be expected to defer to formidable males across all situations such that they would express less outrage against formidable exploiters in both serious and non-serious instances of exploitation. Second, the fact that men on average have 90 percent more upper body strength than women (Bohannon, 1997; Stoll, Huber, Seifert, Michel, and Stucki, 2000) could act as an inhibiting force on female reactions against all male exploiters (whether formidable or non-formidable relative to other males). If this is true, females should express less outrage than males against male exploiters. Finally, given that female third party punishment would only rarely take the form of direct confrontation and rather involve the imposition of indirect social costs, e.g., in the form of gossiping (see Fry, 1998), it is plausible that females simply do not modulate their outrage to the formidability of the exploiter.

Materials and Methods

Participants

To test these possibilities, a sample of 1621 Danish respondents (812 men) in age from 18 to 70 (M = 43.91, SD = 13.91) were recruited via the survey-company Yougov Zapera. To ensure national representativeness, the sample was based on quota sampling and drawn to match the national distribution on gender, sex and geographical region. In addition, the data is weighted on these parameters prior to analysis. The experiment was conducted online in November 2009.

Experimental design

The study used a 2 (formidable vs. non-formidable exploiter) x 2 (serious vs. trivial exploitation) between-subjects experimental design. Subjects were randomly allocated to one of the four conditions. To manipulate formidability, we relied on manipulated head-shot pictures of the exploiter. To manipulate the seriousness of the exploitive act, two fictional scenarios were created – each with an exploitative act being committed. To ensure that the participants paid close attention to the pictured exploiter, participants were first presented with the picture (either the formidable or the non-formidable version) and were asked to rate the man for personality traits. On the second slide, the participants were presented with the picture and the exploitation scenario (either the serious or the trivial exploitation scenario). On the third slide, subjects were presented with the picture and the items measuring moral outrage.

Formidability

To manipulate formidability, we rely on facial features. As testosterone is both the main determinant of physical strength (lean muscle mass) (Griggs et al., 1989; Lassek and Gaulin, 2009; Page et al., 2005) and of sexual dimorphism of the face (Penton-Voak and Chen, 2004; Pound, Penton-Voak, and Surridge, 2009; Weston, Friday, and Lio, 2007), it has been found that facial masculinity is an indicator of actual physical strength (Fink, Neave, and Seydel, 2007; Sell et al., 2009) as well as aggressive and exploitative behavior in men both in real life and in laboratory settings (Carré and McCormick, 2008; Mazur and Booth, 1998; Stirrat and Perrett, 2010). Studies also show that humans consistently associate facial masculinity with aggressive and exploitative behavior (Carré et al., 2009; Perrett et al., 1998), while facial immaturity (baby-faced: large eyes, high eye-brows, small chin) is associated with perceived submissiveness, physical weakness, incompetence, but also honesty and kindness (Keating, 1985; McArthur and Apatow, 1983; McArthur and Berry, 1987). Facial masculinity is related to traits such as a strong jaw, broad cheekbones, lowered eyebrows, and a large facial width-to-height ratio (Weston et al., 2007).

Specifically, we digitally constructed two versions of a head-shot photo of a young man; the two versions only differed in facial masculinity. The photo was selected from the Center for Vital Longevity Face Database (Minear and Park, 2004). Manipulation of facial masculinity was made in Facegen Modeller®, which is a computer program enabling manipulations of distinct and selectable facial characteristics using advanced vector calculations (Facegen Modeller version 3.4; see also Oosterhof and Todorov, 2008, for illustrative material). To increase realism, the two versions of masculinity were, respectively, morphed with the original photo using Magicmorph®.

In addition to the formidability-related trait of dominance (cf. Sell et al., 2009), recent studies in face perception suggest that another main trait spontaneously attributed to human faces is trustworthiness (Oosterhof and Todorov, 2008). To validate our experimental conditions on these two dimensions, we had 20 subjects rate the faces for dominance and trustworthiness. Subjects were presented with one of the faces and asked whether they agreed or disagreed that the terms “dominant” and “untrustworthy,” respectively, applied to the person. Answers were obtained on seven-point scales with “Completely Agree” scored as 1 and “Completely Disagree” scored as 7. As expected, the formidable face was rated as significantly more dominant than the non-formidable face (M_formidable = 2.27, M_{non-formidable} = 3.44, t(18) = 2.24, p = .038, two-tailed), while there was no significant difference in perceptions of untrustworthiness (M_formidable = 4.18, M_{non-formidable} = 3.79, t(18) = −.53, p = .60, two-tailed).

In the actual study, we added hair to the two faces to avoid skinhead-stereotyping (see Bodenhausen and Wyer, 1985). The resulting formidable (figure 1A) and non-formidable (figure 1B) versions of the subject were presented in JPEG format in a 350 × 510 resolution on a white background.

Figure 1.

Pictures used to manipulate the formidability of the exploiter

For additional validation and control, the subjects in the actual study rated the extent to which they perceived the man in the picture to be dominant, physically strong, physically attractive, friendly, trustworthy, extrovert, responsible, laid-back, and open-minded. All nine items were rated on a five-point scale from 1: not at all to 5: extremely. These ratings were performed when subjects were first presented with the picture prior to receiving any information about the pictured person's exploitative tendencies. Using factor analysis on the nine traits on which the faces were rated revealed a clear two-dimensional structure (see Table 1).

Table 1.

Factor analysis of traits ratings

Trait	Factor 1: Trustworthiness	Factor 2: Formidability
Dominant	−.18	.65
Physically strong	.30	.60
Friendly	.81	−.26
Trustworthy	.80	−.07
Open-minded	.74	−.12
Extrovert	.65	.04
Responsible	.65	.12
Physically attractive	.59	.20
Laid-back	.24	.01
Eigen-value	3.68	1.46

Notes: Principal Axis Factoring was used as extraction method and Oblimin was used as rotation method. Only the two displayed factors had eigen-values above 1.

As expected, the two formidability-related traits (dominant, physically strong) loaded on a separate factor, while the rest of the traits loaded much higher on a distinct single common factor. In line with previous research, trustworthiness and friendliness has the highest loadings on this other factor. The factor scores computed from the factor analysis for each subject was saved. Validating the facial manipulations, the formidable individual was judged as significantly higher than the non-formidable individual on the dominance dimension (M_formidable = .25, M_{non-formidable} = −.25, t(1619) = 12.76; p < .001). However, in contrast to the pilot results, the formidable individual was also judged as significantly less trustworthy (M_formidable = −.27, M_{non-formidable} = .28, t(1619) = −12.24; p < .001). Given some previous research (Carré et al., 2008; Stirrat and Perrett, 2010) this is not unexpected. This difference creates a potential confound as research shows that offender trustworthiness has a large influence on third party sanctioning (Baldwin and Kleinke, 1994; Feather and Atchison, 1998; Landy and Aronson, 1969; Weiten, 1980). As this difference between the faces is conceptually and empirically distinct from our main theoretical dimension of interest (i.e., formidability), we include the subjects' perception of the trustworthiness of the exploiter in the analyses as a control variable.

Exploitation scenarios

In order to evoke moral outrage in respondents, two fictional scenarios were created – each with an exploitative act being committed. Fictional scenarios have been used by others investigating context effects on moralistic punishment (Lieberman and Linke, 2007; O'Gorman, Wilson, and Miller, 2005), and fictional scenarios have been found to be quite adept at invoking the same emotional reactions as observations in real-life (Berthoz, Armony, Blair, and Dolan, 2002; Wilson and O'Gorman, 2003). The exploitative acts were constructed with some degree of ambiguity that enabled subjects to react either strongly towards the exploitation or to downplay it. The trivial exploitative act involved a scenario where the individual was cutting in line, while arguing that he was in a hurry. The serious exploitative act involved a scenario where the individual destroyed private property for no apparent reason. The wordings of the scenarios appear in the appendix.

Moral outrage

To measure the degree of third party moral outrage, eight items were constructed based on the theoretical framework from the moral-convention distinction task (see the appendix, table A1 and A2) (Kelly, Stich, Haley, Eng, and Fessler, 2007; Turiel, 1983). All items were measured on a five-point scale from 1: strongly disagree to 5: strongly agree. The items are essentially worded in the same way across the two scenarios with slight variations given the specifics of the different scenarios. To allow for comparisons despite these variations, z-scores for the moral outrage scales are computed independently for each scenario. Hence, a value of 0 on the scale corresponds to the average level of outrage in the given scenario and positive and negative numbers indicate numbers of standard deviations in positive or negative direction from this average. Two items were excluded from the final moral outrage scale, as these items correlated less than r = .20 with the other six items (final α = .83)

Results

Given that we hold distinct expectations for the reactions of males and females, the two sexes are analyzed separately. In the case of males, our theoretical expectations are clear and imply that third party outrage is modulated by the two-way interaction between exploitation severity and exploiter formidability. More specifically, we expect that more outrage is expressed against formidable exploiters committing serious exploitative acts while less outrage is expressed against formidable exploiters committing trivial exploitive acts. Our initial analysis reveals a significant main effect of exploiter formidability on expressed outrage (M_{non-formidable} = −.07, M_formidable = .08, t(809) = 2.06, p = .04, two-tailed). On average, across the two scenarios (i.e., the more and the less serious violations), more outrage is expressed against the formidable exploiter. To investigate our core claim, that this effect is modulated by the severity of the exploitive act, we test the effect of the two-way interaction between the formidability and severity manipulation. Contrary to our expectations, this interaction is insignificant (F_1,808 = 2.11, p = .147, two-tailed). To explore whether this result is confounded by the differences in trustworthiness observed between the face manipulations, we include the perceived trustworthiness of the exploiter as a covariate in the analysis. As shown in Table 2, this analysis reveals the existence of a significant three-way interaction effect between exploiter formidability, exploitation severity and the perceived trustworthiness of the exploiter (F_1,804 = 7.37, p = .007, two-tailed).

Table 2.

Male moral outrage towards exploiters as modulated by seriousness of exploitation and characteristics of the exploiter

	Coefficient
Intercept	.04 (.07)
Formidability of exploiter	−.04 (.11)
Seriousness of exploitation	−.14 (.10)
Perceived trustworthiness of exploiter	−.09 (.08)
Formidability × Seriousness	.23 (.15)
Formidability × Trustworthiness	−.24* (.11)
Seriousness × Trustworthiness	−.17 (.13)
Formidability × Seriousness × Trustworthiness	.47** (.17)
R 2	.05

Notes: N = 812. Entries are OLS regression coefficients with standard errors in parentheses. The dependent variable, moral outrage, and “Perceived trustworthiness of exploiter” are z-scored variables with high values corresponding to high levels of outrage and perceived trustworthiness. “Formidability of exploiter” and “Seriousness of exploitation” are dichotomous experimental variables in a 2*2 factorial design (0 = exploiter is non-formidable / exploitation is trivial, 1 = exploiter is formidable / exploitation is serious).

*

p < .05

**

p < .01.

Based on Table 2, Figure 2A and B provides a graphical display of the three-way interaction effect. The figure is a marginal effect plot with associated 95 percent confidence intervals (see Kam and Franzese, 2007 for statistical procedures). It shows how the marginal effect of exploiter formidability changes as a function of perceived trustworthiness. A positive effect – when statistically different from zero (i.e., zero is not included in the confidence interval) – implies that more outrage is expressed towards the formidable exploiter relative to the non-formidable exploiter, whereas a negative effect – when statistically different from zero – implies less outrage expressed towards the formidable exploiter. Figure 2A displays the marginal effect plot for the trivial exploitive act, while Figure 2B displays the marginal effect for the serious exploitive act.

Figure 2.

Marginal effect of exploiter formidability on male moral outrage in the face of trivial (A) and serious (B) exploitation as a function of the exploiter's perceived trustworthiness. Includes 95 percent confidence interval for the difference.

As can be seen, the predicted modulation of outrage as a function of exploiter formidability and exploitation severity emerges when the exploiter is initially deemed to be trustworthy. For the trivial exploitive act (Figure 2A), subjects begin to discriminate between the formidable and non-formidable exploiter as the perceived trustworthiness of the exploiter increases. Specifically, males become significantly less outraged at formidable exploiters relative to non-formidable exploiters. For the serious exploitive act, the interaction effect is as expected in the opposite direction. As shown in Figure 2B, when exploitation is serious and the exploiter is perceived as trustworthy, males become more outraged when the exploiter is formidable. These results show that while males experience more outrage against formidable individuals when they commit serious acts of exploitation, trivial exploitive acts from formidable individuals activate less outrage than trivial acts from non-formidable others, at least when these individuals are perceived as initially trustworthy. Under this circumstance, the theoretical predictions are supported for males.

The analyses also reveal how exploiter formidability can shape third party outrage among males in another way. In general, there is a negative effect of the perceived trustworthiness on outrage, such that more outrage is expressed towards those who were initially deemed to be untrustworthy (zero-order effect: r = −.17, p < .0001, two-tailed). Given the effect of the formidability manipulation on perceived trustworthiness (cf. the Materials and Methods section), trustworthiness provides an indirect path through which formidability, ceteris paribus, increases the outrage expressed towards an exploiter. In line with this, the initially reported main effect of the formidability manipulation turns insignificant (F_1,808 = .79, p = .37, two-tailed) when controlled for perceived trustworthiness. As the relationships reported in Figure 2A and B do include perceived trustworthiness, any such indirect effects exist in addition to the predicted modulation of outrage as a function of both exploiter formidability and exploitation seriousness.

Given the male-bias for coalitional aggression, we do not expect females to express more outrage (i.e., display a mobilization effect) against formidable exploiters. However, it is less clear whether they, on a priori grounds, will treat formidable exploiters more leniently across scenarios. To explore the different possibilities, we first try to replicate the patterns of effects among males. As expected, both the two-way interaction – between exploiter formidability and exploitation seriousness (F_1,805 = 1.32, p = .25, two-tailed) – and the three-way interaction – between exploiter formidability, exploitation seriousness, and perceived trustworthiness of the exploiter (F_1,801 = .25, p = .61, two-tailed) – are insignificant. For females, the effect of exploiter formidability does not, in other words, depend on the seriousness of the exploitive act. To test whether females treat all formidable exploiters more leniently, independent of exploitation seriousness, we next investigate the existence of a main effect of exploiter formidability on female outrage. Again, we find no such effect (M_{non-formidable} = −.06, M_formidable = .04, t(808) = 1.39, p = .16, two-tailed).

In tandem, these results show that third party outrage among females is not modulated by differences in formidability between male exploiters. One simple explanation for this result could be that females are less attentive to male differences in formidability. However, previous research has found no such sex differences (Sell et al., 2009), and both theory and data suggest that male differences in physical strength is an important parameter in mate selection processes for females (Buss and Schmitt, 1993; Dixson, Halliwell, East, Wignarajah, and Anderson, 2003; Frederick and Haselton, 2007; Little et al., 2010; Maisey, Vale, Cornelissen, and Tovée, 1999). At the outset, it is therefore unlikely that females do not pay attention to differences in formidability. In line with this, our data reveals no sex differences in the effect of our experimental manipulation of exploiter formidability on the ratings of perceived formidability. Specifically, we find an insignificant two-way interaction between the sex of the subject and the formidability manipulation on the subjects' perception of the formidability of the exploiter (F_1,1617 = .22, p = .64, two-tailed). Hence, it is not that females do not attend to male differences in formidability, they just don't modulate their outrage to it. As argued in the theory section, this could reflect the fact that (1) males on average are 90 percent stronger than females, which would imply that even a non-formidable male would be stronger than many females, or (2) females do not reason about third party situations in terms of direct physical confrontations where differences in formidability would be important. The first explanation entails that the lack of effect of male differences in formidability for female outrage reflects that females defer to all male exploiters. This explanation thereby requires that females in general express less outrage than males towards male exploiters. Importantly, this is not the case. There are no sex differences in the outrage individuals express towards the exploiters (M_males = .01, M_females = −.02, t(1620) = .49, p = .63, two-tailed). This leaves us with the second explanation: viz., females do not reason about third party involvement in terms of physical confrontation, presumably because such confrontation has been a male domain over evolutionary history.

Discussion

This study adds to an increasing number of studies demonstrating that third party moral responses to social violations are modulated by the characteristics of the exploiter. Consistent with these studies, the above findings demonstrate that the trustworthiness of the exploiter is a major component in this regard. Individuals, whose personalities are gauged to be geared towards continued exploitation, elicit large levels of outrage. Considering the fact that moral outrage serve as output of systems, in part, designed to deter individuals from transgressing again (Petersen et al., 2010; Price et al., 2002; Sunstein, 2005), the evolutionary importance of exploiter trustworthiness is obvious.

At the same time, the present study goes significantly beyond previous studies. By using very subtle face-manipulations in a nationally representative sample, we have provided evidence that exploiters' formidability is an additional determinant of moral outrage in third parties. When exploiters were perceived as trustworthy, we found that males are less outraged when confronted with formidable offenders having committed a trivial rule violation which is in line with traditional models of animal behavior in conflict situations. Yet, supporting alternative theories about coalitional control of powerful individuals, situations involving more serious transgressions activated heightened outrage among males against formidable exploiters. In other words, if a formidable individual keeps exploitation within limits, he is excused. But if a formidable male reveals massive exploitive tendencies, other males mobilize against him. We also replicated a finding in previous studies that perceptions of trustworthiness are negatively influenced by formidability (Carré et al., 2009; Stirrat and Perrett, 2010). For males, trustworthiness thus serves as an additional channel through which the formidability of exploiters influences moral outrage.

In the case of females, results showed that females did not modulate their expressed outrage in response to exploiters' formidability. The relevance of exploiter formidability for third parties arises, in particular, when engaging in direct physical punishment of the exploiter. Given this, the lack of effect among females is expected and, as indicated in previous research, direct aggression has been a male domain over evolutionary history (e.g., Chagnon, 1983; Van Vugt, De Cremer, and Janssen, 2007; Wrangham and Peterson, 1996; see also Tooby and Cosmides, 2010). As a consequence, we suggest, females do not reason about third party sanctioning in terms of physical confrontation with the exploiter. While the results provide no direct evidence of this, analyses indicate that the lack of formidability-effect on female outrage cannot be explained by alternative interpretations, such as the possibility that females are inattentive to male differences in formidability, or that females defer (i.e., express low levels of outrage) to all male exploiters.

Returning to our main hypothesis, the predicted effects of exploiter formidability on male third party outrage only appeared when subjects initially perceived the exploiters as trustworthy. In terms of selection pressures, the effect of exploiter formidability on moral outrage was predicted based on the fact that formidability would influence the costs associated with future exploitation. In the case of massively untrustworthy individuals such as psychopaths, differences in formidability might be less predictive of future costs as even non-formidable individuals can inflict massive costs if their intentions are malicious enough. The present results could reflect this. In that perspective, a counter-exploitation psychology geared towards estimating the costs of future exploitation would be designed to disregard other person attributes when confronted with highly untrustworthy individuals.

These considerations also point to directions for future research. First of all, it would be important to replicate the results using manipulations that independently manipulated the trustworthiness and formidability of exploiters. In this way, it would be possible to decide whether the present effects of trustworthiness ratings reflect the operations of mechanisms designed to react differently to trustworthy or untrustworthy exploiters or, whether they result from some individual differences among high and low trusters (e.g., Yamagishi, 2003). Second, it would be important to replicate the results using scenarios that more equally match antecedent circumstances. For example, in the present case, one scenario reflects a criminal offense, the other does not. Correspondingly, it would be interesting to investigate whether the results might be generalized to other moral domains (i.e., violations in the domains of purity/sanctity, ingroup/loyalty, or authority/respect) (Haidt, 2007). Finally, in future studies, it would be important to provide additional and independent evidence that the function of moral outrage in the face of a serious formidable exploiter is to coordinate joint coalitional attention. Are people, for example, more inclined to convey information about such exploiters to others? Are they easier to remember? Do women also express outrage towards formidable offenders when given coalitional cues? Affirmative answers to such questions could provide relevant evidence of coalitional mobilization.

All in all, even though moral outrage is designed to fuel moralistic punishment against all social exploiters, this study provides additional support for human outrage being a fine-tuned strategic resource deployed according to subtle cues. Given the involved costs and benefits, male moral outrage is withheld against formidable individuals if they do not excessively exploit their bargaining power. Yet, if that line is crossed, the formidable individual faces severe outrage among male third parties. As the existence of this counter-exploitation circuitry would in itself work as a selective force, we might expect dominant individuals to intuitively recognize this set of dynamics.