Within moral psychology, there has been a shift from a one-sided focus on moral reasoning and the cognitive-developmental perspective of Kohlberg (1984) to a broader view that includes moral personality and moral emotions (Haidt, 2001; Peterson & Seligman, 2004; Walker & Pitts, 1998). One research path for the study of moral personality is to pursue naturalistic conceptions of morality. For example, what do lay people say when asked what attributes a moral prototype or exemplar has, or what the attributes of a good character are (Lapsley & Lasky, 2001; Walker & Pitts, 1998)? An interesting advancement within this area was made by Walker and Hennig (2004). Based on layperson beliefs about morality, these authors distinguish three moral exemplars; namely a caring, a just, and a brave prototype. They demonstrated that one should not speak about “the” moral person, but one should better think of the moral personalities in the plural. Walker and Hennig (2004) argue for an inclusion of three different moral personalities into moral psychology in order to provide a fuller and more balanced view of moral excellence. They see personality as “the flesh vivifying the bare bones of cognition.” (p. 643). So far it is unclear, however, whether these moral prototypes translate into specific types of moral behavior. Nevertheless, it is important to investigate whether and how moral prototypes are connected to moral behavior. Moral reasoning ability alone appears to be a weak and inconsistent predictor of moral behavior, once intelligence has been partialled out (Haidt, 2003). Possibly, moral prototypes as ideal moral personalities have closer relations to moral behaviors. Expanding on Walker and Hennig's (2004) work, we examined the hypothesis that moral prototypes incorporate specific moral behaviors. We then tested whether this incorporated behavioral information could be used to foster moral behavior.
The present studies investigated the extent to which three basic moral prototypes, “just,” “brave”, and “caring”, are related to moral, prosocial behavior. In five studies, we tested (a) whether people would associate three basic types of moral behavior (helping behavior, moral courage, and heroism) with three moral prototypes, and (b) whether specific emotional precursors of moral behavior and moral behavior itself could be promoted by activating the respective moral prototype. As expected, Studies 1–3 revealed that people associated helping behavior with the caring prototype, moral courage with the just prototype, and heroism with the brave prototype. Studies 4 and 5 showed that the activation of the three prototypes differentially influenced emotional precursors of the three types of moral behavior (Study 4) as well as actual moral behavior (Study 5). Thus, the five studies revealed that people associate different moral behaviors with different moral prototypes and that a certain moral behavior can be activated by the priming of the related prototype. Copyright © 2010 John Wiley & Sons, Ltd.
The Caring, the Just, and the Brave Prototype
Walker and Hennig (2004) argue that there might be more than just one single prototype for moral maturity, as it is impossible for any person to embody all of the attributes that lay people list for a moral prototype. Looking closer at lists of attributes held to characterize a highly moral person (e.g., Lapsley & Lasky, 2001) leads to the observation that they include an amalgamation of virtues that are sometimes even incoherent (see also Flanagan, 1991). As Walker and Henning (2004) put it: “Not all virtues are necessarily compatible and […] excellence in one area may come at the expense of development in some other area.” (p. 630). They tested their assumption of three separable moral prototypes in three large-scale studies. It was shown that the caring, the just, and the brave prototype were each characterized by relatively distinct moral personalities that consist of typical descriptors, specific associations with a standard personality inventory, and distinct implicit personality structures. Example attributes for the caring prototype descriptors are: Sympathetic, concerned, good-hearted, helpful; example attributes for the just prototype descriptors are: Fair, moral, truthful, honest; examples for the brave prototype descriptors are: Courageous, heroic, intrepid (for the complete descriptor lists, see Walker & Hennig, 2004, pp. 646–647). Walker and Hennig (2004) provide a useful taxonomic description of three different moral exemplars. This taxonomic description, however, only consists of attributes, and no explicit relations with specific behaviors are assumed. Several authors have argued that prototypes do not only consist of attributes, but that actions are also encoded within them (Carver & Scheier, 1981; Rosch & Mervis, 1975). Furthermore, a variety of studies in the area of health psychology has demonstrated that prototypes play an important role for behavioral intentions, actual behavior, or behavioral changes (e.g., Gibbons, Gerrard, Lando & McGovern, 1991; Oullette, Hessling, Gibbons, Reis-Bergan, & Gerard, 2005; see also the prototype/willingness model (PWM) e.g., Gibbons, Gerrard, Blanton, & Russel, 1998). Therefore, it is reasonable to expect that prototypes also play a role in moral behavior.
Different Types of Moral/Prosocial Behavior
The different types of moral/prosocial behavior1 we assume to be specifically associated with and affected by the moral prototypes are helping behavior, moral courage, and heroism. It only makes sense to suppose specific prototype-behavior associations if not only the prototypes (as demonstrated by Walker & Hening, 2004), but also the related behaviors, are clearly separable. On a theoretical basis, these three moral/prosocial behaviors can be separated along a cost dimension, namely the cost for the helper.
The anticipation of low versus high cost distinguishes helping behavior from heroism and moral courage, respectively. The type of cost (physical vs. social) divides heroism and moral courage (Becker & Eagly, 2004; Greitemeyer, Osswald, Fischer & Frey, 2007). Helping behavior is a low cost behavior, which means that you can do good deeds without suffering any (greater) harm. Heroism, in contrast, is associated with physical costs, because a person takes risks “on behalf of one or more other people, despite the possibility of dying or suffering serious physical consequences” (Becker & Eagly, 2004, p. 164). Finally, moral courage includes social costs: Greitemeyer, Fischer, Kastenmueller, and Frey (2006) define moral courage as a prosocial behavior accompanied by moral outrage, which intends to enforce societal and ethical norms without weighing one's own social costs. Intervention when a person is treated unfairly or unjustly by a perpetrator (where the intervener acts prosocially and might, therefore, suffer negative social consequences, i.e., social costs, as a reaction from the perpetrator) is an example of moral courage. To further empirically test the separation of the three behaviors, an extensive pretest was conducted (see below).
The first hints already exist which suggest an association of the moral prototypes with specific behaviors (see Colby & Damon, 1992; Oliner & Oliner, 1988). In a recent study, Walker and Firmer (2007) examined real brave or caring exemplars. The researchers had chosen their participants based on background information regarding whether the participants had done something heroic (like saving another person's life under considerable danger) and had therefore received the Medal of Bravery, or whether they had provided care, help, and support for individuals or groups and had therefore received the Caring Canadian Award. It seems to be implicitly assumed that different moral prototypes are closely related to different behaviors. So far, however, whether the moral prototypes specifically correspond to different moral behaviors had not been explicitly formulated or empirically tested. Thus, we supposed that the caring prototype was related to helping behavior, the just prototype to moral courage, and the brave prototype to heroism. The present research tested this assumption.
Overview of the Present Research
In the present research, we pursued two goals. First, we examined whether the three moral prototypes of Walker and Hennig (2004) correspond differently to three kinds of moral, prosocial behavior. Concretely, we expected the caring prototype primarily to be connected with helping behavior, the just prototype with moral courage, and the brave prototype with heroism. Second, we tested whether these correspondences can be used to promote specific determinants of moral behavior and moral behavior itself. That is, the activation of the prototypes should produce specific effects on emotional precursors of moral behavior and moral behavior itself.
To pursue these two goals, five studies were conducted. In Studies 1–3, we tested whether the moral prototypes were associated with specific moral behaviors. We examined these associations on the level of attributes (Study 1), with typicality ratings (Study 2), and with behavioral expectancies (Study 3). Studies 4 and 5 addressed the second goal of whether prototype-activation specifically affects precursors of moral behavior as well as actual behavior. In Study 4, all three moral prototypes were activated and their influence on different emotional precursors of moral behavior was tested. Finally, in Study 5 we examined whether the activation of one of the prototypes (the just) would specifically influence behavioral intentions and behavior, namely moral courage, and not helping behavior. Before presenting our studies, a pretest is described which determined whether helping behavior, moral courage, and heroism can be separated empirically.
Pretest: Helping behavior, moral courage, and heroism as three types of moral, prosocial behavior
Via a pretest with 135 students (112 women, M = 23.60, SD = 5.93), we tested whether helping behavior, moral courage, and heroism are understood as three different forms of moral behavior. The participants were given 30 scenarios, ten describing prosocial behavior with low danger and low social costs (helping behavior, e.g., “On a country lane a person is standing beside her defective car. Another person stops his car, looks at the engine of the defective car, and offers to tow her to the next garage.”), ten describing prosocial behavior with moderate/high danger and high social costs (moral courage, e.g., “In a company, almost every day, one of the supervisors shouts at an elderly employee who is not able to answer back. A colleague of the elderly employee notices the situation and addresses the supervisor.”), and ten describing prosocial behavior with high danger and low social costs (heroism, e.g., “A woman is ice-skating on a frozen lake. She overlooks a spot where the ice is thinner and breaks through it. A man realizes the emergency, moves cautiously over the brittle ice, and saves the woman.”). The scenarios had to be evaluated on 7-point scales (1 = not at all; 7 = absolutely) concerning morality, danger, and social costs. Furthermore, participants had to indicate (on a 7-point scale) what kind of behavior (helping behavior, moral courage, or heroism) was presented in the scenarios.
Analysis of variance (ANOVAs) with type of scenario as the within-subject factor revealed that the scenarios were evaluated differently by the participants (for means and standard deviations see Table 1). In fact, moral courage scenarios were evaluated as most moral (M = 6.00), followed by heroism scenarios (M = 5.89) and helping scenarios (M = 5.48), F(2, 268) = 7.77, p < .01, η2 = 0.06. Looking at danger, significant differences between the evaluations of the scenarios types also emerged, F(2, 268) = 386.17, p < .001, η2 = 0.74. Here, heroism scenarios were seen as most dangerous (M = 5.43), followed by moral courage scenarios (M = 5.07), and helping scenarios (M = 2.21). Concerning social costs, results revealed that moral courage scenarios were seen as connected to most social costs (M = 4.53), followed by heroism scenarios (M = 2.66) and helping scenarios (M = 1.96), F(2, 268) = 152.20, p < .001, η2 = 0.53. Furthermore, the behaviors described in the scenarios were recognized as what they were meant to be: Looking at the evaluation as moral courage, moral courage scenarios had the highest mean (M = 6.00), followed by heroism scenarios (M = 5.06) and helping scenarios (M = 3.25), F(2, 268) = 163.36, p < .001, η2 = 0.55. Concerning the evaluation as heroism, heroism scenarios showed the highest mean (M = 5.95), followed by moral courage scenarios (M = 4.87) and helping scenarios (M = 2.60), F(2, 268) = 122.89, p < .001, η2 = 0.48. Looking at evaluation as helping behavior, however, helping and heroism scenarios were likewise evaluated as helping behavior (heroism scenarios M = 6.47, helping scenarios M = 6.41), followed by moral courage scenarios (M = 5.40), F(2, 268) = 70.51, p < .001, η2 = 0.35. This result seems quite plausible, though, because people indeed help when they save another person from drowning or from a similar predicament. In sum, the scenarios were perceived as intended, and most importantly, they could be separated by the dimensions of danger and social costs.
|Helping scenarios||Moral courage scenarios||Heroism scenarios|
|Morality dimension||5.48 (1.87)||6.00 (0.70)||5.89 (1.19)|
|Danger dimension||2.21 (1.04)||5.07 (0.91)||5.43 (1.37)|
|Social costs||1.96 (1.00)||4.53 (0.95)||2.66 (1.89)|
|Evaluation as helping behavior||6.41 (0.71)||5.40 (1.21)||6.47 (0.91)|
|Evaluation as moral courage||3.25 (1.62)||6,00 (0.71)||5.06 (1.74)|
|Evaluation as heroism||2.60 (1.39)||4.87 (1.48)||5.95 (2.61)|
STUDY 1: FREE LISTING
The goal of the first study was to compare the descriptors of the three moral prototypes by Walker and Hennig (2004) to a list of attributes that people freely generated as associated with helping behavior, moral courage, and heroism (to simplify, we will call them behavioral attributes, or descriptors). Research shows that people draw spontaneous trait inferences from behavior (e.g., when you smile you are supposed to be friendly; for a review see Uleman, Newman, & Moskowitz, 1996). This phenomenon was used to generate behavioral descriptors, which were then compared (by counting the concordances) to the descriptors of the three moral prototypes found by Walker and Hennig (2004). We expected that the descriptors of the caring prototype would show the most correspondences with the behavioral descriptors of helping behavior, the descriptors of the just prototype with the behavioral descriptors of moral courage, and the descriptors of the brave prototype with the behavioral descriptors of heroism.
Two hundred twenty individuals (159 women) from 16–60 years of age (M = 26.7; SD = 9.55) took part. Most of them were students (73%) recruited from an undergraduate psychology course. The remaining participants were people of different professions who were asked on the street if they were willing to take part in a study. All participants were Germans.
Procedure and Materials
Participants first completed demographic questions. They were then instructed to write down in a free-response format, a list of attributes, traits, and characteristics (formulated as adjectives) that they associate with a person who shows helping behavior (or moral courage or heroism). We also provided a short definition of the respective behavior to make sure that all participants had roughly the same idea of the relevant behavior. They were given 10 minutes to list the attributes. Participants were asked to provide characteristics for only one type of behavior that was randomly assigned. After listing the attributes, participants were thanked and debriefed.
Preparation of the Data and Comparison Material
The total number of attributes yielded by the free-listing procedure was 491 for helping behavior (73 participants), 496 for moral courage (73 participants), and 510 for heroism (74 participants). The following judgment rules were applied (Walker & Hennig, 2004; Walker & Pitts, 1998): (a) Comparisons and modifiers were dropped (e.g., extremely courageous became courageous), (b) attributes and paraphrases that were judged to be closely synonymous in meaning were reformulated and collapsed (e.g., having a lot of charity and pity for other human beings was reformulated as compassionate), and (c) in order to avoid a change in meaning, information that was paraphrased and not made up of attributes or adjectives was dropped (e.g., does not look away when other persons are in danger).
The descriptor lists were distilled in collaboration by two judges (the first author and a psychologist who was an assistant professor and an expert for qualitative research, but not aware of the hypotheses), who discussed and resolved any differences. The results were three behavioral descriptor lists (one for each behavior) that only consisted of adjectives with frequencies showing how often each attribute was named by the participants. The comparison material was the descriptor list of the three moral prototypes by Walker and Hennig (2004, pp. 646–647). The problem that within the three prototypes some descriptors belong to all three prototypes (with different typicality-ratings, of course) had to be resolved. We decided, therefore, to take the most typical descriptors up to the point at which a double mention in another prototype occurred for the first time. Applying this rule, for the caring prototype 38 descriptors were selected, for the just prototype 37 descriptors were selected, and for the brave 47 descriptors were selected. This conservative strategy guaranteed that only unique and highly prototypical descriptors were entered into the comparison, and double-counts were avoided. The selected descriptors were translated into German by the first author and proofed by a bilingual person.
Results and Discussion
The comparison was a simple counting process: We took the selected prototype descriptors and counted how often each of these descriptors was mentioned by our participants as a behavioral descriptor. Only completely synonymous attributes were considered. Only for two descriptors, differences between the raters emerged which could be traced back to counting mistakes (note that more than 1000 descriptors were counted). Having solved the counting mistakes, the raters provided 100% consistent codes. We first calculated the overall concordances between the prototype and the behavioral descriptors (results are displayed in Table 2). Of the 1497 listed behavioral attributes, 1114 fit with the descriptors of one of the prototypes. This resulted in a total concordance of 74%. To have a closer look at the concordance between the prototypes and the behaviors, a 3 (prototypes) × 3 (behavioral attributes) χ2 test was calculated. Results revealed that the behavioral attributes corresponded differently to the three prototypes, χ2(4, N = 220) = 169.03, p < .001, w = 0.88. Looking at the 467 behavioral attributes that matched the descriptors of the caring prototype, 51% (236) were helping descriptors, 27% (126) were moral courage descriptors, and 22% (105) were heroism descriptors. Of the 130 behavioral attributes that fit the just prototype, 27% (35) were helping descriptors, 52% (68) were moral courage descriptors, and 21% (27) were heroism descriptors. Looking at the 517 behavioral descriptors that matched the brave prototype, 14% (72) were helping descriptors, 44% (229) were moral courage descriptors, and 42% (216) were heroism descriptors.
|Helping descriptors||Moral courage descriptors||Heroism descriptors||Total|
Study 1 offers the first promising support for our hypotheses: The most frequent concordances emerged between the descriptors of the caring prototype and helping behavior, between the just prototype and moral courage, and between the brave prototype and heroism and moral courage. This last finding makes sense insofar as the term moral courage already states that one is not only moral, but also courageous by demonstrating that behavior. On the level of attributes, the moral prototypes and the examined types of moral behaviors share, for the most part, the same pool of descriptors. In Study 2, we aimed to replicate the results with another method.
STUDY 2: TYPICALITY RATINGS OF THE PROTOTYPES' DESCRIPTORS CONCERNING MORAL BEHAVIORS
In Study 2, we remained on the descriptor-level of the moral prototypes and tested our assumptions the other way around: Participants had to rate the typicality of descriptors of the three moral prototypes (Walker & Hennig, 2004) for the three moral behaviors. The procedure we used was adopted from a study by Niedenthal, Cantor, and Kihlstrom (1985). In the present research, the descriptors of the moral prototypes were used as a starting point, and were rated for their typicality when applied to the three moral behaviors. We expected a significant interaction between the descriptors for the three moral prototypes and how typical they were rated for the three different moral behaviors.
The sample consisted of 251 persons (134 women) ranging in age from 15 to 74 years (M = 34.10, SD = 12.13). The study was an online survey; participants (students who had given their e-mail addresses at the university homepage in order to be informed about interesting studies) were personally contacted by e-mail and invited to take part. Included in the mailed invitation was a request to forward the mailing to friends and colleagues and to invite them to participate as well. As an incentive the participants could win small prizes.
Procedure and Materials
To avoid overwhelming the participants with too many descriptors, we used only the German translations (see Study 1) of the first 15 most typical descriptors of the moral prototypes by Walker and Hennig (2004). Each of these 45 descriptors (randomly assigned) had to be rated on an 8-point scale (0 = not typical at all, 7 = absolutely typical) for how typical it was in representing helping behavior, moral courage, and heroism (within-subjects design). The following instructions were given:
“Look at these adjectives and decide for each adjective how typical or atypical it is for helping behavior (or moral courage or heroism). Consider thereby: ‘Is a person who shows helping behavior (or moral courage or heroism) characterized by this attribute?”’
To minimize comparison effects, the participants worked through all of the 45 adjectives three times and gave their evaluations first with regard to helping behavior, then moral courage, and then heroism (the order was counter-balanced). The online study was designed so that all items had to be treated. It took about 10–15 minutes to complete the questionnaire. In the end, participants were thanked and debriefed.
Results and Discussion
We calculated the means of the 15 prototype descriptors for the caring, the just, and the brave types within each behavior. Previous reliability analyses had revealed Cronbach's α as between .88 and .96. With these nine means we conducted a 3 (moral prototype: Caring vs. just vs. brave) × 3 (type of moral behavior: Helping behavior vs. moral courage vs. heroism) factorial ANOVA with repeated measures. Sex and age of the participants as well as order did not produce any significant main effects or interactions for any of the dependent variables. Thus, these variables were not considered further.
Results revealed significant main effects for moral prototype, F(2, 500) = 326.30, p < .001, η2 = 0.57, and for type of moral behavior, F(2, 500) = 15.82, p < .001, η2 = 0.06, which were qualified by the predicted significant interaction, F(4, 1000) = 375.12, p < .001, η2 = 0.60. To illustrate the interaction, three one-way (type of moral behavior) within-subjects ANOVAs for each prototype were conducted that all produced significant effects; caring: F(2, 500) = 277.59, p < .001; η2 = 0.53, just: F(2, 500) = 40.16, p < .001, η2 = 0.14, and brave: F(2, 500) = 276.49, p < .001, η2 = 0.53 (the results are displayed in Figure 1). Follow-up analysis revealed that the descriptors of the caring prototype had the highest typicality ratings for helping behavior (M = 5.81), compared to moral courage (M = 4.83), F(1, 250) = 184.21, p < .001, η2 = 0.42, and compared to heroism (M = 3.92), F(1, 250) = 431.19, p < .001, η2 = 0.63. For the just prototype, follow-up analyses showed that the just descriptors had the highest typicality ratings for moral courage (M = 5.33) compared to helping behavior (M = 4.89), F(1, 250) = 57.09, p < .001, η2 = 0.19, and compared to heroism (M = 4.71), F(1, 250) = 69.95, p < .001, η2 = 0.22. Follow-up analysis for the brave prototype revealed that the brave descriptors had the highest typicality ratings for heroism (M = 6.89), compared to helping behavior (M = 5.14), F(1, 250) = 418.7, p < .001, η2 = 0.63, and compared to moral courage (M = 6.22), F(1, 250) = 121.63, p < .001, η2 = 0.33.
The results of Study 2 further support our hypotheses: The descriptors of the caring type were seen as most typical for helping behavior, the descriptors of the just type for moral courage, and the descriptors of the brave type for heroism. The brave descriptors again played an important role in moral courage, but this time they were more clearly related to heroism, as the simple contrasts between heroism and moral courage were significant.
STUDY 3: ATTRIBUTION OF THE PROTOTYPES TO DIFFERENT MORAL BEHAVIORS
Research about prototypes has demonstrated that people have clear ideas about what a typical person would do in a certain situation (Cantor, Mischel & Schwartz, 1982). In Study 3, we did not utilize behavioral attributes like we did in the previous two studies, but rather, we used situations that described the respective behaviors. We wanted to test whether people expect different moral behaviors from the moral prototypes. Our hypothesis was participants would expect of the caring prototype to most probably show helping behavior, of the just prototype to most probably show moral courage, and of the brave prototype to most probably show heroism.
The sample consisted of 21 female students which were personally approached by the experimenter on the university campus. The age of the participants ranged from 20–48 years (M = 24.76, SD = 5.90). As compensation, sweets were given.
Procedure and Materials
The participants were given a short questionnaire that first asked for demographic information. Then three hypothetical persons were described, each identified with the ten most prototypical descriptors of one of the moral prototypes by Walker and Hennig (2004). We thus constructed an ideally caring, just, or brave person. Following the descriptions of the three hypothetical persons, participants read nine short scenarios that described persons who showed helping behavior, moral courage, or heroism (three scenarios for each behavior). The scenarios were taken from the pretest mentioned above and we therefore ascertained that they clearly expressed one of the three moral behaviors. For each scenario, participants had to indicate which of the three hypothetical persons would most probably show the described behavior, and then which would be second and third.
Results and Discussion
The first step was to organize the raw data; the hypothetical persons were allocated ranks within the scenarios according to the statements of the participants. In other words, the hypothetical person who was seen as most likely to show the behavior described in a scenario had to be given rank 1, the second most likely rank 2, and so on. We then calculated the mean ranks of each moral prototype for the scenarios of helping behavior, moral courage, and heroism, respectively. With these nine mean ranks a Friedman test was conducted, which revealed a significant effect, χ2(2, N = 21) = 111.35, p < .001. Follow-up Friedman tests for each prototype were conducted separately to determine whether the prototypes were allocated differently to the behaviors. They all yielded significant differences concerning the ranking: Caring: χ2(2, N = 21) = 25.58, p < .001, just: χ2(2, N = 21) = 31.90, p < .001, and brave: χ2(2, N = 21) = 38.70, p < .001 (mean ranks are presented in Figure 2). Pairwise comparisons using the Wilcoxon test revealed the following results: For the caring type, helping behavior had the lowest mean rank (Mrank = 1.14), compared to moral courage (Mrank = 2.62) Z = −3.84, p < .001, and to heroism (Mrank = 2.24), Z = −3.97, p < .001. (Remember: The hypothetical person who was seen as most likely to show the behavior described in a scenario was given rank 1; this means the lower the rank, the more probable the participants considered the hypothetical person to engage in the described behavior). For the just prototype, moral courage had the lowest mean rank (Mrank = 1.19), compared to helping behavior (Mrank = 1.90), Z = −2.82, p < .01, and compared to heroism (Mrank = 2.90), Z = −3.85, p < .001. For the brave prototype, heroism had the lowest mean rank (Mrank = 1.00), compared to helping behavior (Mrank = 2.86), Z = −4.08, p < .001, and compared to moral courage (Mrank = 2.14), Z = −4.04, p < .001.
The results of Study 3 confirmed our assumption that people expect different behaviors from the moral prototypes. When the moral prototypes of Walker and Hennig (2004) are presented as person-like concepts, people give very clear statements regarding which of the ideal persons would most probably show a particular behavior: As expected participants indicated that caring prototype would most probably show helping behavior, that the just prototype would most probably show moral courage and that the brave prototype would most probably show heroism.
This gives support to the assumption that the moral prototypes by Walker and Hennig (2004) not only consist of trait-like descriptors, but also include distinct behaviors. When people think of a moral prototype, it is not only the attributes of this prototype that come into their minds, but also the behavior that is most probable and suitable for the relevant prototype.
So far, we have handled the first aim of the present investigation by showing that the moral prototypes suggested by Walker and Hennig (2004) include substantial behavioral information. But more importantly, the three moral prototypes do not all contain the same behavioral information; rather, each moral prototype seems to possess a range of moral reactions that are specifically associated with the prototype. Therefore, the activation of a moral prototype should influence precursors of behavior and behavior itself. More importantly, due to the specific prototype-behavior relations, a moral prototype should primarily promote one type of moral behavior. In the fourth and the fifth studies, we examined whether the activation of a moral prototype influences emotional precursors of moral behavior and behavior.
STUDY 4: ACTIVATION OF THE THREE PROTOTYPES—EFFECTS ON EMOTIONAL PRECURSORS OF MORAL BEHAVIOR
The goal of Study 4 was to examine whether the activation of the three prototypes differently affects precursors of moral behavior. To activate the prototypes, we adopted a procedure by LeBoeuf and Estes (2004): Participants had to compare themselves with one of the prototypes and list similarities. According to Mussweiler (2003), this procedure should lead to assimilation because we suggested similarity testing by our instruction.
Since real behavior is often difficult to measure, we wanted to try the effects of the moral prototypes on precursors of behavior first. We decided in favor of emotional precursors since emotions generally play an important role for behavior (e.g., Fredrickson, 2002; Lazarus, 1991). As an emotional precursor of helping behavior we examined empathy. Numerous studies have demonstrated that empathy is associated with moral, prosocial behavior (e.g., Greitemeyer, 2009). Empathic feelings are a potent source of motivation to help (for meta-analysis and reviews, see Batson, 1998; Eisenberg & Miller, 1987).
Justice sensitivity (JS) was examined as an emotional precursor of moral courage. The concept of JS has been introduced by Schmitt, Gollwitzer, Maes, and Arbach (2005) and is supposed to explain inter-individual differences in emotional reactions (i.e., moral outrage) to unfair situations (which are typical situational preconditions for moral courage behavior, Greitemeyer et al., 2007). JS can be viewed from different perspectives: The victim perspective (you are yourself the victim of an injustice), the observer perspective (you observe the unjust treatment of another person), or the beneficiary perspective (you receive benefits from the unjust treatment of another person). For the purpose of the present article, only JS from the observer perspective (JS observer) is relevant: This perspective was shown to predict the intention to act in a morally courageous fashion (Hauer, Kretteck, Baumer, & Schmitt, 2007).
As an emotional precursor of heroism, we tested fear in dangerous situations. Overcoming of fear has been shown to be positively associated with different forms of moral, prosocial behavior (Woodard & Pury, 2007). According to Becker and Eagly (2004), fearlessness or at least control of fear, acceptance of risk and the courage to take this risk transform moral, prosocial acts into heroism. Even if experimental research regarding a causal relation between control of fear, courage, and moral behavior (like heroism) still stands out, it can be assumed that fearlessness, overcoming of fear, or courage might play an important role in heroic acts.
To further examine whether empathy, JS, and fear(lessness) were primarily related to the assumed behaviors, a pretest was conducted. Twenty participants (12 women) were asked for which behavior (helping behavior, moral courage, or heroism) certain emotional precursors (empathy, JS, and fearlessness) would be most likely, second most likely, or least likely to result. For the most probable behavior, participants had to give four points; for the second, two points; and zero for the least probable. A 3 (emotional precursor) × 3 (type of behavior) within-subjects ANOVA with repeated measures revealed a significant interaction between type of emotional precursor and type of behavior, F(4, 76) = 67.97, p < .001, η2 = 0.78. Participants thought empathy would result most probably in helping behavior, M = 3.80 (moral courage M = 2.10, heroism M = 0.10), JS would most likely lead to moral courage, M = 3.80 (helping behavior M = 1.60, heroism M = 0.60), and fearlessness would most likely result in heroism, M = 3.30 (helping behavior M = 0.10, moral courage M = 2.60).
In order to test the effects of prototypes on behavioral determinants we activated the caring, the just, or the brave prototype in a between-subjects design, but where the dependent variables were measured by a within-subjects solution. Thus, we were able to compare the prototype influences in the best way, such that when our assumption was correct and the prototypes display a specific influence on emotional precursors, we should find an interaction between the activated prototype and the affected emotional precursors. We, furthermore, expected the caring prototype to affect empathy (more than the just and the brave prototype do), the just prototype to affect JS (more than the caring and the brave prototype do), and the brave prototype to affect fear (more than the caring and the just prototype do).
Forty-five persons (21 women) between the ages of 18 and 34 years (M = 22.80, SD = 2.63) volunteered to take part in the study. The participants were recruited from the university campus and received a course credit for participation.
Procedure and Materials
The participants were given a questionnaire packet by the experimenter who ensured that it was filled out in the prescribed order. After demographic questions the manipulation followed. To activate the moral prototypes, participants had to compare themselves either with an ideally caring person (caring prototype), with an ideally just person (just prototype), or with an ideally brave person (brave prototype). The ideally caring, just, or brave person was described by the first five most typical descriptors of each of Walker and Hennig's (2004) prototypes. Participants had to write down in a space of ten lines, situations when these descriptors had applied to them, and had to list further similarities between themselves and a caring/just/brave person.
After the manipulation, the three dependent variables (empathy, JS, and fear) followed, which were given in six different orders (completely mixed to prevent order effects). Empathy was measured via a procedure adopted from De Wall and Baumeister (2006, Studies 4 and 5): Participants were given a handwritten essay which described someone else's suffering in order to evoke empathic responding. In the essay, a person revealed that s/he was left by her boy/girlfriend and that s/he was, therefore, very depressed. After the essay, participants had to report how sympathetic, warm, compassionate, and softhearted they felt toward the author of the essay (on 7-point scales ranging from 0 = not at all to 6 = absolutely). The internal reliability of the adjectives was good (α = .80). Therefore, an empathy index was created by averaging the responses to the four empathy adjectives. To assess JS, the JS observer scale by Schmitt et al. (2005) was used (α = .83). The scale consists of ten items (to be rated on 7-point scales from 0 = not at all to 6 = absolutely) which ask for participants' emotional reactions (moral outrage) on instances of (in)justice. An example item is: “I am upset when someone is being treated worse than others.” To measure fear in dangerous situations, the fear scale of the Woodard-Pury Courage Scale (WPCS-23) was used (Woodard & Pury, 2007) (α = .90). The fear scale of the WPCS-23 consists of 23 different threats with an important outcome (e.g., If called upon during times of national emergency, I would give my life for my country.) and participants are asked to indicate the level of fear they would feel from that threat on a 7-point scale (0 = not at all, 6 = absolutely).
Results and Discussion
To assess the effects on the dependent variables, a 3 (manipulation: Caring vs. just vs. brave) × 3 (emotional precursor: Empathy vs. JS vs. fear) factorial ANOVA2 with repeated measures on the second factor was conducted. Results revealed significant main effects for manipulation, F(2, 42) = 50.43, p < .001, η2 = 0.71, and for emotional precursor, F(2, 84) = 12.54, p < .001, η2 = 0.23, as well as the predicted significant two-way interaction, F(4, 84) = 9.03, p < .001, η2 = 0.30 (see Figure 3). To illustrate the interaction, three one-way ANOVAs with each emotional precursor as the dependent variable were conducted which all reached significance: Empathy, F(2, 42) = 11.72 p < .001 η2 = 0.36; JS, F(2, 42) = 33.48, p < .001, η2 = 0.62; and fear, F(2, 42) = 43.39, p < .001, η2 = 0.67. Since we had specific hypotheses about the influence of the prototypes on the precursors, planned contrasts were calculated: A planned contrast (coded: 2, −1, −1: Caring prototype vs. just and brave prototypes) for empathy revealed that participants were significantly more empathic when they had thought of the caring prototype compared to participants in the just prototype condition and in the brave prototype condition, t(42) = 4.83, p < .001. A planned contrast (coded: 2, −1, −1: Just prototype vs. caring and brave prototypes) for JS revealed that participants were significantly more justice sensitive when they had been in the just prototype condition compared to participants in the caring prototype condition and in the brave prototype condition, t(42) = 5.13, p < .001. A planned contrast (coded: 2, −1, 1: Brave prototype vs. caring and just prototypes) for fear revealed that participants displayed significantly less fear when they had been in the brave prototype condition compared to participants in the just prototype condition and in the caring prototype condition, t(42) = 5.84, p < .001.
Taken together, our hypotheses were confirmed: Activating the caring (relative to the just and the brave) prototype increased empathy, activating the just (relative to the caring and the brave) prototype promoted JS, activating the brave (relative to the caring and the just) prototype caused people to display less fear in dangerous situations. Thus, it can be concluded that an activation of one of the three moral prototypes specifically affects emotional precursors of moral behavior. Since we could demonstrate a specific connection between the prototypes and an emotional precursor of prosocial behavior, the final step was to activate moral behavior via prototype-activation. This was done in Study 5.
STUDY 5: ACTIVATION OF THE JUST PROTOTYPE—BEHAVIORAL INTENTION AND COMMITMENT TO CONDUCT A BEHAVIOR
The goal of Study 5 was to examine whether the activation of a moral prototype would have a specific influence on the related type of moral behavior, but not on the other behaviors. We decided in favor of the just prototype and the related behavior of moral courage because we wanted to examine a behavior that has been rather neglected in previous research. Furthermore, the activation of moral courage is a more conservative test (compared to the activation of helping behavior) because moral courage includes costs. Heroism also includes costs, but the investigation of heroism in a laboratory setting is hardly possible. Plenty of studies have examined priming effects on low-cost moral behavior like helping behavior or cooperating (e.g., Macrae & Johnston, 1998; Nelson & Norton, 2005). To our knowledge there is no study dealing with the activation of costly moral behaviors in general and moral courage specifically. We assumed that the activation of the just prototype would enhance peoples' willingness to show moral courage; the tendency for helping behavior, in contrast, should not be affected. Thus, we expected a significant interaction between the activation of the prototype and the type of moral behavior that is affected.
Participants and Design
Seventy-two (48 female) persons between the ages of 15 and 69 years (M = 33.90, SD = 14.05) volunteered to take part in the study. The participants were recruited from the university campus or from a park near the university. They were individually approached by an experimenter who was not aware of the hypothesis of the study and asked if they were willing to fill out a short questionnaire. The study was set up as a 2 (manipulation: Prototype activation vs. neutral condition) × 2 (kind of prosocial behavior: Moral courage vs. helping behavior) between-subjects design, and participants were randomly assigned to one of the conditions.
Procedure and Materials
The materials consisted of two seemingly independent studies and a request to show one type of moral behavior, everything combined in one questionnaire packet. When participants agreed to take part, they were handed the questionnaire packet by the experimenter who ensured that it was filled out in the prescribed order. After demographic questions, the manipulation followed. To activate the just prototype, we followed the procedure from Study 4 (participants had to compare themselves with an ideally just person). For the neutral condition, participants were asked to think about their last visit to a supermarket. Following the manipulation, participants completed the Positive Affect Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988) (labeled as Study 2). The PANAS ratings were intended as filler items to distract the participants from the following dependent variable, but also to examine possible mood effects.
Participants were then told that the study was complete, but they were invited to read a flyer that was attached to the questionnaire packet. The experimenter was extensively trained to treat each participant the same way in order to prevent experimenter effects. Participants were told that the following request did not have anything to do with the study, but that the university was urgently searching for volunteers and would therefore appreciate it if the participant would read the following page. This was the dependent variable: The request to show moral courage or helping behavior. This request was adopted from Niesta, Greitemeyer, Fischer, and Frey (2009, Study 1) and modified for the present study. It consisted of two different flyers, each promoted participation in a discussion. The discussion was described as part of a project, run by the university, for improved integration of foreigners. In the course of this project, participants were told that the university was working together with laymen who were to take part in discussions on the topic of an efficient integration process of immigrants in Germany. Participants were further informed that the university was urgently searching for volunteers to take part in discussions with a group of high school students (helping condition) or young right-wing extremist delinquents who were in prison because they had committed offenses against foreigners (moral courage condition). Note that a between-subjects design was employed and thus each participant only received one flyer with only one condition on it.
In the helping condition, the audience (i.e., the high school students) was portrayed as polite, and it could be expected that they would have at least a neutral attitude toward immigrants. If participants chose to take part, they would primarily help the university to run the project. This was quite different in the moral courage condition; here, the discussion was described as potentially confrontational. If participants took part they could face being insulted or even threatened when speaking in favor of foreigners. Extensive pre-testing was performed to test whether participants would label the participation in the different conditions as helping behavior or moral courage, respectively, and to investigate whether participants would associate different levels of costs with the respective participation. Results revealed that the participation in discussions with high school students was perceived as helping behavior and associated with less social costs, whereas the participation in discussions with young right-wing extremist delinquents was labeled as moral courage and associated with much social costs. After the description of the discussions and their alleged tasks, participants were asked on an 11-point scale (0 = not at all to 10 = absolutely), if they were generally willing to participate. Then they were asked to decide if they really wanted to take part and if so, to provide their contact information (mailing address or phone number) as well as the period of time during which they could best assist. To underscore the credibility of the project, the participants were informed that a university staff member would contact them in the next few days to arrange an appointment for the discussions. Note that nowhere on the flyer did the words “helping behavior” or “moral courage” appear. After having read the flyer, participants were probed for suspicion (all subjects had believed that the university was really looking for the volunteers and nobody had assumed any relation between the flyers and the manipulation), extensively debriefed, thanked, and excused.
Results and Discussion
Neither sex nor age of the participants had an effect on any of the dependent variables, nor were there any significant interactions. Thus, these variables were not considered further. Furthermore, neither positive nor negative affect as measured by the PANAS was affected by the manipulation, positive affect: F(1, 69) = 0.50, p = .48, negative affect: F(1, 69) = 0.99, p = .32. Thus, mood cannot account for the following results.
General Willingness to Take Part
A 2 × 2 factorial ANOVA with manipulation (prototype activation vs. neutral condition) and behavior (moral courage vs. helping behavior) as between-subject variables revealed the predicted interaction, F(1, 68) = 16.79, p < .001, η2 = 0.20 (for means see Figure 4). No significant main effects for prototype activation, F(1, 68) < 1, p = .90, or for behavior, F (1, 68) = 1.70, p = .20, emerged. Follow-up analysis for helping behavior showed that participants in the neutral condition were more willing to perform the requested behavior (M = 6.83) compared to the prototype activation condition (M = 3.89), t(39) = 2.90, p < .01, Cohen's d = 0.96. With regard to moral courage, however, participants for whom the just prototype was activated were more willing to act in a morally courageous fashion (M = 7.78) than the participants who had been in the neutral condition (M = 4.82), t(33) = 2.90, p < .01, Cohen's d = 0.98. Thus, the activation of the just prototype affected behavioral intentions in the expected way: The intention to show moral courage was enhanced when participants compared themselves with the just prototype. In the neutral condition, however, participants were more willing to show helping behavior.
Even if our hypothesis was affirmed, a result was quite surprising: For helping behavior, the mean in the neutral condition was significantly higher than in the prototype activation condition. One could think of two different explanations: First, the neutral condition primed helping or second, the prototype activation reduced the willingness to help. We tend to the second explanation since there is no reason to assume that a visit in a supermarket would prime helping. Possibly the activation of a just prototype reduces the willingness to help, when the act of helping does not have to do anything with justice or morality; contrast processes might be the reason (Dijksterhuis & Bargh, 2001), but further studies are needed to test these assumptions.
Given Contact Information
Twelve of the 37 participants in the helping behavior condition gave their contact information, as did 16 of the 35 participants in the moral courage condition. Thus, altogether, 28 participants (39%) wanted to take part in the discussions. A 2 (manipulation: Prototype prime vs. neutral condition) × 2 (behavior: Moral courage vs. helping behavior) × (contact information: Given vs. non given) analysis with a log linear model (LOGIT) revealed a significant interaction between manipulation, behavior, and contact information, Z = 2.74, p < .01, SE = 1.02, Est. = 2.78 (results are displayed graphically in Figure 5). Follow-up analyses for helping behavior revealed no significant interaction between manipulation and the contact information provided, χ2(2, N = 72) = 2.31, p < .13, w = 0.18. Of the 12 participants (32% of the 37 participants in the helping condition) who were willing to perform helping behavior, eight (22%) had been in the neutral and four (10%) had been in the prototype priming condition. In contrast to general willingness, the differences for helping behavior between the manipulation conditions did not reach significance. The picture is different for moral courage: Here, a significant interaction resulted between manipulation and the contact information provided, χ2(2, N = 72) = 6.56, p = .01, w = 0.30. Of the 16 participants (45% of the 35 participants in the moral courage condition) who wanted to show moral courage, four (11%) had been in the neutral condition, and 12 (34%) in the prototype condition. Thus, participants in the prototype condition showed more moral courage by giving their contact information in order to take part in discussions with right-wing extremist delinquents compared to participants in the neutral condition. The results argue in favor of our hypotheses: Activating the just prototype led people to show more moral courage, whereas it did not influence helping behavior.
In sum, Study 5 demonstrates that it is possible to activate one of the prototypes by Walker and Hennig (2004) in peoples' minds, and by doing so, to promote specifically the behavior that the prototype is related to (whereas another moral, prosocial behavior that is associated with a different prototype is not affected).
In the present studies, we related the moral prototypes of Walker and Hennig (2004) to moral behavior. In Studies 1–3, specific prototype-behavior correspondences were addressed: The first study demonstrated that free-listed characteristics of a person who acts helpful, morally courageous, or brave were related to the descriptors of the moral prototypes in the expected way. In order to replicate the results of Study 1 in Study 2, a typicality rating procedure was used and again the expected associations emerged. Study 3 focused on concrete behaviors by using scenarios and participants were asked to assign which behavior they would expect from the respective prototype. As predicted, participants primarily expected the caring prototype to help, the just prototype to show moral courage, and the brave prototype to act heroically (always compared to the other prototypes). The specific prototype-behavior associations were used in the fourth and the fifth studies to purposefully activate specific precursors of moral behavior and moral behavior itself. In the fourth study, it was demonstrated that the activation of the moral prototypes specifically affects emotional precursors of moral behavior. Finally, in the fifth study, participants for whom the just prototype was activated showed more moral courage. Helping behavior, however, was not affected. Thus, we were able to promote specific behavioral determinants and the willingness to show a specific (not just any) moral behavior. Taken together, the results of our five studies support our assumptions: The moral prototypes of Walker and Hennig (2004) include specific behavioral information, and because of the included behavioral information, the activation of the prototypes leads to specific behavioral effects.
The results of the first three studies are in line with previous research that demonstrated that prototypes include behavioral information (Cantor & Mischel, 1979; Carver & Scheier, 1981). Going beyond the previous findings, we demonstrated specific associations between certain moral prototypes and moral behaviors (and not only a general prototype-behavior connection). In doing so, the argumentation of Walker and Hennig (2004) was supported and expanded upon: These authors showed on the level of descriptors and traits that people separate the three moral prototypes; we demonstrated that this trichotomy also applies to the level of behaviors. It is not the moral character that shows the moral (or prosocial) behavior, but rather distinct moral prototypes that display different kinds of moral (or prosocial) behaviors. Moreover, we were able to specifically affect moral behavior by activating the related prototype. Studies already exist in the area of health psychology that have demonstrated effects of prototypes on behavior (e.g., Oullette et al., 2005). To our knowledge, our studies are the first to activate moral behavior via moral prototypes. Moreover we demonstrated for the first time a specific influence of prototypes on behavior: If one plans to evoke a specific (and not just any) moral behavior, the associated (and not just any) prototype should be activated.
Limitations and Future Research
One limitation of the studies presented here is that we only investigated three types of moral, prosocial behaviors. Moral or prosocial behavior, however, encompasses a broad range of actions of which helping behavior, moral courage, and heroism are only a subset (Batson, 1998). It would also be interesting to include behaviors like volunteerism or social activism and to assess to which prototypes these would be attributed.
Research in the area of health psychology has demonstrated that even when taking variables from the theory of planned behavior (Ajzen, 1991) and past behaviors into account, prototype perception still significantly enhances the prediction of behavioral intentions (Rivis & Sheeran, 2003; Rivis, Sheeran, & Armitage, 2006). Prototype perception is an important component of the PWM of adolescent health-risk behavior (Gibbons et al., 1998). According to the PWM, two aspects of prototype perception influence health-related decisions: The prototype evaluation (the degree of liking one has for the prototype in question) and prototype similarity (similarity of the prototype to oneself). Prototype perceptions influence people's behavior through the process of social comparison whereby individuals compare themselves with the prototype and its attributes. Since we did not explicitly include prototype perception in our studies, an interesting question would be whether prototype perception not only plays a role in determining health-risk behavior, but moral behavior as well.
Another promising research topic could be derived from the prototype-matching theory by Niedenthal et al. (1985; see also Setterlund & Niedenthal, 1993). Prototype matching is a cognitive strategy for choosing among situations; it involves a comparison of the attributes of the self with the attributes of the typical person associated with each of the situations under consideration, and then choosing the situation in which the self-prototype match is the closest. It would be interesting to examine whether people primarily display the moral behavior for which a match between the corresponding prototype and themselves exists.
Conclusions and Implications
The results of our studies support the involvement of personality and traits as well as the inclusion of lay theories in moral psychology. By relying on lay theories and asking people what moral/prosocial behavior they associate with the prototypes, we were able to demonstrate specific associations between the prototypes and moral behavior, and thus to expand the moral prototypes from the descriptor to the behavioral level. Moreover, we could use these associations to influence determinants of behavior, behavioral intentions, and behavior itself. The use of prototypes and the involvement of personality seem to be good complements to cognitive-developmental approaches that are based on the work of Kohlberg (1984), especially if one plans to activate behavior. As already mentioned, moral reasoning ability alone appears to be weak and inconsistent predictor of moral behavior, once intelligence has been partialled out (Haidt, 2003). We often do not have the time or the cognitive resources to think long, elaborately, and reasonably about a moral problem. Moral exemplars, however, as layperson concepts, quickly come to our mind and have clear relations to behavior. These moral exemplars do not need famous persons like Martin Luther Kind or Mother Teresa; they live among us, as Walker and Firmer (2007) demonstrated. With a moral prototype as model, we might know faster how to behave “good”, “moral”, or “right”. By the use of prototypes, people can be influenced to think and act in a more moral fashion. Thus, prototypes might be utilized in public campaigns that aim to promote moral courage or in moral courage trainings and workshops.
Taken together, the studies conducted here present another case for the trichotomy of moral exemplarity in the caring, just, and brave prototypes since they are not only separable on the level of descriptors, but also on the behavioral level. These prototype-behavior associations add an interesting aspect to the area of moral psychology because they can be used to activate behavioral determinants as well as behavior itself, which in turn contributes to a fuller and more balanced account of moral functioning.
As the terms “moral” and “prosocial” are often used synonymously (see e.g., Fabes, Gustavo, Laible & Kupanoff, 1999); we also did so in the present manuscript.
Sex and age of the participants as well as order did not produce any significant main effects or interactions with any of the dependent variables. Thus, these variables were not considered further.