Social Reasoning Affected by Rank (Mealey, Daood, Krage, 1996)
KeywordsSource Memory Social Rank Dictator Game Social Reasoning Norm Violation
The influence of the social status of a person within a group or society on cheater detection and memory for cheaters.
Most human societies and primate groups are pervaded by social dominance hierarchies. It is therefore plausible to assume that our cognitive system is shaped by selective pressures that require the individuals to adapt to the hierarchical organization of these social environments. Specifically, dominance theory (Cummins 1998, 1999, 2005) implies that the reasoning about social norms is affected by social rank. This hypothesis seems quite reasonable considering that social dominance hierarchies can be defined as a set of explicit or implicit social norms that define who is allowed to gain access to valuable resources (such as mates or food) and who is not. A social dominance hierarchy can only be maintained if the individuals in the group learn and follow the explicit or implicit rules that determine which behaviors are permitted, prohibited, or obligated for each individual, depending on the individual’s social rank (Cummins 1998, 1999, 2005). This theory leads to the hypothesis that social rank may affect social reasoning (Cummins 1999) and, in particular, memory for cheaters (Mealey et al. 1996). This hypothesis needs to be empirically tested.
Following norms and helping others is one way to build alliances that help to gain and to maintain a high social rank in a group or society. High-ranking individuals have to share resources and have to fulfill reciprocal obligations to avoid losing important allies and, thereby, their status. Individuals with a low social rank, in contrast, may try to increase their limited access to valuable resources by cheating and by violating the social norms that deprive them of those resources. Given that the individuals with high social rank usually benefit most from the organization of the society, they are often responsible for enforcing the social norms that grant them privileged access to resources and for punishing the violation of these norms. Low-ranking individuals, in contrast, are lacking the resources to punish high-ranking individuals. Based on this reasoning, dominance theory (Cummins 1998, 1999, 2005) predicts that high-ranking individuals monitor the behavior of low-ranking individuals and are highly vigilant against norm violations of low-ranking individuals, but not vice versa. In fact, the theory implies that cheater detection is a cognitive adaptation that serves to maintain social dominance hierarchies. A clear prediction that can be derived from this framework is that people should be more prone to detect the cheating of lower-ranking individuals in comparison to the cheating of higher-ranking individuals (Cummins 1999).
Although the above hypothesis is well grounded in a functional-evolutionary framework, it would not be right to uncritically accept this hypothesis based on plausible-sounding evolutionary arguments alone. Hypotheses in evolutionary psychology are tested in a similar manner as hypotheses in other areas of psychology and have to receive support from empirical tests before they should be accepted. It is useful to be aware that alternative hypotheses about the relationship between social rank and social reasoning are possible as well. A possibility that should be seriously considered is that social norms apply equally to high-ranking and low-ranking individuals. Consider, for example, the dictator game: In this game, the distribution of power is maximally unequal. One player, the dictator, receives a certain amount of money and has to decide how to split this money between him- or herself and the recipient. The recipient is entirely powerless and has to accept the offer of the dictator no matter how small. The dictator game can be viewed as a model of situations with extreme differences in social rank because the dictator has access to all of the resources and the recipient has nothing. Nevertheless, people generally judge an equal split between the two players to be most socially appropriate (Krupka and Weber 2013), suggesting that fairness norms are equally applicable to persons with high and low social rank. It is even possible to postulate that cheater detection should be higher when tracking the wrongdoings of high-ranking individuals (e.g., politicians with high political power), which would be directly opposed to the predictions outlined above. Some studies have indeed found positive associations between social rank and norm violations (Piff et al. 2012; but see Francis 2012). Despite this evidence to the contrary, high-ranking individuals (e.g., professors) are often expected to be moral role models for lower-ranking individuals (e.g., students) and may therefore be judged by harsher standards. Accordingly, it is also possible to postulate that high-ranking individuals should show a noblesse oblige effect and should be more generous and forgiving against low-ranking individuals than vice versa (Fiddick et al. 2013). Hence, the hypothesis that people should be more vigilant against the cheating of low-ranking individuals in comparison to that of high-ranking individuals is a plausible possibility, but the opposite is equally plausible.
Is Social Reasoning Affected by Social Rank?
In Cummins’ (1999) deontic version of the Wason Selection Task, participants were required to test whether a resident assistant or a student showed compliance with the rule “If someone is assigned to lead a study session, that person is required to tape-record the session.” In a control condition, participants were told to imagine that they overheard a resident assistant’s or student’s statement “If I am assigned to lead a study session, I always tape record the session.”, and were asked to test the truth of the statement. Half of the participants were cued into the perspective of a resident assistant, and the other half were cued into the perspective of a student. Half of each group was instructed that they were checking on a resident assistant, and the other half were told that they were checking on a student. Reasoning was affected by rank, but only in the deontic version of the task. Participants were more likely to show evidence of a “look-for-cheaters” strategy when they were cued into the perspective of a resident assistant checking on students (65 % correct answers) in comparison to all other conditions (15–20 % correct answers). This confirms the prediction of dominance theory that high-status individuals are vigilant against norm violations of low-status individuals. Cummins (1999) interpreted these findings as evidence of a biological adaptation to social dominance hierarchies.
However, a plausible alternative to this interpretation is that Cummins’ (1999) manipulation had served as an implicit cue to monitor the compliance with the rule. Enforcing social norms and confronting norm violations of students is an integral part of the role of a resident assistant, but not of fellow students. It seems possible that the effect was not due to differences in social rank but was rather due to the participants being aware of the fact that resident assistants are supposed to monitor the students’ compliance with norms. It seems therefore unnecessary to postulate a domain-specific mechanism for reasoning about social dominance hierarchies when the results can be attributed so easily to general learning mechanisms that allow individuals to acquire knowledge about the world, including knowledge about specific behaviors required by social roles. Therefore, additional research is needed to test whether these effects are indeed due to the influence of social rank rather than to the specific scenario realized in the study of Cummins (1999).
In a follow-up review, Fiddick and Cummins (2001) summarized several studies in which social status was unintentionally manipulated. Specifically, Gigerenzer and Hug (1992) required participants to take the perspective of high-ranking or low-ranking individuals when reasoning about rules. For example, participants were cued into the perspective of an employer or of an employee when reasoning about the rule “If an employee works on the weekend, then that person gets a day off during the week.” Fiddick and Cummins note that the findings reported by Gigerenzer and Hug (1992) do not provide consistent evidence of an influence of social rank on social reasoning. There was no significant effect of social rank on performance in the day-off problem (75 % detection of high-status cheaters and 61 % detection of low-status cheaters) and in a similar pension problem (64 % detection of high-status cheaters and 70 % detection of low-status cheaters). Only in one of three problems, the subsidy problem (“If a home owner gets a subsidy, then that person must have installed a modern heating system”), a significant effect of perspective was found (81 % detection of low-status cheaters and 59 % detection of high-status cheaters) although the subsidy problem was the problem with the least pronounced differences in social rank (house owner vs. environmental officer). However, the latter effect should not be over-interpreted given that Holyoak and Cheng (1995) have obtained no effect of social rank using the same scenario.
Fiddick and Cummins (2001) used another variant of the Wason Selection Task in which they required participants to look for violations of the rule “If you pay for the gas, I’ll drive you to work.” Participants were either asked to take the perspective of a high-ranking individual (a boss) or of a low-ranking individual (an employee). The study demonstrated a high degree of intercultural variability in the effects of social rank on performance in the Wason Selection Task. In Germany, no significant effects of rank were obtained. In California, participants were better when monitoring the compliance of high-status individuals (70 % correct answers) than when monitoring the compliance of low-status individuals (45 % correct answers). This effect is in the opposite direction of what was predicted by Cummins (1999). The high cultural variability in the findings that has been identified by Fiddick and Cummins (2001) is challenging for the interpretation of the findings in terms of an evolutionary adaptation.
In another series of experiments, Fiddick and Cummins (2007) used a ledger task in which they confronted participants with different degrees of compliance on the part of the gas-paying partner and required them to indicate (1) how likely they would want to continue the car-sharing arrangement and (2) to what degree the arrangement was seen as fair or unfair. Again, participants were required to take the perspective of either a high-ranking individual (a boss) or a low-ranking individual (his or her employee). Participants cued into the perspective of a high-ranking individual were more tolerant of cheating than participants cued into the perspective of a low-ranking individual, which is again inconsistent with the original hypothesis that high-ranking individuals should be more, not less, vigilant against cheating of lower-ranking individuals (Cummins 1999). The results seem to be quite robust (Fiddick et al. 2013), but they are difficult to interpret. Fiddick and Cummins (2001) explained the discrepancy to earlier findings by arguing that the earlier findings (Cummins 1999) are about monitoring the compliance with social norms, while the car-sharing scenario is about a reciprocal personal exchange. However, it is questionable whether these two can be clearly separated when reasoning about work-related activities that concern the employer-employee relationship. Somewhat puzzling, the noblesse oblige effect could not be attributed to an asymmetrical perception of the costs and benefits of the car-sharing arrangement for the employer and the employee, and it is unrelated to differences in income. When the individuals in the scenario were described as boss and employee, but not each other’s boss and employee, the effect vanished. This suggests that the underlying causes for the effect may lie in the specific relationship between a boss and his or her employee when performing work-related activities (paying the other person money for getting back and forth to work) rather than in differences between high-ranking and low-ranking individuals per se. It is an open empirical question whether these findings would generalize to other agreements between a boss and his or her employee and other asymmetrical social relationships.
Is Memory for Cheaters Affected by Social Rank?
Given that the research in the Wason Selection Task does not provide a clear picture, it seems necessary to test the influence of social rank on cheater detection in other paradigms. Fortunately, such evidence is available. Social rank was manipulated in studies of memory for cheaters, which provides an additional test of the hypothesis that processing information about cheaters is affected by social rank. In their pioneering study, Mealey et al. (1996) examined old-new face recognition for faces of cheaters, neutral, and trustworthy persons. They were primarily interested in testing the hypothesis that memory for cheaters is enhanced over memory for other types of faces, but they also included a manipulation of social rank by pairing the faces with high-status professions (e.g., a bishop) or low-status professions (e.g., a cashier). Interestingly, their initial hypothesis was that faces of high-ranking cheaters should be attended, encoded, and recognized with a higher probability than the faces of low-ranking cheaters. Their reasoning was that the behavior of more powerful individuals is less constrained by situational factors, which means that the cheating of those individuals can be more readily attributed to internal character flaws. One may add that the cheating of a high-ranking individual (e.g., of an influential politician) may have more devastating consequences than the cheating of a low-ranking individual, simply because high-ranking individuals are more influential, and their decisions may be more likely to affect a larger number of people to a greater extent than those of low-ranking individuals.
Inconsistent with this a priori hypothesis, Mealey et al. (1996) found that the interaction was in the opposite direction of what was predicted. Furthermore, the results are less compelling than often claimed. Only the recognition of individuals with a low social rank was modulated by whether the face was paired with cheating, neutral, or trustworthy behavior. Even though this effect is often interpreted as evidence for (far) better memory for faces of cheaters (Cummins 1998), the difference between the cheater condition and the neutral control condition was much smaller than the difference between the neutral control condition and the trustworthy condition. Post hoc tests were not reported, but it is possible to conclude from the descriptive pattern of results that the effect was mainly driven by a decreased recognition of trustworthy individuals with a low social rank rather than by an enhanced recognition of cheaters with a low social rank, relative to a low-status neutral control condition, which is hard to reconcile with a functional perspective on memory for social interaction partners. As a post hoc explanation for the absence of any effect of behavior on the recognition of individuals with a high social rank, Mealey et al. proposed that the high attractiveness of the high-ranking individuals may have caused participants to ignore their untrustworthy behaviors. This interpretation was seemingly supported by the fact that female participants showed a descriptive tendency toward a better recognition of trustworthy individuals with a high social rank, but this finding cannot be meaningfully interpreted because the three-way interaction between social status, behavior, and participant gender was not significant. Even if this problem is ignored, it remains debatable whether women should ignore the untrustworthy behaviors of attractive mating partners entirely. More important in the present context is that the pattern of results reported by Mealey et al. has been cited as supporting evidence for the assumption of dominance theory (Cummins 1999) that the cheating of low-ranking individuals should be monitored more closely (Cummins 1998, 1999, 2005).
Given the unexpected nature of Mealey et al.’s (1996) findings, it seems problematic to base strong conclusions on this single study. Fortunately there are several attempts to reproduce Mealey et al.’s findings that provide a clearer picture. Barclay and Lalumière (2006) repeated the study in two experiments (N = 179 and N = 49) and found no effect of social status, no effect of behavior, and no interaction between these two variables on face recognition in either of the experiments. Similarly, Mehl and Buchner (2008) found no evidence of a recognition advantage for faces of cheaters, no effect of social status, and no interaction between these variables in two experiments (N = 96 and N = 123). In a third experiment (N = 64), they found a small effect of social status (the faces of high-ranking individuals were recognized somewhat better than the faces of low-ranking individuals), but the effect was descriptively small. The interaction between social status and behavior was not replicated once again. In light of these failed replications, the findings of Mealey et al. should not be regarded as convincing evidence in favor of dominance theory (Cummins 1999).
Buchner et al. (2009) have argued that the null findings summarized in the previous paragraph are not necessarily problematic for any theory about cheater detection because recognizing faces as familiar cannot help to avoid cheaters and therefore does not provide an evolutionary benefit when the source or context in which the face was encountered is forgotten. What is required instead is the ability to remember that a particular person was associated with a history of cheating. Indeed, source memory for faces of cheaters – that is, the ability to remember that an individual’s face belongs to a cheater – was found to be enhanced in comparison to source memory for trustworthy or irrelevant individuals (see Bell and Buchner 2012, for a review). Most relevant in the present context, the source memory advantage for cheaters was not affected by social rank: Source memory did not differ between high-ranking and low-ranking individuals at all. This finding provides compelling evidence that people are equally interested in the cheating of high-ranking and low-ranking individuals.
In essence, then, there is as yet no reliable evidence that reasoning about social norms or memory for cheaters is affected by social rank. One study (Cummins 1999) demonstrates that high-ranking individuals may be more vigilant against the cheating of low-ranking individuals, but this finding can be attributed to the fact that some social roles with leadership status imply the duty to enforce social norms on others, which may affect the behavior of participants assigned to those roles accordingly. Other studies (Fiddick and Cummins 2007; Fiddick et al. 2013) show that high-ranking individuals are more tolerant of the cheating of low-status persons, but this finding may have to be attributed to the specific relationship between a boss and his or her employee rather than to status differences per se (Fiddick and Cummins 2007). Other lines of research either show a highly inconsistent pattern (Fiddick and Cummins 2001) or have demonstrated that social rank has no effects at all (Barclay and Lalumière 2006; Buchner et al. 2009; Mehl and Buchner 2008). In sum, the results seem to suggest that the relationship between social rank and social reasoning is determined by the specific situation. Although the idea that people may be biologically prepared to deal with social dominance hierarchies sounds quite plausible, the available evidence suggests that this does not necessarily include specialized modules for reasoning about norm violations or memory for cheaters. More research is necessary to gain a deeper understanding of the conditions under which the processing of exchange-related information is affected by social hierarchies.
Effect of Status on Social Reasoning (Cummins 1998)
- Cosmides, L., & Tooby, J. (2005). Neurocognitive adaptations designed for social exchange. In D. M. Buss (Ed.), The handbook of evolutionary psychology (pp. 584–627). Hoboken: Wiley. doi:10.1002/9780470939376.ch20.Google Scholar
- Cummins, D. D. (1998). Social norms and other minds: The evolutionary roots of higher cognition. In D. D. Cummins & C. A. Allen (Eds.), The evolution of mind (pp. 30–50). New York: Oxford University Press.Google Scholar
- Cummins, D. D. (2005). Dominance, status, and social hierarchies. In D. M. Buss (Ed.), The handbook of evolutionary psychology (pp. 676–697). Hoboken: Wiley.Google Scholar