Abstract: Fairness is undoubtedly an essential normative concept in humans and promotes cooperation in human societies. The fact that fairness exists is puzzling, however, because it works against the short-term interest of individuals. Theories of genetic evolution, cultural evolution, and gene-culture coevolution identify plausible mechanisms for the evolution of fairness in humans. Such mechanisms include kin selection, the support of group-beneficial moral norms through ethnic markers, free partner choice with equal outside options, and free partner choice with reputation as well as spite in small populations. Here, we present the results of a common-pool resource game experiment on sharing. Based on data from 37 multiethnic villages in a subsistence agricultural population in Foutah Djallon, Guinea, we show that fair behavior in our experiment increased with increasing ethnic homogeneity and market integration. Group size and kinship had the opposite effect. Overall, fair behavior was not conditional on reputation. Instead, the ability of the different village populations to support individuals’ fairness in situations lacking the opportunity to build a positive reputation varied significantly. Our results suggest that evolutionary theory provides a useful framework for the analysis of fairness in humans.
Discussion
In our experiment, fairness was influenced by some mechanisms that have been
suggested in the literature. First, as expected (Table 1), the subjects’ fairness was
positively influenced by the ethnic homogeneity of villages (Table 3). The predicted
behavior of subjects was on average only fair in ethnically homogeneous villages (Fig.
3) and became more selfish with increasing ethnic heterogeneity. This result corroborates the existing empirical evidence regarding the influence of ethnic homogeneity on
prosociality (Alesina and La Ferrara 2000; Alesina et al. 1999; Anderson and
Paskeviciute 2006; Costa and Kahn 2003; Gustavsson and Jordahl 2008;
Habyarimana et al. 2007; Leigh 2006; Miguel and Gugerty 2005; Newton and
Delhey 2005; Pennant 2005; Putnam 2007).
Second, as expected (Table 1), the subjects’ behavior in our experiment became
increasingly fair with increasing market integration (Table 3, Fig. 3). This result
confirms the positive effect of market integration on fairness found in other experimental studies (Emsinger 2004; Henrich et al. 2001, 2010).
Third, as expected (Table 1), the capacity of villages to support fairness decreased
with population size (Table 3). The behavior of subjects was on average fair only for
villages below 200 individuals and became selfish with increasing population size (Fig.
3). This result adds to the inconsistent experimental evidence related to the effect of
population size on fairness (Henrich et al. 2010; Stahl and Haruvy 2006).
On the other hand, for some other mechanisms generally considered to be important,
we found contradicting evidence. Surprisingly, the predicted positive effect of kinship
on the fairness of subjects (Table 1) was not confirmed: subjects in our experiment
became more selfish with increasing kinship (Table 3). Predicted behavior was on
average only fair in the village populations with the lowest kinship (Fig. 3). This
finding does not match the results of previous studies: although the extensive foodsharing literature widely supports a positive effect of kinship on fair or generous
sharing (Essock-Vitale and McGuire 1980; Gurven 2004; Gurven et al. 2000, 2001;
Nolin 2011; Patton 2005; Wiessner 2002; Ziker and Schnegg 2005), the experimental
literature does not find convincing evidence for any effect of kinship on fairness (Barr
2004; Macfarlan and Quinlan 2008). Interestingly, our measure for kinship at the
individual level, the proportion of the subjects’ kin group in the village population,
did not provide any additional insights into the relationship between kinship and
fairness (Table 3). This means that we cannot resort to mechanisms at the individual
level of the subjects’ kinship to explain this unexpected finding. Two such plausible
mechanisms could be the following: in villages with high kinship at the aggregate level,
either subjects with high individual kinship took more than their share to redistribute it
to kin afterwards, or, from the opposite point of view, subjects with low individual
kinship took more than their share because they were not related to the majority of the
other inhabitants of the village. Both would imply significant estimates of the measure
of kinship at the individual level of the subjects (kin group): the first mechanism would
imply a positive estimate, the second a negative one.
Finally, lack of an opportunity for reputation building did not lead to a
general increase in selfish behavior (Table 1). Instead, the effect on fairness of
such an opportunity varied significantly, depending on the village. We found
the expected increase in selfish behavior in two thirds of the villages. Some individuals were extremely selfish in the treatment condition without observation. Interestingly, our experimental treatment also led to fairer, or even generous behavior (Table 3; A19 in the ESM). This result is not consistent with the
broad experimental literature supporting the notion that observability of behavior triggers prosociality (Bereczkei et al. 2007, 2010; Bull and Gibson-Robinson
1981; Kurzban 2001; Milinski et al. 2002; Satow 1975; Soetevent 2005). We
can only speculate about reasons for this finding. From a methodological point
of view, we can think of two explanations for this finding. First, our experimental stimulus may not have been only a measure of reputation, but also of
the willingness to defy instructions or the fear for punishment. Those different
dimensions might have interacted and led to such a varying effect. Second, our
experimental stimulus may not have worked as expected in all the villages. The
villages where our experimental treatment worked as expected must have
differed in some ways we did not control for from the villages where the
treatment did not work as expected. In theory, an important factor is the feeling
of privacy of the subjects, not only within the village communities in general,
but also after the experiment, when leaving the hut with the salt: in some
instances, others were waiting outside for their turn; in other cases, nobody was waiting. Unfortunately, we did not record this information. The random slopes
for reputation (A19 in the ESM) and the amount of salt (A21 in the ESM) did
capture such situational differences to some extent.
The results of our analysis show that the behavior of subjects in our
experiment was mainly driven by the characteristics of their village (Table 3,
Figs. 2 and 3). Unlike in previous studies (Cronk 2007; Henrich 2000; Paciotti
and Hadley 2003; Roth et al. 1991; Tracer 2003), we did not find an effect of
the subjects’ ethnicity on fairness. We believe that this is because we sampled
several villages containing the studied ethnic groups. Those villages represented
a broad range of different ethnic responses (Fig. 1, Table 2). Furthermore, we
also controlled for the characteristics of the villages our subjects lived in
(Table 3). If we had only selected one Malinke and one Fulbe village for our
experiment (e.g., Boubere and Beleya Koko), we would not have been able to
control for village characteristics and would have estimated a highly significant
effect of ethnicity on fairness. We would have concluded that Malinke are
much more selfish than Fulbe (A24 in the ESM) and probably would have
speculated about the reasons for this ethnic difference in behavior. This finding
highlights the necessity of sampling a multitude of populations for each cultural
group of interest and controlling for contextual factors when conducting cultural
comparative studies (Lamba and Mace 2011; Oosterbeek et al. 2004).
It is obvious that the social structure of the villages (kinship, ethnic homogeneity, market integration, population size, income inequality) shaped the
behavior of subjects (Fig. 3). More research is needed to uncover the underlying mechanisms that work on these variables. This is especially true for kinship
and ethnic homogeneity. Whereas the behavior of subjects in our experiment
was significantly influenced by kinship and ethnic homogeneity at the aggregate
level (the average kinship and the homogeneity of the village), measures at the
individual level (the proportion of subjects’ kin or ethnic group to the total
population) did not influence the behavior of subjects in our experiment
(Table 3). This finding is puzzling. Although the individual and the aggregate
measures correlate strongly (A16), only the aggregate measures influence individual behavior. A similar pattern was found in a study by Putnam (2007):
Diversity does not produce “bad race relations” or ethnically-defined group hostility, our findings suggest. Rather, inhabitants of diverse communities tend to withdraw from collective life, to distrust their neighbours, regardless of the colour of their skin, to withdraw even from close friends, to expect the worst from their community and its leaders, to volunteer less, give less to charity and work on community projects less often, to register to vote less, to agitate for social reform more, but have less faith that they can actually make a difference, and to huddle unhappily in front of the television. Note that this pattern encompasses attitudes and behavior, bridging and bonding social capital, public and private connections. Diversity, at least in the short run, seems to bring out the turtle in all of us (2007:150–51, italics in original).
Finally, we were dealing in our study with traditional, small-scale societies that are partly in transition to more complex and socially stratified societies. The different villages showed substantial variations in population characteristics, and also the inhabitants of the villages differed in key characteristics, although they were all located in the same area (Fig. 1, Table 2). We were able to include a substantial part of this population in our experiment, allowing us to test some general hypotheses related to the biological drivers of human social behavior. We believe that our study population and design allows us to put our results in a broader human context. Some scholars have argued that humans are a unique species because our altruism is not primarily based on kinship. The evolution of fairness and cooperation in large-scale societies cannot be explained solely by genetic evolution and is best accounted for with cultural evolution and gene-culture coevolution. The empirical evidence for these claims is based to an important extent on behavioral experiments (Fehr and Fischbacher 2003; Gintis et al. 2003; Henrich et al. 2010). However, most of these experiments do not include measures of kinship in their analysis. In our experiment, we found a statistically significant effect of kinship on fair behavior. We also found significant effects for all other discussed mechanisms that favor fairness through natural selection. We are therefore inclined to conclude that evolutionary biology is a useful tool to analyze human social behavior and explains the behavior of subjects in our experiment. However, we are well aware that cultural evolution makes the same predictions related to the effect of population size (Forber and Smead 2014; Huck and Oechssler 1999), market integration (Hoel 1987; Roth and Erev 1993; Rubinstein 1982), and reputation (Chiang 2008) as biological evolution. Similarly, ethnic markers can be conceived of as biological, cultural, or both (McElreath et al. 2003), and even kinship has an important cultural dimension (Jones 2000). Moreover, economic game theory, which makes use of rationality and utility maximization assumptions, comes to similar conclusions as evolutionary game theory (Skyrms 1994, 2000). Culture and nature are closely intertwined in humans. More research must therefore be conducted to develop tools allowing to empirically tease apart their contributions to human social behavior in samples that are not made up of twins. Considering the relevance of experimental game theory in this domain of research, these tools should be compatible with a gametheoretical experimental approach.
No comments:
Post a Comment